MEASURING BRAND IMAGE: PERSONIFICATION AND NON ...

MEASURING BRAND IMAGE: PERSONIFICATION AND NON-PERSONIFICATION METHODS

A thesis submitted to the University of Manchester for the degree of Doctor of Philosophy

in the Faculty of Humanities

2018

Melisa Mete

Alliance Manchester Business School

2

Table of Contents

List of Figures ................................................................................................................ 5

List of Tables ................................................................................................................. 6

List of Abbreviations ................................................................................................... 10

Abstract ........................................................................................................................ 12

Declaration ................................................................................................................... 13

Copyright Statement .................................................................................................... 14

Acknowledgement ....................................................................................................... 15

Chapter 1: Introduction ................................................................................................ 16

Research Motivation and Research Design ............................................................. 17

Overview of the Thesis ............................................................................................ 18

Measuring Brand Image: Personification versus Non-Personification Methods . 18

How Best to Measure Employer Brand Image: Personification versus Direct

Method ................................................................................................................. 19

Measuring Brand Image and the Role of Task Difficulty .................................... 21

Thesis Format and Structure .................................................................................... 22

Chapter 2: Brand Image and its Measurement ............................................................. 24

The Notion of Brand Image ..................................................................................... 24

Approaches for Measuring Brand Image ................................................................. 31

The Usage of Brand Personality .............................................................................. 33

Chapter 3: Methodology .............................................................................................. 38

Introduction .............................................................................................................. 38

Research Design and Procedure ............................................................................... 40

Sampling/Data Collection Methods ......................................................................... 43

Statistical/Analytical Techniques and Statistical Software ...................................... 45

Reliability and Validity ............................................................................................ 45

Limitations ............................................................................................................... 46

3

Chapter 4: Measuring Brand Image: Personification versus Non-Personification

Methods ........................................................................................................................ 47

Abstract .................................................................................................................... 48

Introduction .............................................................................................................. 49

Brand Image and Personality ............................................................................... 50

The Stereotype Content Model and Signaling Theory ........................................ 51

Research Method and Hypotheses ........................................................................... 54

Methodology ............................................................................................................ 61

Results and Discussion ............................................................................................ 61

Managerial Implications .......................................................................................... 72

Conclusions and Further Work ................................................................................ 74

Chapter 4.1: Connecting Sub-Chapter 1: Changing Context from Product and

Corporate Brand to Employer Branding when Measuring Brand Image .................... 77

Chapter 5: How Best to Measure Employer Brand Image: Personification versus

Direct Methods ............................................................................................................. 78

Abstract .................................................................................................................... 79

Introduction .............................................................................................................. 80

The Advantages of Employer/employee Branding .............................................. 81

Research Method and Hypotheses ........................................................................... 84

Study 1 ................................................................................................................. 86

Study 2 ............................................................................................................... 103

Managerial Implications ........................................................................................ 115

Conclusions and Further Work ............................................................................... 116

Chapter 5.1 Connecting Sub-Chapter 2: Introduction to Task Difficulty .................. 119

Chapter 6: Measuring Brand Image and the Role of Task Difficulty ........................ 121

Abstract .................................................................................................................. 122

Introduction ............................................................................................................ 123

Literature Review ................................................................................................... 123

Task Difficulty: an Education Perspective ........................................................ 123

The Market Research Perspective on Task Difficulty ....................................... 131

Hypotheses ............................................................................................................. 135

Study 1 ............................................................................................................... 138

4

Study 2 ............................................................................................................... 148

Managerial Implications ........................................................................................ 156

Conclusions and Further Work .............................................................................. 157

Appendix ................................................................................................................ 161

Chapter 7: Conclusion ................................................................................................ 162

References .................................................................................................................. 166

Appendices ................................................................................................................. 196

Appendix 1 Questionnaires .................................................................................... 196

Appendix 1.1.1. M&S Direct Questioning Used ............................................... 196

Appendix 1.1.2. M&S Personification Used ..................................................... 200

Appendix 1.1.3. Pantene Direct Questioning Used ........................................... 203

Appendix 1.1.4. Pantene Personification Used .................................................. 207

Appendix 2. Fisher’s R to Z transformation Tables………………………………...211

Word Count: 37251

5

List of Figures

Chapter 4. Figure 1. Warmth, Competence and Status Dimensions with all Items ..... 70

Chapter 5. Figure 1. The final Model for Warmth Dimension and Its Standardized

Regression Weights ..................................................................................................... 99

Chapter 5. Figure 2. The final Model for Competence Dimension and Its Standardized

Regression Weights ..................................................................................................... 99

Chapter 5. Figure 3. The final Model for Status Dimension and Its Standardized

Regression Weights ................................................................................................... 100

Chapter 5. Figure 4. Warmth Dimension ................................................................... 112

Chapter 5. Figure 5. Competence Dimension with Covariances ............................... 113

Chapter 6. Figure 1. Means Plot for Task Difficulty Score and Age of Respondents

.................................................................................................................................... 144

Chapter 6. Figure 2. Means Plot for Task Difficulty Score and Education ............... 145

Chapter 6. Figure 3. PROCESS Macro Model 1, where Task Difficulty is M, Warmth

or Competence is X, and Satisfaction is Y. ............................................................... 146

Chapter 6. Figure 4. Means Plot for Task Difficulty Score and Age of Respondents

.................................................................................................................................... 153

Chapter 6. Figure 5. Means Plot for Task Difficulty Score and Education ............... 154

6

List of Tables

Chapter 3. Table 1. True Experimental Designs .......................................................... 40

Chapter 4. Table 1. Questionnaire Type Distribution According to Gender and Brand

...................................................................................................................................... 57

Chapter 4. Table 2. Cronbach’s Alpha Values of Dimensions by Groups .................. 61

Chapter 4. Table 3. Means and Levene’s Test for Equality of Variance Values for

Each Group and Dimension ......................................................................................... 62

Chapter 4. Table 4. Adjusted R-Square Values of Dependent Variables by Context . 63

Chapter 4. Table 5 A. Chow Test for Each Dimension and Method when Predicting

Satisfaction ................................................................................................................... 63

Chapter 4. Table 5 B. Chow Test for Each Dimension and Method when Predicting

Purchase ....................................................................................................................... 64

Chapter 4. Table 5 C. Chow Test for Each Dimension and Method when Predicting

Attitude ........................................................................................................................ 64

Chapter 4. Table 5 D. Chow Test for Each Dimension and Method when Predicting

Reputation .................................................................................................................... 65

Chapter 4. Table 6 A. Fisher’s R to Z transformation When Predicting Satisfaction . 66

Chapter 4. Table 6 B. Fisher’s R to Z transformation When Predicting Purchase ...... 67

Chapter 4. Table 6 C. Fisher’s R to Z transformation When Predicting Reputation ... 67

Chapter 4. Table 6 D. Fisher’s R to Z transformation When Predicting Attitude ....... 68

Chapter 4. Table 7. Comparing Direct and Personified Data ...................................... 71

Chapter 4. Table 8. AVE and CR Results According to Dimension and Measurement

Approach ...................................................................................................................... 71

Chapter 5. Table 1. Questionnaire Type Distribution According to Gender for Study 1

...................................................................................................................................... 87

Chapter 5. Table 2. Cronbach’s Alpha Values of Dimensions by Groups .................. 90

7

Chapter 5. Table 3. Mean Scores of Dimensions by Groups ....................................... 90


Each Group .................................................................................................................. 90

Chapter 5. Table 5. Adjusted R-Square Values of Dependent Variables by Context . 92


Satisfaction ................................................................................................................... 93


Intellectual Engagement ............................................................................................... 93


Social Engagement ....................................................................................................... 94


Affective Engagement ................................................................................................. 94

Chapter 5. Table 6 E. Chow Test for Each Dimension and Method when Predicting

Overall Engagement ..................................................................................................... 95

Chapter 5. Table 7 A. Fisher’s R to Z transformation When Predicting Satisfaction . 96

Chapter 5. Table 7 B. Fisher’s R to Z transformation When Predicting Intellectual

Engagement .................................................................................................................. 96

Chapter 5. Table 7 C. Fisher’s R to Z transformation When Predicting Social

Engagement .................................................................................................................. 97

Chapter 5. Table 7 D. Fisher’s R to Z transformation When Predicting Affective

Engagement .................................................................................................................. 97

Chapter 5. Table 7 E. Fisher’s R to Z transformation When Predicting Overall

Engagement .................................................................................................................. 97


Approach .................................................................................................................... 101

Chapter 5. Table 9.Results of Multi Group Analysis for Each Model (factor loadings

constrained) ................................................................................................................ 102

Chapter 5. Table 10. Questionnaire Type and Dimension Distribution According to

Gender ........................................................................................................................ 104

8

Chapter 5. Table 11. Cronbach’s Alpha Values of Dimensions by Groups .............. 105


Each Group ................................................................................................................ 105

Chapter 5. Table 13. Adjusted R-Square Values of Dependent Variables by Context

.................................................................................................................................... 107


Satisfaction ................................................................................................................. 107


Intellectual Engagement ............................................................................................. 108


Social Engagement ..................................................................................................... 108


Affective Engagement ............................................................................................... 109

Chapter 5. Table 14 E. Chow Test for Each Dimension and Method when Predicting

Overall Engagement ................................................................................................... 109

Chapter 5. Table 15. Fisher’s R to Z transformation When Predicting Intellectual

Engagement ................................................................................................................ 111


Approach .................................................................................................................... 114

Chapter 5. Table 17.Results of Multi Group Analysis for Each Model (factor loadings

constrained) ................................................................................................................ 114

Chapter 5. Table 18. Mean Scores and Standard Deviations for the three items ....... 115

Chapter 6. Table 1. Questionnaire Type and Dimension Distribution According to

Gender ........................................................................................................................ 139

Chapter 6. Table 2. Cronbach’s Alpha Values for the Task Difficulty Scale by Method

and Dimension ........................................................................................................... 141

Chapter 6. Table 3. Means of Task Difficulty Score by Questionnaire Types and

Image Dimensions ..................................................................................................... 142

9

Chapter 6. Table 4. Results of the Analysis of Variance (two-way ANOVA with

interaction) of the mean scores of Task Difficulty by Image Dimensions and

Questionnaire Types .................................................................................................. 142

Chapter 6. Table 5. Brand Type and Image Dimension Distribution According to

Gender ........................................................................................................................ 150

Chapter 6. Table 6. Cronbach’s Alpha Values of Brand Types by Image Dimensions

.................................................................................................................................... 151

Chapter 6. Table 7. Cronbach’s Alpha Values of Dimensions by Brand Type for Task

Difficulty Scale (Construct) ....................................................................................... 151

Chapter 6. Table 8. Means of Task Difficulty Score by Brand Types and Image

Dimensions ................................................................................................................ 152

Chapter 6. Table 9. Results of the Analysis of Variance (two-way ANOVA with

interaction) of the mean scores of Task Difficulty by Image Dimensions and Brand

Types……………………………………………………………………………….. 152

10

List of Abbreviations

AFF ENG Affective Engagement

AGFI Adjusted Goodness of Fit Index

ANOVA Analysis of Variance

ATT Attitude

AVE Average Variance Extracted

CFI Comparative Fit Index

CMV Common Methods Variance

CMIN Minimum Chi-Square

CR Composite Reliability

D Direct

DF Degrees of Freedom

DQ Direct Questioning

DV Dependent Variable

GFI Goodness of Fit Index

HOV Homogeneity of Variances

IFI Incremental Fit Index

INT ENG Intellectual Engagement

M&S Marks and Spencer

NASA National Aeronautics and Space Administration

NFI Normed Fit Index

OVR ENG Overall Engagement

P Personification

PUR Purchase

11

RFI Relative Fit Index

REP Reputation

RMSEA Root Mean Square Error of Approximation

SAT Satisfaction

SEM Structural Equation Modelling

SOC ENG Social Engagement

TLI Tucker Lewis Index

TLX Task Load Index

UK United Kingdom

12

Abstract There are several approaches to brand image measurement. The main aim of this thesis is to understand which of the two most common approaches, namely the personification and the direct approach, should be preferred. The personification approach adopts the brand = person metaphor (if the brand came to life as a person would s/he be trustworthy?), while the direct approach simply asks ‘Do you think this brand is trustworthy?’. The main method used is to compare their explanations of typical outcomes (dependent variables) in a series of online surveys. Two different dimensions of brand image (warmth and competence) are considered for different types of brand (product, employer and corporate). The thesis uses the ‘journal ready format’ where a series of related papers form the main part of the work. This thesis adopts a quantitative approach and presents the results from four empirical studies. To compare the two approaches to brand image measurement, Study I (Journal Article I) compared two types of brands (product and corporate) and the two types of brand image measurement approach. In Study II and Study III (Journal Article II), the context was shifted to employer branding, when comparing the two approaches. The analysis of the first and the second studies showed no consistent pattern and no systematic advantage for the personified approach. Indeed the two types of measure appeared quite similar in many respects. When trying to explain the results, task difficulty emerged as a possible explanation and was investigated via Study III and Study IV (Journal Article III). Task difficulty was not lower for the personified approach as expected. While there is a rich body of brand image literature using either personification or direct measurement approaches, there is no research comparing them in the same context/setting to understand any differences between these approaches. Two main conclusions emerged from this research to contribute to the market research literature. This research shows that there is no systematic statistical benefit from adopting the personification approach. Task difficulty varied with age and education, but not as expected from the literature, a finding that might be considered in all survey research, not just that involving brand image. Keywords: Brand image, brand image measurements, brand image dimensions, stereotype content model, warmth, competence, and task difficulty.

13

Declaration

No portion of the work referred to in the dissertation has been submitted in support of

an application for another degree or qualification of this or any other university or

other institute of learning.

14

Copyright Statement

i. The author of this thesis (including any appendices and/or schedules to this thesis)

owns certain “Copyright” or related rights in it and she has given the University of

Manchester certain rights to use such Copyright, including for administrative

purposes.

ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic

copy, may be made only in accordance with the Copyright, Designs and Patents Act

1988 (as amended) and regulations issued under it or, where appropriate, in

accordance with licensing agreements which University has from time to time. This

page must form part of any such copies made.

iii. The ownership of certain copyright, patents, designs, trademarks and other

“Intellectual Property” and any “Reproductions” of copyright works in the thesis, for

example graphs and tables, which may be described in this thesis, may not be owned

by the author and may be owned by third parties. Such Intellectual Property and

Reproductions cannot and must not be made available for use without the prior

written permission of the owner(s) of the relevant Intellectual Property and/or

Reproductions.

iv. Further information on the conditions under which disclosure, publication and

commercialisation of this thesis, the Copyright and any Intellectual Property and/or

Reproductions described in it may take place is available in the University IP Policy

(see http://documents.manchester.ac.uk/display.aspx?DocID=487), in any relevant

Thesis restriction declarations deposited in the University Library, the University

Library’s regulations (see http://www.manchester.ac.uk/library/aboutus/regulations)

and in the University’s Guidance for the Presentation of Thesis.

15

Acknowledgements To quote the one of the greatest authors, who inspired millions of millennials, J. K.

Rowling “ It’s our choices that show what we truly are, far more than our

abilities.”(Rowling, 1999). When I started my PhD journey, I knew that I do not know

much, but I knew I was high on warmth and low on competence. Yet, I also knew

someone who is the ultimate researcher; still learning, and improving himself and the

ones who are around him. I am extremely lucky to have Professor Gary Davies as my

supervisor for bringing wisdom and magic to this journey, just like Dumbledore for

Harry. He was not only the best supervisor ever, but he also introduced me to another

great role model, Dr. Susan Whelan. I feel very grateful for her supervision. I am

eternally thankful to both my supervisors in this journey of aiming to become high on

both dimensions, because I have to have practical implications of my research, surely

(regardless of task difficulty).

I am very appreciative to AMBS as they gifted me a great PhD experience. I would

like to acknowledge the unfailing support given by our doctoral programmes

administrator, Paul Greenham, and my MBS family; Niki Hutson, Vildan Tasli, and

Yusuf Kurt.

Finally and most importantly, I would like to thank the most insightful and visionary

person I know, my mother Nezahat Coskun, for always encouraging me, never failing

to be there for sharing each joy and every sorrow. This thesis, and hopefully every

future contribution of mine are due to you and dedicated to you.

16

Chapter 1: Introduction

This thesis is about comparing two measurement approaches in a brand image

context. It also considers the topic of task difficulty, which emerged during the work.

It is presented in an alternative format; journal format, i.e. the thesis builds on three

articles with four studies. The first two articles mainly focus on comparing the two

main methods of brand image measurement; the direct and personification

approaches, in three contexts, product and corporate brand and employer brand image

respectively. The final article focuses on the role of task difficulty in brand image

measurement.

While brand image measurements are widely made by marketing scholars and

practitioners, the possibility of getting different results when using different

approaches has been overlooked. The two approaches compared in this thesis are

labelled as ‘personified’ and ‘direct’. For example a personified approach might ask a

respondent, ‘if the brand came to life as a person would s/he be trustworthy?, while

the direct approach would ask, ‘Do you think this brand is trustworthy?’. The first

approach is an example of a projective technique and some claim this enables

researchers to acquire better responses since it would make it easier for respondents to

evaluate brand image (Boddy, 2005), and therefore should be preferred over a direct

approach. Other researchers state allocating human associations to inanimate objects,

brands in this research, might be unacceptable (Davies et al., 2001). The first journal

article investigates this issue. The second article specifically investigates this issue in

the case of employer branding. During the analysis of the second study, task difficulty

appeared as a possible explanation for the different results of similar studies in terms

17

of Study I and Study II. Therefore the third journal article focuses on the role of task

difficulty in the context of brand image measurement. To sum, the core of these three

journal articles is brand image measurement; the first two examine the two main types

of brand image measurement to understand the differences between them, and the

final one examines the potential effect of task difficulty when using brand image

measurement.

This introductory chapter explains the research motivation and research design, and

presents an overview of the thesis, its format and structure.

1.1. Research Motivation and Research Design

The main motivation for this research was that while brand image measurements are

well accepted and extensively used both in academic and marketing research, there is

no consensus on which approach of brand image measurement would be a better

fit/choice. Moreover, there is no specific research that compares the two main

approaches in the same context. This lack of previous research comparing the two

methods is the key gap that is identified in the literature for the thesis to fill.

This research is quantitative and often adopts an experimental design. Each of the

journal articles has their own hypotheses, and depending on these hypotheses the data

collection and analyses are carried out accordingly. Further details are explained in

the methodology chapter and in each journal article chapter.

18

1.2. Overview of the Thesis

The following three sub-sections show the abstract and authorship details of the three

journal article style chapters, including the contribution of the PhD candidate for each

one of the article chapters.

1.2.1. Measuring Brand Image: Personification versus Non-Personification

Methods

Abstract from the article: Maintaining a good product or corporate brand image is

considered to be one of the most crucial parts of brand management (eg. Dutton,

Dukerich, and Harquail, 1994; Fombrun and Shanley, 1990). Yet, the methods used

to measure brand image differ between researchers. The two most common

approaches employ either the personification metaphor (e.g. Aaker, 1997; Davies,

Chun, da Silva, and Roper, 2001; Geuens, Weijters, and De Wulf, 2009), or direct

questioning (Hsieh, 2002). Yet, there is no consensus on which method should be

preferred.

This study aims to compare the two approaches by testing their validity and ability to

predict typical dependent variables used in brand image research including

satisfaction and purchase intention.

Using an online survey (n=400) the imagery of two brands Pantene (a leading product

brand) and Marks and Spence (a leading corporate brand) was measured using either a

personified approach or a direct questioning approach. Scale validity and the ability of

competing approaches to predict the dependent variables were tested in a number of

ways.

19

Contrary to expectations, there was no systematic advantage for the personified

approach. The implications for further research are discussed.

Authorship: Melisa Mete, Gary Davies, Susan Whelan

Contribution of PhD candidate: This study was supervised by Gary Davies and Susan

Whelan. The preliminary research and the literature review were conducted by the

PhD candidate. The development of the research design was made by all the authors.

Both of the supervisors helped the PhD candidate with data analysis.

Note: The data collection was carried in the first year of the candidate’s PhD studies.

Parts of the study were presented at the 3rd International Reputation Management

Conference, Reputation Management Institute in Istanbul, Turkey; and at the 10th

Global Conference of Academy of Marketing’s Brand, Identity and Corporate

Reputation SIG in, Turku, Finland.

1.2.2. How Best to Measure Employer Brand Image: Personification versus

Direct Method

Abstract from the article: Two studies (N=221 and N= 440) are reported, both aimed

at identifying whether a personified or a direct form of questioning should be

preferred in the measurement of employer brand image. Two dimensions of brand

image are considered in both, labeled warmth and competence, as suggested by the

application of the stereotype content model (Fiske, Cuddy and Glick, 2007) to the

study of brand image.

20

In both studies members of the public were asked to evaluate their employer. In Study

1 respondents were each asked to evaluate their employer’s image using either a

personified or a direct measure. To test for any possible dimension specific or order

effects, Study 2 uses a between subject factorial design where half of the respondents

evaluated their employer for warmth, half for competence, half using a personified

approach, half a direct approach to measurement.

No systematic benefit for the use of personification was found in either study.

Differences between the predictivity of individual dimensions in Study 1 were not

confirmed in Study 2.

Authorship: Melisa Mete, Gary Davies, Susan Whelan



PhD candidate. The development of the research design involved all the authors. The

data analysis and the conclusions were made by the PhD candidate with guidance

from Gary Davies.

Note: The data collection was carried on the second and the third years of the

candidate’s PhD studies. Parts of the study were presented at the British Academy of

Management Conference 2015 in, Portsmouth, UK; and at the 11th Global Conference

of Academy of Marketing’s Brand, Identity and Corporate Reputation SIG, Bradford,

UK. Parts of the data were analyzed separately and the subsequent paper accepted for

publication in August 2017. Reference:

21

‘Davies, G., Mete, M., and Whelan, S. (2017). When Employer Image Aids Employee

Satisfaction and Engagement. Journal of Organization Effectiveness: People and

Performance. (doi: 10.1108/JOEPP-03-2017-0028)’. The paper included as part of

this thesis is however quite different in focus.

1.2.3. Measuring Brand Image and the Role of Task Difficulty

Abstract from the article: Two studies are conducted to understand the role of task

difficulty in market research and specifically in the context of brand image

measurement. Task difficulty was found to be influential in brand image evaluations

in previous research and its influence is more formally considered here. In order to

understand the influence of task difficulty, several variables such as the age and

education level of the respondents are considered.

In study one, an online survey was made with employees as respondents (N=440) to

evaluate their companies’ brand image using a 2 (Personification vs. Direct) x 2

(Warmth vs. Competence) factorial, between-subjects design.

In study two, the context was changed from employer branding to considering one

brand (Tesco) used in two different contexts, as a corporate/organizational brand and

as a private label/product brand. The respondents were given either warmth or

competence dimension of brand image items to consider,

An adapted version of the TLX measure of task difficulty scale (Hart and Staveland,

1988) was used in both surveys.

Task difficulty did not vary as expected by image dimension or by whether a

projective or direct method was used to measure image. It did not influence the

22

relationship between image and a number of dependent variables, but it did contribute

to an explanation of several variables such as intellectual engagement.

Task difficulty was however found to vary with respondent age and education, but not

in ways implied by existing literature.

Authorship: Melisa Mete, Gary Davies.



PhD candidate. The development of the research design was undertaken by Gary

Davies and the PhD candidate. The data analysis and the conclusions were made by

the PhD candidate with guidance from Gary Davies.

Note: The data collection was carried out in the third year of the candidate’s PhD

studies. Parts of the study were presented at the American Marketing Association

Summer Marketing Educators’ Conference 2016, in Atlanta, USA.

1.3. Thesis Format and Structure

This thesis follows a journal format thesis style. Early in the fieldwork it became clear

that the work fell into a number of separate stages and that it would be logical to

adopt what was then called ‘the alternative thesis style’, now known as the journal

format. Unlike the standard format, the journal format allows chapters which are in a

suitable format for publication in peer-reviewed academic journals. Hence, this thesis

consists of four empirical studies written as three journal article style chapters. Each

23

article chapter has its own sections of literature review, methodology, data analysis

and results, and conclusions. Inevitably there is some overlap between the literature

reviews, particularly in papers one and two. The candidate sometimes refers to a

previous paper in the thesis as if it had been published to make it easier for the reader.

In between major chapters short linking chapters are used again to help the reader. A

final chapter brings together the findings from the three papers.

The use of the journal format for this thesis was formally approved by the

Postgraduate Research Office of Alliance Manchester Business School.

24

Chapter 2: Brand Image and its Measurement

The purpose of this chapter is to discuss the idea of brand image so as to introduce the

debate as to how it should be measured. It examines the different definitions that exist

for brand image and introduces the definition that is relied upon throughout the thesis,

one linked to brand personality. It also examines the different approaches to

measuring brand image to introduce the reader to the main thesis that concerns the

advantages and disadvantages of using brand personality as a measure of brand

image.

The Notion of Brand Image

Since its formal identification in the 1950s, brand image has become a popular topic

in consumer behaviour research for both practitioners and academics due to the reality

that people buy products for something other than their physical attributes and

functions (Dobni and Zinkhan, 1990). Marketers have also realized the strategic

importance of brand image in creating greater value (e.g. Graeff, 1997; Kamins and

Gupta, 1994; Pettijohn, Mellott, and Pettijohn, 1992).

From an academic perspective, the first meaningful reference to brand image was in

Gardner and Levy’s article in 1955 where they explained the motivation behind

purchasing behaviour. They argued that in addition to their physical nature, products

also have a social and psychological role; and also stated that consumers’ feelings,

ideas, and attitudes about and towards brands, or their "image" of brands is vital to

purchase choice (Gardner and Levy, 1955).

25

Our understanding of the importance of brand image has developed. For instance,

instead of their functional qualities, some brands are favored over their competitors’

due to their impact on the buyer/user's status and self-esteem (Levy, 1958). Similarly,

in terms of purchase motivation, when there is congruence between a product’s brand

image and the actual or ideal self-image of the user, a product is more likely to be

consumed and liked (Sirgy, 1985).

As Dobni and Zinkhan (1990) note, some authors have focused on the symbolic

nature of brands and products, (e.g. the "Symbolic utility" notion of Pohlman and

Mudd (1973), "the symbols by which we buy" (Levy, 1958) and "perceived product

symbolism" (Sommers, 1964)) to describe the intangible aspect of consumer product

brands while others, especially those who believe that there is an inseparable link

between an individual’s self-concept and their purchases, have focused on the

humanistic qualities of brands; for example by the use of terms such as "brand

personality" (Hendon and Williams, 1985), "brand character" (Hendon and Williams,

1985), and "personality image" (Sirgy, 1985).

Consequently, there are several different approaches to defining brand image. Gensch

(1978) defined "image" as a purely abstract concept, which unites the influences of

past promotion, reputation and peer evaluation of the brand, and stated that image

indicates the expectations of consumers. Reynolds and Gutman (1984) defined

product brand imagery as ‘the stored meaning in an individual’s memory’, implying

that what is recalled from memory provides the meaning we attribute to image. More

simply brand image has also been defined as the ‘consumer’s perceptions about a

26

brand as reflected by brand associations held in memory’ (Torres and Bijmolt, 2009),

a definition consistent with the associative network model of memory (e.g. Anderson

1983). Accordingly, it is this associative network that constitutes a brand’s image,

identifies the brand’s uniqueness and value to consumers, and explains how a brand’s

equity can be leveraged in the marketplace (Aaker, 1996; Henderson, Iacobucci, and

Calder, 1998; Schnittka, Sattler, and Zenker, 2012).

Other authors criticize such definitions of brand image, as only including attributes or

abstractions (see De Pelsmacker, Geuens, and Van den Bergh, 2007). Such authors

argue that the psychological qualities of both user and brands must be accounted for

(Dobni and Zinkhan, 1990).

Some conceptualizations of brand image are very broad; for instance Levy (1958)

considers brand image consists of not only the physical reality of the product, but also

the beliefs, attitudes and feelings that have come to be associated with it.

Keller, Parameswaran, and Jacob (2011) have provided a more comprehensive

description of brand image. They state “brand image is reflected by the associations

that consumers hold for it. It helps marketers to make a distinction between lower-

level considerations, related to consumer perceptions of specific performance and

imagery attributes and benefits, and higher-level considerations related to the overall

judgments, feelings, and relationships. There is an obvious connection between the

two levels, because consumers’ overall responses and relationship with a brand

typically depend on perceptions of specific attributes and benefits of that brand”

(Keller et al., 2011, p. 379). They also stress the important role of beliefs for brand

27

image due to the fact that they are descriptive thoughts that a person holds about

something. In line with beliefs in general; “brand association beliefs are those

specific attributes and benefits linked to the brand and its competitors. For example,

consumers may have brand association beliefs for Sony Playstation home video

games such as “fun and exciting”, “cool and hip”, “colourful”, “good graphic

quality”, “advanced technology”, “variety of software titles”, and “sometimes

violent” (Keller et al., 2011, p. 379).

It is also important to note that even though the notion of brand image was initially

associated with product brands, three different but related image types have been

identified (Stern, Zinkhan, and Jaju, 2001): product brand imagery (e.g. Levy, 1958;

Sirgy, 1985), company or corporate brand image (Shimp and Bearden, 1982), and

retail or store brand image (Collins-Dodd and Lindley, 2003; Doyle and Fenwick,

1974; Jacoby and Mazursky, 1984; Martineau, 1958). The last two are similar in that

many service organizations, airlines, and hotels for example are mono-brands where

the company and service names are the same. It can be said that corporate brands

differ from product brands due to the fact that they represent the firm, and their image

is potentially constructed by everything a firm is perceived to be doing (e.g. Balmer,

2001; Balmer and Gray, 2003; Harris and de Chernatony, 2001; McDonald, de

Chernatony, and Harris, 2001; Kapferer, 2002). Thus, corporate brand image is

synonymous with a company’s corporate image (Blombäck and Axelsson, 2007).

Furthermore, retail or store brand image can be defined as a combination of the store

as a brand, and the selection of store brands and manufacturer brands offered by the

store (Grewal, Cote, and Baumgartner, 2004; cited in Martenson, 2007), or it can be

28

described as the way consumers view the store as in their impressions or perceptions

(Hartman and Spiro 2005).

Some authors, such as James et al. (1976) and Lindquist (1974/1975) have

emphasized the importance of a broader perspective by arguing that retail or store

image is not only a summation of diverse impressions or perceptions of attributes but

is also a function of the interactions among these attributes in terms of fashionability,

salesmanship, outside attractiveness and advertising (Marks, 1976; cited in Hartman

and Spiro, 2005). For example interactions with the salesman in a store lead

consumers to have a specific store/retail brand image according to their perceptions of

these interactions.

Conversely, some authors such as Bullmore (1984) refute the commonly accepted

assumption that an image belongs to the brand, arguing that an image, like a

reputation, can only exist in the minds of people. Thus, they propose “an image is

projected to the consumer by the marketer, and that it can be selected, created,

implemented, cultivated”, and "managed by the marketer over time” (Dobni and

Zinkhan, 1990).

In summary, regarding the notion of brand imagery, it can be said that brand image is

the concept of a brand that is held by the consumer; and is mainly a subjective and

perceptual phenomenon that is formed through consumer interpretation, that is either

reasoned or emotional. Moreover where brand image is concerned, the perception of

reality is more important than the reality itself. Despite conceptual deviations, it is

apparent that the concept of brand image has been of great significance in consumer

29

behavior research, and has potential to be explored (Dobni and Zinkhan, 1990).

The American Marketing Association formally define brand image as:

“The perception of a brand in the minds of persons. The brand image is a mirror

reflection (though perhaps inaccurate) of the brand personality or product being. It is

what people believe about a brand-their thoughts, feelings, expectations.”

The definition includes the term “brand personality” and suggests that brand image is

closely related to it. It is also the definition that most influenced this thesis.

Researchers have often chosen to focus on the “personification approach” for their

descriptions of brand image. The idea of personifying a brand and imbuing its image

with human characteristics has been researched using mainly two distinct

perspectives.

The first perspective considers the description of the product brand as if it were a

human being, with a distinct personality of its own. A frequently used device is to

associate the brand with an actual human, either fictitious or real (e.g. Betty Crocker,

Uncle Bens). The second focuses on associating the consumer's personality or self-

concept with the image of the product brand; for instance in the fragrance industry,

the association of perfume usage with fulfilled dreams and fantasies (Gardner and

Levy, 1955).

Brand image and brand personality have been defined as both similar concepts (eg.

Hendon and Williams, 1985; Upshaw, 1995) and as separate concepts (eg. Gordon,

1996; Patterson, 1999).

30

Davies and Chun (2003) state that brand image as a metaphor has to be limited to the

idea of a brand being a mental picture, an impression in mind (p.60). They further

argue that brand image as a metaphor has made a limited contribution to our

understanding of what a brand is, since the ideas of brand and image are too

congruent. Contrarily, they claim the “Brand as a person” metaphor is very much

alive; the idea that brands can have personality provides new ways of thinking about

brands and branding (Davies and Chun, 2003, p. 61).

The personification of the brand, or the usage of the brand as person metaphor has

been extensively employed in the marketing literature and research. King (1973), for

instance, claims that the main difference between two similar products of competing

brands is the different personalities that are projected by each brand. Similarly,

Keeble (1991) states that amongst the two competing soap powders Ariel and Persil,

only Persil “had a personality”. Moreover Aaker (1996) found that some brands such

as Hallmark, Fisher Price, AT&T, and Lego are associated with a “warm” and

“caring” personality. Additionally, Fournier (1998) investigated the different types of

relationships that people may have with brands; such as trust and friendship. Similar

to our human interactions, the brands we are involved with mean more to us (Laurent

and Kapferer, 1985, Davies and Chun, 2003). Quite similar to our preferences on

being highly involved with certain people and having a distant approach to other types

of people, we tend to have high and low involvement with products and brands

(Gordon, 1996).

However, some researchers and theorists argue that relating brand image to

personality is not a proper way of describing brand image.

31

Particularly, similar to psychologists’ struggle of defining and measuring personality,

it becomes a problem for those interested in studying brand image as well. Therefore,

foreseeably those who define brand image by reference to personality do not attempt

to define the latter concept in a detailed way. They merely suggest that products have

personality images, or they focus in on some distinctly human descriptor, such as

“gender" image” (Debevec and Iyer, 1986), “age” image (Bettinger and Dawson,

1979), or “social caste” image (Levy, 1958)” (Gardner and Levy, 1955) (cited in

Dobni and Zinkhan, 1990).

On the other hand, associating brand image with personality has been justified on

many grounds. Firstly, both brand imagery and personality are multidimensional, and

seem to work at the same level of abstraction (Gardner and Levy, 1955).

Additionally, some researchers (e.g. Kassarjian and Sheffet, 1975) argue that

personality can be best conceived of as a dynamic whole, which is consistent with the

general sense that many have about brand image.

Approaches for Measuring Brand Image

Not surprisingly given the discussion above on the different definitions of brand

imagery, there is also a lack of consensus on the techniques or approaches for brand

image measurement that should be used.

32

Some researchers have focused on measuring the image of individual dimensions of a

product brand (e.g. “classiness” Pohlman and Mudd, 1973), whereas some have relied

on a single measure for product brand image (Dolich, 1969) e.g. strong or weak.

Boivin (1986) used a brand image measure that focuses on consumer perceptions of a

brand in relation to its competition, whereas Keon (1983) used the TRINODAL

mapping technique to measure product brand image through advertisements in

relation to consumers’ ideal points. The TRINODAL mapping technique is a

multidimensional scaling routine that simultaneously plots brand images and

consumer preferences on a single map and is mostly used to provide insight into the

brand repositioning processes (Keon, 1983).

Sirgy (1985) on the other hand measured product brand image in relation to each of a

person's actual self-image, ideal self-image, the social self-image, and the ideal social

self-image.

Brand image has also been measured as a function of brand conspicuousness and

brand usage (Bird, Channon, and Ehrenberg, 1970), and also assessed from the

perspective of the retailer versus that of the consumer (McClure and Ryans, 1968;

cited in Dobni and Zinkhan, 1990). For instance McClure and Ryans’ research (1968)

concluded that retailers’ views of brand attributes and brand image differ from

consumers’ views; retailers tend to keep the image view that are based on historic

stereotypes, whereas the consumers tend to have a more up-to-date image of brands.

33

The Usage of Brand Personality

Brand personality has been defined as “the set of human characteristics associated

with a brand” (Aaker, 1997, p. 347). Aaker’s approach is to invite respondents to

“imagine a brand has come to life as a human being” and then ask them to assess

his/her personality. This approach, using the brand personality metaphor (brand as

human being) is one of the most commonly used ways to measure brand image (e.g.

Louis and Lombart, 2010).

Metaphors have various roles and forms in our lives (Black, 1962). For instance a

rhetorical usage can be for entertainment and diversion, for example “ Roger is a

Lion” is not about Roger being a Lion literally but this example of metaphor usage

provides us a figurative picture of Roger’s character (Black, 1962; cited in Davies and

Chun, 2003). Likewise academically, metaphors help us to make better sense of

complex ideas, such as brands. When we are trying to understand the complexity of

modern organisations, we could say, “The modern business organization is a machine

or an organism” (Morgan, 1986). Hence, metaphors can be mental models for sense

making (de Chernatony and Dall’Olmo Riley, 1997; cited in Davies and Chun, 2003).

A metaphor works through the associations we can make with something that is better

understood or just easier to understand. In other words, metaphors help us to

“explicate specific phenomena by referring to known properties of objects”

(Cornelissen and Harris, 2001). More fundamentally the use of metaphor invites the

reader to connect two ideas.

34

Going back to the example of “Roger is a Lion”; the target of this metaphor Roger is

seen through the metaphor of the lion, hence it filters and transforms our view of the

target (Davies and Chun, 2003).

Sackmann (1989) states that metaphors are mental pictures, which might “substitute a

thousand words”. Tourangeau (1982) found that we prefer metaphors when the target

and metaphor are not congruent; where the comparison we are asked to make is vast

and therefore the effect on our thinking is greater. Moreover, they are also found to be

more useful when they provide experiential similarities rather than objective

similarities with the target (Lakoff and Johnson, 1980).

Amongst the social sciences, marketing has a more metaphoric language than the

others (Zaltman et al., 1982; O’Malley and Tynan, 1999; Davies and Chun, 2003). In

the case of branding; according to the positivistic approach the brand was originally

seen as an “extended product”, but later the post-modern approach treats a brand as a

“living entity” (Hanby, 1999; cited in Davies and Chun 2003). If brands are

considered as living entities, then people treat them as they treat living entities. For

instance, if personification is used to describe a brand, consumers might or might not

like the humanized entity (Puzakova, Kwak, and Rocereto, 2013) and this depends on

the nature of the personified target (Aggarwal and McGill 2007).

Greater congruity between the features of a product brand and an activated human

schema1 will lead to more positive evaluations; since when consumers see brands as

1 “A schema is a stored framework of cognitive knowledge that represents information about a topic, a concept, or a particular stimulus, including its attributes and the relations among the attributes” (Fiske and Linville 1980; cited in Aggarwal and McGill 2007).

35

humans, it affects their evaluation of that brand. For instance, product brands that are

personified but which lack human features are evaluated less positively than products

that are personified and which have human-like features. The overall processing of

information and evaluation of the products may be influenced both by the degree of

satisfaction from seeing the fit between the product feature and the activated human

schema (Aggarwal and McGill, 2007; Fiske, 1982).

Geuens, Weijters, and De Wulf (2009) justify the personification method by

explaining that consumers tend use brands with a strong brand personality for

building relationships with (see also Fournier, 1998) or as a way of showing their own

personality (e.g. Belk, 1988; Malhotra, 1981). This raises the issue of whether brand

and human personality are the same, rather than brand personality being just a

metaphor for brand image. Geuens et al. (2009) take the former view and argue that

using a brand personality scale that resembles a human personality scale would enable

brand managers to create an appropriate brand personality for their target group.

Parker (2009) claims that associating human personality characteristics with a brand

can be justified due to the fact that people naturally anthropomorphize, in other words

transfer human characteristics to inanimate objects on a regular basis (Bower, 1999;

Boyer, 1996). A very typical example would be when one references an object, such

as a motorboat by saying, “she is a beauty” (Parker, 2009). Additionally, individuals

sometimes consider objects as another person (Boyer, 1996); for instance certain trees

are said to overhear and record conversations between people (James, 1988).

From the perspective of an anthropologist Ellen (1988) argues that: “There is a

general tendency in human relations with the inanimate world to attribute and

36

represent that world in organic terms, and to attribute inanimate objects with the

properties of living things. There is nothing particularly mysterious about this… it

happens because we are bound to model our world directly on those experiences

which are most immediate, and these are experiences of our own body” (Ellen, 1988,

p. 231).

Consequently, there is strong evidence that companies employ brand personality as a

part of their positioning strategy, and that this can affect consumer perceptions in far

more permanent ways than other communication strategies (Burke, 1994).

Furthermore this leads to a simplification of the decision process for consumers and

increases their awareness as well as building loyalty, and improves brand image (Phau

and Lau, 2001; Sutherland, Marshall, and Parker, 2004). Plummer (2000, p. 81)

suggests “brand personality plays a critical role in the “for me” choice, or “I see

myself in that brand” choice” (cited in Parker, 2009). Similarly, Hendon, and

Williams (1985) consider this as an effective way of generating interest because

people favor products that match their own self-image, and (human) personality is one

way for us to see ourselves as well as seeing others.

Moreover, when the consumer is operating without adequate information (for instance

when the consumer has little or no experience with the product, or when the consumer

has insufficient time or interest to evaluate the intrinsic attribute, or when the

consumer cannot readily evaluate the intrinsic attributes (Zeithaml, 1988; cited in

Freling and Forbes, 2005)), they likely to rely on information about a brand’s

personality as a surrogate for intrinsic product attributes.

Hence, brand personality most likely influences product perceptions, particularly

when evaluating intrinsic product attributes is difficult (Freling and Forbes, 2005).

37

Thus, brand personality may allow a given brand to stand out in a crowd.

Additionally, “having information about the brand’s personality may also increase

attention paid to the brand (Sekuler and Blake, 1994) and stimulate active

information processing (Biel 1992)” (cited in Freling and Forbes, 2005).

To summarize, there are various views to understand brand image and how it should

be conceptualized. Brand image and brand personality are seen as being closely

related ideas. Brand personality is used as a way to measure brand image and has been

argued to be a beneficial way to construe brand image.

The next chapter is the methodology chapter. Following methodology chapter comes

the first of three papers that form the core of the thesis. It aims to test the idea that

personification is advantageous as a measurement approach when measuring brand

imagery.

38

Chapter 3: Methodology

1. Introduction

As required by the regulations for a publication ready thesis, this chapter is included

to discuss the methodologies used in this research for the four studies that were

carried out to understand the phenomena of interest. (A more detailed description of

each method is given within the three papers).

The initial main aim of the study was to understand whether there is a difference in

using two main brand image measurement approaches, in terms of personification and

non-personification. The first journal article examines this phenomenon from a

consumer perspective. The second article uses an employer branding perspective.

After conducting the second study, it appeared that ‘task difficulty’ might be a crucial

factor that affected responses. Therefore the third article was dedicated to examining

the role of task difficulty.

All four studies in this research used experimental design. An experiment refers “ to

that portion of research in which variables are manipulated and their effects upon

other variables observed.” (Campbell and Stanlay, 1963 p.1). In other word, it

involves “one or more independent variables to be manipulated to observe their

effects on one or more dependent variables” (Yaremko, Harari, Harrison, and Lynn

,1986, p.72)., and “An experimental design is a plan for assigning experimental units

to treatment levels and the statistical analysis associated with the plan” (Kirk, 1995,

p.1).

39

Experimental research enables researchers to have control over various factors that

influence the phenomenon of interest and to isolate the relationship between

conditions or behaviours they could change and the outcomes they seek (Swanson and

Holton, 2005). When researchers deliberately set out to create certain conditions to

test their theory or propositions, they create specific hypotheses from theory and aim

to test them by experiments (Kirk, 1982; Swanson and Holton, 2005).

The main methodology used in the thesis is then quantitative but quantitative research

can be exploratory; “used to discover relationships, interpretations, and characteristics

of subjects that suggest new theory and define new problems” (Swanson and Holton,

2005, p. 52) as well as confirmatory. Specifically it can help in developing theories

(McCall and Bobko, 1990).

There are various types of experimental design options, Table (1), such as post-test

only control group, pretest-posttest control group, Solomon four-group, and factorial

(Swanson and Holton, 2005).

40

Table 1. True Experimental Designs

Source: Research in Organisations (p. 86, table 6.2) by Swanson and Holton, 2005

This research adopted factorial design, “which enables the researcher to compare two

or more independent variables at the same time”(Swanson and Holton, 2005, p.108).

Factorial design enabled the researcher in this research to examine the independent

effects of variables, such as type of brand image measurement approach, and brand

image dimensions and level of impact, as well as their interaction effects.

2. Research Design and Procedure

Previous research and the literature review lead one to expect a personification

approach to be more advantageous when predicting certain outcomes (dependent

variables) such as consumer satisfaction, consumer perception on brand reputation,

purchase intentions for the product or the corporate brand for the first study. The

41

hypothesis implied is that personified measures should lead to better prediction of

dependent variables. The hypothesis was tested using a 2 X 2 between subjects

factorial design with two brand types (product and corporate brand), two

measurement approaches (personification and non-personification) with a total sample

size of 360.

In the second journal article, two studies were made to test the same hypothesis in the

different context of the employer brand, again comparing the two main brand image

measurement approaches. In the first study (N=221) respondents were randomly

assigned to one of two groups (personification and non-personification). To test for

possible order or context effects, the second study (N=440) adopted a 2 X 2 between

subjects factorial design with two brand image dimensions (warmth and competence),

and two measurement approaches (personification and non-personification). In both

studies, to ensure a large number of brands were being evaluated, respondents were

employees.

For the third journal article, the role of task difficulty was explored when brand image

is measured. Two studies were conducted to understand this phenomenon. The first

study used respondents as employees to understand how they perceive task difficulty

when they evaluate their employer’s brand image.

An online survey was conducted with 440 respondents. A 2 X 2 between subjects

factorial design was used with two brand image dimensions (warmth and

competence), and two measurement approaches (personification and non-

personification). The second study involved a British grocery retailer where

42

respondents assessed their image as either a corporate or product brand using only a

non-personification questioning approach. An online survey was conducted with 663

respondents using a 2 X 2 between subjects factorial design with two brand types

(corporate/organizational brand and private label/product brand) and two brand image

dimensions (warmth and competence).

Procedure. In every study, participants were assigned randomly to one of the groups

(treatment conditions) that are defined in each study. Filter questions were asked to

screen out for example non-consumers and non-UK residents. The market research

company used each time was asked to ensure that no children were included in any

sample to ensure the research fell within the University’s code on ethics in research.

After the filter questions; respondents were told they were participating a survey, in

which their answers would be treated confidentially, and the results of the survey

would be used as a whole, not individually. They were also told there were no right or

wrong answers to any of the questions. Such questions were also included to meet

University policy on research ethics. Additionally, for the second and third studies,

the respondents were told that the survey was about how they see their employer and

their work, and the researchers did not ask for the name of their employer. For the

fourth study, they were told the survey was mostly about their views of Tesco own

brand products/ Tesco as a company/ their local Tesco store.

Then demographics data were collected via age, gender, and education questions.

For studies two and three; how many years the respondents had been working for their

current employer was asked.

43

The same layout was followed in each questionnaire with the dependent variables

(DV’s) first. Following demographics, the satisfaction, reputation, purchase or

involvement questions were asked. Next the brand image questions were asked. The

specific dimensions and the questions/items that create these dimensions were

selected from a list of items that had been used in past research, and they were all

checked for reliability and validity under each new study that was conducted. The

details of the scales used would be further explained in the journal articles.

For study two 2 questions were asked for task difficulty. For study three and four six

questions of the task difficulty scale were asked. Then a final open-ended question

was asked to understand if the respondents had any problems when answering any

parts of the survey in studies 2, 3, and 4.

At the end of each questionnaire the respondent was thanked and informed that that

project was being conducted by staff and students at Alliance Manchester Business

School, and their help was appreciated. Copies of the questionnaires used are included

in the Appendices.

3. Sampling/Data Collection Methods

The entire data for the thesis were collected through self-administered online

surveys/questionnaires. An online survey company (Pureprofile) was used to

distribute the questionnaires. Their panel is representative of the UK population as a

whole. Individuals would be invited to take part in each survey but could choose not

44

to do so or to drop out at any time during the survey. Respondents received payment

in the form of points they could redeem for goods.

The adopted sampling method is convenience sampling, which involves the selection

of sample participants based on availability or accessibility (Swanson and Holton,

2005). This method might be criticised since there is no certainty on how

representative the information collected from the sample, comparing the population as

a whole. However, even though it has been argued it might not be a perfect

representation of the population in question, therefore not the most useful method for

generalizability of the findings, it is one of the most common sampling methods in

published articles (Dooley and Lindner, 2003). What is more important is that

respondents were randomly allocated to each of the experimental cells.

There are certain limitations such as time and budget to lead the researcher to

specify/limit the sample. For instance before conducting a global sample, this study

aims to understand the responses of British respondents and was therefore limited to a

British sample. In order to control the sample, filter questions are used.

Moreover the studies aimed to understand consumers’ or employees’ brand image,

therefore the sample was limited to these groups. The sample was only with actual

consumers when measuring product and corporate brand image, and current workers

when measuring employer brand image. More specifically, when assessing the brand

image of the employer, the respondents were asked to answer the question based on

their perceptions of their current employer. And when assessing the corporate or

product brand, filter questions were asked to make sure the respondents were using

the services or/and products of the brand in question.

45

4. Statistical/Analytical Techniques and Statistical Software

For the analysis of the data, IBM SPSS was used (Version 23, 2015). In order to

understand any moderation effects, PROCESS macro was used (Hayes, 2013). In

order to understand the dimensions and factor loadings the AMOS package of SPSS

was used. In order to understand the data, firstly descriptive statistics were used such

as simple means and averages. Then to compare these means t-tests were used. When

there were more than two groups to compare, the analysis of variance (ANOVA)

technique was used. To check any association between groups, correlations or

regressions were used. No data cleaning was used, as all responses provided by the

company were complete. (The effect of time taken to complete is discussed in the

third paper).

5. Reliability and Validity

For reliability, Cronbach’s alphas were calculated for each latent construct, and all the

resulting alpha values were sufficiently high (Nunnally, 1978; Peterson, 1994).

Then, convergent validity was assessed by determining whether each observed

variable’s estimated maximum likelihood factor loading on its latent construct was

significant (Gerbing and Anderson, 1984). The results showed that the convergent

validity was achieved for each assessment, as all factor loadings were significant (p<

0.05) and within acceptable ranges.

Following, to assess construct validity, Confirmatory Factor Analysis was used

(Jöreskog, 1967) and the Average Variance Extracted (AVE) and Composite

46

Reliability (CR) used to assess the convergent validity of the measurement models

(Fornell and Larcker, 1981, a; Fornell and Larcker, 1981, b)

The details can be found in each chapter.

6. Limitations

Due to the time and budgetary constraints, this research was limited to a UK sample

only. This research was carried out with English speaking respondents, who reside in

the UK. Future research could be conducted with a different language and a culture to

understand whether similar results would be replicated. Another limitation was due to

convenience sampling of the respondents, this research was conducted via online

questionnaires, and therefore this research excluded the population of computer-non-

users.

47

Chapter 4

Measuring Brand Image:

Personification versus Non-Personification Methods

48

Measuring Brand Image: Personification versus Non-Personification Methods

Abstract

Maintaining a good product or corporate brand image is considered to be one of the

most crucial parts of brand management (eg. Dutton, Dukerich, and Harquail, 1994;

Fombrun and Shanley, 1990). Yet, the methods used to measure brand image differ

between researchers. The two most common approaches employ either the

personification metaphor (e.g. Aaker, 1997; Davies, Chun, da Silva, and Roper, 2001;

Geuens, Weijters, and De Wulf, 2009), or direct questioning (Hsieh, 2002). Yet, there

is no consensus on which method should be preferred.

This study aims to compare the two approaches by testing their validity and ability to

predict typical dependent variables used in brand image research including

satisfaction and purchase intention.

Using an online survey (n=400) the imagery of two brands Pantene (a leading product

brand) and Marks and Spencer (a leading corporate brand) was measured using either

a personified approach or a direct questioning approach. Scale validity and the ability

of competing approaches to predict the dependent variables were tested in a number

of ways.

Contrary to expectations, there was no systematic advantage for the personified

approach. The implications for further research are discussed.

49

Introduction

Brand imagery can be measured in a number of ways (Keller, 1998). The most

obvious approach perhaps is to ask respondents direct questions such as, ‘How much

do you trust this brand/company?’ However many studies use the measurement

approach of brand personality to measure both product (e.g. Aaker, 1997; Bosnjak

and Hufschmidt, 2007; Geuens, Weijters, and De Wulf, 2009; Plummer, 1985) and

corporate brand/reputation (Davies, Chun, da Silva, and Roper, 2001; Slaughter,

Zickar, Highhouse, and Mohr, 2004; Whelan, Davies, Walsh, and Bourke, 2010). This

questioning approach typically asks respondents to “imagine that the

company/product has come to life as a human being” and “to rate its personality”.

Rather surprisingly the author could find no previous study that compares the results

of using one or the other approach. The main aim here is then to test whether there is

a difference between using personification or non-personification approaches when

measuring brand image, or is direct questioning (e.g. Hsieh, 2002) perfectly adequate?

Brand personality is an example of a projective technique that has both advantages

and disadvantages (Geuens et al., 2009). The most noted disadvantage is that it could

be considered as unscientific and potentially misleading (Davies et al., 2001). The

use of brand personality relies upon the acceptance of a brand being a person, the

personification metaphor, and the use of metaphor in research has attracted wide

criticism. Supporters of the use of metaphor in research however claim that metaphors

guide our perceptions and interpretations of reality (Ashton et al., 2004). Furthermore

respondents might be willing to reveal attitudes that they are reluctant to admit to

under direct questioning (Boddy, 2005). For instance if you ask questions to

50

employees about their employer directly, they might have some hesitation in

answering questions about how much they trust their employer openly and honestly.

Geuens et al. (2009) justify the use of the personification approach over a direct

questioning approach for measuring brand image, as consumers tend to prefer brands

with a strong personality to enhance their own self-image. The thinking is compatible

with that of (Fournier, 1998) who argues that people see their favorite brands as if

they were people with whom they want a relationship and with that of Belk (1988)

and others who emphasise the role of brands in building one’s identity. Consequently,

Geuens et al. (2009) argue that using a brand personality scale, particularly one that

resembles a human personality scale, allows brand managers to create an appropriate

brand personality for their target group.

Different measurement approaches can be expected to yield different results.

Nevertheless, there is a lack of consensus on what is a valid measurement method,

and this situation has been criticized (e.g. Nguyen and Leblanc, 2001) as a lack of

consensus on validity can lead to ineffective management of both brand image and

corporate reputation (Sarstedt, Wilczynski, and Melewar, 2013) as image has for

example an important effect on consumer loyalty (Nandan, 2005).

Brand Image and Personality

Brand image, be it the image of a product or service or that of a corporate, can be

described as the consumers’ perception and interpretation of the brand’s identity (De

Pelsmacker, Geuens, and Van den Bergh, 2005; Keller, Apéria, and Georgson, 2008).

51

Brand personality in turn can be described as ‘the set of human characteristics

associated with a brand’ (Aaker, 1997, p. 347), and is regarded as a way to measure

brand image (Keller et al., 2008). Brand image and brand personality measures are

both multidimensional in nature (Geuens et al., 2009; Keller et al., 2008; Malhotra,

1988). The original scale of Aaker (1997), for example contained 5 dimensions, but

personality scales published since 1997 have contained sometimes different numbers

of dimensions and even different dimensions altogether (Geuens et al., 2009). As it

would be impractical to research all the dimensions identified in the latter review, a

decision was made to limit the number by looking to two theories of brand imagery

that have been used to identify what might be fundamental or universal dimensions

(Davies, Chun, da Silva, and Roper, 2004). These should be relevant, irrespective of

whether the measurement approach is personified or direct.

The Stereotype Content Model and Signaling Theory

The stereotype content model derives from social cognition theory, and suggests that

people evaluate others on the basis of their ‘warmth’ and ‘competence’. In other

words, humans make their decisions about others on the basis of their perceptions of

whether they are friendly and reliable. When the earliest humans first met another

group, they needed to look initially for indications of warmth (the intentions of the

others) and then their competence to enact their intentions towards them (Fiske,

Cuddy and Glick, 2007; Willis and Todorov, 2006). This sensitivity to potential

threats is argued to be a crucial survival trait historically, such that only those who

made such judgements and did so correctly survived, explaining why humans use

such judgments (often unconsciously) today. According to Fiske et al. (2007), warmth

52

judgments are primary, ‘which reflects the importance of assessing other people’s

intentions before determining their ability to carry out those intentions’ (p.79). The

theory has been applied to brand perception arguing that humans automatically look

to the imagery of a brand on the same two dimensions. Kervyn, Fiske, and Malone

(2012) stated “consumers assess brands’ perceived intentions and abilities, which

elicit certain emotions and drive consumer behaviour.” (Kervyn, Fiske and Malone,

2012, p.165). Prior research has shown that cultivating warmth and competence

results in admiration towards a brand (Aaker, Garbinsky, and Vohs, 2011).

Moreover, a lack of competence leads to negative feelings such as pity, resentment or

anger, and a lack of warmth leads people to experience negative feelings such as envy

or jealousy (Caprariello, Cuddy, and Fiske, 2009; Fiske, Cuddy, Glick, and Xu, 2002).

Signaling theory argues that brands are evaluated for their potential to signal status

(Nelissen and Meijers, 2011) and specifically to the rest of the human population

(Han, Nunez, and Drèze, 2010). According to Han, Nunez, and Drèze (2010), in the

Middle Ages people were bound by rules that specified what each social class was

permitted or forbidden to wear. They argue that even though today, while there are no

such laws, and anyone with adequate wealth can purchase the items that they want to

have, people still tend to try to distinguish themselves from others or try to position

themselves in a certain class through the imagery of the products or brands they

choose to be associated with. Especially, and in order to increase social capital,

people tend to associate themselves with luxury brands (Bouska and Beatty, 1978;

Nelissen and Meijers, 2011) because luxury brand consumption is considered to be a

way of indicating your place in society (Chadha and Husband, 2010).

53

According to Nelissen and Meijers (2011), human beings’ preferences for luxury

products come from the universal tendency for signaling traits that might raise status

(e.g. Cummins, 2005). This tendency is not only relevant for humans but is also

shared by other social primates (de Waal, 1982). Nelissen and Meijers (2011) showed

that people who display luxury brand labels on their clothing are considered to have

more status and this leads to benefits in social interactions. Signaling theory can be

also viewed from an evolutionary perspective. Individuals signal favorable traits with

their possessions, leading to their being preferred as mating partners (Fehr and

Fischbacher, 2003). In a hierarchical society, an ability to recognize dominance

signals (dominance is one aspect of status) can also be a survival skill (Setchell and

Wickings, 2005).

This thesis will focus mainly on two dimensions: Warmth and Competence, partly

because status did not emerged strongly in the first study, and secondly because it was

believed that Warmth and Competence were the most important dimensions to

consider. As Davies, Rojas-Mendez, Whelan, Mete, Loo (2018) note a number of

dimensions of brand personality are common to the vast majority of brand personality

scales including these two. Further, Davies, Chun, daSilva, and Roper (2002) found

that the majority of variation in satisfaction towards brands was explained by these

two dimensions.

The stimulus of a brand name might evoke one of a number of schema-based

stereotypes (Grohmann, 2009). The last mentioned paper for example uses gender as a

framework for brand personality associations. Aaker’s (1997) paper on brand

personality included the dimension of gender. For example a brand like L’Oreal

54

would have more feminine associations than a brand like Lynx, or Gillette. The SCM

argues two rather different dimensions (Warmth and Competence). The implications

for brands are that when faced with a new brand name, for example, potential

customers would look for information relevant to the brand’s warmth and

competence. In other words warmth and competence represent stereotypes in the same

way as masculine and feminine. The schema of these two might even be linked with

each other (Ko, Judd, and Stapel, 2009).

To sum up, because they are common to most image measurements and supported by

theory, three dimensions of brand imagery; Warmth (often labeled as Agreeableness

in brand personality measures), Competence (again frequently apparent in brand

personality measures), and Status were chosen for this study.

Research Method and Hypotheses

Hypotheses

As reviewed earlier, a number of sources recommend the use of projective techniques

when asking respondents to undertake difficult evaluations such as that for brand

image (e.g., Boddy, 2005), others argue that consumers might like the idea of brands

having personality to make it easier to decide on the relevance of buying a particular

brand to their self-image (Geuens et al. 2009). However many criticize the use of

metaphor in research as being unscientific (Davies, Chun, da Silva, and Roper, 2004).

To counter the criticism that personification is not as valid as direct questioning, there

has to be some clear advantage in using it. Hence the following hypothesis was

proposed for testing:

55

Hypothesis 1 (H1): The Personification approach provides a better explanation of

dependent variables such as reputation, satisfaction and purchase than direct

measurement.

Secondly, one might expect personification to be superior to direct questioning when

a brand has more obvious humanistic associations (Geuens et al., 2009). The imagery

of corporate brands (particularly service brands where people’s contact with them is

via other human beings) rather than product brands should be more easily accessible

using the personification approach. Hence:

Hypothesis 2 (H2): The Personification approach provides a better explanation of

dependent variables such as reputation, satisfaction and purchase for corporate brands

than for product brands.

In addition it is necessary to check on other relative measures of validity for the two

approaches.

Methodology

In order to test the hypotheses, a 2 (a corporate brand vs. a product brand) x 2

(personification method vs. non-personification method) factorial, between-subjects

design was used in an online survey. Half the sample would assess brand imagery by

responding to direct questioning, half to personified questions; half would assess a

product brand, half a corporate brand.

56

Between-subjects designs typically serve the researcher well when time is at a

premium or testing/order effects are to be avoided, but only when there are plenty of

participants available, whereas within-subjects designs help to conserve participant

resources and are helpful when the goal is to directly compare multiple products.

Nevertheless, the decision to use a between- or within-subject design implies a trade-

off (Charness et al., 2012); as a within-subject design will limit the number of tasks

that can be examined. In this study, there is more than one treatment to investigate

(personification vs direct on different DVs, and personification vs direct on different

brand types). A within-subject design should be avoided in such studies (Greenwald,

1976), as Poulton (1973, 1974) points out that when using a within-subject design

(repeated measures design), the context provided by exposure to other treatments

(“range effect”) may often alter the effect of a given treatment. The greatest advantage

of using a between-subject design is to eliminate the possibility that an initial stimulus

can influence how respondents perceive and respond to subsequent stimuli (carry-over

effects; Davis and Bremner, 2006). Consequently, a between-subjects experimental

design is adopted for this study.

To maximize responses, it was important to choose brands that were widely known,

and therefore easily assessed. Accordingly, the corporate brand example was chosen

as the retailer Marks and Spencer (M&S), which is one of the leading retailers in the

British market where this study was undertaken. The product brand was chosen as

Pantene (a Procter and Gamble shampoo brand), which was the number one in its

category in the UK at the time of the survey.

57

The respondents for the study were members of a consumer panel and a convenience

sample of 360 of them were randomly assigned to each of the four groups. The panel

was operated by market research company: Pure Profile. Membership of the panel is

fully representative of the population of the UK, where the research was undertaken.

Panel members are rewarded for their participation with points, which they can collect

in exchange for products. The company hosted our questionnaires and contacted panel

members, inviting them to participate. They provided totally anonymous data to us,

and fully completed surveys only. The final sample consisted of 85 (47.5%)

respondents assessing Pantene using the personified version, and 94 (52.5%)

respondents assessing M&S using the same questionnaire. The final sample using the

direct approach was 88 (48.6%) respondents for the Pantene version, and 93 (51.4%)

respondents for the M&S version of the same questionnaire (Table 1).

Questionnaire Type

Gender of Respondents Brand

Number of Respondents

Percentage of Respondents

Direct Approach

Male Pantene 45 47.9 M&S 49 52.1 Total 94 100

Female Pantene 40 47.1 M&S 45 52.1 Total 85 100

Personification Approach

Male Pantene 45 48.4 M&S 48 51.6 Total 93 100

Female Pantene 43 48.9 M&S 45 51.1 Total 88 100

Table 1. Questionnaire Type Distribution According to Gender and Brand

Two filter questions were included to ensure that respondents were responsible for

their own shopping (and specifically for purchasing their own shampoo in the Pantene

58

surveys). A second question ensured respondents had been residents in the UK for at

least a year.

After the filter questions, demographic questions about the respondent’s age, gender

and education level were asked. The final sample consisted of 173 women (48.1%)

and 187 men (51.9%).

Six attitude questions were included (dependent and control variables) to evaluate

respondents’ attitude and behavior towards the brand being assessed and taken from

Davies et al. (2001). To measure satisfaction, the respondents were asked whether

they would recommend the chosen brand to others, whether they would be pleased to

be associated with the selected brand, and whether they would feel an affinity with the

chosen brand. Additionally and to measure reputation, they were asked whether the

selected brand offers good value for money, whether it is a good quality

shampoo/company, whether it has a good reputation as a brand/company.

Satisfaction and purchase intentions are two of the most commonly used outcome

variables for research into both product and corporate brands (eg. Martenson, 2007,

and Davies et al. 2002). In addition two measures of attitude towards to the brand

were included; one, the reputation of the brand/company, secondly a more general

attitude (again reflecting the outcome variables used by Davies et al. 2002). These

outcomes can be expected to be related to each other, for example customer

satisfaction and reputation have been found to be linked, even though they are

conceptually distinct (Walsh, 2009). Moreover, Keller (2013) claims a positive brand

image and brand awareness leads to increased customer satisfaction.

59

After the attitude questions, two questions were included to assess involvement,

adapted from previous research (Laurent and Kapferer, 1985; Krugman, 1977;

Zaichkowsky, 1985; Hupfer and Gardner, 1971) to control for any effects on response

patterns. Specifically the respondents were asked whether they chose where they shop

(for M&S) or the brand of shampoo they buy ‘carefully’, and whether they are

interested in shopping or shampoo brands. (The questionnaires can be found in

Appendix 1 to the thesis).

For similar reasons two questions were included to measure the respondents’

expertise, whether their friends and family tell them that they are good at choosing the

best brands, and whether friends and family ask them for advice about shopping/

shampoo brands (Mitchell and Dacin, 1996; Alba and Hutchinson, 1987). All

questions used the same response scale from 1 to 7 with points 1, 3 and 7 labeled

strongly disagree, neither agree nor disagree and strongly agree.

After the expertise questions, an open ended question was inserted by asking the

respondents to write down their thoughts about either Pantene or M&S. An open-

ended question was placed at this stage to distract from any linkage to the next part of

the questionnaire in which the brand image measuring questions were asked and to

explain any unusual responses.

Then, items for each of the three dimensions of brand image (warmth, competence

and status) were included, selected from published measures (see for example,

Geuens et al., 2009) and chosen to be equally valid in both questioning formats, direct

60

and personification. For the Warmth dimension the brand image items selected were:

friendly, helpful, trustworthy, ethical, sincere, honest, and socially responsible (from;

Aaker, Vohs, and Mogilner, 2010; Davies et al., 2004). For the Competence

dimension: successful, leading, reliable, strong, and intelligent. Finally, for the Status

dimension: sophisticated, prestigious, up-market, and chic (from Aaker, 1997; Davies

et al., 2004).

A five point, Likert type scale was used to assess each item in this part of the survey

with each point labeled from strongly agree to strongly disagree. The scaling

approach in the online survey was varied between question types (sometimes a tick

box, sometimes a sliding scale) to reduce any Common Methods Variance (CMV)

effects. For the personification variants respondents were asked to ‘Imagine that

Marks and Spencer has come to life as a person, what would his/her personality be

like?’ And then to rate the 15 image questions. For the direct measurement versions

respondents were not given any such preamble.

Finally, the respondents were asked two purchase questions in terms of how often

they shop at M&S, or buy Pantene shampoo; and how often they think they will shop

at M&S, or buy Pantene shampoo in the future. Each was assessed on a five-point

scale from ‘never’ to ‘frequently’.

61

Results and Discussion

First, the scales used to assess the three dimensions were checked for reliability with

Cronbach Alpha. The Cronbach Alpha’s for each of the four groups were reliable

(Table 2), all being above 0.8 (Nunnally, 1978; Peterson, 1994).

Group Warmth Competence Status M&S (Personification) .94 .90 .90 M&S (Non-Personification) .96 .95 .86 Pantene (Personification) .95 .95 .95 Pantene (Non-Personification) .93 .93 .91 Table 2. Cronbach’s Alpha Values of Dimensions by Groups

Then the data were tested for the homogeneity of variances assumption (HOV), which

stipulate whether the data has similar variances between measure type (Bryk and

Raudenbush, 1988). This tests whether using one method or the other gives a

different mean score for (as an example) the Warmth scores for Pantene using either

method. Levene (1960) states that comparing the sample means, one should check

that the underlying populations have a common variance, and proposes that in order to

check the homogeneity of variances, the F-test is to applied to the absolute deviations

of the observations from their group means (Gastwirth, Gel, and Miao, 2009).

(For the Levene Test the standard deviation for comparative measures should be equal

if the measures are similar)

62

Method Brand Dimension Mean F value P value Personification Pantene Warmth 3.72 0.12 0.73 Personification M&S Warmth 3.68 Direct Q Pantene Warmth 3.66 1.73 0.19 Direct Q M&S Warmth 3.80 Personification Pantene Competence 3.83 0.53 0.47 Personification M&S Competence 3.75 Direct Q Pantene Competence 3.99 3.62 0.06 Direct Q M&S Competence 3.77 Personification Pantene Status 3.55 0.02 0.90 Personification M&S Status 3.53 Direct Q Pantene Status 3.53 0.18 0.67 Direct Q M&S Status 3.58 Table 3. Means and Levene’s Test for Equality of Variance Values for Each Group and Dimension (If p≤ 0.05, the variances are unequal and one approach gave a significantly different

result. No difference is significant.)

Somewhat surprisingly, when 2 way ANOVA was used to see whether there were any

main or interaction effects from measure type (direct or personified) and brand type

(product or corporate) in predicting the three dimensions of brand image (warmth,

competence, and status) there were no such significant effects.

The data were then tested to see whether either measurement approach predicted

greater variance in the potential dependent variables included in the survey. For this, a

mean score of the items measuring purchase intention, satisfaction and reputation

were used as dependent variables. (Each measure was valid with alphas or inter-item

correlations above 0.8). The predictive ability of the two approaches is compared in

Table (4) using the adjusted R2 for each of the three. The initials in brackets in the

first column indicate whether the respondents had been given the personification

version (P) or the direct questioning version (DQ) of the survey.

63

Model R2 Purchase R2 Reputation R2 Satisfaction M&S (P) .30 .60 .46 M&S (DQ) .43 .63 .66 Pantene (P) .53 .49 .63 Pantene (DQ) .27 .52 .49 Table 4. Adjusted R-Square Values of Dependent Variables by Context

As it can be seen from the table above (Table 4); in some cases personification gave

the higher prediction of variance (measured by R2) in others it was the direct

approach. When covariates were added (age, gender, education, expertise, and

involvement) the picture did not change. The hypotheses imply that the highest R2

would be for M&S with the personified measure and the lowest for Pantene with the

direct measure. Neither was true. The next analysis also considers each of the three

dimensions separately in predicting the three dependent variables. A Chow test can be

used to compare the sum of squared residuals (SSR) at this level.

Dependent Variable

Brand Dimension Method SSR Chow F Statistic

Significant or not

Satisfaction Pantene Warmth Personification 54.40 0.22 Not Significant Satisfaction Pantene Warmth Direct 66.69

Satisfaction M&S Warmth Personification 111.16 0.57 Not Significant Satisfaction M&S Warmth Direct 63.77

Satisfaction Pantene Competence Personification 59.30 3.47 Significant Satisfaction Pantene Competence Direct 83.63 Satisfaction M&S Competence Personification 133.80 0.93 Not

Significant Satisfaction M&S Competence Direct 86.27 Satisfaction Pantene Status Personification 76.63 0.50 Not

Significant Satisfaction Pantene Status Direct 89.11 Satisfaction M&S Status Personification 133.26 0.43 Not

Significant Satisfaction M&S Status Direct 95.85 Satisfaction Pantene All Personification 51.24 1.21 Not

Significant Satisfaction Pantene All Direct 66.83 Satisfaction M&S All Personification 107.60 0.46 Not

Significant Satisfaction M&S All Direct 60.45 Table 5 A. Chow Test for Each Dimension and Method when Predicting Satisfaction

64

Dependent Variable


Significant or not

Purchase Pantene Warmth Personification 73.20 2.79 Not Significant Purchase Pantene Warmth Direct 88.83

Purchase M&S Warmth Personification 102.96 0.42 Not Significant Purchase M&S Warmth Direct 76.55

Purchase Pantene Competence Personification 80.84 4.37 Significant Purchase Pantene Competence Direct 87.60 Purchase M&S Competence Personification 102.29 0.34 Not

Significant Purchase M&S Competence Direct 88.90 Purchase Pantene Status Personification 94.73 3.07 Nearly

Significant Purchase Pantene Status Direct 101.79 Purchase M&S Status Personification 112.67 0.43 Not

Significant Purchase M&S Status Direct 103.64 Purchase Pantene All Personification 71.99 3.68 Significant Purchase Pantene All Direct 85.20 Purchase M&S All Personification 97.15 0.96 Not

Significant Purchase M&S All Direct 75.66 Table 5 B. Chow Test for Each Dimension and Method when Predicting Purchase Dependent Variable


Significant or not

Attitude Pantene Warmth Personification 44.58 0.15 Not Significant Attitude Pantene Warmth Direct 44.69

Attitude M&S Warmth Personification 71.17 0.33 Not Significant Attitude M&S Warmth Direct 42.17

Attitude Pantene Competence Personification 43.82 3.78 Significant Attitude Pantene Competence Direct 50.92 Attitude M&S Competence Personification 87.67 0.16 Not

Significant Attitude M&S Competence Direct 59.91 Attitude Pantene Status Personification 61.14 0.41 Not

Significant Attitude Pantene Status Direct 63.99 Attitude M&S Status Personification 94.15 0.35 Not

Significant Attitude M&S Status Direct 70.71 Attitude Pantene All Personification 39.72 1.80 Not

Significant Attitude Pantene All Direct 41.12 Attitude M&S All Personification 66.04 0.23 Not

Significant Attitude M&S All Direct 38.29 Table 5 C. Chow Test for Each Dimension and Method when Predicting Attitude

65

Dependent Variable


Significant or not

Reputation Pantene Warmth Personification 76.43 0.33 Not Significant Reputation Pantene Warmth Direct 65.87

Reputation M&S Warmth Personification 82.57 0.53 Not Significant Reputation M&S Warmth Direct 59.73

Reputation Pantene Competence Personification 67.31 0.80 Not Significant Reputation Pantene Competence Direct 53.67

Reputation M&S Competence Personification 77.68 1.04 Not Significant Reputation M&S Competence Direct 54.85

Reputation Pantene Status Personification 84.29 0.21 Not Significant Reputation Pantene Status Direct 70.27

Reputation M&S Status Personification 93.67 0.86 Not Significant Reputation M&S Status Direct 92.30

Reputation Pantene All Personification 66.84 1.32 Not Significant Reputation Pantene All Direct 50.60

Reputation M&S All Personification 68.59 0.75 Not Significant Reputation M&S All Direct 46.20

Table 5 D. Chow Test for Each Dimension and Method when Predicting Reputation

The Chow test (Chow, 1960) was initially designed to study the same variables

obtained in two different data sets to determine if they were similar enough to be

combined together (Lee, 2008). Here it is used to test if the coefficients and intercepts

of linear regressions on different data sets are equal e.g. is the regression between the

personified measure and the direct measure and a DV the same?

If the Chow F statistic is greater than the critical F-value, one can conclude the

regression lines of the two data sets are different. The results of the Chow Test show

only some significant differences, the outcome for Pantene for the competence

dimension when predicting satisfaction, purchase, and attitude gives a significant F-

value between the direct and personified approaches (Table 5A-C), and for all three

dimensions when predicting purchase (Table 5B). But this is only four findings

among 32 comparisons, about the frequency that might be expected from random

66

chance. The Chow test however may not be the most appropriate here as the variance

predicted in the three DV’s is quite high for all equations.

Next the Fisher test was used to evaluate, for example, the correlation between

Warmth and Satisfaction is better than that when using a personified measure (P)

compared with using a direct measure (D). First Fisher’s r to z transformation was

applied to the correlations, when predicting four dependent variables; satisfaction

(SAT), purchase (PUR), reputation (REP), and attitude (ATT). The effect of this

transformation is to make the sampling distribution of the transformed coefficient

nearly normally distributed (Kenny, 1987). The critical value of Z is 1.96, when p <

.05. The Fisher’s r to z transformation results for satisfaction and purchase did not

provide a consistent pattern.

DV Brand Dimension Method Pearson

R N Fisher’s z

transformation P value

Significance

SAT Pantene Warmth P 0.79 85 1.32 0.08 Not Significant SAT Pantene Warmth D 0.70 88

SAT M&S Warmth P 0.67 94 -1.92 0.03 Not Significant SAT M&S Warmth D 0.80 93

SAT Pantene Competence P 0.77 85 2 0.02 Significant SAT Pantene Competence D 0.61 88 SAT M&S Competence P 0.59 94 -1.57 0.06 Not

Significant SAT M&S Competence D 0.72 93 SAT Pantene Status P 0.69 85 1.2 0.11 Not

Significant SAT Pantene Status D 0.58 88 SAT M&S Status P 0.59 94 -1.05 0.15 Not

Significant SAT M&S Status D 0.68 93 SAT Pantene All P 0.79 85 1.44 0.07 Not

Significant SAT Pantene All D 0.68 88 SAT M&S All P 0.67 94 -1.9 0.03 Not

Significant SAT M&S All D 0.80 93 Table 6 A. Fisher’s R to Z transformation When Predicting Satisfaction

67


R N Fisher’s z


Significance

PUR Pantene Warmth P 0.70 85 2.49 0.006 Significant PUR Pantene Warmth D 0.45 88 PUR M&S Warmth P 0.54 94 -1.26 0.10 Not

Significant PUR M&S Warmth D 0.66 93 PUR Pantene Competence P 0.66 85 1.92 0.03 Not

Significant PUR Pantene Competence D 0.47 88 PUR M&S Competence P 0.54 94 -0.41 0.34 Not

Significant PUR M&S Competence D 0.58 93 PUR Pantene Status P 0.59 85 2.36 0.009 Significant PUR Pantene Status D 0.30 88 PUR M&S Status P 0.47 94 -0.08 0.47 Not

Significant PUR M&S Status D 0.48 93 PUR Pantene All P 0.68 85 2.35 0.009 Significant PUR Pantene All D 0.44 88 PUR M&S All P 0.56 94 -0.66 0.25 Not

Significant PUR M&S All D 0.62 93 Table 6 B. Fisher’s R to Z transformation When Predicting Purchase


R N Fisher’s z


Significance

REP Pantene Warmth P 0.64 85 -0.13 0.45 Not Significant REP Pantene Warmth D 0.66 88

REP M&S Warmth P 0.71 94 -1.62 0.05 Not Significant REP M&S Warmth D 0.81 93

REP Pantene Competence P 0.70 85 -0.48 0.32 Not Significant REP Pantene Competence D 0.73 88

REP M&S Competence P 0.73 94 -1.67 0.05 Not Significant REP M&S Competence D 0.83 93

REP Pantene Status P 0.60 85 -0.31 0.38 Not Significant REP Pantene Status D 0.63 88

REP M&S Status P 0.66 94 -0.29 0.39 Not Significant REP M&S Status D 0.68 93

REP Pantene All P 0.68 85 -0.75 0.23 Not Significant REP Pantene All D 0.74 88

REP M&S All P 0.76 94 -1.55 0.06 Not Significant REP M&S All D 0.84 93

Table 6 C. Fisher’s R to Z transformation When Predicting Reputation

68

DV Brand Dimension Method Pearson R

N Fisher’s z transformation

P value

Significance

ATT Pantene Warmth P 0.94 85 1.08 0.14 Not Significant ATT Pantene Warmth D 0.92 88

ATT M&S Warmth P 0.91 94 -0.92 0.18 Not Significant ATT M&S Warmth D 0.93 93

ATT Pantene Competence P 0.96 85 2.26 0.01 Not Significant ATT Pantene Competence D 0.92 88

ATT M&S Competence P 0.93 94 -0.26 0.40 Not Significant ATT M&S Competence D 0.93 93

ATT Pantene Status P 0.94 85 2.15 0.02 Significant ATT Pantene Status D 0.89 88 ATT M&S Status P 0.93 94 1.2 0.11 Not

Significant ATT M&S Status D 0.90 93 ATT Pantene All P 0.80 85 0.45 0.33 Not

Significant ATT Pantene All D 0.80 88 ATT M&S All P 0.76 94 -1.65 0.05 Not

Significant ATT M&S All D 0.85 93 Table 6 D. Fisher’s R to Z transformation When Predicting Attitude

Overall there are 5 instances in Tables 6(A-D) when the personified measure has a

significantly higher correlation with a DV than for a direct measure at p<0.05.

However there are two examples where the direct measure gives the higher

correlation and where the significance is almost valid at 0.05. Out of 16 comparisons

only 5 support H1. All 5 are when Pantene was the brand being evaluated. There is

then no support for H2.

Next Model Fit Analysis was carried on for three dimensions using 15 brand image

items and Structural Equation Modelling (AMOS 22) to test whether the differences

between the two types of measure were invariant. Normally such a test is used to

explore, for example, whether a measure differs between two genders. Here it is used

to test whether the same measurement model fits the data from personified and direct

measures. Examination of the Model fit statistics and the modification indices

suggested adding covariances between several errors of the items to obtain a well

fitting model. The final model fitted the combined data well with a CMIN/DF=3.206

69

where the upper threshold is 5, with a significant p-value (due to the large sample

size). GFI= 0.910, AGFI=0.871, CFI=0.965, which are acceptable, Hoelter in 143 and

157 for .05 and .01 indices respectively, and finally RMSEA=. 078. (Figure 1), shows

that SEM model fits data well.

70

Figure 1. Warmth, Competence and Status Dimensions with all Items

Multigroup analysis was used to compare the model fit when using personified and

direct data (Table 7), The models do not differ, suggesting the measures are very

similar.

71

Model DF CMIN P NFI Delta-1

IFI Delta-2

RFI rho-1

TLI rho2

Measurement weights 12 6.286 .901 .001 .001 -.005 -.005 Structural covariances 18 18.132 .447 .003 .003 -.006 -.006 Measurement residuals 37 50.539 .068 .009 .009 -.008 -.009 Table 7. Comparing Direct and Personified Data

In order to assess relative construct validity, Confirmatory Factor Analysis was used

(Jöreskog, 1967). The Average Varience Extracted (AVE) and Composite Reliability

(CR) have been used to assess the convergent validity of the measurement model

(Fornell and Larcker, 1981). The results are found to be very good for all three

dimensions and both measurement approaches (see Table 8) with the direct approach

showing slightly better figures than those for the personified approach.

Dimension Measurement Approach AVE CR

Agreeableness Personification 0.99 1.00 Agreeableness Direct 1.07 1.01 Competence Personification 0.96 0.99 Competence Direct 1.11 1.02 Status Direct 0.94 0.98 Status Personification 1.05 1.01 Table 8. AVE and CR Results According to Dimension and Measurement Approach

There is therefore no support for H1 that the Personification metaphor provides a

better explanation of dependent variables such as reputation, satisfaction and

purchase.

The data support H1 for Pantene for satisfaction and purchase but not reputation, and

provide no support in the case of M&S. Consequently H2, that Personification

72

metaphor is more useful for corporate brands than non-personification methods of

measurement, was not supported.

Finally the idea of personification is more relevant when respondents might be

reluctant to provide responses was examined. While the selection of brands for this

study was intended to be gender neutral and women use Pantene as much as men, the

brand is marketed exclusively at women. When the analysis to predict satisfaction

(Table 2) were repeated but separately for the two genders, the R2 for male

respondents under the direct questioning approach was smaller (0.34) compared to

that for female respondents (0.72) but the figures for the personification approach

were similar, with the regression for males yielding a slightly higher figure (0.66)

compared with that for females (0.61). This suggests that males might have been

reluctant to admit to an affinity with a female oriented brand unless they were giving

responses under personification.

Managerial Implications

Managers can use the findings of the present study to increase their ability to develop

a better understanding of how the two main approaches to brand image measurements

differ. This research investigated two well-known brands in Britain: a shampoo brand

Pantene and a clothing and grocery retailer Marks and Spencer. These results might

not be inclusive of all brand types but offer a useful basis for whether their choice of

measurement approach should differ from one the other. Managers can, then, decide

which approach they find more useful or efficient to measure their brand image.

73

Market research companies often use the projective and personified approaches in

asking questions in the same ways as academic researchers might choose to do. The

findings here suggest that there is little advantage in doing so. Worse, many research

companies have their own measures of brand personality (rather than brand image) as

this helps them market their services to practitioners. How valid such an approach

really is, particularly whether it has advantages over direct questioning, is called into

question here.

The idea that a personification approach might be more relevant when respondents are

reluctant to provide answers was investigated to understand whether either of the

brand image measurement approaches leads to a different response pattern for

respondents. The shampoo brand used in this study has a target group of female

consumers according to their marketing communications strategy and their use of

only female celebrities in their advertisements. Pantene is very much a female

shampoo brand, but there are male respondents to the survey who claim they purchase

and use this brand as well. This study shows that a direct questioning method leads

males to report low levels of satisfaction compared to females, and compared to male

respondents when using the personification approach. This could be due to Pantene’s

positioning strategy as a female brand, and men not wanting to show an affinity with a

female brand under direct questioning. Therefore in case of a sensitive situation as

this, the personification approach might be more useful to gather more fruitful

answers.

74

The research considered a number of outcomes including satisfaction and purchase

intention. While the purpose of the research was to compare measurement

approaches, managers would be interested to see the extent to which these outcomes

are predicted by a brand’s warmth, competence and status, irrespective of how these

are measured. The lowest value of R2 predicted for purchase intention for example

was 0.27 (for Pantene using direct questioning) confirming the importance of

understanding how brand imagery influences outcomes of concern to managers. Even

higher values of R2 were observed for more affective outcomes including satisfaction

with the brand.

Conclusions and Further Work

This study shows that personification, as a measurement approach is not a guarantee

of a better explanation of outcome variables such as brand satisfaction than the, less

controversial, direct questioning approach. There was only limited support for the use

of a personified approach for the measurement of brand image. The more specific

hypothesis that personification might be a more relevant approach in the context of a

corporate or service brand was not supported at all. The findings cannot be used to

‘prove’ that there is no apparent benefit in using personification as there was a

marginal benefit in using the approach at times, one slightly greater than chance. The

analysis was run separately by subgroup (for example gender), and there were no

significant differences between any of the subgroups.

While there is some evidence that respondents might benefit from a personified

approach when evaluating a brand that they find difficulties with (men and Pantene)

the evidence is far from conclusive as the sample sizes were quite small.

75

Perhaps the most compelling evidence came from the multi-group analyses where the

measurement models were found to be indistinguishable between the two approaches.

Given the criticism of the personification approach in that it relies upon the metaphor

that a brand is a person, something that is untrue (Davies et al., 2001), then

researchers may wish to adopt the perfectly adequate, direct style of questioning.

It was challenging to find items that worked for both the personification and direct

questioning contexts. For example items such as ‘supportive, hardworking and open’

appear in measures of brand personality but could be relatively meaningless in a

direct measure of brand image. However this could also provide an advantage for

personification in that more items appear relevant to that context, an effect not tested

here.

The study has its limitations as only two, well-known and generally well liked, brands

were used and the findings may be specific to them and to this context. Furthermore

only a limited number of items were used to measure brand image, as the concern was

to include items that appeared relevant to both personification and direct questioning.

Using a larger number from typical brand personality scales might favour

personification.

Consequently further work might usefully allow the personified variant to be longer

than its direct equivalent or at least include more items that may or may not be

relevant to a direct approach. Similarly further work might consider a larger number

of brands and in a context where respondents might be more reluctant to give open

76

replies. Finally for the most part there is little pattern to the results from this study

other than that there is no difference between the two approaches. It would be good to

explore and test explanations for why the hypotheses were not supported and in

particular why evaluating a corporate brand appears to be more problematic than a

product brand.

77

Chapter 4.1: Connecting Sub-Chapter 1: Changing Context from Product and Corporate Brand to Employer Branding when Measuring Brand Image

The purpose of this section is to bridge between the first and second of the three

papers that form the core of this thesis.

From the first paper, when measuring brand image, there are two main approaches;

namely personification and non-personification. These two approaches were

investigated in the previous chapter with two types of brand: product and corporate.

Three dimensions in terms of warmth, competence, and status were used to

investigate brand imagery. The status dimension was not found to be as relevant as

the first two in predicting outcomes.

It would be interesting to investigate the same issues but in a second context, that of

employer branding, for a number of reasons: to replicate the findings from the first

study; to assess the hypotheses in a different context and; in one where respondents

might be more reluctant to respond to direct questioning. The next study/chapter then

changes the context from product and corporate branding to employer branding to

investigate the same main research questions. The second of two studies study is

limited to considering warmth and competence as image dimensions. The number of

measurement items for both is increased in the second study to consider whether a

larger number might favour personification as well as order effects. By surveying

respondents as employees a much wider range of ‘brands’ can be considered

including those that might not be as well-known as Marks and Spencer and Pantene.

Inevitably, given the nature of this thesis, there is some overlap between the papers,

particularly in the literature reviews.

78

Chapter 5

How Best to Measure Employer Brand Image:

Personification versus Direct Methods

79

How Best to Measure Employer Brand Image: Personification versus Direct Methods Abstract Two studies (N=221 and N= 440) are reported, both aimed at identifying whether a

personified or a direct form of questioning should be preferred in the measurement of

employer brand image. Two dimensions of brand image are considered in both,

labeled warmth and competence, as suggested by the application of the stereotype

content model (Fiske, Cuddy and Glick, 2007) to the study of brand image.

In both studies members of the public were asked to evaluate their employer. In Study

1 respondents were each asked to evaluate their employer’s image using either a

personified or a direct measure. To test for any possible dimension specific or order

effects, Study 2 uses a between subject factorial design where half of the respondents

evaluated their employer for warmth, half for competence, half using a personified

approach, half a direct approach to measurement.

No systematic benefit for the use of personification was found in either study.

Differences between the predictivity of individual dimensions in Study 1 were not

confirmed in Study 2.

80

Introduction Employer branding is crucial for several reasons, such as improving employee

retention (Park, Jaworski, and Maclnnis 1986), and as an internal marketing strategy,

“by systematically exposing workers to the value proposition of the employer brand,

the workplace culture is molded around the corporate goals” (Backhaus and Tikoo,

2004). Job applicants tend to be attracted to organizations with traits similar to their

own personality traits (Lievens, 2007). Moreover, Slaughter, Zickar, Highhouse, and

Mohr (2004) showed that symbolic image dimensions were related to organizational

attractiveness for potential employees.

Employer branding is defined as “a targeted, long-term strategy to manage the

awareness and perceptions of employees, potential employees, and related

stakeholders with regards to a particular firm” (Sullivan, 2004, cited in Backhaus and

Tikoo, 2004, Alniacik and Alniacik 2012). It is also seen as the “sum of a company’s

efforts to communicate to existing and prospective staff that it is a desirable place to

work” (Lloyd 2002; Berthon, Ewing, and Hah, 2005).

Employer branding has arisen as a result of the application of marketing principles to

human resource management (i.e. internal marketing) (Backhaus and Tikoo, 2004;

Cable and Turban, 2001). It was initially argued that the notion of “internal

customers” needed to be introduced (Ewing and Caruana, 1999). This concept claims

that organisations’ employees are indeed the first market for their companies (George

and Gronroos, 1989), as jobs are internal products, and employees are internal

customers (Berthon et al., 2005). Consequently, the internal marketing concept

argues that jobs, as products, must attract and motivate employees, thus satisfying the

81

needs and wants of these internal customers, at the same time addressing the overall

objectives of the organisation (Berry and Parasuraman, 2004).

The Advantages of Employer/employee Branding

Research conducted by Hewitt Associates suggests that the major benefits of

employer branding are enhanced recruitment, retention and employee engagement and

commitment (Barrow and Mosley, 2005, 69, cited in Härkönen, 2015).

“A good employer brand image can be crucial for companies in terms of profits. For

instance in 2012 the Boston Consulting Group together with the World Federation of

People Management Associations (WFPMA) conducted research with 4288 managers

in 102 countries. The results showed a correlation between companies having a strong

employer brand and business growth. Companies which invested in employer

branding, experience double the profit margin growth compared to their previous

results” (Mosley, 2014, cited in Härkönen 2015). A LinkedIn survey in 2011 with

2250 companies around the US showed that having a strong employer brand cuts the

cost per hire by half and reduces the cost of attrition by a quarter (Gultekin, 2011).

One of the greatest advantages of employer branding is that the employer brand

brings out an image showing the organization as a good, desirable place to work

(Sullivan, 2004 cited in Backhaus and Tikoo, 2004, and Llyod, 2002, and Ewing, Pitt,

de Bussy, and Berthon, 2002), consequently many firms either have developed or are

interested in developing employer branding programmes (Conference Board 2001,

cited in Backhaus and Tikoo 2004).

82

Similarly, in terms of corporate branding, it is important for employees to buy into

organizational values and programs, because the corporate brand identity serves as the

link between the organization and the customer. Therefore, it can play a key role in

articulating these elements to employees, retailers, and others who must buy into the

goals and values of the corporate brand and represent them to their customers (Aaker,

2004). Some authors argue that an employer branding programme involves clarifying

what is referred to as the “unique organisational value proposition” (Knox, Maklan,

and Thompson, 2000 and Martin, 2008 and Barrow and Mosley, 2005, cited in

Edwards, 2009).

Consequently, to enhance employees’ identification with the corporate brand and get

their support, corporate brand values have to reflect corporate values and culture

(Yaniv and Farkas, 2005). If there is a gap between the corporate brand values and the

actual corporate values, it will be perceived by the employees as a lie, encourage

cynicism and finally damage their identification with the corporate brand (Yaniv and

Farkas, 2005, Harris and de Chernatony, 2001). “This misbelief on the part of the

employees will be transferred to the customers and undermine their belief in the

corporate brand, which will eventually lead to an increasing gap between the brand

values and the way customers perceive them, and thus decreasing customer loyalty”

(Herman, 2001, cited in Yaniv and Farkas, 2005).

The more employees identify with their employer (that is, incorporate the

organization’s identity into their self-concept), the more internally motivated they are

to engage in behaviors that support organizational brand-building efforts, both on and

off the job (Löhndorf and Diamantopoulos, 2014). It is known that employees who

83

identify with their employer also provide better performance, engage in voluntary

citizenship behaviors, and express lower intentions to leave (Riketta, 2005, cited in

Löhndorf and Diamantopoulos, 2014).

Priyadarshi (2011) used employer brand image to predict employee satisfaction and

affective commitment. The results support a previous study by Davies (2008), where

satisfaction was determined by the friendly and supportive attributes of the

organisation. Similarly, Priyadarshi (2011) claims that maintaining good employer

branding can lead to employee satisfaction and commitment, and Kunerth and Mosley

(2011) argue it also leads to employee engagement as well.

The previously mentioned studies illustrate why it is important to measure the image

employees hold of their employer, so that employers can monitor their employees’

views and preempt any negative spillover onto customers or increased levels of

employee turnover. Researchers too need to be able to explore links between

employer brand imagery and organizational outcomes. This paper aims to contribute

to both needs by investigating which of two methods might be preferred in measuring

employee views of the employer brand.

Since Aaker (1997) formalized the concept of brand personality, many studies have

used the approach to measure (affective) brand image and many different scales have

been published to allow researchers to do so in various contexts (Geuens, Weijters,

and De Wulf, 2009). For example while Aaker’s scale was developed among

consumers and with consumer brands, the measure of Davies, Chun, da Silva, and

Roper (2001) used employees and customers of corporate brands and the Slaughter et

84

al. (2004) scale, potential employees. The approach typically involves asking

respondents to “imagine that the company/product has come to life as a human being”

and “to rate its personality”. However the use of brand personality to measure brand

image has proved controversial for a number of reasons (Davies, Chun, da Silva, and

Roper, 2004) including that using a metaphor (brand = person) is unscientific.

Consequently Mete (2017a) compared this approach with asking the same questions

but without asking the respondent to personify, arguing that there had to be some

advantage in using the personified approach to counter such criticism. She used only

two (well-known) consumer brands and found little difference when using either

approach. The aim here is to extend such work in a context where a larger number of

brands are considered and one where the personified approach might be expected to

be more useful in allowing respondents to evaluate a brand more openly than if asked

to respond to direct questioning.

Research Method and Hypotheses

The personification approach is claimed to be more useful in terms of being able to

provide comparatively richer information than direct questioning (Davies et al., 2001)

Following Mete (2017a) the aim is to test such claims by comparing, among other

metrics, the predictive validity of personified vs. direct questioning in explaining

relevant dependent variables. Employee satisfaction and engagement were chosen as

the dependent variables. Following Soane et al. (2012) three aspects of engagement

were tested: intellectual, social, and affective engagement. From such prior work we

can propose:

85

Hypothesis 1 (H1): Measuring employer brand imagery using a personified measure,

rather than an equivalent direct measure, explains more variance in dependent

variables such as satisfaction, intellectual, social, and affective engagement.

Mete (2017a) focused on three dimensions of brand image/personality (warmth,

competence and status) arguing that each could be supported as relevant to both

methods. In Study 1 the same three dimensions are used but in Study 2 the focus is on

the first two. One reason for this is that, as implied by previous studies (eg. Wojciszke

and Abele, 2008; Wojciszke, Dowhyluk and Jaworski, 1998), warmth and

competence judgments are made differently and as competence judgments (from such

prior work) involve greater processing, they are expected to be able to predict and

explain more variation in the dependent variables. Put another way, it is particularly

relevant to focus on any differences between warmth and competence.

Hence:

Hypothesis 2 (H2): When measuring employer brand imagery, the Warmth

dimension, irrespective of using a personified or direct approach, explains less

variance than the Competence dimension in dependent variables such as satisfaction,

intellectual, social, and affective engagement.

86

Study 1

Methodology

The hypotheses were tested in an online survey. Respondents, as employees, were

asked to evaluate the company that they were working for by using either the

personification approach or direct approach based. Thus, the sample was randomly

split into two and half of the respondents would assess their companies (the

companies that they work for) by responding to direct questioning, the other half by

responding to personified questioning. Both groups would respond to warmth,

competence, and status questions randomly ordered and mixed with other measures of

brand imagery so as not to lead the respondents.

A consumer panel was used with a convenience sample of 221 people randomly

assigned to each of the two groups, so that the number of participants in each group

would be approximately equal. More specifically, 48.9% of the respondents were

given a direct approach based questionnaire (n=108), whereas 51.1% of the

respondents were given a personification approach based questionnaire (n=113).

The surveys started with two filter questions to ensure the respondents were residing

in the UK, and were not self-employed.

Following the filter questions, demographic questions were asked in terms of the age

of the respondents, their gender, their education level, the number of years that they

have been working (in the workforce), and the number of years they had been

87

working for their current employer (company). The sample consisted of 119 males

(53.8 %), and 102 females (46.2%), which can be considered as a good balance

between the genders. Questionnaire type responses according to gender can be seen in

Table 1.

Table 1. Questionnaire Type Distribution According to Gender for Study 1

Then three satisfaction questions were asked. These dependent variable questions

included whether respondents would recommend the company that they work for,

whether they would be pleased to be associated with the company they work for, and

whether they would feel an affinity with the company they work for (adapted from

Davies et al., 2004). The response scale was from 1-7 with points 1, 3 and 7 labeled:


In addition to demographics a number of potential control variables were included.

The expertise of the respondents was measured using four statements for respondents

to rate themselves from 1 to 5 (strongly disagree to strongly agree), namely “ I think I

am good at judging if an organization is a good employer or not”, “I often ask my

friends about their work”, “ I am interested to compare how different employers treat

Questionnaire Type

Gender of Respondents



Direct Approach Male 52 48.1 Female 56 51.9 Total 108 100

Personification Approach

Male 67 59.3 Female 46 40.7 Total 113 100

88

their staff”, and “ My friends, family, and colleagues often ask my advice about work

matters” (adapted from Mitchell and Dacin, 1996; Alba and Hutchinson 1987).

Hovland, Janis, and Kelley (1953) defined expertise as “the extent to which a

communicator is perceived to be a source of valid assertions”. Adjectives such as

“expert”, “knowledgeable”, “experienced”, and “qualified” -all of which have been

found to be clear indications of expertise (Applbaum and Anatol, 1972;Simpson and

Kahler, 1980- 81; Ohanian, 1990).

In order to evaluate the engagement of the respondents, nine questions were included

to review intellectual engagement, social engagement, and affective engagement with

items adopted from Soane et al. (2012)’s Engagement Scale. Three questions concern

intellectual engagement (whether the respondents would focus hard on their work,

whether they would concentrate on, and whether they would pay a lot of attention to

their work), three social engagement (whether the respondents would share the same

work values as their colleagues, whether they would share the same work goals and

the same work attitudes as their colleagues) and three affective engagement (whether

the respondents would feel positive about their work, whether they would feel

energetic, and would be enthusiastic in their work). For each the same response scale

was used from 1 to 7 with points 1, 3 and 7 labeled strongly disagree, neither agree

nor disagree and strongly agree.

After the engagement questions, an open ended question was added by asking the

respondents to write down their thoughts about the company they work for. The open

ended question was placed at this stage to distract from any linkage to the next part of

the questionnaire where questions measuring employer brand image were included.

89

These questions consisted of brand image items to evaluate the three dimensions

(warmth, competence and status) and were selected from published measures, as

equally valid in both questioning approaches, namely direct and personification.

For the Warmth dimension the employer brand image items were selected as friendly,

sincere, agreeable, open, and socially responsible. For the Competence dimension;

successful, reliable, strong, and intelligent were included. Finally, for the Status

dimension; sophisticated, elitist, up-market, and chic were chosen (taken from Davies

et al., 2004; Aaker, Vohs and Mogilner, 2010). A five point Likert scale was used to

assess each item in this part of the survey with each point labeled from strongly agree

to strongly disagree.

For the personification variants respondents were asked “If the organisation you work

for came to life as a person, what would his/her personality be like’ and then to rate

the image items. For the direct measurement versions respondents were not given any

preamble. They were asked directly to rate statements that include the same employer

brand image items; for instance for the “friendly” item they were asked to rate the

following statement: “The organization I work for is a friendly organization” with a

response scale from 1 to 5 with points 1, 3 and 5 labeled strongly disagree, neither

agree nor disagree and strongly agree, accordingly.

Different response scales were used throughout the questionnaire to reduce any

Common methods variance (CMV) issues (Harman, 1967). The scaling approach in

the online survey was also varied between questions (sometimes a tick box,

sometimes a sliding scale).

90


First, the scales used to assess the three dimensions were checked for reliability with

Cronbach Alpha. The Cronbach Alpha’s for each of the two groups were acceptable

(Table 2), (Nunnally, 1978; Peterson, 1994).

Group Warmth Competence Status Personification .89 .87 .76 Direct .88 .89 .73 Table 2. Cronbach’s Alpha Values of Dimensions by Groups Then the mean scores for each dimension were calculated (Table 3), and independent

samples t-tests for each dimension carried out to understand whether the two

approaches lead to statistically different results.

Group Warmth Competence Status Personification 3.58 3.85 2.71 Direct 3.48 3.85 2.90 Table 3. Mean Scores of Dimensions by Groups Group Dimension Mean F value P value Personification Warmth 3.58 1.27 0.26 Direct Q Warmth 3.48 Personification Competence 3.85 1.85 0.75 Direct Q Competence 3.80 Personification Status 2.90 0.44 0.51 Direct Q Status 2.71 Table 4. Means and Levene’s Test for Equality of Variance Values for Each Group

91

The t-test results show that when measuring the Warmth dimension using a direct

approach based questionnaire (3.48± 0.81) leads to no statistically significantly

different mean values compared to the personification approach (3.58± 0.80), t(219) =

-.93, p = . 35.

Similarly, when measuring the Competence dimension using a direct approach based

questionnaire (3.80± 0.85) leads to no statistically significantly different mean values

compared to a personification approach (3.85± 0.75), t(219) = -.43, p = . 67.

Likewise, when measuring Status dimension using a direct approach based

questionnaire (2.70± 0.73) leads to no statistically significantly different mean values

compared to a personification approach used questionnaire (2.90± 0.76), t(219) = -

1.83, p = .07. The standard deviations of each measure were also similar (Table 4).



mean score of the items measuring intellectual engagement (INT ENG), social

engagement (SOC ENG), affective engagement (AFF ENG) and overall engagement

(OVR ENG), and satisfaction (SAT) were used as dependent variables. (Each

measure was valid with alphas or inter-item correlations above 0.8). The predictive

ability of the two approaches (P for Personification, D for Direct) is compared in

Table (5) using the adjusted R2 for each.

92

DV Adjusted

R2 Warmth (P)

Adjusted R2 Warmth (D)

Adjusted R2

Competence (P)

Adjusted R2

Competence (D)

Adjusted R2

Warmth, Competence & Status (P)

Adjusted R2

Warmth, Competence & Status (D)

SAT 0.43 0.64 0.42 0.45 0.48 0.65

INT ENG 0.11 0.06 0.22 0.03 0.21 0.05

SOC ENG 0.36 0.22 0.40 0.09 0.43 0.21

AFF ENG 0.37 0.28 0.34 0.25 0.41 0.31

OVR ENG 0.36 0.26 0.42 0.16 0.45 0.26

Table 5. Adjusted R-Square Values of Dependent Variables by Context As can be seen from Table 5 in some cases the personification approach gave a higher

prediction of variance (measured by R2), in others it was the direct approach. Overall,

direct questioning explained greater variance when the outcome of the equation was

satisfaction. (i.e. in predicting satisfaction), regardless of the brand image dimension

used or if they were used in combination, not supporting H1 . However, when

predicting employee engagement separately or overall, the personification approach

explained more variance than direct questioning for warmth, competence and status

and for the combination of these dimensions (13 instances out of 16), supporting H1.

Warmth and competence predicted different levels of each DV. In 4 instances warmth

predicted more variance, than competence, in one less. There is little support for H2.

The next analysis also involves each of the three dimensions separately, and

collectively in predicting the dependent variables such as employee satisfaction, and

employee engagement. A Chow test can be used to compare the regression residuals

at this level, by running two separate regressions on the two groups (personification

and direct questionnaire respondents) for exactly the same regression equation, as

well as the full sample regression.

93

Dependent Variable

Dimension Method SSR Chow F Statistic

Significant or not

Satisfaction Warmth Personification 117.64 2.72 Not Significant Satisfaction Warmth Direct 86.66

Satisfaction Competence Personification 118.98 0.75 Not Significant Satisfaction Competence Direct 132.80

Satisfaction Status Personification 200.93 1.10 Not Significant Satisfaction Status Direct 219.22

Satisfaction All Personification 104.22 3.87 Significant Satisfaction All Direct 82.75 Table 6 A. Chow Test for Each Dimension and Method when Predicting Satisfaction Dependent Variable


Significant or not

Intellectual Engagement

Warmth Personification 127.00 0.50 Not Significant


Warmth Direct 100.86


Competence Personification 111.44 4.11 Significant


Competence Direct 104.24


Status Personification 141.97 0.05 Not Significant


Status Direct 106.40


All Personification 111.30 4.92 Significant


All Direct 100.67

Table 6 B. Chow Test for Each Dimension and Method when Predicting Intellectual Engagement

94

Dependent Variable Dimension Method SSR Chow F Statistic

Significant or not

Social Engagement


Social Engagement


Social Engagement

Competence Personification 93.95 3.70 Significant

Social Engagement


Social Engagement

Status Personification 151.05 1.13 Not Significant

Social Engagement

Status Direct 228.34

Social Engagement

All Personification 87.40 4.84 Significant

Social Engagement

All Direct 184.19

Table 6 C. Chow Test for Each Dimension and Method when Predicting Social Engagement Dependent Variable


Significant or not

Affective Engagement Warmth Personification 135.63

1.03 Not Significant Affective

Engagement Warmth Direct 122.30

Affective Engagement Competence Personification 141.57


Engagement Competence Direct 128.14

Affective Engagement Status Personification 198.88


Engagement Status Direct 146.34

Affective Engagement All Personification 124.32


Engagement All Direct 114.42

Table 6 D. Chow Test for Each Dimension and Method when Predicting Affective Engagement

95

Dependent Variable


Significant or not

Overall Engagement Warmth Personification 81.67

0.51 Not Significant Overall

Engagement Warmth Direct 86.80

Overall Engagement Competence Personification 73.99

4.01 Significant Overall Engagement Competence Direct 98.66

Overall Engagement Status Personification 121.90

0.22 Not Significant Overall

Engagement Status Direct 108.82

Overall Engagement All Personification 69.34

4.46 Significant Overall Engagement All Direct 85.53

Table 6 E. Chow Test for Each Dimension and Method when Predicting Overall Engagement In several cases, the Chow F statistics are greater than the critical F-value, leading to

the conclusion that the regression lines of the two data sets in terms of personification

and direct approach used questionnaires are different. There are 20 comparisons in

total (Table 6A to 6E), there are 7 cases that result in a greater Chow Statistic, and out

of these 7 cases, 6 of them have greater adjusted R2 values (Table 5) for the

personification approach. Therefore the Chow test results’ findings tend to support

H1.

Next, the Fisher test was used to evaluate whether the correlation between a

dimension and a dependent variable is better than that when using a personified

measure (P) compared with using a direct measure (D). First Fisher’s r to z

transformation was applied to the correlations. The effect of this transformation is to

96

make the sampling distribution of the transformed coefficient nearly normally

distributed (Harrison and Kenny, 1987). The coefficients were compared using

Fisher’s r to z transformation; but the results did not provide a consistent pattern for

the tested dependent variables in terms of satisfaction (SAT), and the separately

evaluated engagement measurement constructs (namely social (SOC ENG),

intellectual (INT ENG), and affective (AFF ENG)), and overall engagement (OVR

ENG) measurement. The critical value of Z is 1.96, when p < .05. There is

significance on the Warmth dimension when predicting satisfaction, whereas the

results favour the Competence dimension when predicting intellectual, social, and

overall engagement variables (Tables 7 A-E).

DV Dimension Method Pearson

R N Fisher’s z


Significance

SAT Warmth P 0.66** 113 2.24 0.01 Significant SAT Warmth D 0.80** 108 SAT Competence P 0.65** 113 0.26 0.40 Not Significant SAT Competence D 0.67** 108 SAT Status P 0.16* 113 1.09 0.13 Not Significant SAT Status D 0.30** 108

**. Correlation is significant at the 0.01 level (1-tailed). *. Correlation is significant at the 0.05 level (1-tailed). Table 7 A. Fisher’s R to Z transformation When Predicting Satisfaction DV Dimension Method Pearson

R N Fisher’s z


Significance

INT ENG Warmth P 0.35** 113 0.73 0.23 Not Significant INT ENG Warmth D 0.26** 108

INT ENG Competence P 0.48** 113 2.35 0.01 Significant INT ENG Competence D 0.20* 108 INT ENG Status P 0.13 113 0.07 0.47 Not

Significant INT ENG Status D 0.14 108 **. Correlation is significant at the 0.01 level (1-tailed). *. Correlation is significant at the 0.05 level (1-tailed). Table 7 B. Fisher’s R to Z transformation When Predicting Intellectual Engagement

97


R N Fisher’s z


Significance

SOC ENG Warmth P 0.60** 113 1.34 0.09 Not Significant SOC ENG Warmth D 0.47** 108 SOC ENG Competence P 0.63** 113 3 0.001 Significant SOC ENG Competence D 0.32** 108 SOC ENG Status P 0.18* 113 0.15 0.44 Not Significant SOC ENG Status D 0.20* 108

**. Correlation is significant at the 0.01 level (1-tailed). *. Correlation is significant at the 0.05 level (1-tailed). Table 7 C. Fisher’s R to Z transformation When Predicting Social Engagement DV Dimension Method Pearson

R N Fisher’s z


Significance

AFF ENG Warmth P 0.61** 113 0.87 0.19 Not Significant AFF ENG Warmth D 0.53** 108

AFF ENG Competence P 0.58** 113 0.83 0.20 Not Significant AFF ENG Competence D 0.50** 108

AFF ENG Status P 0.27** 113 0.90 0.18 Not Significant AFF ENG Status D 0.38** 108

**. Correlation is significant at the 0.01 level (1-tailed). Table 7 D. Fisher’s R to Z transformation When Predicting Affective Engagement DV Dimension Method Pearson

R N Fisher’s z

transformation

P value

Significance

OVR ENG Warmth P 0.60** 113 0.86 0.19 Not Significant OVR ENG Warmth D 0.52** 108

OVR ENG Competence P 0.65** 113 2.40 0.01 Significant OVR ENG Competence D 0.42** 108 OVR ENG Status P 0.23** 113 0.55 0.29 Not

Significant OVR ENG Status D 0.30** 108 **. Correlation is significant at the 0.01 level (1-tailed). Table 7 E. Fisher’s R to Z transformation When Predicting Overall Engagement

Overall there are 4 instances when the personified measure has a significantly higher

correlation with a DV than for a direct measure at p< .05. More specifically, when

predicting intellectual engagement using the competence dimension of brand image,

personification provides a significantly higher correlation than direct measurement.

98

When predicting social engagement using the competence dimension, personification

provides a significantly higher correlation than direct measurement. When predicting

overall engagement using the competence dimension, personification measurement

provides a significantly higher correlation than direct measurement. The findings are

comparable with those comparing the R2 data and tend to support H1 but the pattern is

not totally consistent. Warmth measures showed higher correlations with DV’s on 3

occasions but competence measures higher on 1, again not consistent support for H2.

The next analysis aimed to compare the two types of measure using Structural

Equation Modeling (AMOS 22). First a model for each dimension was tested using

the combined data (personified and direct) and trimmed to exclude any poor fitting

items. As the loadings of both “Leading” and “Elitist” items were less than .50, they

were removed from their respective models (Competence and Status). Subsequently

the error terms of the two measurement items “ethical” and “socially responsible” on

the warmth dimension, and “reliable” and “successful” on the competence dimension

were co-varied to achieve a reasonable fit.

The models for each dimension (warmth, competence and status) resulted in

following fit statistics:

For Warmth, CMIN/DF=1.337, GFI= .969, AGFI=. 919, CFI=. 994, NFI = .977,

Hoelter 271 and 329 for .05 and .01 indices respectively and finally RMSEA=. 039.

99

Figure 1. The final Model for Warmth Dimension and Its Standardized Regression Weights

For Competence, CMIN/DF=2.940, GFI= .987, AGFI=. 870, CFI=. 992, NFI = .988,

Hoelter 225 and 345 for .05 and .01 indices respectively and finally RMSEA=. 064.

Figure 2. The final Model for Competence Dimension and Its Standardized Regression Weights

100

For Status, CMIN/DF=1.411,GFI=. 992, AGFI=. 950, CFI=. 997, NFI= .989, Hoelter

in 466 and 716 for .05 and .01 indices respectively and finally RMSEA=. 043.

Figure 3. The final Model for Status Dimension and Its Standardized Regression Weights

Views differ as to the most appropriate measures of fit in SEM and of their acceptable

values. Bentler and Bonnet (1980) recommend values greater than .90 indicating a

good fit for most commonly used measures. However some researchers suggest that

the cut-off criteria should be NFI ≥.95 (Hu and Bentler, 1999). A major drawback to

this index is that it is sensitive to sample size, underestimating fit for samples less

than 200 (Mulaik et al, 1989; Bentler, 1990), and therefore NFI is not recommended

to be used on its own (Kline, 2005; cited in Hooper, Coughlan, and Mullen, 2008).

The NFI for these models are .977 for Warmth, .988 for Competence, and .989 for

Status dimensions.

101

On the other hand, CFI is one of the most popularly reported fit indices due to its

being one of the measures least effected by sample size (Fan, Thompson, and Wang

1999); since it is a revised form of the NFI which takes into account sample size

(Byrne, 1998) that works well even when sample size is relatively small (Tabachnick

and Fidell, 2007; cited in Hooper et al., 2008). The CFI figures are .994 for Warmth,

.992 for Competence, and .997 for Status models and acceptable (Hooper et al.,

2008).

To assess construct validity, Confirmatory Factor Analysis was used (Jöreskog, 1967)

and the Average Variance Extracted (AVE) and Composite Reliability (CR) used to

assess the convergent validity of the measurement models (Fornell and Larcker, 1981,

a). The average variance extracted score is recommended to be greater than .50

(Fornell and Larcker, 1981,b). For the composite reliability statistic, scores of above

.70 are recommended (Carmines and Zeller, 1979).

The results were found to be good for all three dimensions and both measurement

approaches (see Table 8) with the personification approach showing slightly better

figures than those for the direct approach on the warmth and status dimensions,

whereas the direct approach for the competence dimension shows better results.


Warmth Personification 0.66 0.92 Warmth Direct 0.61 0.90 Competence Personification 0.62 0.87 Competence Direct 0.65 0.88 Status Direct 0.60 0.82 Status Personification 0.66 0.85 Table 8. AVE and CR Results According to Dimension and Measurement Approach

102

The measurement approaches were then compared using multi group analysis in

SEM. The approach is normally used to compare models for different groups of

respondents (e.g. male versus female), but here to compare direct and personified

responses. The multi group analysis does not show any statistically significant

difference between the measurement approaches when each dimension is modeled

separately (Table 9) implying that the two measurement approaches are very similar.

Dimension used in the Model

DF CMIN P

Warmth 5 4.554 .473 Competence 3 6.065 .108 Status 2 2.822 .244 Table 9.Results of Multi Group Analysis for Each Model (factor loadings constrained) The results of study 1 suggest that any advantage from using a personified measure

might be small, smaller than the extensive literature on brand personality and

projective techniques might imply. However the contexts when personification

provided better predictive validity often involved warmth rather than one of the other

two dimensions of personality/image that were considered and prediction was more

significant when using warmth than competence. Hypothesis 2 derived from work on

the stereotype content model, predicted an advantage for competence. This suggests

that any differences between the two approaches might be dimension specific.

However in Study 1 items from the various dimensions were presented to respondents

together. This raises the possibility of a halo effect (Slaughter, Zickar, Highhouse, and

Mohr, 2004) masking any advantage for competence and any dimensional effect on

103

the relative benefit of a personified approach. Consequently, in Study 2 both

Hypotheses are tested again but in the context of respondents using just one

dimension of personality/image.

Because the status dimension did not yield any results of interest, in the next study the

focus is only on the warmth and competence dimensions.

Study 2

Methodology

As in Study 1, respondents, as employees, were asked to evaluate the organization

that they were working for by using either a personification approach or direct

approach based questionnaire. However, in this study the sample was split into 4

groups and respondents randomly assigned to one of these groups. The study adopted

a 2 (Personification (n=222, 50.5%) vs. Direct (n=218, 49.5%)) x 2 (Warmth (n=223,

50.7%) vs. Competence (n=217, 49.3%)) factorial, between-subjects design, with a

sample size of 440 respondents. Therefore, the main difference in this study from

study 1 is that one respondent could only rate their employer only for either

competence or warmth, whereas in the first study all respondents were asked to rate

all dimensions together.

The survey started with three filter questions to ensure the respondents were residing

in the UK, were not currently self-employed, and worked for one employer more than

25 hours per week.

104

After the filter questions, demographic questions were included; such as the age of the

respondents, their gender, their education level, the number of years that they have

been working (in the workforce), and the number of years they had been working for

their current employer (company). Then the respondents were asked to choose the

employer for whom they worked most hours, if they had more than one job.

The sample details are shown in Table 10. Questionnaire Type and Dimension




Direct Approach with Warmth


Direct Approach with Competence


Personification Approach with Warmth


Personification Approach with Competence

Male 79 71.8

Female 31 28.2 Total 110 100

Table 10. Questionnaire Type and Dimension Distribution According to Gender

The order of the questions was kept similar to that in Study 1. Hence, following the

demographics, satisfaction, expertise and engagement questions were asked.

The brand image item questions followed for either warmth or competence. The

number of items was increased from those used in Study 1 to 15 items for each

dimension to ensure that the potential benefit to personification of having a larger

pool of items could be evaluated. For the Warmth dimension the employer brand

105

image items were selected as friendly, honest, sincere, straightforward, pleasant,

trustworthy, reassuring, supportive, agreeable, concerned, socially responsible,

ethical, cheerful, warm, and open (from; Aaker et al., 2010; Davies et al., 2004). For

the Competence dimension; reliable, secure, hardworking, ambitious, achievement

oriented, leading, technical, corporate, effective, efficient, competent, successful,

strong, confident, and intelligent were selected (from Aaker, 1997; Davies et al.,

2004).


The scales used to assess the two dimensions were checked for reliability with

Cronbach Alpha. The Cronbach Alphas for each of the two groups were acceptable

(Table 11), (Nunnally, 1978; Peterson, 1994).

Group Warmth Competence Personification .98 .96 Direct .98 .94 Table 11. Cronbach’s Alpha Values of Dimensions by Groups Then in order to understand whether the data have similar variances between

measurement types (Bryk and Raudenbush, 1988), the data were tested for the

homogeneity of variances assumption (HOV). Levene’s Test results including f-

values and p-values can be seen in Table 12 below.

Method Dimension Mean F value P value Personification Warmth 3.52 0.06 0.81 Direct Q Warmth 3.48 Personification Competence 3.74 1.57 0.21 Direct Q Competence 3.55 Table 12. Means and Levene’s Test for Equality of Variance Values for Each Group

106

The Levine’s test results show that the variances are equal between the direct and

personified approaches.

Then an independent samples T-test was carried on with the means of each dimension

used to see if either method results in a statistically significant difference. The t-test

result shows that measuring Warmth using a direct approach based questionnaire

(3.48± 0.96) leads to no statistically significantly mean values compared to a

personification approach (3.52± 0.94), t (221) = -.359, p =. 809. Similarly, measuring

Competence using a direct approach based questionnaire (3.55± 0.70) leads no

statistically significantly mean values compared to a personification approach (3.73±

0.84), t (215) = -1.1811, p =. 211.



mean score of the items measuring engagement, satisfaction and reputation were used

as dependent variables. (Each DV measure was valid with alphas or inter-item

correlations above 0.8). The predictive ability of the two approaches is compared in

Table (13) using the adjusted R2 for each.

107

Dependent Variables

Adjusted R2 Warmth (P)

Adjusted R2 Warmth (D)

Adjusted R2 Competence (P)

Adjusted R2

Competence (D)

Satisfaction .76 .74 .47 .48 Intellectual Engagement

.36 .13 .38 .20

Social Engagement

.35 .38 .35 .32

Affective Engagement

.64 .63 .50 .46

Overall Engagement

.58 .53 .55 .47

Table 13. Adjusted R-Square Values of Dependent Variables by Context

Next a Chow Test was carried out separately for the warmth and competence

dimensions, by running separate linear regressions for personification and direct

approaches, on each dependent variable: employee satisfaction and employee

engagement.

Dependent Variable


Significant or not

Satisfaction Warmth Personification 63.07 0.43 Not Significant

Satisfaction Warmth Direct 73.55

Satisfaction Competence Personification 113.28 3.01 Not Significant

Satisfaction Competence Direct 139.47

Table 14 A. Chow Test for Each Dimension and Method when Predicting Satisfaction

108

Dependent Variable


Significant or not


Warmth Personification 109.80 5.05 Significant




Competence Personification 90.70 0.66 Not Significant



Table 14 B. Chow Test for Each Dimension and Method when Predicting Intellectual Engagement Dependent Variable


Significant or not

Social Engagement


Social Engagement


Social Engagement


Social Engagement


Table 14 C. Chow Test for Each Dimension and Method when Predicting Social Engagement

109

Dependent Variable


Significant or not









Table 14 D. Chow Test for Each Dimension and Method when Predicting Affective Engagement Dependent Variable


Significant or not

Overall Engagement


Overall Engagement

Warmth Direct 71.88

Overall Engagement


Overall Engagement


Table 14 E. Chow Test for Each Dimension and Method when Predicting Overall Engagement Although the use of a personified approach yielded higher R2 figures (Table 13), out

of 10 Chow test Comparisons there is only one comparison that resulted in a

significant difference (a greater value than the critical F-value) that for the Warmth

dimension when predicting intellectual engagement (Table 14 B).

110

As it can be seen from Tables 13; in some cases personification gave the higher

prediction of variance (measured by R2) in others it was the direct approach. To be

more specific, when predicting employee engagement separately or overall (excluding

social engagement with warmth), the personification approach resulted in explaining a

greater variance than direct questioning, for both warmth and competence dimensions,

supporting H1. Differing from the study 1 results, satisfaction was also predicted

better when using personification for both warmth and competence dimensions,

supporting H1.

Irrespective of which method was used, the competence dimension was expected to

explain more variance than the warmth dimension in dependent variables such as

satisfaction and the engagement constructs. The opposite holds in 4 of the 5 cases

(Table 13) and therefore H2 is not supported.

Next the Fisher test was used to evaluate whether the correlation between a dimension

and a dependent variable is better when using a personified measure compared with

using a direct measure. First Fisher’s r to z transformation was applied to the

correlations. The effect of this transformation is to make the sampling distribution of

the transformed coefficient nearly normally distributed’ (Kenny, 1987). The Fisher’s r

to z transformation results did not provide a consistent pattern for the tested

dependent variables in terms of satisfaction, or the separately evaluated engagement

measurement constructs (social, intellectual, and affective. There is significance on

the warmth dimension when predicting intellectual engagement, as well as the

111

competence dimension on the same DV (Table 15). (Tables for other dependent

variables can be seen in Appendix 2 to the thesis.)


R N Fisher’s z


Significance


Warmth P 0.60** 112 2.16 0.01 Significant


Warmth D 0.38** 111


Competence P 0.62** 110 1.65 0.05 Not Significant


Competence D 0.46** 107

**. Correlation is significant at the 0.01 level (1-tailed). Table 15. Fisher’s R to Z transformation When Predicting Intellectual Engagement

Overall there is 1 instance where the personified measure has a significantly higher

correlation with a DV than for a direct measure at p< .05. Specifically, when

predicting intellectual engagement, with the warmth dimension, a personified measure

shows a significantly higher correlation, than when using a direct measure. However,

out of 12 comparisons only 1 supports H1, close to a result due to random chance of 1

in 20 when using a 0.05 significance test.

Next Model Fit Analysis was carried out for the warmth and competence dimensions

using Structural Equation Modeling (AMOS 22). First, the warmth dimension was

investigated. The loadings of ‘Pleasant’, ‘Supportive’,’ Agreeable’, ‘Ethical’, and

‘Corporate’ were less than .50, and they were removed from the model. Co-variances

were not added. The final model resulted with a CMIN/DF=2.261, GFI=. 935,

AGFI=. 897, CFI=. 983, NFI= .969, Hoelter =140 and 161 for .05 and .01 indices

respectively and finally RMSEA=. 075 (Figure 4). Overall the model fit was

satisfactory.

112

Figure 4. Warmth Dimension Then the competence dimension was investigated and the model trimmed. The

loadings of both ‘Technical’ and ‘Corporate’ brand image items were less than .50,

113

and they were removed from the model. Subsequently in order to achieve a better fit

for the model, several covariances were added. Consequently, the final model had a

CMIN/DF=1.964, GFI=. 920, AGFI=. 877, CFI=. 968, NFI= .959, Hoelter =136 and

152 for .05 and .01 indices respectively and finally RMSEA=. 067 (Figure 5). The fit

indices are overall acceptable.

Figure 5. Competence Dimension with Covariances

114

In order to assess construct validity, Confirmatory Factor Analysis was used

(Jöreskog, 1967). The Average Variance Extracted (AVE) and Composite Reliability

(CR) were used to assess the convergent validity of the measurement model (Fornell

and Larcker, 1981). The results are found to be good for both of the dimensions using

both measurement approaches, being above 0.5 (Fornell and Larcker, 1981, b), (see

Table 16). The personification approach shows better figures than those for the direct

approach for the competence dimension, and vice versa for the direct approach for the

warmth dimension.


Warmth Personification 0.74 0.97 Warmth Direct 0.80 0.98 Competence Personification 0.70 0.97 Competence Direct 0.60 0.95 Table 16. AVE and CR Results According to Dimension and Measurement Approach The data were then compared using multigroup analysis in SEM. The multi group

analysis, which involves a chi-square difference test via measurement weights, does

not show any statistically significant difference between the measurement approaches

for competence. However, for the warmth dimension there is a statistically different

result when comparing personification and direct approaches (Table 17).

Dimension used in the Model

DF CMIN P

Warmth 9 18.986 .025 Competence 12 7.054 .854 Table 17.Results of Multi Group Analysis for Each Model (factor loadings constrained)

115

A further Multi group analysis for warmth with each item loading constrained one at a

time, suggested that “ethical” and “socially responsible” and “reassuring” are where

respondents provide the largest difference in response pattern (Table 18).

This suggests that it might be easier for respondents to assess these (similar) items

under personification, however the mean scores for one are lower under direct

questioning and the standard deviations again do not follow any pattern.

Item Mean

Score (Direct)

Mean Score (Personification)

Standard Deviation (Direct)

Standard Deviation (Personification)

Ethical 3.32 3.46 1.168 1.056 Socially Responsible

3.49 3.44 1.061 1.080

Reassuring 3.44 3.51 1.093 1.013 Table 18. Mean Scores and Standard Deviations for the three items


For the purposes of this study, two different approaches to employer brand image

measurement have been investigated to understand possible differences. By analysing

the responses from employees to their employer brand’s image, the results do not

diverge from each other between each of the methods used. This research provides

management with a valid and reliable data analysis to show, regardless of industry,

there is almost no difference in the usage of either method at least in this context.

In terms of promoting employer brand image, this research clearly shows that there is

a strong correlation between outcomes such as satisfaction and intellectual

engagement of the employees and their perception of their employers brand image for

116

especially the Warmth dimension of brand image. In other words the higher

employees think their employer brand is ‘warm’, the more they are intellectually

engaged with their work and the more satisfied they are. The R2 figures when

predicting satisfaction are quite high. Most companies undertake annual employee

satisfaction surveys but this research suggests they might wish to include questions

about brand image as well as a way to understand what is driving employee

satisfaction. .

The findings of this research have clear implications for market research companies as

well. Market research companies often use projective and personified approaches in

asking questions in the same way as academic researchers might choose to do. The

findings here suggest that there is little advantage in doing so. Worse, many research

companies have their own measures of brand personality (rather than brand image) as

this helps them market their services to practitioners. How valid such an approach

really is particularly whether it has advantages over direct questioning is called into

question here. The idea of an employer brand has become popular in recent years

(Backhaus and Tikoo, 2004; Davies, 2008), and this work questions whether there is

any need to use personification in measuring it.

Conclusions and Further Work

The main aim for this work was to compare the use of a personified and direct

approach to measuring brand imagery and so test and replicate the conclusion from

Mete (2017a) in the context of product and corporate branding, that there is little

difference between the two approaches and no systematic advantage for the

117

personified option. The context this time was of employer branding. Here there were

relatively more examples of the personified approach being more useful, particularly

in predicting employee engagement. However the advantage was not always present

and indeed, as in Mete (2017a), the direct approach sometimes proved superior.

More items were included in Study 2 than in Study 1 and in Mete (2017a) but that

seemed not to change the overall picture. Indeed perhaps the most telling findings

were that the Cronbach alphas were all acceptable for both measure types and, the

means and standard deviations for both measure types were the same. The difference

between the two approaches was only significant once in five comparisons using

multi-group analysis (at p=.025).

This study has also tested whether any conclusions might differ depending upon the

dimension of brand imagery being considered. In Study 1 three dimensions were

considered and in Study 2 two. Differences between the dimensions of warmth and

competence might be expected from the stereotype content model (Fiske, Cuddy and

Glick, 2007). In both studies there was some evidence that warmth is the more useful

dimension in explaining some DV’s. However there was no compelling evidence to

support the idea that personification works better in one context or the other.

Wojciszke and Abele (2008), argue that competence judgments involve greater

processing than warmth judgements, implying that competence evaluations would be

more useful in predicting and explaining DV’s such as engagement. If anything the

opposite was true.

118

Certainly some differences were significant both in terms of image dimension and

question type, particularly when trying to predict engagement, rather than satisfaction.

Further work is needed to clarify why this might be so.

As to the main research question, there was no consistent pattern found to support H1,

leading to the conclusion that there is no systematic benefit for the use of

personification. It is impossible to prove a negative, that there is no advantage in

using personification, but as personification attracts criticism (Mete 2017a) some will

interpret these results as suggesting the use of only direct measures of affective brand

imagery.

119

Chapter 5.1: Connecting Sub-Chapter 2: Introduction to Task Difficulty Investigating the possible effect of task difficulty emerged during work for the second

paper as a possible explanation for the differences that did emerge between direct and

personified measures in the first paper. (A personified measure should be easier for

respondents to use when assessing a complex construct such as brand image). It was

investigated in the first of the studies undertaken for the second paper with only two

questions. Respondents were asked how easy it was to complete the survey; with a

Likert scale ranging from 1 (Very Easy) to 7 (very Difficult). The second question

was to rate the statement “I found it difficult to answer most of the questions”; with a

Likert scale ranging from 1 (Strongly agree) to 7 (Strongly disagree) and 4 was

specified as “Neither agree, nor disagree”.

These two questions were used to test whether there is a task difficulty effect when

brand image is measured. There was a slight difference between the means of brand

image scores for both of the measurement approaches, depending on whether task

difficulty was rated high or low (using a mean split). However, a scale that consists of

only two items invented by the researcher is not an appropriate measure to use in

order to understand a potential statistically significant effect and no results were

reported.

Therefore, in order to be able to test the idea of task difficulty and its potential effect

on brand imagery, a literature review on task difficulty scales was made which lead

the researcher to NASA’s TLX measure in the second study. The next Chapter reports

the findings from using the TLX on both the data collected for the second study

120

reported in the previous Chapter and on that collected for a final study which

considers the possible effects of task difficulty in two contexts.

121

Chapter 6

Measuring Brand Image and the Role of Task Difficulty

122

Measuring Brand Image and the Role of Task Difficulty Abstract Two studies are conducted to understand the role of task difficulty in market research and specifically in the context of brand image measurement. Task difficulty was found to be influential in brand image evaluations in previous research and its influence is more formally considered here. In order to understand the influence of task difficulty, several variables such as the age and education level of the respondents are considered. In study one, an online survey was made with employees as respondents (N=440) to

evaluate their companies’ brand image using a 2 (Personification vs. Direct) x 2

(Warmth vs. Competence) factorial, between-subjects design.

In study two, the context was changed from employer branding to considering one

brand (Tesco) used in two different contexts, as a corporate/organizational brand and

as a private label/product brand. The respondents were given either warmth or

competence dimension of brand image items to consider,

An adapted version of the TLX measure of task difficulty scale (Hart and Staveland,

1988) was used in both surveys.

Task difficulty did not vary as expected by image dimension or by whether a

projective or direct method was used to measure image. It did not influence the

relationship between image and a number of dependent variables, but it did contribute

to an explanation of several variables such as intellectual engagement.

Task difficulty was however found to vary with respondent age and education, but not

in ways implied by existing literature.

123

Introduction

Answering surveys questions can require significant cognitive effort and ability and

cause difficulties to many respondents. This difficulty can lead respondents to adopt

strategies to reduce the ‘task difficulty’ of answering a questionnaire. Task difficulty

has been widely researched in the educational and ergonomics literatures. However, it

has been largely neglected in the marketing and market research literatures.

This article aims to introduce task difficulty and investigate its effect in the context of

research on brand imagery measurement.

Literature Review

Firstly, task difficulty will be explained in a literature review. Then, cognitive load

theory will be introduced and the notion of satisficing will be explained, specifically

that high levels of cognitive demand result in the behaviour of satisficing when

responding to survey questions. The major factors that affect task difficulty, such as

age, gender, and education, will be explained.

Task Difficulty: an Education Perspective

Task difficulty is considered as a decisive factor determining task performance in

education as there is more information processing involved in difficult tasks (Kim,

2006). Task difficulty can be described as “ a subjective perception assessed by task

doers” (Li and Belkin, 2008, cited in Liu, Liu, Yuan, and Belkin, 2011). Additionally,

124

task difficulty can also be defined as task performers’ perception of the complexity of

a task (Kim, 2006).

The influence of task difficulty on respondents when carrying out academic research

has a long history (Aula, Khan, and Guan, 2010; Gwizdka and Spence, 2006; Kim,

2006; Liu, Gwizdka, Liu, Belkin, 2010; Van Der Vaart, Van Der Zouwen, and

Dijkstra, 1995). Task difficulty has been found to be a significant factor influencing

respondents’ performance, such as when requiring more information (Aula, Khan, and

Guan, 2010; Kim, 2006; Liu et al., 2010) and taking longer time if the task is to be

found difficult (Aula et al., 2010; Liu et al., 2010).

Task difficulty vs Task complexity

Some researchers have used the term “ task complexity” interchangeably with the

term “task difficulty” (e.g. Bell and Ruthven, 2004; Kim, 2006). Others, however,

choose to separate the two. For instance, Byström and Järvelin (1995) found that the

more the complexity of a task increased, the more people depended on experts to

provide information. Task complexity can be both objective and subjective (Li and

Belkin, 2008), with subjective task complexity assessed by task performers, and

objective task complexity defined by the number of activities involved in a “work

task” (Ingwersen and Järvelin, 2005, cited in Liu et al., 2011). Some researchers

choose to focus on task difficulty (Cole, Bagic, Kass, and Schneider, 2010; Kim,

2006; Li and Belkin, 2008). There are several studies in which task difficulty was

measured using respondents’ self-reported perceptions of how difficult a task is via

questionnaires.

125

Factors that Affect Task Difficulty

Some researchers emphasise the importance of “prior knowledge” in the issue of task

difficulty (e.g. Vakkari, 1999). Liu et al. (2011) argue that there are other significant

factors, since the pre-task background (e.g. previous experience or knowledge of the

task) could affect task difficulty, although, their research concludes that prior

knowledge has no significant effect on how difficult respondents perceive a task to be.

Researchers have highlighted the time taken on tasks as a way to assess task difficulty

(e.g. Goldhammer et al., 2014) and tried to explain the effect using dual processing

theory, which distinguishes between automatic and controlled mental processes (e.g.

Schneider and Chein, 2003; Schneider and Shiffrin, 1977; Goldhammer et al., 2014).

Automatic processes can be identified as fast and proceduralised, which require little

effort and do not require active control or attention, while controlled mental processes

can be classified as slow, and requiring attention and active control (Ackerman, 1987;

Goldhammer et al., 2014). Individuals are assumed to differ in terms of processing the

same information for a particular task (Carlson, Sullivan, and Schneider, 1989).

Automatic and controlled processes may interact in both reading and problem solving

domains (Goldhammer et al., 2014).

The amount of time taken to read the same material differs for each individual, since

reading a text requires a number of cognitive component processes and related

abilities such as identifying letters and words, establishing coherence between words

(Kintsch, 1998). Moreover, “the speed of semantic integration as well as local

coherence processes is positively related to comprehension (e.g. Naumann, Richter,

Flander, Christmann, and Groeben, 2007: Naumann, Richter, Christmann, and

Groeben, 2008; Richter, Isberner, Naumann, and Neeb , 2013; Goldhammer et al.,

126

2014). There are several longitudinal studies showing that reading performance

continues to improve with education level (e.g. Landerl and Wimmer, 2008).

Additionally, if the individual finds the text difficult, re-reading or engaging in self-

explanations can help comprehension (e.g. Best, Rowe, Ozuru, and McNamara, 2005;

McKeown, Beck and Blake, 2009). Furthermore, effort in strategic reading for skilled

readers positively predicts comprehension (e.g. Sullivan, Gnesdilow and

Puntambekar, 2011); however, this leads to a longer time being spent on the task

(Rosander and Eriksson, 2012). There is also an inclination for respondents to switch

to random guessing, when the task takes a long time (Hornke, 2005).

However, it is also important to note the motivation behind taking a longer time for a

task. There are instances where respondents take more time but when they also think

harder, which results in confusion over the relationship between depth of processing

and time taken (Rosander and Eriksson, 2012). Consequently, some researchers

consider time taken for a task indicates engagement, which includes reading the text

attentively, concentrating on the meaning and sustained cognitive effort (e.g. Guthrie

et al., 2004).

Another issue related to task difficulty is that, when asked a question or for an

opinion, people want to answer “correctly”, peoples’ inclination to conform, which

could be connected to informational conformity (Deutsch and Gerard, 1955), in other

words a desire to answer correctly as a way to protect self-esteem (Cialdini and

Goldstein, 2004; cited in Rosander and Eriksson, 2012). There have been a number of

studies to understand the relationship between task difficulty and conformity (e.g.

Baron, Vandello, and Brunsman, 1996; Gergen and Bauer, 1967; Morris, Miller, and

Spangenberg, 1977).

127

Baron, Vandello, and Brunsman (1996), found that when task difficulty is low,

motivation for accuracy reduces the social impact of confederates, however, when

task difficulty is high, the reverse is true; individuals conform more to an inaccurate

confederate norm when motivations for accuracy is high. They also found women

tend to conform more. Gergen and Bauer (1967) found that with female respondents

there is a curvilinear relationship between self-esteem and conformity in the simple

task condition. Additionally, they argued that the relationship between task difficulty

and conformity increases as the task becomes more complex, up to a certain point;

their data showed that, when exposed to a moderately difficult task, their participants

conformed significantly more than when given an easier task. Morris, Miller, and

Spangenberg (1977) had a similar result in their study and concluded that the

likelihood of conformity depends on the perception of task difficulty.

To conclude, the more important the task is found to be by respondents, the more

important it is to find the right answer for the question given (Rosander and Eriksson,

2012), and the inclination for a higher conformity increases when task difficulty

increases (Baron et al., 1996).

Gender effects on conformity have also been noted (Bond and Smith, 1996). Several

studies show support for a difference in conformity between the two genders, with

women generally inclined to conformity more often than men (Allen and Levine,

1969; Eagly and Carli, 1981; Mori and Arai, 2010). However, one study carried out

in Sweden gave the opposite result, that when faced with a difficult task, men

conformed more than women (Rosander and Eriksson, 2012).

128

Cognitive Load Theory

Cognitive load can be described as a multidimensional construct, which represents the

load that performing a particular task imposes on a learner’s cognitive system (e.g

Meshkati, 1988; Paas and Van Merriënboer, 1994; Yeh and Wickens, 1988). Three

categories of cognitive load have been identified; namely intrinsic cognitive load,

extraneous or ineffective cognitive load, and germane or effective cognitive load

(Paas, Renkl, and Sweller, 2003).

Intrinsic cognitive load is the fundamental level of difficulty associated with a specific

topic; dealing with it involves a combination of working memory and long-term

memory (Paas, Tuovinen, Tabbers, and Van Gerven, 2003). Working memory, in

which all conscious cognitive processing occur, can handle up to 3 interacting

elements at once, and for no longer than 20 seconds, and therefore, this permits only

relatively trivial human cognitive activities (Choi, Van Merriënboer, and Paas, 2014).

Unlike the two major weaknesses of working memory in terms of a severely limited

capacity (Cowan, 2001; Leppink and Van Den Heuvel, 2015; Miller, 1956) and

duration (Peterson and Peterson, 1959), “Long-term memory provides humans with

the ability to vastly expand this processing ability by storing vast numbers of schemas

which are cognitive constructs that incorporate multiple elements of information into

a single element with a specific function” (Paas et al., 2003,p.2).

Secondly, extraneous cognitive load can be explained as an artificially induced

cognitive load, which occurs “when a load is unnecessary and so interferes with

schema acquisition and automation” (Paas et al.,2003, p.3). The third and final form

129

of cognitive load is germane or effective cognitive load. Unlike extraneous cognitive

load, which interferes with learning, germane cognitive load enhances learning. It

results in working memory resources being devoted to schema acquisition and

automation (Sweller, Van Merriënboer, and Paas, 1998).

Cognitive Load Theory, therefore, is contingent upon the notion of a working memory

that is limited in capacity and duration (Sweller, 1988; Cowan, 2001; Leppink and

Van Den Heuvel, 2015). Consequently, “the notion that working memory architecture

and its limitations should be a major consideration when designing instructions” (Paas

et al., 2003). Some factors such as expertise level (e.g. Kalyuga, Ayres, Chandler,

Sweller, 2003), and age (e.g. Paas, Camp, and Rikers, 2001) of respondents are found

to influence cognitive load.

Measurement of Cognitive Load with Subjective Rating Scales

Early measurement approaches for cognitive load conceptualised it in three

dimensions in terms of mental load, mental effort, and performance (Jahns, 1973;

Paas and Van Merriënboer, 1994). Several studies, consequently, attempted to

develop scales for measuring the three dimensions of cognitive load separately

(Ayres, 2006; Cierniak, Scheiter, and Gerjets, 2009; De Leeuw and Mayer, 2008;

Eysink et al., 2009, Galy, Cariou and Mélan, 2012). One of the weaknesses of these

measures was that at least one dimension of cognitive load was represented by a

single item. Leppink, Paas, Van Gog, van Der Vleuten, and Van Merrienboer (2014)

state that in order to be able to generate a more precise measurement, multiple

indicators for each of the dimensions should be considered. Paas (1992) created a

nine-point unidimensional mental effort rating scale that has been widely used to

130

measure the overall cognitive load experienced (Paas et al., 2003; Van Gog and Paas,

2008). Such measures were often derived to be relevant to an educational setting but

no interest in measurement extended to other areas, such as person-machine

interaction and specifically to ergonomics.

The multidimensional NASA-Task Load Index (NASA-TLX) (Hart and Staveland,

1998) is one such instrument that was developed to assess experienced workload

using five seven-point rating scales, “with increments of high, medium, and low

estimates for each point result in 21 gradations on the scales (Hilbert, Renkl, 2009;

Zumbach and Mohraz, 2008)” (p. 34 cited in Leppink et al., 2014).

“Development of the TLX has implied an important and vast program of laboratory

research (Hart and Staveland, 1988), and the instrument’s sensitivity has been

demonstrated using a great variety of tasks. TLX has been applied successfully in

different multitask contexts, as for example in real (Shively, Battiste, Matsumoto,

Pepiton, Bortolussi, and Hart, 1987) and simulated flight tasks (Battiste and

Bortolussi, 1988; Nataupsky and Abbott, 1987; Tsang and Johnson, 1989; Vidulich

and Bortolussi, 1988), in air combat (Bittner, Byers, Hill, and Zacklad, 1989; Hill,

Zacklad, Bittner, Byers, Christ 1988; Hill, Byers, Zacklad, and Christ, 1989), and

using remote-control vehicles (Byers, Bittner, Hill, Zacklad, and Christ, 1988). Sawin

and Scerbo (1995) used the TLX technique to analyse the effects of instruction type

and boredom proneness on vigilance tasks performance”(Rubio, Díaz, Martín, Puente,

2004). The NASA Task Load Index uses six dimensions measuring mental demand,

physical demand, temporal demand, performance, effort, and frustration, to assess

mental workload (Hart and Staveland, 1988).

131

The idea of measuring overall experienced cognitive load, such as when using the

NASA-TLX measure, is widely applicable (Van Gog and Paas, 2008), and the face

validity of several of its items in a market research context high, therefore it was

adopted here in the empirical work that follows.

The Market Research Perspective on Task Difficulty

While the measurement of task difficulty has emerged from research outside of

marketing, concerns have been noted particularly in the market research literature that

parallel those in education and other fields.

Satisficing

When answering a survey question that would optimally require significant cognitive

effort, some respondents might simply choose to provide a satisfactory answer, and

this behaviour is identified as “satisficing” (Krosnick, 1991). Satisficing can take two

forms. The first one occurs when there is incomplete or biased information retrieved,

and/or information integrated. The second form occurs when there is no information

that is retrieved, or no integration at all. “Satisficing may lead respondents to employ

a variety of response strategies, including choosing the first response alternative that

seems to constitute a responsible answer, agreeing with an assertion made by a

question, endorsing the status quo instead of endorsing social change, failing to give

an opinion, and randomly choosing among the response alternatives offered”

(Krosnick, 1991).

Survey respondents are often required to deal with high levels of cognitive effort for

little or no reward, when survey researchers try to gather high-quality data through

many sorts of questions (Krosnick, 1991). As Tourangeau (1984) has described, doing

132

so requires that respondents proceed through four stages of cognitive processing.

They must carefully interpret the meaning of each question, search their memories

extensively or all relevant information, integrate that information carefully into

summary judgements, and report those summary judgements in ways that convey

their meaning as clearly and precisely as possible (e.g. Cannell, Miller, and

Oksenberg, 1981; Tourangeau and Rasinski, 1988; Willis, Royston, and Bercini,

1991). Performing these four steps carefully and comprehensively constitutes what

might be called optimizing.

However, sometimes respondents are motivated to use high levels of cognitive effort

to provide high-quality data for survey researchers due to various different motives

such as desires for self-expression, interpersonal response, and intellectual challenge,

self-understanding, or feeling of altruism (Warwick and Lininger, 1975; Krosnick,

1991). Regardless of their reasons or motivation, respondents probably lose their

vigor/drive relatively quickly during long interviews or questionnaires; they become

increasingly disinterested, impatient, or distracted as the survey progresses. This

distraction leads respondents to change their response strategy. Consequently, instead

of continuing with a high mental effort to generate optimal responses, they are more

likely to compromise, lower their standards and spend less energy thereafter. At first,

respondents probably do so simply by being less thorough in comprehension,

retrieval, judgement, and response selection. They may be less thorough about a

question’s meaning, they may search their memories less thoroughly, they may

integrate retrieved information more carelessly, and they may select a response choice

more haphazardly (see, for example Jabine, 1984, p.19). All four steps are executed,

but each one less diligently and comprehensively than when optimizing occurs. And

instead of attempting to generate an optimal answer, respondents settle for generating

133

merely satisfactory answers. The first answer a respondent considers that seems

acceptable is the one he or she offers. This response behaviour might be termed a

relatively weak form of satisficing” (Krosnick, 1991).

Another reason why satisficing occurs is due to the usage of non-differentiation in

rating scales. “Many survey practitioners believe that answering a series of questions

with the same response alternatives is easier and more enjoyable for respondents and

more efficient than constantly changing response alternatives from question to

question (e.g. Lavrakas, 1993, p. 145-16)”, leading survey designers to group

questions together that offer the same response alternatives. The danger in this

approach is that “satisficing respondents could, for example, simply select a point on

the response scale that appears to be reasonable for the first object, and then rate all of

the remaining objects at that point” (Krosnick, 1991).

Answering a question with “I don’t know” also is another form of satisficing, since

this requires no retrieval or judgment. In order to collect high quality data, and

excluding such responses, filtering questions can be used as remedies. “Mental coin-

flipping” is the final form of satisficing, more notable when the respondents cannot

answer “I don’t know”, they might simply choose randomly from among the response

alternatives offered (Converse, 1964; Krosnick, 1991).

Conditions that Foster Satisficing

In general, the likelihood of satisficing occurring when answering a particular

question is a function of three factors, namely; the inherent difficulty of the task, the

respondent’s ability to perform the required task, and the respondent’s motivation to

perform the task (Krosnick, 1991). To generalize, “the greater the task difficulty, and

the lower the respondent’s ability and motivation to optimize, the more likely

134

satisficing is to occur “(Krosnick, 1991).

Some question stems are difficult to interpret, for instance; “question stems containing

many words require respondents to hold more information in memory in order to

generate a precise answer”, or questions containing rarely used words, or words with

various different meanings. Therefore, in general terms, question stems that are

difficult to interpret are more likely to provoke satisficing (Krosnick, 1991).

Another reason for satisficing can be related to the difficulty of the retrieval process

required by a question. For example, respondents are sometimes asked to report their

current attitudes towards an object, whereas sometimes they are asked what their

attitudes were at some prior time-point. “Reports of current states are presumably

easier than retrospective recall questions because of the relative remoteness of the

relevant information in memory, and questions that require recall of an attitude only a

short time ago are presumably easier than questions that require long-term recall.”

Additionally, the number of objects that are asked to be retrieved from memory is also

another aspect. “A third locus of task difficulty is the judgement stage. Some

questions require relatively simple judgements. It is useful to think of the difficulty of

the judgement phase as a function of the decomposability of the decision to be made;

the more constituent decisions that must be made and integrated into a single

summary judgement, the more difficult this phase will be (see Armstrong, Dennison,

and Gordon, 1975). Also, judgement is more difficult when respondents retrieve many

conflicting pieces of information from memory, as compared to when respondents

recall information that all supports a single judgement. In general, questions entailing

more complex or challenging judgements are more susceptible to satisficing”

(Krosnick, 1991).

135

The Factors that Effect Task Difficulty

Age and Education

The advantage of being older when mediating cognitive performance is having a vast

amount of information in long-term memory to use (Pressley, Borkwski, and

Schneider, 1989). This can be exemplified by the chess expert who remembers more

positions in a mid-game array than the chess novice (e.g., Glaser, Chi and Farr, 1988;

cited in Krosnick, 1990). This might be due to the usage of the prior knowledge on

how to perform a specific task. However, there is evidence to show that not everyone

uses their prior knowledge effectively, such as when university students fail to

activate stored information to mediate their cognitive performance (Pressley et al.,

1989). However, “Since the general cognitive skills, tendencies, and information

stored in long-term memory are fostered by formal education, education can promote

good information processing” (Pressley et al., 1989, pp 863).

The relationship between age and processing speed, and speed and cognition is not

fully understood, but there is a generalisation that increased age is associated with a

slower speed of performing many activities (Salthouse, 1996). It is reported that

cognitive performance is degraded when processing is slow, leading a decrease in

performance on cognition (Salthouse and Babcock, 1991).

Hypotheses

The empirical context for this paper is the measurement of brand image. Mete (2017a

and b) argued that a projective technique such as brand personification can yield

superior results to direct questioning particularly when the respondent is unwilling to

136

provide an answer to questions which might be considered asking for confidential

information (see also Boddy, 2005) such as the image of one’s employer. Projective

techniques, such as personification, have been argued to facilitate response when

questions concern complex ideas such a brand image (Davies, Chun, da Silva, and

Roper, 2001). Therefore, it can be proposed:

H1: Task difficulty will be higher for the direct measurement approach of brand

image compared with a personified measurement approach

Warmth and competence have been argued to be fundamental dimensions of human

judgements (e.g. Expertise (competence) versus trustworthiness (warmth), Hovland,

Janis, and Kelley, 1953; Intellectual (competence) and social (warmth) good-bad,

Rosenberg, Nelson, and Vivekananthan, 1968). Moreover, Wojciszke, Abele, and

Baryla (2009) used various dimensions to show these two dimensions dominate the

judgements of participants. Cuddy, Glick, and Beninger (2011) citing prior research,

argue that warmth judgements have primacy over competence judgements in forming

attitudes about others (Wojciszke and Abele, 2008; Wojciszke, Bazinska, and

Jaworski, 1998), in the identification of words in lexical decision tasks (Ybarra, Chan,

and Park, 2001), when judging faces (Willis and Todorov, 2006), and on children’

judgments generally (Mascaro and Sperber, 2009)). Consequently, having in mind

that warmth judgements are made faster than competence judgements, and the

positive relationship between time and task difficulty, prior work implies that:

137

H2: Task difficulty will be different for evaluations of Warmth and Competence

aspects of brand imagery.

Previous research on task difficulty suggests that education can have a positive effect

on information processing (Pressley, Borkwski, and Schneider, 1989), and therefore

could reduce task difficulty for respondents, but could also produce more challenges.

In terms of gender, there are several studies that show women tend to be more willing

to conform, which could be related to satisficing (Krosnick, 1991), more often than

men (Mori and Arai, 2010). However there is also one study showing men tend to

conform more than women when facing a task that is reported as highly difficult

(Rosander and Eriksson, 2012).

Age is another factor that has been studied within the processing literature (Salthouse,

1996). Age is positively correlated with a decrease in the performance of processing

(Salthouse and Babcock, 1991), and it can be expected therefore that age would be

positively correlated with reported task difficulty.

Consequently, it would be interesting to explore if and how education level, age and

gender affect task difficulty. Based on the previous research mentioned above, it can

be proposed:

H3: For the same task, task difficulty will be higher for (a) older, (b) less educated,

and (c) female respondents

138

Finally, if one or more of these hypotheses hold, then the findings from research

should be influenced by task difficulty:

H4: Task difficulty influences the outcome (findings) of the research questionnaires

Study 1 Methodology

In order to test the hypotheses, an online survey was made with employees as

respondents where the respondents were asked to evaluate the image of the company

that they worked for using either a projective or direct means of questioning. Two

dimensions of brand image were used, warmth and competence. A convenience

sample of 440 respondents from a nationally representative consumer panel was

randomly assigned to one of four groups in a 2x2 factorial, between-subjects design :

(Personification (n=222, 50.5%) vs. Direct (n=218, 49.5%)) x 2 (Warmth (n=223,

50.7%) vs. Competence (n=217, 49.3%)).

In order to evaluate the engagement of the respondents, nine questions were included

to review intellectual engagement, social engagement, and affective engagement with

items adopted from Soane et al. (2012)’s Engagement Scale. Three questions concern

intellectual engagement (whether the respondents would focus hard on their work,

whether they would concentrate on, and whether they would pay a lot of attention to

their work), three social engagement (whether the respondents would share the same

work values as their colleagues, whether they would share the same work goals and

the same work attitudes as their colleagues) and three affective engagement (whether

139

the respondents would feel positive about their work, whether they would feel

energetic, and would be enthusiastic in their work). For each the same response scale

was used from 1 to 7 with points 1, 3 and 7 labeled strongly disagree, neither agree

nor disagree and strongly agree.

Filter questions were asked to confirm the respondents were UK residents and worked

for one employer more than 25 hours per week. The demographic questions followed

the filter questions; such as the age of the respondents, their gender, their education

level, the number of years that they had been working, and the number of years they

had been working for their current company. The sample details are shown in Table1.

Questionnaire Type and Dimension




Direct Approach with Warmth


Direct Approach with Competence


Personification Approach with Warmth


Personification Approach with Competence

Male 79 71.8

Female 31 28.2 Total 110 100

Table 1. Questionnaire Type and Dimension Distribution According to Gender

There were three questions measuring satisfaction, in terms of whether the

respondents would recommend the company that they work for, whether they would

be pleased to be associated with the company they work for, and whether they would

140

feel an affinity with the company they work for, and therefore used to create the

dependent variable named satisfaction (adapted from Davies, Chun, da Silva, and

Roper, 2004). The response scale was from 1-7 with points 1, 3 and 7 labeled:


The brand image item questions followed for either warmth or competence; the two

dimensions being selected as supported as generally applicable for branded entities

(Kervyn, Fiske, and Malone, 2012). For the Warmth dimension the employer brand

image items were selected as friendly, honest, sincere, straightforward, pleasant,

trustworthy, reassuring, supportive, agreeable, concerned, socially responsible,

ethical, cheerful, warm, and open (from; Aaker, Vohs, and Mogilner, 2010; Davies et

al., 2004). For the Competence dimension; reliable, secure, hardworking, ambitious,

achievement oriented, leading, technical, corporate, effective, efficient, competent,

successful, strong, confident, and intelligent were selected (from Aaker, 1997; Davies

et al., 2004).

To measure task difficulty 4 of the 6 items in the NASA Task Load Index (TLX) scale

relevant to the context were chosen, and three items added, drawing upon Bratfisch,

Borg, and Dornic (1972), the Appendix to this paper details the items used. Applying

the TLX normally involves two stages, an evaluation of the task difficulty using the

different scale items and the weighting of each scale item to allow the calculation of

an overall task difficulty score. To avoid the complexity of obtaining weightings, and

following the suggestion of Hendy, Hamilton, and Landry (1993), we limited our

adaption to the use of the items as individual measures without weighting.

141


First the task difficulty scale was tested for reliability using Cronbach Alpha. When

one item was dropped (see appendix) the Cronbach Alpha value for the six item based

scale was .89 for the entire data set (Cronbach, 1951). The Alpha’s for each of the two

groups were also acceptable (Table 2), (Nunnally, 1978; Peterson, 1994).

Method Warmth Competence

Personification .90 .88

Non-Personification .88 .91 Table 2. Cronbach’s Alpha Values for the Task Difficulty Scale by Method and Dimension

Another way of assessment and purification of a scale is to use cut- off points for total

item correlation scores of a proposed scale. Several different cut-off points have been

adopted such as .30 by Cristobal, Flavián and Guinaliu (2007), .40 by Loiacono,

Watson and Goodhue (2002), and with Ladhari (2010) suggesting removing items

with lower correlations from the scale. All six items of the Task Difficulty scale were

found to be acceptable, with corrected item-total correlations ranging from .68 to .79.

After a Task Difficulty Score was created with these six items, a two-way ANOVA

was carried on to understand whether there is a difference in the mean score of

perceived task difficulty between dimensions (warmth and competence) and

questionnaire types (non-personification and personification).

142

When looking at the responses of people who are given ‘warmth’ questionnaires, the

personification approach is rated higher on Task Difficulty (Table 3). Conversely,

when looking at the responses of people who are given competence questionnaires,

the non-personification approach is rated higher on Task Difficulty. Referring back to

H1; the results show that the personification approach has a lower task difficulty score

only when competence dimension is used. However, the outcomes are not statistically

significant. While the task difficulty for warmth respondents was lower in both cases,

as H2 suggests, the difference is not significant and consequently neither H1 nor H2

are supported.

Questionnaire

Type

Dimension Mean N Std. Deviation

Non-

Personification

Warmth 2.38 111 1.17

Competence 2.65 107 1.28

Personification Warmth 2.49 112 1.30


Table 3. Means of Task Difficulty Score by Questionnaire Types and Image Dimensions

Source of

Variation

DF Sum of

Squares

Mean

Squares

F Ratio P Value

Dimension 1 3.48 3.48 2.25 .13

Questionnaire

Type

1 .02 .02 .01 .91

Interaction 1 .93 .93 .60 .44

Error (within) 436 673.9 1.55

Total 440 3473

Table 4. Results of the Analysis of Variance (two-way ANOVA with interaction) of the mean scores of Task Difficulty by Image Dimensions and Questionnaire Types

143

The interaction between questionnaire type and dimension of brand image (warmth or

competence) could not be demonstrated, F (1,436) = .60, p = .44 (Table 4). The R

Squared equals .006 (Adjusted R Squared = .000).

In order to see whether H3 can be supported, several tests were made; firstly, a one-

way ANOVA was conducted for task difficulty to understand if there is a mean

difference between 5 different age groups. The test was significant showing task

difficulty rates differ between the 5 different age groups.(F(5,434) = 3.11, p = .009).

A Tukey post hoc test revealed that the reported task difficulty is significantly lower

for the 46-55 (2.3 ± 1.2, p = .040) and 56-65 (2.3 ± 1.1, p = .035) age groups

compared to the below 25 age group (3.3 ± 1.3). There was no statistically significant

difference between 46-55 and 56-65 age groups (p = 1), see Figure 1.

144

Figure 1. Means Plot for Task Difficulty Score and Age of Respondents

There was a statistically significant difference between groups defined by terminal

education stage as determined by one-way ANOVA (F(3,436) = 7.6, p = .000). A

Tukey post hoc test revealed that the reported task difficulty is statistically

significantly higher for “About 18 (GCSE A level, OND, etc)” (2.8 ± 1.3, p = .000),

“Undergraduate degree (BSc, BA,etc)” (2.6 ± 1.2, p = .004), “Postgraduate degree

(MSc, MA, MBA, PhD, etc)” (2.5 ± 1.2, p = .036) education level groups compared

to “About 16 (GCSE O Level, CSE,etc) “ education level group (2 ± 1.2), see Fig 2.

There was no statistically significant difference between About 18 and Undergraduate

degree groups (p = .324), and between About 18 and Postgraduate (p = .596), and

between Undergraduate and Postgraduate degree groups (p = 1). See also Figure 2.

145

Figure 2. Means Plot for Task Difficulty Score and Education

A T-test was conducted to compare perceived task difficulty by gender. The test

found that male participants reported lower task difficulty (2.5 ± 1.3) compared to

female respondents (2.6 ± 1.1), however this was not statistically significant t(438) =

0.504, p = .61.

Although Task Difficulty did not vary between the four experimental groups, it might

still influence the relationship between variables. To test whether task difficulty

moderates the relationship between the dimensions (warmth or competence) and the

outcome (dependent) variables such as satisfaction and the three engagement scales,

the Process Macro was used (Hayes, 2012). Subsequently, these models were then

enhanced with several covariates such as age, education, and gender.

146

First the relationship between the warmth dimension of employer brand image and

satisfaction was tested with Process Model 1 (Figure 3), with Task Difficulty as a

moderator, (n=223), P= .77 for interaction, P= .54 for Task difficulty. The same test

was conducted for the competence dimension (n=217), P=.20 for interaction, P= .22

for Task difficulty. When predicting satisfaction, there is no statistically significant

moderation effect of Task difficulty for either dimension of brand image.

Figure 3. PROCESS Macro Model 1, where Task Difficulty is M, Warmth or Competence is X, and Satisfaction is Y.

When predicting Engagement (in terms of Social, Affective, and Intellectual, and the

overall Engagement scores), the prediction of Intellectual Engagement by the warmth

dimension was found to be influenced by Task Difficulty (n=223) with p=.004 for

Task Difficulty but not moderated as the interaction term was not significant.

When adding age, gender, and education as covariates, while they proved significant

as covariates, the moderating effect of task difficulty was still not significant. (Task

difficulty did not correlate with any other outcome variable or with any image

measure).

Model Templates for PROCESS for SPSS and SASc⃝2013-2015 Andrew F. Hayes and The Guilford Press

M

YX

Y

b1

eY

1

X

M

XM

b2

b3

Model 1

Conceptual Diagram

Statistical Diagram

Conditional effect of X on Y = b1 + b3M

147

These results lead to the question of whether age could be a moderator in the

relationship between competence and engagement. When these tests were carried out

with age as the moderator to predict overall engagement from the competence

dimension, it proved significant p=.007, with an interaction p-value of .006 (between

Competence and Age).

However, when the warmth dimension was used to test the same relationship, while

age predicted overall engagement (p <001), the interaction term was not significant

(p=.65). The moderating effect of respondent age was then only significant in one of

the eight relationships considered.

To summarize the results from study 1, the expected lower scores for task difficulty

when respondents used a projective technique to assess brand image were not found,

neither did task difficulty moderate the relationship between brand image and the

DV’s tested. However task difficulty did vary by age and education, but not in the

way expected by the literature. Younger, more educated respondents reported higher

task difficulty.

148

Study 2 Methodology The previous study’s results lead the researcher to conclude that there is no reduction

in task difficulty when using a personification approach to measure brand image and

no influence on the relationship between image and typical outcomes.

The hypothesis 2,3, and 4 of Study 1 were retained in this next study (relabeled as H1,

H2, and H5 in this study respectively):

H1: Task difficulty will be different between when Warmth and Competence are used

to evaluate brand image

H2: Task difficulty will be higher for (a) older, (b) less educated, and (c) female

respondents

In addition, there are two more hypotheses proposed for the second study.

Mete (2017a) argued that the assessment of different types of brand (product vs

corporate) would be more or less easier for respondents, hence

H3: Task difficulty will vary with brand type

There are several studies that show that if task difficulty is perceived as higher, it

takes longer to finish the task (Aula, Khan, and Guan, 2010; Liu et al., 2010). Hence:

H4: Task difficulty positively correlates with the time taken to complete the

questionnaire

149

In this study a different context was chosen to test the relevance of task difficulty,

that of respondents as customers. The same brand was chosen, that of Tesco a leading

British grocery retailer, but the context was varied with respondents asked either to

consider Tesco as a corporate brand or as a product brand. (The company has a strong

own brand range)

Particularly if one or more of the hypotheses hold then it follows that

H5: Task difficulty influences the outcome (findings) of the research questionnaires

In this study, a nationally representative consumer panel was used to create

convenience sample of 663 respondents who were randomly assigned to one of four

groups. This study adopted a 2 (Tesco as an Organisational brand (n=444, 67%)) vs.

Tesco as a Product brand (n=219, 33%)) x 2 (Warmth (n=326, 49.2%) vs.

Competence (n=337, 50.8%)) factorial, between-subjects design.

Filter questions were included to ensure the respondents were residing in the UK and

personally shopped for grocery products for their own or others’ use. Following the

filter questions, demographic questions were asked; including the age of the

respondent, their gender, and their education level. Then the respondents were asked

to state how often they shopped at several leading grocery retailers in the British

market such as Tesco, Sainsbury, Morrisons, Asda, Aldi, Waitrose, and Lidl. Next

they were asked if they had a Tesco Club Card.

150

The sample consisted of 313 males (47.2 %), and 350 females (52.8%), which can be

considered as a suitable balance between the genders. The sample details are shown in

Table 5.

Brand Type and Dimension




Organisational Brand with Warmth


Organisational Brand with Competence


Product Brand with Warmth

Male 53 50 Female 53 50 Total 106 100

Product Brand with Competence

Male 49 43.4

Female 64 56.6 Total 113 100

Table 5. Brand Type and Image Dimension Distribution According to Gender The order of the questions was kept similar to that in Study 1. Hence, following the

demographics, satisfaction questions were asked. As in the previous study,

satisfaction was measured with three items (pleased to be associated with, would

recommend to friends and family and overall satisfied).

Respondents assessed either the warmth or competence of the Tesco brand (there are

15 items used for each dimension. For the competence dimension; reliable, secure,

hardworking, ambitious, achievement oriented, leading, technical, corporate,

effective, efficient, competent, successful, strong, confident, intelligent, and for the

warmth dimension: friendly, honest, sincere, straightforward, pleasant, trustworthy,

151

reassuring, supportive, agreeable, concerned, socially responsible, ethical, cheerful,

warm, open are chosen to use. The items were taken as before from Aaker, Vohs, and

Mogilner (2010) and Davies et al. (2004). The same 6 item task difficulty measure

was used as in Study 1.


First the scales used were checked for reliability. The Cronbach Alpha’s for the

groups were found to be acceptable (Table 6), (Nunnally, 1978; Peterson, 1994).

Tesco Brand Type Warmth Competence Organisational Brand .99 .97 Product Brand .98 .97 Table 6. Cronbach’s Alpha Values of Brand Types by Image Dimensions The Alpha values for the task difficulty scale were acceptable (Table 7). Tesco Brand Type Warmth Competence Organisational Brand .86 .86 Product Brand .86 .89 Table 7. Cronbach’s Alpha Values of Dimensions by Brand Type for Task Difficulty Scale (Construct) The total item correlation scores of the task difficulty scale were calculated. All six

items of the Task Difficulty scale were found to be acceptable with corrected item-

total correlations ranging from .58 to .75, above the suggested cuf-off points by

several researchers such as .30 by Cristobal et al. (2007), .40 by Loiacono et al.

(2002), and Ladhari, (2010).

152

A two-way ANOVA was made to understand whether task difficulty would be

different for Warmth and Competence and between corporate brand and product

brand evaluations. This time warmth respondents did not report lower task difficulty

and there was no difference between the measures for brand type. Hence H1 and H3

were not supported.

Brand Type Dimension Mean N Std. Deviation

Organisational

Brand

Warmth 1.97 220 0.94


Product

Brand

Warmth 1.84 106 0.92


Table 8. Means of Task Difficulty Score by Brand Types and Image Dimensions

Source of

Variation

DF Sum of

Squares

Mean

Squares

F Ratio P Value

Dimension 1 0.63 0.63 0.66 .80

Brand Type 1 2.51 2.51 2.61 .11

Interaction 1 .009 .009 .10 .92

Error

(within)

659 633.17 .961

Total 663 3073.61

Table 9. Results of the Analysis of Variance (two-way ANOVA with interaction) of the mean scores of Task Difficulty by Image Dimensions and Brand Types An interaction between task difficulty and dimension of brand image (warmth or

competence) could not be demonstrated as statistically significant, F (1,659) = .11,

p = 0.92 (Table 9). The R Squared equals .004 (Adjusted R Squared = .000).

153

Then a one-way ANOVA was conducted for task difficulty to understand if there is a

mean difference between the 5 different age groups. There was a statistically

significant difference between groups (F(5,657) = 2.58, p = .025), Figure 4.

Figure 4. Means Plot for Task Difficulty Score and Age of Respondents

A Tukey post hoc test revealed that the reported task difficulty is statistically

significantly lower for the 46-55 (2.0 ± 0.9, p = .009) age group compared to the 26-

35 age group (2.1 ± 1.0).

Similarly, There was a statistically significant difference between education groups

(F(3,659) = 4.48, p = .004).

154

A Tukey post hoc test revealed that the reported task difficulty is statistically

significantly higher for “Undergraduate degree (BSc, BA,etc)” (2.04 ± .99, p = .041),

“Postgraduate degree (MSc, MA, MBA, PhD, etc)” (2.20 ± 1.16, p = .012) education

level groups compared to “About 16 (GCSE O Level, CSE,etc) “ education level

group (1.79 ± 0.99) Figure 5. There was no statistically significant difference between

About 16 and About 18 degree groups (p = .828), and between About 18 and

Postgraduate (p = .074), and between Undergraduate and Postgraduate degree groups

(p = .638).

Figure 5. Means Plot for Task Difficulty Score and Education

155

A t test for gender found that male participants reported higher task difficulty (2.0 ±

1.0) compared to female respondents (1.9 ± 1.0), however this was not statistically

significant t(661) = 2.35, p = .125. H2 a and b are supported but not H2c. The

findings confirm those from study 1.

There was no significant correlation between time taken and task difficulty (r=.003,

n=663, p=.934). Even when the respondents were pooled into two groups according to

the amount of time it took them to complete the survey using a median split, while the

participants who spent above average time to complete the survey reported higher task

difficulty (2.0 ± 0.92) compared to below average time takers (1.9 ± 1.1), the

difference was not statistically significant t(661) = -1.15, p = 0.55 .

To test whether task difficulty moderates the relationship between the dimensions

(warmth or competence) and the outcome (dependent) variables such as satisfaction

and involvement, as in study 1 the Process Macro of SPSS was used (Hayes, 2012).

Each model was then enhanced using several covariates such as age, education, and

gender. As in study 1 there was no moderating effect of TD on the relationships

between image and satisfaction, even when age, education and gender were included

as covariates.

When age was used as a moderator when predicting satisfaction from the warmth

dimension, age proved to be a significant moderator (age p= .0001; Warmth p <.001;

interaction term p= .0017). No significant effect was found for Competence.

156

When predicting involvement, for both dimensions, task difficulty on its own was

found to be insignificant related (p=.09) with warmth but less so (p=.19) for

competence.

When adding age, gender, and education as covariates, there was one statistically

significant moderation effect of Task Difficulty, when predicting Involvement with

the warmth dimension (n=326), interaction P= .06, with gender p=.012 and TD=.011,

image p=.82, showing there is some statistically significance.

The same test was conducted for the competence dimension (n=337), but the

interaction term was not significant P=.16, although gender p=.059 and TD=.015

were.


The implications for managers particularly those working for market research

companies, are similar to those for academic researchers. There was no evidence in

this work to suggest that the correlations between dependent and independent

variables depend on task difficulty, however task difficulty influenced scores given to

different constructs and this may be important to marketing practitioners, for example

when comparing the views of younger/older, well-educated/ less well-educated

members of the public. For instance, younger people reported higher task difficulty

compared to older age groups, and well-educated groups reported higher task

difficulty than less well-educated respondents. These findings might give an idea on

how the task difficulty is evaluated in different age and education groups, and might

157

lead to adjustments in market research/ marketing research techniques by including

measures of task difficulty. It is also important to note that two of the studies confirm

that gender has no effect on reported task difficulty, somewhat counter to existing

work.

Younger, more educated respondents report higher task difficulty, which could be

interpreted as because millennials are more likely to exaggerate (see for example,

Yahr and Schimmel, 2013), or that millennials take their work more seriously, and are

trying to be more diligent. Employers in particular might be interested in an

explanation, implying a need for further research. Educators too might consider the

implications for how material is presented to the next generation of students.

Conclusions and Limitations

Prior work shows that there is little difference when using a direct or a personified

(projective) measure of brand image (Mete 2017a and b). One possible explanation is

that the expected influence of personification, by reducing task difficulty, is absent or

irrelevant. In study 1 the task difficulty scores for the personified and direct means

of questioning respondents were similar, suggesting that personification does not

make it easier or respondents to evaluate brand’s image. This finding contradicts prior

work (e.g. Boddy, 2005; Hofstede, van Hoof, Walenberg, and de Jong, 2007).

Prior work also suggests that the benefits of personification might vary by the

dimension of brand image being considered (Mete 2017,a and b). Social cognition

theory further suggests that evaluations of two dimensions of brand imagery (warmth

158

and competence) differ in terms of task difficulty (e.g. Cuddy et al. (2011). Neither

study supports such claims in the context of brand image evaluations and

consequently task difficulty does not explain the small differences found in prior

research to support the benefits of personification (Mete, 2017b).

Task difficulty did however vary as expected with respondent demographics, but not

as some prior work suggests (e.g. Salthouse, 1996; Pressley et al., 1989) based upon

the idea that older, less well educated respondents will have greater difficulty

answering complex questions. Here, task difficulty was significantly higher in both

studies for younger and for more qualified respondents. The age and education of

respondents correlated negatively in both studies (p<.001), reflecting the reality in

society that the younger generation have stayed in full time education for longer than

their parents. The finding here is more compatible with the idea that while cognitive

skills improve with education (Landerl and Wimmer, 2008) recognizing that the task

is difficult leads to a longer time being spent on the task (Rosander and Eriksson,

2012 or that time spent increases the more diligent the respondent (Hornke, 2005).

There are other possible explanations. For example, respondents would not have

known until the end of the questionnaire that the survey was from a University but

might have guessed this and consequently taken the work more seriously if they had

studied for a degree.

Finally Task difficulty sometimes correlated significantly with a dependent variable,

although not as often as might have been expected. However this is worrying as it

implies that research findings can indeed depend upon how difficult respondents find

the task of survey completion to be. Prior work in market research has emphasized

159

task difficulty in the context of making sure a survey is well designed (Krosnick,

1991). The findings here suggest that there may be a wider issue and that researchers

may wish to add a measure of task difficulty as a control variable in a survey

questionnaire.

The work has two main limitations. First, all the research was conducted in English

and in Britain. No attempt was made to replicate the findings in a different language

and culture where personification might have a different influence. For example in

many languages nouns can be either masculine or feminine. Thus different product

types may have different types of brand association due to the gender given to them in

the language. Different cultures have been claimed to be more or less collectivist or

individualistic in nature, and this might influence how people discriminate between

brands. Here the work was conducted only in the UK where the culture is relatively

individualistic. For instance the UK scores highly on Individualism on the Hofstede

measure whereas China scores relatively low (Hofstede, 2018). Second some of the

statistical tests to compare the influences of personified and direct measures might

have been significant if larger samples had been used. It is impossible to prove that

there is no advantage in using a personified measure; however the work does cast

serious doubt on the benefits of its use.

As mentioned above the thesis did not identify any compelling explanation for why

respondents’ replies using either the direct or personified approaches did not differ.

The potential explanation of task difficulty helped somewhat in study 2 but when

more formally assessed in studies 3 and 4 (paper 3) no significant differences

emerged. One possible explanation is empirical, that the sample size we used is too

160

small, and that much larger samples would have provided support for the idea that a

projective technique lowers task difficulty. If however task difficulty cannot be used

to support the use of a projective technique in other words there is no significant

reduction in task difficulty, then thinking on the use of projection generally in market

research needs to be revisited.

Task difficulty did not moderate the relationships in this study between dependent and

independent variables. However what was surprising, and contrary to existing

literature (Ketcham et al., 2002; Craik and McDowd, 1987) was that higher task

difficulty was reported by younger, more educated respondents. We need to

understand why and whether in other contexts task difficulty might moderate

relationships between typical dependent and independent variables. Practitioners

should also be concerned about this possibility. Both academics and practitioners

might then wish to include measures of Task Difficulty in a whole range of different

types of questionnaires to see how significant this previously unexplored variable

might be.

161

Appendix 1. Items Selected for Task Difficulty Measure

(Note: the first added item was dropped in our analyses)

Adapted TLX items and anchored on 1=very low 7=very high

How mentally demanding was it to complete the survey?

How rushed did you feel completing the survey?

How hard did you have to concentrate to complete the survey?

How stressed or annoyed did you feel completing the survey?

Added items:

How easy was it to complete our survey? Anchored on 1=very easy 7= very difficult

(dropped)

I found it difficult to answer most of the questions. Anchored on 1= strongly agree 7=

strongly disagree

I had to think hard in answering the questions. Anchored on 1= strongly agree 7=

strongly disagree

162

Chapter 7: Conclusion

The purpose of this chapter is to synthesize the findings of the work presented in the

main chapters of the thesis beyond the conclusions sections of each individual paper.

The background and context for all three is the wide use of ‘brand personality’ to

measure brand imagery in the marketing and market research literatures. Geuens et

al. (2009) identify 15 published scales of brand personality including the seminal

work of Aaker (1997). At the time of writing the latter paper had 7728 citations in

google.scholar. The various scales share a common approach of inviting survey

respondents to think of the brand they are evaluating as ‘coming to life as a human

being’ and to evaluate its personality. The approach has been criticized because it

evokes the metaphor of brand = person and because it anthropomorphizes an

inanimate object (Davies et al., 2001). The author wanted then to show that there is

some benefit in using such a projective approach that can be used to counter such

criticism and justify its use. Projective techniques can after all help the researcher

because they make it easier for the respondent when faced with the potentially

difficult task of evaluating a brand’s image (Boddy, 2005).

The first paper used a survey of two different brands, Marks and Spencer, a leading

retailer and a corporate brand, and Pantene a leading consumer brand, to compare

the use of a personified measure and its direct equivalent in predicting a number of

dependent variables. Somewhat unexpectedly, there was little in the way of a

systematic advantage in using the projective approach. Sometimes the direct approach

proved more useful statistically. There were no differences between scale validities.

163

The second paper was an attempt to validate the findings from the first by testing

exactly the same issue but in a different context, that of the employer brand. In

addition to being a different context the number of brands being considered would be

far greater than the two investigated in the first study. Some changes were also made

to the measures used but the two studies in the second paper largely replicated the

findings in the first paper, reinforcing the conclusion that there is no obvious

advantage in using a projective approach to measuring a brand’s affective imagery.

For example if the researcher wants to measure how trustworthy a brand is, then there

is little point in asking them first to imagine that the brand has come to life as a

human being.

The second paper went beyond the type of analysis used in the first, especially in its

use of structural equation modelling to compare the personified and direct questioning

approaches. It also considered the possible effects of evaluating different dimensions

of brand image together or separately. The two approaches still proved remarkably

similar when used in these contexts across all types of analysis.

A main contribution of this thesis is then to show that there is no systematic statistical

benefit from adopting the personification approach when measuring brand image. By

doing so it could encourage the wider use of ‘brand personality’ scales to measure

‘brand image’ by both academic and practitioner researchers.

These findings raised the question of why personification did not have an advantage

over direct questioning? In the second study in the second paper a number of

164

questions had been inserted to try to answer this. One item, asking about the difficulty

of answering the survey, moderated the relationship between brand image and some

of the dependent variables. Hence in the final paper the idea was formally tested that

task difficulty could explain when personification was advantageous. It did not and

task difficulty was exactly the same in the first study in paper 3 irrespective of

whether personification was used and irrespective of which of two main dimensions

of brand image were considered.

The absence of any significant difference between the direct and personified approach

raises the question of why? One possible explanation is that the use of a projective

approach did not evoke any different reaction among respondents, because in their

memory brand image and brand personality are very closely associated. The

associative network memory model would imply that that the two memory nodes of

brand personality and brand image are either very close, or very closely linked, in

other words that the associations people make with them are the same (see Keller,

1993).

This led in turn to an interest in the role of task difficulty generally, as three

literatures, education, ergonomics and market research, had considered it. The final

paper then focused on task difficulty as a topic but linked this back to the context of

the thesis. Task difficulty helped explain the findings of the first two papers by

showing that personification does not reduce task difficulty (as implied in the

literature). The paper goes further by showing that task difficulty varies with age and

education, but again not in the way predicted by existing literature (Salthouse and

Babcock, 1991; Paas, Camp, and Rikers, 2001). As such it makes a contribution to the

165

market research literature. Task difficulty was also found to correlate with some of the

dependent variables used in the final paper showing that it can potentially influence

the findings of any questionnaire research.

The work has two main limitations. First, all the research was conducted in English

and in Britain. No attempt was made to replicate the findings in a different language

and culture where personification might have a different influence. Second some of

the statistical tests to compare the influences of personified and direct measures might

have been significant if larger samples had been used. It is impossible to prove that

there is no advantage in using a personified measure; however the work does cast

serious doubt on the benefits of its use.

It would be useful as implied earlier to replicate the work in other cultures and

languages. Further research into task difficulty would also be useful both within the

context of brand image and more generally.

166

References

Aaker, D. A. (1996). Measuring Brand Equity Across Products and Markets.

California Management Review, 38(3), 102-120.

Aaker, D. A. (2004). Leveraging the Corporate Brand. California Management

Review, 46(3), 6-18.

Aaker, J. L. (1997). Dimensions of Brand Personality. Journal of Marketing

Research, 347-356.

Aaker, J. L., Garbinsky, E. N., & Vohs, K. D. (2012). Cultivating Admiration in

Brands: Warmth, Competence, and Landing in the “Golden Quadrant”. Journal of

Consumer Psychology, 22(2), 191-194.

Aaker, J., Vohs, K. D., & Mogilner, C. (2010). Nonprofits are Seen as Warm and For-

Profits as Competent: Firm Stereotypes Matter. Journal of Consumer Research, 37(2),

224-237.

Ackerman, P. L. (1987). Individual Differences in Skill Learning: An Integration of

Psychometric and Information Processing Perspectives. Psychological Bulletin,

102(1), 3.

Aggarwal, P., & McGill, A. L. (2007). Is That Car Smiling at me? Schema Congruity

as a Basis for Evaluating Anthropomorphized Products. Journal of Consumer

Research, 34(4), 468-479.

Alba, J. W., & Hutchinson, J. W. (1987). Dimensions of Consumer Expertise. Journal

of Consumer Research, 13(4), 411-454.

Allen, V. L., & Levine, J. M. (1969). Consensus and Conformity. Journal of

Experimental Social Psychology, 5(4), 389-399.

167

Alnıaçık, E., & Alnıaçık, Ü. (2012). Identifying Dimensions of Attractiveness in

Employer Branding: Effects of Age, Gender, and Current Employment Status.

Procedia-Social and Behavioral Sciences, 58, 1336-1343.

Applbaum, R. F., & Anatol, K. W. (1972). The Factor Structure of Source Credibility

as a Function of the Speaking Situation. Speech Monographs, 39(3), 216-22.

Armstrong, J. S., Denniston Jr, W. B., & Gordon, M. M. (1975). The Use of the

Decomposition Principle in Making Judgments. Organizational Behavior and Human

Performance, 14, 257-263.

Ashton, M. C., Perugini, M., & Vries, R. E. D. (2004). A Six-Factor Structure of

Personality-Descriptive Adjectives: Solutions From Psycholexical Studies in Seven

Languages. Journal of Personality and Social Psychology, 86(2), 356-366.

Aula, A., Khan, R. M., & Guan, Z. (2010, April). How does Search Behavior Change

as Search Becomes More Difficult?. In Proceedings of the SIGCHI Conference on

Human Factors in Computing Systems (pp. 35-44). ACM.

Ayres, P. (2006). Using Subjective Measures to Detect Variations of Intrinsic

Cognitive Load within Problems. Learning and Instruction, 16(5), 389-400.

Balmer, J. M. (2001). Corporate Identity, Corporate Branding and Corporate

Marketing-Seeing Through the Fog. European Journal of Marketing, 35(3/4), 248-

291.

Balmer, J. M., & Gray, E. R. (2003). Corporate Brands: What are They? What of

Them?. European Journal of Marketing, 37(7/8), 972-997.

Baron, R. S., Vandello, J. A., & Brunsman, B. (1996). The Forgotten Variable in

Conformity Research: Impact of Task Importance on Social Influence. Journal of

Personality and Social Psychology, 71(5), 915-927.

168

Battiste, V., & Bortolussi, M. (1988, October). Transport Pilot Workload: A

Comparison of Two Subjective Techniques. In Proceedings of the Human Factors

and Ergonomics Society Annual Meeting (Vol. 32, No. 2, pp. 150-154). SAGE

Publications.

Belk, R. W. (1988). Possessions and the Extended Self. Journal of Consumer

Research, 15(2), 139-168.

Bell, D., & Ruthven, I. (2004). Searcher’s Assessments of Task Complexity for Web

Searching. Advances in Information Retrieval, 57-71.

Bentler, P. M. (1990). Comparative Fit Indexes in Structural Models. Psychological

Bulletin, 107(2), 238.

Bentler, P. M., & Bonett, D. G. (1980). Significance Tests and Goodness of Fit in the

Analysis of Covariance Structures. Psychological Bulletin, 88(3), 588-606.

Berry, L. L., & Parasuraman, A. (2004). Marketing Services: Competing Through

Quality. NY: Simon and Schuster.

Berthon, P., Ewing, M., & Hah, L. L. (2005). Captivating Company: Dimensions of

Attractiveness in Employer Branding. International Journal of Advertising, 24(2),

151-172.

Best, R. M., Rowe, M., Ozuru, Y., & McNamara, D. S. (2005). Deep‐level

Comprehension of Science Texts: The Role of the Reader and the Text. Topics in

Language Disorders, 25(1), 65-83.

Bird, M., Channon, C., & Ehrenberg, A. S. C. (1970). Brand Image and Brand Usage.

Journal of Marketing Research, 7(3), 307.

Bittner, A. C., Byers, J. C., Hill, S. G., Zaklad, A. L., & Christ, R. E. (1989, October).

Generic Workload Ratings of a Mobile Air Defense System (LOS-FH). In

169

Proceedings of the Human Factors and Ergonomics Society Annual Meeting (Vol. 33,

No. 20, pp. 1476-1480). Sage CA: Los Angeles, CA: SAGE Publications.

Blombäck, A., & Axelsson, B. (2007). The Role of Corporate Brand Image in the

Selection of New Subcontractors. Journal of Business & Industrial Marketing, 22(6),

418-430.

Boddy, C. (2005). Projective Techniques in Market Research: Valueless Subjectivity

or Insightful Reality. International Journal of Market Research, 47(3), 239-254.

Boivin, Y. (1986). A Free Response Approach to the Measurement of Brand

Perceptions. International Journal of Research in Marketing, 3(1), 11-17.

Bond, R., & Smith, P. B. (1996). Culture and Conformity: A Meta-Analysis of

Studies Using Asch's (1952b, 1956) Line Judgment Task. Psychological Bulletin,

119(1), 111.

Bower, B. (1999). When Stones Come to Life: Researchers Ponder the Curious

Human Tendency to View All Sorts of Things as Alive. Science News, 155(23), 360-

362.

Bouska, M. L., & Beatty, P. A. (1978). Clothing as a Symbol of Status: Its Effect on

Control of Interaction Territory. Bulletin of the Psychonomic Society, 11(4), 235-238.

Bosnjak, M., Bochmann, V., & Hufschmidt, T. (2007). Dimensions of Brand

Personality Attributions: A Person-Centric Approach in the German Cultural Context.

Social Behavior and Personality: an International Journal, 35(3), 303-316.

Boyer, P. (1996). What Makes Anthropomorphism Natural: Intuitive Ontology and

Cultural Representations. Journal of the Royal Anthropological Institute, 2(1), 83-98.

Bratfisch, O., Borg, G., and Dornic, S. (1972). Perceived Item Difficulty in Three

Tests of Intellectual Performance Capacity (Tech. Rep. No. 29). Stockholm: Institute

of Applied Psychology.

170

Bryk, A. S., & Raudenbush, S. W. (1988). Heterogeneity of Variance in Experimental

Studies: A Challenge to Conventional Interpretations. Psychological Bulletin, 104(3),

396.

Bullmore, J. (1984). The Brand and its Image Re-Visited. International Journal of

Advertising, 3(3), 235-238.

Burke, B. (1994). Position, Personality, Not Price, Should Frame Consumer

Messages. Brandweek, 35(36), 20.

Byers, J. C., Bittner, A. C., Hill, S. G., Zaklad, A. L., & Christ, R. E. (1988, October).

Workload Assessment of a Remotely Piloted Vehicle (RPV) System. In Proceedings

of the Human Factors and Ergonomics Society Annual Meeting (Vol. 32, No. 17, pp.

1145-1149). SAGE Publications.

Byrne, B. M. (1998). Structural Equation Modeling with LISREL, PRELIS, and

SIMPLIS: Basic Concepts, Applications, and Programming. New Jersey: Psychology

Press.

Byström, K., & Järvelin, K. (1995). Task Complexity Affects Information Seeking

and Use. Information Processing & Management, 31(2), 191-213.

Cable, D. M., & Turban, D. B. (2001). Establishing the Dimensions, Sources and

Value of Job Seekers’ Employer Knowledge During Recruitment. In GR Ferris (Ed.),

Research in Personnel and Human Resources Management. (Vol. 20, pp. 115-163).

Emerald Group Publishing Limited.

Campbell, D. T., & Stanley, J. C. (1963). Experimental and Quasi-Experimental

Designs for Research. Boston: Houghton Mifflin.

Cannell, C. F., Miller, P. V., & Oksenberg, L. (1981). Research on Interviewing

Techniques. Sociological Methodology, 12, 389-437.

171

Caprariello, P. A., Cuddy, A. J., & Fiske, S. T. (2009). Social Structure Shapes

Cultural Stereotypes and Emotions: A Causal Test of the Stereotype Content Model.

Group Processes & Intergroup Relations, 12(2), 147-155.

Carlson, R. A., Sullivan, M. A., & Schneider, W. (1989). Practice and Working

Memory Effects in Building Procedural Skill. Journal of Experimental Psychology:

Learning, Memory, and Cognition, 15(3), 517.

Carmines, E. G., & Zeller, R. A. (1979). Reliability and Validity Assessment. Beverly

Hills, CA: Sage Publications.

Chadha, R., & Husband, P. (2010). Cult of the Luxury Brand: Inside Asia's Love

Affair with Luxury. London: Nicholas Brealey Publishing.

Charness, G., Gneezy, U., & Kuhn, M. A. (2012). Experimental Methods: Between-

Subject and Within-Subject Design. Journal of Economic Behavior & Organization,

81(1), 1-8.

Choi, H. H., Van Merriënboer, J. J., & Paas, F. (2014). Effects of the Physical

Environment on Cognitive Load and Learning: Towards a new Model of Cognitive

Load. Educational Psychology Review, 26(2), 225-244.

Chow, G. C. (1960). Tests of Equality Between Sets of Coefficients in Two Linear

Regressions. Econometrica, 28(3), 591-605.

Chun, R., Da Silva, R., Davies, G., & Roper, S. (2002). Corporate Reputation and

Competitiveness. Routledge.

Cialdini, R. B., & Goldstein, N. J. (2004). Social Influence: Compliance and

Conformity. Annual Review of Psychology, 55, 591-621.

Cierniak, G., Scheiter, K., & Gerjets, P. (2009). Explaining the Split-Attention Effect:

Is the Reduction of Extraneous Cognitive Load Accompanied by an Increase in

Germane Cognitive Load?. Computers in Human Behavior, 25(2), 315-324.

172

Cole, M. W., Bagic, A., Kass, R., & Schneider, W. (2010). Prefrontal Dynamics

Underlying Rapid Instructed Task Learning Reverse with Practice. Journal of

Neuroscience, 30(42), 14245-14254.

Collins-Dodd, C., & Lindley, T. (2003). Store Brands and Retail Differentiation: the

Influence of Store Image and Store Brand Attitude on Store Own Brand Perceptions.

Journal of Retailing and Consumer Services, 10(6), 345-352.

Committee on Definitions of the American Marketing Association. (1960). Marketing

Definitions: A Glossary of Marketing Terms. American Marketing Association,

Chicago.

Converse P. E. (1964) The Nature of Belief Systems in the Mass Public. In D. E.

(Ed.), Ideology and Discontent (pp. 206-261). New York, NY: Free Press.

Cornelissen, J., & Harris, P. (2001). The Corporate Identity Metaphor: Perspectives,

Problems and Prospects. Journal of Marketing Management, 17(1-2), 49-71.

Cowan, N. (2001). Metatheory of Storage Capacity Limits. Behavioral and Brain

Sciences, 24(01), 154-176.

Craik, F. I., & McDowd, J. M. (1987). Age Differences in Recall and Recognition.

Journal of Experimental Psychology: Learning, Memory, and Cognition, 13(3), 474.

Cristobal, E., Flavián, C., & Guinaliu, M. (2007). Perceived E-Service Quality (PeSQ)

Measurement Validation and Effects on Consumer Satisfaction and Web Site Loyalty.

Managing Service Quality: An International Journal, 17(3), 317-340.

Cronbach, L. J. (1951). Coefficient Alpha and the Internal Structure of Tests.

Psychometrika, 16(3), 297-334.

173

Cuddy, A. J., Glick, P., & Beninger, A. (2011). The Dynamics of Warmth and

Competence Judgments, and Their Outcomes in Organizations. Research in

Organizational Behavior, 31, 73-98.

Cummins, D. (2005). Dominance, Status, and Social Hierarchies. In D. M. Buss (Ed.),

The Handbook of Evolutionary Psychology (pp. 676–697). Hoboken, NJ: John Wiley

& Sons.

Davies, G. (2008). Employer Branding and its Influence on Managers. European

Journal of Marketing, 42(5/6), 667-681.

Davies, G., & Chun, R. (2003). The Use of Metaphor in the Exploration of the Brand

Concept. Journal of Marketing Management, 19(1-2), 45-71.

Davies, G., Chun, R., da Silva, R. V., & Roper, S. (2004). A Corporate Character

Scale to Assess Employee and Customer Views of Organization Reputation.

Corporate Reputation Review, 7(2), 125-146.

Davies, G., Chun, R., da Silva, R. V., & Roper, S. (2001). The Personification

Metaphor as a Measurement Approach for Corporate Reputation. Corporate

Reputation Review, 4(2), 113-127.

Davis, A., & Bremner, G. (2006). The Experimental Method in Psychology. In G. M.

Breakwell, S. Hammond, C. Fife-Schaw & J. A. Smith (Eds.), Research Methods in

Psychology (3rd ed.) (pp. 64-87). London, England: Sage.

De Leeuw, K. E., & Mayer, R. E. (2008). A Comparison of Three Measures of

Cognitive Load: Evidence for Separable Measures of Intrinsic, Extraneous, and

Germane Load. Journal of Educational Psychology, 100(1), 223.

De Pelsmacker, P., Geuens, M., & van den Bergh, J. (2005). Foundations of

Marketing Communications: A European Perspective. London: Pearson Education.

174

De Pelsmacker, P., Geuens, M., & Van den Bergh, J. (2007). Marketing

Communications: A European Perspective. London: Pearson Education.

De Waal, F. (1982). Chimpanzee Politics: Sex and Power Among Apes. London:

Jonathan Cape.

Deutsch, M., & Gerard, H. B. (1955). A Study of Normative and Informational Social

Influences upon Individual Judgment. The Journal of Abnormal and Social

Psychology, 51(3), 629.

Dobni, D., & Zinkhan, G. M. (1990). In Search of Brand Image: a Foundation

Analysis. ACR North American Advances.

Dolich, I. J. (1969). Congruence Relationships Between Self Images and Product

Brands. Journal of Marketing Research, 6(1), 80-84.

Dooley, L. M., & Lindner, J. R. (2003). The Handling of Nonresponse Error. Human

Resource Development Quarterly, 14(1), 99-110.

Doyle, P., & Fenwick, I. (1974). How Store Image Affects Shopping Habits in

Grocery Chains. Journal of Retailing, 50(4), 39-52.

Dutton, J. E., & Dukerich, J. M. (1994). Organizational Images and Member

Identification. Administrative Science Quarterly, 39(2), 239-263.

Eagly, A. H., & Carli, L. L. (1981). Sex of Researchers and Sex-Typed

Communications as Determinants of Sex Differences in Influenceability: A Meta-

Analysis of Social Influence Studies. Psychological Bulletin, 90(1), 1-20.

Edwards, M. R. (2010). An Integrative Review of Employer Branding and OB

Theory. Personnel Review, 39(1), 5-23.

Ellen, R. (1988). Fetishism. Man, 23: 213-235.

175

Ewing, M. T., & Caruana, A. (1999). An Internal Marketing Approach to Public

Sector Management: the Marketing and Human Resources Interface. International

Journal of Public Sector Management, 12(1), 17-29.

Ewing, M., Pitt, L., & De Bussy, N. (2002). Employment Branding in the Knowledge

Economy. International Journal of Advertising, 21(1), 3-22.

Eysink, T. H., de Jong, T., Berthold, K., Kolloffel, B., Opfermann, M., & Wouters, P.

(2009). Learner Performance in Multimedia Learning Arrangements: An Analysis

across Instructional Approaches. American Educational Research Journal, 46(4),

1107-1149.

Fan, X., Wang, L., & Thompson, B. (1999). Effects of Sample Size, Estimation

Methods, and Model Specification on Structural Equation Modeling Fit Indexes.

Structural Equation Modeling, 6(1), 56-83.

Fehr, E., & Fischbacher, U. (2003). The Nature of Human Altruism. Nature,

422(6928), 785-791.

Fiske, S. T. (1982). Schema-Triggered Affect: Applications to Social Perception. In

Affect and Cognition: 17th Annual Carnegie Mellon Symposium on Cognition (pp. 55-

78). Hillsdale: Lawrence Erlbaum.

Fiske, S. T., Cuddy, A. J., & Glick, P. (2007). Universal Dimensions of Social

Cognition: Warmth and Competence. Trends in Cognitive Sciences, 11(2), 77-83.

Fiske, S. T., Cuddy, A. J., Glick, P., & Xu, J. (2002). A Model of (Often Mixed)

Stereotype Content: Competence and Warmth Respectively Follow From Perceived

Status and Competition. Journal of Personality and Social Psychology, 82(6), 878-

902.

176

Fombrun, C., & Shanley, M. (1990). What's in a Name? Reputation Building and

Corporate Strategy. Academy of Management Journal, 33(2), 233.

Fornell, C., & Larcker, D. F. (1981). Evaluating Structural Equation Models with

Unobservable Variables and Measurement Error. Journal of Marketing Research,

18(1), 39-50.

Fornell, C., & Larcker, D. F. (1981). Structural Equation Models with Unobservable

Variables and Measurement Error: Algebra and Statistics. Journal of Marketing

Research,18(3), 382-388.

Fournier, S. (1998). Consumers and Their Brands: Developing Relationship Theory in

Consumer Research. Journal of Consumer Research, 24(4), 343-373.

Freling, T. H., & Forbes, L. P. (2005). An Empirical Analysis of the Brand

Personality Effect. Journal of Product & Brand Management, 14(7), 404-413.

Galy, E., Cariou, M., & Mélan, C. (2012). What is the Relationship between Mental

Workload Factors and Cognitive Load Types?. International Journal of

Psychophysiology, 83(3), 269-275.

Gardner, B. B., & Levy, S. J. (1955). The Product and the Brand. Harvard Business

Review, 33(2), 33-39.

Gastwirth, J. L., Gel, Y. R., & Miao, W. (2009). The Impact of Levene’s Test of

Equality of Variances on Statistical Theory and Practice. Statistical Science, 24(3),

343-360.

Gensch, D. H. (1978). Image-Measurement Segmentation. Journal of Marketing

Research, 15(3), 384.

George, W. R., & Gronroos, C. (1989). Developing Customer-Conscious Employees

at Every Level: Internal Marketing. In C. A. Congram, & M. L. Friedman (Eds.).

Handbook of Services Marketing(pp. 29-37). New York, NY: AMACOM.

177

Gerbing, D. W., & Anderson, J. C. (1984). On the Meaning of Within-Factor

Correlated Measurement Errors. Journal of Consumer Research, 11(1), 572-80.

Gergen, K. J., & Bauer, R. A. (1967). Interactive Effects of Self-esteem and Task

Difficulty on Social Conformity. Journal of Personality and Social Psychology, 6(1),

16-22.

Geuens, M., Weijters, B., & De Wulf, K. (2009). A New Measure of Brand

Personality. International Journal of Research in Marketing, 26(2), 97-107.

Goldhammer, F., Naumann, J., Stelter, A., Tóth, K., Rölke, H., & Klieme, E. (2014).

The Time on Task Effect in Reading and Problem Solving is Moderated by Task

Difficulty and Skill: Insights from a Computer-Based Large-Scale Assessment.

Journal of Educational Psychology, 106(3), 608-626.

Graeff, T. R. (1997). Consumption Situations and the Effects of Brand Image on

Consumers' Brand Evaluations. Psychology & Marketing, 14(1), 49-70.

Greenwald, A. G. (1976). Within-Subjects Designs: To Use or not to Use?.

Psychological Bulletin, 83(2), 314.

Grohmann, B. (2009). Gender Dimensions of Brand Personality. Journal of

Marketing Research, 46(1), 105-119.

Gultekin, E. (2011, December 1). What’s the Value of Your Employment Brand?

[Blog Post]. Retrieved from: https://business.linkedin.com/talent-

solutions/blog/2011/12/whats-the-value-of-your-employment-brand.

Guthrie, J. T., Wigfield, A., Barbosa, P., Perencevich, K. C., Taboada, A., Davis, M.

H., ... & Tonks, S. (2004). Increasing Reading Comprehension and Engagement

Through Concept-Oriented Reading Instruction. Journal of Educational Psychology,

96(3), 403-423.

178

Gwizdka, J., & Spence, I. (2006). What Can Searching Behavior Tell Us about the

Difficulty of Information Tasks? A Study of Web Navigation. Proceedings of the

Association for Information Science and Technology, 43(1), 1-22.

Han, Y. J., Nunes, J. C., & Drèze, X. (2010). Signaling Status with Luxury Goods:

The Role of Brand Prominence. Journal of Marketing, 74(4), 15-30.

Härkönen, E. (2015). Employer Branding: Case Company Stadium Finland.

Harman, H. H. (1976). Modern Factor Analysis. Chicago, IL: University of Chicago

Press.

Harris, F., & De Chernatony, L. (2001). Corporate Branding and Corporate Brand

Performance. European Journal of Marketing, 35(3/4), 441-456.

Hart S. G., Staveland L. E. (1988). Development of a NASA-TLX (Task Load Index):

Results of Empirical and Theoretical Research. In Hancock P. S., Meshkati N. (Eds.),

Human Mental Workload (pp. 139–183). Amsterdam: Elsevier.

Hartman, K. B., & Spiro, R. L. (2005). Recapturing Store Image in Customer-Based

Store Equity: a Construct Conceptualization. Journal of Business Research, 58(8),

1112-1120.

Hayes, A. F. (2013). Introduction to Mediation, Moderation, and Conditional Process

Analysis: A Regression-Based Approach. New York, NY: Guilford Press.

Hayes, A. F. (2012). PROCESS: A Versatile Computational Tool for Observed

Variable Mediation, Moderation, and Conditional Process Modeling [White Paper].

Retrieved from http://www.afhayes.com/ public/process2012.pdf

Henderson, G. R., Iacobucci, D., & Calder, B. J. (1998). Brand Diagnostics: Mapping

Branding Effects Using Consumer Associative Networks. European Journal of

Operational Research, 111(2), 306-327.

179

Hendon, D. W., & Williams, E. L. (1985). Winning the Battle for Your Customer.

Journal of Consumer Marketing, 2(4), 65-75.

Hendy, K. C., Hamilton, K. M., & Landry, L. N. (1993). Measuring Subjective

Workload: When Is One Scale Better Than Many?. Human Factors: The Journal of

the Human Factors and Ergonomics Society, 35(4), 579-601.

Hill, S. G., Byers, J. C., Zaklad, A. L., & Christ, R. E. (1989, October). Subjective

Workload Assessment during 48 Continuous Hours of LOS-FH Operations. In

Proceedings of the Human Factors and Ergonomics Society Annual Meeting (Vol. 33,

No. 16, pp. 1129-1133). SAGE Publications.

Hill, S. G., Zaklad, A. L., Bittner, A. C., Byers, J. C., & Christ, R. E. (1988, October).

Workload Assessment of a Mobile Air Defense Missile System. In Proceedings of the

Human Factors and Ergonomics Society Annual Meeting (Vol. 32, No. 16, pp. 1068-

1072). SAGE Publications.

Hofstede, G. (2017, November 30) Compare Countries. Retrieved from

https://www.hofstede-insights.com/product/compare-countries/

Hofstede, A., van Hoof, J., Walenberg, N., & de Jong, M. (2007). Projective

Techniques for Brand Image Research: Two Personification-Based Methods

Explored. Qualitative Market Research: An International Journal, 10(3), 300-309.

Hooper, D., Coughlan, J., & Mullen, M. R. (2008). Structural Equation Modelling:

Guidelines for Determining Model Fit. Electronic Journal of Business Research

Methods, 6(1), 53-60.

Hornke, L. F. (2005). Response Time in Computer-Aided Testing: A" Verbal

Memory" Test for Routes and Maps. Psychology Science, 47(2), 280.

Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and Persuasion:

Psychological Studies of Opinion Change . New Haven, CT: Yale University Press

180

Hsieh, M. H. (2002). Identifying Brand Image Dimensionality and Measuring the

Degree of Brand Globalization: A Cross-National Study. Journal of International

Marketing, 10(2), 46-67.

Hu, L. T., & Bentler, P. M. (1999). Cutoff Criteria for Fit Indexes in Covariance

Structure Analysis: Conventional Criteria versus New Alternatives. Structural

Equation Modeling, 6(1), 1-55.

Hupfer, N., Gardner, D. M. (1971). Differential Involvement with Products and

Issues: An Exploratory Study. In D. M. Gardner (Ed.) Proceedings: Association of

Consumer Research (2, 262-269) College Park: MD

IBM Corp. Released 2015. IBM SPSS Statistics for Windows, Version 23.0. Armonk,

NY: IBM Corp.

Jabine, T. B. (1984). Cognitive Aspects of Survey Methodology. In Advanced

Research Seminar on Cognitive Aspects of Survey Methodology (1983-1984: Saint

Michaels, Md. and Baltimore, Md.). Washington, DC: National Academy Press.

Jacoby, J., & Mazursky, D. (1984). Linking Brand and Retailer Images: Do the

Potential Risks Outweigh the Potential Benefits?. Journal of Retailing, 60(2), 105-122

Jahns, D. W. (1973). Operator Workload: What is It and How should It be Measured?

In K. D. Gross & J. J. McGrath (Eds.), Crew System Design. (pp. 281–288). Santa

Barbara, CA: Anacapa Sciences.

James, W. (1999). The Listening Ebony: Moral Knowledge, Religion, and Power

among the Uduk of Sudan. Oxford: Oxford University Press.

Jöreskog, K. G. (1967). Some Contributions to Maximum Likelihood Factor Analysis.

Psychometrika, 32(4), 443-482.

181

Kamins, M. A., & Gupta, K. (1994). Congruence between Spokesperson and Product

Type: A Matchup Hypothesis Perspective. Psychology and Marketing, 11(6), 569.

Kalyuga, S., Ayres, P., Chandler, P., & Sweller, J. (2003). The Expertise Reversal

Effect. Educational Psychologist, 38(1), 23-31.

Kapferer, J. N. (2002). Corporate Brand and Organizational Identity. In B. Moingeon,

& G. Soenon (Eds.), Corporate and Organizational Identities: Integrating Strategy,

Marketing, Communication and Organizational Perspectives (pp. 175-194). London:

Routledge.

Kapferer, J. N., & Laurent, G. (1985). Measuring Consumer Involvement profiles.

Journal of Marketing Research, 22(1), 41-53.

Kassarjian, H. H., & Sheffet, M. J. (1975). Personality and Consumer Behavior: One

More Time. Combined Proceedings (No. 37, pp.197-201). Chicago, IL: AMA,

Keller, K. L. (1993). Conceptualizing, Measuring, and Managing Customer-Based

Brand Equity. the Journal of Marketing, 1-22.

Keller, K. L. (2013). Strategic Brand Management: Global Edition. Pearson Higher

Ed.

Keller, K. L. (1998). Strategic Brand Management: Building, Measuring, and

Managing Brand Equity. New Jersey: Pearson Education.

Keller, K. L., Apéria, T., & Georgson, M. (2008). Strategic Brand Management: A

European perspective. New Jersey: Pearson Education.

Keller, K. L., Parameswaran, M. G., & Jacob, I. (2011). Strategic Brand

Management: Building, Measuring, and Managing Brand Equity. Pearson Education

India.

182

Kenny, D. (1987). Statistics for the Social and Behavior Sciences. Boston: Little

Brown & Co.

Keon, J. W. (1983). Product Positioning: Trinodal Mapping of Brand Images, Ad

Images, and Consumer Preference. Journal of Marketing Research, 20(4), 380-392.

Kervyn, N., Fiske, S. T., & Malone, C. (2012). Brands as Intentional Agents

Framework: How Perceived Intentions and Ability Can Map Brand Perception.

Journal of Consumer Psychology, 22(2), 166-176.

Ketcham, C. J., Seidler, R. D., Van Gemmert, A. W., & Stelmach, G. E. (2002). Age-

Related Kinematic Differences as Influenced by Task Difficulty, Target Size, and

Movement Amplitude. The Journals of Gerontology Series B: Psychological Sciences

and Social Sciences, 57(1), P54-P64.

Kim, J. (2006, April). Task difficulty as a predictor and indicator of web searching

interaction. In CHI'06 Extended Abstracts on Human Factors in Computing Systems

(pp. 959-964). ACM.

Kintsch, W. (1998). Comprehension: A Paradigm for Cognition. Cambridge:

Cambridge University Press.

Kirk, R. E. (1982). Experimental Design. Hoboken, NJ: John Wiley & Sons, Inc.

Kirk, R. E. (1995). Experimental Design: Procedures for the Behavioral Sciences (3rd

edn.). Pacific Grove, CA: Brooks/Cole.

Ko, S. J., Judd, C. M., & Stapel, D. A. (2009). Stereotyping Based on Voice in the

Presence of Individuating Information: Vocal Femininity Affects Perceived

Competence but not Warmth. Personality and Social Psychology Bulletin, 35(2), 198-

211.

Kristin, B., & Surinder, T. (2004). Conceptualizing and Researching Employer

Branding. Career Development International, 9(5), 501-517.

183

Krosnick, J. A. (1990). Expertise and Political Psychology. Social Cognition, 8(1), 1-

8.

Krosnick, J. A. (1991). Response Strategies for Coping with the Cognitive Demands

of Attitude Measures in Surveys. Applied Cognitive Psychology, 5(3), 213-236.

Krugman, H. E. (1977). Memory without Recall, Exposure without Perception.

Journal of Advertising Research, 17(4), 7-12.

Kunerth, B., & Mosley, R. (2011). Applying Employer Brand Management to

Employee Engagement. Strategic HR Review, 10(3), 19-26.

Ladhari, R. (2010). Developing E-Service Quality Scales: A Literature Review.

Journal of Retailing and Consumer Services, 17(6), 464-477.

Lakoff, G., & Johnson, M. (1980). Metaphors We Live by Chicago. Chicago

University.

Landerl, K., & Wimmer, H. (2008). Development of Word reading Fluency and

Spelling in a Consistent Orthography: An 8-year Follow-up. Journal of Educational

Psychology, 100(1), 150-161.

Lavrakas, P. J. (1993). Telephone Survey Methods: Sampling, Selection, and

Supervision. Washington, DC: Sage Publications, Inc.

Lee, H. B. (2008). Using the Chow Test to Analyze Regression Discontinuities.

Tutorials in Quantitative Methods for Psychology, 4(2), 46-50.

Leppink, J., & van den Heuvel, A. (2015). The Evolution of Cognitive Load Theory

and Its Application to Medical Education. Perspectives on Medical Education, 4(3),

119-127.

184

Leppink, J., Paas, F., Van Gog, T., van Der Vleuten, C. P., & Van Merrienboer, J. J.

(2014). Effects of Pairs of Problems and Examples on Task Performance and

Different Types of Cognitive Load. Learning and Instruction, 30(3), 32-42.

Levene, H. (1960). Robust Tests for Equality of Variances. Contributions to

Probability and Statistics, 1, 278-292.

Levy, S.J. (1958). Symbols By Which We Buy. In L. Stockman (ED.), Advancing

Marketing Efficiency (pp. 409-416). Chicago: American Marketing Association.

Li, Y., & Belkin, N. J. (2008). A Faceted Approach to Conceptualizing Tasks in

Information Seeking. Information Processing & Management, 44(6), 1822-1837.

Lievens, F. (2007). Employer Branding in the Belgian Army: The Importance of

Instrumental and Symbolic Beliefs for Potential Applicants, Actual Applicants, and

Military Employees. Human Resource Management, 46(1), 51-69.

Liu, J., Liu, C., Yuan, X., & Belkin, N. J. (2011). Understanding Searchers'

Perception of Task Difficulty: Relationships with Task Type. Proceedings of the

Association for Information Science and Technology, 48(1), 1-10.

Liu, J., Gwizdka, J., Liu, C., & Belkin, N. J. (2010). Predicting Task Difficulty for

Different Task Types. Proceedings of the Association for Information Science and

Technology, 47(1), 1-10.

Lloyd, S. (2002).Branding from the Inside Out. Business Review Weekly, 24(10), 64-

66.

Loiacono, E. T., Watson, R. T., & Goodhue, D. L. (2002). WebQual: A Measure of

Website Quality. Marketing Theory and Applications, 13(3), 432-438.

Louis, D., & Lombart, C. (2010). Impact of Brand Personality on Three Major

Relational Consequences (Trust, Attachment, and Commitment to the Brand). Journal

of Product & Brand Management, 19(2), 114-130.

185

Löhndorf, B., & Diamantopoulos, A. (2014). Internal Branding: Social Identity and

Social Exchange Perspectives on Turning Employees into Brand Champions. Journal

of Service Research, 17(3), 310-325.

Malhotra, N. K. (1981). A Scale to Measure Self-Concepts, Person Concepts, and

Product Concepts. Journal of Marketing Research, 18(4), 456.

Malhotra, N. K. (1988). Self Concept and Product Choice: An Integrated Perspective.

Journal of Economic Psychology, 9(1), 1-28.

Martenson, R. (2007). Corporate Brand Image, Satisfaction and Store Loyalty: A

Study of the Store as a Brand, Store Brands and Manufacturer Brands. International

Journal of Retail & Distribution Management, 35(7), 544-555.

Martineau, P. (1958). The Personality of the Retail Store. Journal of Retailing,

52(Fall), 37-46.

Mascaro, O., & Sperber, D. The Moral, Epistemic, and Mindreading Components of

Children's Vigilance towards Deception. Cognition, 112(3), 367-380.

McCall, M. W., & Bobko, P. (1990). Research Methods in the Service of Discovery.

In M. D. Dunnete & L. Hough (Eds.), Handbook of Industrial and Organizational

Psychology (pp. 381–418). Palo Alto, CA: Consulting Psychologist’s Press.

McClure, P. J., & Ryans, J. K. (1968). Differences between Retailers' and Consumers'

Perceptions. Journal of Marketing Research, 5(1), 35.

McDonald, M. H., De Chernatony, L., & Harris, F. (2001). Corporate Marketing and

Service Brands-Moving Beyond the Fast-Moving Consumer Goods Model. European

Journal of Marketing, 35(3/4), 335-352.

186

McKeown, M. G., Beck, I. L., & Blake, R. G. (2009). Rethinking reading

Comprehension Instruction: A Comparison of Instruction for Strategies and Content

Approaches. Reading Research Quarterly, 44(3), 218-253.

Mete, M., Davies, G., & Whelan, S. (2017). Measuring Brand Image: Personification

versus Non-Personification Methods.

Mete, M., Davies, G., & Whelan, S. (2017). How to Best Measure Employer Brand

Image: Personification versus Direct Methods.

Meshkati, N. (1988). Toward Development of a Cohesive Model of Workload.

Advances in Psychology, 52, 305-314.

Miller, G. A. (1956). The Magical Number Seven, Plus or Minus Two: Some Limits

on Our Capacity for Processing Information. Psychological Review, 63(2), 81.

Mitchell, A. A., & Dacin, P. A. (1996). The Assessment of Alternative Measures of

Consumer Expertise. Journal of Consumer Research, 23(3), 219-239.

Morgan, G., Gregory, F., & Roach, C. (1997). Images of Organization. (Sage

Publications, 1997, Pp. 355-373)

Mori, K., & Arai, M. (2010). No Need to Fake it: Reproduction of the Asch

Experiment without Confederates. International Journal of Psychology, 45(5), 390-

397.

Morris, W. N., Miller, R. S., & Spangenberg, S. (1977). The Effects of Dissenter

Position and Task Difficulty on Conformity and Response Conflict. Journal of

Personality, 45(2), 251-266.

Mulaik, S. A., James, L. R., Van Alstine, J., Bennett, N., Lind, S., & Stilwell, C. D.

(1989). Evaluation of Goodness-of-Fit Indices for Structural Equation Models.

Psychological Bulletin, 105(3), 430.

187

Nandan, S. (2005). An Exploration of the Brand Identity–Brand Image Linkage: A

Communications Perspective. Journal of Brand Management, 12(4), 264-278.

Nataupsky, M., & Abbott, T. S. (1987, September). Comparison of Workload

Measures on Computer—Generated Primary Flight Displays. In Proceedings of the

Human Factors Society Annual Meeting (Vol. 31, No. 5, pp. 548-552). Sage CA: Los

Angeles, CA: SAGE Publications.

Naumann, J., Richter, T., Christmann, U., & Groeben, N. (2008). Working Memory

Capacity and Reading Skill Moderate the Effectiveness of Strategy Training in

Learning from Hypertext. Learning and Individual Differences, 18(2), 197-213.

Naumann, J., Richter, T., Flender, J., Christmann, U., & Groeben, N. (2007).

Signaling in Expository Hypertexts Compensates for Deficits in Reading Skill.


Nelissen, R. M., & Meijers, M. H. (2011). Social Benefits of Luxury Brands as Costly

Signals of Wealth and Status. Evolution and Human Behavior, 32(5), 343-355.

Nunnally, J. (1978). Psychometric Methods (Second ed.). New York, NY: McGraw

Hill.

Nguyen, N., & Leblanc, G. (2001). Corporate Image and Corporate Reputation in

Customers' Retention Decisions in Services. Journal of Retailing and Consumer

Services, 4(8), 227-236.

Ohanian, R. (1990). Construction and Validation of a Scale to Measure Celebrity

Endorsers' Perceived Expertise, Trustworthiness, and Attractiveness. Journal of

Advertising, 19(3), 39-52.

O'malley, L., & Tynan, C. (1999). The Utility of the Relationship Metaphor in Consumer Markets: a Critical Evaluation. Journal of Marketing Management, 15(7), 587-602.

188

Paas, F. G. (1992). Training Strategies for Attaining Transfer of Problem-Solving

Skill in Statistics: A Cognitive-Load Approach. Journal of Educational Psychology,

84(4), 429-434.

Paas, F., Camp, G., & Rikers, R. (2001). Instructional Compensation for Age-Related

Cognitive Declines: Effects of Goal Specificity in Maze Learning. Journal of

Educational Psychology, 93(1), 181-186.

Paas, F., Renkl, A., & Sweller, J. (2003). Cognitive Load Theory and Instructional

Design: Recent Developments. Educational Psychologist, 38(1), 1-4.

Paas, F., Tuovinen, J. E., Tabbers, H., & Van Gerven, P. W. (2003). Cognitive Load

Measurement as a Means to Advance Cognitive Load Theory. Educational

Psychologist, 38(1), 63-71.

Paas, F. G., & Van Merriënboer, J. J. (1994). Variability of Worked Examples and

Transfer of Geometrical Problem-Solving Skills: A Cognitive-Load Approach.


Park, W. C., Jaworski, B. J., & MacInnis, D. J. (1986). Strategic Brand Concept-

Image Management. Journal of Marketing, 50(4), 135-145.

Parker, B. T. (2009). A Comparison of Brand Personality and Brand User-Imagery

Congruence. Journal of Consumer Marketing, 26(3), 175-184.

Peterson, R. A. (1994). A Meta-Analysis of Cronbach's Coefficient Alpha. Journal of

Consumer Research, 21(2), 381-391.

Peterson, L., & Peterson, M. J. (1959). Short-Term Retention of Individual Verbal

Items. Journal of Experimental Psychology, 58(3), 193-198.

Pettijohn, L. S., Mellott, D. W., & Pettijohn, C. E. (1992). The Relationship Between

Retailer Image and Brand Image. Psychology and Marketing, 9(4), 311-328.

189

Phau, I., & Lau, K. C. (2001). Brand Personality and Consumer Self-Expression:

Single or Dual Carriageway?. The Journal of Brand Management, 8(6), 428-444.

Plummer, J. T. (1985). How Personality Makes a Difference. Journal of Advertising

Research, 24(6), 27-31.

Pohlman, A., & Mudd, S. (1973). Market Image as a Function of Consumer Group

and Product Type: A Quantitative Approach. Journal of Applied Psychology, 57(2),

167-171.

Pressley, M., Borkwski, J. G., & Schneider, W. (1989). Good Information Processing:

What It is and How Education can Promote It. International Journal of Educational

Research, 13(8), 857-867.

Priyadarshi, P. (2011). Employer Brand Image as Predictor of Employee Satisfaction,

Affective Commitment & Turnover. Indian Journal of Industrial Relations, 46(3),

510-522.

Puzakova, M., Kwak, H., & Rocereto, J. F. (2013). When Humanizing Brands Goes

Wrong: The Detrimental Effect of Brand Anthropomorphization Amid Product

Wrongdoings. Journal of Marketing, 77(3), 81-100.

Reynolds, T. J., & Gutman, J. (1984). Advertising is Image Management. Journal of

Advertising Research. 24(1), 27-38.

Richter, T., Isberner, M. B., Naumann, J., & Neeb, Y. (2013). Lexical Quality and

Reading Comprehension in Primary School Children. Scientific Studies of Reading,

17(6), 415-434.

Rosenberg, S., Nelson, C., & Vivekananthan, P. S. (1968). A Multidimensional

Approach to the Structure of Personality Impressions. Journal of Personality and

Social Psychology, 9(4), 283-294.

190

Rosander, M., & Eriksson, O. (2012). Conformity on the Internet–The Role of Task

Difficulty and Gender Differences. Computers in Human Behavior, 28(5), 1587-1595.

Rubio, S., Diaz, E., Martin, J., & Puente, J. M. (2004). Evaluation of Subjective

Mental Workload: A Comparison of SWAT, NASA-TLX, and Workload Profile

Methods. Applied Psychology: An International Review, 53(1), 61-86.

Sackmann, S. (1989). The Role of Metaphors in Organization Transformation. Human

Relations, 42(6), 463-485.

Salthouse, T. A. (1996). The Processing-Speed Theory of Adult Age Differences in

Cognition. Psychological Review, 103(3), 403-428.

Salthouse, T. A., & Babcock, R. L. (1991). Decomposing Adult Age Differences in

Working Memory. Developmental Psychology, 27(5), 763- 776.

Sawin, D. A., & Scerbo, M. W. (1995). Effects of Instruction Type and Boredom

Proneness in Vigilance: Implications for Boredom and Workload. Human Factors:

The Journal of the Human Factors and Ergonomics Society, 37(4), 752-765.

Sarstedt, M., Wilczynski, P., & Melewar, T. C. (2013). Measuring Reputation in

Global Markets—A Comparison of Reputation Measures’ Convergent and Criterion

Validities. Journal of World Business, 48(3), 329-339.

Schneider, W., & Chein, J. M. (2003). Controlled & Automatic Processing: Behavior,

Theory, and Biological Mechanisms. Cognitive Science, 27(3), 525-559.

Schneider, W., & Shiffrin, R. M. (1977). Controlled and Automatic Human

Information Processing: I. Detection, Search, and Attention. Psychological Review,

84(1), 1-66.

Schnittka, O., Sattler, H., & Zenker, S. (2012). Advanced Brand Concept Maps: A

New Approach for Evaluating the Favorability of Brand Association Networks.

International Journal of Research in Marketing, 29(3), 265-274.

191

Setchell, J. M., & Wickings, E. J. (2005). Dominance, Status Signals and Coloration

in Male Mandrills. Ethology, 111(1), 25-50.

Shimp, T. A., & Bearden, W. O. (1982). Warranty and Other Extrinsic Cue Effects on

Consumers' Risk Perceptions. Journal of Consumer Research, 9(1), 38-46.

Shively, R. J., Battiste, V., Matsumoto, J. H., Pepitone, D. D., Bortolussi, M. R. &

Hart, S. G. (1987, April). Inflight Evaluation of Pilot Workload Measures for

Rotorcraft Research. In Proceedings of International Symposium on Aviation

Psychology (pp. 637-643). Columbus, OH: Ohio State University

Simpson, E. K., & Ruel, C. Kahler (1980-81),“A Scale for Source Credibility,

Validated in the Selling Context,”. Journal of Personal Selling and Sales

Management, 12(1), 17-25.

Sirgy, M. J. (1985). Using Self-Congruity and Ideal Congruity to Predict Purchase

Motivation. Journal of Business Research, 13(3), 195-206.

Slaughter, J. E., Zickar, M. J., Highhouse, S., & Mohr, D. C. (2004). Personality Trait

Inferences About Organizations: Development of a Measure and Assessment of

Construct Validity. Journal of Applied Psychology, 89(1), 85-103.

Soane, E., Truss, C., Alfes, K., Shantz, A., Gatenby, M., & Rees, C. (2012).

Development and Application of a New Measure of Employee Engagement: the ISA

Engagement Scale. Human Resource Development International, 15(5), 529-547.

Sommers, M. S. (1964). Product Symbolism and the Perception of Social Strata. In

Proceedings of the American Marketing Association (Vol. 22, pp. 200-216).

Stern, B., Zinkhan, G. M., & Jaju, A. (2001). Marketing Images: Construct Definition,

Measurement Issues, and Theory Development. Marketing Theory, 1(2), 201-224.

192

Sullivan, S., Gnesdilow, D., & Puntambekar, S. (2011). Navigation Behaviors and

Strategies Used by Middle School Students to Learn from a Science Hypertext.

Journal of Educational Multimedia and Hypermedia, 20(4), 387-423.

Sunderland, J., Marshall, S., & Parker, B.T. (2004). Real, Ideal, and Undesired Self-

Concepts and Their Effects onViewer Preferences: Who do You Love? In

Proceedings of the American Academy of Advertising (pp. 118-129), Baton Rouge,

LA.

Swanson, R. A., & Holton, E. F. (2005). Research in Organizations: Foundations and

Methods in Inquiry. San Francisco, CA: Berrett-Koehler Publishers.

Sweller, J. (1988). Cognitive Load During Problem Solving: Effects on Learning.

Cognitive Science, 12(2), 257-285.

Sweller, J., Van Merrienboer, J. J., & Paas, F. G. (1998). Cognitive Architecture and

Instructional Design. Educational Psychology Review, 10(3), 251-296.

Torres, A., & Bijmolt, T. H. (2009). Assessing Brand Image Through Communalities

and Asymmetries in Brand-to-Attribute and Attribute-to-Brand Associations.

European Journal of Operational Research, 195(2), 628-640.

Tourangeau, R. (1982). Metaphor and Cognitive Structure, in Miall, D.S. ( Ed.),

Metaphor: Problems and perspectives, 14-35, Harvester Press, Brighton, Sussex.

Tourangeau, R. (1984). Cognitive Sciences and Survey Methods. In National

Research council Committee on National Statistics. In T.B. Jabine, M. L. Straf, J. M.

Tanur, & R. Tourangeau (Eds.), Cognitive Aspects of Survey Methodology: Building a

Bridge Between Disciplines (pp. 73-100). Washington, DC: The National Academies

Press.

Tourangeau, R., & Rasinski, K. A. (1988). Cognitive Processes Underlying Context

Effects in Attitude Measurement. Psychological Bulletin, 103(3), 299.

193

Tsang, P. S., & Johnson, W. W. (1989). Cognitive Demands in Automation. Aviation,

Space, and Environmental Medicine, 60(2), 130.

Vakkari, P. (1999). Task Complexity, Problem Structure and Information Actions:

Integrating Studies on Information Seeking and Retrieval. Information Processing &

Management, 35(6), 819-837.

Van Der Vaart, W., Van Der Zouwen, J., & Dijkstra, W. (1995). Retrospective

Questions: Data Quality, Task Difficulty, and the Use of a Checklist. Quality &

Quantity, 29(3), 299-315.

Van Gog, T., & Paas, F. (2008). Instructional Efficiency: Revisiting the Original

Construct in Educational Research. Educational Psychologist, 43(1), 16-26.

Vidulich, M. A., & Bortolussi, M. R. (1988, October). A Dissociation of Objective

and Subjective Workload Measures in Assessing the Impact of Speech Controls in

Advanced Helicopters. In Proceedings of the Human Factors and Ergonomics Society

Annual Meeting (Vol. 32, No. 19, pp. 1471-1475). Washington, DC: SAGE

Publications.

Walsh, G., Mitchell, V. W., Jackson, P. R., & Beatty, S. E. (2009). Examining the

Antecedents and Consequences of Corporate Reputation: A customer Perspective.

British Journal of Management, 20(2), 187-203.

Warwick, D. P., & Lininger, C. A. (1975). The Sample Survey: Theory and Practice.

New York, NY: McGraw-Hill.

Whelan, S., Davies, G., Walsh, M., & Bourke, R. (2010). Public Sector Corporate

Branding and Customer Orientation. Journal of Business Research, 11(63), 1164-

1171.

Willis, J., & Todorov, A. (2006). First Impressions: Making Up Your Mind After a

100-Ms Exposure to a Face. Psychological Science, 17(7), 592-598.

194

Willis, G. B., Royston, P., & Bercini, D. (1991). The Use of Verbal Report Methods

in the Development and Testing of Survey Questionnaires. Applied Cognitive

Psychology, 5(3), 251-267.

Wojciszke, B., & Abele, A. E. (2008). The Primacy of Communion over Agency and

Its Reversals in Evaluations. European Journal of Social Psychology, 38(7), 1139-

1147.

Wojciszke, B., Abele, A. E., & Baryla, W. (2009). Two Dimensions of Interpersonal

Attitudes: Liking Depends on Communion, Respect Depends on Agency. European

Journal of Social Psychology, 39(6), 973-990.

Wojciszke, B., Bazinska, R., & Jaworski, M. (1998). On the Dominance of Moral

Categories in Impression Formation. Personality and Social Psychology Bulletin,

24(12), 1251-1263.

Yahr, M. A., & Schimmel, K. (2013). Comparing Current Students to a Pre-

Millennial Generation: Are They Really Different?. Research in Higher Education

Journal, 20.

Yaniv, E., & Farkas, F. (2005). The Impact of Person-Organization Fit on the

Corporate Brand Perception of Employees and of Customers. Journal of Change

Management, 5(4), 447-461.

Yaremko, R. M., Harari, H., Harrison, R. C., & Lynn, E. (2013). Handbook of

Research and Quantitative Methods in Psychology: For Students and Professionals.

Hillsdale, NJ: Psychology Press.

Yaremko, R. M., Harari, H., Harrison, R. C., & Lynn, E. (1986). Handbook of

Research and Quantitative Methods in Psychology: For Students and Professionals.

Hillsdale, NJ: Lawrence Erlbaum Associates.

Ybarra, O., Chan, E., & Park, D. (2001). Young and Old Adults' Concerns about

Morality and Competence. Motivation and Emotion, 25(2), 85-100.

195

Yeh, Y. Y., & Wickens, C. D. (1988). Dissociation of Performance and Subjective

Measures of Workload. Human Factors, 30(1), 111-120.

Zaichkowsky, J. L. (1985). Measuring the Involvement Construct. Journal of

Consumer Research, 12(3), 341-352.

Zaltman, G., LeMasters, K., & Heffring, M. (1982). Theory Construction in

Marketing: Some Thoughts on Thinking. John Wiley & Sons.

196

Appendix 1. Questionnaires

Appendix 1.1.1 M&S Direct Questioning Used

Filter Questions

Do you personally do shopping for your own or others’ use? Yes No (If no don’t

continue)

Do you normally live in the UK and have been here more than 1 year? Yes No (If no

don’t continue)

Survey Questions

Your answers will be treated confidentially, as we will only be using the results of the surveys as a whole, not individually. There are no right or wrong answers to any of our questions. PART A May we ask your age? 25 or under 26-35 36-45 46-55 56-65 Over 65 And your gender? Male Female

197

____________________________________________________________________ Instructions: Please READ each statement carefully and CIRCLE the appropriate box as follows:

(5) Strongly Agree (2) Disagree (4) Agree (1) Strongly Disagree (3) Neutral / No opinion (if you don't understand the meaning of the word, please mark no.3)

STRONGLY DISAGREE

DISAGREE

NEITHER AGREE

NOR DISAGREE

AGREE

STRONGLY AGREE

1. Marks & Spencer are a trustworthy organisation

1 2 3 4 5

2. Marks & Spencer are a friendly organisation

1 2 3 4 5

3. Marks & Spencer are an ethical organisation

1 2 3 4 5

4. Marks & Spencer are a sincere organisation

1 2 3 4 5

5. Marks & Spencer are an honest organisation

1 2 3 4 5

6. Marks & Spencer are a socially responsible organisation

1 2 3 4 5

7. Marks & Spencer are a successful organisation

1 2 3 4 5

8. Marks & Spencer are a leading organisation

1 2 3 4 5

9. Marks & Spencer are a reliable organisation

1 2 3 4 5

198

10. Marks & Spencer are a strong organisation

1 2 3 4 5

11. Marks & Spencer are a intelligent organisation

1 2 3 4 5

12. Marks & Spencer are a sophisticated organisation

1 2 3 4 5

13. Marks & Spencer are a prestigious organisation

1 2 3 4 5

14. Marks & Spencer are a up market organisation

1 2 3 4 5

15. Marks & Spencer are a chic organisation

1 2 3 4 5

16. Please write down your thoughts about Marks & Spencer

................................................................................................................................

STRONGLY DISAGREE

DISAGREE

NEITHER AGREE NOR DISAGREE

AGREE

STRONGLY AGREE

17. Marks & Spencer offer good value for money

1 2 3 4 5

18. Marks & Spencer products are good quality

1 2 3 4 5

19. Marks & Spencer has a good reputation as a company

1 2 3 4 5

20. Marks & Spencer’s prices are often expensive

1 2 3 4 5

199

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

21. I would recommend Marks & Spencer to a friend or colleague

1 2 3 4 5

22. I would be pleased to be associated with Marks & Spencer

1 2 3 4 5

23. I feel an affinity with Marks & Spencer

1 2 3 4 5

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

24. I choose where I go shopping carefully

1 2 3 4 5

25. I like to go shopping

1 2 3 4 5

NEVER FREQUENTLY 26. How often do you shop at Marks & Spencer?

1 2 3 4 5

Thank you very much for your time. This project is being conducted by staff and students at Manchester Business School. We appreciate your help.

200

Appendix 1.1.2 M&S Personification Used

Filter Questions


continue)


don’t continue)

Survey Questions


201

____________________________________________________________________ Instructions: Please READ each statement carefully and CIRCLE the appropriate box as follows:


Question: "If Marks & Spencer came to life as a person, what would his/her personality be like?" For example,

1. Friendly: If Marks & Spencer came to life as a person, do you think he/she would be friendly? You are able to choose from 1-5 depending on how strongly you disagree (1) or agree (5). PLEASE ANSWER EVERY LINE

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

1. Trustworthy 1 2 3 4 5 2. Friendly 1 2 3 4 5 3. Ethical 1 2 3 4 5 4. Sincere 1 2 3 4 5 5. Honest 1 2 3 4 5 6. Socially Responsible

1 2 3 4 5

7. Successful 1 2 3 4 5 8. Leading 1 2 3 4 5 9. Reliable 1 2 3 4 5 10. Strong 1 2 3 4 5 11. Intelligent 1 2 3 4 5 12. Sophisticated 1 2 3 4 5 13. Prestigious 1 2 3 4 5 14. Upmarket 1 2 3 4 5 15. Chic 1 2 3 4 5

16. Please write down your thoughts about Marks & Spencer

................................................................................................................................

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

17. Marks & Spencer offer good value for money

1 2 3 4 5

18. Marks & Spencer products are good quality

1 2 3 4 5

202

19. Marks & Spencer has a good reputation as a company

1 2 3 4 5

20. Marks & Spencer’s prices are often expensive

1 2 3 4 5

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

21. I would recommend Marks & Spencer to a friend or colleague

1 2 3 4 5

22. I would be pleased to be associated with Marks & Spencer

1 2 3 4 5

23. I feel an affinity with Marks & Spencer

1 2 3 4 5

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

24. I choose where I go shopping carefully

1 2 3 4 5

25. I like to go shopping

1 2 3 4 5

NEVER FREQUENTLY 26. How often do you shop at Marks & Spencer?

1 2 3 4 5


203

Appendix 1.1.3 Pantene Direct Questioning Used

Filter Questions


continue)


don’t continue)

Survey Questions


204

_____________________________________________________________________ Instructions: Please READ each statement carefully and CIRCLE the appropriate box as follows:


STRONGLY DISAGREE

DISAGREE

NEITHER AGREE

NOR DISAGREE

AGREE

STRONGLY AGREE

1. Pantene is a trustworthy brand

1 2 3 4 5

2. Pantene is a friendly brand

1 2 3 4 5

3. Pantene is an ethical brand

1 2 3 4 5

4. Pantene is a sincere brand

1 2 3 4 5

5. Pantene is an honest brand

1 2 3 4 5

6. Pantene is a socially responsible brand

1 2 3 4 5

7. Pantene is a successful brand

1 2 3 4 5

8. Pantene is a leading brand

1 2 3 4 5

9. Pantene is a reliable brand

1 2 3 4 5

10. Pantene is a strong brand

1 2 3 4 5

11. Pantene is an intelligent brand

1 2 3 4 5

12. Pantene is a sophisticated brand

1 2 3 4 5

13. Pantene is a prestigious brand

1 2 3 4 5

205

14. Pantene is an up market brand

1 2 3 4 5

15. Pantene is a chic brand

1 2 3 4 5

16. Please write down your own thoughts about Pantene Shampoo

................................................................................................................................

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

17. Pantene offer good value for money

1 2 3 4 5

18. Pantene Shampoo are good quality

1 2 3 4 5

19. Pantene has a good reputation as a brand

1 2 3 4 5

20. Pantene’s price is often expensive

1 2 3 4 5

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

21. I would recommend Pantene to a friend or colleague

1 2 3 4 5

22. I would be pleased to be associated with Pantene

1 2 3 4 5

23. I feel an affinity with Pantene

1 2 3 4 5

206

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

24. I choose what I shop for shampoo carefully

1 2 3 4 5

25. I am interested in shampoo shopping

1 2 3 4 5

NEVER FREQUENTLY 26. How often do you buy Pantene?

1 2 3 4 5


207

Appendix 1.1.4. Pantene Personification Used

Filter Questions


continue)


don’t continue)

Survey Questions


208

_____________________________________________________________________ Instructions: Please READ each statement carefully and CIRCLE the appropriate box as follows:


Question: "If Pantene came to life as a person, what would his/her personality be like?" For example, 1. Friendly: If Pantene came to life as a person, do you think he/she would be friendly? You are able to choose from 1-5 depending on how strongly you disagree (1) or agree (5). PLEASE ANSWER EVERY LINE

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

1. Trustworthy 1 2 3 4 5 2. Friendly 1 2 3 4 5 3. Ethical 1 2 3 4 5 4. Sincere 1 2 3 4 5 5. Honest 1 2 3 4 5 6. Socially Responsible

1 2 3 4 5

7. Successful 1 2 3 4 5 8. Leading 1 2 3 4 5 9. Reliable 1 2 3 4 5 10. Strong 1 2 3 4 5 11. Intelligent 1 2 3 4 5 12. Sophisticated 1 2 3 4 5 13. Prestigious 1 2 3 4 5 14. Upmarket 1 2 3 4 5 15. Chic 1 2 3 4 5

16. Please write down your own thoughts about Pantene Shampoo

................................................................................................................................

209

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

17. Pantene offer good value for money

1 2 3 4 5

18. Pantene shampoo are good quality

1 2 3 4 5

19. Pantene has a good reputation as a brand

1 2 3 4 5

20. Pantene’s prices are often expensive

1 2 3 4 5

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

21. I would recommend Pantene to a friend or colleague

1 2 3 4 5

22. I would be pleased to be associated with Pantene

1 2 3 4 5

23. I feel an affinity with Pantene

1 2 3 4 5

STRONGLY DISAGREE

DISAGREE


AGREE

STRONGLY AGREE

24. I choose my shampoo carefully

1 2 3 4 5

25. I am interested in shampoo shopping

1 2 3 4 5

210

NEVER FREQUENTLY 26. How often do you buy Pantene?

1 2 3 4 5


211

Appendix 2. Fisher’s R to Z transformation Tables

DV Dimension Method Pearson Correlation


P value

Significance

Satisfaction Warmth Personification 0.87 112 0 0.5 Not Significant Satisfaction Warmth Direct 0.87 111

Satisfaction Competence Personification 0.69 110 0.14 0.44 Not Significant Satisfaction Competence Direct 0.70 107

Table 1 A. Fisher’s R to Z transformation When Predicting Satisfaction



P value

Significance

Expertise Warmth Personification 0.48 112 1.41 0.08 Not Significant Expertise Warmth Direct 0.32 111

Expertise Competence Personification 0.43 110 0.18 0.43 Not Significant Expertise Competence Direct 0.41 107

Table 1 B. Fisher’s R to Z transformation When Predicting Expertise



P value

Significance

Engagement Warmth Personification 0.77 112 0.67 0.25 Not Significant Engagement Warmth Direct 0.73 111

Engagement Competence Personification 0.74 110 0.6 0.27 Not Significant Engagement Competence Direct 0.70 107

Table 1 C. Fisher’s R to Z transformation When Predicting Overall Engagement



P value

Significance


Warmth Personification 0.60 112 2.16 0.015 Significant


Warmth Direct 0.38 111


Competence Personification 0.62 110 1.65 0.05 Not Significant


Competence Direct 0.46 107

Table 1 D. Fisher’s R to Z transformation When Predicting Intellectual Engagement

212



P value

Significance

Social Engagement

Warmth Personification 0.62 112 0.23 0.41 Not Significant

Social Engagement


Social Engagement


Social Engagement


Table 1 E. Fisher’s R to Z transformation When Predicting Social Engagement



P value

Significance


Warmth Personification 0.80 112 0.12 0.45 Not Significant







Table 1 F. Fisher’s R to Z transformation When Predicting Affective Engagement *Z-critical is 1.96 for p < .05 **If ra is greater than rb, the resulting value of z will have a positive sign; if ra is smaller than rb, the sign of z will be negative.

MEASURING BRAND IMAGE: PERSONIFICATION AND NON ...

Documents