Chapter 5 Research Methodology

166

Chapter 5

Research Methodology

5.1 Introduction

This chapter deals with the details that are critical in executing a meaningful and accurate

report on this study. As marketing research covers a wide range of phenomena, not all

areas will be discussed, however, general concepts, any relevant information and

rationalisation of chosen approaches will be. This chapter will start off with a look at the

research objectives and design, followed by the fieldwork, data preparation and analysis

and finally, limitations of the research.

5.2 Research objectives

In a bid to understand the broader nature of garage convenience stores and traditional

cafés, with regard to the Millennial Generation, specific objectives were identified at the

start of the research project and would assist the researcher in drawing up more accurate

and prudent conclusions. These issues were explored and evaluated, bearing in mind that

some were more relevant. The specific objectives of this study are:

• to determine the current store patronage of the Millennial Generation with regard to

garage convenience stores and traditional cafés;

• to determine the similarities and differences of garage convenience stores and

traditional cafés as perceived by the Millennial Generation; and

• to determine the critical factors that need to be implemented by the two retail types in

order to appeal to the Millennial Generation.

Once the purpose of the study has been identified and examined, the next step is to

formulate ways in which to achieve research objectives. This area is addressed in the

section on research design.

167

5.3 Research design

Whilst research design is defined in numerous ways [Emory & Cooper, 1991:32;

Malhotra, 1996:86; McDaniel & Gates, 1998:30; Tull & Hawkins 1993:50; Churchill,

1999:36; and Zikmund, 2000:65], a suitable adaptation for this report, from the latter

sources, would refer to research design as a framework or blueprint with specifications of

procedures for collecting and analysing the relevant data, from various sources, necessary

to help identify or react to a research problem or opportunity, so that the difference

between the cost of acquiring various levels of accuracy and the expected value of the

information associated with each level of accuracy is maximised.

Generally speaking, experts [Malhotra, 1996:86; McDaniel & Gates, 1998:25; Cravens,

2000:149; Brierty et al., 1998:141; Sudman & Blair, 1997:18; Tull & Hawkins 1993:52;

Walker et al., 1999:123; and Kotler, 2000:106] regard a successful research design to

comprise of the following tasks or components:

• define the research problem;

• determine the problem-solving information that is needed and when it is needed;

• design the exploratory, descriptive or causal phases of the research;

• specify the measurement and scaling procedures;

• construct and pre-test a questionnaire or an appropriate form for data collection;

• specify the sampling process and sample size;

• develop a plan of data analysis and tabulation;

• evaluate the ethics of the research;

• specify the time and financial constraints; and

• follow-up on the completed research study.

Bearing in mind that caution should be exercised to avoid defining a symptom rather than

the underlying problem [Cravens, 2000:148], it is imperative that a framework be

identified and implemented throughout the research design. Subsequently this is

dependent on the nature of the research problem and ultimately can be executed through

the use of exploratory, descriptive or causal research. The following section takes a closer

look at the three previously mentioned categories.

168

5.3.1 Exploratory research

Exploratory research’s primary objective is to provide insights into and comprehension of

the problem situation confronting the researcher. The results, however, are not hoped to

be concise determinations but rather guidelines in achieving a better understanding of the

problem situation at hand, simultaneously encouraging the discretion and creativity of the

researcher [Malhotra, 1996:86; Zikmund, 2000:137; and Sudman & Blair, 1997:22].

Hazlitt [in Daintith, 1997:346] simply put it as: “rules and models destroy genius and art.”

The broad-spectrum cites exploratory research for utilisation for any of the following

reasons [Mariampolski, 2001:23; Churchill & Peter, 1998:122; Zikmund, 2000:137;

Malhotra, 1996:88; and Brassington & Pettitt, 1997:205]:

• to develop hypotheses;

• when little is known about a research issue;

• to isolate key variables and relationships for supplementary examination;

• for positioning studies;

• to familiarise with consumers;

• in establishing precedence for auxiliary research;

• for gaining new ideas;

• in describing processes;

• to formulate a problem or define a problem more specifically; and

• in gaining insights for developing an approach to the problem.

As the general nature of a research problem and the relevant variables are investigated,

exploratory research is typically inevitable when the required information is loosely

defined, resulting in an unstructured working format. This deviation of formal procedures

and protocols allows greater versatility and flexibility. This, however, does not mean that

the results are non-systematic but rather of a qualitative nature. Instead of focusing on

numbers, qualitative research focuses on observations and words: evocative

characterisations, stories, visual depictions, interpretations and other expressive

169

descriptions [McDaniel & Gates 1998:27; Tull & Hawkins, 1993:57; Malhotra, 1996:88;

and Walker et al., 1999:125].

Four general categories of exploratory research methods are at marketers’ disposal (all are

used in this study) namely, expert surveys, case studies, secondary data analysis and pilot

studies. Expert surveys refer to any person who has the knowledge or experience within a

specific area that can be consulted in carrying out exploratory research. Case studies are

past researchers’ results that may be used for the present research study. Secondary data

analysis is an economical approach that looks at existing information based on another

research area, which might appeal to the existing research study. This information may be

ready to use or may require further processing. Secondary data can help this research

study by [Malhotra, 1996:117]:

• identifying and defining the research problem more clearly;

• developing an approach and research design to the research problem;

• answering certain research questions; and

• interpreting primary data more astutely.

Pilot studies (also known as qualitative research) refer to any small-scale exploratory

research technique that uses sampling to gain an understanding of the research problem

setting [Malhotra, 1996:88 and Zikmund, 2000:140].

Qualitative research can either be classified as being direct or indirect with regard to

respondents knowing the actual reason behind the research study. A direct approach is not

disguised whilst an indirect approach disguises the real reason of the study. This research

study has opted for the direct approach by making use of focus groups. A focus group is

an interview conducted by a moderator among a tiny group of respondents in an

amorphous and free-flowing manner [Malhotra, 1996:165 and Tull & Hawkins,

1993:165].

Focus groups are considered to have the following advantages [Malhotra, 1996:172 and

Zikmund, 2000:146]:

170

• they are much quicker when compared to interviewing individuals on a separate basis;

• the use of a specialised (and often expensive) moderator is justified;

• the level of excitement increase as the topic under discussion stimulates respondents;

• analyses can take place even after the discussion ends thanks to video and tape

recordings;

• the moderator has a lot of control on the discussion;

• respondents generally feel safe and comfortable in group discussions and will

therefore express their ideas and opinions more freely;

• the lack of direct questioning to individuals allows more spontaneity;

• a group of respondents in one session can produce a wide range of information and

ideas;

• often one idea can trigger many other ideas and opinions amongst respondents; and

• ideas are more likely to pop up out of the blue in group discussions as opposed to

individual thinking.

Potential disadvantages of focus groups are [Malhotra, 1996:172 and Zikmund,

2000:151]:

• they are sometimes difficult to moderate;

• an unstructured format can lead to messy coding and interpretations;

• they can be misjudged due to research and client biases;

• results may be misused if thought to be conclusive; and

• can be considered not to be representative of the general population.

Although the centre of attention may not be on numbers, reasons for making use of

exploratory research in this report include [Zikmund, 2000:136; Sudman & Blair,

1997:22; Malhotra: 1996:165; Tull & Hawkins, 1993:335; and Mariampolski, 2001:48]:

• it is not always possible, or desirable, to use fully structured methods to obtain

information from respondents;

• people may be unwilling or unable to answer certain questions;

• to provide a greater understanding of a concept;

• studies may be deemed not credible without exploratory usage;

171

• people are unwilling to give truthful answers to questions that invade their privacy,

embarrass them or have a negative impact on their ego;

• to discover new ideas;

• screen alternatives; and

• people may be unable to provide accurate answers to questions that tap their

subconscious.

5.3.2 Descriptive research

Descriptive research’s chief intention is to describe something pertaining to marketing

characteristics, functions or any phenomena [Malhotra, 1996:90 and Zikmund, 2000:56].

Churchill & Peter [1998:122] state that it also studies the relationship (if any) between two

variables. Kent [1999:6], however, notes that the term ‘descriptive’ should only be used

when dealing with research that measures and presents variables one at a time and does

not try to evaluate the relationships between them.

Although a fair amount of understanding of the research problem is generally present, all

attempts with regard to descriptive research have to be as accurate as possible [Malhotra,

1996:90 and Tull & Hawkins, 1993:57]. Researchers [McDaniel & Gates, 1998:30;

Bennett, 1998:228; Malhotra, 1996:91] claim how successful descriptive studies are able

to answer any question with as much precision as possible.

By and large descriptive research is used for any of the following reasons [Malhotra,

1996:90; Brassington & Pettitt, 1997:205; Zikmund, 2000:57; and Walker et al.,

1999:126]:

• to determine the perceptions of product characteristics;

• to understand the decision-making process and influences;

• to make specific predictions such as the sales of a specific product in a specific store;

• to describe the characteristics of relevant groups such as consumers, organisations,

market areas and salespeople;

• to determine the extent of differences in needs, attitudes and opinions among

subgroups;

172

• to estimate the percentage of units in a specified population exhibiting a certain

behaviour; and

• to determine the degree to which marketing variables are associated.

Descriptive research is characterised by the preceding formulations of explicit hypotheses,

therefore stressing the importance of clearly defined research problems. This

subsequently leads to a research design that is more structured, consisting of numerous

planning and statistical methodologies. Specific instructions with regard to estimating

things and large representative samples are also typical characteristics of descriptive

research [Malhotra, 1996:91; Kent, 1999:6; and Kotler, 2000:106].

Whilst any source is available to use, the four major methods that enable the execution of

descriptive studies include: panels, observational data, secondary data and surveys. Panels

refer to groups of people who have agreed to respond to a number of telephone, mail or

personal interviews over a period of time. Observation involves recording the behavioural

patterns of consumers or objects in a methodical way to obtain information about an

incident of interest. Secondary data analysis is a cost-effective approach that examines

existing information based on another research area, which might appeal to the existing

research study. Surveys involve interviews with a large number of respondents using a

pre-designed questionnaire and may be generally classified on the basis of their content.

[Tull & Hawkins, 1993:57 and Malhotra, 1996:92].

This research study has used the survey technique, by making use of questionnaires, as it’s

predominant source in executing its descriptive research. Whilst three types of surveys

can be made use of (telephone, personal or mail), this study is focused on personal

surveys. Personal surveys can be administered almost anywhere such as shopping malls,

at home or on the Internet and generally yield an acceptable response rate [Malhotra,

1996:92].

5.3.3 Causal research

The foremost goal of causal research (also referred to as experimental research) is to

recognise and obtain evidence with regard to cause-and-effect relationships between

173

variables in the problem model. Exploratory and descriptive research usually paves the

way for cause-and-effect relationship studies. Whilst descriptive research is able to

examine associations between variables, causal research has the potential to illustrate that

a change in one variable causes some predictable change in another variable. In the latter

research type, researchers typically have an anticipation about the relationship to be

explained and must thus be quite knowledgeable about the subject in order to attempt to

explain why things happen. [Walker et al., 1999:126; Malhotra, 1996:97; Tull &

Hawkins, 1993:57; Kotler, 2000:106; and McDaniel & Gates, 1998:196]

Roughly speaking, causal research is suitable to be used in any of the following scenarios

[Malhotra, 1996:97 and Sudman & Blair, 1997:24]:

• to realise which variables are the cause (independent variables) and which variables

are the effect (dependent variables) of a phenomenon; and

• to verify the nature of the relationship between the causal variables and the effect to be

predicted.

Causal research is characterised by a structured design and a considerable amount of

planning. This design sees how various independent variables are manipulated in order to

check how a dependent variable is affected within a relatively controlled environment –

something which other research approaches are not capable of doing [Malhotra, 1996:97].

There are, however, disadvantages of causal research. Some of the most common issues

include it being expensive, having administration problems, not always being a good

market representation and that competitors are able to audit the prevailing results [Walker

et al., 1999:126 and Brassington & Pettitt, 1997:206].

This research study does not warrant the use of causal research.

5.3.4 Choosing a research approach

Bearing all of the three previously mentioned research approaches in mind a few questions

still loom: how is it possible to construct the best research design; which is the correct

approach; and who is the undisputed authority on the subject at hand?

174

As George Bernard Shaw [in Daintith, 1997:346] simply put it: “The golden rule is that

there are no golden rules.” From this one sees how there is no clear-cut method as to

which research design is best - it often depends on the nature of the study and discretion of

the researcher [Zikmund, 2000:69]. However, being prudent one should remember that

historical works could help with direction, after all “better one safe way than a hundred on

which you cannot reckon” [Aesop in Ehrligh & de Bruhl, 1996:430].

As the preceding research classifications have been dealt with separately, they are by no

means isolated and sovereign from each other, though it has been noted that research that

restricts itself to creating ideas can be justifiably characterised as ‘exploratory’ and only

research that limits itself to univariate analysis is solely ‘descriptive.’ The required

information needed to help solve a research problem usually depicts which research

approach will be used and to what extent [Malhotra, 1996:98 and Kent, 1999:7]. A

suggested general directive in handling this task is discussed next.

With regard to every changing research needs, it is often not essential to start every

research design using exploratory research, it is however, often the opening measure,

particularly if the research problem is of an ambiguous nature. When researchers are

aware of the problem but still require some more understanding, they turn to descriptive

research. On the other hand, if the research problem is clearly defined, causal research

seems appropriate. As exploratory research subsequently leads to the forming of

hypotheses, it is recommended that descriptive or causal research be used to statistically

test them. It should also be noted that once the latter two research methods have produced

some findings, it does not guarantee a fully comprehendible understanding from the

researcher’s point of view. In these scenarios, exploratory research can be used afterwards

in attempting to shed some light onto the subject [Malhotra, 1996:99 and Zikmund,

2000:59].

When one considers the relationship between the three previously discussed types of

research methods, deciding on which research approach to use is always a tricky decision.

The best way to rationalise the chosen research method(s) is to examine the situation at

hand in logical steps.

175

The purpose of this study is to determine and compare the factors that influence the

preferences of the Millennial Generation when it comes to garage convenience stores and

traditional cafés. From this one can reason what type of information is sought in order to

address the research problem. Since the concept of garage convenience stores, coupled

with the Millennial Generation is a relatively new topic with regard to South Africa, not

much literature is available or dedicated to the subject matter (as discussed in chapter

one). In other words, marketers know very little about these research issues and are in

need of new ideas and approaches in familiarising themselves with the respective young

consumers and their behaviours. Marketers also have to remember that tackling a new

marketing arena in a professional and ethical manner is supreme in setting the precedence

for future research studies. In addition, both types of retailers need to adjust and position

themselves in such a way that they appeal to and are able to influence the youth – the

obstacle is finding out how to.

The previous paragraph has distinctly outlined some of the essential details that are

required in solving the research problem. More interestingly enough is that these details

are all able to converge under the umbrella of exploratory research, but what about in-

depth issues and statistical data?

Adults and children do not think alike and therefore should receive different marketing

treatment (as proposed in chapter one). To facilitate this, a look at personal

characteristics, attitudes, perceptions, opinions, needs and wants have to be examined.

This in turn should try to help explain consumer behaviour and prepare for future trends.

In addition, any marketing variables that forge ahead with regard to extracting this

required information also have to be accounted for.

In light of all these reasons and previous research [Rutenberg, 2000], it is believed that the

most successful research approach taken on in this study would encompass the use of both

exploratory and descriptive research. As this study does not examine any cause-and-effect

relationships with regard to either retailers or the Millennial Generation, it seems that

causal research does not seem to be appropriate.

Now that the information to be obtained is specified and the type of research design has

been determined, focus can be given to the measurement and scaling procedures.

176

5.4 Measurement & scaling procedures

In order to examine and produce accurate and meaningful data, care has to be exercised

regarding measurement and scaling. Measurement refers to assigning numbers or symbols

to characteristics of objects according to certain pre-specified rules [Malhotra, 1996:271].

This study makes use of this concept as numbers consent statistical analysis of the

children’s responses. Furthermore, numbers assists the communication of measurement

rules and results. Scaling refers to generating a continuum upon which measured objects

are located [Malhotra, 1996:271]. For this study, it played a small role since the data

collection format had to be of a simplistic nature due to having been used by children.

There are a variety of primary scales of measurement available to marketers. This study

has made use of two namely, nominal and ordinal scaling. A nominal scale has numbers

or letters that are assigned to an object for identification and classification purposes. For

example, ‘BP Express Shop’ was assigned the number 3 when listing the garage

convenience stores. This nominal scale number does not reflect the amount of

characteristic possessed by the object. However, an ordinal scale arranges objects

according to their magnitude in an ordered relationship [Malhotra, 1996:275 and Tull &

Hawkins, 1993:303]. Question 12 in the questionnaire (Appendix C) illustrates this

concept as the children were asked to rank a list of issues according to their preferences.

With regard to scaling techniques it was decided that a combination of comparative and

non-comparative scales were to be used. A comparative scale indicates a direct

comparison of stimulus objects with one another, whilst a non-comparative scale refers to

each stimulus object being scaled independently of the other objects in the stimulus set

[Malhotra, 1996:276]. Examples of the former and latter are the paired comparisons

between the garage convenience store and café in questions 21 through to 27 (Appendix

C). The idea behind using non-forced choices was to find out which one of the retailers

they prefer more. If the respondents felt that they were impartial to the two retailers, they

could then subsequently express it.

Generally speaking, the greater the number of scale categories, the finer the discrimination

among stimulus objects. Bearing this in mind, it was felt that the number of scale

177

categories should not exceed 4 (such as question 7 for instance). This decision was

influenced by the ability of children to differentiate options and how the data would be

analysed [Zikmund, 2000:374].

Endeavours in assisting good measurements were crucial to the research study. The three

main criteria in accomplishing this were reliability, validity and sensitivity. Reliability is

the degree to which measures are free from random errors and promote consistent results.

Validity refers to the ability of a scale to measure the initial measurement intention whilst

sensitivity is a measurement’s ability to accurately measure variability in responses

[Malhotra, 1996:304 and Tull & Hawkins, 1993:314].

These challenges were addressed using at test-retest method as the sampling and

evaluation process was carried out at different periods and still yielded very similar

results. The focus on sensitivity had a minimal influence due to the nature of keeping

things simple as young respondents were used. Three potential problems - that of

familiarising participants to the study, changing attitudes over long periods and

generalisability were kept in mind at the time. [Malhotra, 1996:308].

Once the measurement and scaling techniques have been given attention, it is important to

determine the objectives required when choosing to either use a questionnaire or

observation form and its respective design.

5.5 Questionnaire design

This study as opted to make use of a questionnaire. The benefits of this data collection

choice include the ability to ensure comparability of the data, increase speed and accuracy

of recording and facilitate data processing [Kotler, 2000:110 and Malhotra, 1996:318].

As always, it was important to consider the right type of the questions and their ordering.

To assist this rationalisation, the use of a few focus groups helped reveal hot topics and

problem areas with regard to this research study’s objectives. Furthermore, ensuring that

the needed information would be retrieved from respondents and that response errors were

minimised, it was decided to construct a preliminary questionnaire that would paint a

small picture of what was to be expected from a final questionnaire. A look at how the

focus groups were conducted now takes place.

178

5.5.1 Focus group sampling process

Qualitative research was needed to gain an understanding of the underlying reasons and

motivations for children’s attitudes and buying behaviour as well as to provide some

credibility to the study. Focus groups can be used successfully with children over the age

of five [Tull & Hawkins, 1993:451]. The exploratory research made use of non-disguised

focus groups. The duration, however, was less than that of an adult focus group. This

accommodated the shorter attention span of children.

The eight focus groups consisted of 6 or 7 children that fell within the sampling frame.

The groups comprised of sample population units that were of the same age and were

familiar with each other. For example, Group A consisted of 7 boys all aged 11 years old.

From previous studies regarding children of this age, researchers prefer sampling groups

of friends. The children are then less prone to lying and more open in telling what is

really on the their mind [Bainbridge, 1999:1]. Mixing various age groups together often

sees how the older children dominate the discussion [Gray, 1997:26]. The flexibility and

versatility is generally useful but a semi-structured moderator’s guide was used.

Consideration was made in ensuring a pleasant ambience for the focus group. The

children were seated around a table and offered refreshments. They were told prior to the

discussion that a tape recorder would be used to record the session. The children were

also informed about the nature of discussion but not given the exact objectives as laid out

in the problem statement. This procedure is warranted by the ethical code of marketing

research practices [Zikmund, 2000:100].

The location of the discussion was either in a school classroom or school hall, depending

on the respective circumstances. This subsequently affected the actual time of the

discussion. Where possible, the researcher attempted to carry out all focus group

discussions in the morning as too avoid having a sample of tired children. Generally,

participants are given a small sum of money or an incentive but as this research study was

not funded by external sources, no monetary incentives were offered to the focus group

members [Kotler, 2000:113].

179

The focus group cassette tapes were played back numerous times in order to efficiently

assess the children’s feedback. The moderation notes were also consulted, and tied up

with any relevant body behaviour portrayed during the session. Where needed, written

dialogues were drawn up.

With the use of the valuable information and creative ideas that were borne from the focus

groups, the researcher was able to draw up a preliminary questionnaire.

5.5.2 Preliminary questionnaire design and testing

Special attention was given to the actual wording and layout of the proposed

questionnaire, attempting to make it easy for children to understand and respond, whilst

still being able to extract vital research information. The proposed questionnaire was

subjected to a sequence of pilot tests, which involved every question being scrutinised and

edited where necessary, resulting in the materialisation of a preliminary questionnaire.

As pre-testing allows one to highlight the weak areas, the preliminary questionnaire

comprising of 34 questions, was given to a test sample of ten children to complete and

evaluate. Any question that seemed to be ambiguous or troublesome to the children was

adjusted accordingly. The necessary final adjustments that were made to the preliminary

questionnaire are shown in table 5.1.

180

Table 5.1: Questionnaire adjustments

Source: Final questionnaire (Appendix C)

With regard to table 5.1, the preliminary questionnaire testing revealed the following

interesting issues, which were taken into account when drawing up the final questionnaire:

• Children were more comfortable “crossing” the questionnaire option boxes as opposed

to “ticking” or “colouring” them in.

• The rank order question (question 12) did not cause any difficulties in the test sample.

• All the children knew what a “landline telephone” was. There was no need to call it a

“Telkom telephone.”

• The font, font size and response spaces seemed to be suitable.

Preliminary satilliteFinal satellitePreliminary What do you generally spend your money on?Final What do you generally spend your pocket money on?Final "make-up / accessories" addedPreliminary How often do you go to a garage convenience store?Final How often do you go inside a garage convenience store?Preliminary How often do you go to a café?Final How often do you go inside a café?

Preliminary (tick as many as you like)

Preliminary How much time do you spend inside the shop?

19 Final ( X as many as you like)Preliminary About 5 minutesFinal About 5 minutes or less

Which of these facilities do you or your family normally use at a garage?

12, 28, 29, 30, 31, 33

5, 8, 13, 16, 17,

18, 32, 34

15, 19(b), 20(b)

Final

Final

Which of the following garage convenience stores have you regularly visited this year?

Which of the following garage convenience stores have you visited this year?

Which of these facilities do you or your parents normally use at a garage?

Final ( X as many as you like)

How much time do you usually spend inside the café or convenience store?

34

33

15

Preliminary

Final

Preliminary

Final

Numbering response lines

Question Number

Question Type Content and changes

9

5

10

11

181

• There were no complaints about the questionnaire being too long or too difficult to

understand.

• All preliminary questionnaires were completed in less than twenty minutes.

5.5.3 Questionnaire content and format

The final questionnaire used (Appendix C) easily fits onto a single A4 paper and consists

of 3 sections and 34 questions. These sections are based on the following:

i. general information for respondent profile assessment (questions 1 to 5);

ii. children’s pocket money (questions 6 to 8); and

iii. comparisons between garage convenience stores and cafés (questions 9 to 34).

The questionnaire was set in English and the types of question formats included the use of

multiple-choice questions, a rank order question and a few open-ended questions (refer to

Appendix C). Again, the wording of the questionnaire was kept simple, thus allowing the

children to easily respond.

Now that the questionnaire design, content and format have been looked at, the next step

is to design suitable sampling procedures.

5.6 Sampling design and process

According to Malhotra [1996:358], the sampling design has to answer a few elementary

questions such as:

i. should a sample be taken?;

ii. if so, what process should be followed?;

iii. what kind of sample should be taken?;

iv. how large should it be?; and

v. what can be done to control and adjust for non-response errors?

This research study has addressed the above questions and the chosen sampling process

that resulted will now be discussed.

182

The population used, consisted of school-going children, aged between 9 and 12 years

(Grade 4 to 7), living in the geographic sampling frame. Taking the limitations and

preliminary research [Rutenberg, 2000] into account, the sampling frame was made up of

children, aged 9 to 12 years (Grade 4 to 7), living between the following boundaries in

Johannesburg, Gauteng:

• The horizontal boundaries are the N1 highway (Northern bypass) and R24 in the

south.

• The vertical boundaries are the N1 highway (Western bypass) and the M1 highway in

the east.

There are approximately 85 schools that cater towards primary school children within the

chosen boundaries. Only schools in the assigned area, that welcomed the research study,

took part. Once the sample size objective (at least 500 respondents) was achieved, it was

not necessary to continue sampling. The exact details regarding the sample size will be

addressed later on in this section. The filling out of questionnaires took place during

school time (morning sessions), thanks to the kind permission of the respective

Headmasters. The actual locations were held either in a classroom or in the school hall.

With regard to choosing which schools and children to use, non-probability sampling

without replacement was the selected sampling technique to be used in collecting samples.

Within this method, judgmental sampling was used.

Non-probability sampling relies on the personal judgement of the researcher rather than on

chance to select sample elements. The probability of any particular member of the

population being chosen is unknown [Zikmund, 2000:474 and Tull & Hawkins,

1993:543]. This approach may yield good estimates of the population characteristics, but

does not allow for objective evaluation of the precision of the sample results. The nature

of this study (exploratory) and the low variability in the population makes this approach

rather favourable. In addition, heavy statistical considerations are not as influential as

opposed to operational considerations [Malhotra, 1996:379].

183

Sampling without replacement indicates that each respondent’s questionnaire and

discussion contribution is unique and cannot be repeated (each respondent completes only

one questionnaire and takes part in one focus group discussion).

Judgmental (or purposive) sampling is appropriate as it costs less, is convenient and is not

time-consuming. The nature of the research objectives, when compared to similar studies,

makes this approach valid and potentially useful [Zikmund, 2000:475 and Malhotra,

1996:366]. Problems of subjectivity were tried to be overcome. One potential problem of

this method is that it does not allow generalisation [Malhotra, 1996:376]. The following

section discusses how the sample size was determined.

Generally speaking the larger and more diverse the sample, the more accurate the

research. Random sampling error varies with samples of different sizes [Zikmund,

2000:513 and Hawkins et al., 2000:743]. Tull & Hawkins [1993:545] believe that as a

sample size increases, judgment becomes less trustworthy compared to random selected

procedures. The five most important issues that have played a role in determining the

sample size include resource constraints, previous studies, the variance of the population,

the magnitude of acceptable error and the confidence interval.

Traditional studies suggest that for research of this temperament, sample size of no fewer

than five hundred entries for the questionnaire is required. A minimum of two groups is

required when dealing with focus groups [Malhotra, 1996:363]. Zikmund [2000:518] is of

the opinions that methodically drawn up tables are essential. Applying his tables to this

research study would result in the following two scenarios:

Assuming that 85% of the targeted population in the designated geographic sampling

frame prefer one retail entity over the other, would result in needing a sample of 322, in

order to have no more than 5 percentage points of error.

The second scenario assumes that 70% of the targeted population in the designated

geographic sampling frame prefer one retail entity over the other, resulting in acquiring a

sample of 306, in order to have no more than 5 percentage points of error.

184

Tull & Hawkins [1993:567] suggest that research that proposes to deal with a regional

population consisting of only a few subgroups, should consider a sample size ranging

between 200 and 500. An average number of subgroups, however, would result in

needing between 500 and 1000 sample units.

An exact figure could not be determined prior to the actual sampling stage, as the

fieldwork relied heavily on the children’s responsiveness and co-operation. Subsequently,

precautions had to have been made in order to achieve the projected target. The benefit of

not mailing the questionnaires was one of the measures taken. Personal administration

and making allowance for random sampling errors were in turn put to full effect.

However, determining suitable sampling sizes is not an exact science and there are factors

that may influence the appropriate size of a sample. An example would see how a

relatively detailed analysis on the data was performed, thus requiring a moderately larger

sample. It should also be noted that a small group of people would not be a representative

sample - no matter how carefully they are recruited. Focus groups cannot take the place of

quantitative studies but are used to complement them [Zikmund, 2000:145].

This concludes the section on research design formulation. The next step in the marketing

research process is that of data collection. It involves deciding on an appropriate means to

gather information and how to record it, whilst eventually allowing it to be processed and

analysed.

5.7 Data collection

The people who design the research rarely collect marketing research data. Generally

speaking, researchers have two options for collecting their data: they can contact a

fieldwork agency or they can develop their own organisations. Irrespective of the chosen

option, some kind of field force will be needed [Zikmund, 2000:531]. Bearing in mind

that this is an individual academic study, the actual design and fieldwork was carried out

by the researcher.

The advantage of this piece of information eliminates the need of having to select, train,

supervise and evaluate field workers. In addition, the chances of violating ethical

185

concerns is also far less [Malhotra, 1996:445]. The major disadvantage is that the

researcher bears all the responsibility. The data collection phase is generally the most

expensive and the most prone to error [Kotler, 2000:113]. The actual process and finer

details of this research study’s fieldwork will now take place.

5.7.1 Fieldwork

• The final questionnaire was administered to the sample population in November 2002.

• The researcher personally administered questionnaires to the children to ensure an

efficient, accurate and effective sample was obtained.

• School teachers played a passive assistance role as requested by the researcher.

• Simple and easy to understand instructions were given to the children and any queries

were answered before and during the execution of the questionnaires.

• As an incentive, children were informed beforehand that honest and accurate work

would be rewarded (sweets), as well as the fact that the questionnaire was an

anonymous one.

• Questionnaires were filled out in English.

• Children were asked to take their time in completing the questionnaires.

• The children were asked to provide their own stationery.

• The researcher emphasised the fact that each child was to give their own answers and

to refrain from coping their friends’ answers.

• Extra questionnaires were available for those children that felt they had made

significant errors and wanted another attempt in filling out their questionnaire.

• Any complaints or compliments were addressed once the children had completed

filling in their questionnaires.

In total 514 questionnaires were completed during the fieldwork period. This figure

represents a 100% return rate due to the researcher personally administering the

questionnaire. There was only one spoilt questionnaire whilst the rest were all acceptable

and usable. There were, however, questionnaires with questions that were left out or

incorrectly marked and is reflected in the analysis in chapter 6.

186

As there are usually many spoilt questionnaires in any survey, this accurate and high

response rate is rather welcomed. It is believed that proper pre-testing, sound

questionnaire designing and a high level of motivation amongst the children, were

responsible for the positive outcome.

Once the fieldwork has been completed, the researcher then starts to prepare the raw data.

The manner in which this is done has to be accurate and meaningful, allowing the required

information to be analysed. The following section takes a look at this course of action.

5.8 Data preparation and analysis

The researcher initially received the questionnaires back from the sample group, which

were then sorted and organised appropriately. With the help of the South African Institute

of Race Relations, the raw data was subjected to various analyses, producing objective-

achieving information. Subsequently this information was then converted into graphical

summaries, where possible, to make direct comparisons between the garage convenience

stores and the traditional cafés. These results will be dealt with in chapter 6.

Peripherals that were used for the data preparation and analysis consisted of:

• the researcher’s custom-designed computerised questionnaires;

• interviews with the targeted group and various study-related professionals;

• making use of the Internet and University in-house sourcing;

• statistical computer programmes and techniques for analysis of collected data; and

• various textbooks, magazines and academic journals.

5.9 Limitations of the research

It is important to be aware that exploratory research (such as focus groups) should never

take the place of quantitative research, which is far more conclusive. Doing so, could lead

to misinterpretations and poor judgements. For example, a child’s body language could be

totally misread by an adult. Another concern is that of making projections from the

187

findings. For example, this study’s sample may not be seen as being representative with

regard to the Millennial Generation due to the chosen method of sampling. The greatest

concern, however, is that of rejecting a good idea during the initial stages of exploratory

research, thus voiding it from being analysed and targeted correctly [Zikmund, 2000:159].

As secondary data has been collected for purposes other than those outlined in this

research study, its usefulness may be restricted in a few ways. To ensure that any

potential hurdles were prevailed over, certain criteria were used to evaluate the data,

which included error and accuracy; dependability; currency; specifications and

methodology; and research nature and objectives [Malhotra, 1996:418 and Arens,

1999:194].

Efforts were made to warrant a minimal impact with regards to research results that were

not biased. Biased research transpires when the research process is executed improperly,

resulting in incorrect findings [Malhotra, 1996:824]. Errors of this nature can occur at

almost any stage and are numerous in variations. Typical issues included whether children

could or wanted to respond; how the interviews were conducted and recorded; deliberate

and unconscious falsification / misrepresentation; and administrative errors. The

preceding section on the sampling process highlighted the major concerns and how they

were overcome.

5.10 Summary

This chapter started out by setting the foundation of the research design. The two chosen

approach angles were that of exploratory and descriptive research. Exploratory research

aims to provide insights and a better understanding of the problem situation whilst

descriptive research tries to describe certain marketing characteristics or functions. The

combination of these two approaches lead to the use of expert surveys, case studies,

secondary data analysis and pilot studies. Furthermore, a surveys technique, that of

questionnaires, was chosen as the primary means in collecting data.

The culmination of the foregoing paragraph on research approaches and techniques

intends to satisfy the research objectives of determining current and future store patronage;

188

similarities and differences of the respective retailers; and what the retailers can do to

ensure better business whilst keeping young consumers happy.

The section on measurement looked at how numbers and symbols were assigned to

object’s characteristics whilst bearing in mind any pre-determined rules. Scaling,

however, was kept to a minimum in order to assist the response rate from the children. In

keeping accurate and user-friendly results, validity, reliability and sensitivity issues were

incorporated. This warranted that measurement intentions were met, that measures were

free from random errors and produced steady results.

It was decided that non-random, judgmental sampling without replacement would be the

suitable method of collecting data as it was relatively inexpensive and not all that time

consuming. The sampling process started out by testing out a preliminary questionnaire.

The rationale behind this move was to see whether the chosen research design proved to

be effective in satisfying objectives as well as determining any potential problems when

carrying out a bigger sampling process. After careful, rewording and other adjustments, a

final questionnaire was produced. This questionnaire was consequently administered to

children that made up the sampling frame. The desired number of questionnaire and focus

group samples were achieved, whilst still maintaining high ethical practices. In addition,

consideration was given to external and situational variables that would lend a hand in a

good response rate and accurate information.

Data analysis was executed with the aid of professional software programmes. Other

sources of reference, such as textbooks and academic journals, were also used. Any

limitations that could influence the research process at any stage, whether at the beginning

of sampling or even after data tabulation, was anticipated and dealt with within reason.

This draws the empirical chapter to an end. The next chapter reveals the findings from for

the research study.