i University of Sheffield A critical analysis of the quality of research of eight randomized controlled trials with objective outcomes supporting the Broaden and Build Theory of Emotions MSc Psychotherapy Studies August 2013
i
University of Sheffield
A critical analysis of the quality of research of eight randomized controlled trials with objective outcomes supporting the Broaden
and Build Theory of Emotions
MSc Psychotherapy Studies
August 2013
ii
Title
A critical analysis of the quality of research of eight randomized controlled trials with objective outcomes supporting the Broaden
and Build Theory of Emotions
Name
Degree
MSc Psychotherapy Studies
School
Health and Related Research
Supervisor
Chris Blackmore
Submission Date
August 2013
Word Count
20,692
iii
Abstract
The purpose of this dissertation was to employ a critical analysis methodology on the
quality of research of eight randomized controlled trials with objective outcomes
supporting the Broaden and Build Theory of Emotions. This study employed a
systematic review to search for RCTs, which included outcome measures that did not
rely solely on self-evaluation measures. The search found eight studies. These eight
studies were then critically analysed based on the following: sample composition, self-
assessment, objective metrics, blinding, randomisation, emotion induction, neutral
induction and neutral control. The critical analysis uncovered areas of concern. These
areas of concern included: over reliance on samples of convenience, measurement
issues related to self-assessment, under reporting of blinding, faulty randomization
and allocation, obscure reporting of active controls, and issues of measurement when
using a neutral control. The objective of this study was to highlight these deficiencies
so that future researchers would be aware of them and thus limit their reoccurrence.
iv
Dedication Page
To Charlotte Molly
and
Christopher Grantham
v
Table of Figures
Figure 1 Filter Selection..........................................................................................................- 27 -
Figure 2 Author Journal Abstract............................................................................................- 28 -
Figure 3 Study Selection.........................................................................................................- 29 -
Figure 4 Validity......................................................................................................................- 30 -
Figure 5 Sample Composition.................................................................................................- 33 -
Figure 6 Reliance on Self-assessment.....................................................................................- 36 -
Figure 7 Objective measures..................................................................................................- 39 -
Figure 8 Blinding.....................................................................................................................- 42 -
Figure 9 Randomization in RCTs.............................................................................................- 44 -
Figure 10 Induction................................................................................................................- 46 -
Figure 11 Induction control....................................................................................................- 50 -
vi
Table of Contents
Abstract....................................................................................................................................... iii
Dedication Page........................................................................................................................... iv
Table of Figures............................................................................................................................v
Table of Contents.........................................................................................................................vi
1. Introduction.........................................................................................................................- 1 -
1.1 Aim of the Study.............................................................................................................- 1 -
1.2 Overview........................................................................................................................- 1 -
1.3 Literature Review-Historical Context..............................................................................- 3 -
1.31 Positive Psychology..................................................................................................- 3 -
1.32 The Development of the Broaden and Build Theory................................................- 7 -
1.33 Emotions.................................................................................................................- 12 -
1.34 Self-Evaluation: Likert scale and Measurement......................................................- 14 -
1.35 Issues in self-reporting scales and measurement...................................................- 15 -
1.36 Blinding in RCTs......................................................................................................- 16 -
1.37 Randomization in RCTs...........................................................................................- 18 -
1.38 Mood Induction......................................................................................................- 20 -
2. Methods...........................................................................................................................- 25 -
2.1 Methodology for Search Strategy.................................................................................- 25 -
2.2 Methodology for Analysis.............................................................................................- 25 -
2.3 Search Strategy.............................................................................................................- 25 -
2.4 Final Search Strategy....................................................................................................- 26 -
2.41 Rationale................................................................................................................- 26 -
2.42 Flow chart-Selection-filters.....................................................................................- 27 -
2.43 Chart: Study Selection Criteria form......................................................................- 29 -
2.44 Chart: Sample of Validity Assessment Chart...........................................................- 30 -
2.5 Criteria used for a Summary of Study Chart.................................................................- 31 -
2.6 Methodological Issues..................................................................................................- 31 -
vii
3. Findings and Results...........................................................................................................- 32 -
Chart 3.1 Critical Analysis: Sample Composition................................................................- 33 -
Chart 3.2 Critical Analysis: Self-Assessment.......................................................................- 36 -
Chart 3.3 Critical Analysis: Objective Outcomes.................................................................- 39 -
Chart 3.4 Critical Analysis: Blinding....................................................................................- 42 -
Chart 3.5 Critical Analysis: Randomization in RCTs.............................................................- 44 -
Chart 3.6 Critical Analysis: Induction..................................................................................- 46 -
Chart 3.7 Critical Analysis: Neutral Induction as Control....................................................- 50 -
4. Conclusions........................................................................................................................- 56 -
4.1 Summary of Findings and Remedial Recommendations..............................................- 56 -
4.2 Recommendations for further research.......................................................................- 59 -
4.3 Limitations to the study................................................................................................- 60 -
4.4 Summary......................................................................................................................- 61 -
References.............................................................................................................................- 62 -
Appendix................................................................................................................................- 67 -
Please note: The information contained in some of the charts should be considered as
paraphrasing or as direct quotes from the actual studies. This process allows the integrity of
the authors’ statements to be maintained.............................................................................- 67 -
Study 1...................................................................................................................................- 68 -
Study Selection Form..........................................................................................................- 68 -
Summary of study..............................................................................................................- 69 -
Schemata of experiment....................................................................................................- 70 -
Assessment of risk of bias for RCTs....................................................................................- 71 -
Email correspondence........................................................................................................- 71 -
Study 2...................................................................................................................................- 72 -
Study selection form..........................................................................................................- 72 -
Summary of study..............................................................................................................- 73 -
Schemata of experiment....................................................................................................- 74 -
Assessment of risk of bias for RCTs....................................................................................- 75 -
Email correspondence........................................................................................................- 75 -
viii
Study 3...................................................................................................................................- 76 -
Study selection form..........................................................................................................- 76 -
Summary of study..............................................................................................................- 77 -
Schemata of experiment....................................................................................................- 78 -
Assessment of risk of bias for RCTs....................................................................................- 79 -
Email correspondence........................................................................................................- 79 -
Study 4...................................................................................................................................- 80 -
Study selection form..........................................................................................................- 80 -
Summary of study..............................................................................................................- 81 -
Schemata of experiment....................................................................................................- 82 -
Assessment of risk of bias for RCTs....................................................................................- 83 -
Email correspondence: None.............................................................................................- 83 -
Study 5...................................................................................................................................- 84 -
Study selection form..........................................................................................................- 84 -
Summary of study..............................................................................................................- 85 -
Schemata of experiment....................................................................................................- 86 -
Assessment of risk of bias for RCTs....................................................................................- 87 -
Email Correspondence.......................................................................................................- 87 -
Study 6...................................................................................................................................- 89 -
Study selection form..........................................................................................................- 89 -
Summary of study..............................................................................................................- 90 -
Schemata of experiment....................................................................................................- 91 -
Assessment of risk of bias for RCTs....................................................................................- 92 -
Email Correspondence.......................................................................................................- 92 -
Study 7...................................................................................................................................- 94 -
Study selection form..........................................................................................................- 94 -
Summary of study..............................................................................................................- 95 -
Schemata of experiment....................................................................................................- 96 -
Assessment of risk of bias for RCTs....................................................................................- 97 -
ix
Email correspondence........................................................................................................- 97 -
Study 8...................................................................................................................................- 98 -
Study selection form..........................................................................................................- 98 -
Summary of study..............................................................................................................- 99 -
Schemata of experiment..................................................................................................- 100 -
Assessment of risk of bias for RCTs..................................................................................- 101 -
Email correspondence......................................................................................................- 101 -
1
1. Introduction
1.1 Aim of the Study This study looks at the quality of research developed in Applied Positive Psychology
(APP). Originally APP research centred on correlational studies and self-evaluation
measures for outcomes. This quality of research was satisfactory for building research
questions and developing testable hypothesis, but it was limited as the research was
correlational. RCTs were then utilized in APP studies in order to test for causality.
However, the fact that the final outcomes still relied on self-evaluation measurements
limited the rigour of the studies. In this dissertation, a systematic search was used to
find studies that had objective outcomes that were free from the subjective influence
of self-evaluations. Eight studies were found to meet the search criteria. The aim of
this study was to critically analyse the quality of the research of these eight studies
based on the following seven criteria:
sample composition (population) reliance on self-assessment (study integrity) objective metrics (outcomes) the execution of blinding (study integrity) the quality of each of the randomly controlled trials (RCT) (study design) the use of induction (intervention) the use of neutral induction as control (control)
These seven criteria were selected based on information presented in the literature
review.
The research question is: Are there issues of quality in the research of these eight
studies, and if so, what are they, and how can they be reduced or eliminated?
The objective of this dissertation is to highlight current problems in APP research in
order to raise the quality of future APP research.
1.2 Overview Chapter 1 includes a brief introduction, the aims, the research question and the
objective of the study. Chapter 1 also includes the literature review, which is much
more than simply a presentation of the literature as the literature review presents the
historical context that shaped and developed the science relating to APP and BBT. This
context is vital when using a critical analysis methodology as there needs to be an
2
understanding of the personalities and the motivations of the time and how this links
to the present.
The numerous sections in the literature review also broaden understanding of specific
aspects of research listed in the aims of the review including: the nature and
characteristics of emotions, the challenge of defining the term emotions, the concepts
of state and trait, intensity and duration, discrete and collaborative emotions and
positive and negative emotions. In addition, issues of measurement were considered
including such tools as Likert Scales, Likert-Type Scales, Multi-Scales and objective
measures. A review of studies on emotions highlights their complexity and the
challenge in measuring them.
Chapter 2 is the Methodology. Two methodologies were used. The first methodology
is based on a systematic review as presented in The Cochrane Handbook for
Systematic Reviews of Interventions Version 5.1.0 edited by Higgins and Green (2011).
Once the studies were selected, the information was then analysed using the second
methodology: a critical analysis approach defined by Scriven and Paul (2003).
Chapter 3 is the Analysis. Here the findings were unearthed using a critical analysis
approach. In this chapter, each of the eight studies was analysed on the bases of:
sample composition, reliance on self-assessment, use of objective metrics, the
execution of blinding, the quality of the randomization, the use of mood induction and
the role of the neutral control. The analysis takes the form of comprehensive charts
followed by a written presentation.
Chapter 4 is the Conclusion. This chapter includes: issues and deficiencies in the
research, suggestions on mitigating deficiencies in the research, recommendations for
future research, limitation of the study and the summary.
3
1.3 Literature Review-Historical Context This section presents an historical narrative from the roots of Applied Positive
Psychology (APP) to the development of the Broaden and Build Theory (BBT) This is
followed by a review of the development of the BBT based on specific studies that
charted its inception as a model to its full development as a central theory within APP.
It should be stated that there are other theories that have been developed with
regard to emotions, but it is not within the scope of this work to include them or to
critique the BBT. This literature review helps to place these eight studies and APP
within the broader context of social science research methods.
1.31 Positive Psychology
Although Martin Seligman popularized the term APP, the term ‘positive psychology’
was presented decades earlier in Abraham Maslow’s 1954 book, Motivation and
Personality. In this book, he uses the term ‘positive psychology’ as a chapter title,
writing:
The science of psychology has been far more successful on the negative than on the positive side. It has revealed to us much about man’s shortcomings, his illness, his sins, but little about his potentialities, his virtues, his achievable aspirations, or his full psychological height. It is as if psychology has voluntarily restricted itself to only half its rightful jurisdiction, and that, the darker, meaner half (Maslow 1954, p. 354).
Even before Maslow, ‘William James argued that in order to study optimal human
functioning thoroughly, one has to consider the subjective experience of an individual,
for that belief and others James, is considered by some to be America’s first positive
psychologist’ (Taylor 2001, p. 15).
However, PP in its current form was created on August 21, 1998, when then President
of the American Psychological Association, Martin Seligman announced:
Our mission is to utilize quality scientific research and scholarship to reorient our science and practice toward human strength …. To encourage and foster the growth of the new science and profession of positive psychology…. The creation of a new science of positive psychology can be the Manhattan Project for the social sciences (Seligman 1998, pp. 561- 562).
4
Following this pronouncement, Csikszentmihalyi & Seligman went to great lengths to
differentiate positive psychology from humanistic psychology. They criticised
humanistic psychology for its lack of scientific rigour, including ‘its use of unscientific
methodologies and its inadequate empirical foundation’ (Csikszentmihalyi & Seligman
2000, p. 7).
They went even further in their position, stating, ‘Unfortunately, humanistic
psychology did not attract much of a cumulative empirical base’ (Csikszentmihalyi &
Seligman 2000, p. 7). However, according to Froh, ’Seligman defined research in a very
limiting way, as it was based solely on positivistic methodologies; phenomenological
methods were discounted’ (Froh 2004, p. 19).
This schism between humanism and Positive Psychology was both clear and
deliberate. Seligman was going much further than a mere rebranding; he was creating
a new powerful entity—positive psychology. It would not be subsumed by pre-existing
psychology or become a branch of humanistic psychology. This was a stand-alone
edifice that would chart new horizons, unencumbered by pathologies of the past, but
only if a substantial body of sound empirical research was established. If the research
was not based on empirical studies Positive Psychology could simply be dismissed as a
fad.
The importance of Seligman to the formation of PP cannot be ignored. He was an
eminent researcher and his work in learned helplessness was instrumental in
undermining the dominance that behaviourism had on much of the research of the
time (Mair & Seligman 1976). His larger than life persona, his academic
accomplishments, his vision, his energy and his connections moved the disparate
strands of positive psychology to coalesce and develop into a coherent entity. Others
were involved, but none, at that time, seemed as integral to the process as Seligman.
In the early days, his persona permeated the very essence of PP to the point that he
appeared, at times, to be the Guru. This was most clearly seen by the hero worship of
some of his students who would breathlessly run from their classes and blog in awed
tones about what their awesome professor had just said (Ruark 2009, p. 2).
This frothy enthusiasm, highly prevalent in the early years, tainted the significance of
the science of PP. With little knowledge of the subject, but armed with access to blogs,
5
and social media such as YouTube and Facebook, the world was bombarded by a PP
that seemed little more than a series of quick fixes, lists, smiley faces and gushing
praise, all based on affirmations and being happy. National Public Radio did a special
on positive psychology. ‘Its prime example of the field was the best seller, The Secret
in which television producer Rhonda Byrne argues that everything in the universe
vibrated on a particular frequency. If people attune their thoughts to the same
frequency as… money…they will attract wealth’ (Ruark 2009, p. 2). Barbara
Fredrickson, a significant researcher and developer of the BBT lamented, ‘The curse of
working in this area is having to distinguish it from Chicken Soup for the Soul’ (Ruark
2009, p. 2).
APP ran the risk of being delegitimised through the process of Pop Psychology. Caught
up in the hype, PP was seen as a source of income instead of a gold mine of research
to help humankind. Internet sites would pop up with the ubiquitous lists on seven
steps to happiness or seven steps to resiliency, all accomplished in seven minutes.
Fortunately, APP is much more than emoticons and instant happiness. At its inception,
APP was seen as an extension of the current-day psychology: taking up were it left off.
While most current psychology was defined within limits articulated in Freud’s maxim:
‘Much will be gained if we succeed in transforming…hysterical misery into common
unhappiness’ (Breuer & Freud 1895, p. 305). APP saw its role in helping clients move
from common unhappiness to some aspect of happiness. APP was the upside to the
downside of regular psychology.
Although current psychology has much to recommend it and had in developing new
therapeutic processes, improved many people’s lives, it was limited to minimizing
misery. Unfortunately, the ability to relieve emotional distress is not the same as
creating well-being. APP critiqued current psychology as it was, for the most part,
based on pathologies and the medical model. Clients were patients who had
deficiencies that needed to be fixed. They were victims seeking to be cured. Gillham
and Seligman, stated, ‘these fixations on pathology led to psychology becoming
“victimology”; instead of viewing individuals as proactive, creative, self-determined
beings, psychologist viewed individuals as passive individuals subjected to external
forces’ (Gillham & Seligman 1999, pp. 163-164).
6
An excellent example of client as victim was presented in a biography on Franklin
Eleanor Roosevelt. In analysing Roosevelt’s motives for helping minorities, the author
presents it as a means to compensate ‘for her father’s alcoholism and mother’s
narcissism’. This negative perspective dismissed any idea of virtuous behaviour or
strength of character and presented ‘a pessimistic view of human nature’ (Gillham &
Seligman 1999, pp. 168).
One of the criticisms levelled at APP was that it was unrealistic and overly optimistic to
the point of being ‘Pollyannaish’, yet APP’s research in depression indicated a firm
grasp on reality. The increase in depression was, to a certain extent, a function of
increased awareness and testing. However, APP saw the rise in depression as a result
of the inability of individuals to cope with failure and unrealised expectations. These
two issues were a function of the unremitting, unauthentic, unbridled use of self-
esteem building. Self-esteem does not prepare students for the inevitable failures in
life or for unrealised expectations that had no bases in reality. When girls were told
they could do anything, and classrooms have posters plastered all over the walls about
how special one is: failure to meet expectations became almost inevitable. The
inability to deal with failure and the resultant lack of resilience, created by unrealistic
self-esteem building, resulted in significant hardship. Failure inevitably happens,
disappointment follows, and for some depression is the result (Gillham & Seligman
1999, pp. 163-173).
One of the greatest strengths of APP is its focus on being an applied social science. In
that regard, it has researched and developed programs that help people use their
strengths to develop resilience in order to prevent depression from developing and in
using an individual’s strengths to create greater well-being. APP has researched and
developed ‘undoing’ processes that help individuals reduce unpleasant emotions and
increase pleasant emotions. Much of the work done with the BBT is focussed on the
nature of emotions and this ‘undoing’ process (Cohen & Fredrickson 2010, pp. 355-
366).
Given the relative newness of APP (1998) there have been teething problems,
including the wholesale dismissal of humanistic and health psychology, and the
confusing terminology occasionally used in APP. One of the most unfortunate choices
was to use the word ‘happiness’ when the term actually meant was ‘well-being’. Well-
7
being is a much more comprehensive term than happiness. Using this happiness
moniker was to invite derision.
In addition, the terms ‘negative emotions’ and ‘positive emotions’ are confusing.
Simply stated, people may see negative emotions as making them feel bad and
positive emotions making them feel good. However, the terms negative and positive
emotions in the Broaden and Build theory and in APP research in emotions, have very
specific attributes.
Another ongoing problem is the actual definition of PP. Some definitions of PP are lists
of characteristics, others refer to the ancients and use terms such as ‘hedonics
(pleasure) and ‘eudaemonia’ (well-being), and others use their own personal
definition. To add to the confusion, Seligman published the book Flourishing in 2011,
and changed his original definition of PP. In this dissertation, the definition of PP is as
follows: ‘Positive Psychology is the scientific study of optimal human functioning’
(Linely 2006, p. 8).
Given the newness of PP and the fear of it becoming nothing more than glib pop-
psychology Seligman made it abundantly clear that the randomly controlled study was
the way to develop a sound empirical base. Not only did this methodology separate
APP from most other forms of psychology, especially humanist, it also reflected his
positivistic background and his desire/need for positive psychology to appear to be a
hard science. (Linley 2006, p. 11).
1.32 The Development of the Broaden and Build Theory
The BBT was included in this study because it was used as a limiting filter in the
search, and more importantly, it was integral to all the selected studies. The impact of
the Broaden and Build Theory is significant. Its growth in importance is clearly
illustrated in a two Google searches. The first search was from 1957-1997 (the year
before Frederickson published her first article on broadens and build, and a year
before Seligman formally announced the formation of APP), using the term ‘broaden
and build’ the search generated two journal articles: both miscategorised. The second
search, using the same term for the years 1998-2013 generated 3830 journal articles.
It is also interesting to see how the scientific method on building theories was
reflected in the development of the BBT. Fredrickson did not create this theory in a
vacuum. At her disposal were the previous research studies of many scientists that
8
had explored events relating to emotions. Fredrickson also worked with a dedicated
peer group that did much to add to this theory. In order to show the development of
the BBT it is important to show the current research (1998-present day) and
acknowledge the past research (before 1998). This section looks at five studies by
Fredrickson that track the development of the BBT, as well as past works of others
that Fredrickson incorporated into her studies. It is this body of historical data that
allowed her to create her theory in such a relatively short time.
The first time the term, ‘Broaden and Build’ was introduced it was presented as a
model, not a theory. The Broaden and Build Model was unveiled in Fredrickson’s
paper (sole author), What Good are Positive Emotions? (1998b). Prior to this was
another paper authored by Fredrickson and B. Levenson entitled, Positive Emotions
Speed Recovery from the Cardiovascular Sequelae of Negative Emotions, (1998a). This
earlier paper, 1998a had been rejected numerous times over a span of seven years
(Fredrickson 2013, p. 5).
Possible reasons for this long delay included that negative emotions dominated the
research. Negative emotions dominated because they tended to reflect the medical
model of psychology. Fear, anger and depression highlighted deficiencies that needed
to be addressed. Positive emotions such as joy and gratitude were not seen as
deficiencies, so there was little interest in studying something that was not a
deficiency. The other issue was that most of the scientists studying emotions would
try to shoehorn positive emotions into the existing model designed for negative
emotions. When it was seen that positive emotions did not fit that model, there was
little interest in pursuing this research any further (Fredrickson 1998b, pp. 301-303).
It is clear that negative emotions elicit actions. Anger is associated with attack, fear
with flight and sexual jealousy with domestic violence. All of these negative emotions
trigger the autonomic nervous system (ANS). On the other hand, joy may be
associated with a smile but then so can a number of other positive emotions and none
of the positive emotions triggered the ANS. So if positive emotions did not elicit
actions or trigger the ANS, what was their function (Fredrickson 1998b, p. 300)?
Ekman (1994, pp. 15-19) suggested that it was possible that the structure of the
negative emotion model was not suitable. Given the different characteristics of the
emotions, a different model needed to be developed for positive emotions. Levenson
9
(1998) supported this position: ‘The evolutionary meaning of positive emotions such
as happiness might be to function as efficient “undoer” of states of ANS’ (Levenson
1988, p. 25). The hypothesis was that negative emotions elicit actions, which then
activate the ANS in order to accommodate the requirements of the emotion. When
the elicit action is completed positive emotions will help adapt the ANS back to its
prior state before the action was elicited.
The study done by Fredrickson and B. Levenson (1998a) indicated that positive
emotions did calm the ANS. The experiment consisted of showing a fear-eliciting film
to 60 participants. All were connected to a series of measures of the ANS. After the
induction of fear, each group was shown a second film meant to induce a second
emotion, which was either, contentment, amusement, neutral or sadness. ‘Compared
to Ss [subjects] who viewed the neutral or sad secondary films, those who viewed the
positive films exhibited more rapid returns to pre-film levels of cardiovascular
activation’(Fredrickson & Levenson 1998a, p. 191).
Their paper had suggested an additional answer to the question on what good are
positive emotions. The authors believed ‘it is not a good idea to assume the adaptive
value of positive emotions is simply isomorphic to the adaptive value of the negative
emotions’ (Fredrickson & Levenson 1998a, p. 200).
Many of Alice Isen’s studies (1984, 1985, 1987, 1992) related to aspects of the BBT.
They dealt with the relationship of positive emotions and cognition, research design
for inducing emotions using a neutral control and studies in the area of positive
emotions and cognition. These studies were instrumental in Frederickson’s
development of the BBT. Isen’s study looked at the influence of effect on selection.
The study indicated, ‘positive words create a more complex cognitive context by
cueing more associates relative to individuals in a neutral control condition’ (Isen et al.
1985, p. 1414). This reflects an important axiom of the BBT, that positive affect
broadens cognition. Other studies using word association in Mednick’s Remote
Associates Test indicated that individuals in a positive affect scored better on this task
than people in a neutral or negative emotional affect (Isen et al. 1987, pp. 145-154).
Another of her studies indicated that people experiencing positive affect used more
inclusive categories in problem solving (Isen et al. 1992, pp. 65-78). Isen’s work gave
Fredrickson the empirical support that underwrote the broaden aspect of the BBT.
10
The build part of Fredrickson’s model refers to the building of personal resources. As
with the broaden part of her model, previous researchers guided her work and helped
shape her hypotheses. An example of building resources is demonstrated in the rough
and tumble of children at play. The importance of this interaction around play as
explained by ethnologists Boulton and Smith was a universal activity for both humans
and animals. Rough and tumble play helped strengthen and increase agility, and being
able to avoid danger, a key personal resource (Boulton &Smith, 1992, cited in
Fredrickson 1998b, p. 309).
Not only do positive emotions build physical resources they also enhance intellectual
resources. In studies relating to intrinsic interest in learning it has been shown that
intrinsic learning ‘is linked to greater conceptual understanding, higher levels of
academic achievement, lower drop-out rates, and greater psychological adjustment’
(Deci et al. 1991, p. 337). Studies with school age children asked to recall a joyous
event or a time of contentment were able to master a math problem more quickly
than the neutral or negative groups (Bryan 1991, p. 493). In an experiment by
Carnevale and Isen, (1986) they determined that:
The use of positive affect may be a very useful tactic that may help negotiators discover optimal solutions. These results also imply that basic cognitive processes, such as those recently found to be facilitated by positive affect…those involved in categorization, similarity judgments, associations, and creative problem solving-are important (Carnevale & Isen 1986, p. 12).
At this stage in 1998, there was some indication that positive emotions had a broaden
and build function but ‘this empirical base remains thin, and direct tests of hypothesis
derived from the broaden-and-build model of positive emotions are needed,
especially experimental tests that can establish causality’ (Fredrickson 1998b, pp. 311-
312).
These two papers by Fredrickson and Frederickson and Levenson, published in 1998,
illustrate the dynamics of the scientific method. Published studies by numerous
scientists working at different labs and in different decades leave a body of knowledge
for others to pursue. Fredrickson was able to coalesce different strands of research
into a model. In the next few years and after significant empirical study from
Fredrickson’s lab and other labs, the model evolved into the Broaden and Build
Theory.
11
Part of building a theory is the need to replicate the studies, Fredrickson et al. (2000),
replicated the study of 1998a with The Undoing Effect of Positive Emotion (2000). This
study not only confirmed the previous findings but also added to the body of
knowledge. The study had a different demographic mix, so the undoing effect could be
expanded to a larger population. Also in the original study, the positive emotions may
have replaced the negative emotions. However, in this study, the neutral and positive
groups showed no change in ANS arousal. Only the group that had negative emotional
arousal showed ANS activation; therefore, the undoing process of positive emotions is
only elicited in the presence of negative emotional arousal (Frederickson 2000, pp.
237-255).
In the introduction of the paper entitled, The Role of Positive Emotions in Positive
Psychology: The Broaden-and-Build Theory of Positive Emotions, Fredrickson (2001)
summarises the BBT. ‘The Broaden-and-Build Theory posits that experiences of
positive motions broaden people’s momentary thought-action repertoires, which in
turn serves to build their enduring personal resources, ranging from physical and
intellectual resources to social and psychological resources’ (Fredrickson 2001, p. 218).
Fredrickson’s research not only indicates that positive emotions create resources she
identifies one of the main concepts in Positive Psychology—resilience— as one of
these resources. It was found that ‘more resilient participants exhibited significantly
faster returns to baseline levels of cardiovascular activation…this difference in time
needed to achieve cardiovascular recovery was mediated by differences in positive
emotions’ (Fredrickson 2001, p. 216). Fredrickson makes the observation that the use
of positive emotions ‘could build psychological resilience, not just reflect it’. She goes
on to suggest, ‘The broadened attention and cognition triggered by earlier experiences
of positive emotion should facilitate coping with adversity, and this improved coping
should predict future experiences of positive emotion’ (Fredrickson 2001, p. 216-217).
The paper, Positive Emotions Trigger Upward Spirals Toward Emotional Well-Being, by
Fredrickson and Joiner, was cited as an unpublished manuscript in 2000 and referred
to in the 2001 article. However, for the purposes of this work the paper is cited as
2002, to reflect the year it was published.
In this study, Fredrickson and Joiner (2002) introduce another key component to the
BBT: the spiral. This dynamic is a key factor that encompasses a major aspect of
12
emotion and mood regulation. Studies conducted by Basso (1996) and Derryberry and
Tucker (1994), indicated that participants who had negative emotions such as anxiety
and or depression had an attentional focus that was narrowed. This spiralling process
reflected research on depression by Peterson and Seligman (1984). Their studies
indicated a downward spiral due to constant negative emotions that created a
constant narrowing of view with increasing pessimism and eventually resulting in a
chain reaction that over time created clinical depression (Basso, 1996; Derryberry &
Tucker, 1994; Peterson & Seligman, 1984; cited in Fredrickson & Joiner 2002, pp. 172-
175).
Frederickson and Joiner posed the question: ‘Does the broaden-and-build theory of
positive emotions predict that positive emotions broaden the scope of attention and
cognition, and, by consequence, initiate upward spirals toward increasing emotional
well-being’ (Fredrickson & Joiner 2002, p. 175)? The study used the Positive and
Negative Affect Schedule (PANAS) to measure positive and negative affect. They then
used the Coping Responses Inventory (CRI) to ascertain what coping measures
participants used to deal with the most challenging problem they faced during the
past year. The scale consisted of eight measures: cognitive analysis, positive
reframing, seeking alternative rewards, problem solving, emotional discharge,
avoidance, acceptance/resignation and seeking counsel. At T1, information was taken
and PANAS and CRI were completed. Five weeks later at T2, the same measures were
repeated. The study found that Positive Affect (PA) and not Negative Affect (NA) led to
enhanced broad-minded coping. It was also found that at T1 broad-minded coping
significantly predicted increased PA. ‘Taken together, this sequence of findings
indicates that PA predicts itself partly via broad-minded coping, and that broad-
minded coping predicts partly via PA, PA and broad-minded coping, then, mutually
build on one another’ (Fredrickson & Joiner 2002, p. 174).
These five published studies present the development of the BBT. The article: What
Good are Positive Emotions? has been cited in 2319 publications, and the article: The
Role of Positive Emotions in Positive Psychology: The Broaden and Build Theory of
Positive Emotions, has been cited 3281 times. Since the inception of the BBT, over
three thousand academic articles have been published on this theory and a substantial
amount of that research is devoted to creating a sound empirical body of evidence for
the BBT.
13
1.33 Emotions
In order to understand the tools for measuring emotions it is necessary to understand
the nature and characteristics of emotions. The following is a brief overview on the
nature and characteristics of emotions. It is not exhaustive and is limited to some of
the basic characteristics of emotion, as it relates to the BBT.
There are over ninety definitions for the term ‘emotions’ (Larsen & Fredrickson 1999,
p. 41). This number reflects the complexity of the subject; therefore, it is not
surprising that there is a disagreement on the definition of this term. In order to
eliminate confusion scientist usually present their working definition of emotions
when writing up their study. It is hoped that these working theories will, overtime, be
refined and developed into a coherent theory.
Although research into emotions is now prevalent in psychology, this was not always
the case. In the mid-1900s, the dominance of behaviourism in the social sciences
limited research done on emotion. Behaviourists such as Skinner dismissed emotions
‘as irrelevant and misleading epiphenomena and derided those who studied them as
mentalists’ (cited by Fredrickson 2013, p. 4).
It was not until the 1980s that the study of emotions was recognized as a subspecialty.
Even then, the study of emotions focussed on such emotions as fear or anger. The
idea of studying joy or gratitude would have seemed to be frivolous, and funding
would have been difficult to get. Since the advent of APP, the study of emotions has
grown exponentially. Studying positive emotions such as joy and gratitude is
generating significant findings and having a direct positive influence on peoples’ lives.
Emotions have a number of characteristics that are important to understand. The first
is duration and intensity. The duration of emotions can be fleeting and hard to
measure, or it can last for a significant time. Intensity of emotions can be considered
on the bases of such measures as blood pressure and heart rate.
Another characteristic of emotions is that they may be discrete or diffuse. Discrete
emotions refer ‘to classical emotion theory which emphasizes the importance of…
specific types of affect, such as anxiety, hostility and depression’ (Watson & Clark
1997, p. 268). However, the trend in emotional theory is to see emotions as diffuse.
Diffuse (also known as dimensional) looks at emotions in which there are sets of
associated emotions such as fear and guilt or love and joy.
14
Emotional characteristics also include the concepts of state and trait: ‘Long lasting
responses that may not be associated with anything specific, are seen as traits,
whereas, a state is associated with something specific and is shorter in duration’
(Lucas et al. 2009, p. 142). Although state is an emotion and trait is a mood, common
usage is to use the term emotion for both state and trait and differentiate between
the two when it is required (Lucas et al. 2009, p. 142).
1.34 Self-Evaluation: Likert scale and Measurement
The Likert scale and the Likert type scale are ubiquitous in social science research.
There are three reasons for its prevalence: the first is that it is cheap, the second is it is
simple to present and the third reason is it is the only way, without equipment, to
measure directly, aspects of attitude and emotion.
There are some concerns with this form of testing. First, there is the problem of
terminology. When searching for the distinction between the Likert and Likert type
there was little clear agreement. One researcher posits that it is as Likert presented it
in his 1932 paper, which is a horizontal scale. Yet if one checks Likert’s 1932 paper, he
use four examples, two are horizontal scales and the other two are multiple-choice
questions. The formats are different, but all the constituent parts were incorporated in
both examples (Likert 1932, p. 236).
Secondly, there are structural issues with the Lickert scale that influences how
answers are selected. If a Lickert scale has an odd number of selections then the
middle choice will be the one that many may select in order to avoid stress or thinking
(central variance). If there are an even number of choices in a Likert scale the
individual might be forced to give an answer that does not allow for neutrality when
that would be the participant’s preference (Larsen & Prizmic-Larsen 2004, p. 346). On
the other hand, an individual may take the extreme position: reporting the highest or
lowest number on the scale (extreme responding). This has been seen with children as
they tend to take extreme positions, but as they grow older, they take the extreme
position less often (Chambers & Johnson 2000, p. 33).
Thirdly, is the question about the type of data presented and what type of statistical
tests can be performed from that data? On the one hand, the argument is that the
data is ordinal which means that the numbers of the scale usually 1-5 or 1-7 are an
order of ranking but not a measure of distance. Therefore, one cannot say the
difference in value between 1 and 2 is the same as the difference in value between 4
15
and 5. Given this limitation only certain statistical manipulations can be done including
calculating the median the mode and the range. Inferential techniques would include
only non-parametric methods (Jamieson 2000, p. 1212).
There are those who argue: ‘The weight of empirical evidence…clearly supports the
view and position that Likert scales…produce interval data, particularly if the scale
meets the standard psychometric rule-of-thumb criterion of comprising at least eight
reasonably related items’ (Carifio & Perla 2008, p. 1150). In that case, more powerful
parametric processes, such as factor analysis, structural equation models and ANOVA,
can be used. On the other hand, the non-parametric techniques related to ordinal
data have a very limited ability as the ‘rank methods like Spearman’s Rho, Kruskal-
Wallis, appear frozen in time and are used rarely’ (Norman 2010, p. 627). This is a
significant issue, as the Likert scale is used heavily in the social sciences, and the use of
the more robust parametric measures allows for analysis that is more powerful.
Carifio and Perla argue that there is significant empirical evidence to show that
parametric analysis can be used with a collection of Likert items, but not with
individual Likert items (Carifio & Perla 2008, pp. 1150-1151). This situation remains
unresolved.
1.35 Issues in self-reporting scales and measurement
As with many measurement tools there are concerns that may limit the effectiveness
of its data. Some of the more obvious issues for self-reporting would include the
ability to complete the test properly, due to age or language restrictions. Cultural
issues, other than language, would include experiencing different emotions and
responding to emotions in a way that may be culturally different from expected. For
example, reacting in anger may be culturally inappropriate and the need to answer in
a conciliatory manner may be culturally appropriate. Individuals may react in a way
that reflects a traumatic experience in their past or a need to repress certain emotions
for their well-being.
Certain patterns of answering questions may be formed and the pattern may influence
which answer the participant selects (non-content variance). Test repetition of
emotion elicitation and measurement may create a pattern of answers, which do not
change from one test to the next (stereotypic responding) (Larsen & Prizmic-Larsen
2006, p. 346).
16
Another form of non-content variance would be social desirability, which is a frequent
issue in the use of questionnaires. In this situation, a participant answers the
questions on the bases of making a positive impression, rather than the actual
emotion that they were experiencing. This may be a conscious choice or an
unconscious process. In addition, it can be based on the participant presenting what
she/he would like to be, rather than on what he or she actually is.
The period for assessment also may have a significant impact on what the scale is
measuring. If the event is precipitated by a specific stimulus and measured over a
short time, then an emotion will most likely be measured. If measurement relates to
long-lasting feelings that tend to be unrelated to specific objects and events then a
mood is most likely being measured (Mauss 2009, p. 212). Another aspect of time is
the dynamic nature of emotions. If one’s anger is precipitated because of a particular
event, usually the anger will subside over time, but will still have an effect on mood.
However, the intensity of the emotion will most likely have lessened within 24 hours.
The longer the time interval is from the event to the measurement, the more
problematic the result. In addition, because of the length of time involved before
measurement of an emotion, it is possible that a person’s underlying personality will
begin to shape and change the emotional response (Lucas et al. 2009, p. 145).
Self-report is also open to measurement reactivity if the process of measurement
shapes the outcome. For example, a participant completing a self-evaluation may view
a set of emotions in the scale and on seeing this list they might, in some form,
reprioritize or replace the actual emotions originally elicited.
It is important to understand: ‘Reliability is not an intrinsic property of an instrument
but rather a quality that varies depending on the characteristics of the instrument in
combination with the particular sample in which it is used’ (Messick 1995, p. 741). So
when using self-referencing scales do measurements properties change based on the
composition of the sample? It would appear from the limited research that is the case.
Youngstrom and Green’s paper (2003) concludes: ‘It is imperative that future studies
not assume that the published reliabilities based on college samples (or other
middle/high SES groups) will generalize to low-income samples’. Not only are there
questions about Socioeconomic Status (SES) but also with age, as older individuals’
emotions tend to be more moderate than young peoples’ emotions. There are also
17
differences in gender (Youngstrom & Green 2003, pp. 280-281). This issue will be
discussed further in Ogden and Lo’s study (2011) in Chapter 3, Chart 3.2.
1.36 Blinding in RCTs
Blinding refers to ‘keeping trial participants, investigators (usually health-care
providers), or assessors (those collecting outcome data) unaware of the assigned
intervention, so that they will not be influenced by that knowledge’ (Schultz & Grimes
2002b, p. 696). The purpose of blinding in clinical trials is to reduce effect bias, ‘an
overview of empirical studies of bias in meta-analyses found that beneficial effects on
subjective outcomes were, on average, 25% lower in trials labelled “double blind”
compared with similar trials that were not “double blind”’(Hrobjartsson et al. 2009, p.
967). In addition,’knowledge of the intervention received, and perceptions of that
treatment, can affect the psychological or physical responses of the participants.
Knowledge of treatment allocation can also affect compliance and retention of trial
participants’ (Schultz & Grimes 2002b, p. 696). Participants are not the only individuals
who may react to prior knowledge; it is also possible that investigators in the study
may be influenced this could include trial designers, participant enrollers,
randomization implementers, health care providers, intervention counsellors and data
collectors. ‘If investigators are not blinded, their attitudes for or against an
intervention can be directly transferred to participants’ (Shultz & Grimes 2002b, p.
696). This transfer of attitudes could be of particular concern when dealing with the
study and measurement of emotions.
Given that the correct use of blinding reduces effect bias and the lack of its correct use
increases effect bias, it is imperative that the individuals structuring the research
understand when and who should be blinded in the study. It is also important for the
value of the research that the researcher correctly report not only that there was
blinding but who was blinded.
Unfortunately, it becomes apparent in both meaning and application that blinding is
neither well understood nor accurately reported. The confusion with regard to
blinding occurs in text books and with instructors, which is then reflected in their
studies. In one survey responders provided 15 different meanings for the term ‘double
blind’, and believed that their definition was the most commonly used definition
(Haahr & Hrobjartsson 2006, p. 362). In another study on blinding terminology the
authors found, ‘83 studies reported as double-blind, 41 made no mention of which
18
groups were blinded, 29 studies identified 1 group as blinded, 11 studies identified 2
groups as blinded, 1 study identified 3 groups as blinded, and 1 study identified 4
groups as blinded’ (Devereaux et al. 2001, p. 2000).
Another frustration is created when blinding has been incorporated in the
researchdesign, stated in the protocol and followed in the study, but is not written up
in the final paper. This not only unnecessarily limits the value of the study, but when
other researchers might wish to include their work in a systematic review, the review
authors have to spend time trying to track down the primary author and/or the
protocol. In fact a useful study may even be discarded because it does not conform to
the parameters set in the systematic review because blinding was not reported.
1.37 Randomization in RCTs
Stolberg states:
Randomised controlled trials are quantitative, comparative, controlled experiments in which a group of investigators studies two or more interventions by administering them to a group of individuals who have been randomly assigned to receive each intervention (Stolberg 2004, p. 1539).
Although random allocation is a common occurrence in clinical trials , much like
blindness, it seems to be both misunderstood and misreported. This undermines the
very core of clinical research. In fact, the combined issues relating to blindness
coupled with the issues relating to randomization would suggest that many findings
using RCT are suspect (Schulz 1995, p. 410).
As with the section on blindness, it is necessary to place the research in positive
psychology within the framework of current research in clinical trials and the use of
randomization. Clinical trials using random allocation were first developed in the
1940s under the guidance of Sir Austin Bradford Hill and became commonplace in the
1970s in creating evidence based research (Stolberg 2004, p. 1539).
Although random allocation may seem to be quite straight forward, a lot can go wrong
between the development of the study design and the completion of the study. Some
areas of concern include, study design, sequence generation, form of randomisation
allocation selected, sample composition, allocation concealment, blindness and report
write up.
19
Random allocation is the only method that can suggest causality, as it has the ability to
deal with unknown bias in samples. The use of randomisation is to create groups that
have similar characteristics. Other processes can eliminate known differences or they
can be taken into account because they are known. However, random allocation,
because it is random, is able to randomise the unknown differences between/amongst
groups. In so doing random allocation will create similar groups; therefore, the
difference between control and experimental groups will not be based on the
confounding of differences in sampling. The importance of this process is that the
difference between the groups will, if all else is done correctly, be due to the
intervention which would indicate causality. In addition, the strength of RCTs is that
the numerical results can be statistically manipulated in order to ascertain the
significance of the results (Shultz & Grimes 2002a, p. 516).
There are numerous methods of randomization, from free randomization, which is as
simple as flipping a coin, to more complex systems using stratification and multiple
blinded block sizes. Although flipping a coin might seem like the most elegant solution,
there are problems with this type of randomisation. Unfortunately, this could create
very unevenly sized groups, and if there was a small sample size this would limit the
ability to see small differences between sample groups. It is also possible that the
researchers could see an imbalance in the groups and decide to change the outcome
of the toss (Shultz & Grimes & 2002a, p. 517).
In order to set and control sample size, restricted types of random allocation can be
used if the number of participants is limited. This can be done by blocking, but the
blocking itself must be blinded in order not to expose the allocation process. The
random allocation rule can be employed when a predetermined sample size is desired.
An example of this is to have 25 black and 25 white balls drawn blind from a hat; in
that way sample size can be controlled (Schultz & Grimes 2002a, p. 517).
It is unfortunate, given the importance of randomization to the integrity of the study
and to those researchers who wish to build on their research, especially those
developing systematic reviews and meta-analysis, the reporting of information about
random allocation in scientific studies is problematic.
Reports and quotes by authorities in the field indicate a significant problem in the
understanding and write up of the randomization and allocation process. In one study
20
on allocation concealment, 96 of 102 trials had unclear allocation concealment. In 39
of these trials, ‘neither the protocol nor the publication provided any attempt to
conceal allocation’ (Pidel 2005, p. 1.) To compound this issue, ‘empirical studies have
shown that publication of trials in which allocation of concealment is unclear or
inadequate are associated with, on average, 20-30% exaggeration of the treatment
affect…compared with trials of the same interventions with adequate concealment’
(Pidel 2005, p. 2). ). Here the lack of blindness or the lack of reporting blindness or of
listing those who are blind in the study severely limits the usefulness of the study.
Not only is concealment of allocation limited the actual randomisation process is also
questionable. In many cases, there is not enough information to be able to know how
randomization was developed. ‘In one study of 206 studies 129 or 63% of the reports
did not specify the randomization process which would then hide not only the process
but the value of the process’ (Schultz & Grimes 2002a, p. 516).
Some of this confusion comes from using the oxymoron quasi-random, which has
been compared to quasi-pregnant. These studies are in fact non-random studies and
calling them quasi-random is obfuscation. Some researchers use systematic allocation
thinking that the process is random, when that is not the case: ‘Groups using such
methods as groups based on preintervention tests, last names, or admission date, are
systematic and not random.’ The problem is amplified as, ‘ systematic allocation
usually precludes adequate concealment, since it results in previous knowledge of
treatment assignment among those who recruit participants to the trial” (Schultz &
Grimes 2002a, p. 516).
1.38 Mood Induction
When studying the development of mood induction, a logical place to begin is with
Emmet Velten’s study titled: A Laboratory Task for Induction of Mood States. The
following is a summary of his study (1968, pp. 473-482). Velten’s experiment is the
foundation for mood induction processes as used in Positive Psychology research and
was developed from his PhD dissertation. Obviously since this report was published in
1968, there have been numerous studies dealing with mood induction. It is interesting
to see the scientific method at work, supporting some hypothesise and challenging
others as gradual changes and refinements helped develop and improve mood
induction research. Velten’s study is important, as it was the most popular induction
21
process from the late 1960s until the mid-1990s. It is the foundation for the
methodology of current practice in mood induction.
His study of 100 female college students included an initial assessment of suitability
for mood induction. Seventeen of the women were removed, as they scored low on
the hypnotic induction scale. The remaining 83 participants were given a pretest for
baseline data. These tests consisted of an ability to make decisions and a perceptual
ambiguity test.
Participants were randomly assigned to one of five groups. The first three groups
consisted of elation (EL), depression (DE), and neutral (NU). For these three groups the
induction process consisted of reading silently and then aloud their induction
statements, which consisted of 60 self-reference statements. All started with neutral
statements and gradually moved to elation, depression or remaining neutral
statements. An example of an EL statement was: ‘This is great. I really do feel good—I
am elated about things.’ For DE statements an example was: ‘Every now and then I
feel so tired and gloomy that I’d rather just sit then do anything.’ For NU an example
was, ‘Utah is the Beehive State’ (Velten 1968, p. 475).
The fourth and fifth groups respectively were the elated demand characteristics (EDC)
group and the depressive demand characteristics (DDC) group. The two groups ‘serve
as controls for conscious role-playing or unconscious influence of the S. by the obvious
demand characteristics’. During their induction process they received the same
statements as their non-demand group as well as a few synonyms for elation or
depression and read the instructions to, ‘believe that way’. These two groups were
also shown samples of the words used in the EL or DE groups. The last instruction was,
‘to always remember to act as if I were (elated) or (depressed)’ (Velten 1968, p. 475).
Seven tests were given to the participants to see if induction had taken place. These
tests were, writing speed, distance approximation, decision time, perceptual
ambiguity, word association, (MAACL) Multiple Affect Adjective Check List, and
Spontaneous Verbalization. This was followed by a post-experimental questionnaire
‘designed to indicate awareness of effects induced by the treatment and awareness of
E’s hypotheses regarding the critical measures’. Results were that EL, NU and DE
groups performed as anticipated, but EDC and DDC groups’ dominant emotion was
22
limited. In two groups subjects tended to overplay there subjective mood but
underperformed on the objective tests (Velten 1998, 476).
This difference in behaviour to the mood induction was a demand bias, as the
participants were acting in a manner to please the experimenters. The nature of
demand effect and its influence on mood induction continues to be a challenge.
Building on Velten’s methods, Izzard presented a more complex issue on mood
induction. He questioned the idea that Velten’s study was actually dealing with
discrete emotions. It was Izzard’s contention that these were not discrete emotions
but associated emotions. ‘One emotion can almost instantaneously elicit another
emotion that amplifies, attenuates, inhibits or interacts with the original emotional
experience’ (Izzard 1972, p. 177). The issue of discrete versus associated emotions is
an ongoing debate but the tendency currently is to state in each study the author’s
definition of emotions.
Polivy’s 1981 study, employing Velten’s methodology, studied the question of discrete
and associated emotions following up on Izzard’s contention that the emotions are
not discrete. Her results ‘indicate that inducing elation or depression causes
significant increases in tranquillity or anxiety and goodwill or hostility’ (Polivy 1981, p.
808). Another study based on fear of a painful electric shock not only indicated fear
but also depression and hostility (Polivy 1981, p. 811). She concludes her paper
stating: ‘We should be aware that our investigations of “an emotion” are most
probably investigations of several stimulus emotions’ (Polivy 1981, p. 816).
Frost and Green (1982) then question the duration of induced emotions and the
apparent lack of debriefing and removal of latent emotional residue: ‘(1) Do subjects
leave VMIP (Velten Mood Induction Process) experiments with residual negative
moods? And (2) Do VIMP effects last long enough for investigators to rely on them’
(Frost & Green 1982, p. 342)? Frost’s concern highlighted the fact that his research
indicated that 13 of 17 studies using VMIP did not indicate ‘any strategies to remove
negative moods’ (Frost 1982, p. 342). Clearly, there would be some concern of letting
participants leave the study in a depressive state due to an induction experiment.
Other discussions centered on the nature of the participant and their individual
susceptibility to depressive mood induction. This individual susceptibility using the
Velten induction method was a significant concern as, ‘between 30-50% of subjects
23
failed to respond to it’ (Clark 1985, p. 45). This low rate raised ‘doubts about the
generalizability of any results that may be obtained using it’ (Clark 1985, p 45). This
percentage of failed inductions was too high. It was becoming apparent that the
Velten method had a failure rate that was higher than many other forms of induction.
Another concern was on how to administer the induction process. During the
induction phase in VMIP, should the induction just be on the sixty phrases or the
twenty phrases and should the explicit instructions to feel the emotion deeply be
included in the instructions or should there be no mention of the emotion?
This last issue dealt with demand or the social desirability bias. The issue here was
that some participants were faking the emotion in order to satisfy the demands of the
experimenters. In addition, the participants would try to enhance the mood induction
by creating thoughts and expressions that would deliberately enhance the mood so
efficacy of the mood induction process was minimised by the demand and the
participants’ active participation. Fortunately it was becoming clear that demand bias
was easily identified, as participants who ‘fake it’ overemphasized the
expected/known behaviours, but underscored in the unknown measures associated
with that particular mood induction (Kenealy 1986, pp. 321-327).
For Blackburn et al. (1990), the viability of the Velten’s mood induction was
questionable. Blackburn’s study investigated Velten’s mood induction and the
individual differences and responses to the Velten Mood Induction Procedure.
Although this study added to the evidence that cognition was associated with
depression, it also confirmed ‘the variable and low rate of response to a mood
induction procedure’ (Blackburn et al. 1990, p. 730). Given that the variable rate of
response was based on the nature of the participants, it was clear that the reliability
of the Velten induction process was low and difficult to generalize.
Astrid Gerrards-Hesse’s (1994) article, Experimental inductions of emotional states and
their effectiveness: A review, looked at mood induction procedures from 1979-1994.
This study ranked the effectiveness of an induction process to create an effective
mood induction. The ratings of effectiveness of an elation induction were, film/story
with explicit instructions as to mood (100%), Drugs (100%), film/story no instructions
to mood (92%), and gift (83%). For depression: imagination (100%), film/story (94%),
success and failure (92%), and Velten, which includes specific mood instructions (80%)
24
(Astrid Gerrards-Hesse 1994, p. 66). In conclusion: ‘If the same procedure is to be used
for both elation and depression induction the film/story MIP should be the first
choice’. Velten’s mood induction ranked fourth in depression and did not rank for
elation. (Astrid Gerrards-Hesse et al. 1994, p. 70).
Westermann’s study, Relative effectiveness and validity of mood induction procedures:
a meta-analysis focuses on 250 induction studies, points out that, ‘effects are
especially large when subjects are explicitly instructed to enter the specified
mood….Effects tend to be smaller when demand characteristics are controlled or
subjects are not informed about the purpose of the experiments’ (Westermann et al.
1996, p. 557). This study concludes by stating: ‘If demand effects are considered to be
a serious threat to the validity of an experiment, MIPs that do not inform subjects may
be preferable’ (Westermann et al. 1996, p. 578). As the Velten mood induction was
not effective for elation, and it required stating specific self-referent statements about
mood the demand bias could not be removed resulting in lower validity. Because of
this, more effective MIP’s replaced the VMIP.
25
26
2. Methods
In this chapter the two methodologies are presented. The first methodology is the
systematic search strategy based on the Cochrane Handbook. This methodology is
implemented in this chapter in order to select the appropriate studies. The second
methodology critical analysis will be used in Chapter 3.
2.1 Methodology for Search StrategyThe search strategy for the studies was based on the Cochrane Handbook for
Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. A systematic
review has clearly stipulated inclusion and exclusion criteria combined with an explicit
search strategy, as well as systematic structures including: flow charts, data extraction
charts, assessment of RCTs charts, selection criteria charts, and key search indicators
(Higgins & Green 2011).
2.2 Methodology for AnalysisOnce the studies were selected, the information was analysed using the second
methodology: a critical analysis approach. A critical analysis methodology is: ‘ the
intellectually disciplined process of actively and skilfully conceptualising, applying,
analysing, synthesising and/or evaluating information gathered from, or generalised
by, observation, experience, reflection, reasoning or communication, as a guide to
belief or action [or argument]’ (Scriven & Paul 2003, p. 3).
2.3 Search StrategyA number of exploratory searches, including Data Bases such as MedLine, Psych Net
and search engine Google Scholar were used in order to help refine search parameters
and terms. Google Scholar was found to be more comprehensive in its search and in
some circumstances more flexible than the other databases. Search parameters for
this study included articles from peer reviewed journals, translated from any language
into English, from 1998-2011.
The selection of 1998 was when Frederickson published her first paper on the
‘broaden and build’ concept. In addition, this was the year that Martin Seligman
became President of the APA and announced, at their annual general meeting the
new focus to be placed on APP. The final search was initiated May 5th 2012. The end
of 2011 was the cut-off date as numerous previous searches had been completed with
that cut-off date.
27
Only RCTs were selected, as causality was important (Seligman 2011, p. 32). These
RCTs had to be about positive emotions within the BBT, given that this study was
concerned with research on the BBT. An additional criterion was to have at least one
outcome measure that was not based on self-assessment. In this work, outcome
measures refer to an outcome that can be used to support or refute a hypothesis and
the outcome must be in measurable units. The unit of measurement must be specific
and concrete, not a subjective construct. This layer of measurable objective outcomes
added depth to the studies, as most research in this area is self-assessment. In order
to clarify information on blinding and randomization all authors were contacted.
Searching Google Scholar, the phrase ‘Broaden and Build’ was mandatory as were the
terms: ‘random’, ‘controlled’, ‘experiment’ and ’induction’. These terms were chosen,
as they were integral to the process of creating an RCT study. Broaden and Build was
the theory and random, controlled and experiment were integral parts of a RCT.
Induction was included as it was used as an intervention and as an active control.
Other terms relating to RCT including ‘blind’ and ‘double blind’ were tried but found to
be too limiting. It would have been useful to add ‘objective metrics’ to limit the
search; however, this term was not found in any of the studies and a manual search
was carried out.
2.4 Final Search Strategy
2.41 Rationale
Google Scholar search returned 153 studies. The term ‘Broaden and Build’ was
included in the literature review section of many studies, but was not integral to the
paper. This led to the elimination of 54 studies. Two duplications in the search were
also eliminated. Books were clearly labelled in the search results and all 30 were
eliminated, as they did not meet the criteria of being in a peer-reviewed journal.
Forty, theses, dissertations and working papers were also eliminated for the same
reason. Three studies had functional magnetic resonance imaging f(MRI) outcomes
and were not included, as they did not have objective outcomes as defined in this
study. Twenty-four papers made it into the final selection. At this stage, a hand search
using a basic data extraction form was utilized and the objective outcome criteria
were included. In this exclusion: one study did not deal with BBT, two were not
experiments, one did not have a control group, and twelve studies did not have
objective outcomes. The total studies remaining were eight RCTs with objective
measures. After receiving the emails from the researchers’ it was found the one study
28
that was reported as random was in fact a systematic process. This study was not
eliminated, as it was a good example of the issues associated with appropriate
reporting of randomization.
2.42 Flow chart-Selection-filters
A diagram that shows the filters for inclusion and exclusion
Figure 1 Filter Selection
Literature Search
Search engine Google Scholar
Terms: ‘Broaden and Build’, ‘random, controlled’, ‘experiment’, ‘induction’, 1998-2011, No language limits, Peer reviewed journals.
Articles screened on bases of
title and abstract
Excluded (n=129)
Duplicates=2
fMRI Studies=3
Books=30
Not in a Peer Reviewed Journal=40
Off Topic= 54
Total Search Results (n= 153)
Included (n=24)
Manuscript review and
application of inclusion criteria
Included (n=8 papers)
(n=8 studies)
Excluded (n=16)
Not Broaden and Build=1
Not experiment= 2
No random control=1
Subjective outcomes= 12
29
2.43 Author Journal Abstract
Authors and Journal AbstractAnderson, E. et al. (2011)Journal of experimental social psychology, 47(4), 856-860.
This experiment, demonstrates that the affective state of a perceiver influences the contents of visual awareness.
Biss, R. & Hasher, L. (2011) Emotion, 11(6), 1474-1478.
In this study the authors examine the impact of emotional state on irrelevant information.
Falkenstern, M. et al., (2009)The Journal of Positive Psychology, 4(5), 365-371.
The goal of the current study was to extend the undoing hypothesis to the cognitive domain to determine whether positive emotion can undo the cognitive effects of negative emotion.
Fredrickson, B., et al., (2000) Motivation and Emotion, 24(4), 237-258.
Positive emotions are hypothesized to undo the cardiovascular after effects of negative emotions.
Lee, G. (2011)Personality and Individual Differences, 50(5), 717-722.
This study examined relationships between positive affect and jumping to conclusions (JTC) in delusional thinking.
Sakaki, M. & Kazuhisa, N. (2011) Cognitive, Affective, & Behavioural Neuroscience, 11(4), 526-540.
To study the effects of the brief viewing of positive and negative pictures on reaction times for understanding solutions to insight problems.
Shmueli, D. & Judith, P. (2012) Psychological Addiction Behaviour, 26, 157-161.
Positive affect elicited with a video was able to counteract the detrimental effects of self-control depletion on smoking behaviour.
Wadlinger, H. & Isaacowitz, D. (2006)Motivation and Emotion, 30(1), 87-99.
The goal of the study was to evaluate whether a positive mood induction would broaden the visual attention of individuals using images that varied on level of emotional valence from very negative to neutral to very positive.
Figure 2 Author Journal Abstract
30
2.43 Chart: Study Selection Criteria form
For all study selection forms please see Appendix Study Selection Criteria Form
Reviewer Andrew SmithDate April 15-16 2012Details of PublicationAuthor: Kuhbander, Christof, et al.Title: Effects of Mood on the speed of conscious perception: behavioural and electrophysiological evidenceJournal: SCAN (2009) 4, 286-293
Filters If yes, continue If no, remove study
Published 1998-2011Yes
Peer reviewed journal Yes
Has positive affect/effect/emotion as part of the intervention Yes
Has a control Yes
Has random assignment Yes
At least one outcome must be an objective outcome
No
One or more outcomes supports or contradicts the Broaden and Build theory
Figure 3 Study Selection
31
2.44 Chart: Sample of Validity Assessment Chart
For all study selection forms please see Appendix All RCT studies were appraised for study quality after the initial selection process.Figure 4 Validity
Random sequence generation Low risk Stated that computer randomized participants into one of two conditions
Allocation concealment Low risk This was a web-based experiment so controlled allocation.
Blinding of participants and personnel
Low risk This was a web based experiment so not likely that participants aware of intervention. As limited interaction with personnel little chance of influencing participants
Blinding of outcome assessment Low risk There was no mention of blinding but given the form of the experiment, outcomes unlikely to be influenced. In addition, outcome, number of questions asked, was an objective outcome.
Incomplete outcome data Low risk No missing outcome dataSelective reporting Unclear Insufficient information to make a
judgementOther bias Unclear Sample is one of convenience consisting of
young participants average age 22.8 S.D. (6.0) attending university.
As these forms, combined with the information in the studies did little to clarify the
validity of the studies. The primary researchers were contacted by email. They were
asked the following two questions: ‘What was the process of randomization’? and
‘Were the participants and experimenters blind’?
With the information from the emails, there were now three levels of information
relating to randomization and blinding. These levels of information are: what was
written in the study (using a key word search), what could be deduced from
Cochrane’s protocol and charts, and finally the actual email replies from the scientists.
All three levels presented different evidence.
32
2.5 Criteria used for a Summary of Study ChartFor all summary of study charts please see Appendix.
Identification and Citation
Location
Abstract
Study Design
Hypothesis
Sample Composition
Control
Blind
Induction
Intervention
Outcomes
2.6 Methodological IssuesProblems of searching using key terms. Certain terms, although part of the criteria for
the search, were not included in the searchable information. For example, searching
the phrase ‘objective outcomes’ returned no hits, and putting the words in individually
resulted in numerous papers with the word ‘objective’ or ‘outcomes’ but never
‘objective outcomes’. The search term ‘blind’ was not effective as the concept of
‘blinding’ or ‘masking’ was, in most cases, either non-existent or presumed but not
stated in the studies. The term ‘Positive Psychology’ was too broad until other filters
were applied and the term ‘Broaden and Build’ was too broad as many studies
included this theory in their literature search of their experiment without using it
further.
Another issue was the use of the term random control. This was a straightforward
search item; however, in reading the studies, filling in the Cochrane classification chart
and then contacting the researchers there appeared to be a number of discrepancies.
The use of the term ‘control’ was unclear. In many of these studies, the control group
was given an induction designed to create a neutral emotional state. There is one
study that had both a neutral induced control as well as a no action taken control.
Another had only a no action taken control and no neutral control. In some cases, the
control was simply categorised as a neutral induction.
33
3. Findings and Results
In this chapter a critical analysis methodology is used to analyse the quality of
research of each of the eight studies. This critical analysis is based on seven specific
criteria developed from the literature review. Each chart covers one of the following
categories:
Chart 3.1: Sample composition (population) Chart 3.2: The use of self-assessment measure (study integrity) Chart 3.3: The use of objective measurement (outcomes) Chart 3.4: The nature of emotions and blinding (study integrity) Chart 3.5: The quality of the RCTs (study design) Chart 3.6: The use of induction (intervention) Chart 3.7: The use of neutral induction as control (control)
Each chart contains information on all eight studies. Although there is a significant amount of information included in each chart, it is descriptive information. The critical analysis of this study is presented directly following each of the charts.
34
Chart 3.1 Critical Analysis: Sample CompositionLead Author
Sample Composition
Anderson 50 undergraduates from Boston College.35/15 female/malemean age 20.64 (range 17-32)sample of convenience
Biss 64 students from the University of Toronto (Canada)43/21 female /malemean age 19.4 (SD 2.7; range 17-33) 12.9 yrs. of educ. (SD 1.2)sample of convenience
Falkenstern 86 undergraduates from a liberal arts college (U.S.)41/21 female/male, 72 Caucasian 84%mean age 19 (range 18-22),sample of convenience
Fredrickson Sample 1 Sample 295 university students (U.S.) 75 university students (U.S.)48/47 female/male 38/37 female/male71 European 75% 58 European 77%sample of convenience sample of convenience
Lee 189 students from University of Manchester (U.K.)140/49 female/male), 140 white 74%mean age 22.8 (range 18-55, SD 6.0)sample of convenience
Sakaki 31 undergraduates from U of Tokyo (Japan)19/12 female/malemean age 20.2 (SD 0.83)sample of convenience
Shmueli 193 smokers from the San Francisco area (U.S.) local ads118/67/8 female/male/transgender, 69 white 26%, 85 African 44%heterogeneous sample from the San Francisco area
Wadlinger 58 undergrads from Psych course at Brandeis University (U.S.)30/28 male/femalemean age 19.69 (range 18-26, SD 2.06)sample of convenience
Figure 5 Sample Composition
When looking at dominant themes in sampling of these studies there were some clear
areas of concern. Of the eight studies, five were from the United States. The other
countries were Canada, the United Kingdom and Japan. All studies except Shmeuli’s
were in lab settings. All studies, except Shmeuli’s, used convenience sampling, were in
a university setting, and were predominately comprised of undergraduate students
enrolled in a psychology class. All studies, except Sakaki’s had more females in the
studies. The largest imbalance in a single study was a sample size of 189 students, 140
females and 49 males. This gender imbalance reflects, to an extent, the changing
demographics in enrolment, as nearly 72 per cent of new PhD and PsyDs entering
35
psychology are women (Cynkar 2007, p. 1). The total count from all studies totalled
607 females and 339 males. From this information, the dominant participant would
be an eighteen to twenty-two year old middle class, white female psychology
undergraduate university student living in the United States.
This very narrow demographic band skews outcomes. Differences in Socioeconomic
Status (SES), age, ethnicity, and gender create different outcomes even when using
identical testing methods. These differences in results are amplified when the testing
has a subjective or self-evaluation component, as life experience shapes an
individual’s point of view. This will be readily apparent in the study by Jane Ogden and
Jessica Lo (2011).
Choosing convenience sampling was both cost and time efficient, but as a non-
probability study it excluded random selection and severely limited generalization to a
larger population. Because of these choices, the studies had a limited external validity.
Also, all but Shmueli’s study were clinical studies, so there were concerns relating to
ecological validity.
However, as in any scientific study there have to be trade-offs relating to time, money
and external and internal validity. Although not explicitly stated, the majority of these
studies were clinical trials developed in order to add information to Frederickson’s
Broaden and Build theory. For example in Wadlinger and Derek’s (2010) paper the
abstract opens with the following statement: ‘ In an attempt to investigate the impact
of positive emotions on visual attention within the context of Fredrickson’s (1998)
broaden-and-build model, eye tracking was used in two studies to measure visual
attentional preferences of college student’s’ (Wadlinger & Derek 2006, p. 87). Another
study concludes: ‘The current study provides preliminary support for the application
of the undoing hypothesis (Fredrickson & Levenson, 1988) to the cognitive domain’
(Falkenstern et al. 2009, p. 370). This is repeated in another study: ‘This relative
positive congruence effect is consistent with the research in the attention literature
showing that positive affect biases attention toward rewarding information
presumably in the service of approach-related decision making and behaviour
(Frederickson, 2001)’ (Anderson et al. 2011, p. 859). Given this focus on developing
studies related to the BBT, external validity was not the priority.
36
Unfortunately, many of the papers did not acknowledge the issue of having a limiting
homogeneous grouping. In some papers the only time the composition of the sample
was stated was in the methods section under the heading ‘Participants’. Group
composition was never stated in any of the titles so you have an article titled: Positive
mood broadens visual attention to positive stimuli. More accurately the title for this
paper should have been: Positive mood broadens visual attention to positive stimuli in
undergraduate students. This lack of clarity was reflected in the discussion part of
many of the papers. Some studies simply referred to participants or made no mention
of the participants at all. Sahaki’s paper did not even mention the term participants
but concludes: ‘These results suggest that the effects of emotional pictures on
understanding solutions to insight problems do not depend on similar mechanisms as
do the effects of emotional pictures on memory’ (Sakaki & Niki 2011, p. 535). The
sample used in this study was thirty-one undergraduate students. Other studies did
acknowledge that their samples were problematic. Two authors went further than
that. Falkenstern’s (2009) paper stated:
Although the current study provides tentative support for the undoing hypothesis on cognitive processing, it utilized a homogeneous convenience sample of general psychology students at a small liberal arts college. Therefore, it is difficult to determine how the results of this study would generalize to different populations (Falkenstern et al. 2009, p. 370).
Lee’s study added another limitation. He wrote: ‘Replication in real world settings
would be critical to supporting the ecological validity of present results…. From this
point of view; the following suggestions are presented speculatively’ (Lee et al. 2011,
p. 721). Clearly some of the authors realized the need to acknowledge the limitations
of their findings, other authors may have known the limitations but failed to make
them clear.
37
Chart 3.2 Critical Analysis: Self-Assessment Lead Author
Subjective self-assessment
Anderson Nine-point Likert scale: After each induction condition T1,T2,T3,T4, participants reported their core affective state in terms of valence and arousal using a 9-point scale
Biss Likert Scale: 4 self-ratings for mood and arousalFalkenstern Emotion Report Form (ERF): Participants completed an Emotion Report Form (ERF)
developed by Fredrickson. The ERF contained nine different emotions including amusement, anger, anxiety, contentment, disgust, fear, happiness, sadness, and serenity. Participants rated the amount of each emotion they felt in response to the films they watched on a 9-point scale ranging from 0 (none) to 8 (a great deal). The ERF was used after two different inductions. Format of ERF is a self- reporting Lickert-like scale test.
Fredrickson Emotion Report Form (ERF): At the end of the experiment participants completed one (ERF) to describe how they felt while preparing their speech and a second ERF to describe how they felt viewing the randomly assigned film clip. Format of ERF is a self- reporting Lickert like scale test. Tests were done at the end of the experiment.
Lee The Positive and Negative Affect Schedule (PANAS): Consists of twenty adjectives describing current affect. There are ten adjectives forming a positive affect scale (e.g., “interested” and “proud”) and ten for a negative affect scale (e.g., “nervous” and “jittery”). Each item is rated 1–5, reflecting the degree to which that adjective is felt (1 = “very slightly or not at all”, 5 = “extremelyThe Peters Delusion Inventory (PDI-21): consists of 21 items measuring delusional–proneness in non-clinical populations. Each item is answered yes/no with, three subscales rated: distress (1 = “not distressing at all”, 5 = “very distressing”), preoccupation (1 = “hardly ever think about it”, 5 = “think about it all the time”) and conviction (1 = “don’t believe it’s true”, 5 = “believe it’s absolutely true”).Two single-item Likert scales Self-evaluation of performance/task difficulty consists of two single-item Likert scales that measured how well participants thought they had performed compared to other people, and how difficult they had found the JTC and creativity tasks. Both items were scored 0–10 (0 = “performed much worse than others/very easy”, 10 = “performed much better than others/very difficult”). These evaluations were included on two occasions (t1 and t2.)
Sakaki No self-assessment.Shmueli Participants completed a set of pre-experimental measures.
Immediately following the cue exposure task, participants completed a manipulation check assessing perceived interest, difficulty, and pleasantness of the task, two items each.Type of test unknown.Participants then underwent the affect induction portion of the study using one of two procedures—video or writing exercise—allowing for examination of the impact of different forms of positive affect induction.
WadlingerStudy 1
The Positive and Negative Affect Schedule (PANAS): The following measures tested for positive and negative affect, anxiety, optimism, and depressive symptoms: the PANAS inventory was used as a self-report measure of trait positive and negative affect.CES-D depression scale. Depression was measured with the CES-D depression scaleSpeilberger State/Trait Anxiety Scales (STAI-S and STAI-T): Anxiety was measured12-item Life Orientation Test: Measured optimism
Figure 6 Reliance on Self-assessment (Information in this chart taken directly from the written studies)
38
The Reliance on Self-Assessment chart lists and explains the self-assessment measures
used in the eight studies. The Likert scale was used by Anderson, Bliss and Lee. In all
three cases, the test measurements preformed, fell within the capabilities of the Likert
scale single item measure of mood in Anderson and Bliss’s studies and the single
measure of performance in Lee’s study. Sakaki’s study only had an objective
measurement and no self-assessment. Shmueli described a number of
preexperimental questionnaires and after the mood induction relied on an objective
measure to study the effect. This was an interesting study design, but the inability to
measure the mood induction created issues when it was demonstrated in the results
that aspects of the mood induction had been ineffective.
The self-reporting Likert like scale test- Emotion Report Form (ERF) was used by both
Falkenstern and Fredrickson. Both Lee and Wadlinger used the Positive and Negative
Affect Schedule (PANAS.) Lee also used the Peters Delusion Inventory (PD-21) and
Wadlinger used the CES-D depression scale, the Spielberger State/Trait Anxiety Scales
(STAI-S and STAI-T) and the 12-item Life Orientation Test. All of these tests were multi-
scale self-assessment tests.
However, a paper written by Jane Ogden and Jessica Lo, How meaningful are data
from Likert scales? An evaluation of how ratings are made and the role of the response
shift in the socially disadvantaged, admirably highlights the problems of using Likert
scale on different SES groups. Ogden and Lo’s (2011) study includes both quantitative
data from the scales, and qualitative responses initiated by open questions. The
qualitative responses were explanations that the participants gave in order to explain
their quantitative selection. This study focussed on quality of life and includes three
samples: a homeless group N=75, first year students living away from home N=301
and individuals that live in the town, N=72. Questions were the same for all groups
and focused on mood and health ‘right now’ and satisfaction in the ‘last few days’. The
authors used a five point scale ranging from: not at all (1), to rarely (2), somewhat (3),
fairly (4), very much (5). Open questions related to a description of their
accommodation and their eating behaviour. The qualitative data were summarized
into two groups: accommodation, mood and social interaction, and eating drinking
and feeling healthy (Ogden & Lo 2011, pp. 350-361).
39
In summation, the homeless quantitative data ranked higher than either the students
or individuals living in the town. The people living rough reported feeling less tired and
more healthy that the others and in a better mood than the others, in relation to
frustration, loneliness and anger (Ogden & Lo 2011, p. 356). However, their qualitative
answers reflected problems with sleeping and a host of medical and health issues
including a wide range of psychological problems such as anxiety, depression and
problems with drug and alcohol addiction. ‘Such inconsistencies between different
forms of data may reflect measurement error and the psychometric limitations of the
Likert scales’ (Ogden & Lo 2011, p. 359).
These discrepancies reflect a difference in the interpretation of the focus of the
question. Students discussed their accommodation on the bases of social aspects,
whereas the homeless focused on the physical aspects of their lodging. On the subject
of food both townies and students focused on the regularity and healthiness of their
eating, the homeless responded with descriptions such as: ‘I never get hungry…but I
haven’t eaten for days’. ‘I don’t really need to eat much’. ‘I don’t get hungry…I haven’t
eaten since Friday’ (Ogden & Lo 2011, p. 359).
The conclusion of this study presents three issues that may have influenced decision
making. The first issue is ‘the frame of reference’ which reflects how an individual’s
answers are shaped on the bases of what is important to that individual. The second is
‘the different methods of comparison’. In this case the homeless group were using
within-subject comparisons and not between subject comparisons. For the homeless
they would refer to situations that had been worse for them. The third aspect is
‘different time frames’. In the case of the homeless their positive ranking of their
experience could be that: ‘Participants who have been deprived of an aspect of life for
a sustained period of time, habituation rather than sensitization may be the response’.
In the conclusion the authors suggest Likert scales ‘have their limitations and that data
derived from their use should be understood within the broader context of
participants’ decision making processes’ (Ogden & Lo 2011, p. 360).
40
Chart 3.3 Critical Analysis: Objective OutcomesLead Author
Objective Outcomes
Anderson Participants completed a series of binocular rivalry trials in which a face (smiling, scowling, or neutral) was presented to one eye and a house to the other. The percepts ‘competed’ for dominance in visual consciousness.
Biss Participants ability to recall and complete primed word fragmentsFalkenstern The dependent variable was response latency as measured by the time in
milliseconds that the participant took to press the button after the letter appeared on the screen.
Fredrickson Cardiovascular Measures(1) Heart rate (HR (ECG)(2) Finger pulse amplitude (FPA)(3) Pulse transmission times to the finger (PTF)(4) Pulse transmission time to the ear (PTE )(5) diastolic blood pressure (DBP(6) systolic blood pressure (SBP)
Lee The key variable was the number of comments requested before a decision was made.
Sakaki Participants were faster to report that they understood the solutions following positive images, and were slower to report it following negative images.
Shmueli Participants were given a 10-min recess. Whether or not participants smoked during the recess, assessed by self-report and biochemical verification, served as the primary dependent variable.
Wadlinger The eye tracker specifically records the duration and location of the participant’s left eye sixty times per second. The total viewing time can be calculated. The data derived from the eye tracker was recorded as percent gaze fixation times
Figure 7 Objective measures
In the early stages of research in APP many studies were correlational studies using
subjective self-evaluation measures. These studies were useful in helping to develop
potential testable hypothesis. From this process then developed research that
included RCTs, but as in most psychological studies, much of the data was still
generated from self-evaluations. As has been discussed, self-evaluations are useful but
they have limitations.
The inclusion of objective metrics in these eight studies is very useful as the major
focus on objective metrics is to add a process of measurement that does not use self-
assessment in the outcome. In this work, outcome measures refer to an outcome that
can be used to support or refute a hypothesis and the outcome must be in measurable
units. In the following discussion, the unit of measurement and the method of
obtaining that measurement are presented.
41
In Anderson’s study (2011) the unit of measurement was time in milliseconds.
Participants were given a mood induction and then completed a series of binocular
rivalry trials in which a face (smiling, scowling, or neutral) was presented to one eye
and a house to the other eye. When subjects were induced into a pleasant affective
state, the faces dominated over the house and the participants spent more time
looking at the pleasant faces. When participants were induced into an unpleasant
affective state the pictures of the faces dominated over the house and the participants
spent more time looking at all pictures but scowling faces dominated.
In Bliss’s study (2011) the objective metric was the number of word fragments
remembered. After mood induction, participants were shown a series of pictures and
overlapping words. They were instructed to focus on the pictures. Following a timed
interval, they were then given a word completion task on the words that were in the
original pictures. Those with a positive induction remembered more words than those
in the neutral induction.
In Falkenstern’s study (2009) the unit of measurement was time in milliseconds. When
a letter appeared on the screen, participants were to press that letter on the
keyboard. The time from seeing the letter to pressing the keyboard was measured.
After a negative induction to all participants, the letter recognition test was repeated.
Then three randomly selected groups were given a positive, a negative or a neutral
induction. Participants were then given a letter identification task for the final time.
Negative induction decreased reaction times.
Frederickson’s study (2000) used six measurements related to the cardiovascular
system to measure time in seconds to recover to base line. An initial stress induction
was given to all participants. Then a neutral, positive or negative mood induction was
given to each group. The time for the individual to return to base line was measured.
Those in the positive mood induction were the fastest to return to base line, followed
by the neutral group and then the negative group.
In Lee’s study (2011), the unit of measurement was the total number of questions that
each participant asked before reaching a decision. Mood induction was induced by
positive or neutral feedback. The task was to reach a conclusion by asking questions.
Participants given the positive induction asked more questions than the neutral group.
42
In Sakai’s study (2011), the unit of measure was the time to understand a riddle.
Participants were told a riddle. Then they were shown a positive, neutral or negative
image, after which they were given the solution. They were then asked to push a
button when they understood the riddle. Participants were faster in understanding
the riddle when shown a positive image.
In Shmueli’s study (2012), the objective measure was the number of participants (all
smokers) who went out for a cigarette during a break in the study. This study dealt
with the self-regulation and depletion. A depletion process took place as participants
were exposed to either cookies or vegetables but were not allowed to eat them. After
that, participants were then given a positive or neutral induction. After that there was
a break. The positive video group, who were depleted because of the cookies, were
nearly as likely not to smoke as the non-depletion vegetable group.
In Wadlinger’s study (2006), the unit of measurements consisted of percentage of
time in gaze fixation on peripheral images and the number of saccades (fixing on a
point) participants made per slide. Units of measurement were generated from an eye
tracker. The positive induction group fixated more on peripheral images with high
positive valence and made significantly more saccades than the neutral group.
43
Chart 3.4 Critical Analysis: BlindingLead Author
Blinding
Anderson Participants were completely blind to condition. The staff were not blind to the condition because as part of the "induce positive" condition, participants were presented with a small gift immediately prior to eye-tracking (thus staff knew the condition, but were instructed not to deviate from the prepared scripts). Control participants received the gift immediately after tracking. Participants were not debriefed until after the study and from informal conversations afterward none connected that the candy gift related to their attention patterns (that we were tracking).
Biss Participants were blind to the purpose of the experiment and to the fact that we were manipulating mood, however, as our mood manipulation conditions involved listening to emotional music and viewing emotional pictures, it is always possible that some figured out which condition they were in (although none reported this to the experimenter).Staff were not blind to condition
Falkenstern Participants obviously knew what video they saw, but they were not told that the purpose of the video was to induce a specific mood and were not aware that there were other conditions seeing a different video until they were debriefed.Staff were not blind to the condition, but all of the data were collected on a computer, so there really wasn’t room for bias in data collection.
Fredrickson No informationLee Participants and staff were blind to assignment condition.Sakaki Participants were run by my undergraduate RAs that did not know the
hypothesis.Shmueli The groups were not blind to the depletion condition, in the sense that they
knew they would be asked to resist eating either vegetables or sweets, and were then asked to resist one of those foods. However, they were not aware of the intent of this manipulation (i.e., to deplete self-control). The groups were unaware of the replenishment manipulation, or that there was more than one condition (i.e., positive vs. neutral).
Wadlinger No informationFigure 8 Blinding (Information in this chart is verbatim from email correspondence with researchers)
Using the results from a computer generated key word search none of the studies
mentioned blinding or masking. As there was no information contained in the studies
about blinding, Cochrane’s protocol and chart were used in order to create a
systematic approach for gathering information. These charts are in the appendix of
this work. Unfortunately, in this case, the charts were little more than a systematic
way of guessing and did not provide information that was reliable. The only viable
process was to communicate directly with the researchers.
Most researchers kindly responded to the author’s email requesting clarification about
blinding. This information is contained in Chart 3.4. The actual copies of the emails are
listed in the Appendix. To summarise: Anderson, Biss, Folkenstern, Lee, and Shmuli
44
stated that the participants were blind to the condition or purpose. Anderson, and
Biss, stated that none of the staff were blind as they were responsible for inducing the
mood. Falkenstern stated staff were not blind, but as data was collected by computer
there would not be data bias. Lee was the only one that stated that both participants
and staff were blind to assignment condition. Sakaki’s response to the question of
blinding was unclear: ‘participants were run by my undergraduate [Research
Assistants] RAs that did not know the hypothesis’.
Although when asked, all researchers’ stated that the participants were blind, it would
be much more useful if this was recorded in the study. The fact that blinding is not
reported in the actual studies, when it is actually used in the studies, is unfortunate, as
it could bias the study and limit its use in systematic reviews.
It is possible that there was less concern for blinding as the objective outcomes were
opaque to the participant. Also if the research was self-contained on a computer and
there was no direct interaction between participant and researcher, blinding would be
built into the research design. However, it should not be up to the reader to search
this out, and this would not answer the question if other areas of the study were
blind.
45
Chart 3.5 Critical Analysis: Randomization in RCTsLead Author RCTAnderson Randomization of trial order was done using the presentation
software, Eprime.Biss In our study, participants were randomly assigned to condition using a
randomized block design. The blocks were randomized using the random function in Excel.
Falkenstern We just used a “draw the numbers out of a hat” approach to ensure roughly equal numbers of participants across conditions.
Fredrickson No replyLee Randomisation was achieved by using the first letter of the person's
surname: if it began with a letter between A-M, they were assigned to the neutral condition and N-Z to the positive condition. When the first 100 participants had fully completed the study, the assignment was reversed (i.e. A-M assignment to positive condition, N-Z to neutral).
Sakaki Randomization in Study 1 was performed by random numbers generated by Excenumber l.
Shmueli Predetermined randomization order was based on an online random number generator: http://www.randomizer.org/form.htm
Wadlinger I believed we used a random online number generator (i.e. ResearchRandomizer - http://www.randomizer.org/) to assign experimental conditions.
Figure 9 Randomization in RCTs (Information in this chart is verbatim from email correspondence with researchers)
Using the results from a computer generated key word search all of the studies
mentioned some form of the term ‘random’ with the most common phrase being,
‘randomly assigned’. What was lacking was an explanation on how the randomisation
was accomplished. Without this information, the process of randomization is suspect.
In order to rectify this deficit the next course of action was to use Cochrane’s protocol
and chart in order to create a systematic approach in gathering information. These
charts are in the Appendix. Unfortunately, in this case, the charts were little more than
a systematic way of guessing and did not provide information that was reliable. The
only viable process was again to communicate directly with the researchers.
According to their emails, both Anderson and Shmueli used a computer generated
random number system and included a link to the site. Biss and Sakaki used Excel
software as a random number generator. Wadlinger used an unnamed random
number generator, and Falkenstern used ‘draw out of a hat’ in order to develop equal
groups. The use of recognized number generating software has its advantages. Not
only is it replicatable, and reliable, it also effectively limits the number of people who
would know the allocation procedure. Citing the software or the link adds another
layer of transparency. Falkenstern’s ‘draw out of a hat’ is effective when size of groups
46
needs to be balanced; however, its hands on procedure is open to manipulation of the
allocation procedure.
Lee’s study states, ‘The computer randomised participants into one of two conditions’
(Lee et al. 2011, p. 719). This appears to be at odds with his email: ‘Randomisation was
achieved by using the first letter of the person's surname’ (Appendix p. 105). It is
possible that the computer was programmed to randomise on the bases of last
names. If this was the case, what appears as a randomisation process in the study
appears to be a systematic process in the email. ‘Groups using such methods as
groups based on pre-intervention tests, last names, or admission date, are systematic
and not random’ (Schultz & Grimes 2002a, p. 516). In addition, using alphabetization
of the participant’s last name makes the allocation process highly visible and open to
manipulation.
47
Chart 3.6 Critical Analysis: InductionLead Author Induction
TypeInductionProcess
AndersonWhat you feel influences what you see
PleasantNeutralUnpleasant
Ten pleasant images were shown for each inductionTen neutral images were shown for each inductionTen unpleasant images were shown for each induction
BissDelighted and Distracted
PositiveNeutral
Pictures and music-jazzed up version of Brandenburg ConcertoPictures and music-ambient street noise
FalkensternMood over matter
SadnessPositiveNeutralNegative
Shown to all. The ChampHumorous film clipTwo men talking film clipFriend’s death movie clip
FredricksonThe undoing effect
Stress
ContentmentAmusementNeutralSadness
Presented to all. Video instructions write a speech that will be evaluated.Film clip of waves breaking on a beachFilm clip of puppies playingFilm clip abstract scene of sticks piling upFilm clip from The Champ death scene
LeeThe influence of positive affect on jumping to conclusions
PositiveNeutral
False taped positive feedback was presentedFalse taped neutral feedback was presented
SakakiEffects of the brief viewing of emotional stimuli
PositiveNeutralNegativeControl
Positive picture shownNeutral picture shownNegative picture showNo picture was shown
ShmueliA test of positive effect induction for countering self-control depletion
Positive (v)Neutral (v)Positive (w)Neutral (w)
Five minute comedy videoFive minute video about building bridgesWriting and emotionally feeling the happiest time in one’s lifeWriting a detailed description of one’s room
WadlingerPositive mood broadensVisual attention
PositiveControl
Given a small bag of candy.No action (not aware of the candy)
Figure 10 Induction
48
The studies included in this particular research used various forms of induction
including: giving candy, film and video clips, still pictures, false feedback and pictures
combined with music. The emphasis in this section is on the induction process and an
analysis of the failed inductions.
In Wadlinger’s study, the induction process was reasonably straightforward. In order
to induce a positive mood a small bag of candy was given to a randomly selected
group. They were told that they had to wait until the completion of the test before
they could eat the candy. This may have been a precaution as the ingredients in
chocolate could give a chemically induced positive mood and not a cognitively induced
mood. Control did not receive any candy until after the experiment was over. Giving
candy is a straight forward way of creating a positive induction as it ‘has been found to
be an effective means of inducing state positive effect in past research’ (Carnevale &
Isen, 1986; Isen et al., 1987).
In Shmueli’s induction process one of her inductions failed. Her study included two
sets of positive or neutral inductions. The first set of inductions used a positive video,
which showed a five-minute comedy skit, or a neutral video, which showed a five-
minute video on building bridges. The use of video clips is quite common in mood
induction studies. The other two inductions were induced by writing about a time the
participant was most happy, for the positive induction, and writing a description of the
room for the neutral induction.
Shmueli’s study looked at self-control depletion and replenishment. It was thought
that the positive induction group’s self-control would be replenished, but the neutral
control would indicate no replenishment:
Participants were presented with a large plate of either tempting desserts…or raw vegetables and instructed to resist eating the food while the experimenter left the room. Participants followed a 5-min pre-recorded tape with a series of bell rings, every time the bell rang, they were to lift the plate and smell the food while thinking about the temptation involved in resisting eating the food (Shmueli 2012, p. 159).
This process is not ideal as there is no way one could ascertain if the participants
followed instructions. Some did not follow instructions, as six of the participants ate
49
some of the food during the five-minute cueing. Clearly, this group had a difficult time
with self-control. Data from this group was deleted, as it was incomplete.
The findings of the study indicated that in the groups that received the positive
induction of a comedy sketch video, only 10.5 per cent of the dessert group went out
for a cigarette and of those in the vegetable group, 0 per cent went for a cigarette
break. In the groups that received the positive induction of writing about their
happiest event, 69.7 per cent of the dessert group went out for a cigarette and those
in the vegetable group, 68.8 per cent went for a cigarette break. This was a higher rate
than even the neutral groups with 65 per cent for desserts and 50 per cent for the
vegetables, going for a cigarette (Shmueli 2012, p. 161).
One explanation for the failure of the writing induction was that having to write a
paper about an event was stressful and stress increased the desire for a cigarette.
Another issue could be that watching a video is a passive exercise whereas writing is
an active process. In addition, it is possible that there were members in the group that
did not like to write or felt that what they wrote might be studied, which could cause
embarrassment for them.
Sakaki and Anderson used images for induction. Both selected their images from the
International Affective Picture System (IAPS; Lang, Bradley, & Cuthbert, 1997, 2008).
Anderson used pleasant and unpleasant affect. Sakaki’s inductions consisted of a
positive, negative, neutral and control. Standardized images were rated for valence
and arousal.
Although Bliss used images based on valence ratings in his study, following the same
protocol as Anderson and Sakaki, he added a sound track to the showing of the
pictures. For the positive induction he used ‘a jazzed up version of Bach’s Brandenburg
Concerto No.3’ (Bliss 2011, p. 1474), which had been used by Rowe (2007, p. 387). For
the neutral induction he added ambient street noise. There was no supporting history
of using this soundtrack. Bliss gave no explanation on why he added the sound track,
although one would assume he thought it enhanced the induction process. It would
have been helpful if Bliss had gone into this in more detail. In using two processes of
induction for the same mood at the same time, is it not possible that instead of
reinforcing the induction it could interfere with the induction?
50
Lee used feedback as the inductive process. For the positive induction he gave a
recorded positive feedback message, for his neutral induction he used the same
process but the message was neutral.
Falkenstern’s study used movie/film clips. An initial induction mood was played to all
participants. This general induction was then used as a base line and then other
inductions were introduced. From T1 to T2 speed of word recognition was faster for all
groups. From T2 to T3 both the negative and neutral groups the speed in word
recognition was faster compared to T1 to T2; however, for the positive induction
group their T2 and T3 times were the same. This was not expected.
Fortunately, Falkenstern spends sometime on this issue. She presents two
possibilities. First is that although the positive induction was statistically significant
‘the ratings of the intended emotions were only slightly above or at the midpoint on
the response scale’ (Falkenstern 2009, p. 369). The other factor relates to the
sadness induction process, which did induce sadness but also induced anger and fear
which the positive induction did not contain: ‘It is possible that the presence of anger,
fear or disgust, is responsible for the faster reaction times associated with the
negative affect condition’ (Falkenstern 2009, p. 370).
51
Chart 3.7 Critical Analysis: Neutral Induction as ControlLead Author
Neutral Induction as Control
Anderson Ten neutral images were shown using International Affective Picture System (IAPS) based on normative ratings
Biss Neutral images-( IAPS selected) and music-ambient street noiseCavanagh Neutral aquatic film clip shown to all.
Neutral induction clip of Gosford Park for neutral group. All film clips had been previously evaluated for specific mood induction.
Falkenstern Neutral two men talking from All the President’s Men film clip. Specific film clips selected as they had been used in previous experiments
Fredrickson Neutral film clip abstract scene of coloured sticks piling up.All film clips were played without sound.
Lee Neutral false, taped feedback was presented. Tone was pleasant and the participant was told they outperformed 50% of the public.
Sakaki Control no picture shown.Neutral pictures IAPS selected of mundane items such as mushrooms shown. Within-participants study.Only study with both a control and neutral film clips
Shmueli Neutral (v) Five minute video about building bridges.Neutral (w) Writing a detailed description of one’s room.
Wadlinger Control, no action taken.Figure 11 Induction control
In clinical trials ‘the randomized, placebo controlled trial is the gold standard of clinical
research’ (Freedman 1996, p. 243). This particular form of orthodoxy has, until
recently, been a mandated requirement for clinical trials. Without the use of a placebo
control, experimenters would jeopardize their research. However, this orthodoxy of
the placebo has been challenged on both moral and statistical grounds in that an
active control relating to best practice should be utilized as the participants are then
not denied treatment (Freedman 1996, p. 243).
There are a number of types and categories of control. The largest category is the
‘active’ control in which the participants are given an intervention leading to
treatment that is different from the experimental group. There is the ‘best available
practice’ control when participants are given the best available treatment while the
experimental group is treated with the experimental procedure. There is a ‘do
nothing’ control in which one might just sit for ten minutes looking at a blank screen.
Clearly there are different methods of control that reflect different situations and
requirements. So where does neutral control fit into this research design? A neutral
control which is commonly used in studies on emotions is a process that creates an
active control. The reason it is called a neutral control is to define its purpose, which is
to induce a temporary neutral emotional state.
52
The argument for using a neutral control group is presented in two articles entitled:
Do Positive Emotions Facilitate Recovery More Efficiently than ‘Nothing’? (Fredrickson
& Levenson 1998); and a follow up article; The Undoing Effect of Positive Emotions
(Fredrickson et al. 2000).
In these articles the authors suggested that giving the control an induction process in
exactly the same manner as the experimental group would facilitate blinding as there
would be less perceived difference in the process amongst the different groups. It was
also felt that an induction process would override the control participant’s normal
emotional process. If the participants were left to do nothing each of the individuals
in the control group would be in varying emotional states; therefore, the
experimenters would be faced with trying to measure the relatively homogeneous
experimental groups against a relatively emotionally heterogeneous control group.
Inducing a neutral emotional state to the control group would assist in creating more
uniform control measurements (Fredrickson 1998, p. 206-207) and (Fredrickson et al.
2000, p. 244).
To clarify further, the neutral induction is considered to be midpoint between
languishing and thriving. This central position is reflected in Waugh and Fredrickson’s
2008 paper, Adapting to life’s slings and arrows: Individual differences in resilience
when recovering from an anticipated threat. They used a measuring device that had
an affective scale that translated to discrete integers from (0-9) and, ‘there were labels
for “negative”, “positive”, and for each number from 0 to 9. There was no label for
‘neutral; however, neutral implicitly corresponds to a rating of about 4.5‘(Waugh
&Fredrickson 2008, p. 1037).
In Fredrickson and Branigan’s 2005 study, Positive emotions broaden the scope of
attention and thought action repertoires, this midpoint of 4.5 is categorised as
‘emotional neutrality’ and reflects an emotional report eliciting ‘zero for all nine
emotional terms, in this study.’ The study goes on to differentiate that ‘positive
emotions broaden the scope of attention relative to a neutral state’ and that
‘comparisons to neutral states provide more stringent tests of the broaden hypothesis
and less ambiguous results than do comparisons to negative states’ (Fredrickson &
Branigan 2005, pp. 8-9).
53
Before moving on to the use of the neutral state in the eight studies under
consideration, it is necessary to consider and clarify both the meaning and implication
of the neutral induction. The first is the midpoint of 4.5. The authors’ stated: ‘neutral
implicitly corresponds to a rating of about 4.5’ (Waugh &Fredrickson 2008, p. 1037).
One of the concerns with this is that the number can give a sense of precision when
there is no such precision. Emotional neutrality refers to the absence of positive
emotional characteristics and the absence of negative emotional characteristics; it
does not refer to a midpoint equidistant from the negative and positive. Also by
placing neutral in the midpoint it gives the appearance that this is a form of emotional
continuum with a positive pole, a negative pole and a neutral central point. Individuals
have the ability to hold many emotions with differing characteristics and intensity
simultaneously. One can be having a pleasant thought about one’s childhood yet at
the same be sad because one’s parents are no longer living. A continuum is not a
representative model for emotions. Admittedly the need to measure and quantify
requires simplification, but at times there is a risk of oversimplification and
misdirection.
Another area of concern is based on the positivity bias. For example,’ 86% of the 43
nations for which nationally representative samples are available, mean subjective
well-being (SWB) response was above neutral’ (Diener & Diener 1996, pp. 181-182). A
possible reason for this bias is the need to be primed in order to get through the day.
If we were not positively primed we would have difficulty being productive and in
some cases more prone to depression. The two of the most effective positive primers
are optimism and hope.
However, when a neutral induction is given this positive bias would be neutralised for
a brief time, as previously mentioned in Fredrickson and Branigan (2005). So it could
be said that in the majority of cases the neutral induction is likely to lower the positive
affect of most of the participants in a study. In many of the studies the neutral
induction group is considered the control. Presenting the control as neutral gives the
impression of the control being benign when in fact the induction has lowered the
overall positivity of the sample group. The consequence for measurement would be
that the difference between the positive group and the neutral control group would
be greater because the positivity of the neutral control group had been decreased by
induction and the positivity of the positive group had been increased by induction.
54
However, in written studies, the increase in the positivity of the positive group
compared to the neutral control group is only explained on the bases of the positive
induction and not with regard to the decrease in positivity of the neutral control. The
reduction of positivity by the neutral control needs to be stated.
Given the discussion earlier, the neutral induction was decided upon because the ‘no
film’ ‘do nothing’ control was not appropriate (Fredrickson et al. 2000, p. 244) so the
active control, the neutral induction, was utilized. This use of the neutral induction
was employed in seven of the eight studies in this paper. One of the concerns is that
the authors’ of these studies do not explicitly state that the neutral induction is an
active control. For some of the studies, it is only by reading the studies carefully that it
becomes clear that the specific emotional inductions are the interventions and the
neutral induction is the control.
Wadlinger’s study used a gift of candy to induce a positive mood. The control group
here was not an active control as they did not receive candy until after the experiment
as a thank you. Current affective state was measured using a Lickert Scale. The group
that received the candy as an induction showed a significantly higher positive state
than did the control. This was the only study that did not have a neutral induction or
other active control (Wadlinger & Isaacowitz 2006, pp. 87-99).
In Shmueli’s study there were two neutral inductions. Shmueli never refers to the
neutral inductions as controls even though she does refer to this study as a ‘controlled
randomised experiment’ the inference being that the neutral inductions were the
controls. All interventions were categorised as either neutral inductions or positive
inductions. One neutral induction was to write a description of the room, the other
neutral induction was a video on building bridges. (Dikla & Prochaska 2012, pp. 157-
161).
Sakaki’s study referred to the terms: positive, negative, neutral and control. She was
making a distinction between a control and a neutral induction. So in this situation
there was a ‘take no action’ control. That is, a picture was not shown. In the neutral
condition, neutral photographs were shown. For other studies this neutral state
would have been considered the control. The objective measure was time of reaction.
The neutral induction slowed the reaction time compared to the non-active control. In
this case a neutral induction control did not create the same result as a no action
55
control. The difference between these two was not addressed in the study, perhaps
because the sample size of 31 participants was too small to draw conclusions. The
difference might be explained by Frederickson’s understanding of the function of a
neutral control. She stated that the neutral category was designed to be midway
between depression and thriving. There is, however, still an issue which is not made
clear. Does the author view both the neutral induction and the no action control as
controls, or is she seeing the no action control as the control and the neutral induction
as an induction related to a balanced emotional state and not as a control (Sakaki, &
Niki 2011, pp. 526-540)?
In Lee’s study the use of a computer programme standardised the delivery of the
induction process. The positive induction was a preprogramed positive feedback. The
neutral induction was accomplished in the same manner, but with a neutral feedback
delivered in a neutral voice saying participants had done as well as 50% of the
population. Although the delivery was standardised the nature of the feedback was
open to wide interpretation by the participant. As most people believe they are better
than average (positivity bias) someone who was presented with a 50% feedback
would see this as a negative result and not a neutral result. If the idea of a neutral
control is to be midway between depression and thriving could this induction have
been a negative induction and not a neutral induction (Lee et al. 2011, pp. 717-722)?
Falkenstern’s study contained a neutral induction created by watching a 65-second
video clip of two men talking. The neutral induction group were significantly less
happy than the positive induction group but significantly less sad than the negative
group. There is no mention of the neutral group as being the neutral or active control
group. As with other studies, however, the write up was focussed on the results of the
positive and negative induction with no mention of the neutral induction except in
relation to the other two inductions (Falkenstern et al. 2009, pp. 365-371).
Biss’s induction used pictures. The neutral condition pictures had a valence ranging
from 4.5-5.5. Sound clips were also used in the induction. In the case of the neutral
induction the sound clip was of ambient street noise. The discussion refers to the
positive induction group and there is no mention of the neutral group being an active
control (Biss & Hasher 2011 pp. 1474-1478).
56
In Anderson’s study all participants looked at three series of pictures. Each series of
pictures represented a positive, negative or neutral affect mood induction. As in the
other studies the neutral affective mood fell between the positive and negative
affective moods. Anderson did spend some time on the neutral state, but findings
were not significant. There was no mention of the neutral state being an active control
(Anderson et al. 2011, pp. 856-860).
57
4. Conclusions
This introduction to Chapter 4 restates the intentions of this dissertation. First, the
aim of the study was to critically analyse the eight studies based on seven criteria
developed from the literature review. Second, this analysis was done in order to
answer the following research question: Are there issues of quality in the research of
these eight studies, and if so, what are they, and how can they be reduced or
eliminated? Third, the objective of this dissertation was to highlight current problems
in APP research in order to raise the quality of future APP research.
4.1 Summary of Findings and Remedial RecommendationsIn this section of the conclusion, the summary of individual findings (SF) is followed by
suggested remedial recommendations (RR). These findings and recommendations
reflect both the literature review and the critical analysis of the seven aims of the
study listed in the aims section of Chapter 1.
Chart 1: Sample composition:
(SF) The use of samples of convenience is a significant issue. As with all samples of
convenience there is a problem with external validity. But the issue goes further than
that as most of the samples are not only homogenous within samples but also
between samples, as the sample population is predominantly comprised of universitiy
students. Given that personal experience shapes a participant’s reality and many
studies rely on self-evaluations, a very limited and privileged segment of society
unknowingly shapes outcomes and disenfranchises others.
(RR) The issue of sample composition could be mitigated if researchers would move
out of the lab and out of the university. Take the research to the participants. Go to
seniors’ residences, community centres, sports arenas; advertise in Craigslist or social
media. One would still get a sample of convenience, but it would be less homogenous.
If there is a need to stay on campus then broaden the search and give a cash reward.
Soliciting Psychology students on the bases of giving them credit borders on the
incestuous. There are different SES groups on campus. Even moving outside of the
Psychology department would be useful.
58
Chart 2: The use of self-assessment measure:
(SF) It was clear from Ogden and Lo’s powerful 2011 study, that self-evaluations have
limitations. In their study the statistical measures using the Likert scale indicated that
people analysed their situation in context with their experiences. The study also
demonstrated that statistical self-evaluation differed significantly from written self-
evaluation.
(RR) There is a greater need to understand the limitations of Likert scales and self-
assessment. It is best, when using self-evaluations, to use an RCT approach going
from a single emotion Likert measure to a multi mood scale and then to an objective
measure. It would also be a good idea to use both quantitative and qualitative
measures when using an emotion scale. This would give greater context and meaning
to what appears to be an arbitrary selection of a number.
Chart 3: The use of objective measurement:
(SF) In this work, objective outcome measures refer to an outcome that can be used to
support or refute a hypothesis and the outcome must be in measurable units. The
inclusion of objective metrics in these eight studies is very useful as the major focus on
objective metrics is to add a process of measurement that does not use self-
assessment in the outcome.
(RR) More emphasis should be placed on developing study designs that have objective
outcomes.
Chart 4: The nature of emotions and blinding:
(SF) Considering the discussion in the literature review on the importance of the
correct use of blinding, the lack of reporting is cause for concern. The fact that
blinding is not mentioned in the studies is a problem of poor reporting, but it could
also indicate a lax attitude to the blinding process. If blinding is not rigorously followed
the whole study could be compromised. Even the appearance of lack of blinding limits
the possibility of the study being used by other studies. Another frustration is created
when blinding has been incorporated in the research design, stated in the protocol
and followed in the study, but is not written up in the final paper. This not only
unnecessarily limits the value of the study, but when other researchers might wish to
include the study in a systematic review, the review authors have to spend time trying
59
to track down the primary author and/or the protocol. Unfortunately, a useful study
may even be discarded because it does not conform to the search parameters set in
the systematic review because blinding was not reported. This wastes everyone’s time
and money.
(RR) Clear protocols on the use of blinding have been developed in the CONSORT 2010
Explanation and Elaboration (Moher et al., 2010) and CONSORT 2010 Statement
(Schulz, et al. 2010). Given that these protocols are constantly ignored, voluntary
compliance seems to be lacking. As many journals are peer reviewed, one of the
responsibilities of the reviewer should be to see that these protocols have been
followed. This should be a precondition for publishing.
Chart 5: The quality of the RCTs:
(SF) Unlike blinding, randomization is stated in every study. Unfortunately,
information about how the randomisation was accomplished is seriously
underreported. All researchers had to be contacted by email, as neither the study nor
the Cochrane assessment of risk of bias presented sufficient information. In addition,
some researchers have a relaxed view of what randomisation is and use systematic
methods, which may appear to be random but are not random. This use of a
systematic method can then expose the random allocation process. Without adequate
information, randomisation is suspect, as is the random allocation process and the
results of the study. This limits the studies use for other studies, and wastes
everyone’s time and money.
(RR) As with blinding there are clear protocols on the use of randomisation presented
in the CONSORT 2010, Explanation and Elaboration (Moher et al., 2010) and CONSORT
2010, Statement (Schulz et al., 2010). Given that these protocols are constantly
ignored, voluntary compliance seems to be lacking. As many journals are peer
reviewed, one of the responsibilities of the reviewer should be to see that these
protocols have been followed. This should be a precondition for publishing.
Chart 6: The use of induction:
(SF) The use of inductions in these studies was proficient. There were a few inductions
that were not effective. In one case, a neutral induction was presented at the same
60
time ambient street noise was added to the induction. There was no supporting
history of using this form of induction soundtrack and no explanation on why it was
added. In using two processes of induction for the same mood at the same time, it is
possible that instead of reinforcing the induction it interfered with the induction. In
another case because of the research design, the effectiveness of the induction was
not measured until the intervention had taken place. Results in two cases were
unreliable, as the inductions had failed.
(RR) The effectiveness of a mood induction should be measured immediately following
the induction in order to test the effectiveness of the induction. Where possible, mood
inductions should be passive such as watching a video. Using written mood inductions
can cause stress and anxiety for the participants and is a biased towards higher SES
groups such as university students. Induction methods should have a record of
accomplishment cited from previous studies. If unusual inductions are presented the
author should explain them.
Chart 7: The use of neutral induction as control:
(SF) The reporting and the use of neutral induction in some studies was not clearly
reported or explained. The use of the term ‘neutral induction’ is not synonymous with
the term ‘neutral control’ yet some studies did not make that clear. The most
significant concern, which seems to have been overlooked, is that the neutral
induction, given the positive bias of most individuals, tends to reduce the positive bias
of the induction group. This has measurement implications if the neutral induction is
the control. There has been no acknowledgement of this issue in the literature.
(RR) The implementation of a neutral control is understandable; however, there needs
to be more research and understanding of its meaning and its implications for
measurement. There also needs to be clearer reporting in the studies and a clear
identification of the control.
4.2 Recommendations for further research In general there should be more research carried out on the under twenty, the
over forty, and the socially and economically disadvantaged.
There need to be more studies done in the area of emotions and gender. The
research in this area is contradictory. Some of this type of research is agenda
driven.
61
The use of technology should be developed or enhanced in order to help
facilitate blinding and randomization.
As positive psychology is an applied science there is an expectation that research
will be focussed on studies in the field in order to increase effectiveness of
interventions.
The relationship generated by positive psychology, neuro-plasticity and neuro-
imaging is of great interest. It is challenging to see just how much potential there
is in this triad. There should be a significant amount of research done in this
area.
4.3 Limitations to the study A possible limitation in this study may have occurred in the initial search process.
One of the deciding factors for inclusion was the use of a control. It is possible
that in my selections process, I eliminated studies that had a control. This may
have happened because the control was reported as a neutral induction and not
as a control.
Another concern was that this study started off as a systematic review. This was
found to be very restrictive. A hybrid study was developed, in which much of the
search and format reflected a systematic review. This was then followed by a
critical analysis based on summary, synthesis and evaluation for the data
analysis.
I can be faulted for not doing a statistical analysis. There are a number of reasons
for my choice. There seems to be a propensity to focus on the statistics over the
written word. This might explain some of the imprecise reporting in the studies.
It should also be understood that if randomisation or blinding is not correctly
implemented, or the study design is flawed the statistics become meaningless.
Words shape meaning; statistics reflect meaning.
62
4.4 SummaryIn answering the research question, there is little doubt that there has been
significant improvement in the depth of the research in Positive Psychology. Seven of
the studies included here were RCTs incorporating multi-layered measuring protocols
using single and multi-scale self-evaluation tests and supported by objective
measurements. On the other hand, this study has shown that there are serious
deficiencies in the quality of the research especially in randomization and blinding.
However, these deficiencies are not isolated to just these eight studies. According to
the literature review in the Introduction, these types of deficiencies are prevalent in
the larger research arena. Although there are numerous protocols in place for correct
research procedures, these protocols seem to be underutilised.
The objective of this dissertation was to highlight current problems in APP research in
order to raise the quality of future APP research. This dissertation has highlighted
current problems in APP research. But will this raise the quality of future APP
research? It seems unlikely.
63
References
Alicke, M. et al., (1995) Personal contact, individuation, and the better-than-average effect. Journal of Personality and Social Psychology, 68: 804–825.
Anderson, E. et al., (2011) What you feel influences what you see: the role of affective feelings in resolving binocular rivalry. Journal of experimental social psychology, 47(4), 856-860.
Biss, R. & Hasher, L. (2011) Delighted and distracted: Positive affect increases priming for irrelevant information. Emotion, 11(6), 1474-1478.
Blackburn, I. M. et al., (1990) Individual differences and response to the Velten Mood Induction Procedure. Personality and Individual differences, 11(7), 725-731. shttp://dx.doi.org/10.1016/0191-8869(90)90258-S
Boulton, M. J. & Smith, P. K. (1992) The social nature of play fighting and play chasing: Mechanisms and strategies underlying cooperation and compromise. In: J. H. Barkow & L. Cosmides (Ed.) The adapted mind: Evolutionary psychology and the generation of culture. New York, Oxford University Press. pp. 429-444.
Bryan, T. & Bryan, J. (1991) Positive mood and math performance. Journal of Learning Disabilities, 24, 490–494.
Carifio, J. & Perla, R. (2008) Resolving the 50-year debate around using and misusing Likert scales. Medical Education, 2, 1150–1152. doi: 10.1111/j.1365-2923.2008.03172.x 1150
Carnevale, P. & Isen, A. (1986) The influence of positive affect and visual access on the discovery of integrative solutions in bilateral negotiation. Organizational Behaviour and Human Decision Processes, 37, 1–13.
Chambers, C. & Johnston, C. (2002) Developmental differences in children's use of rating scales. Journal of Paediatric Psychology, 27 (1), 27-36.
Clark, D. (1983) On the induction of depressed mood in the laboratory: Evaluation and comparison of the Velten and musical procedures. Advances in Behaviour Research and Therapy, 5(1), 27–49. http://dx.doi.org/10.1016/0146-6402(83)90014-0
Cohn, M. A. & Fredrickson, B. L. (2010) In search of durable positive psychology interventions: Predictors and consequences of long-term positive behaviour change. The Journal of Positive Psychology, 5, 355-366.
Cynkar, A. (2007) The changing gender composition of psychology. APA Monitor, 38(6), http://www.apa.org/monitor/jun07/changing.aspx
Deci, E. et al., (1991) Motivation and Education: The Self-Determination Perspective. Educational Psychologist, 26(3 & 4), 325-346.
Devereaux, P. et al., (2001) Physician interpretations and textbook definitions of blinding terminology in randomized controlled trials. JAMA, 285(15), 2000-2003.
Diener, E. & Diener, C. (1996) Most People are Happy. Psychological Science, 7(3), 181-185.
64
Ekman, P. (1994) Moods Emotions and Traits. In: P. Ekman & R. Davidson (Ed.) The Nature of Emotion: Fundamental Questions. New York, Oxford University Press. 15-19.
Falkenstern, M. et al., (2009) Mood over matter: can happiness be your undoing? The Journal of Positive Psychology, 4(5), 365-371. DOI:10.1080/17439760902992415n
Fredrickson, B. & Levenson, R. (1998) Positive emotions speed recovery from the cardiovascular sequelae of negative emotions. Cognition and Emotion, 12, 191-220.
Fredrickson, B. (1998) What good are positive emotions? Review of General Psychology, 2, 300-319.
Fredrickson, B. (2000) The Undoing Effect of Positive Emotions. Motivation and Emotion, 24(4), 237-258.
Fredrickson, B. (2001) The role of positive emotions in positive psychology: The broaden-and-build theory of positive emotions. American Psychologist, 56, 218–226.
Fredrickson, B. & Joiner, T. (2002) Positive emotions trigger upward spirals toward emotional well-being. Psychological Science, 13, 172-175.
Fredrickson, B. & Branigan, C. (2005) Positive emotions broaden the scope of attention and thought-action repertoires. Cognitive Emotions, 19(3) 313–332. doi: 10.1080/02699930441000238
Fredrickson, B. et al., (2010) Upward spirals of positive emotions counter downward spirals of negativity: Insights from the broaden-and-build theory and affective neuroscience on the treatment of emotion dysfunctions and deficits in psychopathology. Clinical Psychology Review, 30(7), 849-864. http://dx.doi.org/10.1016/j.cpr.2010.03.002
Fredrickson, B. L. (in press) Positive emotions broaden and build. In: E. Ashby Plant & P.G. Devine (Ed.) Advances on Experimental Social Psychology, Volume 47.
Freedman, B. & Weijer, C. (1996) Placebo orthodoxy in clinical research I: Empirical and methodological myths. Journal of Law, Medicine & Ethics, 24(3), 243-251.
Froh, J. (2004) The History of Positive Psychology: Truth Be Told. NYS Psychologist, May-June, 18-20.
Frost, R. & Green, M. (1982) Velten Mood Induction Procedure Effects Duration and Postexperimental Removal. Personal Social Psychology Bulletin, 8(2), 341-347. doi: 10.1177/0146167282082024
Gerrerds-Hesse, A. et al., (1994) Experimental inductions of emotional states and their effectivness: A review. British Journal of Psychology, 85, 55-78.
Gillham, J.E. & Seligman, M. (1999) Footsteps on the road to positive psychology. Behaviour Research and Therapy, 37, S163-S173.
Grimes, D. & Schulz, K. (2002) An overview of clinical research: the lay of the land. The Lancet, 359(9300), 57-61. DOI: 10.1016/S0140-6736(02)07283-5
Haahr, T. & Hróbjartssona, A. (2006) Who is blinded in randomized clinical trials? A study of 200 trials and a survey of authors. Clinical Trials, 3(4), 360-365.
65
Higgins, J. & Green, S. (2011) Cochrane Handbook for Systematic Reviews of Interventions. The Cochrane Collaboration.
Hróbjartssona, A. et al., (2009) Reporting on blinding in trial protocols and corresponding publications was often inadequate but rarely contradictory. Journal of Clinical Epidemiology, 62(9), 967-973. http://dx.doi.org/10.1016/j.jclinepi.04.003
Isen, A. (1984) The Influence of Positive Affect on Decision Making and Cognitive Organization. Advances in Consumer Research, 11, 534-537.
Isen, A. (1985) The influence of positive affect on the unusualness of word associations. Journal of Personality and Social Psychology, 48(6), 1413-1426. doi: 10.1037/0022-3514.48.6.1413
Isen, A. (1987) The influence of positive affect on acceptable level of risk: The person with a large canoe has a large worry. Organizational behaviour and human decision processes, 04(39), 145 – 154. DOI: 10.1016/0749-5978(87)90034-3
Isen, A. et al., (1992) The influence of positive affect on social categorization. Motivation and Emotion, 16, 65–78.
Izard, C. E. (1972) Patterns of emotion: A new analysis of anxiety and depression. New York, Academic Press.
Jamieson, S. (2004) Likert scales: how to (ab)use them. Medical Education, 38, 1212–1213. doi: 10.1111/j.1365-2929.2004.02012.x1217
Kenealy, P. (1986) The velten mood induction procedure: A methodological review. Motivation and Emotion, 10(4), 315-335.
Larsen, R. J. & Fredrickson, B. L. (1999) Measurement issues in emotion research. In: D. Kahneman, E. Diener & N. Schwarz (Ed.) Well-being: Foundations of hedonic psychology. New York, Russell Sage. pp. 40-60.
Larsen, R. & Prizmic-Larsen, Z. (2006) Measuring emotions: Implications of a multimethod perspective. In: E. Diener (Ed.) Handbook of multimethod measurement in psychology. Washington, DC, American Psychological Association. pp. 337–351.
Lee, G. (2011) The influence of positive affect on jumping to conclusions in delusional thinking. Personality and Individual Differences, 50(5), 717-722.
Levenson, R. (1988) Emotion and the autonomic nervous system: A prospectus for research on autonomic specificity. In: H,Wagner (ed) Social psychophysiology and emotion: Theory and clinical applications. London, Wiley. pp. 17–42.
Likert, R. (1932) A technique for the measurement of attitudes. Archives of Psychology, 140, 1-55.
Linley, A. (2006) Positive psychology: Past, present, and (possible) future. The Journal of Positive Psychology, 1(1) 3-16. DOI:10.1080/17439760500372796
Lucas, R. E. et al., (2009) Measuring positive emotions. In: E. Diener (Ed.) Assessing Well-Being: Social Indicators Research Series. Springer, Netherlands. pp. 139–155.
66
Maslow, A. H. (1954) Motivation and personality. New York, Harper.
Mauss, I. & Robinson, M. (2009) Measures of emotion: A review. Cognitive Emotion, 23(2), 209-237. doi:10.1080/02699930802204677
Messick, S. (1995) Validation of psychological assessment: Validation of inferences from persons’ responses and performances as scientific inquiry into score meaning. American Psychologist, 50(9), 741-749. doi:10.1037/0003-066X.50.9.741
Moher, D. et al., (2010) CONSORT 2010 Explanation and Elaboration: updated guidelines for reporting parallel group randomised trial. BMJ 2010; 340:c869
Norman, G. (2010) Likert scales, levels of measurement and the “laws” of statistics. Advances in Health Sciences Education, 15 (5), 625-632.
Ogden, J. & Lo, J. (2011) How meaningful are data from Likert scales? An evaluation of how ratings are made and the role of the response shift in the socially disadvantaged. Journal of Health Psychology, 17(3), 350-361. doi: 10.1177/1359105311417192
Pildal, J. et al., (2005) Comparison of descriptions of allocation concealment in trial protocols and the published reports: cohort study. BMJ, 330(7499), 1049. doi: 10.1136/bmj.38414.422650.8F
Polivy, J. (1981) On the induction of emotion in the laboratory: Discrete moods or multiple affect states? Journal of Personality and Social Psychology, 41(4), 803-817.
Rowe, G. et al., (2007) Positive affect increases the breadth of attentional selection. Proceedings of the National Academy of Sciences, 104(1), 383–388.
Ruark, J. (2009) An intellectual movement for the masses: Ten years after its founding, positive psychology struggle with its own success. The Chronicle of Higher Education, August 3rd. http://chronicle.com/article/An-Intellectual-Movement-for/47500/
Sakaki, M. & Kazuhisa, N. (2011) Effects of the brief viewing of emotional stimuli on understanding of insight solutions. Cognitive, Affective, & Behavioural Neuroscience, 11(4), 526-540.
Schulz, K. (2010) CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMC Medicine 2010, 8:18.
Schulz, K. et al., (1995) Empirical Evidence of Bias: Dimensions of Methodological Quality Associated With Estimates of Treatment Effects in Controlled Trials. JAMA, 273(5), 408-412. doi:10.1001/jama.1995.03520290060030
Schulz, K. & Grimes, D. (2002a) Generation of allocation sequences in randomised trials: chance, not choice. The Lancet, 359(9305), 515-519. DOI: 10.1016/S0140-6736(02)07683-3
Schulz, K. & Grimes, D. (2002b) Blinding in randomised trials: hiding who got what. The Lancet, 359(9307), 696-700. http://dx.doi.org/10.1016/S0140-6736(02)07816-9
Scriven, M. & Paul, R. (2003) Defining Critical Thinking. http://www. criticalthinking.org/University/univclass/Defi ning.html
67
Seligman, M. (1967) Failure to escape traumatic shock. Journal of Experimental Psychology, 74(1), 1-9.
Seligman, M. (1998) The APA 1998 Annual Report. American Psychologist, 54(8), 537-568. doi: 10.1037/0003-066X.54.8.537
Seligman, M. & Csikszentmihalyi, M. (2000) Positive psychology: An introduction. American Psychologist, 55(1), 5-14. doi: 10.1037/0003-066X.55.1.5
Seligman, M. et al., (2005) Positive Psychology Progress: Empirical Validation of Interventions. American Psychologist, 60(5), 410-421. doi: 10.1037/0003-066X.60.5.410
Seligman, E. (2011) Flourish: A visionary new understanding of happiness and well-being. New York, Free Press.
Shmueli, D. & Judith, P. (2012) A test of positive affect induction for countering self-control depletion in cigarette smokers. Psychological Addiction Behaviour, 26, 157-161.
Stolberg, H. et al., (2004) Randomized Controlled Trials Fundamentals of Clinical Research for Radiologists. American Journal of Roentgeneology, 183(6), 1539-1544.
Taylor, E. (2001) Positive psychology and humanistic psychology: A reply to Seligman. Journal of Humanistic Psychology, 41, 13-29.
Vacha-Haase, T. & Thompson, B. (2011) Score Reliability: A Retrospective Look Back at 12 Years of Reliability Generalization. Studies Measurement and Evaluation in Counseling and Development, 44(3), 159-168. doi: 10.1177/0748175611409845
Wadlinger, H. & Isaacowitz, D. (2006) Positive mood broadens visual attention to positive stimuli. Motivation and Emotion, 30(1), 87-99.
Watson, D. & Clark, L. (1997) Extraversion and its positive emotional core. In: R. Hogan, J. Johnson, & S. Briggs (Ed.) Handbook of personality psychology. San Diego, CA., Academic Press. pp.767-793.
Waugh, C. (2008) Adapting to life’s slings and arrows: Individual differences in resilience when recovering from an anticipated threat. Journal of Research in Personality, 42(4), 1031-1046. http://dx.doi.org/10.1016/j.jrp.2008.02.005
Westermann, R. et al., (1996) Relative effectiveness and validity of mood induction procedures: A meta-analysis. European Journal of Social Psychology, 26(4), 557-580. http://www.interscience.wiley.com/jpages/0046-2772/
Youngstrom, E. & Green, K. (2003) Reliability generalization of self-report of emotions using the differential emotions scale. Educational and Psychological Measurement, 63, 279–295.
68
Appendix
Please note: The information contained in some of the charts should be considered
as paraphrasing or as direct quotes from the actual studies. This process allows the
integrity of the authors’ statements to be maintained.
69
Study 1
Study Selection FormReviewer Andrew Smith
Date April 15-16 2012
Details of Publication
Authors: Anderson E., Siegel, E., Barrett. L.Title: What you feel influences what you see: The role of affective feelings in resolving binocular rivalry
Journal: Journal of Experimental Social Psychology, 47, 856-860. (2011)
Filters If yes, continue If no, remove study
Published 1998-2011 Yes
Peer reviewed journal Yes
Has positive affect/effect/emotion as part of the intervention
Yes
Has a control Yes
Has random assignment Yes
At least one outcome must bean objective outcome
Yes
One or more outcomes supports or contradicts the Broaden and Build theory
Yes
70
Summary of studyIdentification and Citation
Authors: Anderson E., Siegel, E., Barrett, L.Journal: Exp Soc Psychol. 2011 Jul; 47(4):856-860.MID:21789027[PubMed]
Location Corresponding author at: Department of Psychology, Northeastern University, Boston, MA 02115, USA. Fax: +1 617 373 8714.E-mail address: [email protected] (L.F. Barrett). Language: English
Abstract This experiment, demonstrates that the affective state of a perceiver influences the contents of visual awareness.
Study Design and Location
RCT in university lab
Hypothesis That the affective state of a perceiver influences the contents of visual awareness.
Sample Composition
Participants were 50 (15 male) naïve young adults ranging in age from 17 to 32 (Mean=20.64years).
Control As this was an induction process the control was an induction designed to be neutral. This was done by showing a selection of pictures selected as creating neutral mood.
Blind There was no mention of any participants being blind. The authors state that given the nature and complexity of the intervention it would be most unlikely that the subjects would have any understanding of the process.
InductionCategories
Unpleasant affective state, Neutral affective state, Pleasant affective state
Intervention Used ten images to initiate induction of unpleasantness or pleasantness. Subjects looked at pictures through a stereoscope. Pictures were of a house and a face that was neutral, scowling or smiling. Visual dominance of the image was indicated by pushing a button.
Outcomes When subjects are in an unpleasant affective state there was face dominance especially of the scowling face. In a pleasant state smiling faces dominated. In a neutral state neither types of faces dominated. This is the first direct evidence that a perceiver’s affective state helps to select the contents of consciousness.
(This chart contains close paraphrasing or direct quotes from the actual study)
71
Schemata of experiment
What you feel influences what you see: The role of affective feelings in resolving
binocular rivalry.
1-----2-----R-----3-----4-----5-----R-----6-----7-----8-----R-----9-----10-----11-----R-----12
Clarification: All participants had three inductions: positive, neutral and negative.
Faces depicted as neutral, smiling, or scowling face were randomized across the
three trials.
1. Induction-IN
2. Core affective state-CAS
3. Binocular rivalry trial-BRT
4. Induction
5. Core affective state
6. Binocular rivalry trial
7. Induction
8. Core affective state
9. Binocular rivalry trial
10. Induction
11. Core affective state
12. Binocular rivalry trial
72
Assessment of risk of bias for RCTs
Domain Reviewer’s judgement Support for judgmentRandom sequence generation
unclear Insufficient information
Allocation concealment unclear Insufficient informationBlinding of participants and personnel
Low risk Outcome not likely to be influenced by lack of blinding
Blinding of outcome assessment
Low risk Outcome is on the selection of particular pictures which is an objective outcome
Incomplete outcome data Low risk Nine participants removed from study due to poor eye sight. This was likely done at the beginning of the study so no data generated.
Selective reporting Unclear risk Insufficient informationOther sources of bias Unclear risk Insufficient information
Email correspondenceAnderson, Eric 27 Feb to AR, Lisa
Hi Andrew,
I'm happy to answer your questions.
What method was used for randomization?
Randomization of trial order was done using the presentation software, Eprime.
What groups were blind in the study?
Not exactly sure what you mean by this question. There was no between participant treatment condition, so there was no double blinding. PPs were blind to the hypothesis.
How were the participants selected, and where were they from?
PPs were Boston College undergraduates. They signed up to take part in the study through campus advertising.
Best of luck with your project. Eric~
73
Study 2
Study selection formReviewer Andrew Smith
Date April 15-16 201
Details of PublicationAuthors: Biss & HasherTitle: Delighted and distracted: Positive affect increases priming for irrelevant informationJournal: Emotion, 11(6), 1474-1478. (2011)Filters If yes, continue If no, remove
studyPublished 1998-2011
Yes
Peer reviewed journal Yes
Has positive affect/effect/emotion as part of the intervention Yes
Has a control Yes
Has random assignment Yes
At least one outcome must be an objective outcome Yes
One or more outcomes supports or contradicts the Broaden and Build theory YesIdentification and Citation
Author: Renée, Biss and Hasher, LynnJournal: Emotion. Vol. 11(6), Dec, 2011. P. 1474-1478
Location Correspondence concerning this article should be addressed to Rene´e K. Biss, Department of Psychology, University of Toronto, 100 St. George Street, Toronto, Ontario, M5S 3G3 Canada. E-mail: [email protected]
Abstract Emotional states are known to influence how people process relevant information. In this study the authors examine the impact of emotional state on irrelevant information
Study Design and Location
Participants were randomly assigned to a neutral or positive mood induction, and then completed a task that involved viewing a sequence of overlapping pictures and words. Lab setting.
Hypothesis The purpose of the present study was to determine whether positive affect has a causal effect on the broadening of attention to distracting information, ultimately resulting in facilitated transfer of this information to a later task.
Sample Composition 64 students from the University of Toronto (Canada)43/21 female /malemean age 19.4 (SD 2.7; range 17-33) 12.9 yrs. of educ. (SD 1.2)sample of convenience
Control As this was an induction process the control was an induction designed to be neutral. Pictures used for the neutral condition
74
had valence ratings between 4.5 and 5.5. Sound clips in the neutral condition, consisted of ambient street noise
Blind There was no mention of any individuals being blind.InductionCategories
There were four combinations of induction in two categories:For desserts: positive video, neutral video, positive written, neutral written. For vegetables same as for desserts.
Intervention Both subjects in the neutral group (control) and the positive induction group completed a task that involved viewing asequence of overlapping pictures and words. They were instructed to attend to the pictures and ignore the distracting words. Following a filled interval, implicit memory for the distracting words was tested using a word fragment completion task
Outcomes Priming for distraction was calculated by subtracting control completion rates for each group from each participant’s primed completion. Priming for previous distraction was greater for the positive group compared to the neutral mood group. This was consistent with predictions that positive mood broadens attention
Summary of study(This chart contains close paraphrasing or direct quotes from the actual study)
75
Schemata of experiment
Delighted and distracted: Positive affect increases priming for irrelevant information
2a----3----4----5----6----7----8----9----10
1----Randomization--
2b----3----4----5----6----7----8----9----10
1. Self-rated current mood and arousal
2. Six minute induction: 2a positive mood, 2b neutral
3. Self-rated current mood and arousal after induction
4. 1-back task (pictures randomized)
5. Nonverbal filler task
6. Self-rated current mood and arousal
7. Fragment completion task
8. Self-rated current mood and arousal
9. Awareness questionnaire completed by students
10. Filled in background questionnaire and debriefed
76
Assessment of risk of bias for RCTsDomain Reviewer’s judgement Support for judgment
Random sequence generation
Unclear Insufficient information
Allocation concealment
Unclear Insufficient information
Blinding of participants and personnel
High risk Four participants admitted they were able to ascertain the second part of the experiment. Their data was removed.
Blinding of outcome assessment
Low risk Completion of word fragment test was an objective outcome
Incomplete outcome data
Low risk No missing outcome data
Selective reporting Unclear Insufficient informationOther sources of bias
High risk Sample of convenience
Email correspondenceDear Andrew,
Thank you for your interest in our article. In our study, participants
were randomly assigned to condition using a randomized block design.
The blocks were randomized using the random number function in Excel.
Participants were blind to the purpose of the experiment and to the
fact that we were manipulating mood, however, as our mood manipulation
conditions involved listening to emotional music and viewing emotional
pictures, it is always possible that some figured out which condition
they were in (although none reported this to the experimenter). Staff
were not blind to condition.
Regards,
Renee Biss
77
Study 3
Study selection formReviewer Andrew Smith
Date April 15-16 2012
Details of PublicationAuthors: Falkenstern, M., Schiffrin, H.
Title: Mood over matter: can happiness be your undoing?
Journal: Journal of Positive Psychology, 4(5), 365-371. (2009)
Filters If yes, continue If no, remove study
Published 1998-2011Yes
Peer reviewed journalYes
Has positive affect/effect/emotion as part of the intervention Yes
Has a controlYes
Has random assignmentYes
At least one outcome must be an objective outcome Yes
One or more outcomes supports or contradicts the Broaden and Build theory Yes
78
Summary of studyIdentification and Citation
Mood over matter: can happiness be your undoing?Falkenstern, M., Schiffrin, H.,The Journal of Positive PsychologyVol. 4, No. 5, September 2009, 365–371
Location Department of Psychology, University of Mary Washington,Fredericksburg, VA, USALanguage: English
Abstract The goal of the current study was to extend the undoing hypothesis to the cognitive domain to determine whether positive emotion can undo the cognitive effects of negative emotion.
Study Design and Location
RCT in Lab
Hypothesis The goal of the current study was to extend the undoing hypothesis to the cognitive domain to determine whether positive emotion can undo the cognitive effects of negative emotion.
Sample Composition
86 undergraduate students from a liberal arts college in the mid-Atlantic region. Participants included 41 females and 21 males between the ages of 18 and 22 with a mean age of 19. Approximately 83.9% of the participants wereCaucasian, 1.6% African-American, 6.5% Hispanic,and 8.1% Asian or Pacific Islander, 6.5% ‘Other.’
Control The control group was shown a 65-second segment of two men talking in a courtroom from the film, All the President’s Men. This was used to induce aneutral mood.
Blind No mention of blindnessInductionCategories
First Induction: The negative emotion of sadness was induced in all participants by showing a 171 second video clip from the film The Champ where a son sees his father dying. Second Induction: Randomly assigned groups were shown one of three additional video clips. Sadness followed by one of: humour, sadness, neutral
Intervention Participants were asked to press the letter on the keyboard that corresponded to the letter that appeared on the screen. The dependent variable was response latency as measured by the time in milliseconds that the participant took to press the button after the letter appeared on the screen.
Outcomes All participants responded faster on the letter-identification task when induced into a negative affective state by a clip from The Champ. On second induction the negative and neutral affect conditions continued to demonstrate significantly faster reaction times, while the positive group’s reaction time did not change.
(This chart contains close paraphrasing or direct quotes from the actual study)
79
Schemata of experiment
Mood over matter: can happiness be your undoing?
--5a-----6------7-----8
--1------2------3------4--Randomization ---5b-----6-----7------8
--5c-----6-----7------8
1. LRTt1, Letter Recognition Task, familiarization
2. Induction
3. LRTt2, creating baseline
4. ERF, Emotion Report Form
5. Induction: 5a positive, 5b negative, 5c neutral
6. LRTt3, measure response time
7. ERF
8. LOT-R, Revised-Life Orientation Task, trait optimism
80
Assessment of risk of bias for RCTsDomain Reviewer’s
judgementSupport for judgment
Random sequence generation
Unclear Only mentions randomization but nothing about how it was generated
Allocation concealment Unclear Insufficient information to make a judgment
Blinding of participants and personnel
Low risk Letter identification task was computer generated
Blinding of outcome assessment
Low risk Objective outcome was the response time
Incomplete outcome data Low risk No missing outcome dataSelective reporting Unclear Insufficient information to make a
judgmentOther sources of bias Unclear Insufficient information to make a
judgment
Email correspondenceDear Dr. Schiffrin :
I am contacting you in regard to the article, "Mood over matter: can happiness be your undoing?" I am completing a systematic review for my MSc thesis and am working on a narrative synthesis relating to Fredrickson's broaden-and-build theory. When you have a moment, could you please let me know what process was used for selecting randomization of treatment and also if participants and/or staff were blind. Thank you, your help is greatly appreciated.
All the best,
Andrew Smith
Hi Andrew:
Participant signed up for group sessions (n = 9 if everyone showed) and the sessions were randomly assigned to one of the three film conditions. We just used a “draw the numbers out of a hat” approach to ensure roughly equal numbers of participants across conditions. Staff were not blind to the condition, but all of the data were collected on a computer, so there really wasn’t room for bias in data collection. Participants obviously knew what video they saw, but they were not told that the purpose of the video was to induce a specific mood and were not aware that there were other conditions seeing a different video until they were debriefed. Hope that helps!
Holly
p.s. I’d love to see the synthesis when you complete it!
81
Reviewer Andrew Smith
Date April 15-16 2012
Study 4
Study selection formDetails of PublicationAuthors: Fredrickson, B., Mancuso, R., Branigan, C., and Tugade, M.Title: The Undoing effect of positive emotionsJournal: Motiv Emot, 24(4), 237-258 (2000)Filters If yes, continue If no, remove
studyPublished 1998-2011 Yes
Peer reviewed journal Yes
Has positive affect/effect/emotion as part of the intervention
Yes
Has a control Yes
Has random assignment Yes
At least one outcome must be an objective outcome
Yes
One or more outcomes supports or contradicts the Broaden and Build theory
Yes
82
Summary of studyIdentification and Citation
The Undoing Effect of Positive EmotionsAuthors: Fredrickson, B., Mancuso, R., Branigan, C., and Tugade, M.Motiv Emot. 2000 December; 24(4): 237–258.
Location Department of Psychology, University of Michigan, 525 East University Avenue, Ann Arbor, Michigan 48109-1109; Email: [email protected]
Abstract Positive emotions are hypothesized to undo the cardiovascular aftereffects of negative emotions. Contentment-eliciting and amusing films produced faster cardiovascular recovery than neutral or sad films did.
Study Design and Location
RCT in Lab
Hypothesis That positive emotion would be unique in their ability to speed recovery from the cardiovascular reactivity that lingers after a negative emotion. Also that the undoing effect would generalise across sexes and ethnicity
Sample Composition
Two samples of university students were tested. Sample 1 included 95 university students (50% women) recruited for a study on emotions. 71 were European American (50% women) and 24 were African American (50% women). Sample 2, 75 students: 58 were Eur. American, 8 Asian, 3 African American, and 2 Hispanic and 4 other.
Control The Sticks film provides a neutral condition because as it holds cognitive demands constant, yet is devoid of emotion.
Blind No information on blindnessInductionCategories
First Induction Anxiety induction: to prepare a 3 minute speech on a to-be-determined topic. Second Induction: 1. Waves, breaking on a beach for contentment. 2. Puppy, playing with a flower for amusement. 3. Neutral control. Sticks, an abstract dynamic display of sticks piling up. 4. Sad, Boy crying watching father dying for sadness
Intervention Study 1 tests the undoing effect. Participants (n = 170) experiencing anxiety-induced cardiovascular reactivity. Then viewed a film that elicited (a) contentment, (b) amusement, (c) neutrality, or (d) sadness. Reactivity to baseline measured by six cardiovascular measures.
Outcomes Contentment-eliciting and amusing films produced faster cardiovascular recovery than neutral or sad films. Study 1 also demonstrated undoing effect such that the undoing effect occurs when the initial negative emotion generates a clear pattern of heightened sympathetic cardiovascular reactivity that is typical of anxiety, fear, and other health-damaging emotions.
(This chart contains close paraphrasing or direct quotes from the actual study)
83
Schemata of experiment
The Undoing Effect of Positive Emotions
5a----6----7----8
5b----6----7----8
----1----2----3----4--Randomization--- 5c----6----7----8
5d----6----7----8
1. Attachment of physiological sensors (continuous measurement)
2. Five minute adaption period
3. Two minute baseline measures
4. Stress induction tape played
5. Induction 5a contentment, 5b happy, 5c neutral, 5d sadness
6. Three minute blank screen and sound
7. Emotional report form 1 preparing speech
8. Emotional report form 2 viewing induction 5
84
Assessment of risk of bias for RCTsDomain Reviewer’s judgement Support for judgmentRandom sequence generation
Unclear Insufficient information
Allocation concealment Unclear Insufficient informationBlinding of participants and personnel
Low risk Lack of blinding unlikely to influence outcome
Blinding of outcome assessment
Low risk Cardiovascular measurements are objective outcomes
Incomplete outcome data Low risk No missing outcome dataSelective reporting Unclear Insufficient informationOther sources of bias High risk Sample composition
Email correspondence: None
85
Study 5
Study selection form
Reviewer Andrew SmithDate April 15-16 2012Details of PublicationAuthors: Lee, G., Lobban, F., Barrowclough C.Title: The influence of positive affect on jumping to conclusions in delusional thinkingJournal: Personality and Individual Differences, 50, 717-722. 2011Filters If yes, continue If no, remove
studyPublished 1998-2011
YesPeer reviewed journal
YesHas positive affect/effect/emotion as part of the intervention
YesHas a control
YesHas random assignment Originally considered
random but may in fact be systematic. Waiting for response to email for confirmation. Please see email correspondence.
At least one outcome must be an objective outcome Yes
One or more outcomes supports or contradicts the Broaden and Build theory Yes
86
Summary of studyIdentification and Citation
Authors: Lee, G., Lobban, F., Barrowclough C.Journal: Personality and Individual DifferencesVolume 50, Issue 5, April 2011, Pages 717–722
Location Gary Lee: School of Psychological Sciences, University of Manchester, Zochonis Building, Manchester M13 9PL, United Kingdom
Abstract This study examined relationships between positive affect and jumping to conclusions (JTC) in delusional thinking.
Study Design and Location
RCT. The computer randomized participants into one of two conditions (positive or neutral condition), differing only in the type of feedback provided. In university lab.
Hypothesis Participants exposed to the positive affect induction (positive condition) would require more information before making decisions than those receiving no mood induction (neutral condition).
Sample Composition
Gender Male 49 25.9, Female 140 74.1Ethnicity White 140 74.1%, Asian 7 3.7%, Black 3 1.6%, Chinese 2 1.1%, Other 8 4.2%Mean age22.8 (S.D.) 6, Min 18 Max 55
Control As this was an induction process the control was an induction process designed to be neutral. This was accomplished by giving neutral feed- back on a task.
Blind No mention of blindingInductionCategories
Induction: Positive moodTo induce positive affect with a meaningful stimulus, in the form of performance feedback that might be encountered within social situations
Intervention Those in a positive condition performed online creativity tasks and received bogus positive feedback as part of a positive affect induction procedure, whilst a neutral condition received neutral feedback. Both groups were subsequently assessed on a survey task for changes in JTC
Outcomes Those in a positive condition performed online creativity tasks and received bogus positive feedback as part of a positive affect induction procedure, whilst a neutral condition received neutral feedback. Both groups were subsequently assessed on a survey task for changes in JTC
(This chart contains close paraphrasing or direct quotes from the actual study)
87
Schemata of experiment
The influence of positive affect on jumping to conclusions (JTC) in delusional
thinking.
7p---8---9---10---11---12---13
1---2---3---4---5---6---Randomization
7n---8---9---10---11---12---13
1. PANAS t1, positive and negative affect schedule
2. Creative tasks
3. JTCn Trial 1 Jumping To Conclusions-majority negative comments
4. JTCp Trial 2 Jumping To Conclusions-majority positive comments
5. Self-evaluation t1
6. Difficulty rating t1
7. 7p-Positive feedback; 7n-Neutral feedback
8. PANAS t2
9. JTCp Trial 3
10. JTCn Trial 4
11. Self-evaluation t2
12. Difficulty rating t2
13. PDI-21, Peters Delusion Inventory
88
Assessment of risk of bias for RCTsDomain Reviewer’s judgement Support for judgment
Random sequence generation
Low risk Stated that computer randomized participants into one of two conditions
Allocation concealment
Low risk This was a web based experiment so controlled allocation.
Blinding of participants and personnel
Low risk This was a web based experiment so not likely that participants aware of intervention. As limited interaction with personnel little chance of influencing participants
Blinding of outcome assessment
Low risk There was no mention of blinding but given the form of the experiment, outcomes unlikely to be influenced. Also outcome, number of questions asked, was an objective outcome.
Incomplete outcome data
Low risk No missing outcome data
Selective reporting Unclear Insufficient information to make a judgement
Other bias Unclear Sample is one of convenience consisting of young participants average age 22.8 S.D. (6.0) attending university.
Email CorrespondenceGary Lee
to AR
Dear Andrew,
Thanks for your email. Randomisation was achieved by using the first letter of the
person's surname: if it began with a letter between A-M, they were assigned to the
neutral condition and N-Z to the positive condition. When the first 100 participants
had fully completed the study, the assignment was reversed (i.e. A-M assignment to
positive condition, N-Z to neutral). Participants and staff were blind to assignment
condition. Hope this helps! Bw
Follow up: 27 Jun/13
Dear Dr. Lee:
Could I take a few moments of your time for a follow up question please? I have been
questioned about the randomization process using last names. On the face of it, this
appears to be a systematic process rather than a randomization process. The study
89
states, 'The computer randomized participants into one of two conditions (positive or
neutral)...." Is selection for intervention by the computer based on number of
participants and alphabitization of last names? Is random allocation concealed due to
use of the computer? Any clarification would be greatly appreciated as one aspect of
my study looks at the process of randomization in research design. Thank you very
much for your time.
Cheers,
Andrew Smith
90
Study 6
Study selection form
Details of PublicationAuthor: Sakaki, M. (2011)Title: Effects of the brief viewing of emotional stimuli on understanding of insight solutionsJournal: Cogn Affect Behav Neurosci, 11, 526-540.Filters If yes, continue If no, remove studyPublished 1998-2011 Yes
Peer reviewed journal Yes
Has positive affect/effect/emotion as part of the intervention
Yes
Has a control Yes
Has random assignment Yes
At least one outcome must be an objective outcome
Yes
One or more outcomes supports or contradicts the Broaden and Build theory
Yes
Reviewer Andrew Smith
Date April 15-16 2012
91
Summary of studyIdentification and Citation
Authors: Sakaki, M., & Niki, K.Cogn Affect Behav Neurosci (2011) 11:526–540DOI 10.3758/s13415-011-0051-0
Location M. Sakaki University of Southern California,3715 McClintock Ave.,Los Angeles, CA 90089, USA e-mail: [email protected]. Niki National Institute of Advanced Industrial Science and Technology,1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan e-mail: [email protected]
Abstract To study the effects of the brief viewing of positive and negative pictures on reaction times for understanding solutions to insight problems
Study Design and Location
RCT in Lab 72 trials were randomly divided into three blocks of 24 trials. Each block included six positive, six negative, six neutral, and six control trials in a random order. Eighteen trials were used for each condition.
Hypothesis If presentation of emotional images modulates attentional scope, participants should be able to understand the solutions more quickly following positive modulators as compared with neutral modulators or the fixation cross. In contrast, response times should be slower following negative modulator images compared to neutral modulators or the fixation cross.
Sample Composition
Thirty-one Japanese undergraduates at the University of Tokyo participated in the experiment (12men; Mage = 20.2, SD = 0.83).
Control There was a neutral mood induction consisting of 18 pictures of mundane activities. There was also a no action control.
Blind No mention of blindnessInduction Catagories
The induction type (positive, negative, neutral, or control) was manipulated as a within-participants Each block included six positive, six negative, six neutral, and six control trial pictures in a random order
Intervention For each trial, participants were first presented with an insight problem and then briefly viewed a task irrelevant positive, negative, or neutral image (660ms) ,then subject was to give their solution to the problem.
Outcomes Positive and negative pictures had different impacts on subsequent understanding of the solutions to problems requiring insight. Brief viewing of positive pictures led to faster response times to report understanding the solutions, whereas viewing negative pictures for the same duration increased response times.
(This chart contains close paraphrasing or direct quotes from the actual study)
92
Schemata of experiment
Effects of the brief viewing of emotional stimuli on understanding of insight
solutions
--1-----randomisation of riddle---2-----3-----4a------5-----6-----7
4b
1. Familiarization
2. Presentation of riddle
3. Fixation cross
4. 4a Modulator Picture 4b control, continued fixation cross no picture
5. Fixation cross
6. Answer to riddle shown
7. Participants respond understand/not understand
93
Assessment of risk of bias for RCTsEffects of the brief viewing of emotional stimuli on understanding of insight solutions
Domain Reviewer’s judgement
Support for judgment
Random sequence generation
Unclear risk Insufficient information
Allocation concealment Unclear risk Insufficient informationBlinding of participants and personnel
Low risk Outcome not likely to be influenced by lack of blinding
Blinding of outcome assessment
Low risk Outcome was an objective measure. It was the length of time to respond to understanding the solution. Also follow up used neuroimaging which tended to confirm to mood state
Incomplete outcome data High risk Data from two participants were excluded because they did not finish the experiment.Five individual trials producing extreme outliers, were excluded.
Selective reporting Unclear risk Insufficient informationOther sources of bias High risk Sample size and composition
Email CorrespondenceDear Dr. Sakaki:When you have a moment, could you clarify two items with regard to your paper please: Effects of the brief viewing of emotional stimuli on understanding of insight solutions.1) What method was used for randomisation?2) What groups were blind in the study?I am a student working on my MSc at the University of Sheffield and this information is needed for my research.Thank you very much for your time and consideration; it is greatly appreciated.Cheers!Andrew SmithDear Andrew
Thank you for your interests in our study.
1) Randomization: In Study 1, riddles were randomly assigned to one of the four conditions. Each of the riddles assigned to the negative, neutral and positive conditions was randomly paired with one of the emotional modulator pictures used in the corresponding condition. The riddle-condition assignment and riddle-picture pairing were counterbalanced across participants. In Study 2, we first asked each participant to solve as many riddles as possible. We then picked up riddles that the participants could not solve for a later fMRI session. Those selected riddles were randomly assigned to the conditions.
2) We did not have any groups in the study. The experiment employed a within-participants design both for a behavioral (Study 1) and fMRI study (Study 2).
Please feel free to let me know if you have further questions.Best,MichikoAR Smith <[email protected]> 1 Mar
94
to Michiko
Dear Dr. Sakaki:Thank you for your speedy response. I apologise that my questions were not well written.The randomisation question should have referred to how the numbers for randomization were generated for Study 1. Did you use computer generated numbers or some other form of number allocation process.?With regard to the blinding, other than the participants in Study 1, were any other individuals blinded in the study?.I greatly appreciate your understanding. Best wishes,Andrew Smith
On 1 March 2013 09:20, Michiko Sakaki <[email protected]> wrote:Dear Andrew Smith,Thank you for your interests in our study.1) Randomization: In Study 1, riddles were randomly assigned to one of the four conditions. Each of the riddles assigned to the negative, neutral and positive conditions was randomly paired with one of the emotional modulator pictures used in the corresponding condition. The riddle-condition assignment and riddle-picture pairing were counterbalanced across participants. In Study 2, we first asked each participant to solve as many riddles as possible. We then picked up riddles that the participants could not solve for a later fMRI session. Those selected riddles were randomly assigned to the conditions. 2) We did not have any groups in the study. The experiment employed a within-participants design both for a behavioral (Study 1) and fMRI study (Study 2). Please feel free to let me know if you have further questionsBest wishes, Michiko
95
Study 7
Study selection form
Reviewer Andrew Smith
Date April 15-16 2012
Details of PublicationAuthor: Shmueli,DTitle: A test of positive affect induction for countering self-control depletion in cigarette smokersJournal: Psychology of Addictive Behaviors, 26 (1), 157-161. . (2012)Filters If yes, continue If no, remove studyPublished 1998-2011 Yes
Peer reviewed journal Yes
Has positive affect/effect/emotion as part of the intervention
Yes
Has a control Yes
Has random assignment Yes
At least one outcome must be an objective outcome
Yes
One or more outcomes supports or contradicts the Broaden and Build theory
Yes
96
Summary of studyIdentification and Citation
Author: Shmueli, D., Judith J., P.Journal: Psychology of addictive behaviorsISSN: 0893-164X Date: 2012 Volume: 26 Issue: 1 Page: 157DOI: 10.1037/a002370
Location Correspondence concerning this article should be addressed to Judith J. Prochaska, University of California, San Francisco, 401 Parnassus Avenue – TRC 0984, San Francisco, CA, 94143-0984. Email: [email protected] University of California-San Francisco, San Francisco, CA, US
Abstract The self-control strength model posits that exerting self-control on one task, such as resisting temptations, will deplete self-control and impair subsequent self-regulatory performance, such as controlling smoking… Positive affect elicited with a video was able to counteract the detrimental effects of self-control depletion on smoking behavior
Study Design and Location
Empirical Study: Qualitative Study RCT 2x2 design. Experiment took place in the field.
Hypothesis That positive affect induction would counteract depletion of self-control among participants assigned to resist tempting desserts versus raw vegetables and whether induction type produced different effects.
Sample Composition
The study sample consisted of smokers from the San Francisco Bay Area. Recruitment efforts included flyers and advertisements.118 women 67 men 8 transgender 36% white,44% African, 20% mixed,10% Hispanic 193 brought cigarettes (n=193)
Control The control is created by playing a neutral video induction for one group and a neutral writing exercise for the second group.
Blind No mention of blind.InductionCategories
There were four combinations of induction in two categories:Positive video, neutral video, positive written, neutral written for desserts and for vegetables The positive video was a 5-minute clip of a comedy stand-up routine; The neutral video was a 5 minute video about building bridges. The positive written exercise was writing about the happiest time in your life and to recreate the emotional feeling. The neutral exercise was to write a detailed description of a room in your house.
Intervention Randomized to one of four conditions (food exposure × affect induction), Participants were to try and resist either carrots or cakes that were within their reach for 1 hour. A ten minute break. Many participants went out for a smoke.
Outcomes Examination of the interaction demonstrated that it was only the positive video condition that counteracted the effects of depletion. Individuals exposed to desserts (10.5%) or vegetables (0%) that received the positive video condition were less likely to smoke relative to the other groups, range 20 to 85%.
(This chart contains close paraphrasing or direct quotes from the actual study)
97
Schemata of experiment
A test of positive affect induction for countering self-control depletion in cigarette
smokers
3a----4----5----6a----7----8----9
3b----4----5----6b----7---8----9
--1----2----Randomization
3c----4----5----6c----7----8----9
3d----4----5----6d----7----8----9
1. Pre-experimental measures
2. CO2 measures t1
3. 3a desert, 3b vegetables, 3c desert, 3d vegetables
4. Five minute exposure to food
5. Perceived interest in task assigned measured
6. Affect induction using either a positive 3a video or 3b writing or a neutral 3c
video or 3d writing
7. Break
8. CO2 measures t2 to find out who smoked during the break
9. Debriefing
98
Assessment of risk of bias for RCTs
Selective reporting Unclear Insufficient information to make a judgement
Random sequence generation
unclear Stated randomization but not how this was accomplished
Allocation concealment unclear Insufficient informationBlinding of participants and personnel
Low risk Outcome not likely to be influenced by lack of blinding
Blinding of outcome assessment
Low risk Outcome, number of people smoking measured by biochemical measurement, was an objective outcome.
Incomplete outcome data
Low risk 19 individuals who did not bring cigarettes were excluded decision was made a priori.Six ate the food and one did not receive the affect induction due to technical malfunction. Intent -to-treat analyses removing or keeping these individuals in did not alter the outcome so excluded individuals sample size n=193
Email correspondence
Hi Andrew,
I’d be happy to answer your questions. Predetermined randomization order was based on an online random number generator: http://www.randomizer.org/form.htm The groups were not blind to the depletion condition, in the sense that they knew they would be asked to resist eating either vegetables or sweets, and were then asked to resist one of those foods. However, they were not aware of the intent of this manipulation (i.e., to deplete self-control). The groups were unaware of the replenishment manipulation, or that there was more than one condition (i.e., positive vs. neutral). I hope that answers your questions. Please feel free to contact me if there is anything else I can help with. I will be starting maternity leave tonight but will check my emails weekly.Thank you,Dikla
99
Reviewer Andrew SmithDate April 15-16 2012Details of PublicationAuthor: Wadlinger, H. (2006)Title: Positive mood broadens visual attention to positive stimuliJournal: Motiv Emot. 30 (1), 87-99.Filters If yes, continue If no, remove study
Published 1998-2011 Yes
Peer reviewed journal Yes
Has positive affect/effect/emotion as part of the intervention
Yes
Has a control Yes
Has random assignment Yes
At least one outcome must be an objective oucome
Yes
One or more outcomes supports or contradicts the Broaden and Build theory
Yes
Study 8
Study selection form
100
Summary of studyIdentification and Citation
Authors: Wadlinger, H., A. and Isaacowitz, D., M.Motiv Emot. 2006 March 1; 30(1): 87–99.doi: 10.1007/s11031-006-9021-1
Location Department of Psychology, Brandeis University, MS 062, Waltham, MA 02454-9110 Email: [email protected]: English
Abstract The goal of the study was to evaluate whether a positive mood induction would broaden the visual attention of individuals using images that varied on level of emotional valence from very negative to neutral to very positive
Study Design RCT in lab. Participants were assigned either to be in the control or experimental condition using a random number generator.
Hypothesis Utilizing eye tracking technology, the purpose of the present studies was to investigate the broaden-and-build theory's claim that positive emotion increases the breadth of visual attention
Sample Composition
Fifty-eight young adult (undergraduate students) ranging from age 18 to 26, (M=19.69; SD=2.06) served as participants in this study (28 females, 30 males). Participants were recruited from an Introductory Psychology class at Brandeis University or through flyers posted on campus.
Control The control group was not given any candy and was unaware that there was any candy until the completion of the experiment when they were given some candy.
Blind No mention of blinding.Induction Positive mood was induced in participants in the experimental group by
presenting the participants with a small bag of candy (approximately five chocolate pieces) as a token of appreciation for their participation in the study after completion of the questionnaire packet and immediately prior to the slide presentation.
Intervention The main intervention related to timing the eye tracking of the subjects. A mixed-model analysis tested for overall effects of the mood manipulation on the first measure of attentional breadth that is the percentage viewing time to various types of peripheral emotional stimuli.
Outcomes Findings suggest that participants experiencing induced positive mood broaden their visual attention through increased viewing time to peripheral images of high positive valence and by making more frequent saccades to images of varying levels of positive and neutral valence.
(This chart contains close paraphrasing or direct quotes from the actual study)
101
Schemata of experiment
Positive mood broadens visual attention to positive stimuli
3a----4----5----6----7
---1-----2-----Randomization--
3b----4----5----6----7
1. Snellen Visual Acuity Test
2. Six demographic questions. Self-evaluation tests: CES-D, STAI, PANA and LOT
3. 3a positive mood, given candy; 3b control, no candy
4. Current affective state
5. Calibration to eye tracker
6. Slide presentation and tracking (Images not controlled for arousal or emotional
valence)
7. Debriefing
102
Assessment of risk of bias for RCTsDomain Reviewer’s
judgementSupport for judgment
Random sequence generation
Low risk Used random number generator
Allocation concealment Unclear Insufficient information
Blinding of participants and personnel
Unclear Study did not address this issue
Blinding of outcome assessment
Low risk Eye tracker measured gaze fixation duration times: all objective measures
Incomplete outcome data Low risk Potential participants that wore hard contact lenses or visual abnormalities were excluded from the experiment. Four participants data not included due to high reflective glare
Selective reporting Unclear Insufficient informationOther sources of bias Low risk Study appears to be free from
other sources of bias
Email correspondence: From: AR Smith [mailto:[email protected]]
Sent: Tuesday, October 23, 2012 1:08 PM
Subject: Follow up: Positive mood broadens
Dear Dr. Wadlinger:
I am contacting you in regard to the article, "Positive mood broadens visual attention to
positive stimuli." I am completing a systematic review for my MSc thesis and am working on a
narrative synthesis relating to Fredrickson's broaden-and-build theory. When you have a
moment, could you please let me know what process was used for selecting randomisation of
treatment and also if participants and/or staff were blind. Thank you, your help is greatly
appreciated.
All the best,
Andrew Smith
103
Dear Andrew:
My apologies for replying a bit tardy, this was forwarded to me promptly, but sometimes I
check my non-work emails less frequently.
Thanks for your inquiry and inclusion in your review! I actually conducted this study for my
Masters thesis. I believed we used a random online number generator (i.e.
ResearchRandomizer - http://www.randomizer.org/) to assign experimental conditions.
Participants were completely blind to condition. The staff was not blind to condition because
as part of the "induce positive" condition, participants were presented with a small gift
immediately prior to eye-tracking (thus staff knew the condition, but were instructed not to
deviate from the prepared scripts). Control participants received the gift immediately after
tracking. Participants were not debriefed until after the study and from informal conversations
afterward none connected that the candy gift related to their attention patterns (that we were
tracking).
Please feel free to get in touch if you have any additional questions. Please use this email or
my work email ([email protected]). Good luck with the review definitely sounds very
promising! Look forward to reading it in the future!
Cheers,
Heather