When labor leads to persuasion: The Ikea effect in persuasive texts
By
Ayellet Pelled
A dissertation submitted in partial fulfillment of
the requirements for the degree of
Doctor of Philosophy
(Mass Communication)
at the
UNIVERSTIY OF WISCONSIN-MADISON
2023
Date of final oral examination: 12/04/2023
The dissertation is approved by the following members of the Final Oral Committee:
Shah, D. Faculty, Professor, Mass Communication
McLeod, D. Faculty, Professor, Mass Communication
Wagner, M. Faculty, Professor, Mass Communication
Niedenthal, P. Faculty, Professor, Psychology
Kaplan, D. Faculty, Professor, Education Psychology
i
Acknowledgements
This dissertation is dedicated to Nathan Walter,
who helped me navigate through my darkest hours.
While many are fortunate to call him their colleague,
I count myself luckier to call him my friend.
I would also like to express the deepest gratitude to my advisor, Prof. Dhavan Shah, for his
invaluable guidance, unwavering support, and extraordinary patience throughout this journey. His
mentorship was indispensable for dissertation. Thank you!
ii
Table Of Contents
List of tables................................................................................................................................................ iv
List of figures ............................................................................................................................................... v
General Introduction .................................................................................................................................. 1
The IKEA Effect ....................................................................................................................................... 1
Study Rational: From Effort to Persuasion (or The Idea Behind Ikea) ..................................................... 4
The applicability of text as a DIY product ............................................................................................ 5
Text assembly as a method of self-persuasion ...................................................................................... 6
Chapters Summary .................................................................................................................................. 10
CHAPTER 1. If You Build It: Meta-Analysis of the Ikea Effect .......................................................... 12
Contextual Moderators ............................................................................................................................ 14
Self-Expression via Product Customization........................................................................................ 14
Extent of effort and freedom of choice ............................................................................................... 15
Product Tangibility ............................................................................................................................. 16
Product Usage: Utilitarian vs. Hedonistic ........................................................................................... 17
Method .................................................................................................................................................... 19
Results ..................................................................................................................................................... 25
Discussion ............................................................................................................................................... 32
CHAPTER 2. From the Ikea Effect to Self-Persuasion ......................................................................... 38
Self-persuasion ........................................................................................................................................ 38
The influence of behaviors on attitudes .............................................................................................. 39
Self-persuasion via role-playing ......................................................................................................... 42
The power of self-generated arguments .............................................................................................. 43
Self-persuasion and Social Media ....................................................................................................... 45
Integrating the IKEA Effect into a Self-Persuasion Model .................................................................... 47
Moderators of self-persuasion and the Ikea effect .............................................................................. 50
Final Notes for Study Design .............................................................................................................. 53
CHAPTER 3: The IKEA Effect in Persuasive Healthcare Messages .................................................. 56
Study 1 .................................................................................................................................................... 56
Study 2 .................................................................................................................................................... 73
Study 3 .................................................................................................................................................... 79
GENERAL DISCUSSION ....................................................................................................................... 99
Key Findings and Theoretical Implications .......................................................................................... 100
Study Limitations and Directions for Future Research ......................................................................... 106
iii
References ................................................................................................................................................ 114
Appendix .................................................................................................................................................. 131
iv
List of tables
Table number
Title
Page
Table 1
Summary of article coding for the meta-analysis
23
Table 2
The effect of self-invested effort on outcome variables
27
Table 3
The IKEA effect by contextual moderators
29
Table 4
Summary statistics – Study 1
70
Table 5
Summary statistics – Study 2
77
Table 6
Summary statistics – Study 3
90
Table 7
Regression models for post attitudes - Study 3
98
Table 8
Codebook for meta-analysis
132
Table 9
Regression models for post attitudes - Study 1
134
Table 10
Regression models for text evaluation - Study 1
135
Table 11
Regression models for post attitudes - Study 2
136
v
List of figures
Figure number
Title
Page
Figure 1
Distribution of SDs in Means for the effect of labor on valuation.
26
Figure 2
Distribution of SDs in Means for the effect of labor on liking.
27
Figure 3
The influence of effort on valuation by tangibility and
customization
30
Figure 4
Funnel Plot for the Detection of a Potential Publication Bias
31
Figure 5
IKEA effect model - main mechanisms
48
Figure 6
Self-persuasion model - main mechanisms
48
Figure 7
Integrated model for the IKEA effect and Self-persuasion
48
Figure 8
Model proposal for mediation hypotheses testing
66
Figure 9
Multiple mediation model - Experiment 1
68
Figure 10
Multiple mediation model - Experiment 2
76
Figure 11
Negative emotions by effort
92
Figure 12
Influence of effort on evaluation via negative emotion
93
Figure 13
Valuation by text engagement and webpage customization
94
Figure 14
Multiple mediation model - Experiment 3
96
Figure 15
Search strategy flow-chart for meta-analysis
131
Figure 16
Stimuli material study 3 – Icons for color scheme choices
140
Figure 17
Stimuli material study 3 – Mock webpage design
141
1
General Introduction
The IKEA Effect
The IKEA effect, a term coined in economic psychology, is a common cognitive bias that
refers to the tendency of people to overestimate the value of things that they have personally
invested effort in creating or assembling (Norton et al., 2012). People often appraise greater value
to, and are willing to pay more for, products that they assemble themselves, even though those
items may not be objectively more valuable than similar items that were assembled by someone
else. This bias persists even compared to products of better quality that were created by
professionals (Norton, et al., 2012). The phenomenon was demonstrated in a series of studies in
which participants valued their “creations” with greater regard compared to observers who did
not engage in the assembly, whether it was storage boxes or Lego models they assembled, or
even origami crane that they folded (Mochon, et al., 2012; Norton et al., 2012). Not only did they
value their creations on the same level as experts, but they also believed that others would
similarly value their craftsmanship.
Main Mechanisms of the IKEA Effect
The existing literature proposes three main pillars on which the IKEA effect stands: self-
invested effort, positive self-concept, and psychological ownership. These factors and their
related psychological processes are not mutually exclusive, but rather, their impact is integrated
in conjunction with each other, as they collectively cultivate a positive association between the
individual and the self-assembled object, rendering it of greater perceived value.
The value of effort. The initial conception was that the IKEA effect stems from the
rationale of labor equals love (Norton et al., 2012) which is based on the effort heuristic. The
effort heuristic predicts that perceived effort leads to greater value and appreciation (Kruger, et
al., 2004) and, indeed, people often rely on effort to infer the quality of items (Inzlicht et al.,
2
2018). In general, people value their time and effort, thus investing effort into a task will often
lead to greater valuation of the task and its outcome. A classic study by Aronson and Mills
(1959) demonstrated the relationship between effort and valuation by requiring female
participants to undergo an initiation of varying severity, prior to joining a discussion group. The
more severe and embarrassing the initiation they underwent, the more positively the participants
rated the discussion group, even though they all attended the same “dull” discussion (Aronson &
Mills, 1959). The authors suggested that the positive valuation may be a psychological
mechanism intended to reduce cognitive dissonance (Festinger, 1957) as a manner of making
sense of the personal investment. This reasoning process, which stems from cognitive dissonance
is also referred to as effort justification. It is the tendency of people to attribute greater value to
the outcome of a task they engaged in as a means of rationalizing their effort (Aronson & Mills,
1959; Festinger & Carlsmith, 1959).
The value of accomplishment. Another explanation for the relationship between effort
and valuation relates to our concept of self. From this perspective, the IKEA effect is driven by
the inherent satisfaction and feelings of accomplishment that individuals derive from completing
a task or creating something on their own (Mochon et al., 2012). This sense of accomplishment
and pride may be the reason leading individuals to place a higher value on items that they have
successfully created. Self-concept is a dynamic web of thought and knowledge we have about
ourselves; comprised of our Self-image—the cognitive aspect of how we perceive ourselves and
how we think others see perceive us, and Self-esteem—The affective aspect of how we generally
feel about ourselves (Alsaker & Kroger, 2006; Baumeister; 1999). The successful completion of
a task can lead to a feeling of accomplishment and positive self-efficacy (Bandura, 1982), which
in turn improves our self-concept. The more difficult and demanding a task, the greater the
3
psychological reward (Bandura, 1982; Talsma et al., 2018), that is, if the task is accomplished
successfully. It is the successful effort that leads to positive valuation, as the relation between
labor and positive valuation only occurs under fruitful attempts. Failure to complete the task or
destroying the creation diminishes the effect (Norton et al., 2012; Sarstedt et al., 2017). These
studies indicate that effort on its own may not be sufficient to generate positive valuations, it is
also the outcome of the effort that matters.
The value of [psychological] ownership. Finally, a psychological sense of ownership
over our creations has also been demonstrated to moderate the relationship between labor and
valuation (Sarstedt, et al., 2017), as effort devoted to creating an item may foster an emotional
attachment and via personal sentiment. A sense of ownership over a product is induced by
becoming familiar with it, controlling it, and self-investing into it (Pierce, 2001). People tend to
place a greater value on things they own, a phenomenon known as the endowment effect
(Kahneman, et al., 1990). This cognitive bias occurs when people perceive more value to
something simply because they own it, and they demand a price much higher than non-owners
may be willing to pay for it. The endowment effect is presumed to be driven by a combination of
factors, including the sunk cost fallacy, loss aversion, and feelings of attachment and familiarity
(Kahneman et al., 1990; Marzilli et al., 2014). Deriving from the logic of loss aversion, people
may be willing to pay more just to hold on to what they perceive is their property (Kahneman et
al., 1990). People may also perceive greater value in things they own because it gives them a
sense of identity and self-expression. From this perspective, positive product evaluations may be
the result of a broader sense of ownership in addition to emotional attachments (Walasek et al.,
2017). However, a later study examined the cognitive biases underlying the IKEA effect via a
developmental approach and contested the role of ownership (Marsh, et al., 2018). Marsh et al.,
4
(2018) concluded that a sense of ownership is not likely a main factor but rather, that there is a
strong connection between our creations and our concept of self-concept, further supporting the
idea of a correlation between the psychological reward of task completion and positive valuation
of the self. Pride has been demonstrated to operate as an antecedent to psychological ownership
and it can amplify its influence on valuation (Kirk et al., 2015). Thus, a sense of accomplishment
and psychological ownership may be somewhat interrelated.
Study Rational: From Effort to Persuasion (or The Idea Behind Ikea)
From a communication perspective, this line of research raises relevant questions as to
whether the IKEA effect would operate similarly in the context of text. Would the cognitive
processes that occur during the assembly of consumer goods apply to the “assembly” of a text?
Would this process potentially result in a more positive valuation of the text someone labored on
compared to a text created by someone else? Developing this line of research one step further, it
becomes imperative to delve into the potential implications of text assembly were the text in
question to be comprised of persuasive content. Within this context of persuasive messages, it
would then be of scholarly interest to investigate whether the correlation between effort and
valuation could be extended to encompass a correlation between effort and persuasion. In
essence, this inquires whether self-invested effort into assembling a persuasive text could lead,
not only to elevated valuation of the text, but also, to a more pronounced shift in attitude that
aligns with the message conveyed within the text.
Furthermore, exploring whether individuals who invest personal effort into assembling a
text are not only more likely to value it positively, but also more likely to align their attitudes
with the arguments comprising the text, would contribute to the scholarly discourse by
suggesting a potential mediation path of influence. More explicitly, if the data indicates that
5
engaging in text-assembly results in both enhanced valuations and increased persuasion, it could
suggest that the potential impact of effort on attitude change is mediated via text valuation.
Summing up the questions presented above, this dissertation peruses a two-fold research
question: Would laboring on the assembly of a persuasive text lead to greater valuation of the
text, and in turn, would this greater valuation translate into greater support of its persuasive
content?
The applicability of text as a DIY product
This first part of the main research question inquires whether the relationship between
self-invested effort and product valuation would operate similarly pertaining to text assembly, or
in other words: Would assembling a text, as opposed to solely reading it, result in a more
positive valuation of the text? It is a rather straight forward question, asking to examine whether
the IKEA effect long with its underlying mechanisms, manifests when the self-assembled
product is a text. This taps into a deeper inquiry aiming to assess the applicability of text as a
DIY product. The IKEA effect has primarily been studied in the context of consumer products,
such as household items, fashion, toys, food, and art. The pertinence of the IKEA effect to text
may appear less straightforward due to the specific features of text that distinguish it from
consumer goods. However, recent studies have successfully tested the IKEA effect with regards
to products that are not necessarily physical nor strictly consumeristic in nature, such as music
playlists (Cloots, 2019), informative presentations (Rusho & Raban, 2021), and software
programing (Fink & Geldman, 2017), suggesting that the phenomenon may be applied to text-
based products as well as virtual environments and non-physical items.
Furthermore, assembling a text, much like assembling consumer products, would involve
self-investment of time and cognitive resources. In the process of assembling a product from
6
given materials, as in the common DIY product, one must follow specific instructions and place
all the pieces in the correct location and order to successfully construct a functioning final
product. Similarly, in the process of assembling a text from given statements and terms, one must
follow specific grammatic rules and logical flow to construct a meaningful and readable text.
Through the effort invested into this process, individuals may develop a sense of ownership and
emotional attachment to the text they labored on, just as they would for other self-assembled
products. Thus, while there are unique aspects to texts in comparison to common consumer
goods, it is plausible to assume that we could find support for the applicability of the IKEA
effect in the context of text assembly.
Text assembly as a method of self-persuasion
The second part of the main research question seeks to determine whether assembling a
persuasive text, and greater valuation of the text, would translate into greater attitude alignment
with the arguments conveyed in the text. More specifically, it first inquires whether individuals'
attitudes would be influenced by the content of a message they assembled more than the
influence of reading the same message without engaging in its construction. It then asks whether
this influence would be mediated via valuation of the text. Therefore, it is less straightforward, as
it demands us to address multiple concerns, beyond assessing a potential mediation path via
valuation. Mainly, it directs us to determine whether assembling a text offers sufficient
engagement in order to activate the key underlying psychological mechanisms that lead to self-
persuasion. In general, we can presume that prompting people to increase the perceived valuation
of their own fruits of labor would be an easier task than prompting them to shift their attitudes,
let alone their behaviors. Thus, while it is plausible to hypothesize, consistent with the principles
of the IKEA effect, that self-invested effort in constructing a text may increase its perceived
7
valuation, it remains less certain as to whether such investment and increased valuation would
result in attitude change, especially considering factors that affect information processing and
text assessment such as reading fluency and disruption to information consumption.
When examining the potential functionality of the framework particularly with persuasive
content, there are unique features relating to language and information processing that should be
considered. As noted above, both fluency of the text readability and perceived message strength
can impact the evaluation of the text and its processing path. Processing fluency refers to the
ease with which information is consumed and understood by the individual (Song & Schwarz,
2008). Texts that are easy to read and follow a clear flow are more likely to be evaluated
positively and possess persuasive power, as they require less cognitive effort to process
(Winkielman et al., 2003). Fluency has been found to serve as an indicator for believability, as it
increases the likelihood that a statement will be judged as true (Reber & Unkelbach, 2010).
Similarly, rhyming in aphorisms has been found to increase their perceived accuracy, rendering
them as more believable because it is easy to anticipate the continuation of the sentences, which
increases their processing fluency (McGlone, & Tofighbakhsh, 2000). Within the context of
persuasive health information, materials that are processed fluently and easily tend to elicit more
positive evaluations such as liking, and they can impact an individual’s willingness to undertake
the treatment (Okuhara et al., 2017). Similarly, interruptions to fluency are generally perceived
as negative for communication as they are likely to inhibit persuasion, because the individual's
ability to process of the message may be diminished, resulting in reduced engagement with the
message (Petty & Cacioppo, 1986). This is relevant for a text assembly task, where the
interruption occurs simultaneously as part of the task, because this distraction imposes cognitive
load and may deplete individuals' processing capacity (Kupor & Tormala, 2015). However,
8
reduced cognitive capacity would also result in lower ability to counterargue with the message in
a systematic manner, potentially resulting in less opposition to the arguments in the text—a
desired factor for persuasion goals.
Message strength refers to the quality and persuasiveness of the message itself rather than
the flow of information. A well-researched, logical, and emotionally appealing message is more
likely to be persuasive than a message that is poorly constructed or lacks substance, although the
level of engagement with the message depends on more than message strength (Eagly &
Chaiken, 1993; Petty, Cacioppo & Goldman 1981). The characteristics of the audience also play
a critical role in determining the perceived strength and persuasiveness of a text. A message that
resonates with an individual's values and beliefs is more likely to be persuasive than a message
that contradicts them, and the processing path is determined by their motivation and ability to
systematically engage with the message (Chaiken, 1980). Additionally, the audiences’ prior
knowledge of the topic can also influence the perception of the message's strength rendering it as
more or less persuasive because they are more attuned to inaccuracies in the message, and they
are more likely to have a developed arsenal of counterarguments to potentially inhibit any of
persuasion attempt. If the message is perceived as strong, informative, and unbiased, there is a
greater likelihood of accepting it.
To conclude, in contemplating the potential integration of the IKEA effect framework
into the realm of persuasive textual content, the intricacies of cognitive mechanisms involved in
information processing highlight significant factors that warrant consideration. Among others,
elements such as reading fluency, text readability, and perceived message strength intertwine to
shape how individuals evaluate and interact with persuasive texts. Processing fluency, reflecting
the ease of information comprehension, not only affects the positive evaluation of texts but also
9
carries implications for their persuasive potency. Similarly, the inherent strength and appeal of
the message itself influence its persuasive impact. The alignment of the audiences’ values and
pre-existing beliefs with the message, coupled with their prior knowledge and ability to engage,
further contributes to the intricate matrix of persuasion. The exploration of these complexities
underscores the need for a comprehensive understanding of the interplay between these
components as we progress to test the synergy between the IKEA effect, text assembly, and the
dynamics of persuasion. Given the distinctive properties of textual information, a more
comprehensive investigation of relevant assumptions, with particular attention to information
processing and persuasive communication, is warranted.
To address the questions and concerns raised above, and to assess the potential
application of the IKEA effect as a method of self-persuasion in text-based contexts, the current
dissertation draws on theories and empirical findings from the domains of the IKEA effect, self-
persuasion, and information processing. It begins with an extensive meta-analysis of the IKEA
effect that assesses its overall effect size, robustness, and generalizability of the phenomenon,
while highlighting its underlying mechanisms and potential boundary conditions. This meta-
analysis is conducted in order to attain a better grasp on the phenomenon and its potential
moderators, as well as serve as a steppingstone for the following chapters. The dissertation then
proceeds to review the extant literature on self-persuasion, with a focus on processes that are
relevant to the main research questions. This review provides the basis for integrating the IKEA
effect into a self-persuasion model within the context of text assembly and enables us to assess
the similarities and distinctions between the two frameworks. Finally, the model is developed
and tested in a series of empirical studies comparing the valuation and persuasion outcomes
10
between individuals who assembled persuasive messages, and those who merely read the same
message without engaging in its creation.
Chapters Summary
Chapter 1. If you build it: Meta-analysis of the IKEA effect. This chapter presents data
from a meta-analysis aggregating data from empirical studies assessing the relationship between
self-invested effort and valuation (n = 31, k = 55, N = 5454). It estimates the average effect size
of the IKEA effect and highlights the main underlying mechanisms that moderate the
phenomenon. In addition to the three main moderators noted by existing literature—effort, self-
concept, and psychological ownership—the analysis tests potential moderators including
customization, product tangibility, product usage type, liking the product and positive affect.
Beyond offering implications for future research on the IKEA Effect, this meta-analysis serves as
a steppingstone for the consecutive chapters, as it assesses vital assumptions about the
phenomenon, and enables the adaptation of the framework as an application to communication
research.
Chapter 2. From the IKEA Effect to Self-persuasion: An integrated model. This chapter
begins with an extensive review of relevant self-persuasion literature, covering main concepts
such as cognitive dissonance, self-perception, and the power of self-generated arguments. It
incorporates the advantages and limitations of self-persuasion methods compared to direct
persuasion methods and highlights the varied mechanisms through which behavior can influence
attitudes. Following the review, the chapter contrasts the similarities and differences between the
IKEA effect and self-persuasion frameworks within the context of persuasive texts, followed by
a discussion concerning the proposed integrated model which merges the two. This model is later
tested in a series of empirical studies, examining the influence of assembly effort on text
11
valuation [IKEA effect] and the influence of both effort and valuation on attitude change [Self-
persuasion].
Chapter 3. The IKEA Effect in persuasive healthcare messages presents data from three
empirical studies testing the proposed model. Experiment 1 tested the influence of self-labor on
attitude change concerning a healthcare issue, presenting data from a 2-task type (assemble vs.
read) by 2-healthcare practice (teeth flossing and sunscreen application) between subject design.
Experiment 2 replicated the design of Study 1 with minor changes to the stimuli and
questionnaire aimed to address potential issues that arose from the original materials. Experiment
3 replicated the model from the previous studies, while incorporating a major adjustment, as the
stimuli were embedded within a [mock] health website. Furthermore, this study included two
additional manipulations, with the intention of testing the influence of (a) self-expression via
aesthetic customization of the web page (customization vs no customization) and (b) the extent of
effort by altering the difficulty level of the assembly task (assemble-difficult vs. assemble-easy
vs. control). This resulted in a 2 by 3 randomized experiment between subject design. The goal
of Experiment 3 was to hone everything together by incorporating findings from the meta-
analysis presented in Chapter 1 as well as addressing questions and concerns that arose from the
preceding experiments.
General Discussion. This section summarizes the main findings and inferences from all
the studies and addresses theoretical and methodological implications of utilizing the IKEA
effect in persuasive messages. It then addresses potential study limitations and proposes practical
and theoretical implications for future research.
12
CHAPTER 1. If You Build It: Meta-Analysis of the Ikea Effect
Over the last two decades, the IKEA effect has garnered significant attention from
researchers across a range of disciplines, including psychology, marketing, and consumer
behavior. Studies have explored the underlying mechanisms of the effect, its boundary
conditions, and its implications for consumer behavior and decision-making, testing this
phenomenon with relation to a wide array of consumer goods, including toys (e.g., Marsh et al.,
2018), fashion (e.g., Sarstedt, et al., 2017), food (e.g., 2014; Lefebvre & Orlowski, 2020;
Raghoebar et al., 2017), arts and crafts (e.g., Erol, 2022; Straffon et al., 2022), 3D-prints (e.g.,
Wiecek, et al., 2020), software coding (e.g., Fink & Geldman, 2017; Shmueli et al, 2016),
chatbots (Wald et el., 2021), household items (e.g., Atakan et al., 2014b), robotics (Sun &
Sundar, 2016); investment portfolios (e.g., Ashtiani et al., 2021; Brunner, et al., 2022) and
information (e.g., Rusho & Raban, 2020). However, despite the growing interest in the IKEA
effect, there is a lack of consensus as to its robustness and generalizability, and considerable
work remains to be done to untangle and clarify the specific mechanisms and boundary
conditions under which the IKEA Effect operates. Although most studies are informed by similar
frameworks and employ akin designs, the results vary considerably. While many studies yield
strong support for the effect, others exhibit weak or null effects on product valuation, and several
have identified important moderators that limit its scope. In light of the discrepancies, this meta-
analysis conducts a comprehensive synthesis of the existing literature by systematically
examining the primary mechanisms that have been theorized to explain the IKEA Effect—self-
invested effort, psychological ownership, and self-concept. The meta-analysis also tests
additional contextual mechanisms that may explain variability in the results.
13
In the first step, it sets out to evaluate the overall effect of self-invested effort on product
valuation. Accordingly, the following research question is proposed:
RQ1: What is the average effect of self-invested effort on valuation of the target
product?
To answer this question, we conduct a meta-analysis that aggregates data from empirical
studies assessing the relationship between effort and valuation of DIY products. Specifically, we
focus on studies that manipulate the type or extent of effort invested into creating or designing a
product, and then measure the influence of that effort on valuation of the product.
In succession to valuation, the meta-analysis aims to assess the overall effect that effort
elicits on additional outcomes of interest including self-concept, psychological ownership, liking
of the product, and positive affect. The first two factors—self-concept, and psychological
ownership—have been highlighted in the introduction as two identified key psychological
mechanisms of the phenomenon. The latter two—liking and positive affect—are associated with
valuation more generally. Liking of the product is closely related to valuation as people's positive
feelings towards a product can lead to a higher perceived valuation, and accordingly, several
studies that tested the IKEA Effect indeed measured liking as an outcome of effort (e.g., Erkin et
al., 2018; Norton et al., 2012). Similarly, the experience of positive affect is closely associated
with a sense of accomplishment and pride, both of which relate to emotional attachment and
psychological ownership. Although these four outcomes are typically treated as secondary to
valuation, closer attention to these constructs not only may help explain the main influence of the
IKEA Effect but also elucidate its underlying mechanisms. Accordingly, the two following
research question was posed:
14
RQ2: What is the average effect of self-invested effort on (a) liking; (b) self-
concept; (c) psychological ownership; and (d) positive affect?
RQ3: Is the effect of effort on product valuation contingent upon these four
mechanisms?
Contextual Moderators
In addition to the three main moderators presented in the introduction, there is a myriad
of contextual moderators that influence the relationship. Therefore, we next gauge the more
nuanced factors, highlighting additional moderators that inhibit or enhance the relationship
between effort and valuation. Having laid out the main aspects of the IKEA effect and the
theories from which it stems, we progress via a theory-based approach to address potential
boundary conditions that may explain the variances in the findings across studies, including self-
expression, extent of effort, product tangibility and product usage.
Self-Expression via Product Customization
The ability to customize a product renders it unique and more likely to satisfy a person’s
individual preferences and tastes. It offers a form of identity expression through personal design
choices (Franke & Schreier 2010; Franke et al. 2010). Customization of a product may also
increase its perceived value based on the mental effort invested into designing and personalizing
it. The type of effort required for the production can moderate its influence on valuation,
pertaining to the extent of mental effort one is required to invest (Atakan et al., 2014a).
Additionally, customizing a product increases the emotional bonding with it as a result of the
time spent on the product and the mental effort invested into its appearance (Mugge et al., 2009).
It could develop a feeling of ownership and control, which increases its perceived value.
Furthermore, according to some studies, consumers experience greater enjoyment and value the
15
assembly experience when the customization decisions and the assembly process are integrated
(Buechel & Janiszewski, 2013). Thus, the following hypothesis is posed:
H1: Self-invested effort bears stronger influence on product valuation when self-
customizing the item as opposed to solely assembling it.
Extent of effort and freedom of choice
When the process of creating and designing a product offers a greater freedom of choice,
as in more choice options, it inevitably offers a greater opportunity for self-expression via
customization. In turn, expression of self develops attachment and sense of ownership, which
may lead to increased valuation (Ling, et al., 2020). Similarly, when there are more restrictions
on creativity, as in less choice options, the task of creation is not as mentally taxing or engaging.
In this manner, restriction levels can be used as an indicator for mental effort. Challenging tasks
develop greater product attachment, as the level of emotional attachment is influenced by both
the extent of invested effort and the degree of self-expression (Atakan et al., 2014a; Mugge, et al.
2009). From an alternative perspective, having an abundance of options is not always positive.
Too many options can over complicate the customization, due to the requirement of processing a
simultaneous overload of information and attributes (Dellaert & Stremersch 2005; Valenzuela et
al. 2009). From this perspective, constraints can both enhance creativity, as well as inhibit it
(e.g., Dahl & Moreau 2007). Ideally, one could compare the level of effort between different
studies to assert whether conditions which elicit greater effort result in greater effect sizes.
However, one cannot assume the extent of effort without it being consistently measured across
different studies, especially considering distinct features such as product types and design
processes. What we can assume, however, is that the lowest level of elicited effort, whether
mental or physical, may be associated with studies in which the item was created via 3D-
16
printing. After all, the “creation” of the product in such cases merely involved pressing a button
to instruct the program to begin printing. Accordingly, the second hypothesis is posed:
H2: Self-invested effort has a weaker influence on product valuation when the
item is created via 3D-printing.
Product Tangibility
Tangible products are physical objects that can be touched and handled. Touching and
holding a physical object can increase the sense of ownership and attachment to it (Peck & Shu
2009). Even sensory information derived from touching a product can influence its perceived
valuation (Atakan, 2014). This explains why simply touching items may increase people's
willingness to pay for them (Wolf, et al., 2008). Because tangible products can be experienced
through multiple senses, they may create a stronger emotional connection than intangible
products (e.g., digital products or services. The sense of attachment may be increased by
touching and handling the product, as it can create a physical connection and make the product
feel more "real" and personal to the consumer. Product tangibility, however, does not guarantee
increased attachment. It depends on the personal preferences and the context of the product. For
example, a consumer may not feel attached to a product if they did not have any involvement in
its creation or purchase decision, or if the product does not meet their expectations.
Some of the evidence concerning the IKEA effect, have been retrieved from studies
where the assembly and design efforts were conducted virtually on a computer via a digital
design program (e.g., Franke et al., 2010; Ling et al., 2020), while other studies required
participants to physically create the product. In most studies where the item was 3D-printed,
participants received the item at the end of the process, offering them the experience of holding
the final creation (e.g., Walaseck et al., 2017). In other studies, the product was digital in nature,
17
such as a music playlist (e.g., Cloots, 2019), or an informative presentation (e.g., Rusho &
Raban, 2020), thus experiencing a final physical product was not feasible. Based on literature,
one can expect that due to a stronger sense of attachment and ownership of tangible products, as
opposed to digital products, the former will result in greater valuations.
H3: Self-invested effort has stronger influence on item valuation when creating a
physically tangible item versus creating it virtually.
Furthermore, there may be an interaction between the two variables relating to the
creation process–product tangibility and customization–as they may enhance or inhibit one
another. Put differently, creating a tangible item that requires customization could result in
stronger effects because it integrates both self-invested effort and self-expression via
customization. Similarly, we could expect that creating an intangible item that does not require
customization would derive weaker effects due to lower involvement in the product creation
paired with lower sensory arousal.
H4: There is an interaction between product tangibility and customization, such
that (a) greater valuation emerges for tangible customized products, while (b)
lower valuation emerges for non-tangible non-customized products.
Product Usage: Utilitarian vs. Hedonistic
People may value products differently based on whether their intended usage is utilitarian
or hedonic. Utility products are those that are primarily used to fulfill a functional need or solve a
practical problem. They tend to be purchased because they provide a specific benefit and are
typically evaluated logically, based on their performance and effectiveness in fulfilling their
intended purpose (Dhar & Wertenbroch, 2000; Voss, et al.,2003). Hedonic products, on the other
18
hand, are primarily used for pleasure or enjoyment. They tend to be purchased for their
emotional or psychological benefits, and are typically evaluated based on their design, brand, and
overall affective appeal (Dhar & Wertenbroch, 2000; Voss et al., 2003).
The two usage types are clearly not mutually exclusive, as some products can have both
hedonic and utilitarian benefits. For example, a smartphone can be used for communication,
information, and entertainment, and at the same time, it can be an indicator of social status and
its casing design can indicate personal aesthetic preference in addition to the utility of protecting
the phone. Although some products offer both utilitarian and hedonic uses, people tend to
categorize them as primarily one or the other (Dhar & Wertenbroch, 2000). More specifically,
hedonic products are often associated with subjective preferences and aesthetics, as they lend
themselves to a form of self-expression and extension of self (Maimaran & Simonson, 2011),
which, in turn, increases a sense of product attachment (Wiecek et al., 2020). In comparison to
utilitarian items, hedonic products foster greater feelings of product ownership because they
trigger emotional cues (Shu & Peck, 2011). Thus, one can assume that people experience a
greater sense of ownership for hedonic products compared to utilitarian products (Chan, 2015;
Shu & Peck 2011; Wiecek et al., 2020), likely because the former invokes a stronger emotional
attachment which in turn develops a stronger endowment effect (Chan, 2015).
The question arising from the IKEA Effect literature is whether self-created products
would produce a similar pattern of valuation favoring hedonistic products, or rather, would the
mechanisms that elicit emotional attachment and sense of ownership via the creation process
render self-created utilitarian products valuable as hedonistic ones. A recent study tested this
assumption and found that co-production increased positive attachment and valuations of
hedonic products, but not utilitarian products (Wiececk et al., 2020). However, the products they
19
tested in their study were 3D printed items, meaning that there was not much engagement on
behalf of the participants. Furthermore, the original study by Norton et al., (2012) found a
significant effect even for an explicitly utilitarian product such as storage boxes. Given the
limited literature assessing whether the IKEA effect is mainly relevant for hedonistic items or
whether it operates with utilitarian items alike, the final research question is posed:
RQ4: Does usage type (Utilitarian vs Hedonistic) of self-produced items moderate
the influence of effort on product valuation?
Method
Literature search process.
First, we searched for relevant empirical studies via electronic databases that concerned
the IKEA Effect, including Google Scholar, ProQuest, ResearchGate and the UW library catalog.
The queries identified academic journal publications, conference papers, book chapters, and
doctoral dissertations from a range of domains and disciplines. The terms in the queries that were
used to search the databases included: IKEA Effect, Willingness to pay (WTP), effort
justification, valuation, DIY, self-customization, and co-creation. In a second step, after
constructing an initial corpus, we conducted secondary queries within the articles that cited the
main articles in the initial corpus. Finally, we contacted authors (k = 11) for supplemental files in
case information was missing from their manuscripts which was critical for the models.
Inclusion criteria.
To be included in the meta-analysis, studies had to meet the following criteria. First, the
stimuli in the study had to require subjects to construct or customize a target object, whether it
was done physically or digitally via a computer program. We did not include studies with solely
envisioned efforts, such as asking subjects to imagine that they constructed an item (e.g., Atakan
20
et al., 2014b study 1; Erkin, et al., 2018 study 2). Second, the study had to assess valuation of the
target object as the dependent variable. While the focus was on assessments of monetary
valuation, we also included studies that measured other forms of assessments, such as liking,
perceived valence and preference fit. Third, studies had to include at least one of the following
comparisons: (1) a comparison of valuations between subjects who engaged in the creation of the
target object (i.e., builders) and subjects who reviewed a readymade object (i.e., non-builders), or
alternatively, (2) a comparison between two groups of “builders,” where the effort type was
manipulated. Four, each study had to report on the statistics necessary for calculating effect sizes,
(e.g., group means, standard deviations, sample sizes, t-values or F-values, counts, frequencies,
or exact p-values). In cases where the original manuscript was missing the necessary data to
evaluate effect sizes (k = 11), corresponding authors were contacted, and the relevant data were
obtained. After the screening process, 31 research reports that documented results of 55 separate
studies were included in the meta-analysis, with a total sample size of 5,454 participants [See
Appendix 1.A for literature search workflow]
Measures
Predictor: Customization/Effort manipulation
The studies were initially coded for whether self-invested effort was manipulated by
requesting subjects to 1= assemble a product, 2 = customize and design a product, or 3 = both
customize and assemble a product. Categories 2 and 3 were then combined into one, creating a
binary variable that indicated whether the process of item creation required customization or not.
Outcome Variables
Valuation. The effect sizes for valuation, including monetary value assessments,
willingness to pay (WTP), willingness to accept (WTA) and product evaluations, were combined
21
into the main meta-analysis model. In addition to valuation, we tested four main outcome
variables. These were (a) liking a product, which included liking and preference fit, (b) self-
concept, which included competence, pride, accomplishment, self-affirmation, self-concept, and
efficacy, (c) psychological ownership, which included perceived ownership, psychological
ownership, product attachment, product identity and affective commitment, and (d) positive
affect, which included, emotion, mood, satisfaction, and enjoyment. After testing these outcome
variables, we imputed them as moderators into the main meta-regression model to assess whether
they operate as significant mechanisms of the IKEA Effect—the influence of effort on valuation.
Contextual Moderators
In addition to the main mechanisms, the analysis focused on a series of moderators
relating to the process of effort and characteristics of the product that was created. These
included: customization, 3D printed items, product tangibility, and product usage type. The first
three were coded as binary variables, indicating whether they were present in the study or not.
For usage type, products were categorized as to whether their intended use was more hedonic (k
= 27), more utilitarian (k = 16), or equally both (k = 7). This coding was based on the validated
HED/UT scale developed by Voss et al. (2003). The scale includes ten semantic differential
response items, five of which refer to the hedonic dimension and five of which refer to the
utilitarian dimension of consumer attitudes. [See Appendix 1.B for codebook].
Inter-coder reliability
A subset of the final corpus (k = 24) was coded by two independent coders to evaluate
ICR, employing weighted Kappa to assess the level of agreement. The reliability coefficient
reached .878 or higher for all variables, indicating a sufficient level of reliability among the
coders. Disagreements were resolved through discussion between the coders and when needed, a
22
second round of coding was conducted on a different subset of data. The ICR scores were self-
invested effort (kappa = .919), product valuation (kappa = .919), liking (kappa = 1), self-concept
(kappa = 1), Psychological ownership (kappa = .893), positive affect (kappa = 1),customization
(kappa = .919), 3D printed item (kappa = 1), product tangibility (kappa = 1), and product usage
type (kappa = .878). Table 1 presents a summary of the article included in the analysis and a list
of outcome variables measured for each study
Data analysis
Effect sizes and correlation coefficients (Cohen’s d) were assessed via the
Comprehensive Meta-Analysis Software (CMA v.4; Borenstein, et al., 2014.), with weights
being assigned to primary studies based on the inverse of the variance. This study reports on
uncorrected effect sizes based on assumption of random effects (Hedges & Vevea, 1998). A
separate analysis was conducted for each of the outcome variables: (1) valuation, (2) liking, (3)
self-concept, (4) psychological ownership, and (5) positive affect (RQ1-2). After analyzing main
effects, statistical heterogeneity among the studies was estimated by Chi-square test at the p <
.05 level (Higgins & Thompson, 2002) and by using the I
2
statistic to quantify the heterogeneity
of the results (Higgins et al., 2019). Potential underlying mechanisms (i.e., RQ3) were tested
with a meta-regression (Rosenthal, 1991). The potential role played by contextual moderators
(i.e., H1-H4 and RQ4 ) was probed with Q statistic. Due to concerns over a potential file-drawer
problem, a combination of publication bias indices was used. These tests are described in detail
at the end of the results section.
25
Results
Efforts’ Influence on Product-Related Outcomes
Regarding the influence of self-invested effort on valuation (RQ1), across 45 individual studies,
the average effect was positive, moderate, and significant (d = 0.57, SE = .06, p < .005, 95% CI
[.45, .69]), such that builders tended to report on higher valuation compared to their nonbuilders
counterparts. Additionally, there was a substantial heterogeneity in effect sizes (Q(44) = 206.32,
I
2
= 78.67%, p < .005). Similarly, across 17 individual studies, the effect of self-invested effort
on liking (RQ2a) was positive, moderate, and significant (d = 0.57, SE = .10, p < .005, 95% CI
[.38, .75) with substantial heterogeneity in effect sizes (Q(16) = 57.48, I
2
= 72.17%, p < .005).
Moreover, the effect of effort on self-concept (RQ2b) was positive, strong, and significant, albeit
only across four studies (d = 1.17, SE = .30, p < .005, 95% CI [.58, 1.77) with substantial
heterogeneity in effect sizes (Q(3) = 21.24, I
2
= 85.88%, p < .005). With respect to psychological
ownership (RQ2c), across 13 studies, the IKEA Effect was again positive, strong, and significant
(d = 0.85, SE = .11, p < .001, 95% CI [.63, 1.07]) with substantial heterogeneity in effect sizes
(Q(12) = 55.70, I
2
= 78.46%, p < .001). Finally, an examination of the influence of effort on
positive affect (RQ2d) yielded a positive, strong, and significant effect (d = 0.76, SE = .19, p <
.005, 95% CI [.40, 1.12), with limited heterogeneity in effect sizes (Q(3) = 8.75, I
2
= 65.71%, p <
.005). [See table 2 for summary data of the five models; See figures 1-2 for a complete outline of
effect sizes on valuation and liking by study]. Taken together, the results associated with RQ1-
RQ2 paint a rather consistent portrait of the IKEA Effect, such that there seems to be substantial
influence on various product-related outcomes. With that in mind, however, there is also
noticeable variance in effect sizes that can be potentially explained with a closer look at theory-
driven moderators.
26
Figure 1 . Distribution of standard deviation in Means for the effect of labor on valuation.
27
Figure 2 . Distribution of standard deviation in Means for the effect of labor on liking.
Table 2.
The effect of self-invested effort on valuation and secondary outcome variables
Heterogeneity Statistics
Random Effects
k
I
2
Q
b
d(se)
95% CI
Valuation
45
78.67%
206.32**
0.57(.06)
0.45, .69
Liking
17
72.17%
57.48**
0.57(.10)
0.38, .75
Self-Concept
4
85.88%
21.24**
1.17(.30)
0.58, 1.77
Psychological Ownership
13
78.46%
55.70***
0.85(.11)
0.63, 1.07
Affect
4
65.71%
8.75**
0.76(.19)
0.40, 1.12
Note: CI = confidence interval. *p < .05, **p < .01, ***p < .001.
28
Four Potential Mechanisms of the IKEA Effect
Next, a meta-regression assessed whether the IKEA Effect on valuation (RQ3) is contingent on
the level of (a) liking, (b) self-concept, (c) psychological ownership, and (d) positive affect. To
begin, the meta-regression model did not support the role played by liking as a moderator of the
IKEA Effect on valuation (b = .17, SE = .36, p = .646, k = 17; Q(1) = 0.21, p = .646, R
2
= .00).
Likewise, self-concept did not emerge as a significant moderator of the IKEA Effect on valuation
(b = .22, SE = .33, p = .504, k = 4; Q(1) = 0.45, p = .504, R
2
= .00). Interestingly, however,
psychological ownership significantly amplified the IKEA Effect (b = .75, SE = .30, p = .012, k =
13; Q(1) = 6.26, p = .012, R
2
= .44). Each increase in psychological ownership translated into a
considerable increase in the impact of self-invested effort on valuation. Similarly, the meta-
regression pointed to the significant role played by experienced positive affect in enhancing the
influence exerted by the IKEA Effect on valuation (b = .57, SE = .23, p = .011, k = 4; Q(1) =
6.44, p = .011, R
2
= .64).
Contextual Moderators
We now progress to test the contextual moderators in our data and assess whether they
significantly impact the IKEA Effect–the influence of effort on valuation. Table 3 summarizes
these models. These moderators pertain to the process of item creation, characteristics of the
product, and demographic characteristics of the sample. Counter to the assumption of H1, the
meta-regression model did indicate that customization is a significant moderator of effort on
valuation (Q(1) = 0.027, p = .870). Similarly, counter to H2 and H3, the meta-regression model
did not find support for either 3D printing (Q(1) = 1.874, p = .171), nor product tangibility (Q(1)
= 0.769, p = .381), as playing significant roles in the IKEA Effect. Finally, RQ4 asked whether
there would be variance in valuations between usage types of products. The meta regression
29
indicates that there is no significant difference based on the intended usage of the product as
being more hedonistic or utilitarian (Q(1) = 0.083, p = .773), indicating that the IKEA Effect
operates with utilitarian products and hedonistic products alike.
Table 3.
The IKEA effect by contextual moderators (H1, H2, H3 & RQ4)
Variable
d
k
n
Q
p
95% CI
Main effect
.57
45
5,454
[.45,.69]
Customization
With
Without
.58
.56
26
19
3,141
2,313
0.027
.870
[.44,.72]
[.34,.77]
3D Printing
With
Without
.34
.61
7
38
785
4,669
1.874
.171
[-.02,.71]
[.48,.71]
Tangibility
Tangible
Intangible
.63
.53
18
27
1,790
3,664
0.769
.381
[.48,.78]
[.36,.70]
Usage type
Hedonic
Utilitarian
.60
.56
23
18
2,683
2,159
0.083
.773
[.47,.73]
[.30,.81]
Additional analyses tested H4 for a possible interaction between tangibility and
customization, resulting in a significant difference in retrieved effect sizes (Q(3) = 8.67, p =
.034). However, it was the tangible products without customization that elicited the greatest
effects, not the tangible with customization as initially expected. As illustrated in Figure 3, the
strongest IKEA Effect emerged for those who engaged with a physical product that was not
customized (d = 0.75, SE = .10, p < .001, 95% CI [.56, .95]), followed by a digital product that
was customized (d = 0.63, SE = .09, p < .001, 95% CI [.50, .81]), a physical product that was
customized (d = 0.44, SE = .10, p < .001, 95% CI [.24, .64]), and a digital product that was not
Customized (d = 0.29, SE = .16, p = .066, 95% CI [-.02, 59]). Critically, the post hoc
30
comparisons across conditions, further supported H4, showing a significant difference between
the physical assembly of a customized and a non-customized product (Q(1) = 4.71, p = .030).
The difference between the digital creation of a customized and a non-customized product was
not significant within the 95% CI, (Q(1) = 3.61, p = .058).
Figure 3. The influence of effort on valuation moderated by tangibility and customization (H4). Values by
bracket represent post-hoc tests for group differences; *= p< .05
Publication bias
We employ multiple methods to assess whether our corpus suffered from publication
bias. First, we calculate Kendall’s Tau to estimate the relation between standardized effects and
their variances (Begg & Mazumdar, 1994). By juxtaposing the effect sizes with their standard
error for each study, we produce a funnel plot that indicates a considerable clustering of studies
in the upper-left part of the funnel [See figure 4]. This rather a-symmetrical distribution in
the data suggests that studies with large samples that demonstrated stronger effects may be
underrepresented. Similarly, we see another, albeit smaller, cluster in the lower-right quarter of
the plot, indicating that small-sample studies with weak effects may also be underrepresented in
0.443
0.628
0.752
0.286
0
0.2
0.4
0.6
0.8
1
Tangible Non-Tangible
With customization Without customization
3.61
4.71*
31
the meta-analysis. Alternatively, we can employ the trim-and-fill technique, proposed by Duval
and Tweedie (2000). This nonparametric method is used to assess publication bias numerically
by compiling how many studies would need to be added and removed to achieve “ideal”
symmetry. Initially, small-sample studies are withdrawn from the model as they are usually the
culprits for asymmetry. In a second step, the plot-center is filled with omitted studies, then a
more accurate averaged effect is estimated. According to this method, 13 studies should be
trimmed from our model, resulting in a weaker average IKEA effect (d = 0.38, 95% CI [.25, .51])
compared to the initial effect observed. Finally, we can also conclude there is a possible
Figure 4. Funnel Plot for the Detection of a Potential Publication Bias
publication bias based on the fact that both Egger’s (1997) test for asymmetry (p < .001) and
Begg and Mazumdar’s rank-correlation test (p = .006) reached significance. Each of the different
methods noted a similar conclusion, that adding more unpublished research would result in a
32
lower average effect size. When considered collectively, the various publication bias estimates
suggest that the IKEA effect literature may consistently under-report weak effect outcomes.
Discussion
This is the first study to synthesize the collective data accumulated about the IKEA effect
and the impact of effort on valuations of DIY products. Overall, the IKEA effect represents an
important phenomenon within the domain of cognitive bias, that has significant implications for
both consumer behavior and decision making. Via the deep examination of extant literature, we
find that the average effect size of self-invested effort on valuation and liking of self-made
products is moderate, however, multiple tests of publication bias suggest that research may
systematically fail to report weaker effects and after adjusting the models for greater accuracy, the
average IKEA may not be as strong.
Underlying Mechanisms
In the second iteration, we found that self-invested effort was a strong predictor of
secondary outcome variables including self-concept, a sense of psychological ownership and
attachment, and experienced positive affect. However, when we utilized these variables as
underlying mechanisms in the meta-analysis model, only sense of ownership emerged as a
reliable significant predictor that the IKEA effect is contingent upon. Because both valuation and
liking of a product were similarly influenced by self-invested effort, and based on extant
literature, we intuitively presumed that they would be correlated with one another. However, the
data suggests that liking is not a significant predictor of valuation, indicating that we do not
necessarily like what we value, or more precisely, that we do not estimate value of a product
solely based on whether we like it.
33
Contrary to expectations, the data did not find support for effect being contingent positive
self-concept. We should note, however, that our analysis model only included four studies that
measured self-concept, therefore, more studies are needed for an accurate and generalizable
inference about the role of self-concept in the IKEA Effect. Furthermore, even though self-
concept did not emerge as significant, it does not suggest that self-concept does not play a major
role, but rather, that there may be an interplay with other factors impeding its significance which
should be addressed in future research. For instance, the relevance of self-concept as a
mechanism for the phenomenon may be limited, as it relies on feelings of success and
accomplishment. In the original study design, success and failure are possible outcomes as there
is a single correct solution, and no subjective preferences and aesthetics involved. However,
computer-generated products and designs do not allow for the possibility of failure, therefore the
completion of a final product may not indicate personal success. Self-concept may play a more
vital role when the task at hand is difficult and cognitively taxing because greater effort is needed
to trigger the psychological rewards associated with success. To test this assumption, future
IKEA effect studies should employ a task with varying levels of difficulty between conditions to
test whether self-concept is significant in the difficult level but not in the easier level and enable
the option to fail.
In support of expectations, the comprehensive meta-analysis suggests that the overall
effect of effort on valuation was contingent upon psychological ownership, indicating this
mechanism significantly facilitates the phenomenon across studies. Our results fall in line with
assumptions related to the endowment effect which links ownership with inflated valuation, and
the results highlight sense of ownership as a key factor leading effort to valuation. This runs
counter to a study by Marsh et al. (2018) who concluded ownership was not a major mechanism
34
but rather that self-concept is likely the main component linked to valuation. A potential
explanation for this discrepancy is that Marsh et al., (2018) examined the IKEA effect via a
developmental approach, that is, their sample was composed of children aged 3 to 6-years-old,
which were excluded from our analysis. Even though children develop a sense of ownership and
attachment very early on, it may be argued that a sense of accomplishment and affirming their
self-concept is more prominent at that stage of development. Furthermore, since our corpus was
limited to studies that systematically measured self-concept, we can only attest to the role played
by psychological ownership. To address this, future studies should compare a similar task
between groups of different ages in order to test whether different levels of mental-cognitive
development are bound by the same processes. Additional studies that systematically measure
these concepts are needed to fine-tune the optimal conditions for each distinct mechanism.
Contextual Moderators
Customization did not emerge as a significant predictor of valuation though there was an
interesting interaction with tangibility which will be discussed shortly. Though we expected that
the ability to customize a product would increase its perceived value due to greater product
attachment and self-expression, the data indicated no difference between the two forms of
participation, suggesting that customization efforts may not be necessary in order to increase
valuation outcomes. These findings indicate the need for more studies that specifically test the
impact of customization on valuation of self-made products, specifically since research has
exhibited discrepancies in outcomes. For instance, Fink and Geldman (2017) tested the IKEA
effect in the context of software and found that willingness to pay (WTP) was lower when
subjects were required to self-design the program but found no difference in WTP between
builders and non-builders. Atakan et al. (2014a), on the other hand, found that customization
35
resulted in greater valuation, even under low-effort requirements. Diefenbach et al., (2018)
examined differences between self-designers and self-creators of tote-bags and found that both
forms of involvement lead to greater emotional attachment and valuation of the final product.
Ling et al, (2020) found that offering greater customization choices lead to increased valuations
but that this effect was moderated by an individual’s aesthetic preferences and level of self-
esteem. Future studies that are interested in the influence of self-expression via customization
may want to test different contexts and specifically compare self-designers to self-builders
Counter to expectations, tangibility, and customization, on their own, were not significant
predictors of valuation. However, there was a significant interaction between the two, which may
further explain these results. Customization of physical products seems to reduce valuation,
while an opposite pattern emerged for non-tangible items, that is, customization increased
valuation for digital products that were created online. This may be explained by two aspects.
First, there is a spectrum of design capabilities and self-efficacy among people. Some individuals
may accurately assess their designs as of lower quality. However, when designing products
online via computerized programs then personal talent and capabilities do not matter. The image
of the final product will seem professional either way because there is no possible way to error.
Another way to look at the data is that when customizing the product, there is not a significant
difference between tangible and intangible items, however, when there is no customization
involved, then tangible items are more positively valued, while the valuation of non-tangible
items decreases. Put it simply, if we focus solely on studies without product customization, then
the data supports H3 since tangibility of products indeed elicited more positive valuations.
The use of 3D printing as a proxy for assessing low effort did not emerge as a significant
moderator for the relationship between effort and valuation. We expected that this form of low
36
effort would not prompt the extent of consumer-product engagement which is required to foster
an emotional bond a DIY product. Perhaps there is a halo effect concerning 3D printing, which
amplifies the valuation of the relatively novel technology that has rapidly developed over the last
few years. Its increasing popularity is a manifestation of a larger trend of emerging technologies
that signal a fundamental shift in the way people participate in co-producing goods and services
(Fisher , 2022; Wiececk et al., 2020). Even though on paper we could argue that these 3D studies
required the lowest level of effort and engagement, what matters is how the participants
perceived their engagement. More specifically, though they were only required to press a button
as part of their creation effort, 3D printing technology enables people to design and create their
own product at home, reducing their dependence on traditional manufacturers, which renders
them as active co-producers, which is the opposite of passive consumers.
Finally, the data indicated no significant difference between the valuation of hedonistic
and utilitarian usage, and none of the demographic variables emerged as significant predictors of
the relation between self-invested effort and product valuation.
Concluding Implications
Approaching the results of this meta-analysis from a broader perspective, a more
disconcerting interpretation arises. Apart from the significant interaction between tangibility and
customization, none of the contextual moderators in the meta-regression yielded significant
findings. According to these data it can be assumed that the consistency of the phenomenon
across studies and products is indicative of its robustness traversing cultural, generational, and
geographical contexts, however, this view may be oversimplified. Alternatively, one could claim
that if the IKEA Effect is in everything, then it may be in nothing, or that the studies are picking
up on another process. While effort is presumed to enhance valuation, the data may, in fact, be
37
demonstrating that a sense of ownership is what drives the phenomenon, and that effort is
peripheral, meaning that effort, whether physical or mental, may simply serve as a trigger for a
prompting sense of ownership, rendering the process as a form of the endowment effect. This is
important because many studies that claim to test the IKEA effect may be manipulating different
biases. To reconciles the two potential inferences, and to enhance the specificity and accuracy of
studies investigating the IKEA effect, we suggest that future research should consider narrowing
the definition phenomenon, and perhaps focus on adhering more closely to the original
experimental design (e.g., Mochon et al., 2012; Norton et al., 2012), and continue to test
additional boundary conditions highlighting when the IKEA effect does and does not occur
optimally.
End Note
1. The models assessing whether the IKEA effect was contingent upon self-concept and positive
affect found the former insignificant and later as significant. However, it should be noted that
these models only included four studies for each of these mechanisms. Therefore, these results
should be taken with a pinch of salt, and rather than indicating them as definite major or
peripheral factors, this is a testimony for the need for additional studies that systematically assess
and compare the potential role that these mechanisms play in the relationship between invested
effort and valuation of self-created products.
38
CHAPTER 2. From the Ikea Effect to Self-Persuasion
The preceding chapter systematically deconstructed the phenomenon known as the IKEA
effect, illustrating the fundamental mechanisms and moderating variables that shape the impact
of self-invested effort on product valuation. With this deeper understanding of the IKEA effect
and its boundary conditions, the current chapter can begin to address the main inquiry of this
dissertation, which is assessing the IKEA effect framework within the context of persuasive texts
as a method for self-persuasion. Thus, in order to integrate the IKEA effect into a self-persuasion
model, it befits that we first undertake a parallel exposition of the extant scholarship on self-
persuasion. The following section reviews the main theories and underlying mechanisms
associated with self-persuasion research, focusing on the aspects relevant for the current research
goals. The review is followed by a discussion on the similarities and distinctions between the
classic self-persuasion methods and the proposed integrated model, highlighting its contribution,
as well as potential weaknesses that should be addressed in the consecutive study design.
Self-persuasion
The theory of self-persuasion was developed with the notion that receivers of a message
play an active role in influencing their own attitudes (Aaronson, 1999). Persuasion literature, in
general, examines the various factors that influence attitude change, including characteristics of
the message, of the source, and of the target audience (Hovland et al., 1953). Within this vast
literature, self-persuasion focuses on instances in which people’s behaviors and thoughts
influence themselves, that is, instances in which the target of persuasion is also its source. As a
method of influence, self-persuasion entails creating conditions in which individuals are
motivated to modify their own beliefs or behavior (Aaronson,1995). This approach represents a
potent means of stimulating attitude change and compliance, with the potential for long-term
39
lasting results (Aaronson, 1995). Even stable states such as self-esteem have been shown to be
malleable though self-persuasion techniques (Jones, 1990). Furthermore, self-persuasion may
prompt greater attitude and behavioral changes compared to traditional forms of direct
persuasion, as individuals perceive the motivation for change to be internal (Aaronson, 1995;
Hovland et al., 1953). It may seem counter intuitive that behavior can influence attitudes,
because we tend to think of the relation between attitude and behavior in a manner that the
former determines the latter. Knowledge about attitudes is often used to predict behavior, and if
we wish to change someone’s behavior, we will likely attempt to influence their attitude and
thoughts regarding the matter. Self-persuasion studies, however, have demonstrated the existence
of a complementary direction of influence, in which an attitude is deducted from, or adjusted to,
a behavior. For example, patients who received a diagnosis in a noisy clinic environment and had
to actively lean toward the physician to concentrate on their words, subsequently rated the
diagnosis and treatment plan as more valuable compared to patients who received their diagnosis
in a quiet room (Reference, ??). People who tried to persuade a fellow colleague to quit smoking
reported more negative attitudes towards smoking compared to people who listened to someone
else advocating against smoking (Elms, 1966). Even experienced debaters who were randomly
assigned to advocate for one side or the other of a topical motion, later indicated greater
confidence and support for the position they had advocated for (Schwardmann et al., 2019)
The influence of behaviors on attitudes
The mechanisms by which behaviors influence attitudes are complex and multifaceted,
with two prominent psychological processes being self-perception and cognitive dissonance.
People desire to understand the cause of their behavior. At times, they use their own behavior as
an indicator to determine their thoughts and feelings (Olson & Stone, 2005). Self-perception
40
occurs when individuals interpret their own behavior after-the-fact, as a means of understanding
their attitude towards a particular target object or issue (Bem, 1967; 1972). If people repeatedly
engage in a certain behavior and experience positive outcomes, they may conclude that they are
doing so because they must like that activity, and infer their attitude accordingly (Bem, 1965).
Even involuntary behavior is used as an indicator of attitudes, as demonstrated by
research showing that people who were instructed to nod their head vertically while reading
persuasive arguments (Briñol & Petty, 2003), or listening to an editorial radio broadcast (Wells
& Petty, 1980), tended to be more agreeable with the arguments compared to individuals who
were instructed to shake their head horizontally. People who watched cartoons and were
instructed to hold a pen in their mouth in a manner that activated muscles associated with
smiling, later rated the cartoons as funnier (Strack et al., 1988). Children who were punished for,
or restricted from, playing with a certain toy later rated it less favorably, even though they had
rated the same toy as desirable prior to the restriction (Zanna, et al., 1973). In general, people are
not very good at determining the cause of their own actions. They often do not consider external
factors that influence their decision-making, and they tend to misattribute their behavior to
internal motivation, believing it is a result of personal choice (Brock, 1968). Insufficient
justification refers to instances in which external factors are not registered by individuals as
plausible causes for their behavior and it plays a significant role in attitude change via self-
perception (Brock, 1968; Festinger & Carlsmith, 1959).
Insufficient justification is also a key factor in attitude change via cognitive dissonance,
which is the second main process by which behavior can influence attitudes (Lawrence &
Festinger,1962). This theory posits that people seek to maintain internal consistency between
their cognitions and behaviors, and when they perceive a discrepancy between the two, they
41
experience a state of discomfort (Festinger, 1957). The discomfort is proportional to the
magnitude of the inconsistency and can be heightened when the behavior is made public with
strong commitment, as it increases the sense of scrutiny over one's actions (Brehm & Cohen,
1962). Public deliberation, for instance, can trigger a sense of commitment based on the principle
of consistency, which stems from the psychological desire of individuals to maintain consistency
in the way that others perceive them, and in the way they see themselves (Cialdini, 2009; Jensen
& Carter, 1981). Cognitive dissonance can lead to attitude changes as individuals seek to reduce
the discomfort caused by inconsistencies between their behavior and attitudes. One of the
primary strategies for reducing cognitive dissonance is to add a consonant cognition that explains
the discrepancy, also known as external justification. When there is not sufficient external
justification for one's behavior, the next best strategy is changing the attitude that is inconsistent
with the behavior (Festinger & Carlsmith, 1959; Harmon-Jones & Mills, 2019). Festinger and
Carlsmith (1959) asked participants to engage in a tedious task and then to tell prospective
participants that the study was interesting, for which they were randomly rewarded with either $1
or $20. Though all the participants likely experienced a discrepancy between their behavior
(lying about the study) and their positive self-concept (someone who does not lie), participants
who received the lower reward were more likely to rate the study as enjoyable. Those who
received $20 could justify their behavior by attributing it to a worthwhile external cause.
Conversely, $1 was not sufficient to justify the behavior, which resulted in attitude change.
In summary, insufficient justification is a crucial factor in both processes because
individuals perceive their behavior as an indicator about themselves. However, the self-
perception process entails deducing an attitude based on behavior, whereas cognitive dissonance
involves adjusting an existing attitude to a behavior, thus, a main difference between the two is
42
that self-perception does not require a prior existing attitude, while cognitive dissonance can only
occur if there is an existing attitude (Suedfeld, 1968; Wilson, 1990).
Self-persuasion via role-playing
The use of role-playing, or improvisation, has been demonstrated to be a powerful tool in
self-persuasion research in a series of studies during the 1950’s (Hovland et al., 1953; Janis &
King, 1954; King & Janis, 1956). Individuals who play a role of advocating for a certain position
often end up convincing themselves in the process (Aronson, 1999; Janis & King, 1956). In one
of the studies from their series, Janis and King (1954) asked participants to take part in a
discussion group on different issues. In the high involvement condition participants were asked
to give a persuasive speech while improvising their own arguments [self-persuasion]. In the low
involvement condition participants listened to a speech by someone else in the group [direct
persuasion]. Participants who generated their own arguments showed greater attitude change in
line with the position they had advocated for. In a follow-up study, King and Janis (1956)
instructed participants to advocate for a position oppositional to their own view concerning
mandatory military service. Again, in the high involvement condition [self-persuasion], they
were asked to improvise their own arguments, while in the low involvement condition [direct
persuasion] they read a speech that had been prepared by someone else. Though the speeches in
both conditions were comprised of relatively similar arguments, the results indicated that
participants who self-generated arguments experienced greater attitude change. The authors
asserted it noteworthy that participants’ attitudes underwent a considerable shift toward a
direction contrary to their previous standpoint, solely due to their active role playing of
advocating for that particular side of the issue (King & Janis, 1956). Subsequent studies
replicated this design to examine the influence of behavior on attitude change in a diverse array
43
of issues, including political attitudes (e.g., Watts, 1967); education related policies (e.g.,
Greenwald & Albert, 1968), health-related behavior such as smoking or eating habits (e.g., Elms,
1966; Janis & Mann, 1965; Li, et al., 2019; 2020b); environmental awareness (Damen, et al.,
2015); pro-social behavior such as tipping (Bernritter, et al., 2017), and prevention measures to
reduce sexual aggression (Wong, 2018). More recently, this method has been tested in promoting
motivation for voluntary social isolation during the covid-19 pandemic (Drążkowski et al. 2020),
and reduce alcohol consumption e.g., (Loman, 2019). Role-playing methods have been found to
exert influence not merely on attitudes, but also on behavioral intentions (e.g., Lee & Muller,
2020), and actual real-world behavior (e.g., Müller et al., 2009).
The power of self-generated arguments
A key factor in all the role-playing studies mentioned above, is people generating their
own arguments, whether by actively attempting to persuade someone else via discussion, writing
a persuasive essay, listing arguments, or simply thinking about arguments in response to an
open-ended question. Why are self-generated arguments so successful at generating attitude
change? Self-persuasion facilitated by self-generate arguments offers several advantages over
traditional direct-persuasion methods. Generally, people do not like being told what to do, let
alone feel like they are being manipulated. Conventional persuasion attempts, which involve
externally imposed influence, can trigger individuals to engage in self-guarding strategies that
may hinder the effectiveness of the persuasive message (Dillard & Shen, 2005; Mussweiler &
Neumann, 2000). These strategies may include internal resistance to persuasion, such as
reactance, or bolstering and counter-arguing (Briñol et al., 2004; Crawford et al., 2002).
Reactance, a cognitive-affective response to perceived threats to our freedom of choice,
may lead to intentional avoidance of the position advocated for in the persuasive message
44
(Brehm, 1966; Dillard & Shen, 2005). The degree of response is affected by both the perceived
level of threat to one's freedom and an individual's susceptibility to reactance as a personality
trait (Dillard & Shen, 2005). However, self-persuasion methods are less likely to elicit reactance
because the target of influence is also the source of influence (Briñol, et al., 2012). As such,
individuals are unlikely to perceive external limitations to their freedom when they generate their
own arguments.
Counter-arguing is another self-guarding strategy in which one elicits arguments that
counter the persuasive message in an attempt to correct and oppose information provided by
external sources. Counter-arguing is especially likely if the issue of debate is relevant and
important to the message receiver because he or she is more attentive and knowledgeable on the
topic (Petty & Cacioppo, 1986). However, similar to reactance, people are unlikely to feel the
need to counter-argue with themselves as there is less motivation to correct self-generated
information (Mussweiler & Neumann, 2000). Additionally, individuals preparing to advocate for
a particular side of an issue tend to systematically seek arguments that support their position,
often overlooking, filtering out, or dismissing oppositional arguments that counter their position
(Janis & Gilmore, 1965; Janis & King, 1954). Biased scanning may lead to a lack of cognitive
effort spent on counter arguments that challenge the position, while focusing solely on
supporting arguments, consequently rendering the latter more salient (Janis & Gilmore, 1965).
Self-tailored messaging is the third main advantage of self-generated arguments within
the context of self-persuasion. When individuals self-generate arguments, they tend to come up
with reasoning that is most compelling to them, based on their personal cognitive process and
values (Briñol, et al., 2012, Zimbardo, 1963). Within this internal debate, they also formulate
arguments that challenge the vulnerabilities that they are most familiar with (Zimbardo, 1963).
45
Though the intended goal may be to persuade other people, by self-generating arguments,
individuals effectively customize arguments to persuade themselves. Through this process of
thought construction, individuals partake in an internal dialog in which they may formulate new
reasoning and cognitions that further influence their own attitudes (Pingree, 2007).
In summary, the literature suggests that the power of self-persuasion stems from dynamic
advocacy and self-generated arguments, underscoring the substantial impact of active
engagement on shaping attitudes and behavioral intentions. By allowing individuals to internally
motivate attitude change through role-playing and self-generated arguments, self-persuasion
methods overcome the challenges associated with conventional direct-persuasion techniques,
such as reactance and counter-arguing, ultimately leading to more compelling persuasion without
activating self-guarding strategies. The personalized nature of self-persuasion, driven by
individual cognitive processes and values, enhances the effectiveness of persuasion while
minimizing resistance. It should be noted that while self-generated arguments are clearly a
powerful tool to induce attitude change, their effectiveness, as with most forms of persuasion, are
influenced by various personal, cultural, and environmental factors. The mere salience of others’
attitudes versus one’s own attitude can influence attitude change (Shaffer & Tabor, 1980). These
relevant moderators will be discussed in greater detail later in the chapter, when comparing the
similarities and differences between self-persuasion and the IKEA effect frameworks.
Self-persuasion and Social Media
Over the past two decades, we have witnessed vast developments in communication
technologies, with a steadily growing share of social communication migrating into online
spaces and mobile connectivity. It becomes important to adjust research environments
accordingly, as we attempt to pragmatize findings from lab experiments and adapt them into the
46
real word settings. Social media emerges as a prominent environment for inducing self-
persuasion, and several factors contribute to its efficacy in this regard. Firstly, the visibility of
content to others on social media aligns seamlessly with the principles of commitment and
consistency (Cialdini, 2009). When individuals engage in self-persuasion by creating and sharing
self-generated arguments online, the visibility of this behavior enhances its influence on
attitudes, as it is observed by others within their social network. This unique feature of social
media distinguishes it from other platforms, providing a potential advantage for self-persuasion
methods. Second, the interactivity along the visibility among audience members on social media
further amplify its impact compared to other media platforms. Research suggests that self-
persuasion effects may be more pronounced online than offline due to the specific characteristics
of the online environment (Boyd, 2011; Valkenburg, 2017). Valkenburg (2017) argues that
various affordances of social media make self-effects more frequent and influential, and thus,
more likely to be included as part of a communication processes compared to similar offline
settings. Empirical studies support the idea that self-persuasion in online settings can lead to
significant attitude changes. For instance, studies examining negative behavior in online
interactions found that behaving negatively towards a discussion partner resulted in negative
attitudes toward the discussed issue (Walther et al., 2010). Similarly, role-playing in online
environments, such as video games, demonstrated that participants' attitudes were positively
influenced by the group they were assigned to represent (Alhabash & Wise, 2015). Self-
generating arguments and posting them on an interactive mobile app has even been found to
promote healthier eating habits (Stavrositu & Kim, 2018). The act of posting online reviews has
been shown to influence the poster's evaluations, particularly when the meaning associated with
47
review post holds high validity (Kim et al., 2021). Collectively, these factors underscore the
unique and potent role of social media in fostering self-persuasion processes.
Integrating the IKEA Effect into a Self-Persuasion Model
Building upon the foundations of this review, we can now progress to examine the
similarities and differences between self-persuasion and the IKEA effect frameworks, while
exploring the potential utilization of the latter as a means for former. As elaborated earlier, the
IKEA effect refers to a cognitive bias in valuation of DIY consumer goods toward which the
individual develops a sense of accomplishments, psychological ownership and emotional
attachment following the personal investment of time and mental effort. On the other hand, self-
persuasion via roll playing refers to the phenomenon in which individuals tend to adjust or infer
their own attitudes following a scenario in which they were instructed to generate arguments or
actively attempt to persuade others in support a specific position concerning an issue of conflict.
Thus, upon an initial examination, the two frameworks may seem unrelated, giving the
impression that their integration would be arbitrary [figures 5 and 6]. However, when
considering the substitution of the DIY product with a persuasive text, as depicted in Figure 7,
the integration of the two frameworks begins to reveal its rationale. In this model, assumptions of
the IKEA effect would encompass the processes influencing the valuation of the text with
relation to the effort invested into its construction, while assumptions associated with self-
persuasion would encompass the influences concerning attitudes toward the issue mentioned in
the text as well as scrutiny of the arguments comprising the text.
A wider perspective further reveals that the two frameworks share a few intersecting
underlying mechanisms. Personal investment and effort justification are relevant to the two
domains as both rely on the effort heuristic. In the case of the IKEA effect, an individual invests
48
Self-generating
arguments
Self-perception
Inhibition of
self-guarding strategies
Attitude change
Cognitive dissonance
Figure 6. Self-persuasion via role playing – main mechanisms
Self-assembling a
product
Attachment &
Psychological ownership
Accomplishment &
Self-concept
Product valuation
Effort justification
Figure 5. IKEA effect model - main mechanisms
Self-assembling a
persuasive Text
Self-persuasion
mechanisms
IKEA effect
mechanisms
Attitude change
Text (product)
Valuation
Figure 7. Integrated model for the IKEA effect and Self-persuasion in persuasive texts
49
time and effort in assembling a product, which may lead to a sense of ownership and attachment
to the product. Similarly, in self-persuasion, an individual invests time and effort in persuasion
efforts, which may lead to a sense of ownership and commitment to the expressed belief or
behavior. Since individuals strive for consistency between their beliefs and their behavior, this
can lead to attitude change.
Additionally, both the IKEA effect and self-persuasion rely on the concept of cognitive
dissonance, which occurs when an individual experiences conflicting thoughts or beliefs
(Festinger, 1957). In the case of the IKEA effect, the effort invested into creating or assembling a
product may generate a dissonance between the effort invested and the perceived value of the
product, leading to an overestimation of its value. Similarly, in self-persuasion, an individual
may experience cognitive dissonance when their attitudes or beliefs are inconsistent with their
behavior, leading them to change their attitudes or beliefs to align with their behavior. Engaging
in certain activities such as advocating for a certain position can also trigger attitude change as a
strategy to reduce the experience of cognitive dissonance (Festinger, 1959), for instance,
laboring on the assembly of a text with which content one does not agree with. Thus, a task that
would produce the greatest shift of attitude should elicit effort that requires psychological
justification, and the content of the persuasive text would need to be dissonant to an extent from
the individual’s pre-existing attitude.
While there are similarities and overlapping processes between the two, there are also a
number of critical distinctions between the common methods of self-persuasion which task
participants to generate arguments in advocacy of a certain position, and the proposed method
for text assembly from given arguments. How does “assembling” a text differ from self-
persuasion studies in which participants were instructed to write a persuasive essay or list
50
arguments in favor of a certain position? Most studies on self-persuasion via role-playing
compare attitude change between conditions in which subjects generate their own arguments,
and conditions of direct persuasion in which subjects are exposed to arguments from an external
source. The IKEA effect framework would focus on the creation of a product by following
specific instructions and selecting features from a predetermined list of preferences and
components. By tasking subjects to assemble a paragraph from predetermined arguments, this
design offers an integration between the two. In this manner, the IKEA effect can introduce an
additional perspective by offering a midpoint between self-generated arguments and externally
provided arguments. Individuals would engage in the creation of a persuasive message, without
actually generating their own arguments. Exploring the IKEA effect through the lens of self-
persuasion allows for a nuanced distinction between two key factors that are often intertwined in
self-persuasion research: self-expression and engagement.
Moderators of self-persuasion and the Ikea effect
We should also consider factors that moderate persuasion and how they may impact the
processes in the proposed integrated model. There is an abundance of potential moderators,
though this section will focus on factors that are relevant for testing the IKEA effect within the
context of persuasive texts. To begin, the impact of self-generated arguments on attitude change
is contingent upon the subjective relevance and perceived importance of the issues of debate, as
in as with most contexts of persuasion (Sarup, 1981). Motivation and involvement with the issue
are important factors in persuasion as they determine the path of information processing and
therefore impact the level of scrutiny given to arguments (Petty & Cacioppo, 1986). In role-
playing studies, it is likely that self-persuasion occurs through a high-elaboration mechanism
because participants actively generate and express their own arguments and hold a desire to
51
perform well (reference, year). However, it is less clear which path of processing will occur via
assembly of a text from predetermined arguments, and whether engaging with the arguments to
construct a paragraph is sufficient to prompt systematic processing of their content.
Another key moderator is positive self-esteem, which can reduce the likelihood of
attitude change by mitigating the negative feeling triggered by cognitive dissonance, even if the
reason for increased self-esteem is not related to the behavior or the topic of persuasion (Steele,
1988). When an individual’s feeling of self-worth increases, they feel less need to justify their
behavior or attitude, and therefore less likely change their attitude (Steele, 1988). Conversely,
threats to self-esteem can increase the magnitude of the dissonance-reduction effect, and thus
increase the probability of attitude change (Heine & Lehman, 1997). From the lens of the IKEA
Effect, however, a sense of accomplishment and positive self-concept following a successful
product assembly, have been associated with greater valuation of the self-constructed product.
Therefore, if the content of the persuasive text stands in opposition with an individual’s world
view and values, the experience of cognitive dissonance may be diminished by the positive
feelings deriving from the success of completing the task.
Need for expression is a relevant personality trait for positive self-esteem in this context
because people who have a high need for expression may find it easier and more rewarding to
generate persuasive arguments, leading to greater confidence in their beliefs and potentially
greater persuasion (Susmann, 2021). Alternatively, positive emotions have been associated with
greater ease of persuasion and processing information via the peripheral path because individuals
in a positive affective state are not motivated to scrutinize arguments systematically and tend to
assess information from a more holistic view (reference). According to the emotions-as-frames
perspective, emotions guide us to focus our attention on information associated with a certain
52
emotion and process it accordingly (Nabi, 2003). However, in the proposed utilization of the
IKEA effect framework, there is no self-expression because individuals will not be generating
their own arguments, nor will they communicate the arguments to a potential audience.
Therefore, we may expect this model might result in weaker attitude change effects compared to
more traditional self-persuasion methods. Furthermore, the extent of effort invested into the
persuasion attempt can potentially mitigate its success. Brinol et al., (2012) found opposite
results when participants were asked to advocate for pro-attitudinal as opposed to counter-
attitudinal views and when the target of persuasion was self or others. The authors concluded that
the results stem from the extent of effort people think they need to invest compared to the actual
effort they put in when trying to convince others.
Cultural differences may also play a role in the effectiveness of self-persuasion, as studies
have demonstrated a variance in experience of cognitive dissonance across cultures (Li, 2020a).
In collectivistic cultures, self-persuasion may be less effective due to a greater emphasis on
group harmony and social norms, which can lead to a reluctance to change one's own attitudes in
favor of group norms (Li, 2020a). Individuals from collectivistic cultures show dissonance
effects when they focus on their relationships with others rather than on evaluation of themselves
(Kitayama et al., 2004). Furthermore, the strive for positive self-concept may not be a universal
phenomenon and could be culturally variant. A study of Japanese culture found that a self-
critical focus is more characteristic of Japanese individuals, suggesting that the need for positive
self-concept stems from distinctive aspects of western cultures (Heine et al., 1999). In
comparison to self-persuasion, the IKEA effect has been demonstrated to be a robust global
phenomenon. A recent study compared the occurrence of the IKEA effect between children from
the UK and India and found no significant differences among the two samples (Marsh, 2022).
53
Additionally, the meta-analysis of the IKEA effect (chapter 1) compared the effect sizes among
55 studies from around the globe and found that the relationship between effort and valuation
was not contingent upon geographical region. Thus, while there may be variance in the
experience of cognitive dissonance, the tendency to over-valuate and feel attached to objects we
created spans across cultures.
Final Notes for Study Design
In order to design a study that adequately tests the model, we need to consider the long
list of aforementioned mechanisms and assumptions that guide the two frameworks. Originally
applied in the context of consumer products, the IKEA effect involves the process of assembling
an explicit product from predesigned components following specific instructions. In the context
of persuasive texts, the “product” refers to a cohesive and grammatically sound passage adhering
to linguistic rules. Based on the original study design (Norton et al., 2012), the process would
require constructing a specific text using predesigned arguments, comparable to constructing a
puzzle made from fragments of text. Consequently, this means that individuals would not
generate their own arguments nor determine the order in which the arguments are organized
within the logical flow of the text as its structure is predetermined. This design is crucial from
the perspective of the IKEA effect framework, as it enables us to compare valuations of the exact
same product, where the main variance between subjects is whether they participated in its
assembly or not. The absence of self-generated arguments diverges from the aforementioned
studies on self-persuasion and the strengths associated with role-playing because the individual is
not the direct source of the arguments. Nonetheless, the act of constructing the arguments into a
meaningful text could render individuals as co-creators of the text. The text assembly task fosters
active engagement with the text as it requires individuals to read the arguments in order to
54
reconstruct a meaningful and grammatically correct text, meaning that they would be exposed to
the arguments and actively engage in the creation of the persuasive text, albeit not generating the
arguments themselves. From the perspective of self-persuasion literature, this level of
engagement may be sufficient to induce attitude change, because ultimately, the efficacy of self-
persuasion methods derives from people genuinely considering arguments supporting the
opposing side (Greenwald, 1969).
The impact of self-generated arguments on attitude change is contingent upon the
subjective relevance and perceived importance of the issues of debate (Sarup, 1981). Motivation
and involvement with the issue are important factors in self-persuasion as they determine the
path of information processing and therefore impact the level of scrutiny given to arguments
(Petty & Cacioppo, 1986). In role-playing studies, it is likely that self-persuasion occurs through
a high-elaboration mechanism because participants actively generate and express their own
arguments, especially in instances in which participants individuals expect to express their
arguments in front of an audience. However, it is less clear which path of processing will occur
via assembly of a text from given arguments, because individuals will not conjure arguments
themselves. Moreover, although individuals would not be self-generating arguments, they would
be investing mental effort and time into the cognitive task of reconstructing the text and
organizing the arguments into the correct order. As noted earlier, investing effort into the
creation of an item can cultivate a sense of psychological ownership and emotional attachment
with it, which arises from the process of becoming acquainted with the item, exerting control
over it, and investing oneself in it (Pierce, 2001). By developing a sense of attachment to the text
they labored on, they might feel like the message was partially generated internally. A final
factor to consider is that the task is cognitively taxing, meaning that their cognitive capacity is
55
likely to be occupied with the structure of the text and completing the task, leaving fewer
cognitive resources available for self-guarding strategies. Thus, it can be presumed that laboring
on the assembly of a persuasive text would not trigger self-guarding strategies such as reactance
and counterarguing, at least not to the same extent that would emerge following direct persuasion
attempts. Since self-guarding strategies are a main inhibitor of persuasion, it may be
consecutively presumed that inhibiting them is likely to result in greater persuasion in alignment
with the content of the persuasive text.
To conclude, the integration of the IKEA effect into a self-persuasion model may offer a
compelling approach to understanding the cognitive processes involved in attitude change. While
initially seeming unrelated, the DIY nature of the IKEA effect can be effectively translated into
the domain of persuasive texts. The common threads of personal investment, effort justification,
and cognitive dissonance weave a connection between the two frameworks, providing a basis for
their integration. The shared reliance on the effort heuristic, personal investment, and cognitive
dissonance suggests an overlap in the underlying mechanisms of the two frameworks. However,
critical distinctions arise in the methods of self-persuasion, particularly when comparing the
generation of arguments through role-playing to the assembly of a text from predetermined
arguments. This distinction introduces a unique perspective, offering a midpoint between self-
generated and externally provided arguments. The study design considerations highlight the
importance of constructing a specific text using predetermined arguments, akin to assembling a
puzzle. While the absence of self-generated arguments differs from traditional self-persuasion
methods, the active engagement required in text assembly may foster attitude change. The
cognitive effort invested in reconstructing the text, coupled with a potential sense of attachment,
could mitigate self-guarding strategies, thereby enhancing persuasion.
56
CHAPTER 3: The IKEA Effect in Persuasive Healthcare Messages
This chapter presents data from three empirical studies designed to test primary research
goals of this dissertation which inquired: Would laboring on the assembly of a persuasive text
lead to greater valuation of the text, and in turn, would this greater valuation translate into
greater support of its persuasive content? Therefore, the initial task is to assess whether
individuals would indeed value a text more positively if they engage in its assembly compared to
the valuation of the same text had it been crafted by someone else, and whether they differ in
their support for the position advocated for in the text.
Study 1
Self-invested Effort and Text Valuation
The IKEA effect model encompasses psychological mechanisms that explain how and
why effort invested in the assembly of a DIY product leads to a corresponding increase in the
valuation of that product. The current study aims to test the integrated model within the context
of persuasive messages. As extrapolated in the previous sections, is possible to surmise that
individuals who successfully construct a text from pre-given arguments would value the outcome
of their effort more positively based on the effort heuristic (Kruger, et al., 2004) and specifically
the IKEA effect (Norton et al., 2012). Although text may not be easily commodified in the same
manner as consumer goods, it can still be evaluated based on its logical coherence,
informativeness and ability to engage the reader, thus we could expect that actively engaging in
its construction would elevate perceptions regarding the content of the text and its structure.
Moreover, though text in itself is not easily commodifiable, the time and effort invested into the
construction of the text can be readily translated into monetary value, similar to assessing the
value of editing labor. This logic suggests that individuals who engage in the assembly of the
57
text would perceive their labor as more valuable compared to individuals who read a text that
was constructed and edited by someone else. Based on the literature concerning the influence of
effort on valuation, the following hypothesis are posed:
H1a.: Assembling a text, as opposed to solely reading it, will result in a more positive
evaluation of the text.
H1b. Assembling a text, as opposed to solely reading it, will result in a more positive
valuation of the assembly labor.
Text Engagement and Attitude change
Assembling a text from individual arguments and sentence fragments requires a certain
extent of cognitive and mental investment in order to create a meaningful passage with logical
flow and correct grammar. The ensuing consequences of such an investment, beyond the
presumed positive valuation of the text, is the potential enhancement of its persuasive efficacy.
Research on persuasion and message characteristics suggests that a message which is perceived
as strong and coherent has a greater probability of influencing attitudes (Dillard et al., 2007).
Furthermore, if an individual’s attitude does not reside with the message they labored on, they
may experience cognitive dissonance associated with investing effort into assembling a message
they disagree with (Festinger, 1957). This discomfort, depending on its extensity, could
potentially prompt individuals to adjust their attitudes to align with the message to justify their
investment of effort and reduce dissonance. Drawing on these theoretical perspectives, it can be
postulated that individuals who invest effort into assembling a persuasive text will be more likely
to manifest attitude change consistent with the message conveyed in the text, compared to those
who simply read the same text without engaging in its assembly. Accordingly, the following
hypotheses are posed.
58
H2. Assembling a persuasive text, opposed to solely reading it, will result in
greater attitude alignment with the conveyed message.
H3. The influence of effort on persuasion will be mediated by valuation.
Text Engagement and Self-Guarding Strategies
The efficacy of self-generated arguments in persuasion arises from two principal
advantages over conventional direct-persuasion methods. First, self-persuasion methods are less
inclined to provoke self-guarding strategies such as reactance or counter-arguing because the
individual generating the arguments also serves as the influence source, thus minimizing the
likelihood of perceiving external limitations on personal choice (Dillard & Shen, 2005). Second,
when individuals generate arguments, they tailor the message to themselves, developing
reasoning that they find most convincing based on their personal values and distinctive cognitive
processes (Briñol, et al., 2012). The IKEA effect framework tested in this study is not parallel to
common self-persuasion studies in which individuals are asked to generate their own arguments
in advocacy of a certain position. Nor is it equivalent to direct-persuasion methods in which
individuals are simply exposed to persuasive content from an external source. Rather, it resides
on the intersection between the two, drawing on both their strengths and weaknesses. To clarify,
the stimulus of this study requires participants to assemble a deconstructed paragraph by
reordering the fragments back into a grammatically correct text with a logical flow. This means
that they do not come up with arguments themselves, however they are engaged with the text
creation rather than consuming it passively. From the aspect of the IKEA effect, if individuals
invest effort into the creation of a text, they are likely to develop a greater sense of psychological
ownership and responsibility towards it, even though they do not “design” the arguments
59
themselves. From the aspect of self-persuasion, engaging in the text creation may inhibit the
sense of an external persuasion effort.
Taken together, one can assume that engaging in the assembly of a persuasive text, as
opposed to simply reading the same text as in direct persuasion, is less likely to trigger self-
guarding strategies like reactance and counterarguing, which are known to impede the
effectiveness of persuasive messages. After all, the efficacy of self-persuasion methods lies in
their ability to prompt individuals to genuinely consider arguments supporting the opposing
viewpoint (Greenwald, 1969). People do not like being told what to do, thus if individuals feel
like the text is trying to persuade or manipulate their opinion, they are more likely to experience
negative affect while reading the text, and rate high levels of induction, or threat to freedom of
choice (Dillard & Shen, 2005). In this context, induction refers to the provocation of a defensive
or resistant response, where individuals feel their freedom of choice is being threatened or
undermined, leading to an increase in counter-arguing or resistance against the persuasive
message or influence (Dillard & Shen, 2005). This would likely result in negative evaluations of
the text as well as reduced likelihood of agreeing with the message conveyed in the text.
According to the literature, the following three hypotheses are posed:
H4. Assembling a text, opposed to solely reading it, will result in lower levels of
perceived threat to freedom of choice
H5. Greater levels of perceived freedom threat will be associated negatively with
attitude alignment with the persuasive message.
H6. Greater levels of perceived freedom threat will be associated negatively with
text evaluations.
60
Method
Sample & Study Design
A sample of 153 undergraduate students (81.5% female), ages ranging from 18 to 24 (M
= 19.47, SD = 0.9) participated in the study for extra credit. The study employed an online
randomized experiment with a 2 (task type: Assemble vs Read) by 2 (healthcare topics:
(Sunscreen or Flossing), between subject design. In the read condition (i.e., control group),
participants read a short paragraph advocating one of the two healthcare practices. In the
assemble condition (Ikea effect), participants viewed one of the same two paragraphs, however,
the order of the sentences was distorted, and participants were tasked to reconstruct a coherent
paragraph with a logical flow by dragging the sentences into the correct order. The outcome
variables included evaluations of the text (e.g., logical; persuasive) and monetary valuations of
the editing work, as well as attitude towards the healthcare practice advocated for in the text that
they read or assembled. [See appendix for comprehensive stimuli material and questionnaire].
Developing the stimuli material
Two pilot studies (N
1
= 40; N
2
= 147) were conducted among undergraduate students to
test the stimuli material as well as determine the preferred method for the text assembly task.
Originally, three potential texts were developed for the stimuli, pertaining to different healthcare
practices: applying sunscreen, teeth flossing and getting the Covid-19 vaccination. The
information comprising the texts were taken from the Skin Cancer Foundation website
(www.skincancer.org) , and the CDC webpages (www.cdc.gov/coronavirus/2019-ncov). The
data from Pilot 2 (79.8% female; M
age
= 18.95 (0.93)), indicated a ceiling effect for attitudes
concerning the Covid-19 vaccine for both pre- and post-attitude measures. This likely occurred
for two reasons. First, it was a student sample which means the vast majority was vaccinated as
61
part of the requirement to attend in-person classes. Second, the data was collected during the
height of the pandemic thus attitudes toward the vaccination, and relief from the lockdown, were
already positive to begin with. Thus, it is unsurprising that nearly all respondents marked the
highest value on all scales. Due to the lack of variance among participants, this healthcare issue
was not included in the stimuli of the main studies.
For the text assembly task two methods were tested. To complete the text, participants
were asked to either fill in missing words from the paragraph, or alternatively, reorder sentences
to reconstruct a disassembled paragraph. There was not a significant difference between the two
assembly methods concerning valuation and other outcome variables. Though filling in missing
words could potentially offer more pragmatic use for future development, especially when
considering adaptation of the method to contemporary media environments, it was eventually
determined to use the sentence-reordering method for the main studies because it resembled the
process of product assembly more closely to that used in IKEA effect studies.
Procedure
Participants received an email with an informed consent form and a link directing them to
the study on Qualtrics. After clicking the link, all participants filled a pre-attitude questionnaire
comprised of items concerning the two healthcare practices that will appear in the stimuli (teeth
flossing and sunscreen application), as well as filler items relating to other healthcare practices
such as the importance of reducing alcohol consumption, wearing a seatbelt, and getting the
Covid-19 vaccine. In the next step, participants were randomly assigned to one of the four
conditions, to either read or assemble a paragraph concerning one of the two health care
practices. Upon completing the assembly task, participants received a notification saying
“Success! You constructed the paragraph correctly.” The notification of success was
incorporated for two reasons. First, it was intended to boost a sense of accomplishment following
62
a successful task completion, as this is one of the main mechanisms associated with the IKEA
effect (e.g., Cloots, 2019; Mochon et al., 2012). Second, and more critically, it attempted to
resolve an issue from the pilot studies in which participants were not certain whether the text
they submitted was correct. In pilots, the stimuli offered multiple possible solutions which were
all accepted as correct. It was initially assumed that multiple solutions would make the task
easier, however, for some participants this turned out to be the opposite. Because there was no
indication as to which solution that they submitted was correct, participants who were engaged
with the task longer were less certain about their choice. They rated the assembly task as more
difficult and perceived their performance as worse compared to others. They also tended to rate
the text more negatively compared to their read-only counterparts. In the main experiments the
settings were changed so that only one solution was accepted. Following the stimuli, all
participants filled the same questionnaire assessing the dependent variables and the long list of
relevant control variables. Participants in the assemble conditions responded to an additional
question asking them how difficult they found the task of assembling the paragraph. The time
spent on the stimuli was tracked in all four conditions, whether reading or assembling the
paragraph. This value was used as an indicator of engagement and effort invested in the task.
Measures
Pre-attitude. These items assessed attitudes toward the healthcare treatments from the
stimuli (i.e., flossing teeth, applying sunscreen), as well attitudes toward filler issues, which were
intended to distract from of main issue (e.g., reducing alcohol consumption or wearing a
seatbelt). Participants responded to Likert-type items such as “applying sunscreen is effective,”
and “daily application of sunscreen is important” on a 7-point scales, ranging from 1 “Strongly
disagree,” to 7 “Strongly agree.”
63
Post-attitude. Post-stimuli attitudes toward the healthcare practice advocated for in the
text were assessed via similar Likert-like items on a 7-point scale, ranging from 1 “Strongly
disagree,” to 7 “Strongly agree.” These items included “daily application of sunscreen or
flossing) is good,” “daily application of sunscreen is unnecessary,” “daily application of
sunscreen is harmful,” “daily application of sunscreen is foolish,” “daily application of sunscreen
is desirable.” Negative items such as “wearing sunscreen is harmful” were reverse coded.
Parallel items were included for post attitudes toward flossing. These items were adapted from
Dillard and Shen (2005) who assessed these as semantic-differential scales. The scales were split
and converted into Likert type scales so that the pre-attitude and post-attitude could be easily
compared without creating a salient repetition that could result in participants adjusting their post
items based on their responses to the pre items. A factor analysis indicated that the items for
sunscreen application loaded onto one dimension, except for “_____ is desirable.” The related
items were averaged into a single value of post attitude toward sunscreen (Cronbach’s α = .81).
A parallel process was done to create a post attitude score for flossing (Cronbach’s α = .76)
Attitude change. This measure was calculated by deducting the average value of the pre-
attitude measure from the average value of the post attitude measure. Greater positive values
indicate a positive attitude change in line with the persuasive message, while negative values
indicate an attitude change oppositional to the message. Zero would indicate no change.
Text Evaluation. Evaluation of the message was assessed via six Likert items on a 7-
point scale, ranging from “Strongly disagree,” to “Strongly agree.” The items asked participants
to rank their level of agreement as to whether the paragraph they read or worked on was logical,
interesting, believable, confusing, informative, and irritating. The items read “Do you agree or
disagree that the paragraph you read is ________.” Negative items (i.e., confusing, irritating)
64
were reverse coded. A factor analysis indicated all items loaded onto one dimension, except for
“the paragraph was interesting.” The five remaining items were averaged into a text evaluation
scale (Cronbach’s α = .82).
Labor Valuation. Participants were asked to rate the appropriate payment for the editing
work put into assembling the paragraph, on a 5-point multiple-choice question with response
options ranging from $1 to $20. For the participants in the assembly condition the items read:
“Imagine you were working as an editor for an online health magazine. What would be the
appropriate payment for the editing work you invested into constructing the paragraph?” For
participants in the direct persuasion condition, the item read: “This paragraph was edited for an
online health magazine. What would be the appropriate payment for the editing work invested
into constructing the paragraph?”
Freedom threat. This scale was used to assess whether there was a difference between
the conditions of assembly and direct persuasion in the experience of threat to personal choice.
This index is based on the validated measure from Dillard and Shen (2005) where respondents
ranked the extent to which they agree or disagree on four 7-point-Likert scales ranging from
“Strongly disagree,” to “Strongly agree,” including the statements such as “The message in the
paragraph tried to make a decision for me,” or “…tried to manipulate me.” A factor analysis
indicated that three of the items loaded onto one dimension. These were then averaged into a
perceived freedom threat value (Cronbach’s α = .725)
Political ideology. Participants indicated where they place themselves on two political
ideology scales concerning social and economic values. The scales range from 1 “strongly
liberal” to 7 “strongly conservative”. The score from both scales was averaged into a combined
political ideology scale. The sample was skewed to the liberal ideology with a median of 2.5, and
65
nearly 80% ranking themselves between strong liberal to somewhat liberal. The scale was then
folded onto itself, creating a 4-item political polarization scale, greater values indicating a more
polarized political ideology.
Issue relevance. Participants stated whether they, or someone they know, suffered from
health issues relating to the topic raised in the persuasive message that they read. Response
options were yes, no, and I don’t know. The wording for this item was “Have you or anyone of
you family members or friends suffered from health issues related to sun exposure/gum health?”
Demographic variables. Basic demographic data was collected including age, gender,
and ethnicity. Education was not assessed because this was an undergraduate student sample.
Data analysis
The mediation hypotheses were assessed using Hayes (2022) Process Macro version 4.0,
a widely used tool in the field of mediation and moderation analysis which focuses on assessing
the strength and significance of varied indirect effects in mediation models and moderation
effects in interaction models. Process does not have strict assumptions about the distribution of
data and generally estimates robust standard errors. The syntax allows for dichotomous
independent variables, and also enables users to define multicategorical independent variables
and moderators if necessary. However, it is always important to consider the general
assumptions related to regression and mediation analysis such as linearity, heteroscedasticity and
normal distribution of the residuals. To test the hypotheses, the current study utilizes Process
models 4 and 6, which enable assessment of parallel mediation and sequential mediation tests
accordingly. The bootstrap sample iterations were set 5000 within a 95% confidence interval.
For all mediation models, the experimental condition is set as the independent variable, and post-
attitude as the dependent variable. Monetary valuation, text evaluation, and perceived threat to
66
freedom of choice are set as mediators. Demographic variables, including political ideology and
gender, are then imputed as control variables alongside issue relevance and pre-attitudes, (See
figure 8 for proposed mediation model).
Results
For clarity purposes, and to maintain consistency with the IKEA effect literature,
participants in the text-assembly treatment group are referred to as "builders" (or “b” for short).
Similarly, participants in the control group who solely read the text without engaging in its
assembly are referred to as "non-builders" (or “nb”). All the models were first tested for the
entire sample, and then separately for each healthcare subsample.
Entire Sample
The analysis revealed a significant influence of assembly effort on text evaluation, (t =
2.948, MSE = 0.35, p < 0.001), however, the direction of influence was negative, contrary to
H4
H1b
Assembly of
Persuasive Text
Freedom threat
Post Attitude
Text evaluation
Figure 8. Model for mediation hypotheses testing
H2
H3b
H5
H6
Labor valuation
H1a
H3a
67
expectations of H1a. Participants in the assembly condition (builders) tended to rate the text
slightly less positively compared to their non-builder counterparts (M
b
= 4.89 (0.57), M
nb
,=
5.14(0.65)). Interestingly though, the data reveal support for H1b which presumed greater
monetary valuation for self-invested effort, (M
b
= $9.49; M
nb
= $7.06; t = 1.73, MSE = 31.42, p =
.032), indicating that participants in the assembly condition valued the labor invested into
creating the text more than their non-builder counterparts. Contrary to H2, neither the direct
effect (p = .081), nor the total effect (p = .222) of assembly effort on attitudes was significant.
The path predicting posts-attitude was significant (t = 2.46, MSE = 0.76, p < .001). Self-invested
effort was not a statistically significant predictor (p = .11), nor was valuation (p = .647). Rather,
it was text evaluation that emerged as a significant predictor of attitudes (t = 2.64, se = .12, p =
.009), with more positive text evaluations associated with greater attitude alignment.
Additionally, even though the total and direct effects were not significant, the model indicated
that there was a statistically significant indirect effect of effort on post-attitudes via text
valuation, offering partial support for H3. The next two hypotheses assumed that assembly
efforts would reduce the perception of threat to freedom of choice, which in turn, should
translate to greater agreeableness with the message. The data did not indicate that assembly
efforts influence perceived freedom threat (p = .64), but perceived freedom threat did emerge as
a significant predict of post attitudes, with greater levels of perceived threat associated with less
message agreeableness (t = 2.85, se = .06, p = .005), thus offering support for H5 but not for H4.
Finally, perceived freedom threat emerged as a significant predictor of text evaluation (t = -3.24,
se = .04, p = .002) as greater levels of perceived threat were associated negatively with text
evaluations, offering support for H6. Figure 9 presents the correlation coefficient and statistical
significance produced by the multiple mediation model .
68
Upon incorporating the relevant control variables into the model, a few interesting
changes were observed, though the indirect influence of effort on attitudes via text evaluations
remained significant (t = 2.78, MSE = 0.679, p = .000). The path predicting perceived freedom
threat was now significant (t = 2.4, MSE = 1.30, p < .001), though it was not due to self-invested
effort (p = .88), but rather, it was political ideology that emerged as a significant predictor of
perceived threat to freedom of choice, revealing that participants who identify more strongly as
liberal tended to perceive less freedom threat from the text they read (t = 4.58, se = 0.08, p =
.000). Considering post attitudes as the outcome, both gender (t = -3.96, se = 0.18, p = .0001)
and political ideology (t = -2.47, se = 0.06, p < .05) emerged as significant predictors, with
individuals identifying as liberal and female expressing greater support for the healthcare
practice advocated for in the text (t = 2.69, MSE = .74, p = .000). It should be noted that the
sample was predominantly female (78.4%). Furthermore, the sample was heavily skewed as the
.093
2.3*
Assembly of
Persuasive Text
Freedom threat
Post Attitude
Text evaluation
Figure 9. Mediation model of entire sample - Experiment 1
Note. *p < .05, **p < .01, 95% confidence interval
-.18
0.006
-.128**
-.172**
Labor valuation
-0.24*
.319**
69
vast majority identified as left leaning, with 79.3% categorizing themselves from very Liberal to
somewhat Liberal, and an additional 10% identifying as moderate (or neither Liberal nor
Conservative). Based on this distribution on the political spectrum, we cannot make concrete
inferences comparing conservatives to liberals, however, the data would suggest that there is a
meaningful difference between moderate liberals and those who identify strongly with liberal
ideology. A more representative sample would be required to make accurate inferences about
gender and political ideology as moderators. Notably, issue relevance did not emerge as a
significant predictor of text evaluations or attitudes (p = .07 ; p = .63). Unsurprisingly, the
strongest predictor of post attitudes was pre-attitudes (t = 7.11, se = 0.06, p = .000). Upon adding
pre-attitudes as control variable, the influence of gender and political ideology on attitudes were
no longer significant (p = .08 ; p = .06), and neither was the indirect path of influence of effort on
attitudes. In a final model iteration, post attitudes were replaced with attitude change as the
outcome variable, but this model was not significant. Taken together, these results suggest that
there may be more nuanced influences at work which deserve further investigation. Furthermore,
it should be noted that even though some differences between the conditions are statistically
significant, the variations are rather small, and the averages for both groups reside in the upper
half of the scales. Next the model was replicated for each of the two healthcare practices.
By healthcare subsample
For Sunscreen application (n = 75), the significant patterns of influence were parallel to
that of the full sample. Assembling the text was a significant predictor of valuations (t = 2.08,
MSE = 0.83, p < 0.01), with individuals who assembled the text tending to value the editing of
the text more positively than their non-builder counterparts (M
b
= $10.84 ; M
nb
,= $6.87), while, at
the same time, expressing lower evaluations for the text itself (M
b
= 4.90; M
nb
,= 5.19). Likewise,
70
assembling the text was not a significant predictor of freedom threat (p = .243), while freedom
threat was a significant predictor of text evaluation (t = -2.48, p = .015). Higher text evaluations
and lower levels of freedom threat were both associated with greater attitude alignment with the
persuasive text (t = 2.88, p = .005; t = -2.405, p = .019). Adding the control variables to the
model yielded similar results, as gender (t = 2.87, p = .005), as well as pre-attitudes (t = 6.78, p =
.000), significantly predicted post attitudes. Concerning the Flossing sub-sample (n = 78) similar
patterns emerged for the influence of text assembly on valuation, however, the model was not
significant, (t = 1.1, MSE = 0.673, p < 0.01). The total effect model only appeared significant for
this subsample after including pre-attitudes as a control variable , (t = 1.57 , MSE = 0.617, p <
0.05), which was the only significant predictor of post attitudes, and even that effect was
relatively weak small. Table 4 presents a summary of descriptive statistics by treatment group.
Table 4
Sample means and standard deviations of outcome variables by treatment group – Study 1
Full Sample
Sunscreen
Floss
Total
assemble
control
assemble
control
assemble
control
Labor valuation
$9.49
($6.78)
$7.06
($4.12)
$10.84
($7.33)
$6.87
($4.47)
$8.21
($6.04)
$7.26
($3.80)
$8.27
($5.72)
Text evaluation
4.89
(0.57)
5.14
(0.65)
4.90
(0.45)
5.19
(0.61)
4.89
(0.67)
5.10
(0.69)
5.02
(0.62)
Freedom threat
3.49
(1.21)
3.39
(1.21)
3.46
(1.13)
3.14
(1.24)
3.51
(1.30)
3.65
(1.15)
3.44
(1.21)
Post attitude
5.88
(0.92)
6.14
0.92)
5.74
(0.97)
6.07
(1.05)
6.01
(0.86)
6.21
(0.78)
6.01
(0.93)
Attitude change
0.25
(1.16)
0.13
(0.75)
0.09
(0.82)
0.16
(0.87)
0.39
(1.40)
0.11
(0.61)
0.19
(0.97)
n
76
77
37
38
39
39
153
71
Discussion Study 1
Considering evaluation of the text, results were significant but in the opposite direction of
expectations. Those who invested effort into assembling the paragraph tended to evaluate it more
negatively and they also tended to disagree with its persuasive message compared to participants
who solely read the paragraph in the direct persuasion condition. At the same time, however,
those same individuals who engaged in the text assembly tended to value the editing labor more
positively, offering partial support for the efficacy of the IKEA effect with relation to text as a
DIY product. This discrepancy resonates with a similar disconnect in the meta-analysis (Chapter
1) indicating that liking is not necessarily correlated with valuation of a self-labored item.
It is particularly intriguing that many of the relationships between induction and attitude
change were significant but in the opposite direction of what was expected, as it stands in
contrast with assumption grounded in self-persuasion literature. which suggests that direct
persuasion methods would lead to lower evaluations and less agreement. A straightforward
interpretation of these results would be that this form of message creation may not be perceived
as significantly different from direct persuasion methods and might not be sufficient to induce
compliance because it triggers self-guarding strategies. It is also possible that factors such as
interruptions to processing fluency hinder attitude change in this context. However, prior to
making such a conclusion, it would make sense to delve into methodological factors, aiming to
identify and address potential confounding factors that emerged from the study design. That way
the follow-up studies could rectify these issues and gain a deeper understanding of the
underlying dynamics. This approach would contribute to resolving the contradictions and
enhancing the overall coherence of the findings.
72
Notably, pre-attitudes emerged as the strongest predictor of post-attitudes, regardless of
experimental condition, indicating that any observed attitude change was relatively minor. There
are several possible explanations for this finding. Firstly, it is plausible that the persuasive
message employed in the study was not effective enough to induce substantial attitude change. .
Alternatively, it could be that participants' attitudes were already firmly established and
relatively stable on the issue at hand. This is supported by the high level of agreement found in
positive support for healthcare treatment among participants. While the data did not exhibit a
complete ceiling effect, there was a notable concentration of responses in the upper half of the
attitude scale. Another contributing factor could be the presence of consistency bias, where
participants strive to maintain consistency between their responses to the pre- and post-attitude
items. Ideally, collecting pre-attitude data at an earlier time point would minimize potential
biases introduced by participants attempting to maintain consistency. Unfortunately, due to
constraints with the student sample used in this study, collecting data at an earlier time was not
feasible. In the subsequent studies this issue was addressed by either omitting the pre-attitude
assessment (Study 2) or replacing it with questions pertaining to real-life behavior as an
alternative indicator of pre-attitude assessment methods that do not resemble the post-attitude
measure too closely (Study 3).
Another potential issue with the study design is the difference in the text format between
the control and treatment groups. The absence of a "final product" in the assembly condition may
have hindered the influence of accomplishment on valuation and persuasion. Unlike their non-
builder counterparts, participants in the assembly condition did not see a coherent final version
of the paragraph after submitting their solution Although they read the sentences to reorder them
correctly as part of the assembly task, they did not have the opportunity to view a completed
73
version of the paragraph they created. Furthermore, even after reorganizing the arguments,
participants may have experienced a lack of fluency due to interruptions in reading. As noted
earlier, fluency plays a significant role in text evaluation, and it could potentially explain why
self-invested effort in text assembly resulted in lower evaluations of the text. This potential issue
will be addressed in Study 2 by showing participants the finalized version of the paragraph after
they receive notification of their success.
Study 2
The objective of this study, as the preceding one, is to evaluate the applicability of the
IKEA effect in persuasive texts. The design and procedure closely replicate those of Study 1,
with minor, albeit essential adjustments made to the stimuli and empirical process to address
concerns identified in the previous experiment. The current investigation tests the same
hypotheses using a new dataset collected specifically for the present study. Consequently, Study
1 could be considered as Pilot 3, representing the concluding exploratory studies before initiating
the primary research studies.
Method
Sample, Design & Procedure
Ninety-five undergraduate students (76.3% female), in age range of 19 to 23 (M
= 19.72,
SD = 0.87) participated in the study for extra credit. The study employed an online randomized
experiment with 2 (text-tasks: Assemble vs Read) by 2 (healthcare: Flossing teeth or Sunscreen
application), between subject design. The modifications from the previous study were
implemented in the stimuli of the two assembly conditions. First, instead of breaking up the
paragraph into individual sentences, the paragraph was broken in the middle of the sentences,
resulting in text fragments that were not stand-alone arguments. This created an actual puzzle
74
with only one possible solution, as all the pieces of the paragraph needed to be placed in the
correct order for the text to appear grammatically correct and be coherently readable. This
revision was necessary to address the issue from the previous stimuli, which offered more than
one solution that could be assessed as grammatically correct. The second revision was prompted
following the assembly task. After receiving the “success” notification, participants were
presented with the complete paragraph, parallel to the version viewed in the direct-persuasion
conditions. This was done to exemplify the finalized “product” in the form of a complete
paragraph, and to ensure that the only difference between the control and treatment conditions
was the assembly manipulation. Even if participants did not read the entire paragraph again
front-to-back, it offered them an opportunity to visualize the successfully completed paragraph
on which they labored, rather than a clump of separate sentence pieces.
Measures
This study utilized the same measures as the preceding study, encompassing post
attitudes (Sunscreen Cronbach’s α = .859; Floss Cronbach’s α = .745), text evaluation
(Cronbach’s α = .768), labor valuation, perceived freedom threat (Cronbach’s α = .56), issue
relevance, political ideology (single item), and the basic demographic variables. However, there
were a couple of adjustments—attitude change was not assessed due to the omission of the pre-
attitude measure. Furthermore, the item assessing the valuation editing labor underwent revision.
Instead of a 5-point multiple-choice question, participants were presented with a slider, enabling
a continuous selection ranging from $1 to $21.This adjustment aimed to provide a more precise
assessment of perceived monetary valuation.
75
Results
As in Study1, the model was initially tested for the entire sample comparing between all
participants in the assembly and read conditions. Then the model was tested again for each of the
healthcare subsamples separately.
Full Sample
Figure 10 presents the correlation coefficients produced by the mediation model. In
accordance with expectations of H1b, text assembly was a significant predictor of labor valuation
(t = 3.92, p < .001), with participants in the assembly condition rating the editing labor more
positively than their non-builder counterparts (M
b
= $12.07, sd = $4.69; M
nb
,= $8.60, sd = $4.35;
t = 2.09 , MSE =.54, p = .016). Text assembly was also a significant predictor of perceived
freedom threat (t = -2.61, p < .05), with individuals in the assembly condition expressing lower
levels of perceived threat compared to non-builders (M
b
= 3.09, sd = 0.8; M
nb
,= 3.63, sd = 1.16; t
= 2.61 , MSE = 1.02, p < 0.05), providing additional support for H4. The model predicting text
evaluation was significant, (t = 2.09 , MSE = 0.54, p = .016), though the direct path from text
assembly to text evaluation was not significant (p = .197), offering no support for H1a. Rather, it
was perceived freedom threat that significantly predicted text evaluations (t = -2.9, p < .01), with
greater levels of perceived threat associated with less positive text evaluations (t = 2.09 , MSE =
0.54, p < .05), indicating support for H6. Concerning post attitudes as the outcome variable, the
influence of text evaluation was significant (t = 3.56, p < .001), with more positive text
evaluation associated with stronger support for the message conveyed in the text (t = 2.25 , MSE
= 0.56, p = .001), offering partial support for H3. The model found a significant indirect path of
influence via perceived freedom threat and text evaluations, in a manner that text assembly
resulted in less perceived threat, which lead to greater text evaluations, which in turn, were
76
associated with greater text-attitude alignment, offering additional support for H5. However,
counter to the assumption of H2, neither the total effect nor the direct effect of text assembly on
attitudes were significant (p = .2; p = .23).
Upon incorporating the relevant control variables into the mediation model, we find that
issue relevance was a significant predictor of perceived freedom threat, with those whom for the
healthcare issue was relevant, perceiving less threat to their freedom of choice (t = 2.35, p < .05).
Gender was a significant predictor of text evaluation, with female participants rating the text
more favorably (t = 2.37, p < .05). Finally, political ideology was a significant predictor of post
attitudes, with those identifying more strongly as liberal also expressing more support for the
benefits and importance of the healthcare practice (t = -2.05, p < .05). As in the previous model,
the total and direct effect of text assembly on attitudes remained insignificant, but more than that,
the indirect path via perceived freedom threat and text evaluation was no longer significant,
.-.499*
3.303 **
Assembly of
Persuasive Text
threat to freedom
Post Attitude
Text evaluation
Figure 10. Mediation model of entire sample - Study 2
Note. *p < .05, **p < .01, ***p < .001, 95% confidence interval
-.242
-.007
-.222*
-.099
Labor valuation
-.243
.0.374 ***
77
suggesting that the variance in attitudes was mostly explained by the control variables, or
alternatively, that there were contradictory influences that cancelled each other out.
By Sub-sample
The model testing the mediation for the Sunscreen subsample was not as strong. Text
assembly still significantly predicted monetary valuation (t = 2.64, p < .05). Likewise, text
evaluation was still a significant predictor of post attitudes (t = 2.02 ,p < .05), however, other
than these two paths of influence, the data did not reveal any significant effects, neither for the
direct nor indirect effects of text assembly on post attitudes. For the subsample assessing
Flossing, the observed patterns closely mirrored those of the overall model. Assembly effort
significantly predicted levels of perceived threat to freedom of choice, and increased freedom
threat associated with less favorable evaluations of the text. As expected, higher text evaluations
Table 5
Sample means and standard deviations by treatment group – Study 2
Entire sample
Sunscreen
Floss
Total
assemble
control
assemble
control
assemble
control
Labor
valuation
$12.07
($4.69)
$8.60
($4.35)
$11.56
($4.15)
$8.65
($4.06)
$12.70
($5.32)
$8.56
($4.65)
$10.24
($4.81)
Text
evaluation
5.57
(0.81)
5.65
(0.72)
5.66
(0.75)
5.77
(0.67)
5.45
(0.89)
5.55
(0.77)
5.61
(0.76)
Freedom
threat
3.09
(0.80)
3.63
(1.16)
3.26
(0.80)
3.52
(1.19)
2.87
(0.77)
3.72
(1.15)
3.37
(1.04)
Post
attitudes
5.85
(0.86)
6.07
(0.77)
5.77
(0.93)
6.17
(0.75)
5.95
(0.76)
5.98
(0.78)
5.97
(0.81)
n
45
50
25
23
20
27
95
78
correlated with greater alignment of attitudes with the persuasive text. While the correlations that
emerged align with those of the full sample, it is noteworthy that the indirect influence of text
assembly on attitudes was not significant. This could be attributed to a limited statistical power
stemming from the sample size. Table 5 provides summary statistics by sample.
Discussion Study 2
The findings so far support the assumption that the IKEA effect occurs within the context
of text assembly, with individuals perceiving higher valuations for the editing labor of text they
assembled compared to the same text edited by an external source. However, assembling the text
does not seem to result in greater evaluations of the text itself. With regard to attitude change,
text assembly efforts did not result in greater alignment with the message conveyed in the text.
Notably, the influence of assembly on induction was significant and in the expected direction,
suggesting that that those who labored on text assembly perceived less threat to their freedom of
choice compared to those in the direct persuasion conditions. Even though the influence on
attitudes was not significant, this is a meaningful finding because inhibition freedom threat
perceptions is one of the main strengths of self-persuasion methods. It consequently results in
reduced self-guarding strategies such as reactance and counterarguing, which enables the
individual to be more open to consider alternative information. Furthermore, as in experiment 1,
responses for many of the outcome variables were concentrated in the upper half of scales, which
would also contribute to the insignificant effects. This lack of variance could derive from a
combination of a relatively homogeneous and a non-controversial topic which they mostly
agreed upon. Since the respondents preexisting attitudes already tended to align in support of the
healthcare practice, it would be difficult to induce attitude change in greater support. The results
and inferences of this study are elaborated on in further detail in the general discussion.
79
Study 3
This concluding experiment provides a comprehensive investigation of the primary
research questions, drawing upon the insights and considerations discussed in the all the
preceding sections of this dissertation. In addition to addressing the main research questions, the
current study enhances the model’s applicability within a more real-world setting and develops
the design by incorporating two additional manipulations concerning the extent of effort and
self-expression via customization. In this manner, the current study explores potential boundary
conditions that can enhance our understanding of the interplay between engagement with
persuasive content, text valuation, and attitude change.
Effort and Valuation
As previously discussed, the IKEA Effect is a phenomenon in which individuals tend to
assign higher value to products that they assembled or co-created. This effect can be attributed to
the Effort Heuristic, which posits that the extent of effort invested into a task directly influenced
the perceived value of its outcome (Kruger et al, 2004). Effort is often used as an indicator for
quality (Inzlicht et al., 2018), and self-investing effort into the creation of a product can develop
emotional attachment and a sense of psychological ownership which is associated with greater
perceived valuations (Kahneman, et al., 1990; Sarstedt, et al., 2017). Hence, it is plausible to
assume that engaging tasks tend to foster stronger product attachment and the more challenging
the task, the greater likelihood of evoking a sense of accomplishment and success. Accordingly,
we can anticipate that engaging tasks of higher difficulty would higher valuations.
While several studies have demonstrated the path of influence between effort and
valuation of self-created products, there is limited literature specifically examining the effect of
extent of effort on this relationship. Diefenbach et al, (2018) and Atakan et al, (2014b) compared
80
valuations between two types of effort, by contrasting the outcomes of assembling the product
versus customizing it. In two additional studies, (Atakan et al., 2014a; Ling et al., 2020) the
number of customization choice options was manipulated as an indicator for the level of mental
effort involved in creating the product. However, it can be argued that this type of mental effort
does not bear any consequences for task accomplishment as there is no possibility of “failing”
the task of product creation; it may simply take longer to select the preferred customization
options. Furthermore, as indicated in the meta-analysis on the IKEA Effect (Chapter 1), we could
not compare the extent of effort between studies without it being consistently measured. Thus,
the current study aims to bridge this gap in extant literature by comparing two conditions with
tasks that differ in their difficulty level. In this manner, we can progress beyond the mere
expectations that those who assemble the text will perceive greater value compared to those who
solely read the text and examine whether the extent of self-invested effort leads to its greater
valuation. This refers to both (a) evaluations of the text, as well as (b) monetary valuation of the
editing labor Accordingly, the following hypotheses are posed:
H1. The difficulty level of the text assembly task will influence its perceived
valuation in a manner that the highest (a) evaluations, and (b) valuations, will
be associated with the difficult assembly task, followed by the easy assembly,
and lowest for the no-assembly condition.
Effort and Attachment
If the IKEA Effect manifests in a similar manner with texts as it does with consumer
products, it is likely to operate through the same psychological mechanisms. One of the key
underlying mechanisms of the IKEA Effect is developing a sense of psychological ownership
and emotional attachment during the process of product creation. The meta-analysis detailed in
81
Chapter 1 further corroborated attachment as a significant moderator influencing the effect of
effort on product valuation across studies. Given the existing literature, it could be presumed that
individuals would develop a comparable attachment to a text they actively assemble, similar to
their attachment to any product they engaged in creating. Similarly, it could be presumed that
greater effort and engagement with the text would result in greater attachment, though it is
unclear whether simply reordering arguments into a coherent paragraph is enough to generate an
attachment to a text in the same manner as to a physical object. The current study attempts to
answer this question. Finally, since attachment is a key mechanism of the Ikea effect, we may
expect that greater attachment to the text will result in more positive valuations associated with
the text. Accordingly, the following two hypotheses are posed:
H2. The extent of effort will influence the emotional attachment to the text in a
manner that stronger attachment will be associated with the difficult assembly
task, followed by the easy assembly, and lowest for the no-assembly condition.
H3. A greater sense of attachment to the text will correlate with a more positive (a)
text evaluation, and (b) monetary valuation
Affect and Persuasion
Emotions play a crucial role in shaping our cognitive processes, influencing the way
we perceive, interpret, and evaluate incoming information. Whether it be positive
emotions like happiness and excitement, or negative emotions such as anger and sadness,
our emotional state can significantly impact the way we engage with information.
Research suggests that individuals in a positive emotional state may be more susceptible
to persuasion compared to those in a neutral or negative emotional state, because they tend
to expand their attention, encompassing a broader range of information (Fredrickson &
82
Branigan, 2005). Additionally, positive emotions can enhance motivation and reduce
resistance to persuasion, making individuals more open to attitude change. The influence
of negative emotions on persuasion is more complex and can vary depending on several
factors. While negative emotions can narrow attentional focus and lead to more focused
processing of information, they can also increase skepticism and resistance to persuasion.
When individuals are experiencing negative emotions such as fear, or anger, their
attentional focus may narrow, leading to a more selective processing of information,
which may result in a heightened scrutiny of persuasive messages, making individuals
more critical and less easily persuaded.
Emotional states can influence the processing route individuals take when
evaluating persuasive information. Positive emotions often promote heuristic processing
via the peripheral route, where individuals rely on mental shortcuts and quick judgments.
Negative emotions, on the other hand, can lead to more systematic processing, where
individuals engage in careful evaluation and scrutiny of the persuasive message.
Additionally, emotional states can affect the level of scrutiny individuals apply to
persuasive messages. When in a positive emotional state, people may be more inclined to
process information in a less critical and more favorable manner. A positive mood can
lead to less message scrutiny when a depressing message is encountered, as long as the
message is not mood threatening (Wegener et al., 1995). Negative emotions, however, can
heighten skepticism and increase the scrutiny of persuasive messages, leading to more
careful evaluation of the arguments presented. Emotional states can also impact resistance
to persuasion. Positive emotions can make individuals more receptive to persuasive
appeals and increase their willingness to change attitudes or behaviors. Negative emotions
83
may enhance resistance to persuasion, making individuals more resistant to changing their
existing attitudes or behaviors. However, the impact of emotions on persuasion is not
uniform and can be influenced by factors such as message content, personal relevance, and
individual differences. For example, if a message offers a solution to the negative state, it
may be more persuasive, based on the (Nabi, 2003). Individuals who are highly involved
or motivated to process information may still be influenced by persuasive messages, even
in the presence of negative emotions.
A task requirement itself can prompt individuals to experience certain emotions which in
turn may influence the evaluation of the task and its outcome. Concerning the text-assembly task,
an overly complicated and difficult task can lead to feelings of frustration, irritation, or even
anger, consequently resulting in negative assessments of the persuasive text. Alternatively, the
successful completion of the task may spur a positive emotion relating to success and
competence, leading to a more positive evaluation of the text. Based on this logic it can be
assumed that a difficult task may produce more negative emotions such as annoyance compared
to an easy-to-solve task which is more likely to result in positive emotions such as satisfaction
and pride. Accordingly, the arousal of such positive and negative task related emotions is likely
to be associated with positive and negative evaluations of the text. Alternatively, solving a
difficult task may lead the individual to experience a stronger positive emotion compared to
solving a simple task that could be solved easily by anyone.
Finally, the experience of negative emotions is related to reactance, a psychological self-
guarding strategy that individuals may take when they perceive a threat to their freedom of
choice. It often leads to a motivational state aimed at restoring or asserting that freedom. When
individuals feel that their freedom is being restricted or pressured, it can trigger a defensive
84
emotional which is a key component of reactance. The experience of negative emotions while
encountering information serves as a signal to a potential threat to freedom and will likely result
in rejection of the arguments conveyed in the message, as well as lower evaluation of the
message itself.
H4a. Experiencing negative emotions while reading the text will correlate with less
favorable evaluations of the text?
H4b. Experiencing negative emotions while reading the text will correlate with less
support for the message conveyed in the text
Customization as Self-expression
Enabling customization options of products has the potential to influence perceived
valuation as it offers a form of identity expression and personalization (e.g., Franke et al., 2010;
Ling et al., 2020), as it taps into the psychological desire for self-expression and personalization,
which can positively influence the perceived valuation of the product By allowing individuals to
tailor the product to their unique tastes or preferences, it enhances their sense of ownership and
emotional connection to the product. Furthermore, it engages the individual with the process of
creation, requiring more investment of thought and active choice, potentially increasing the
emotional attachment and sense of ownership of the product. However, research has exhibited
discrepancies in outcomes concerning the valuation of self-customized products. While some
found that enabling customization decisions in the creation process leads to greater satisfaction
and amplified valuations (Atakan et al, 2014; Buechel & Janiszewski, 2013; Ling et al, 2020),
others found that requiring self-customizing led to lower valuations (Fink & Geldman, 2017) or
no difference between valuations of those who customized the product and those who solely
assembled it (Diefenbach et al., 2018). The results from the meta-analysis on the IKEA Effect
85
(Chapter 1) suggest that the influence of customization on valuation may be moderated by
product tangibility, in a manner that it reduces valuations of tangible products while increasing
the valuations of products that were not tangible during the creation process. As the task at hand
involves assembly of text via a digital program, adding the option to customize the webpages
appearance may bare a positive impact on both valuation of the text and emotional attachment to
the creation. Accordingly, the following hypotheses are posed
H5. Participants who customize the webpage will: (a) evaluate the text more
positively, (b) perceive rate greater monetary value to the editing labor, (c) express
higher levels of attachment to the text
Online Health Communication and the IKEA Effect
It has been suggested that Mobile healthcare apps can motivate users to monitor their
habits and set goals to help them be healthier (Wang et al., 2016). Enabling self-customization of
the app, could help maintain engagement and encourage long-term commitment (Wang et al.,
2016). Within the context of the present study, this addition could address the question as to
whether any effort related to the task would influence appreciation and persuasion, or does the
effort need to be invested directly in the construction of the persuasive message itself. We could
also ask whether taking part in the design choices of the app interface predicted more usage of
the app. Based on the literature, the following hypotheses are posed:
H6. Participants who customize the website’s color will express greater intent to
use the website in the future.
H7. Participants who customize the website’s interface will express lower
perceived threat to freedom from content they consume on the website.
86
Method
Sample & Study Design
This study used a representative sample of the US general population, which was
recruited via Qualtrics online samples (N = 534), ages ranging between 18 and 82 (M
age
=42.25).
Of them, 264 identified as female, 260 as male, and 10 marked other. Approximately 78%
identified as White or Caucasian (418), 81 as Black or African American, 15 Asian, 16, Native
American or Alaskan native, 5 as Native Hawaiian or Pacific Islander and 36 as Other or
preferred not to say. Of these, 31 identified as being of mixed race. Originally 569 subjects
responded to the study but 35 were removed from the dataset because they sped through the
stimulus material without reading it, as this defeats the purpose of the study which compares
those who read and those who assemble the text. The threshold for “speeders” was set as those
who spent less than three seconds on the reading task page.
Building upon the stimulus material developed for sunscreen application, this study
enhances the model’s applicability in a more realistic context by integrating the stimuli into a
[mock] healthcare website named “On your health.” This modification goes beyond asserting
that the persuasive text originates from a healthcare website, by providing context for the
message, and bolstering the external validity of the results. Additionally, the study expands the
experimental design by incorporating two additional manipulations. The first manipulation
explores whether effort needs to be invested directly into the construction of the text to produce
message-related attitude change or whether customization of aesthetic features alone is
sufficient. Half of the participants were asked to select their preferred color scheme for the health
information website out of three options: green, red, or blue [see appendix for design options].
This offers the ability to test the influence of self-expression via aesthetic customization on the
87
relevant outcomes. The second manipulation altered the difficulty of the assembly task by
fragmenting the paragraph into more segments making it slightly more difficult to reassemble.
By comparing conditions of varying difficulty, this study examines whether the extent of effort
further moderates the influence of text assembly on attitude change. This resulted in a 2
customization (customization vs no customization), by 3 task-type (assemble-difficult; assemble-
easy; read) randomized experiment with a between subject design. [See Appendix 3.A-D for all
stimuli material].
Procedure
Respondents from Qualtrics sample pools were sent a recruitment link provided by the
company. They first read a consent form providing information about the study, and after
clicking to indicate they consent to participate, they were redirected to the study page. All
participants answered the same pre-attitudinal survey in which they were asked about their
everyday practices concerning application of sunscreen as well as filler items related to other
healthcare practices such as tendency to floss teeth and vegetable consumption. This was
collected to both distract them from the main study goal, as well as to assess their overall health
habit inclination. Participants were then randomized into one of the six conditions. Those in the
three customization conditions were asked to select their preferred color scheme for a health
information website out of three options. Following the color selection, members of the read-
only condition (control) were presented with a website page in the color they selected, containing
the persuasive text. Participants in the other two groups (easy and difficult) were directed to the
assembly task. After completing it successfully and receiving the “success” notification, they
were presented with the same website page in color scheme that they selected, containing the text
they had just worked on. Participants in the three non-customization conditions (Read/Assemble-
88
easy/Assemble-difficult) were not prompted to select their preferred color scheme prior to the
stimuli. After the pre-assessment questionnaire they were directed to the text stimuli and the
same mock webpage containing the text. Following the stimuli, all participants filled the same
questionnaire assessing the dependent variables and relevant control variables.
Measures
The questionnaire was comprised of the measures and scales from Study 1. These
measures and correlation values for the indexes created are as detailed: Post-attitude (Cronbach’s
α = .896), text evaluation (Cronbach’s α = .835), monetary valuation; perceived freedom threat
(Cronbach’s α = .0.845); political ideology (single item); issue relevance; and basic demographic
variables. In addition to these, the current study also includes the measures detailed below.
Healthcare habits. Participants rated the frequency in which they tend to apply
sunscreen. Responses were on a five-point scale, ranging from “never” to “multiple times a day.”
The questionnaire also contained filler items inquiring about their tendency to partake in other
healthcare practices, including wearing a seatbelt, flossing teeth, and consuming vegetables in
their daily meals. The average of these items was used as a proxy for overall healthcare tendency
by the respondent, which can indicate the general attitude toward self-healthcare habits.
Perceived effort. Participants in the experimental groups were asked to indicate how
difficult their assembly task was in their opinion. Response options on a 5-point scale ranged
from 1 (extremely easy) to 5 (extremely difficult). This item was also used as a manipulation
check to verify that the difficult assembly task was indeed perceived as more difficult.
Attachment. This measure was comprised of three Likert items ranging on a 7-point scale
from strongly agree to strongly disagree. Two items assessing emotional attachment were
adapted from Sarstedt et al., (2017). These included “ This paragraph has no special meaning for
89
me,” and “This paragraph moved me.” A third item pertaining attachment as an extension of self
“If someone praised this paragraph, I would feel somewhat praised myself,” was adapted from
Fuchs et al., (2010). The original indexes by were comprised of 28 items assessing different
aspects of attachment, however, because these measures were developed for contexts of physical
goods, most rendered incompatible for assessing attachment to a text. (Cronbach’s α = .0.714)
Behavior intention. In addition to assessing post attitudes toward the healthcare practice
as in Studies 1 and 2, this study included items assessing intentional behavior, concerning the
likelihood of (a) applying sunscreen on a daily basis, as well as (b) the likelihood of using the
website On Your Health in the future. These items used a 7-point Likert scale ranging from
strongly agree to strongly disagree.
Negative Affect. Participants ranked the extent to which they experienced each of three
following emotions while constructing/reading the text.: “irritated,” “angry,” annoyed,” on a 5-
point response scale ranging from 1 “none of this feeling” to 5 “a great deal of this feeling.”
(Cronbach’s α = .0.847); This scale is based on the validated measure by Dillard & Peck (2000).
Their scale includes a fourth item “aggravated,” which was not included in this study.
Education. A basic question concerning education was added to the demographic items,
asking participants what the highest level of education they had completed so far is. This
measure was a multiple-choice item with 8 options ranging from “some high school” to
“graduate /MB” . Since this study uses a representative sample from a wide spectrum of
education levels, unlike the preceding studies, it was imperative to assess this as an additional
control variable. Around 11% had a graduate or professional degree, 30% bachelor’s or technical
degree, 18.5% some college, 29% high school diploma or GED and 9% some high school
education or less.
91
Results
Manipulation check
There was a significant difference between builders in the difficult versus easy assembly
conditions (t = 6.18, p < 0.001), with the former rating the task as slightly more difficult
compared to the later (M = 2.31, sd = 0.96; M = 1.71, sd = 0.86), suggesting the
stimuli worked as expected. Participants in the difficult assembly condition also spent more
time on average attempting to reconstruct the paragraph (M = 87(197), M = 186(416)), further
corroborating the additional effort required for completing the difficult assembly task.
Hypothesis testing
See table 6 for sample means and standard deviations by treatment group for variables. Of
interest. The first hypothesis presumed that (a) evaluation of the text, and (b) valuation of the
labor invested into the text assembly, will both be highest in the difficult assembly condition,
followed by the easy assembly condition, and lowest for the no-assembly condition. A simple
one-way ANOVA indicated that the main effect of text assembly task on text evaluation was
significant (t = 1.81, p = 0.04), with scores distributed in the expected direction (M = 6.04, sd =
0.78; M = 5.95, sd = 0.83; M = 5.79, sd = 1.13). A post hoc test reviled that the difficult
assembly condition was significantly different than the control group (p < .05), but the easy
assembly condition was not significantly different than either group. The model testing
monetary valuation as the outcome was also significant (t = 2.28, p < 0.01), however, it was the
easy assembly task that had the highest score (M = $13.21, sd = 0.43), followed by the difficult
and control group (M = $12.2, sd = 0.44; M = $11.16, sd = 0.46). The post hoc test reviled that
the easy-assembly condition was significantly different than the control group (p < .01), but not
from the difficult assembly group. Thus, H1 was partially supported by the data. It should be
92
noted that although both models were significant, the variance among the groups was rather
small as the averages of all groups tended to cluster in the upper portion of the scale.
The next set of hypotheses presumed that greater effort invested into the text assembly
will result in stronger emotional attachment to the text they labored on. In turn, greater
attachment will correlate with more positive evaluations of the text and monetary valuations.
The effect of assembly effort on attachment was not significant (p > 0.05). The effect of
attachment on text evaluation was significant with greater attachment levels predicting a more
positive evaluation of the text (t = 9.57, B = .38, sd = .04, p < .001). The effect of attachment on
monetary valuation was not significant (p > .05). Therefore, H2 and H3b were not supported,
while H3a was supported by the data.
Concerning affect, H4 assumed that the experience of negative emotions while reading
the text would result in less favorable evaluations of the text, as well as less support for the
message conveyed in the text. First, it was necessary to test whether text assembly impacted the
negative effect. The influence of assembly task on the experience of negative emotions was
Figure 11. Estimated means of negative emotions by effort.
93
significant (t = 2.67, p > .001), as illustrated in figure 11, with those in the difficult assembly
condition experiencing least negative emotions (M = 1.09, sd = 0.29), Followed by the easy
assembly condition (M = 1.17, sd = 0.45) , and the control group (M = 1.3, sd = 0.74). A post
hoc test revealed that the difficult assembly condition and the control group were significantly
different than each other (p = .002), but not from the easy assembly condition.
Next, a mediation analysis assessed the influence of assembly effort on text evaluation
via negative emotions (See figure 12). The effect of negative emotions was significant, as
stronger negative emotions correlated with less favorable evaluations of the text (p < .001). The
direct effect of effort on text evaluations was not significant (p = .11) but the total effect of the
model was significant (p = .012). While H4a is supported by the data, it should be noted that the
portion of variance that was explained by the mediation model was rather small (R
2
= .11).
The last set of hypotheses posed assumptions for outcomes relating to webpage
customization, and indeed, including customization as an additional predicting variable resulted
in interesting patterns among the six experimental groups. The influence of customization on
attachment was not significant (p > .05), offering no support for H6. The model assessing the
outcome of monetary valuation was significant (t = 2.01 , p < .001). Assembly effort remained a
-.104***
Assembly Effort
Text evaluation
Figure 12. Influence of effort on text evaluation via negative emotion
Note. *p < .05, ***p < .001, 95% confidence interval
c’ .08
-.45***
Negative emotions
C .12 *
94
significant predictor of perceived valuation. The main effect of customization on valuation was
also significant (p < .05), with subjects who selected the color scheme rating the text editing
labor as less valuable compared to their non-customizing counterparts (M = $11.66, sd = 5.89; M
= $12.8, sd = 6.01). Among the six experimental groups, the highest valuation was given by the
easy-assembly condition without customization, followed by difficult-assembly without
customization (M = $14.32, sd = 0.61; M = $12.89, sd = 0.62). The lowest valuation was offered
by the control group without customization (M = $10.83, sd = 0.9). A post hoc test revealed that
the easy-assembly condition was significantly different than both non-builder groups (p = 0.01; p
= 0.001), as well as the difficult-assembly condition with customization (p = 0.014).
Interestingly, the average valuations of three groups that customized the webpage were
Figure 13. Estimated means of Valuation by text engagement and webpage customization
$10.83
$14.32
$12.89
$11.47
$12.10
$11.50
$9
$10
$11
$12
$13
$14
$15
Read (Control) Assemble Easy Assemble Difficult
Valuation
Task Engagement
without
with
customization
95
not significantly different than one another, regardless of task difficulty, suggesting that
incorporating customization into the task dissipated the effects of text assembly effort. Figure
13 illustrates the estimated means of valuation by text engagement task and customization.
These results stand counter to the expectations of H7.
Similarly, the model estimating text evaluations by treatment group was significant (t =
1.84, MSE = 0.21, p < .01), with both assembly effort (p < .05) and customization (p < .01)
significantly predicting text evaluation. those who did not customize the color of the web page
rated the text as slightly more positive than those who did customize (M = 6.05, sd = 0.82; M =
5.8, sd = 1). A post hoc revealed that only the difficult assembly condition and the control group
(non-builders) without customization were significantly different than each other (p = .05),
which is not surprising as the averages of all groups were relatively close. Again, this result is
counter to the expectation of H8 which assumed that those who customize will evaluate the text
more positively. Another model tested whether customizing the color of the webpage would
suffice to influence intentions of visiting the website in the future. While the slope was in the
expected direction, with those who selected the color indicating a slightly higher likelihood of
visiting the website compared to those who did not customize, the gap between the two was not
significant, (p > .05) offering no support for H6.
Finally, in order to address the main research questions pertaining to self-persuasion,
this study concludes with a mediation analysis, similar to that which was conducted in the two
previous studies, employing Hayes’ process Macro, model 6 for serial mediation. Effort was
imputed as the predicting variable and post attitude as the outcome variable. Perceived freedom
threat, text evaluations a labor valuation were imputed as the mediating variables, accordingly,
as illustrated in figure 14. The model predicting perceived freedom threat was not significant (p
96
=.29), indicating no meaningful influence of assembly effort on this outcome. The model
regressing text evaluation was significant (t = 7.2, p < .001), with both assembly effort (p = .02),
and perceived freedom threat (p < .001) significantly predicting evaluation. As would be
expected, greater effort translated into higher evaluations, while greater perceived freedom
threat predicted less favorable evaluations. The model regressing labor valuation was not
significant (p = .091), though text evaluation did emerge a significant positive predictor for
valuation (p = .05). Finally, for the outcome of post attitude toward sunscreen application, the
model was significant (t = 9.25, p < .001). Both assembly effort (p = .004), and perceived
freedom threat (p = .006), negatively predicted post attitudes, while text evaluation, positively
predicted post attitudes (p < .001). The total effect of the model was not significant. However,
-.079
Assembly Effort
Post Attitude
Figure 14. Multiple mediation for the influence of Effort on Post Attitudes
Note. *p < .05, **p<.01, ***p < .001, 95% confidence interval
.6*
Freedom threat
Labor valuation
Text evaluation
.086
.763***
.104**
-.262***
.006
.432
-.089**
-.148**
97
the indirect effect via text evaluation was significant and positive, indicating that greater effort
invested into the text assembly resulted in more positive evaluations of the text, which in turn,
translated to a stronger agreement with the message conveyed in the text.
Next, a series of regression models tested the impact of the relevant control variables on
post attitudes (See table 7 for detailed regression model results). Adding customization as a
predicting variable did not alter the significance of the paths of influence. Negative emotions (p
< .05), and attachment (p < .001) significantly influenced post attitudes, with those who
experienced fewer negative emotions, as well as those who indicated greater attachment to the
text, tending to report greater agreement with support for the message. Issue relevance was not a
significant predictor of post attitudes ((p > .05). Unsurprisingly, pre-behavior was a significant
predictor of post attitudes, with those who already tend to apply sunscreen frequently also
indicating greater support for the benefits and importance of sunscreen application. Importantly,
though, even after inclusion of pre-behaviors as a control variable, the influence of text
evaluation on post-attitudes remained significant (p < .001). Upon examining the demographic
control variables, political ideology did not emerge as a significant predictor of post attitudes,
unlike the results of Study 2. Gender and education did, however, emerge as significant
predictors, female participants, and those with higher levels of education, tending to indicate
greater support for the importance and benefits of sunscreen application.
In a final iteration, the models were replicated by replacing the outcome variable of post
attitudes with behavioral intentions, pertaining to their likelihood of applying sunscreen on a
daily basis in the future. Similar patterns emerged for behavioral intentions; however, the
models were not significant, indicating that text assembly in this context was not sufficient for
inducing a meaningful shift in behavioral intentions.
98
Table 7
Regression models for post attitudes by text assembly and control variables – Study 3
Model 1
Model 2
Model 3
Model 4
Model 5
Constant
1.39***
(.35)
1.66***
(.39)
1.6***
(.41)
1.39***
(.324)
1.88***
(.39)
Assembly effort
-.148**
(.051)
-.148**
(.051)
-.161**
(.051)
-.057
(.048)
-.122*
(.051)
Freedom threat
-.089**
(.033)
-.094**
(.033)
-.061
(.034)
-.101***
(.03)
-.101**
(.032)
Text evaluation
.763***
(.049)
.752***
(.049)
-.658***
(.052)
.574***
(.049)
.721***
(.049)
Valuation
.006
(.007)
.006
(.007)
.004
(.007)
.005
(.006)
.003
(.007)
Customization
--
-.123
(.084)
-.123
(.082)
--
--
Negative emotions
--
--
-.205*
(.086)
--
--
Attachment
--
--
.216***
(.047)
--
--
Issue relevance
--
--
--
-.001
(.001)
--
Pre-behavior
--
--
--
.327***
(.03)
--
Political ideology
--
--
--
--
-.05
(.025)
Gender
--
--
--
--
-.372***
(.084)
Age
--
--
--
--
.006*
(.003)
Education
--
--
--
--
.087***
(.026)
Adjusted R
2
.385
.394
.453
.504
.421
t
9.25
8.31
7.43
9.31
7.01
p
.000
.000
.000
.000
.000
Note: n = 534, *p < .05, **p < .01, ***p < .001. Bootstrap = 5000, CI = 95%
99
GENERAL DISCUSSION
The present dissertation sets out to investigate the potential application of the Ikea effect
as a method for inducing attitude change within the domain of persuasive texts. The empirical
investigation of this inquiry is approached through the lens of self-persuasion, guided by two
primary objectives. First, it aims to ascertain whether the cognitive investment in text assembly
elicits heightened valuation of said texts, analogous to the psychological processes associated
with the assembly of consumer goods. Second, it seeks to evaluate the potential utility of the Ikea
effect framework as a tool for self-persuasion. The empirical foundation of this project resides on
four studies, encompassing a meta-analysis and three experiments. These studies serve the dual
purpose of assessing the psychological processes moderating the impact of self-invested effort on
valuation, and empirically scrutinizing the conceptual proposition regarding the relevance of the
Ikea effect as a self-persuasion mechanism. The meta-analysis of the Ikea effect offers a
comprehensive understanding of the phenomenon, shedding light on its boundary conditions, key
aspects, major strengths as well as its inconsistencies. In doing so, it transcends its preliminary
contribution by offering a resource to scholars interested in this aspect of economic psychology
and marketing research as a whole. Additionally, it provides a steppingstone for the empirical
studies conducted in this project by which we can progress to a broader academic landscape,
bridging a gap between the domain of consumer behavior and the realms of communication and
information processing. The studies presented here contribute to the extant literature by
introducing an additional layer to our comprehension of the influence of effort and engagement
on valuation. Importantly, they propose a potential utilization for the effect as a tool to capable of
not only inducing attitude change but also influencing the underlying psychological processes
related to persuasion. While Studies 1 and 2 test the proposed model, addressing the primary
100
research questions, Study 3 extends the scholarly discourse by examining potential boundary
conditions and addressing issues raised in the preceding sections. This comprehensive
examination seeks to clarify and explain the factors that may either enhance or inhibit the
influence of self-invested effort on attitude change, thereby enriching the investigation of the
Ikea effect in the context of persuasive texts and contributing to the literature of self-persuasion.
Key Findings and Theoretical Implications
RQ1. The primary research question sought to ascertain whether the Ikea effect could
manifest when the target product is a text rather than consumer goods. Findings across all three
studies affirmatively indicate that the phenomenon can indeed operate with text as the self-
assembled product. However, certain distinctions emerged between the initial student-sample
studies and the subsequent comprehensive study with a representative sample. In studies 1 and 2,
the assembly of a text from predetermined argument led participants to perceive their “editing”
labor as more valuable than the labor of someone else who edited the text, particularly in terms
of monetary valuation. However, when evaluating the text’s contents and structure, the data did
not indicate assembly efforts resulted in more positive assessment. In fact, in Study 1, those who
merely read the text without engaging in its assembly actually gave slightly more positive ratings
compared to their assembling counterparts. In study 2 there was no significant difference in
evaluations among participants from the different treatment groups. It is essential to mention
that, on average, both builders and non-builders tended to rate the text on the positive side of the
scale, indicating a general positivity in assessments. The discrepancy between the two forms of
valuation is interesting because it stands contrary to the assumption that they would be
interrelated. These results align with the literature on the IKEA effect, which posits that we tend
to value products we actively assemble due to an inherent appreciation of our own labor and
101
time. Thus, even if we may think less of a product, we may still overvalue it merely because we
invested out selves into it.
However, insights from the final study prompt consideration of alternative explanations
for this discrepancy. In study 3, the effort individuals invested into restructuring the paragraph
significantly influenced their evaluation of the text. This suggests that the extent of effort
invested into assembling a text, even when arguments are provided from an external source, can
influence one’s assessment of the message’s qualities. Notably, in this study, the participants in
the assembly conditions valued the editing labor more than those who only read the text, but
there was no significant difference between the two assembly groups, suggesting that the extent
of effort did not predict the monetary valuations for the editing labor. Rather, it was the
evaluation of the text itself that correlated with monetary valuations, indicating that more
positive text assessments resonated with higher monetary valuations. Thus, the persistence of
disconnect between text evaluations and monetary valuations only among the student samples
suggests a potential generational aspect influencing perceptions of labor and financial
expectations. In summary, the response to the first research question is affirmative. These
collective findings indicate that investing effort into assembling a text yields more positive
assessments concerning the text’s quality. Additionally, it leads to higher valuations of the labor
involved in the text’s creation, even when the arguments were not self-generated.
RQ2. The second research question guiding this project inquired whether invested effort
into the assembly of a persuasive text would result in greater attitude change, and whether this
relationship would be mediated via valuation. Once again, the three studies yielded different
results. The data from the student-sample studies did not indicate that effort invested into
assembling persuasive texts translated into greater advocacy for the arguments conveyed in the
102
text. In fact, Study 1 suggested a notable influence in the opposite direction, as participants who
assembled the text tended to exhibit less agreement with the message compared to the control
group. Rather than effort and evaluation, the main predictors of post attitudes in this data were
preexisting attitudes, as well gender, and political ideology, with women, and stronger liberals
showing more greater agreement with the message. In Study 2, text assembly did not influence
post-attitudes, though it did influence perceived freedom threat, with those who engaged in the
text assembly indicating less perceived threat compared to the control group. This is a notable
finding in the realm of persuasion, as a diminished sense of threat to personal freedom of choice
can mitigate self-guarding strategies such as reactance. The alleviation of perceived threat is a
substantial obstacle to overcome in direct persuasion methods. These results suggest that
engaging in the creation of the persuasive message, even when the arguments are not self-
generated, may be sufficient to diminish the perception of an external source attempting to assert
influence. Thus, even though there was no significant influence on attitudes, there was a
significant influence on a key underlying mechanism of persuasion.
In contrast to Studies 1 and 2, the insights from Study 3 provide support for the
hypothesized mediating relationship. Both the direct path from effort to post attitudes, as well as
the indirect path via text evaluation were significant and aligned in the expected direction.
Though the difference between the scores of the two builder groups (difficult and easy assembly,
were not significant, they were both greater than the control group. Consequently, grounded in
the findings of the concluding study, the answer to the second inquiry is also affirmative. Greater
effort invested into the text assembly task resulted in greater evaluations of the text’s qualities,
which then translated into greater agreement with the persuasive message. The distinction in
results between the concluding study and the two preceding studies warrants a deeper discussion
103
as it may be explained on both methodological and theoretical levels. First, in the final study the
text was deconstructed in such a manner that there was only one possible solution which could
be read correctly from front to end. This eliminated frustrations associated with a seemingly
correct solution not being accepted. While in the preceding studies only one solution was
accepted, there were potentially two other solutions that could be considered grammatically
correct, even though the logical flow was weaker. This may have resulted in frustration among
participants who were denied a seemingly correct submission. A second major difference is that
embedding the text within a health information website may have bestowed a pragmatic utility
on the text which was lacking in the two initial studies. In other words, it is more readily
perceived as a product, rather than a mere black on white text. This change also offered the text
and website a characteristic of utility which could have further enhanced the sense of
contribution among the participants. Third, the variance in results may also be explained by the
composition of the samples, in combination with the topic of the persuasive text. The student
samples are more homogenic with regards to education and healthcare-practices. Furthermore, as
they are younger and tend to be healthier, the topic of self-care and health issues may not be of
great importance or consequence to them compared to the older, and more representative sample.
This extends the discussion beyond the immediate focus of exploring the primary research
questions, as it also sheds light on a broader consideration of the suitability and unsuitability of
using student samples for certain research endeavors. On some topics there may be less variance
in responses due to certain homogeneity and convergence of values, education level, worldviews,
and shared experiences.
Expression versus Engagement. While the primary objective of this research endeavor
was to evaluate the practical utilization of the Ikea effect framework as a method for inducing
104
attitude change, the empirical studies presented here offer a gateway to a broader theoretical
exploration of the underlying psychological mechanisms inherent to self-persuasion. Classic
studies on self-persuasion attribute the efficacy of these methods to the active engagement in
role-playing and the formulation of self-generated arguments. In addition to the well documented
benefits of self-tailored messages and inhibition of self-guarding strategies, the success of self-
persuasion stems from people genuinely contemplating arguments that support the opposing
viewpoint. However, in the concluding study, individuals who reconstructed a paragraph from
pregiven arguments also indicated greater attitude alignment with the arguments. Thus, one of
the unique contributions of the current research to the existing literature is that, under certain
conditions, consideration of the alternative perspectives can occur even when the arguments are
provided by an external source. This notion could be expanded to a distinction between
expression and engagement. Within traditional self-persuasion techniques, the act of generating
one's own arguments may be considered a form of expression. In the context of self-persuasion,
individuals often partake in the process of constructing their own arguments or reasons to
support a particular perspective or belief. This act of self-generation is likened to a mode of
expressing one's thoughts, opinions, or rationale, emphasizing the active and personal nature of
the persuasive process. Conversely, the act of assembling a text from externally sourced
arguments represents a different method of influence, which centers on engagement rather than
expression. It challenges individuals to comprehend the content in order to reconstruct it, without
the process of producing arguments in favor of the issue or seeking counter arguments to refute
the message. This distinction invites a deeper exploration of the cognitive processes underlying
attitude change within the realm of self-persuasion, especially contexts that invite engagement
with alternative or opposing opinions that, simultaneously, do not trigger self-guarding strategies
105
Perceived influence versus actual influence. In all three studies, evaluation of the text
was a significant predictor of attitude alignment with the message conveyed in the text. The
relationship between perceived message strength and agreement with a message can vary, and it
often depends on factors such as subjective differences in values and knowledge, the nature of
the message, and the context in which one is exposed to the information. Research on persuasion
and attitude change suggests that both directions of influence can occur. It is possible is that
individuals first perceive a message as strong, credible, or persuasive. This positive perception
can then lead to agreement or acceptance of the message. This aligns with the idea that people
are more likely to be persuaded by messages they perceive as well-reasoned, credible, or
compelling. On the other hand, individuals may initially agree with a message, and this
agreement can subsequently influence their perception of the message's qualities. The act of
agreeing with a message may lead individuals to view it more positively, reinforcing the
perception of its strength. Since we cannot concretely determine the order of influence between
these two factors, one could potentially argue that the path of influence may be in the
complementary direction, suggesting that greater agreement with the message resulted in a more
positive evaluation of the text. However, a series of studies by Dillard et al., (2007) concerning
fear appeals demonstrated more support for the influence of perceived message effectiveness on
actual effectiveness in terms of attitude change, offering more support for the path of influence
tested in these models. In practice, both directions are likely to occur in different situations. The
interplay between perceived strength and agreement is complex and can be influenced by various
psychological and contextual factors. Thus, it is imperative to account for potential factors, such
as preexisting attitudes and tendencies, as well as prior knowledge on the issue of conflict.
106
Study Limitations and Directions for Future Research
Classic self-persuasion methods. A peripheral goal of this study was to compare the
IKEA effect method of text assembly with that of direct persuasion, or in other words, engaged
versus passive conditions. However, none of the studies compare the results of text assembly to a
classic self-persuasion condition. Traditional self-persuasion studies which request subjects to
generate their own arguments by listing arguments, writing an essay, or attempting to persuade
another individual, tend to integrate engagement and argument production in a single bundle.
The model in this study manipulates engagement and effort by asking participants to reconstruct
a persuasive paragraph from pregiven arguments. Thus, this model design has the potential to
separate effort from the act of self-generation of arguments, which would enable scholas to
compare between the two and assess which has a greater impact on attitude change. If, as this
dissertation claims, employing the IKEA effect with texts can offer a mid-point between direct
persuasion and self-persuasion methods, then it is imperative to conduct such a comparison.
However, the studies in this manuscript did not include a condition in which subjects self-
generated their own arguments, thus such a comparison cannot be conducted. A future study
could easily replicate the experimental design and incorporate an additional condition in which
participants are asked to write a short paragraph or list arguments in favor of daily sunscreen
application, or any other issue of debate, and compare post attitudes between conditions of self-
generated versus self-assembled persuasive texts. Such a comparison would contribute to the
self-persuasion literature by disentangling two theoretically confounded variables. This would
also enable scholars to compare the results between the three groups and indicate how the
proposed model fairs relative to traditional self-persuasion methods versus direct persuasion
methods.
107
Low variance and Ceiling effects. Even though the data indicated support for some of the
assumptions, and there were a number of interesting significant findings, it is notable that there
was little variance for many of outcome variables and the models only explained a small portion
of that variance, especially in studies 1 and 2. There are a few potential explanations for these
results which could easily be addressed in future studies. Namely, the healthcare issues selected
for the studies are not highly controversial. Most people tend to have a relatively positive attitude
toward both of the healthcare treatments and, if asked, would likely claim that it is important and
beneficial to apply sunscreen and floss daily. Whether they actually floss and apply sunscreen is
another question. To resolve this matter, future studies can employ one of three potential
solutions. First, they should consider testing lesser-known healthcare practices or topics with
which participants are less familiar. It is easier to influence attitudes when they are less
formulated, and the individual does not have much experience with or knowledge of them.
Attitudes toward lesser-known topics tend to be more susceptible to influence due to factors such
as limited prior knowledge and reduced cognitive dissonance. Individuals often have less
formulated attitudes on these topics, providing a greater receptivity to external information and
persuasive messages. The absence of well-established beliefs reduces resistance to change,
rendering attitudes more flexible. Additionally, social influence dynamics also play a role
because lesser-known topics lack established social norms, further rendering individuals more
open to external information. The current studies were able to nudge attitudes on well-known
and socially accepted practices using this method, thus if we were to witness greater attitude
change with lesser-known topics it would serve as further testimony to its potential utility.
Alternatively, subsequent research could use the same healthcare topics but advocate for
the opposing stance, highlighting the negative aspects of applying sunscreen or flossing teeth.
108
The stimuli employed in the present studies exclusively presented arguments in favor of the two
healthcare practices. As noted above, these arguments likely aligned with the pre-existing
attitudes of the participants, which resulted in a ceiling effect, as it is difficult to shift attitudes to
greater support when they are already near the positive pole of the scale. Thus, to elicit more
significant attitude change, future studies might consider presenting arguments that emphasize
potential harm associated with the healthcare products, such as the presence of benzine and other
carcinogens in the components of sunscreens, or the usage of plastics in traditional dental floss
products. By presenting arguments that challenge pre-existing attitudes, the model could
potentially enhance its effectiveness in inducing attitude change, avoiding near-ceiling effects
that emerged in the data sets. Albeit, while seeking to assert greater power for this potential
method of influence, it is worth noting that even incremental shifts toward promoting better self-
care, should be acknowledges as meaningful achievements, underscoring the potential impact of
this method of influence.
Artificiality and generalizability. While the findings from the three studies present
promising insights, this line of research would benefit from the development of a more pragmatic
methodology for the application of the IKEA effect framework beyond controlled experimental
environments. The ability to translate findings from empirical studies into practical applications
for the real world holds paramount importance. Although empirical research provides valuable
insights within controlled settings, its true impact is realized when these findings are made
applicable and relevant to address real-world challenges, particularly in domains such as
healthcare and education. The current design provided a tightly controlled environment to
investigate the interplay between the invested effort and persuasive texts. While the selected
design focused on isolating the participants' engagement with the text—an integral aspect of the
109
study—it introduced a measure of artificiality inherent in the text-assembly task. This task,
involving the assembly of paragraphs from predetermined arguments, akin to constructing a
puzzle, adhered to the IKEA effect framework, but simultaneously, sacrificed the authenticity of
natural argumentation reducing its external validity. Specifically, the task failed to capture the
complexity and spontaneity characteristic of persuasive processes in everyday communication,
and it is improbable that individuals would encounter such a task outside of the lab setting.
While the controlled manipulation facilitated the focused exploration, the absence of a real-world
analog for the text assembly task hinders the pragmatic utilization of the findings. Presenting
such a complex and time-consuming task within contemporary communication outlets would be
both unnatural, and impractical. Thus, future research seeking to refine this method should aim
to simplify the task with practical considerations such as the complexity and time constraints, as
well as embedding the assembly task in a more natural venue. One possibility for simplifying the
task could be significantly shortening the text. Rather than requiring participants to reconstruct a
full paragraph, they could be presented with a single argument with scrambled word order.
Alternatively, the design could be adapted from the original pilot, where participants filled in
missing words. This adaptation would render the method more feasible for integration into a
social media post or even a poster, aligning with the approaches employed in recent studies on
self-persuasion and open-ended questions (e.g., Loman, 2018;2019; Müller, 2016).
An alternative approach to mitigating the inherent artificiality in the current design
involves the development of a distinct form of text assembly, potentially deviating from the
IKEA effect framework and exploring alternative strategies to engage individuals with
predetermined arguments. Future research endeavors could delve into methodologies that strike a
balance experimental control and tasks that authentically reflect natural information processing.
110
The incorporation of tasks that allow for some participant-driven expression may enhance the
external validity of the findings. For instance, rather than assembling a text akin to a puzzle,
participants could select the arguments from an argument pool and craft the paragraph
themselves. Alternatively, they could be tasked with ordering the arguments based on
importance. In both of these scenarios, the arguments are provided from an external source, but
participants gain agency in the construction process, allowing for a measure of subjective
preference and self-expression, which would offer a more authentic engagement with the
persuasive content. This would also add to the discussion on the continuum between engagement
and self-expression, both of which play a role in processes relating to persuasion.
Cause for negative affect. A potential limitation of this study pertains to the measurement
of negative affect, particularly as it related to the emotional component of reactance as outlined
in the framework proposed by Dillard and Shen (2005). Reactance theory posits that individuals
are inclined to restore their freedom of choice when they perceive a threat to that freedom. Such
perceived threats can elicit a negative emotional response that serves as an alert signal to the
potential infringement on one’s freedom. In the context of Study 3, participants were asked with
rating the extent to which they experienced feelings of anger, frustration, and irritation during the
process of assembling and/or reading the text. However, the measurement that was employed
did not differentiate between negative emotions arising from reading the text and those stemming
from frustration in irritancy associated with the text assembly task. Consequently, it remains
indeterminant whether the stated level of negative emotions primarily resulted from frustration
induced by the task or prompted by the content of the message itself, representing a crucial
distinction associated with the reactance component. A future study aiming to further investigate
reactance and assess the extent to which such an assembly task mitigated perceived threats to
111
freedom of choice should refine the measurement to disentangle these two potential causes,
ensuring greater clarity in the evaluation and consequences of the emotional response. The data
may provide insight into this debate, given that participants in the difficult assembly condition
tended to report slightly lower levels of negative emotions compared to those in the easy
assembly condition. This outcome is particularly noteworthy considering that the difficult
assembly would likely elicit greater frustration associated with the task. The observed
discrepancy in reported negative emotions suggests that they stemmed from the perceived threat
to freedom of choice. This finding warrants further examination as it may contribute to the
discourse surrounding the nuanced relationship between task difficulty and emotional response.
A related potential issue inquires whether the stimuli and design actually manipulated
engagement with the text. Indeed, there was a certain extent of cognitive effort invested into the
reordering of the arguments, however, was that sufficient in order to engage individuals with the
text to produce a meaningful shift in attitudes? A simple way to assess whether individuals
engaged with the text would be to measure their memory by conducting a simple recall
assessment on issues mentioned in the text, or asking them to write the text from memory
Boundary conditions and potential moderators. Finaly, to further examine the
compatibility of the IKEA effect in the realm of persuasive texts, it is imperative to investigate
the phenomenon’s boundary conditions. Evaluating whether factors that shape or hinder the
IKEA effect in consumer goods extend to text assembly enhances the potential utility of this
model for self-persuasion. One of the primary objectives underpinning this study was to assess
the interplay between text assembly, text evaluation and attitude change. While the current
investigation systematically examined select potential moderators, including emotional
attachment, the experience of negative emotions, perceived freedom threat, it is important to
112
acknowledge the existence of an extensive array of potential moderators and underlying
mechanisms that may further outline the boundary conditions surrounding the Ikea effect and
self-persuasion. Due to inherent constraints within the study scope, a comprehensive exploration
of all the conceivable mechanisms and moderators was unfeasible. Future research endeavors
seeking to expand and build upon these findings are encouraged to examine additional
mechanisms that were outlined in the meta-analysis of the Ikea effect and in the review of self-
persuasion literature. For instance, a noteworthy joint underlying mechanism is the experience of
cognitive dissonance, a psychological phenomenon that has been suggested to result in attitude
change under certain circumstances. However, the topics employed in the current study may not
have been sufficiently controversial as to evoke the discomfort associated with dissonance. To
address this, a future study intent on eliciting cognitive dissonance could select more contentious
issues rather than healthcare practices. It may be the case that dissonance did play a role in
attitude change for participants due to the fact that the task itself was meaningless or boring, yet
they invested time and effort, leading them to perceive the fruit of their labor and as more
valuable. However, due to constraints of the study scope, the experience of dissonance was not
empirically assessed. A future research endeavor could assess preexisting attitudes toward
specific issues and explicitly prompt those participants to assemble a text that conflates their
established worldview, increasing the likelihood that participants experience cognitive
dissonance as a result from the position they “advocated” for . Furthermore, future research
could delve into the potential influence of an audience or lack thereof on the self-persuasion
process. The awareness of an audience may intensify concerns related to external judgment and
scrutiny of one’s actions and words, thereby adding supplementary layers to the dynamics of
self-persuasion. Another example could focus on self-concept which is a key mechanism as the
113
IKEA effect has only been found to occur under successful completion. In their original study,
Norton at al., (2012), when subjects were prevented from completing the product assembly, or
were instructed to deliberately destroy their creation, the phenomenon did not occur. Their
findings suggest that these higher estimates are associated with higher levels of self-concept and
feelings of accomplishment. Thus, future studies could manipulate this factor by introducing a
random “failure” or “success” notification after participants submit their text, or perhaps adding
a time limit in which some people do not succeed to reconstruct the original paragraph in time.
From another aspect, the three studies primarily focused on aspects related to western
cultures, specifically North America. Pursuing this line of research with a comparison of cultural
divergences would be of scholarly significance. More explicitly, the meta-analysis on the Ikea
effect yielded non-significant difference across cultures and geographical regions. Nevertheless,
extant research indicates to the differences in the efficacy of self-persuasion and the experience
of cognitive dissonance between Western and Eastern cultures, specifically outlining the
dynamics within individualistic and collectivistic cultures. Thus, it would be of empirical interest
to examine whether cultural nuances exert influence on the relationship between effort,
engagement, and persuasion. Comparative studies across diverse cultures can unveil whether the
Ikea effect would manifest and guide persuasive approaches for specific cultural contexts.
To conclude, the theoretical implications of the research presented here extend beyond
the immediate focus of the Ikea effect, prompting considerations for overlapping underlying
psychological mechanisms of self-persuasion. This dissertation, which is anchored in the Ikea
effect framework, transcends traditional consumer psychology into communication studies, and
invites a continued exploration of the nature of self-persuasion with a broader interdisciplinary
approach.
114
References
* Note. An asterisk precedes references that were included in the meta-analysis of Chapter 1
Alhabash, S., & Wise, K. (2015). Playing their game: changing stereotypes of Palestinians and
Israelis through videogame play. New media & society, 17(8), 1358-1376.
Alsaker, F. D. and J. Kroger (2006). Self-concept, self-esteem and identity. In S. Jackson & L.
Goosens (Eds.), Handbook of Adolescent Development (pp. 90-117). New York,
Psychology Press.
Aronson, E. (1999). The power of self-persuasion. American Psychologist, 54(11), 875-884. doi:
10.1037/h0088188
Aronson, E., & Mills, J. (1959). The effect of severity of initiation on liking for a group. The
Journal of Abnormal and Social Psychology, 59(2), 177-181. Doi: 10.1037/h0047195.
Ashtiani, A. Z., Rieger, M. O., & Stutz, D. (2021). Nudging against panic selling: Making use of
the IKEA effect. Journal of Behavioral and Experimental Finance, 30, 100502.
Atakan, S. S. (2014). Consumer response to product construction: The role of haptic stimulation.
International Journal of Consumer Studies, 38(6), 586-592.
* Atakan, S. S., Bagozzi, R. P., & Yoon, C. (2014a). Consumer participation in the design and
realization stages of production: How self-production shapes consumer evaluations and
relationships to products. International Journal of Research in Marketing, 31(4), 395-
408.
* Atakan, S. S., Bagozzi, R. P., & Yoon, C. (2014b). Make it your own: How process valence
and self‐construal affect evaluation of self‐made products. Psychology & Marketing,
31(6), 451-468.
115
Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37(2),
122–147.
Baumeister, R. F. (1999). Self-concept, self-esteem, and identity. In V. J. Derlega, B. A.
Winstead, & W. H. Jones (Eds.), Personality: Contemporary Theory and Research (pp.
339–375). Nelson-Hall Publishers.
Begg, C. B., & Mazumdar, M. (1994). Operating characteristics of a rank correlation test for
publication bias. Biometrics, 50, 1088–1101. doi:10.2307/2533446
Bem, D. J. (1972). Self-perception theory. In Advances in Experimental Social Psychology (Vol.
4, pp. 1-62). Academic Press
Bem, D. J. (1967). Self-perception: An alternative interpretation of cognitive dissonance
phenomena. Psychological Review, 74(3), 183-200.
Bem, D. J. (1965). An experimental analysis of self-persuasion. Journal of Experimental Social
Psychology, 1(3), 199–218. Doi:10.1016/0022-1031(65)90026-0
Bernritter, S. F., van Ooijen, I., & Müller, B. C. (2017). Self-persuasion as marketing technique:
the role of consumers’ involvement. European Journal of Marketing.
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2014). Comprehensive Meta-Analysis
Version 4.0, Englewood, NJ: Biostat
Brehm, J. W., (1966). A theory of psychological reactance. New York, N.Y: Academic Press.
Brehm, J. W., & Cohen, A. R. (1962). Explorations in cognitive dissonance. John Wiley & Sons
Inc. doi: 10.1037/11622-000
Briñol, P., McCaslin, M. J., & Petty, R. E. (2012). Self-generated persuasion: effects of the target
and direction of arguments. Journal of Personality and Social Psychology, 102(5), 925-
940. Doi:10.1037/a0027231
116
Briñol, P., & Petty, R. E. (2003). Overt head movements and persuasion: A self-validation
analysis. Journal of Personality and Social Psychology, 84(6), 1123–1139. doi:
10.1037/0022-3514.84.6.1123
Briñol, P., Rucker, D. D., Tormala, Z. L., & Petty, R. E. (2004). Individual differences in
resistance to persuasion: The role of beliefs and meta-beliefs, in E, S. Knowles & J. A.
Linn (Eds.) Resistance and Persuasion, 83-104.
Brock, T. C. (1968). Relative efficacy of volition and justification in arousing dissonance.
Journal of Personality, 36(1), 49–66.
* Brunner, F., Gamm, F., & Mill, W. (2022). MyPortfolio: The IKEA effect in financial
investment decisions. Journal of Banking & Finance, 106529.
Buechel, E. C., & Janiszewski, C. (2013). A lot of work or a work of art: How the structure of a
customized assembly task determines the utility derived from assembly effort. Journal of
Consumer Research, 40, 960–972.
* Bühren, C., & Pleßner, M. (2020). IKEA Effect vs. Trophy Effect-An Experimental
Comparison (No. 37-2020). MAGKS Joint Discussion Paper Series in Economics.
Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source
versus message cues in persuasion. Journal of Experimental Social Psychology, 39(5),
752-766. doi: 10.1037/00223514.39.5.752
Chan, E. Y. (2015). Endowment effect for hedonic but not utilitarian goods. International
Journal of Research in Marketing, 32, 439– 441.
Cialdini, R. B. (2009). Influence: Science and practice. Boston: Pearson Education.
117
* Cloots, E. (2019). The 'i' in IKEA leads consumers' investments in a digital platform to more
positive outcomes? [Thesis, University of Ghent, Belgium]. ProQuest Dissertations &
Theses Global.
Dahl, D. W., & Moreau, C. P. (2007). Thinking inside the box: Why consumers enjoy
constrained creative experiences. Journal of Marketing Research, 44(3), 357-369.
Damen, T. G. E., Müller, B. C. N., van Baaren, R. B., & Dijksterhuis, A. (2015). Re-examining
the agentic shift: The sense of agency influences the effectiveness of (self)persuasion.
PloS ONE, 10(6). Doi: 10.1371/journal.pone.0128635
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0128635
Dellaert, B. G., & Stremersch, S. (2005). Marketing mass-customized products: Striking a
balance between utility and complexity. Journal of marketing research, 42(2), 219-227.
Dhar, R., & Wertenbroch, K. (2000). Consumer choice between hedonic and utilitarian goods.
Journal of Marketing Research, 37(1), 60-71.
* Diefenbach, S., Jung, S., Diller, T., Franze, C., & Maciejczyk, S. (2018). The secret of self-
made: The potential of different types of consumer participation for product attachment
and commercial value. Social Sciences, 7(4), 52, 1-19.
https://doi.org/10.3390/socsci7040052
Dillard, J. P., & Peck, E. (2000). Affect and persuasion: Emotional responses to public service
announcements. Communication research, 27(4), 461-495.
Dillard, J. P., & Shen, L. (2005). On the nature of reactance and its role in persuasive health
communication. Communication Monographs, 72(2), 144-168. Doi:
10.1080/03637750500111815
118
Dillard, J. P., Shen, L., & Vail, R. G. (2007). Does perceived message effectiveness cause
persuasion or vice versa? 17 consistent answers. Human Communication Research,
33(4), 467-488.
* Dohle, S., Rall, S., & Siegrist, M. (2016). Does self-prepared food taste better? Effects of food
preparation on liking. Health Psychology, 35(5), 500-508.
https://doi.org/10.1037/hea0000315
Drążkowski, D., Trepanowski, R., Chwiłkowska, P., & Majewska, M. (2020). Self-persuasion
increases motivation for social isolation during the COVID-19 pandemic through moral
obligation. Social Psychological Bulletin, 15(4), 1-20
Duval, S., & Tweedie, R. (2000). A nonparametric “trim and fill “method of assessing
publication bias in meta-analysis. Journal of the American Statistical Association, 95, 89–
98. https://doi.org/10.1080/01621459.2000.10473905
Eagly, A. H., & Chaiken, S. (1993). The Psychology of Attitudes. Fort Worth, TX: Harcourt,
Brace, Jovanovich.
Egger, M., Davey Smith, G., Schneider, M., & Minder, C. (1997). Bias in meta-analysis detected
by a simple, graphical test. BMJ, 315, 629-634.
Elms, A. C., & Janis, I. L. (1965). Counter-norm attitudes induced by consonant vs. dissonant
conditions of role-playing. Journal of Experimental Research in Personality, 1(1), 50-60.
* Erkin, A., Piller, F. T., & Wentzel, D. (2018). The boundaries of co-production: How the
interplay of self-printing and branding affects product valuation (Doctoral dissertation,
Universitätsbibliothek der RWTH Aachen).
119
* Erol, K. (2022). Looking at the IKEA effect with someone else's labor: Do the values we set
for our work and someone else's work differentiate? Journal of Accounting, Finance and
Auditing Studies, 8(4), 82-97.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press.
Festinger, L., & Carlsmith, J. M. (1959). Cognitive consequences of forced compliance. The
journal of abnormal and social psychology, 58(2), 203-210. Doi: 10.1037/h0041593
* Fink, L., & Geldman, D. (2017). The effects of consumer participation in product construction
and design on willingness to pay: The case of software. Computers in Human Behavior,
75, 903-911. https://doi.org/10.1016/j.chb.2017.06.039
Fisher, G. J. (2022). How Will Self-Manufacture and the Maker Movement Reshape Consumer
Preferences? Research-Technology Management, 65(4), 18-26.
https://doi.org/10.1080/08956308.2022.2071063
* Franke, N., Keinz, P., & Steger, C. J. (2009). Testing the value of customization: when do
customers really prefer products tailored to their preferences? Journal of marketing,
73(5), 103-121.
* Franke, N., & Piller, F. (2004). Value creation by toolkits for user innovation and design: The
case of the watch market. Journal of Product Innovation Management, 21(6), 401-415.
https://doi.org/10.1111/j.0737-6782.2004.00094.x
* Franke, N., & Schreier, M. (2010). Why customers value self‐designed products: The
importance of process effort and enjoyment. Journal of Product Innovation Management,
27(7), 1020-1031. https://doi.org/10.1111/j.1540-5885.2010.00768.x
120
* Franke, N., Schreier, M., & Kaiser, U. (2010). The “I designed it myself” effect in mass
customization. Management science, 56(1), 125-140.
https://doi.org/10.1287/mnsc.1090.1077
Fredrickson, B. L., & Branigan, C. (2005). Positive emotions broaden the scope of attention and
thought‐action repertoires. Cognition & Emotion, 19(3), 313-332.
Fuchs, C., Prandelli, E., & Schreier, M. (2010). The psychological effects of empowerment
strategies on consumers’ product demand. Journal of marketing, 74(1), 65-79.
Glock, S., Müller, B. C. N., & Ritter, S. (2013). Warning labels formulated as questions
positively influence smoking-related risk perception. Journal of Health Psychology,
18(2), 252-262. doi. 10.1177/1359105312439734
Greenwald, A. G. (1969). The open-mindedness of the counter attitudinal role player. Journal of
Experimental Social Psychology, 5(4), 375-388. doi: 10.1016/0022-1031(69)90031-6
Greenwald, A.G., & Albert, R. D. (1968). Acceptance and recall of improvised arguments.
Journal of Personality and Social Psychology, 8(1), 31-34. doi: 10.1037/h0021237
Harmon-Jones, E., & Mills, J. (2019). An introduction to cognitive dissonance theory and an
overview of current perspectives on the theory. In E. Harmon-Jones (Ed.), Cognitive
dissonance: Reexamining a pivotal theory in psychology (pp. 3–24). American
Psychological Association. https://doi.org/10.1037/0000135-001
Hayes, A. F. (2022). Introduction to mediation, moderation, and conditional process analysis: A
regression-based approach (3rd edition). New York: The Guilford Press.
Hedges, L. V., & Vevea, J. L. (1998). Fixed- and random-effects models in meta-analysis.
Psychological Methods, 3(4), 486–504. https://doi.org/10.1037/1082-989x.3.4.486
121
Heine, S. J., & Lehman, D. R. (1997). Culture, dissonance, and self-affirmation. Personality and
Social Psychology Bulletin, 23(4), 389-400.
Heine, S. J., Lehman, D. R., Markus, H. R., & Kitayama, S. (1999). Is there a universal need for
positive self-regard? Psychological Review, 106(4), 766-794. https://doi.org/10.1037/0033-
295X.106.4.766
* Herd, K., & Moreau, C. P. (2012). The Influence of Identity on Creative Outcomes. ACR
North American Advances.
Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and Persuasion. Yale
University Press.
Inzlicht, M., Shenhav, A., & Olivola, C. Y. (2018). The effort paradox: Effort is both costly and
valued. Trends in cognitive sciences, 22(4), 337-349.
Janis, I. L., & Gilmore, J. B. (1965). The influence of incentive conditions on the success of role
playing in modifying attitudes. Journal of Personality and Social Psychology, 1(1), 17-
27.
Janis, I. L., & Mann, L. (1965). Effectiveness of emotional role-playing in modifying smoking
habits and attitudes. Journal of experimental research in personality, 1(2), 84–90.
Janis, I. L., & King, B. T. (1954). The influence of role playing on opinion change. The journal
of abnormal and social psychology, 49(2), 211-218. doi: 10.1037/h0056957
Jensen, K., & Carter, D. A. (1981). Self‐persuasion: The effects of public speaking on speakers.
Southern Journal of Communication, 46(2), 163-174.p
Jones, E. E. (1990). Constrained behavior and self-concept change. In Self-inference processes:
The Ontario symposium (Vol. 6, pp. 69-86). Hillsdale, NJ: Erlbaum.
Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1990). Experimental tests of the endowment
effect and the Coase theorem. Journal of political Economy, 98(6), 1325-1348.
122
Kirk, C. P., Swain, S. D., & Gaskin, J. E. (2015). I’m proud of it: Consumer technology
appropriation and psychological ownership. Journal of Marketing Theory and Practice,
23(2), 166-184.
Kim, T., Duhachek, A., Briñol, P., & Petty, R. E. (2021). How posting online reviews can
influence the poster’s evaluations. Personality and Social Psychology Bulletin, 47(9),
1401-1413.
King, B. T., & Janis, I. L. (1956). Comparison of the effectiveness of improvised versus non-
improvised role-playing in producing opinion changes. Human relations, 9(2), 177-186.
Kitayama, S., Snibbe, A. C., Markus, H. R., & Suzuki, T. (2004). Is there any “free” choice? Self
and dissonance in two cultures. Psychological science, 15(8), 527-533.
Kruger, J., Wirtz, D., Van Boven, L., & Altermatt, T. W. (2004). The effort heuristic. Journal of
Experimental Social Psychology, 40(1), 91-98.
Kupor, D. M., & Tormala, Z. L. (2015). Persuasion, interrupted: The effect of momentary
interruptions on message processing and persuasion. Journal of Consumer Research,
42(2), 300-315.
Lawrence, D. H., & Festinger, L. (1962). Deterrents and reinforcement: The psychology of
insufficient reward. Stanford University Press.
Lefebvre, S., & Orlowski, M. (2020). Preparation (mis) perception: effects of involvement on
food attributes and desirability. British Food Journal, 123(2). 739-753.
https://doi.org/10.1108/BFJ-03-2020-0166
Li, S., Ritter, S. M., Bi, C., van Baaren, R. B., & Müller, B. C. (2019). Does smokers' self-
construal moderate the effect of (self-) persuasion on smoking?.
123
Li, S., van Baaren, R. B., & Müller, B. C. (2020a). The influence of culture and close others on
the effectiveness of (self)-persuasion. The Journal of General Psychology, 1-30.
Li, S., van Halen, C., van Baaren, R. B., & Müller, B. C. (2020b). Self-persuasion increases
healthy eating intention depending on cultural background. International journal of
environmental research and public health, 17(10), 3405. doi:10.3390/ijerph17103405
Ling, I. L., Liu, Y. F., Lin, C. W. W., & Shieh, C. H. (2020). Exploring IKEA effect in self-
expressive mass customization: Underlying mechanism and boundary conditions. Journal
of Consumer Marketing, 37(4), 365-374. https://doi.org/10.1108/JCM-09-2017-2373
Loman, J. G. B., de Vries, S. A., Kukken, N., van Baaren, R. B., Buijzen, M., & Müller, B. C.
(2019). Quick question or intensive inquiry: The role of message elaboration in the
effectiveness of self-persuasive anti-alcohol posters. Plos one, 14(1), e0211030.
Loman, J. G. B., Müller, B. C. N., Oude Groote Beverborg, A., van Baaren, R. B., & Buijzen, M.
(2018). Self-persuasion in media messages: Reducing alcohol consumption among
students with open-ended questions. Journal of Experimental Psychology: Applied, 24(1),
81–91. doi: 10.1037/xap0000162
Maimaran, M., & Simonson, I. (2011). Multiple routes to self-versus other-expression in
consumer choice. Journal of Marketing Research, 48(4), 755-766.
Marsh, L. E., Gil, J., & Kanngiesser, P. (2022). The influence of collaboration and culture on the
IKEA effect: Does co-creation alter perceptions of value in British and Indian children?
Developmental Psychology, 58(4), 662-670.
Marsh, L. E., Kanngiesser, P., & Hood, B. (2018). When and how does labour lead to love? The
ontogeny and mechanisms of the IKEA effect. Cognition, 170, 245-253.
124
Marzilli Ericson, K. M., & Fuster, A. (2014). The endowment effect. Annual REview of
Economics, 6(1), 555-579.
McGlone, M. S., & Tofighbakhsh, J. (2000). Birds of a feather flock conjointly (?): Rhyme as
reason in aphorisms. Psychological science, 11(5), 424-428.
* Mochon, D., Norton, M. I., & Ariely, D. (2012). Bolstering and restoring feelings of
competence via the IKEA effect. International Journal of Research in Marketing, 29(4),
363-369.
Monnier, A., Lim, S., Latour, K., & Van Osselaer, S. M. (2022). Baking Your Own Cookies:
Does Food Self-Production Increase Consumption? Journal of the Association for
Consumer Research, 7(4), 501-508.
Mugge, R., Schoormans, J. P., & Schifferstein, H. N. (2009). Emotional bonding with
personalised products. Journal of Engineering Design, 20(5), 467-476.
Müller, B. C., Ritter, S. M., Glock, S., Dijksterhuis, A., Engels, R. C., & van Baaren, R. B.
(2016). Smoking-related warning messages formulated as questions positively influence
short-term smoking behavior. Journal of Health Psychology, 21(1), 60-68. doi:
10.1177/1359105314522083
Müller, B. C., van Baaren, R. B., Ritter, S. M., Woud, M. L., Bergmann, H., Harakeh, Z., Engels.
R. C. M. E., & Dijksterhuis, A. (2009). Tell me why… The influence of self-involvement
on short term smoking behavior. Addictive Behaviors, 34(5), 427-431.
Mussweiler, T., & Neumann, R. (2000). Sources of mental contamination: Comparing the effects
of self-generated versus externally provided primes. Journal of Experimental Social
Psychology, 36(2), 194-206. doi: 10.1006/jesp.1999.1415
125
Nabi, R. L. (2003). Exploring the framing effects of emotion: Do discrete emotions differentially
influence information accessibility, information seeking, and policy preference?
Communication Research, 30(2), 224-247.
* Norton, M. I., Mochon, D., & Ariely, D. (2012). The IKEA effect: When labor leads to love.
Journal of Consumer Psychology, 22(3), 453–460. doi:10.1016/j.jcps.2011.08.002
Olson, J. M., & Stone, J. (2005). The Influence of Behavior on Attitudes. In D. Albarracín, B. T.
Johnson, & M. P. Zanna (Eds.), The Handbook of Attitudes (pp. 223–271). Mahwah, N.J.:
Lawrence Erlbaum Associates Publishers.
Okuhara, T., Ishikawa, H., Okada, M., Kato, M., & Kiuchi, T. (2017). Designing persuasive
health materials using processing fluency: a literature review. BMC Research Notes,
10(1), 1-10. Doi: https://doi.org/10.1186/s13104-017-2524-x
Peck, J., & Shu, S. B. (2009). The effect of mere touch on perceived ownership. Journal of
Consumer Research, 36(3), 434-447.
Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral
routes to attitude change. New York: Springer.
Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement as a determinant of
argument-based persuasion. Journal of personality and social psychology, 41(5), 847.
Pierce, J. L., Kostova, T., & Dirks, K. T. (2001). Toward a theory of psychological ownership in
organizations. Academy of Management Review, 26(2), 298-310.
Pingree, R. J. (2007). How messages affect their senders: A more general model of message
effects and implications for deliberation. Communication theory, 17(4), 439-461.
* Predmore, C., Topyan, K., & Trabold Apadula, L. (2021). Impact of process misconception in
becker-degroot-marschak single response value elicitation procedures: An experimental
126
investigation in consumer behavior using the IKEA effect. Economies, 9(4), 173, 1-13.
https://doi.org/10.3390/economies9040173
Raghoebar, S., van Kleef, E., & de Vet, E. (2017). Self-crafting vegetable snacks: testing the
IKEA-effect in children. British Food Journal, 119(6), 1301-1312. doi.10.1108/BFJ-09-
2016-0443
Reber, R., & Unkelbach, C. (2010). The epistemic status of processing fluency as source for
judgments of truth. Review of philosophy and psychology, 1(4), 563-581.
Rosenthal, R. (1991). Meta-analysis: a review. Psychosomatic Medicine, 53(3), 247-271.
* Rusho, Y., & Raban, D. R. (2021). Join the club? Peer effects on information value perception.
Journal of the Association for Information Science and Technology, 72(2), 156-172.
* Rusho, Y., & Raban, D. R. (2020). Hands on: information experiences as sources of value.
Journal of the Association for Information Science and Technology, 71(6), 671-684.
* Rusho, Y., & Raban, D. R., (2018) The effects of information production process on
experience and evaluation. AoIR. https://doi.org/10.5210/spir.v2018i0.10504
* Sarstedt, M., Neubert, D., & Barth, K. (2017). The IKEA effect. A conceptual replication.
Journal of Marketing Behavior, 2(4), 307-312.
Sarup, G. (1981). Role playing, issue importance, and attitude change. Social Behavior and
Personality, 9(2), 191-202.
* Schreier, M. 2006. The value increment of mass-customized products: An empirical
assessment. Journal of Consumer Behaviour, 5(4), 317-327.
Schwardmann, P., Tripodi, E., & Van der Weele, J. J. (2019). Self-persuasion: Evidence from
field experiments at two international debating competitions. CESifo Working Paper No.
7946. Available at https://www.aeaweb.org/articles?id=10.1257/aer.20200372
127
Shaffer, D. R., & Tabor, C. (1980). Salience of own and others' attitudes as determinants of self-
persuasion. Journal of Social Psychology, 111(2), 225-236.
Shmueli, O., Pliskin, N., & Fink, L. (2016). The dark side of emotional involvement in software
development: A behavioral economics perspective. Asia Pacific Journal of Information
Systems, 26(2), 322-337.
Shu, S. B., & Peck, J. (2011). Psychological ownership and affective reaction: Emotional
attachment process variables and the endowment effect. Journal of Consumer
Psychology, 21(4), 439-452.
Song, H., & Schwarz, N. (2008). If it's hard to read, it's hard to do: Processing fluency affects
effort prediction and motivation. Psychological science, 19(10), 986-988.
Straffon, L. M., Agnew, G., Desch-Bailey, C., van Berlo, E., Goclowska, G., & Kret, M. (2022).
Visual attention bias for self-made artworks. Psychology of Aesthetics, Creativity, and the
Arts. https://doi.org/10.1037/aca0000451
Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self.
Advances in experimental social psychology, 21, 261-302. Doi: org/10.1016/S0065-
2601(08)60229-4
Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facilitating conditions of the
human smile: a nonobtrusive test of the facial feedback hypothesis. Journal of
personality and social psychology, 54(5), 768-777.
Stavrositu, C., & Kim, J. (2018). Self-persuasion through mobile applications: Exploring
different routes to health behavioral change. Cyberpsychology, Behavior, and Social
Networking, 21(8), 516-522.
128
Sun, Y., & Sundar, S. S. (2016). Psychological importance of human agency: How self-assembly
affects user experience of robots. In 2016 11th ACM/IEEE International Conference on
Human-Robot Interaction, 189-196. doi.10.1109/HRI.2016.7451751
Susmann, M. W., Xu, M., Clark, J. K., Wallace, L. E., Blankenship, K. L., Philipp-Muller, A. Z.,
Luttrell, A., Wegener, D. T.& Petty, R. E. (2021). Persuasion amidst a pandemic: Insights
from the Elaboration Likelihood Model. European Review of Social Psychology, 1-37.
Talsma, K., Schüz, B., Schwarzer, R., & Norris, K. (2018). I believe, therefore I achieve (and
vice versa): A meta-analytic cross-lagged panel analysis of self-efficacy and academic
performance. Learning and Individual Differences, 61, 136-150.
Valenzuela, A., Dhar, R., & Zettelmeyer, F. (2009). Contingent response to self-customization
procedures: Implications for decision satisfaction and choice. Journal of marketing
research, 46(6), 754-763.
Valkenburg, P. M. (2017). Understanding self-effects in social media. Human Communication
Research, 43(4), 477-490.
Voss, K. E., Spangenberg, E. R., & Grohmann, B. (2003). Measuring the Hedonic and Utilitarian
Dimensions of Consumer Attitude. Journal of Marketing Research, 40(3), 310–320.
https://doi.org/10.1509/jmkr.40.3.310.19238
* Walasek, L., Rakow, T., & Matthews, W. J. (2017). When does construction enhance product
value? Investigating the combined effects of object assembly and ownership on valuation.
Journal of Behavioral Decision Making, 30(2), 144-156.
Wald, R., Heijselaar, E., & Bosse, T. (2021). Make your own: The potential of chatbot
customization for the development of user trust. In Adjunct Proceedings of the 29th ACM
129
Conference on User Modeling, Adaptation and Personalization, 382–387.
https://doi.org/10.1145/3450614.3463600
Wang, Y., Pfeil, U., & Reiterer, H. (2016). Supporting self-assembly: The IKEA effect on
mobile health persuasive technology. In Proceedings of the 2016 ACM Workshop on
Multimedia for Personal Health and Health Care. 19-22. Retrieved from
https://kops.uni-konstanz.de/bitstream/handle/123456789/38188/Wang_0-
399780.pdf?sequence=3&isAllowed=y
Watts, W. A. (1967). Relative persistence of opinion change induced by active compared to
passive participation. Journal of Personality and Social Psychology, 5(1), 4–15
Wegener, D. T., Petty, R. E., & Smith, S. M. (1995). Positive mood can increase or decrease
message scrutiny: the hedonic contingency view of mood and message processing.
Journal of Personality and Social Psychology, 69(1), 5-15.
Wells, G. L., & Petty, R. E. (1980). The effects of overt head movements on persuasion:
Compatibility and incompatibility of responses. Basic and Applied Social Psychology,
1(3), 219–230.
* Wiecek, A., Wentzel, D., & Erkin, A. (2020). Just print it! The effects of self-printing a
product on consumers’ product evaluations and perceived ownership. Journal of the
Academy of Marketing Science, 48(4), 795-811.
Wilson, T. D. (1990). Self-persuasion via self-reflection. In Self-inference processes: The
Ontario symposium (Vol. 6, pp. 43-67). Hillsdale, NJ: Erlbaum.
Winkielman, P., Schwarz, N., Fazendeiro, T. A., & Reber, R. (2003). The hedonic marking of
processing fluency: Implications for evaluative judgment. In J. Musch & K. C. Klauer
130
(Eds.), The psychology of evaluation: Affective processes in cognition and emotion (pp.
189-217). Mahwah, NJ: Lawrence Erlbaum
Wolf, J. R., Arkes, H. R., & Muhanna, W. A. (2008). The power of touch: An examination of the
effect of duration of physical contact on the valuation of objects. Judgment and Decision
Making, 3(6), 476-482.
Wong, Y. J., McDermott, R. C., Zounlome, N. O., Klann, E. M., & Peterson, Z. D. (2020). Self-
persuasion: An experimental evaluation of a sexual aggression preventive intervention for
US college men. Journal of Interpersonal Violence, 1-25. doi:
10.1177/0886260520936369
Zanna, M. P., Lepper, M. R., & Abelson, R. P. (1973). Attentional mechanisms in children's
devaluation of a forbidden activity in a forced-compliance situation. Journal of
Personality and Social Psychology, 28(3), 355-359.
Zimbardo, P. G. (1965). The effect of effort and improvisation on self-persuasion produced by
role-playing. Journal of Experimental Social Psychology, 1(2), 103-120.
131
Appendix
Appendix 1.A.
Figure 15. Search strategy flow-chart
Records identified through initial database
searches.
(n = 1730)
Records after duplicates removed.
(n =1718)
Full-text articles assessed for eligibility
(n = 67)
Additional studies obtained from reference
lists and secondary searches
(n = 53)
Emailed 12 authors for missing statistics
(n = 40)
Studies included in the final meta-analysis
corpus.
(n = 31, k = 55)
Records excluded based on
title or abstract
(n =1651)
Full-text articles excluded:
1.
Not randomized experiment (n = 12 )
2.No relevant outcome (n = 9)
3.IKEA Effect not predictor (n = 3)
(n = 24)
Full-text articles excluded:
1. Not randomized experiment (n = 4)
2. No relevant outcome (n = 5 )
3. Children based sample (n = 4 )
(n = 13 )
Full-text articles excluded:
1. Missing statistics (n = 6)
2. Consumption as outcome (n = 3)
(n = 9)
132
Appendix 1.B
Table 8
Codebook for meta-analysis
Identifiers
Study ID #
Study title
Authors
Year
Main Outcome
Var 2
Value Outcome
1 = monetary value
• estimated value.
• willingness to pay (WTP)
• willingness to accept (WTA)
2 = ranked value (e.g., Likert like scale)
3 = consumption
Effort manipulation
Var 3
customization
1 = builders assembled the product
2 = builders designed the product
3 = both assembled and designed
Var 4
Freedom of
choice
Builder groups had a different number of choice options
Product
Var 5
Tangible design
0 = design/assemble product digitally
1 = design/assemble product physically
Var 6
Tangible Final
0 = No physical final product (all digital)
1 = participants received a physical final product
Var 7
Usage type
1 = more utilitarian
2 = more hedonic
• based on Voss hedonic/utilitarian scale
Moderators – if measured in the study code 1, if not code 0
Var 8
liking
Includes: Liking, preference fit
Var 9
Consumption
133
Var 10
Ownership/
attachment
Includes: Perceived ownership, psychological ownership
Var 11
competence
Includes: Competence, pride, accomplishment, self-affirmation
Sample
Var 12
Population
0 = general population
1 = students (academic)
2 = children (k12)
Var 13
Size
Var 14
Region
1= North America; 2 = West Europe; 3 = Oceania; 4 = East Asia;
5 = Other
Var 15
Mean age
Note for Codebook. Initially, we coded for consumption as an additional outcome variable, and conducted
a separate model to test the influence of effort on consumption. Several studies examined food as a self-
made product and measured its valuation based on consumption rather than a monetary estimate or product
evaluation (e.g., Monnier et al., 2022; REFER). Consumption of food, however, does not necessarily
indicate a greater perceived value, as consumption may be influenced by different factors such as cravings,
coping mechanisms, taste, and personal preferences, thus, consumption may be more akin to usage rather
than an indicator of perceived product value. One study also measured perceived naturalness (Dohle, Rall
& Siegrist 2014), and another assessed caloric perception (Lefebvre & Orlowski, 2020), both of which
could potentially serve as indicators for positive assessment of the food as healthier, but these do not
necessarily translate into greater valuation. People may consume junk food or unhealthy food despite
knowing it is not nutritious. After tightening the inclusion criteria, we decided to remove studies that only
tested consumption as an outcome variable, to avoid comparing apples to oranges. Our omission of studies
that only tested consumption as the main outcome variable does not indicate that the IKEA Effect is not
relevant for food products, but rather it indicates the need to develop a systematic work-frame for future
studies to follow. The outcome variables of food related studies were varied, as, some tested for
consumption or liking of the food, others tested for caloric perception, perceptions of portion and food
desirability. In order to compare these, future studies should consider a consistent set of measures that are
potentially more aligned with the IKEA Effect assumptions, as consumption in itself is not parallel to
valuation as an outcome.
134
Appendix 2.A.1
Table 9
Regression models for post attitudes by text assembly and control variables – Study 1
Model 1
Model 2
Model 3
Model 4
Constant
6.14***
(.105)
5.04***
(.724)
7.49
(.356)
3.347
(.777)
Text Assembly
-.261*
(.149)
-.18
(.147)
-.146
(.139)
-.036
(.121)
Freedom threat
--
-.172**
(.061)
-.142*
(.061)
-.143**
(.053)
Text Evaluation
--
.319**
(.121)
.306**
(.116)
.23*
(.101)
Valuation
--
.006
(.013)
.002
(.012)
-.000
(.011)
Political ideology
--
--
-0.65
(.063)
-.025
(.055)
Gender
--
--
-.725***
(.175)
-.318
(.162)
Issue relevance
--
--
.039
(.143)
.11
(.124)
Pre-attitude
--
--
--
.392***
(.055)
Adjusted R
2
.014
.117
.214
.409
t
1.76
2.46
2.59
3.77153
p
.000
.000
.000
.000
Note: n = 153, *p < .05, **p < .01, ***p < .001. Bootstrap = 5000, CI = 95%
135
Appendix 2.A.2
Table 10
Regression models for text evaluation by text assembly and control variables – Study 1
Model 1
Model 2
Model 3
Model 4
Constant
5.144***
(.069)
5.641***
(.167)
5.978***
(.27)
5.52***
(.447)
Text Assembly
-.25*
(.098)
-.22*
(.098)
-.207*
(.098)
-.188
(.099)
Freedom threat
--
-.131**
(.04)
-.102*
(.042)
-.101*
(.043)
Valuation
--
-.007
(.009)
-.006
(.009)
-.006
(.009)
Political ideology
--
--
-.068
(.045)
-.061
(.045)
Gender
--
--
.026
(.125)
.086
(.133)
Issue relevance
--
--
-.182
(.101)
-.17
(.101)
Pre-attitude
--
--
--
.058
(.045)
Adjusted R
2
.034
.092
.105
.109
t
2.54
2.45
1.97
1.89
p
.012
.000
.001
.001
Note: n = 153, *p < .05, **p < .01, ***p < .001. Bootstrap = 5000, CI = 95%
136
Appendix 2.A.1
Table 11
Regression models for post attitudes by text assembly and control variables – Study 2
Model 1
Model 2
Model 3
Constant
6.03***
(.122)
4.38***
(.804)
5.46***
(.866)
Text Assembly
-.232
(.177)
-.247
(.187)
-.223
(.179)
Freedom threat
--
-.116
(.087)
-.092
(.086)
Text Evaluation
--
.375**
(.115)
.312**
(.113)
Valuation
--
-.005
(.019)
-.006
(.018)
Political ideology
--
--
-.145*
(.068)
Gender
--
--
-.410*
(.205)
Issue relevance
--
--
.046
(.138)
Adjusted R
2
131
.133
.222
t
1.31
2.15
2.17
p
.193
.002
.000
Note: n = 95, *p < .05, **p < .01, ***p < .001. Bootstrap = 5000, CI = 95%
137
Appendix 2.B.1
Persuasive text for stimuli – pilot 1 & 2
Sunscreen Application
Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk of
developing skin cancer. By wearing sunscreen, you are reducing ultraviolet exposure, which is the
main cause of skin damage such as aging, wrinkling, and loss of elasticity, as well as skin cancer
such as melanoma and carcinoma. Furthermore, studies show that sunscreen is important for all
skin types. Although skin cancer is more prevalent in people with lighter skin, it can still affect
those with darker skin. Remember that a sunburn is an immediate reaction, but critical sun damage
accumulates over a lifetime.
Covid-19 Vaccine
Getting vaccinated for COVID-19 is an important practice with several health benefits. By getting
vaccinated you are reducing your likelihood of getting or spreading the COVID-19 virus.
Furthermore, if you do get infected, then you are far less likely to develop severe symptoms. Many
aren't aware that COVID-19 can affect people of all ages. Though it is more dangerous for risky
populations such as older and ill people, it can still affect those who are young and in good health.
A final point to remember that the vaccine is safe and side effects are unlikely. A final point to
remember is that side effects are temporary, and thus are better than long-term damage from the
disease itself.
Flossing Teeth
Flossing regularly can dramatically reduce your risk for cavities and gum disease. By flossing your
teeth, you are scraping excessive plaque, which is known to contribute to inflamed, bleeding, and
painful gums. Furthermore, studies show that flossing is beneficial for most people. This is because
although only 30% of the population may be genetically susceptible to gum disease, the other 70%
can still develop gum disease if plaque builds up over time and inflammation is left untreated.
Therefore, it is important to maintain a daily flossing routine in order to sustain a healthy smile
that will last a lifetime.
138
Appendix 2.B.2
Persuasive text for stimuli – Study 2
Sunscreen
Full paragraph
Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk of
developing skin cancer. By wearing sunscreen, you are reducing ultraviolet exposure, which is the
main cause of skin damage such as aging, wrinkling, and loss of elasticity, as well as skin cancer
such as melanoma and carcinoma. Furthermore, studies show that sunscreen is important for all
skin types. Although skin cancer is more prevalent in people with lighter skin, it can still affect
those with darker skin. Remember that a sunburn is an immediate reaction, but critical sun damage
accumulates over a lifetime.
Assemble condition
• Although
• skin cancer is more prevalent in people with lighter skin, it can still affect those with
darker skin.
• Remember that a sunburn is an immediate reaction, but critical sun damage accumulates
over a lifetime.
• Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk
of developing skin cancer.
• By wearing sunscreen, you are reducing ultraviolet exposure, which is the main cause of
skin damage such as aging and wrinkling.
• Furthermore,
• studies show that sunscreen is important for all skin types.
139
Appendix 3.A
Stimuli material for easy and difficult conditions - Study 3
Read/Control Group (Direct persuasion)
Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk of
developing skin cancer. By wearing sunscreen, you are reducing ultraviolet exposure, which
is the main cause of skin damage such as aging, wrinkling, and loss of elasticity, as well as
skin cancer such as melanoma and carcinoma. Furthermore, studies show that sunscreen is
important for all skin types. Although skin cancer is more prevalent in people with lighter
skin, it can still affect those with darker skin. Finally, remember that a sunburn is an
immediate reaction, but critical sun damage accumulates over a lifetime.
Assemble- easy (IKEA effect 1)
• skin cancer. By wearing sunscreen, you are reducing ultraviolet exposure, which is the
main cause of skin damage such as aging, wrinkling, and loss of elasticity, as well as
• Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk
of developing skin cancer such as melanoma and carcinoma. Furthermore,
• remember that a sunburn is an immediate reaction, but critical sun damage accumulates
over a lifetime.
• studies show that sunscreen is important for all skin types. Although skin cancer is more
prevalent in people with lighter skin, it can still affect those with darker skin. Finally,
Assemble-difficult (IKEA Effect 2)
• skin cancer. By wearing sunscreen, you are reducing ultraviolet exposure, which is the
main cause of skin damage such as aging, wrinkling, and loss of elasticity, as well as
• Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk
of developing
• remember that a sunburn is an immediate reaction, but critical sun damage accumulates
over a lifetime.
• skin cancer such as melanoma and carcinoma. Furthermore,
• skin cancer is more prevalent in people with lighter skin, it can still affect those with
darker skin. Finally,
• studies show that sunscreen is important for all skin types. Although
140
Appendix 3.B
Three color scheme choices for website - customization manipulation
Appendix 3.C
Website top banner based on the color scheme selection.
141
Appendix 3.D
Mock web page
142
Appendix 3.E
Questionnaire study 3
Qualtrics questionnaire - On Your Health
Start of Block: consent
Start of Block: demographic screeners
What is your birth year?
▼ 2007 (77) ... 1939 (530)
Skip To: End of Block If What is your birth year? = 2006
Skip To: End of Block If What is your birth year? = 2007
English Are you fluent in English
o Yes (Highly fluent) (1) ,
o Moderately fluent (2)
o No (3)
End of Block: demographic screeners
Start of Block: preBehave
preBehaveInstruct
In the first section you will be asked about your personal practices. There are no right or wrong answers.
Select the answer that is closest to yours.
Click NEXT to begin.
preBeFlo How often do you floss your teeth on average?
o Never (0)
o Multiple times a day (5)
preBeSun How often do you apply sunscreen on average?
o Never (0)
o Multiple times each day (5)
143
preBeVeg How often do you eat vegetables?
o Never (0)
o During every meal (5)
End of Block: preBehave
Start of Block: dpSun 1
instructS
On the next page you will read a short paragraph concerning a random health issue, from the website On Your Health
Read the paragraph thoroughly.
You will answer questions about this paragraph in the following sections of the survey.
Click "next" once you are ready to begin.
1articleGreen Why sunscreen is good for you!
Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk of developing skin cancer. By wearing
sunscreen, you are reducing ultraviolet exposure, which is the main cause of skin damage such as aging and wrinkling, as well as
skin cancer such as melanoma and carcinoma. Furthermore, studies show that sunscreen is important for all skin types.
Although skin cancer is more prevalent in people with lighter skin, it can still affect those with darker skin. Finally, remember
that a sunburn is an immediate reaction, but critical sun damage accumulates over a lifetime.
Start of Block: dpSunDesign 2
Customize - In your opinion, which color scheme would be a good fit for a health-information website?
Select your preference from the three options, then click next.
o Green (1)
o Blue (2)
o Red (3)
instructSD
On the next page you will read a short paragraph concerning a random health issue, from the website On Your Health, in the
color scheme that you selected.
Read the paragraph thoroughly.
You will answer questions about this paragraph in the following sections of the survey.
Click "next" once you are ready to begin.
Page Break
144
instructSH
On the next page you will read a scrambled paragraph concerning a random health issue from the website On Your Health.
Your task is to reorder the sentences back into a logical paragraph.
Important: You can only proceed after the paragraph is constructed correctly. You have as many attempts as you need.
Click "next" once you are ready to begin.
orderSH
Drag and reorder the items below to recreate the original paragraph
• skin cancer such as melanoma and carcinoma. Furthermore, (1)
• Although (11)
• skin cancer. By wearing sunscreen, you are reducing ultraviolet exposure, which is the main cause of of skin damage
such as aging and wrinkling, as well as (12)
• Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk of developing (2)
• Finally, remember that a sunburn is an immediate reaction, but critical sun damage accumulates over a lifetime. (9)
• studies show that sunscreen is important for all skin types. (3)
• skin cancer is more prevalent in people with lighter skin, it can still affect those with darker skin. (4)
Start of Block: ikeaSunEasy 5
instructSE On the next page you will read a scrambled paragraph concerning a random health issue from the website On Your
Health.
Your task is to reorder the sentences back into a logical paragraph.
Important: You can only proceed after the paragraph is constructed correctly. You have as many attempts as you need.
Click "next" once you are ready to begin.
orderSE
Drag and reorder the items below to recreate the original paragraph
• skin cancer. By wearing sunscreen, you are reducing ultraviolet exposure, which is the main cause of skin damage (12)
• Daily application of sunscreen, even on cloudy days, can dramatically decrease your risk of developing (2)
• skin. Finally, remember that a sunburn is an immediate reaction, but critical sun damage accumulates over a lifetime.
(9)
• such as aging and wrinkling, as well as skin cancer such as melanoma and carcinoma. Furthermore, studies show that
sunscreen is important (3)
• for all skin types. Although skin cancer is more prevalent in people with lighter skin, it can still affect those with
darker (4)
successSE
Success!
You have constructed the paragraph correctly
Click next to see the paragraph on the website "On your health"
145
WTA Imagine you are working as an editor for the website On Your Health. What would you consider an appropriate payment
for the work put into constructing the paragraph?
o $3 (1)
o $6 (2)
o $9 (3)
o $12 (4)
o $15 (5)
o $18 (6)
o $21 (7)
Start of Block: difficult
How difficult or easy was the task of reconstructing the paragraph?
o very easy (1)
o somewhat easy (2)
o neither easy nor difficult (3)
o somewhat difficult (4)
o very difficult (5)
Start of Block: valueOthers
WTP The paragraph you just read was edited for the website On Your Health. What would you consider an appropriate
payment for the editing work put into constructing the paragraph?
o $3 (1)
o $6 (2)
o $9 (3)
o $12 (4)
o $15 (5)
o $18 (6)
o $21 (7)
Start of Block: postSun
146
postInstruct
In the next sections you will rate the extent to which you agree or disagree with the several statements, from Strongly disagree
to Strongly agree
Click NEXT to begin
Wearing sunscreen daily is necessary
Wearing sunscreen daily is good
Wearing sunscreen daily is harmful
Wearing sunscreen daily is foolish
Wearing sunscreen daily is important
Wearing sunscreen daily is effective
o Strongly disagree (1)
o Strongly agree (7)
Start of Block: postBehave
How likely are you to apply sunscreen every day in the next month?
o Extremely unlikely (1)
o Extremely likely (5)
How likely are you to visit the website "On your health" ?
o Extremely unlikely (1)
o Extremely likely (5)
End of Block: postBehave
Start of Block: textEvaluation
textEval
In the next section you will evaluate the paragraph you read.
Rate the extent to which you agree or disagree with each statement from "Strongly Disagree" to "Strongly Agree"
Click Next to begin
The paragraph was logical
The paragraph was interesting
The paragraph was believable
The paragraph was confusing
The paragraph was informative
o Strongly disagree (1)
o Strongly agree (7)
147
The paragraph was logical
The paragraph was interesting
The paragraph was believable
The paragraph was confusing
The paragraph was informative
o Strongly disagree (1)
o Strongly agree (7)
Start of Block: attachment
If someone praised the paragraph, I would feel somewhat praised myself
I feel l emotionally connected to the message in the paragraph
This paragraph has no special meaning for me
o Strongly disagree (1)
o Strongly agree (5)
Start of Block: Freedom threat
In the next section you will answer questions concerning the message of the paragraph.
The paragraph tried to pressure me
The paragraph tried to make a decision for me
The paragraph tried to manipulate me
o Strongly disagree (1)
o Strongly agree (7)
Start of Block: negative emotion
Next rate to what extent you experienced each emotion
Reading the paragraph made me feel
Angry
o Not at all angry (1)
o Slightly angry (2)
o Moderately angry (3)
o Very angry (4)
148
Irritated
o Not at all irritated (1)
o Slightly irritated (2)
o Moderately irritated (3)
o Very irritated (4)
Annoyed
o Not at all annoyed (1)
o Slightly annoyed (2)
o Moderately annoyed (3)
o Very annoyed (6)
End of Block: emotion
You are almost done!
In the final section you will answer a few questions about yourself.
Where would you place yourself on a Political Ideology scale, from "Very Liberal" to "Very Conservative" ?
o Very Liberal (1)
o Very Conservative (7)
Start of Block: relevance
Have you or anyone of your close family and friends suffered from health issues related to sun exposure?
o Yes (1), No (2) , I don't know (99)
Start of Block: Demographics
gender With which gender do you identify?
o Female (1) , Male (2) , Other (99)
149
Education = What is the highest level of education you have completed
o Some high school or less (1)
o High school diploma or GED (2)
o Some college, but no degree (3)
o Associates or technical degree (4)
o Bachelor's degree (5)
o Graduate or professional degree (MA, MS, MBA, PhD, JD, MD, DDS etc.) (6)
o Prefer not to say (7)
Race - Choose one or more races that you consider yourself to be
▢ White or Caucasian (1)
▢ Black or African American (2)
▢ American Indian/Native American or Alaska Native (3)
▢ Asian (4)
▢ Native Hawaiian or Other Pacific Islander (5)
▢ Other (6)
▢ ⊗Prefer not to say (7)
End of Block: Demographics
Start of Block: End of Block
disclosure
Thank you for your participation!
Click next to submit the survey
The purpose of this research is to study how effort may influence attitude. All participants were randomly assigned to one of six
different tasks. The article and website were made up for the purpose of this study. While the website isn’t real, the
information used for the article is real, and was taken from the Skin Cancer Foundation Website.
Results from this research will be used for academic purposes only and your responses will remain strictly confidential. If you
are interested in learning more about this research, please contact Ayellet at [email protected].