ArticlePDF Available

Abstract and Figures

Despite the voluminous evidence in support of the paradoxical finding that providing individuals with more options can be detrimental to choice, the question of whether and when large assortments impede choice remains open. Even though extant research has identified a variety of antecedents and consequences of choice overload, the findings of the individual studies fail to come together into a cohesive understanding of when large assortments can benefit choice and when they can be detrimental to choice. In a meta-analysis of 99 observations (N = 7,202) reported by prior research, we identify four key factors—choice set complexity, decision task difficulty, preference uncertainty, and decision goal—that moderate the impact of assortment size on choice overload. We further show that each of these four factors has a reliable and significant impact on choice overload, whereby higher levels of decision task difficulty, greater choice set complexity, higher preference uncertainty, and a more prominent, effort-minimizing goal facilitate choice overload. We also find that four of the measures of choice overload used in prior research—satisfaction/confidence, regret, choice deferral, and switching likelihood—are equally powerful measures of choice overload and can be used interchangeably. Finally, we document that when moderating variables are taken into account the overall effect of assortment size on choice overload is significant—a finding counter to the data reported by prior meta-analytic research.
Conceptual model of the impact of assortment size on choice overload. Note.-The four antecedents of choice overload are operationalized as follows: (1) The complexity of the choice set describes the aspects of the decision set associated with the particular values of the choice options: the presence of a dominant option in the choice set, the overall attractiveness of the options in the choice set, and the relationship between individual options in the decision set (alignability and complementarity); (2) The difficulty of the decision task refers to the general structural characteristics of the decision problem: time constraints, decision accountability, and number of attributes describing each option; (3) Preference uncertainty refers to the degree to which individuals have articulated preferences with respect to the decision at hand and has been operationalized by two factors: the level of product-specific expertise and the availability of an articulated ideal point; and (4) The decision goal reflects the degree to which individuals aim to minimize the cognitive effort involved in making a choice among the options contained in the available assortments and is operationalized by two measures: decision intent (buying vs. browsing) and decision focus (choosing an assortment vs. choosing a particular option). In this context, we expect higher levels of decision task difficulty, greater choice set complexity, higher preference uncertainty, and a more prominent, effort-minimizing goal to produce greater choice overload.
… 
Content may be subject to copyright.
Research Review
Choice overload: A conceptual review and meta-analysis
Alexander Chernev
a,
, Ulf Böckenholt
a
, Joseph Goodman
b
a
Kellogg School of Management, Northwestern University, USA
b
Compass Lexecon, Chicago, USA
Received 5 October 2012; received in revised form 18 August 2014; accepted 23 August 2014
Available online 29 August 2014
Abstract
Despite the voluminous evidence in support of the paradoxical nding that providing individuals with more options can be detrimental to
choice, the question of whether and when large assortments impede choice remains open. Even though extant research has identied a variety of
antecedents and consequences of choice overload, the ndings of the individual studies fail to come together into a cohesive understanding of when
large assortments can benet choice and when they can be detrimental to choice. In a meta-analysis of 99 observations (N= 7202) reported by
prior research, we identify four key factorschoice set complexity, decision task difculty, preference uncertainty, and decision goalthat
moderate the impact of assortment size on choice overload. We further show that each of these four factors has a reliable and signicant impact on
choice overload, whereby higher levels of decision task difculty, greater choice set complexity, higher preference uncertainty, and a more
prominent, effort-minimizing goal facilitate choice overload. We also nd that four of the measures of choice overload used in prior research
satisfaction/condence, regret, choice deferral, and switching likelihoodare equally powerful measures of choice overload and can be used
interchangeably. Finally, we document that when moderating variables are taken into account the overall effect of assortment size on choice
overload is signicantanding counter to the data reported by prior meta-analytic research.
© 2014 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved.
Keywords: Choice overload; Assortment; Decision complexity; Meta-analysis
Contents
Introduction ............................................................... 334
The pros and cons of large assortments ................................................. 334
Choice overload in consumer decision making ............................................. 335
Choice overload in individual decision making ........................................... 335
Conceptualizing the impact of assortment size on choice overload .................................. 336
Decision task difculty ........................................................ 337
Choice set complexity ........................................................ 337
Preference uncertainty ........................................................ 338
Decision goal ............................................................. 339
Method ................................................................. 339
The data ............................................................... 339
The model .............................................................. 340
Results .................................................................. 344
Model t............................................................... 344
Effects of the specic moderators of choice overload ........................................ 344
Corresponding author. Fax: +1 847 491 2498.
E-mail address: ach@kellogg.northwestern.edu (A. Chernev).
http://dx.doi.org/10.1016/j.jcps.2014.08.002
1057-7408/© 2014 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved.
Available online at www.sciencedirect.com
ScienceDirect
Journal of Consumer Psychology 25, 2 (2015) 333 358
The mean effect of assortment size on choice overload ........................................ 346
The effect of assortment size across individual experiments ..................................... 348
Publication bias ............................................................ 350
Discussion ................................................................ 351
Key ndings ............................................................. 351
Prior meta-analytic research ...................................................... 352
Future research ............................................................ 353
Appendix A. Overview of the analyzed studies .............................................. 354
Appendix B. Reanalyzing the data from prior meta-analytic research .................................. 356
References ................................................................ 354
Introduction
The importance of assortment decisions for both retailers and
manufacturers has been underscored by numerous research
articles, marketing textbooks, and the popular press (Iyengar,
2010; Levy & Weitz, 2006; Schwartz, 2003). Because of its
importance, the topic of how product assortment influences
consumer choice has generated a substantial amount of interest
across different research domains, including economics, analytical
and empirical modeling, individual and group decision making,
and social psychology (Broniarczyk, 2008; Chernev, 2012; Kahn,
1999; Kahn, Weingarten, & Townsend, 2013; Lancaster, 1990;
Lehmann, 1998; Simonson, 1999).
Within assortment research, the topic of the negative
consequences of large assortments has attracted a disproportion-
ate amount of interest among researchers. This interest can be
attributed largely to the paradoxical finding that variety can be
detrimental to choice, which challenged the conventional wisdom
that providing consumers with more options always facilitates
choice (Iyengar & Lepper, 2000; Reibstein, Youngblood, &
Fromkin, 1975). Building on these findings, recent research has
moved beyond simply documenting choice overload to identify-
ing its antecedents and boundary conditions. In doing so,
researchers have identified a number of important moderators
of choice overload, such as attribute alignability (Gourville &
Soman, 2005), consumer expectations (Diehl & Poynor, 2010),
availability of an ideal point (Chernev, 2003b), personality traits
and cultural norms (Iyengar, Wells, & Schwartz, 2006), option
attractiveness (Chernev & Hamilton, 2009), decision focus
(Chernev, 2006), construal level (Goodman & Malkoc, 2012),
time pressure (Haynes, 2009), product type (Sela, Berger, & Liu,
2009), consumer expertise (Mogilner, Rudnick, & Iyengar,
2008), and variety seeking (Oppewal & Koelemeijer, 2005).
Despite the voluminous evidence that large assortments can
lead to choice overload, the question of whether and when large
assortments are detrimental to choice remains open. Indeed, even
though extant research has identified a variety of antecedents and
consequences of choice overload, the individual studies use
diverse independent and dependent variables. As a result, the
findings of these studies fail to come together in a cohesive
understanding of whether and when assortment size is likely to
lead to choice overload. The goal of our research, therefore, is to
identify factors that reliably moderate the impact of assortment
size on choice overload and generalize them into an overarching
conceptual framework. To achieve this goal, we abstract from the
specific variables and manipulations reported in the individual
studies to identify the key drivers of choice overload, quantify the
effect sizes associated with these factors, and evaluate their
impact on choice overload.
Our analysis is organized as follows.First,wediscussthepros
and cons of large assortments, focusing on how assortment size
influences individual decision processes. This is followed by a
conceptual analysis of the antecedents of choice overload, in
which we identify four key drivers that are likely to influence the
impact of assortment size on choice overload. We then present our
methodology in more detail, followed by a summary of our key
findings. This research concludes with a discussion in which we
highlight our theoretical contributions, discuss the managerial
implications, and outline directions for future research.
The pros and cons of large assortments
Offering consumers a large variety of options to choose from
can have a two-pronged impact on choice: It can both benefit and
hinder choice. The most intuitive benefit, featured prominently in
economics research, is that the greater the number of options in the
choice set, the higher the likelihood that consumers can find a close
match to their purchase goals (Baumol & Ide, 1956; Hotelling,
1929). A related economic explanation of consumer preference for
larger assortments involves the greater efficiency of time and effort
involved in identifying the available alternatives in the case of
one-stop shopping associated with retailers offering larger
assortments (Betancourt & Gautschi, 1990; Messinger &
Narasimhan, 1997).
It has also been proposed that larger assortments might lead to
stronger preferences because they offer option value (Reibstein et
al., 1975), allow consumers to maintain flexibility in light of
uncertainty about future tastes (Kahn & Lehmann, 1991;
Kahneman & Snell, 1992; Kreps, 1979), and accommodate
consumers' future variety-seeking behavior (Inman, 2001; Kahn,
1995; Levav & Zhu, 2009; Ratner, Kahn, & Kahneman, 1999;
Van Herpen & Pieters, 2002). It has further been argued that
consumers might experience additional utility simply from having
multiple items in the choice set because it creates the perception of
freedom of choice (Kahn, Moore, & Glazer, 1987), enhances the
enjoyment of shopping (Babin, Darden, & Griffin, 1994), and
strengthens overall choice satisfaction (Botti & Iyengar, 2004).
Finally, it has been proposed that larger assortments influence
consumer preferences by reducing the uncertainty of whether the
357
334 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
choice set at hand adequately represents all potentially available
options. Prior research has shown that consumers may delay their
purchasing because they are uncertain about the degree to which
the available set is representative of the entire roster of possible
options (Greenleaf & Lehmann, 1995; Karni & Schwartz, 1977).
To illustrate, consumers might feel more confident when
selecting from a retailer that offers a larger assortment because
it is less likely that a potentially superior alternative is not
represented in the available choice set.
Despite their multiple benefits, larger assortments have a
number of important drawbacks. Prior research has shown that the
benefits of greater variety are, at least partially, offset by a
corresponding increase in the cognitive costs associated with
choosing from a larger assortment. In this context, it has been
shown that reducing the size of an assortment can actually increase
the purchase likelihood from that assortment. For example, it was
shown that consumers were more likely to make a purchase when
presented with an assortment comprising 6 flavors of jam than
with an assortment comprising 24 flavors (Iyengar & Lepper,
2000). Similar findings have been documented in a variety of
product categories, such as chocolates (Berger, Draganska, &
Simonson, 2007; Chernev, 2003b), consumer electronics
(Chernev, 2003a), and mutual funds (Huberman, Iyengar, &
Jiang, 2007; Ketcham, Lucarelli, Miravete, & Roebuck, 2012;
Morrin,Broniarczyk,Inman,&Broussard,2008).
Recent research also has argued that consumer preference
for larger assortments is likely to be subject to diminishing
returns because the marginal benefits from each additional
alternative tend to decrease with the increase in assortment size
(Chernev & Hamilton, 2009; Oppewal & Koelemeijer, 2005).
Given that the increase in benefits accrues at a decreasing rate,
at some point it is likely to be offset by the additional costs of
evaluating the available alternatives (Roberts & Lattin, 1991).
Thus, it has been shown that the probability of purchasing a
brand, reflected in the brand's market share, tends to decrease
after the product line achieves a certain size (Draganska & Jain,
2005).
It has further been argued that larger assortments tend to
shift consumers' ideal points in a way that makes them more
difficult to attain (Chernev, 2003b; Schwartz et al., 2002).
Larger assortments have also been found to inflate consumers'
expectations of finding their ideal option in the available
assortment and the degree of preference match they can achieve
(Diehl & Poynor, 2010). Consequently, it has been proposed
that choices from larger assortments can lead to disconfirmation
of consumer expectations, resulting in greater choice deferral
and lower satisfaction with the chosen option.
In this research we focus on the negative consequences of
large assortments, specifically those factors that are likely to
influence whether and how larger assortments will produce
choice overload. We discuss the antecedents and consequences
of choice overload in more detail in the following sections.
Choice overload in consumer decision making
We start by discussing prior research examining the impact of
assortment size on choice overload to underscore the importance
of developing a theory-based approach to generalizing the
findings of the individual studies. We then propose a conceptual
model that identifies four key factors that influence the impact of
assortment size on choice overload.
Choice overload in individual decision making
The term choice overloadalso referred to as overchoice
is typically used in reference to a scenario in which the
complexity of the decision problem faced by an individual
exceeds the individual's cognitive resources (Simon, 1955;
Toffler, 1970). In this research, our main focus is on a particular
type of choice overloadone in which the decision complexity
is caused, at least partially, by the (large) number of available
decision alternatives (Iyengar & Lepper, 2000).
Because choice overload is a mental construct describing
the subjective state of the decision maker, it cannot be
directly observed; instead, it is reflected in a series of
objective indicators, which, in turn, are used to measure
choice overload. In this context, two types of indicators of
choice overload can be identified: process-based indicators
describing the subjective state of the decision maker and
outcome-based indicators reflecting the decision maker's
observable behavior.
As a subjective state, choice overload is captured by
changes in consumers' internal states, such as decision
confidence, satisfaction, and regret, whereby higher levels
of choice overload are likely to produce lower levels of
satisfaction/confidence and higher levels of regret. Thus,
compared to individuals not experiencing choice overload,
those experiencing overload are (1) less likely to be satisfied
with their decisions (Botti & Iyengar, 2004), (2) less
confident that they have chosen the best option (Haynes,
2009), and (3) prone to more post-decision regret (Inbar,
Botti, & Hanko, 2011).
Behavioral consequences of choice overload, on the other
hand, include factors that capture consumer actions such as the
likelihood of deferring choice, the likelihood of reversing an
already made choice, the preference for larger assortments, and
the nature of the ultimately chosen option. In this context,
greater levels of choice overload are associated with greater
probability of choice deferral, greater switching likelihood,
decreased preference for larger assortments, and greater
preference for easily justifiable options. Thus, compared to
individuals not experiencing choice overload, those experienc-
ing overload are (1) less likely to make a choice from a
particular assortment (Iyengar & Lepper, 2000), (2) more
likely to reverse their initial choice (Chernev, 2003b), (3) less
likelytodisplayapreference for larger assortments (Chernev,
2006),and(4)morelikelytochooseanoptionthatcanbe
easily justified (Sela et al., 2009).
Note that the above indicators of choice overload do not
represent a complete list of all viable measures of choice
overload; rather, these are the measures that represent the most
common decision scenarios and have been frequently utilized
by prior research.
335A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
Conceptualizing the impact of assortment size on choice
overload
Recent assortment research has identified a number of
important antecedents and consequences of choice overload.
While this plethora of predictors and measures facilitates
understanding the impact of assortment size on choice
overload, it also complicates generalization of the findings of
the individual studies because different experiments use diverse
independent and dependent variables. Indeed, whereas numer-
ous studies identify a variety of factors that are likely to
influence choice overloadincluding attribute alignability,
attribute complementarity, ideal point availability, option
attractiveness, consumer expertise, variety seeking, time
pressure, product type, and need for cognitionconvergence
of these findings is difficult to achieve because there is little
overlap between the study-specific independent variables. In
the same vein, different studies use different dependent
variables to measure choice overload, including satisfaction,
confidence, the likelihood of deferring choice, and the
likelihood of switching to an alternative option. Therefore, in
order to make meaningful cross-study comparisons, one must
generalize the study-specific moderators and measures into
theoretically meaningful constructs and examine whether and
how these constructs influence choice overload.
Generalizing the effects of study-specific factors involves
identifying higher level constructs and linking each study factor to
one of these constructs. Accordingly, in this research we identify
four such constructs and then test their validity by examining the
ability of these constructs to explain the variance in prior studies.
On a more general level, we view the impact of assortment size on
consumer decision processes as a function of two types of factors:
(1) extrinsic factors that define the decision problem and are
similar across individuals and (2) intrinsic factors that reflect
individuals' idiosyncratic knowledge and motivation and are
particular to each decision maker.
Building on prior research in the domain of behavioral
decision theory and choice, we divide extrinsic factors into two
categories: task factors and context factors (Payne, Bettman, &
Johnson, 1993). Here, task factors describe the general structural
characteristics of the decision problem, including number of
alternatives, number of attributes describing each option, time
constraints, decision accountability, and information presentation
mode. In contrast, context factors describe the aspects of the
decision associated with the particular values of the choice
options, including the similarity and the overall attractiveness of
the alternatives. In this research we refer to the task factors and
their impact on choice overload as decision task difficulty and to
the context factors as choice set complexity.
Unlike extrinsic factors, which refer to the characteristics of
the decision problem and are similar across individuals,
intrinsic factors are particular to the decision maker. Two
specific factors have been discussed in prior research examin-
ing the impact of assortment size on choice overload:
preference uncertainty and decision goals. Here, preference
uncertainty refers to the degree to which individuals have
articulated preferences with respect to the decision at hand and
includes factors such as the level of product-specific expertise
and the availability of an articulated ideal point (Chernev,
2003b). The decision goal, on the other hand, reflects the
degree to which a consumer's goal involves choosing among
the options in a given assortment (Chernev & Hamilton, 2009).
Our theorizing is summarized in Fig. 1. The four factors
discussed abovedecision task difficulty, choice set com-
plexity, preference uncertainty, and decision goalcomprise
the key moderators that could potentially influence the impact
Number
of options
Choice set
complexity
Subjective
state
Choice
satisfaction
Decision
regret
Choice
overload
Decision task
diculty
Preference
uncertainty
Decision
goal Behavioral
outcome
Choice
deferral
Switching
likelihood
Assortment
choice
Antecedents of choice overload Consequences of choice overload
Option
selection
Decision
condence
Extrinsic (objective) factors
Intrinsic (subjective) factors
Fig. 1. Conceptual model of the impact of assortment size on choice overload. Note.The four antecedents of choice overload are operationalized as follows: (1) The
complexity of the choice set describes the aspects of the decision set associated with the particular values of the choice options: the presence of a dominant option in
the choice set, the overall attractiveness of the options in the choice set, and the relationship between individual options in the decision set (alignability and
complementarity); (2) The difficulty of the decision task refers to the general structural characteristics of the decision problem: time constraints, decision
accountability, and number of attributes describing each option; (3) Preference uncertainty refers to the degree to which individuals have articulated preferences with
respect to the decision at hand and has been operationalized by two factors: the level of product-specific expertise and the availability of an articulated ideal point; and
(4) The decision goal reflects the degree to which individuals aim to minimize the cognitive effort involved in making a choice among the options contained in the
available assortments and is operationalized by two measures: decision intent (buying vs. browsing) and decision focus (choosing an assortment vs. choosing a
particular option). In this context, we expect higher levels of decision task difficulty, greater choice set complexity, higher preference uncertainty, and a more
prominent, effort-minimizing goal to produce greater choice overload.
336 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
of assortment size on choice overload. Choice overload is
then measured as a subjective state of the decision maker
(satisfaction, confidence, and regret) and/or as a specific
behavioral outcome (choice deferral, switching likelihood,
assortment choice, and option selection).
We discuss the impact of each of the four factors outlined in
Fig. 1 in more detail and make directional predictions regarding
their impact in the following sections. We then proceed to test
the validity of the proposed model by examining its ability to
account for the findings reported by the existing empirical data.
When discussing the specific measures encompassed by each of
the above factors, we focus on measures that have been used by
prior research and, thus, are available for the purposes of
meta-analysis. In the general discussion section, we address
some of the additional measures that are likely to moderate the
impact of assortment on choice overload but have not yet
gained empirical support.
Decision task difficulty
The difficulty of the decision task reflects the general
structural characteristics of the decision problem without
influencing the values of the particular choice options (Payne et
al., 1993). Prior research has argued that a number of
decision-task factorsincluding time constraints,decision
accountability,number of attributes describing each option,
and presentation formatare likely to influence the impact of
assortment size on choice overload. We discuss the effects of
these three factors on choice overload in more detail below.
One decision task factor that is likely to influence the impact of
assortment size on choice overload is the imposition of time
constraints. Specifically, it has been argued that an external limit
on the length of the evaluation period increases the cognitive
challenge associated with making a choice and forces consumers
to engage in a less systematic evaluation of the available
alternatives (Bettman, Luce, & Payne, 1998). This nonsystematic
processing of the available alternatives, in turn, has been shown to
lower consumer satisfaction with the chosen alternatives and
diminish consumers' confidence in their decisions (Dhar &
Nowlis, 1999; Haynes, 2009). It has further been shown that the
impact of the assortment size on decision regret is a function of the
time pressure experienced by individuals, such that the subjective
feeling of being rushed accounts for greater regret when choosing
from larger sets (Inbar et al., 2011).
Another decision task factor that could influence the impact
of assortment size on choice overload is decision accountability
(e.g., requiring consumers to justify their choices). Related
research has shown that preference for larger assortments tends
to increase when consumers expect to have to justify their
choice of an assortment to others but tends to decrease when
consumers have to justify the choice of a particular option from
the available assortments (Chernev, 2006; Ratner & Kahn,
2002; Scheibehenne, Greifeneder, & Todd, 2009). It has further
been shown that when making a choice from a given
assortment, decision accountability can decrease the likelihood
of making a choice from larger (relative to smaller) assortments
(Gourville & Soman, 2005).
Yet another decision task factor likely to influence the
impact of assortment size on choice overload is the number of
attributes describing the available options (Chernev, 2003b;
Greifeneder, Scheibehenne, & Kleber, 2010; Hoch, Bradlow, &
Wansink, 1999). Specifically, it has been argued that the more
dimensions on which products are differentiated, the more
complex a choice becomes since consumers need to sift through
additional information to compare the options before ultimately
making a choice. Thus, choosing from a set of items described
along a single dimension (e.g., color) is likely to be less
cognitively taxing than choosing from a set of items described
along multiple attributes (e.g., color, design, durability). In
addition to increasing the amount of information that needs to
be evaluated, increasing the number of attributes describing the
available options also increases the number of dimensions on
which each of the available options is inferior to the other
options in the choice set, further complicating choice.
Prior research has further argued that the impact of
assortment size on choice overload can also be influenced by
the presentation format of the individual options. Thus,
ordering options in a given assortment has been found to
decrease search costs, thus decreasing the difficulty of choosing
an item from larger assortments (Diehl, 2005; Diehl, Kornish,
& Lynch, 2003; Mogilner et al., 2008). In the same vein,
research by Hoch et al. (1999) has documented that consumers
are more satisfied with and are likely to choose assortments that
offer a high variety of options displayed in an organized rather
than random manner. Recent research has further shown that
choice overload is likely to be a function of whether the
assortments are presented in a visual or verbal format, whereby
visual presentation is associated with less systematic processing
and is more likely to lead to overload in the case of larger
assortments compared to verbal (text-based) presentation
(Townsend & Kahn, 2014).
Building on the above findings, we expect that decision task
difficulty can have a significant impact on the way assortment
size influences choice overload. Specifically, we expect that
higher levels of decision task difficultyoperationalized in
terms of time constraints, decision accountability, number of
attributes describing each option, and the complexity of the
presentation formatwill lead to greater choice overload.
Choice set complexity
The complexity of the choice set reflects the aspects of the
decision task that influence the values of the particular choice
options without necessarily influencing the structural aspects of
the decision problem at hand (Payne et al., 1993). Prior research
has argued that a number of choice-set-complexity factors,
including the presence of a dominant option,theoverall
attractiveness of the choice options,aswellasthealignability
and the complementarity of the options, are likely to influence the
impact of assortment size on choice overload. We discuss the
effects of these factors on choice overload in more detail below.
One of the determinants of the complexity of the choice set is
whether it contains a dominant optionthat is, an option superior
to all other available options for a given individual (Huber, Payne,
337A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
& Puto, 1982). In this context, it has been shown that consumers
are more likely to make a purchase from an assortment when it
contains a dominant option than when such an option is absent
(Boatwright & Nunes, 2001; Broniarczyk, Hoyer, & McAlister,
1998; Chernev, 2006; Oppewal & Koelemeijer, 2005). Similarly,
adding an inferior option that enhances the dominance of one of
the existing options has been shown to increase the likelihood that
a choice will be made from an assortment (Dhar, 1997), whereas
adding equally attractive optionshasbeenreportedtohavethe
opposite effect, increasing the likelihood that choice will be
deferred (Dhar, 1997; Dhar & Simonson, 2003; Tversky & Shafir,
1992). Furthermore, adding an inferior option has been shown to
increase the share of the dominant option, a finding commonly
referred to as the attraction effect (Huber et al., 1982; Simonson,
1989; Simonson & Tversky, 1992). Thus, the finding that
consumers are more likely to make a purchase from an assortment
containing a dominant option is consistent with the notion that the
availability of a dominant option decreases choice overload,
consequently increasing probability of purchase.
The impact of assortment size on choice overload is also
influenced by the attractiveness of the choice options.Some
assortments comprise options that are of higher quality and, hence,
are likely to be perceived as more attractive, whereas other
assortments comprise options that are of lower quality and are
likely to be perceived as relatively less attractive. In this context,
prior research has argued that option attractiveness influences the
way consumers choose among assortments, such that smaller
assortments are preferred to larger ones when these assortments
are composed of more attractive rather than less attractive options
(Chernev & Hamilton, 2009). Thus, consumers are more likely to
prefer smaller assortments when these assortments are curated to
include the most attractive options from larger assortments.
Prior research has further shown that the impact of assortment
size on choice overload is a function of the alignability of the
attributes describing the options in the assortment. Here alignability
describes the relationships among the attribute levels of the options
in a given assortment. Nonalignable attributes describe a scenario
in which a given feature is present in one of the options and absent
in the others, whereas alignable attributes describe a scenario in
which objects have different (but nonzero) levels of a given
attribute (Markman & Medin, 1995). In this context, it has been
argued that increasing the size of assortments whose options are
differentiated by alignable attributes reportedly can lead to an
increase in purchase probability from that assortment, whereas
increasing the size of assortments differentiated by options with
nonalignable attributes has been shown to have the opposite effect
of decreasing purchase probability (Gourville & Soman, 2005).
Moreover, research has linked attribute alignability to satisfaction
with choice, following an inverted U-shape for options differen-
tiated on nonalignable (but not alignable) attributes (Griffin &
Broniarczyk, 2010).
A related argument has been advanced by Chernev (2005),who
shows that the impact of assortment size on choice overload is also
a function of feature complementarity, defined as the extent to
which features complement one another with respect to their ability
to fulfill a particular consumer need. Thus, increasing a product
assortment by adding options differentiated by complementary
features tends to lower the attractiveness of all alternatives in that
assortment. In this context, purchase likelihood from a given
assortment is shown to be a function of the complementarity of its
options, such that choice deferral is greater for assortments
comprising complementary rather than noncomplementary op-
tions. Furthermore, increasing assortment size by adding noncom-
plementary options tends to increase purchase likelihood from the
assortment, whereas increasing assortment size by adding
complementary options tends to decrease purchase likelihood
from the assortment.
The above research suggests that choice set complexity can
have a significant impact on whether and how assortment size
influences choice overload. Specifically, we expect that higher
levels of choice set complexityoperationalized in terms of the
presence of a dominant option, as well as the overall
attractiveness, alignability, and complementarity of the choice
optionswill lead to greater choice overload.
Preference uncertainty
Preference uncertainty refers to the degree to which
individuals have articulated preferences with respect to the
decision at hand, meaning that they understand the benefits of
the choice options and can prioritize these benefits when
trading off the pros and cons of the choice options (Chernev,
2003b). This factor has been operationalized in prior research in
two ways: in terms of the level of product-specific expertise and
in terms of the availability of an articulated ideal point.
Research has argued that the impact of assortment size on
choice overload is a function of consumers' expertise and, in
particular, their knowledge about the attributes and attribute
levels describing the available alternatives. In this context, it has
been shown that for consumers who are unfamiliar with the
product category, choices from larger assortments are more likely
to lead to choice deferral and weaker preferences for the selected
alternative than choices from smaller assortments. In contrast, for
expert consumers, the impact of assortment size is reversed,
leading to greater likelihood of choice deferral and weaker
preferences for the chosen alternative in the context of smaller
rather than larger assortments (Chernev, 2003b; Mogilner et al.,
2008; Morrin, Broniarczyk, and Inman, 2012).
Choice overload is also likely to be a function of the degree
to which consumers have an articulated ideal point. Whereas
product expertise implies knowledge of the product category
and the product at hand, the availability of an articulated ideal
point implies that consumers have well-defined preferences
within a given category. Thus, the availability of an ideal point
goes beyond product expertise and implies a hierarchical
attribute structure and already articulated attribute trade-offs
(Carpenter & Nakamoto, 1989; Dhar, 1997; Wansink, Kent, &
Hoch, 1998). Because the articulation of attribute trade-offs is
essential for choice, the availability of an ideal attribute
combination effectively increases the compatibility of consum-
er preference structures with the decision task, thus reducing
the structural complexity of the decision. Therefore, prefer-
ences based on an articulated ideal point will be more effective
in reducing the structural complexity of the decision than will
338 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
preferences that do not involve an ideal point. Because large
assortments are generally associated with more complex
decisions, the differential impact of ideal point articulation is
likely to be more pronounced for larger than for smaller
assortments (Chernev, 2003b). Accordingly, consumers with an
available ideal point are more likely to have stronger
preferences for and make a purchase from larger assortments
than consumers without an available ideal point, who are more
likely to have stronger preferences for and make a purchase
from smaller assortments.
Building on the above research, we expect that individuals'
preference uncertainty can influence the impact of assortment size
on choice overload. Specifically, we predict that a greater degree
of choice overload will result from higher levels of preference
uncertainty, defined in terms of the level of product-specific
expertise and the availability of an articulated ideal point.
Decision goal
The decision goal reflects the degree to which individuals aim
to minimize the cognitive effort involved in making a choice
among the options contained in the available assortments. The
importance of the decision goal as a factor contributing to choice
overload is underscored by the fact that overload is, at least in
part, driven by consumers' inability to make a tradeoff among the
available optionsan effect that is more pronounced for choices
from larger than from smaller assortments (Chernev, 2003b).
Whereas most assortment research has focused on scenarios in
which consumers' goals involve making a choice from the
available assortments, this is not always the case, and on many
occasions consumer decisions do not involve such choices. In this
research, we identify three factors that could lead to scenarios in
which consumers might not aim to minimize cognitive effort:
decision intent (buying vs. browsing), decision focus (choosing
an assortment vs. choosing a particular option), and level of
construal (high vs. low).
Consumers' decision intent reflects whether they approach
the decision task with the explicit goal of making a choice (i.e.,
buying) or merely to consider the available alternatives without
the explicit goal of selecting one or more of the available
options (i.e., browsing). Indeed, on some occasions, consumers
approach the available assortments with the cognitive goal of
learning more about the available options and/or their own
preferences. In the same vein, consumers might approach the
available assortments with the affective goal of deriving
pleasure from the evaluation process itself (Kahn & Ratner,
2005; Kahn & Wansink, 2004). In this context, one can argue
that decisions associated with a browsing goal are less likely to
lead to cognitive overload compared with decisions that involve
the goal of making a choice.
Consistent with this line of reasoning, prior research has
shown that consumers who evaluate assortment options with a
goal of browsing rather than making a choice are less likely to
face cognitive overload when presented with more extensive
assortments (Chernev & Hamilton, 2009; Choi & Fishbach,
2011; Hamilton & Chernev, 2010). In the same vein, research
has shown that experienced consumers who seek variety are
more satisfied with larger assortments than with smaller ones
(Oppewal & Koelemeijer, 2005). Indeed, because the search
itself is utility enhancing for these consumers, there are fewer
(if any) negative consequences associated with larger assort-
ments and, hence, less likelihood of choice overload.
Even when consumers approach assortments with the intent of
making a choice, their decision might not necessarily involve
choosing among the available options but might involve choosing
among the assortments themselves (Arentze, Oppewal, &
Timmermans, 2005; Kahn & Lehmann, 1991). Because choices
among assortments do not necessarily involve evaluating the
individual options in these assortments and trading off their pros
and cons, larger assortments are not necessarily associated with
greater cognitive effort and hence choice overload. Therefore, one
can argue that the impact of assortment size on choice overload is
likely to be a function of consumers' decision focus and,
specifically, whether they consider choosing among assortments
or choosing an item from a given assortment.
In this context, focusing consumerattentiononchoosingan
assortment tends to enhance the benefits of variety while
de-emphasizing the cognitive costs associated with making a
choice, thus strengthening the preference for larger assortments. In
contrast, focusing individuals' attention on choosing a specific item
from a given assortment tends to make the difficulty of choosing
from larger assortments more prominent, consequently increasing
the preference for smaller assortments (Chernev, 2006;seealso
Sood, Rottenstreich, & Brenner, 2004; Huffman & Kahn, 1998).
Therefore, varying the decision focus can systematically vary the
choice overload experienced by consumers, such that larger
assortments are more likely to lead to choice overload when the
goal involves choosing an option from an assortment rather than
choosing among assortments.
Research examining choice among assortments has further
shown that the impact of assortment size on choice overload is a
function of the level of construal. Specifically, the construal-level
theory predicts that the way individuals conceptualize the
decision processas a high-level, abstract, and distant process
or a low-level, concrete, and proximate processcan signifi-
cantly influence decision outcomes (Trope & Liberman, 2010).
In this context, it has been shown that varying psychological
distance can influence consumers' awareness of the difficulty of
the decision task (choice feasibility), which in turn is likely to
influence their preference for larger versus smaller assortments
(Goodman & Malkoc, 2012).
The above reasoning suggests that the impact of assortment
size on choice overload is likely to be a function of individuals'
decision goals. Specifically, we expect that choice overload is
likely to be more pronounced in cases when consumers aim to
minimize the effort involved in making a choice from a given
assortment.
Method
The data
The data utilized in the meta-analysis were collected through a
literature review of articles published in refereed psychology and
339A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
marketing journals. We also examined the recently published
review papers in the domain of choice overload (Broniarczyk,
2008; Chernev, 2012; Scheibehenne, Todd, & Greifeneder, 2010)
to ensure comprehensive coverage of the published data. Given
our focus on identifying factors that moderate the impact of
assortment size on choice overload, we sought out studies that (1)
included assortment size as an independent variable and (2)
included a dependent variable that has been linked by prior
research to choice overload (e.g., choice satisfaction, decision
confidence, decision regret, choice deferral, switching likelihood,
assortment choice, and option selection).
Our units of analysis are the individual conditions of the studies
reported in the relevant articlesan approach common for
meta-analyses (Becker, 2001; Rosenthal, 1995; Sánchez-Meca,
Marín-Martínez, & Chacón-Moscoso, 2003). For example, a
study featuringa 2 (moderator: high vs. low) × 2 (assortment size:
high vs. low) design yields two meta-analytic observationsone
for each level of the moderating factor. In the same vein, an article
reporting three studies, each featuring a 2 × 2 design, yields six
observations (3 studies × 2 conditions).
Overall, the meta-analysis includes 99 observations derived
from 53 studies published in 21 articles across 7202
participants. Consistent with our theorizing, the observations
were assigned to one of the four experimental factors: choice
set complexity, decision task difficulty, preference uncertainty,
and decision goal. Specifically, the choice set complexity factor
involved 18 observations across 1336 respondents, the decision
task difficulty factor involved 17 observations across 1409
respondents, the preference uncertainty factor involved 22
observations across 1092 respondents, and the decision goal
factor involved 24 observations across 1778 respondents. In
addition, there were 18 observations across 1587 respondents
that captured only the main effect of assortment size and, thus,
did not involve any moderating factors.
Each observation was also coded to reflect the correspond-
ing dependent variablesatisfaction/confidence, decision re-
gret, choice deferral, switching likelihood, assortment choice,
and option selection. Because we were interested in estimating
the differential impact of choice overload using different
metrics, in studies with multiple dependent variables we
recorded each dependent variable as a separate observation.
All coding directly followed from the information reported by
the authors of the individual studies.
A summary of the experimental data is given in Table 1,which
outlines the underlying observations, studies, and articles and
indicates how each observation corresponds to our conceptual
model, presented in Fig. 1. A detailed overview of the individual
observations and the corresponding papers and an outline of the
rationale for their selection are offered in Appendix A.
The model
We use a theory-based meta-analytic approach (Becker, 2001)
to explore the drivers of choice overload. This approach facilitates
both quantifying and testing the effects of theoretically derived
moderators. The key benefit of this approach is that rather than
focusing on the presence or absence of a main effect of assortment
size on choice overload, it enables us to validate our conceptual
model by evaluating the relative impact of the four conceptually
derived antecedents of choice overload. This is important because
the published studies document choice overload in some
conditions but not in others. In this context, we aim to test the
validity of our conceptual model with respect to its ability to
account for the conceptual moderators that have been advanced
andtestedinpriorresearch.
As informed by our conceptual model, we identify four
factorschoice set complexity, decision task difficulty,
preference uncertainty, and decision goalthat are likely to
influence the impact of assortment size on choice overload. We
further operationalize choice overload by measuring choice
deferral, switching likelihood, option selection, assortment
choice, decision regret, and satisfaction/confidence (given the
low number of studies measuring satisfaction and confidence,
we combined these two factors into a single variable).
To integrate the individual studies into a format suitable for
meta-analysis, we transformed the differences between the small
and large assortments within individual studies into effect size
measures represented by Cohen's dan approach commonly
used in meta-analysis (Cohen, 1988). In this context, a positive
d-value is associated with a negative impact of larger assortments
(choice overload), whereas a negative d-value is associated with a
positive impact of larger assortments. For the decision outcomes
measured on a continuous scale (e.g., regret, satisfaction, and
confidence), Cohen's dis calculated as the difference between the
two means divided by the combined standard deviation
1
(Cohen,
1988). For the decision outcomes measured on a binary scale
(choice deferral, switching likelihood, option selection), Cohen's
dis calculated using the arcsine transformation
2
(Lipsey &
Wilson, 2001; Scheibehenne et al., 2010). Using the log-odds
ratio, which has been shown to perform well for binary outcomes
(Sánchez-Meca et al., 2003) yielded nearly identical results.
To illustrate, consider the data by Iyengar and Lepper (2000;
Study 3), which involves both continuous and binary measures.
Specifically, the data measured on a continuous scale show that
participants in the small assortment condition were more
satisfied than participants in the large assortment condition
(M= 6.28, SD = 0.54 vs. M= 5.46, SD = 0.82), and the data
measured on a binary scale show that participants given the
smaller assortment condition were more likely to make a purchase
compared to those given the larger assortment (16 out of 33,
versus 4 out of 34). Accordingly, we calculate d=1.18 for
satisfaction and d= .84 for the choice likelihood measure,
3
whereby the positive dindicates the presence of choice overload.
To analyze the data, we use a meta-analytic model that
regresses effect sizes (Cohen's deffects) of the explanatory
variables on the observed measures of choice overload. Given
the nested nature of the data (99 observations derived from 53
studies published in 21 separate articles), we use a three-level
1
d¼meansmallmeanlarge
s, where s ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
nsmall1
ðÞ
s2
smallþnlarge 1
ðÞ
s2
large
nsmallþnlarge 2
r.
2
d¼2arcsine ffiffiffiffiffiffiffiffiffiffi
Psmall
p2arcsine ffiffiffiffiffiffiffiffiffiffi
Plarge
p, where P
small
and P
large
are the
proportions of participants who make a selection from each assortment.
3
d¼6:285:46
:696 ¼1:18, where s¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
331ðÞ:542þ341ðÞ:822
33þ342
q¼:696 and d¼2
arcsine ffiffiffi
16
33
q2arcsine ffiffiffi
4
34
q¼1:54:70 ¼:84.
340 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
Table 1
Overview of the studies included in the Meta-Analysis.
Study ID Product
category
Sample
size
Moderators of choice overload
(independent variables)
Indicators of choice
overload
Assortment size Effect size
Author/article Conceptual factor Factor
level
Condition (dependent variables) Small Large Cohen's dS
2
Weight
Iyengar and Lepper (2000),When Choice Is
Demotivating: Can One Desire Too Much of a
Good Thing?
1 1 Jam 249 No moderators –– Choice deferral 6 24 0.82 0.02 0.03
2 2 Essays 193 No moderators –– Choice deferral 6 30 0.30 0.02 0.02
2 3 Essays 193 No moderators –– Satisfaction 6 30 0.44 0.02 0.02
3 4 Chocolate 67 No moderators –– Choice deferral 6 30 0.84 0.06 0.01
3 5 Chocolate 67 No moderators –– Satisfaction 6 30 1.18 0.07 0.01
Chernev (2003a),Product Assortment and
Individual Decision Processes.
1 6 Chocolate 50 Preference uncertainty Low Ideal point available Option selection 8 20 1.34 0.08 0.01
1 7 Chocolate 51 Preference uncertainty High Ideal point not available Option selection 8 20 0.37 0.08 0.01
1 8 Chocolate 50 Preference uncertainty Low Ideal point available Satisfaction 8 20 0.15 0.08 0.01
1 9 Chocolate 51 Preference uncertainty High Ideal point not available Satisfaction 8 20 0.91 0.09 0.01
Chernev (2003b),When More Is Less and Less Is
More: The Role of Ideal Point Availability and
Assortment in Consumer Choice.
1 10 Chocolate 45 Preference uncertainty Low Ideal point available Switching likelihood 4 16 0.36 0.09 0.01
1 11 Chocolate 43 Preference uncertainty High Ideal point not available Switching likelihood 4 16 0.72 0.09 0.01
2 12 Chocolate 34 Preference uncertainty Low Ideal point available Switching likelihood 4 16 0.35 0.12 0.00
2 13 Chocolate 41 Preference uncertainty High Ideal point not available Switching likelihood 4 16 0.57 0.10 0.01
3 14 Assorted products 81 Preference uncertainty Low Ideal point available Switching likelihood 6 24 0.29 0.05 0.01
3 15 Assorted products 86 Preference uncertainty High Ideal point not available Switching likelihood 6 24 0.47 0.05 0.01
4 16 Assorted products 84 Preference uncertainty Low Ideal point available Switching likelihood 6 24 0.13 0.01 0.01
4 17 Assorted products 84 Preference uncertainty High Ideal point not available Switching likelihood 6 24 0.16 0.01 0.01
4 18 Assorted products 84 Preference uncertainty Low Ideal point available Satisfaction 6 24 0.30 0.01 0.01
4 19 Assorted products 84 Preference uncertainty High Ideal point not available Satisfaction 6 24 0.17 0.01 0.01
Chernev (2005),Feature Complementarity and
Assortment in Choice.
2 20 MP3 players, toothpaste 88 Choice set complexity Low Noncomplementary options Choice deferral 2 5 0.33 0.02 0.01
2 21 MP3 players, toothpaste 88 Choice set complexity High Complementary options Choice deferral 2 5 0.31 0.02 0.01
Gourville and Soman (2005),Overchoice and
Assortment Type: When and Why Variety
Backfires.
1 22 Microwave ovens 120 Choice set complexity Low Alignable options Option selection 23560.43 0.03 0.02
1 23 Microwave ovens 120 Choice set complexity High Nonalignable options Option selection 2356 0.37 0.03 0.02
2 24 Camera 102 Decision task difficulty Low Small number of attributes Option selection 2 3 0.59 0.04 0.01
2 25 Camera 102 Decision task difficulty High Large number of attributes Option selection 2 3 0.48 0.04 0.01
3 26 Golf balls 240 Decision task difficulty Low No justification required Option selection 2 3 0.42 0.02 0.03
3 27 Golf balls 240 Decision task difficulty High Justification required Option selection 2 3 0.27 0.02 0.03
Oppewal and Koelemeijer (2005),More Choice
Is Better: Effects of Assortment Size and
Composition on Assortment Evaluation.
1 28 Flowers 741 Decision (effort-minimizing)goal Low Browsing Satisfaction 561112 0.82 0.00 0.09
Chernev (2006),Decision Focus and Consumer
Choice among Assortments.
1 29 Vending machines 57 Decision (effort-minimizing) goal Low Assortment selection Assortment choice 6 36 2.61 0.04 0.01
1 30 Vending machines 54 Decision (effort-minimizing) goal High Product selection Assortment choice 6 36 0.60 0.04 0.01
2 31 Chocolate 46 Decision (effort-minimizing) goal Low Assortment selection Assortment choice 24 88 2.55 0.04 0.01
2 32 Chocolate 92 Decision (effort-minimizing) goal High Product selection Assortment choice 24 88 1.48 0.02 0.01
3 33 Pens 52 Decision (effort-minimizing) goal Low Assortment selection Assortment choice 12 60 2.47 0.04 0.01
3 34 Pens 36 Decision (effort-minimizing) goal High Product selection Assortment choice 12 60 1.31 0.06 0.00
3 35 Pens 52 Decision (effort-minimizing) goal Low Assortment selection Satisfaction 12 60 1.52 0.05 0.01
3 36 Pens 36 Decision (effort-minimizing) goal High Product selection Satisfaction 12 60 0.82 0.06 0.00
4 37 Hotel resorts 41 Decision (effort-minimizing) goal Low Assortment selection Assortment choice 6 24 2.51 0.05 0.01
4 38 Hotel resorts 44 Decision (effort-minimizing) goal High Product selection Assortment choice 6 24 0.65 0.05 0.01
4 39 Hotel resorts 41 Decision (effort-minimizing) goal Low Assortment selection Satisfaction 6 24 1.54 0.06 0.01
4 40 Hotel resorts 44 Decision (effort-minimizing) goal High Product selection Satisfaction 6 24 0.66 0.05 0.01
Lin and Wu (2006),The Effect of Variety on
Consumer Preferences: The Role of Need for
Cognition and Recommended Alternatives.
1 41 Chocolate 43 Decision (effort-minimizing) goal Low High need for cognition Regret 6 16 1.69 0.13 0.01
Shah and Wolford (2007),Buying Behavior as a
Function of Parametric Variation of Number of
Choices.
1 42 Pens 80 No moderators –– Choice deferral 612 1420 0.77 0.05 0.01
3 43 Coffee 20 Preference uncertainty Low Ideal point available Satisfaction 5 50 0.30 0.20 0.00
(continued on next page)
341A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
Table 1 (continued)
Study ID Product
category
Sample
size
Moderators of choice overload
(independent variables)
Indicators of choice
overload
Assortment size Effect size
Author/article Conceptual factor Factor
level
Condition (dependent variables) Small Large Cohen's dS
2
Weight
Mogilner et al. (2008),TheMereCategorization
Effect: How the Presence of Categories
Increases Choosers' Perceptions of
Assortment Variety and Outcome
Satisfaction.
3 44 Coffee 39 Preference uncertainty High Ideal point not available Satisfaction 5 50 1.21 0.12 0.00
Chernev and Hamilton (2009),Assortment Size
and Option Attractiveness in Consumer Choice
Among Retailers.
1 45 Sandwich shops 30 Choice set complexity Low Less attractive options Assortment choice 9 38 1.65 0.07 0.00
1 46 Sandwich shops 30 Choice set complexity High More attractive options Assortment choice 9 38 0.40 0.07 0.00
2 47 Assorted products 126 Choice set complexity Low Less attractive options Assortment choice 8 24 0.78 0.02 0.02
2 48 Assorted products 118 Choice set complexity High More attractive options Assortment choice 8 24 0.44 0.02 0.01
3 49 Jam 69 Choice set complexity Low Less attractive options Assortment choice 9 54 1.58 0.03 0.01
3 50 Jam 72 Choice set complexity High More attractive options Assortment choice 9 54 0.00 0.03 0.01
4 51 Assorted products 19 Choice set complexity Low Less attractive options Assortment choice 9 54 1.58 0.03 0.00
4 52 Assorted products 23 Choice set complexity High More attractive options Assortment choice 9 54 0.02 0.02 0.00
5 53 Chocolate stores 47 Choice set complexity Low Less attractive options Assortment choice 9 54 1.02 0.04 0.01
5 54 Chocolate stores 47 Choice set complexity High More attractive options Assortment choice 9 54 0.21 0.04 0.01
Fasolo et al. (2009),The Effect of Choice
Complexity on the Time Spent Choosing.
1 55 Mobile phones 64 No moderators –– Satisfaction 6 24 0.06 0.06 0.01
2 56 Mobile phones 60 No moderators –– Satisfaction 6 24 0.52 0.07 0.01
2 57 Mobile phones 60 No moderators –– Satisfaction 6 24 0.11 0.07 0.01
Haynes (2009),Testing the Boundaries of the
Choice Overload Phenomenon: The Effect of
Number of Options and Time Pressure on
Decision Difficulty and Satisfaction.
1 58 Various prizes 36 Decision task difficulty Low Low time pressure Satisfaction 3 10 0.28 0.11 0.00
1 59 Various prizes 33 Decision task difficulty High High time pressure Satisfaction 3 10 0.64 0.13 0.00
1 60 Various prizes 36 Decision task difficulty Low Low time pressure Regret 3 10 0.17 0.11 0.00
1 61 Various prizes 33 Decision task difficulty High High time pressure Regret 3 10 0.04 0.12 0.00
Scheibehenne et al. (2009),What Moderates the
Too-Much-Choice Effect.
1 62 Restaurant coupons 80 No moderators –– Choice deferral 5 30 0.11 0.05 0.01
2a 63 Charity 60 Preference uncertainty Low High expertise Choice deferral 2 30 0.32 0.07 0.01
2a 64 Charity 57 Preference uncertainty High Low expertise Choice deferral 5 40 0.13 0.07 0.01
2b 65 Charity 75 No moderators –– Choice deferral 5 79 0.18 0.05 0.01
2c 66 Charity 80 Decision task difficulty High Justification required Choice deferral 5 80 0.37 0.05 0.01
3a 67 Music 80 Preference uncertainty Low High expertise Satisfaction 6 30 0.25 0.03 0.01
3b 68 Music 87 Preference uncertainty Low High expertise Satisfaction 6 30 0.05 0.02 0.01
Sela et al. (2009),Variety, Vice, and Virtue: How
Assortment Size Influences Option Choice.
1a 69 Ice cream 121 No moderators –– Option selection 2 10 0.38 0.03 0.02
1b 70 Food 75 No moderators –– Option selection 4 12 0.43 0.05 0.01
2 71 Printers, MP3 players 50 No moderators –– Option selection 4 12 0.89 0.08 0.01
3 72 Printers, MP3 players 156 No moderators –– Option selection 4 12 0.35 0.03 0.02
342 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
4 73 Laptops 86 Choice set complexity Low Dominant (justifiable)
option available
Option selection 4 12 0.47 0.05 0.01
4 74 Laptops 85 Choice set complexity High Dominant (justifiable)
option not available
Option selection 4 12 0.45 0.05 0.01
5 75 Printers, MP3 players 84 Choice set complexity Low Dominant (justifiable)
option available
Option selection 4 12 0.46 0.05 0.01
5 76 Printers, MP3 players 84 Choice set complexity High Dominant (justifiable)
option not available
Option selection 4 12 0.48 0.05 0.01
Diehl and Poynor (2010),Great Expectations?!
Assortment Size, Expectations and Satisfaction.
2 77 Camcorder 165 No moderators –– Satisfaction 8 32 0.33 0.02 0.02
3 78 Computer wallpaper 65 No moderators –– Satisfaction 60 300 0.54 0.06 0.01
Greifeneder et al. (2010),Less May Be More When
Choosing Is Difficult: Choice Complexity and Too
Much Choice.
1 79 Pens 40 Decision task difficulty Low Low number of attributes Satisfaction 6 30 0.12 0.10 0.01
1 80 Pens 40 Decision task difficulty High High number of attributes Satisfaction 6 30 0.81 0.11 0.01
2 81 MP3 players 52 Decision task difficulty Low Low number of attributes Satisfaction 6 30 0.28 0.08 0.01
2 82 MP3 players 52 Decision task difficulty High High number of
attributes
Satisfaction 6 30 0.54 0.08 0.01
Inbar et al. (2011),Decision Speed and Choice
Regret: When Haste Feels Like Waste.
1 83 DVDs 27 No moderators –– Regret 5 30 1.22 0.18 0.00
2 84 DVDs 78 Decision task difficulty Low Low time pressure Regret 15 45 0.31 0.05 0.01
2 85 DVDs 78 Decision task difficulty High High time pressure Regret 15 45 0.68 0.05 0.01
Goodman and Malkoc (2012),Choosing Here and
Now Versus There and Later: The Moderating Role
of Psychological Distance on Assortment Size
Preferences.
1A 86 Restaurant coupons 63 Decision (effort-minimizing) goal Low Low level of construal Assortment choice 7 14 0.55 0.03 0.01
1A 87 Restaurant coupons 67 Decision (effort-minimizing) goal High High level of construal Assortment choice 7 14 0.15 0.03 0.01
1B 88 Ice cream 78 Decision (effort-minimizing) goal Low Low level of construal Assortment choice 6 18 1.53 0.03 0.01
1B 89 Ice cream 82 Decision (effort-minimizing) goal High High level of construal Assortment choice 6 18 0.80 0.02 0.01
2 90 Vacation packages 51 Decision (effort-minimizing) goal Low Low level of construal Assortment choice 6 18 1.14 0.04 0.01
2 91 Vacation packages 47 Decision (effort-minimizing) goal High High level of construal Assortment choice 6 18 0.30 0.04 0.01
3 92 Blenders 42 Decision (effort-minimizing) goal Low Low level of construal Assortment choice 6 18 0.99 0.05 0.01
3 93 Blenders 45 Decision (effort-minimizing) goal High High level of construal Assortment choice 6 18 0.04 0.04 0.01
4 94 Blenders 56 Decision (effort-minimizing) goal Low Low level of construal Assortment choice 4 24 1.49 0.04 0.01
4 95 Blenders 41 Decision (effort-minimizing) goal High High level of construal Assortment choice 4 24 0.54 0.05 0.01
Morrin et al. (2012),Plan Format and Participation
in 401(k) Plans: The Moderating Role of Investor
Knowledge.
1 96 Mutual funds 77 Preference uncertainty Low High expertise Choice deferral 9 21 0.56 0.05 0.01
1 97 Mutual funds 73 Preference uncertainty High Low expertise Choice deferral 9 21 0.33 0.05 0.01
Townsend and Kahn (2014),The Visual Preference
Heuristic: The Influence of Visual versus Verbal
Depiction on Assortment Processing, Perceived
Variety, and Choice Overload.
5 98 Crackers 129 Decision task difficulty Low Verbal presentation format Choice deferral 8 27 0.32 0.03 0.02
5 99 Crackers 107 Decision task difficulty High Visual presentation format Choice deferral 8 27 0.37 0.04 0.01
Note.Study indicates the number of the study in the corresponding paper. ID is the identification number assigned to each observation for the purposes of the meta-analysis. Conceptual factors are the four factors
decision task difficulty, choice set complexity, preference uncertainty, and decision (effort-minimizing) goaldepicted in Fig. 1,operationalized by a variety of study-specific variables. Assortment size refers to the
number of options used to represent small and large assortments in each study. Cohen's d is the measure of the effect size, S
2
is the corresponding variance, and weight is the relative weight of each observation as a
function of the sample size. Studies using multiple dependent variables are reported in the table multiple times; this replication is controlled for in the statistical analysis.
343A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
meta-regression model, in which the first two levels capture the
specific effect sizes across studies and the third level captures the
underlying articles. This approach enables us to account for the
fact that some of the individual observations are part of the same
experiments, a number of which belong to the same articles.
Controlling for interdependencies across experiments is impor-
tant because effect sizes reported in the same article are likely to
be more similar compared to effect sizes in experiments reported
in articles following different research paradigms. Thus, the
three-level meta-regression approach allows us to reduce the risk
of spurious findings that may result from treating effect sizes that
are reported in the same article as independent sources of
information. The resulting model is as follows:
ES
ia
=β
0NM
+β
1
SetComplexity + β
2
TaskDifficulty + β
3
PreferenceUncertainty + β
4
DecisionGoal + β
5
ChoiceDeferral +
β
6
SwitchingLikelihood + β
7
OptionSelection + β
8
Assortment
Choice + β
8
Satisfaction + β
10
DecisionRegret +υ
0a
+η
ia
Here, the intercept β
0NM
represents the overall effect of
assortment size in the absence of moderators. The next four factors
(β
1
β
4
) capture the four hypothesized moderators of choice
overload (choice set complexity, decision task difficulty, prefer-
ence uncertainty, and decision goal). The following six factors
(β
5
β
10
) reflect the six outcome measures used in different
experiments (satisfaction/confidence, choice deferral, switching
likelihood, assortment choice, regret, and option selection).
The model further includes two random sources of variance
that capture effects not explained by the four moderators. Thus,
the between-study variability η
ia
estimates random deviations
from the intercept and moderator effects, whereby its size reflects
the degree to which the intercept and moderators can capture any
observed differences among the effect sizes. We also estimate the
variance of υ
0a
, which capturesthe fact that that some experiments
stem from the same article and reflects the degree to which effect
sizes in these experiments differ from one another.
Results
We present our findings starting with describing the fit of
our model with the existing empirical data and its ability to
account for the variance across existing studies. We then
proceed to test the effects of the specific drivers of choice
overload identified by our conceptual model. Finally, we test
for the effects of additional factors that might influence choice
overload, as well as for the presence of publication bias in the
data.
Model fit
The data show that the moderators and outcome measures
capture 68% of the residual variances estimated at the study and
article levels (Huedo-Medina, Sanchez-Meca, Marin-Martinez,
& Botella, 2006) compared to the intercept-only model. The
fit difference between the model including the conceptual
moderators and outcome measures outlined in the previous
section and the intercept-only model is highly significant
(χ
2
(10) = 118.1, p b.001), indicating that our conceptual
framework receives support by accounting for a substantial
variation in the effect sizes. We further estimate the residual
variances as 0.05 at the study level and as 0.13 at the article
level. Overall, these data suggest that the observed heteroge-
neity in the reported effect sizes of the studies are reasonably well
accounted for by the moderators as derived by our theoretical
framework.
To further test the validity of our conceptual model, we
investigated whether the effects of the moderating factors
interacted with any of the outcome measures. The interaction
test showed that the effects of each moderator across the
outcome measures did not vary significantly. The correspond-
ing test statistics for the interaction effect of choice set
complexity, decision task difficulty, preference uncertainty,
and decision goal are χ
2
(2) = 3.6, p = .16, χ
2
(3) = .6, p=
.90, χ
2
(3) = .5, p= .92, and χ
2
(2) = 4.3, p= .12, respec-
tively. These results for each of the four factors are consistent
with our conceptual model presented in Fig. 1, showing that
different moderators yield similar effects across different
outcome measures. Another test of the validity of our grouping
of empirical factors into conceptual factors is the degree to
which each of the conceptual factors captures the variability
among the effect sizes across its different operationalizations.
In this context, the fact that different operationalizations lead
to similar effect sizes would be reflected in a nonsignificant
interaction between the individual operationalizations of each
conceptual factor and the impact of assortment size on choice
overload. The test statistics for the grouping hypothesis of the
four moderators are: χ
2
(3) = 3.7, p= .29 for choice set
complexity; χ
2
(3) = 4.4, p= .23 for decision task difficulty;
χ
2
(1) = 1.9, p= .17 for preference uncertainty; and χ
2
(2) =
6.8, p= .03 for decision goal. The absence of significant
interactions for three of the four factors is consistent with the
notion that these different operationalizations lead to similar effect
sizes. For example, in the case of choice set complexity, this
means that its different operationalizationsthe availability of a
dominant option, option attractiveness, option alignability, and
option complementarityhave similar explanatory power. For
the decision goal factor, we find that decision focus has greater
impact on the overload effect compared to the other two factors
(decision intent and level of construal). We also tested for
multicollinearity by examining the correlation matrix of the
independent variables and found no evidence for it; all of the
correlations were small and most were nonsignificant.
Effects of the specific moderators of choice overload
To examine the effect of the four factors identified as the
potential drivers of choice overload, we first examine a model
that does not include these moderators and then compare this
Fig. 2.Forest plot of the overall effect of assortment size on choice overload in a model without conceptual moderators.Note.The Forest plot depicts effect sizes and their respective condence intervals for
individual observations. The individual observations are arranged by effect size, starting with the strongest positive effect at the top and ending with the strongest negative effect on the bottom. The data show
that the effect sizes vary from 4.9 to 1.6, suggesting signicant variability in the experimental results. The large negative effect sizes in the four observations at the bottom of the chart are consistent with
individuals' natural preference for large assortments in the context of the assortment choice task.
344 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
345A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
model with one that includes the four moderators implied by
our conceptual modelan approach consistent with prior
meta-analytic research (Scheibehenne et al., 2010). Specifical-
ly, the simple model involves the intercept β
0
and the two
random sources of variance described earlierυ
0a
and η
ia
as
shown below.
ESia ¼β0þυ0a þηia
The overall effect of assortment size for each observation
can be visually represented using a Forest plot that depicts
effect sizes (Cohen's d) and their respective confidence
intervals for each specific observation (Fig. 2). The data show
that in the absence of the conceptual moderators, the mean
effect of assortment size on choice overload is nonsignificant
(t(20) = .10; p= .48)a finding consistent with the findings
reported by prior research (Scheibehenne et al., 2010). At the
same time, there was a significant unexplained variability of the
effect size across individual articles and studies not accounted
for by the simple model, suggesting that the simple model is not
representative of the data (χ
2
(78) = 665.5, pb.001 at the
study level and χ
2
(20) = 130.3, pb.001 at the article level).
Given the significant amount of unexplained variance by the
simple model, we then proceed with testing a more compre-
hensive model that includes the four conceptual moderators
stemming from our theorizing. The results of the statistical
analysis are summarized in Table 2, which reports that all four
factorschoice set complexity, decision task difficulty,
preference uncertainty, and decision goalidentified by our
conceptual model as potential antecedents of choice overload
are statistically significant (p b.001) and have relatively strong
effects on choice overload. This finding lends support to our
conceptual model depicted in Fig. 1, showing that each of these
four factors has a significant impact on choice overload,
whereby higher levels of decision task difficulty, greater choice
set complexity, higher preference uncertainty, and a more
prominent effort-minimizing goal lead to a greater choice
overload.
With respect to the dependent variables used to capture
choice overload, we find that the effect estimates of four of the
six outcome measures reflecting overload effects above and
beyond the moderator effects are nonsignificant. Thus, because
satisfaction/confidence, regret, choice deferral, and switching
likelihood do not produce an effect that is not already captured
by the four moderating variables, we conclude that they are
equally powerful in capturing the impact of assortment size on
choice overload. The two outcome measures producing a
significant effect are assortment choice and option selection,
and we conjecture that their ability to capture unique aspects of
the impact of assortment size on choice overload is a function
of the specifics of the decision task associated with these
outcome measures. Specifically, the assortment-choice task
typically involves a very strong preference for large assort-
ments, which is likely to skew individuals' choices in favor of
the larger assortment (Chernev, 2006). The option-selection
task, on the other hand, employed assortments that were on
average significantly smaller than those used with the other
dependent variables (the mean assortment sizes in the
option-choice task were 4 and 12, significantly smaller than
the corresponding assortment sizes of 8 and 34 in the other
decision tasks). The fact that the remaining four dependent
variablessatisfaction/confidence, regret, choice deferral, and
switching likelihooddo not produce a significant effect
above and beyond the four conceptual moderators, is important
because it suggests that these measures could be used
interchangeably to capture the impact of assortment size on
choice overload.
The effect sizes and the respective confidence intervals for
the specific observations can be visually represented using
Forest plots, as shown in Fig. 3. Given our focus on identifying
the differential impact of the four moderating factors, we
organized the Forest plot by effect size within each moderator.
The data represented by the Forest plots show that varying each
of the four factors influences the direction of choice overload,
such that the effect is positive in conditions where these factors
are present and negative in conditions where these factors are
less pronounced or absent.
The mean effect of assortment size on choice overload
Following the examination of the individual effects of the
four moderators included in our conceptual model, we examine
the mean overload effect in studies that do not include
moderators (identified in Table 1). The data show that for
studies that do not include moderators, the intercept term
Table 2
A summary of the meta-analysis results.
Effect Estimate SE T P
Intercept (no moderators) .41 .14 3.0 .01
Moderators of choice overload
Choice set complexity .55 .07 7.9 b.001
Decision task difficulty .37 .08 4.7 b.001
Preference uncertainty .32 .07 4.5 b.001
Decision goal .56 .06 8.8 b.001
Measures of choice overload
Satisfaction .12 .12 1.1 .29
Regret .17 .25 .69 .49
Choice deferral .08 .15 .50 .62
Switching likelihood .23 .22 1.1 .28
Assortment choice .72 .16 4.4 .001
Option selection .45 .21 2.10 .04
Note.Individual cells represent the estimates of the meta-regression model.
The data further show that for the model that includes the conceptual
moderators the intercept term reflecting the main effect of assortment size is
significant (t(67) = 3.0; p= .004), indicating that studies that do not include
moderators tend to exhibit a systematic overload effect. All four moderators
identified by our conceptual model (Fig. 1) are significant and the
corresponding coefficients indicate comparable effects, with the decision goal
and choice set complexity having the strongest effects and preference
uncertainty the weakest effect on choice overload. Thus, higher levels of
decision task difficulty, greater choice set complexity, higher preference
uncertainty, and a more prominent, effort-minimizing goal tend to produce
greater choice overload. The last six rows indicate the differential impact of the
individual response variables. The data show that four of the six factors
satisfaction/confidence, regret, choice deferral, and switching likelihooddo
not have a significant impact, suggesting that they capture the impact of
assortment size on choice overload equally well.
346 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
reflecting the main effect of assortment size is significant
(t(67) = 3.0; p= .004), indicating that studies that do not
include moderators tend to exhibit a systematic overload effect
(Fig. 3E).
In addition to examining the mean overload effect in studies
without moderators, we also examine the mean effect of choice
overload across studies both with and without moderators. In
this analysis, we exclude studies in which the decision goal
involved choice among assortments. Indeed, whereas studies
examining assortment choice are relevant for capturing the
effect of the conceptual moderators (e.g., decision goal) on the
impact of assortment size on choice overload, decisions that
involve choices among assortments (rather than choices of an
option from a given assortment) tend to display an inherent
strong preference for larger assortments (Chernev, 2006).
Accordingly, to examine the mean effect of choice overload
we excluded observations that involved assortment choice
(2940, 4554, and 8695). The data show that across all studies
Fig. 3a. Forest plot showing how choice set complexity influences the impact of assortment size on choice overload. Note.The Forest plots 3A3E depict effect
sizes and their respective confidence intervals for individual observations for each of the four moderating factors identified in Fig. 1, as well as for observations from
studies without moderators. Here, the individual observations are grouped into one of the four moderating factors, with the individual observations within each factor,
as well as the observations without moderators, arranged by effect size. Varying each factor influenced the direction of the effect of assortment size, such that choice
overload is present in conditions where these factors are present/more pronounced and is in the opposite direction when these factors are less pronounced/absent.
Within each of the plots depicting moderators (3A3D), the level of each factor (high vs. low) is correlated with the direction of the effect, such that high levels are
associated with choice overload whereas lower levels are associated with the opposite (more is better) effect. The mean values of the effect size for each of the two
levels of a particular moderator are indicated with a diamond, whereby the size of the diamond reflects the uncertainty associated with estimating these effects. The
overall effect (across the two conditions of each moderator) of assortment size in these plots is given by the difference between the dashed vertical lines. Studies
without moderators (plot 3E) show the tendency of larger assortments to produce choice overload. Studies examining choice among assortments (Chernev &
Hamilton, 2009 in plot 3A and Chernev, 2006 in plot 3D) display a less pronounced ability to engender choice overloada finding that likely stems from the lower
decision costs associated with choice among assortments.
347A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
that examine choices from an already given assortment,
assortment size has a significant impact on choice overload,
such that in comparison to smaller assortments larger assortments
are more likely to produce choice overload (t(39) = .17,
pb.001). We also tested the mean effect of assortment size
on choice overload by isolating the effects of studies with and
without moderators (i.e., using a model with separate
intercepts for the studies with and without moderators rather
than using a single intercept). This test yielded a
non-significant effect for studies with moderators (t(39) =
.04, p = .41) and a significant effect for studies without
moderators (t(39) = .41, p b.001)a finding suggesting that
aggregating the observations across different levels of a given
moderating factor is likely to attenuate (or eliminate) the
choice overload effect (Chernev, Böckenholt, & Goodman,
2010).
The finding that assortment size can have a significant
main effect on choice overload is counter to the data
reported by prior meta-analytic research, which finds this
effect to be nonsignificant (Scheibehenne et al., 2010). This
discrepancy suggests that the mean effect of assortment size
on choice overload is likely to be contingent on the subset
of studies included in the meta-analysis and/or the
conceptual model tested (Chernev et al., 2010). We address
this discrepancy in more detail in the discussion section of
this research.
The effect of assortment size across individual experiments
Most studies investigating the impact of assortment size on
choice overload have treated assortment size as a binary
variable, differentiating between small and large assortments.
However, the operationalization of small versus large assort-
ments has varied across these studies, enabling us to examine
whether choice overload is indeed more pronounced in studies
that utilize relatively larger assortments. Analysis of the choice
set sizes of small and large assortments across the individual
studies shows that the most common comparison involved 6
options, representing a small assortment, and 24 options,
representing a large assortment (6 and 24 are the median
values; the corresponding means are M
Small
= 5.4, SD = 18.6
and M
Large
= 27.8, SD = 3.7). This assortment size selection
Fig. 3b. Forest plot showing how decision task difficulty influences the impact of assortment size on choice overload. (continued).
348 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
*
*The values reported in this section as t-statistics are b coefficients; the complete statistics are b=.17, t(39)= 4.5; b=.04, t(39)= .9; b=.41, t(39)=5.3
was likely influenced by the methodology used for the
choice set sizes in the pioneering article reporting empirical
evidence of choice overload (Iyengar & Lepper, 2000). The
study-specific assortment sizes are given in Table 1 and the
distribution of the sizes of small and large assortments across
individual experiments is shown in Fig. 4.
To examine the effect of assortment size across studies, we
added assortment size to our model as a variable. For the
purposes of this analysis we excluded one observation (78) as
an outlier because it used choice sets that were disproportion-
ately larger than those used in the other studies. The data show
that the effect of the larger assortment size across studies was
not significant (b=.005, t(37) = 1.5, p=.13).We also
examined whether the effect of assortment size was likely to
decline as the number of options in the choice set increased
(Chernev & Hamilton, 2009; Ratner et al., 1999)bytestingfor
a quadratic trend. However, the data showed no evidence that
this effect diminishes as the size of the larger assortment
increases (b= .002, t(36) = 1.4, p=.17).
The lack of a significant monotonically increasing relation-
ship between assortment size and choice overload is consistent
with the findings reported by prior research (Scheibehenne et
al., 2010). This finding is further consistent with the argument
made by Chernev et al. (2010), who reason that although
conceptually such a relationship should exist, the failure of a
meta-analysis to document it is not surprising, since there are a
number of intervening factors (e.g., decision-maker's expertise,
the composition and the organization of the assortment, and the
nature of the decision task) that ultimately determine whether
increasing assortment size will result in choice overload.
Because the experiments included in the meta-analysis vary on
a number of dimensions (e.g., option complexity, organization of
the choice set, and product familiarity) that likely contribute to
choice overload, the absence of a monotonic (linear or curvilinear)
effect of assortment size is not inconsistent with the choice
overload hypothesis. This line of reasoning is consistent with the
distribution of small and large assortment sizes across individual
experiments illustrated in Fig. 4. Thus, some studies demonstrate
choice overload in choice sets with relatively few options
(Chernev, 2005; Gourville & Soman, 2005; Haynes, 2009)
whereas other studies show that overload is unlikely to be
prominent even in relatively large choice sets (Chernev, 2006;
Fig. 3c. Forest plot showing how preference uncertainty influences the impact of assortment size on choice overload. (continued).
349A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
Diehl & Poynor, 2010). The overlap between the two distribu-
tions suggests that some assortment sizes have been used across
studies as both large and small assortmentsa finding consistent
with the existence of significant moderators that influence the
impact of assortment size on choice overload.
Publication bias
An important consideration when conducting meta-analytical
studies is to control for the tendency of authors and journals to
overemphasize significant findings and underreport nonsignificant
or inconclusive findings. This publication bias is likely to lead to
asymmetric relationships between the effect sizes and their
standard errors in the absence of moderators (Sterne & Egger,
2001). Such asymmetric relationships can be diagnosed with a
funnel plot, which offers a visual tool for identifying publication
bias in studies (Rothstein, Sutton, & Borenstein, 2005; Sterne,
Becker, and Egger, 2005).
In the absence of moderators, the funnel plot typically displays
effect sizes and their standard errors. However, when the tested
model includes moderators, effect sizes are heterogeneous. Thus,
it is more informative to examine the relationship between
residual effect sizes and their standard errors after the effect of the
moderators has been accounted for rather than between the effect
sizes and their standard errors as they appear in the model
that includes moderators. In addition to being directly compara-
ble, the residual effect sizes provide information about the
meta-regression's ability to account for the observed effect-size
differences (Sutton et al., 2011).
To test for the existence of publication bias in the data, we
examined the dispersion of the residual effects as a function of
the size of the studies' standard errors. In general, the absence
of publication bias is indicated by the fact that residual
estimates with larger standard errors (located at the bottom of
the funnel plot) are more widely dispersed than estimates with
smaller standard errors (located at the top). These dispersions
are expected to be symmetric and scattered around both sides
Fig. 3d. Forest plot showing how decision goal influences the impact of assortment size on choice overload. (continued).
350 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
of zero, with the size of the scatter increasing with less
precision in the estimates. Deviations from a symmetric
scatter of the residual effects and the studies' standard errors
suggest publication bias because one would expect the residual
effects to be randomly distributed in the absence of publication
bias.
The funnel plot of the studies included in the meta-analysis
is presented in Fig. 5. The pattern of dispersion of the residual
effects in the considered studies shows little evidence for
publication bias since the funnel plot appears to be symmetric
and a regression test for funnel plot asymmetry yielded
nonsignificant results. Specifically, we used a weighted
regression model with a multiplicative dispersion term and the
studies' standard errors as predictor. The results show that the
residual effect size estimates cannot be predicted by the
standard errors (z = .34, p= .74). This finding is consistent
with the visual inspection of the funnel plot, suggesting a lack
of systematic relationship between the standard errors of the
effect sizes and their residual estimates. Thus, one can
reasonably expect that the results of the meta-analysis are
unlikely to be systematically influenced by publication bias.
Discussion
Key findings
Despite the plethora of prior studies examining whether and
when large assortments are likely to lead to choice overload, there
have been few attempts to develop an integrative, overarching
model that characterizes the impact of assortment size on choice
overload. In this context, our research contributes to the literature
by identifying the conceptual drivers of the impact of assortment
size on choice overload. To the best of our knowledge, this is the
first attempt to identify the key conceptual drivers of choice
overload, empirically test their validity, and quantify their relative
effects.
Specifically, in this research we identify four key factors that
can reliably predict whether, when, and how assortment size is
Fig. 3e. Forest plot showing the main effect (no moderators) of assortment size on choice overload. (continued).
351A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
likely to influence choice overload: (1) the difficulty of the
decision task, which reflects the structural properties of the
decision task operationalized in terms of time constraints,
decision accountability, number of attributes describing each
option, and the complexity of the presentation format; (2) the
complexity of the choice set, which reflects the value-based
relationships among the choice alternatives, including the
presence of a dominant option, as well as the overall
attractiveness, alignability, and complementarity of the choice
options; (3) consumers' preference uncertainty, which reflects
the degree to which consumers can evaluate the benefits of the
choice options and have an articulated ideal point; and (4)
consumers' decision goal, which reflects the degree to which
individuals aim to minimize the cognitive effort involved in
making a choice among the options contained in the available
assortments. More important, we show that each of these four
factors has a directionally consistent and significant impact on
choice overload, such that higher levels of decision task
difficulty, greater choice set complexity, higher preference
uncertainty, and a more prominent, effort-minimizing goal
facilitate choice overload.
In addition to identifying the key factors that moderate the
impact of assortment size on choice overload, we identify
several common outcomes of choice overload used as
dependent measures in prior researchsatisfaction/confidence,
regret, choice deferral, switching likelihood, assortment choice,
and option selectionand examine the ability of these
measures to capture unique aspects of the impact of assortment
size on choice overload. In this context, we find that four of the
six dependent measuressatisfaction/confidence, choice de-
ferral, switching likelihood, and regretare not significantly
different from one another, suggesting that these four measures
capture the impact of assortment size in a similar way and
hence can be used interchangeably. To the best of our
knowledge, this is the first systematic attempt to compare the
degree to which different measures are able to capture the
impact of assortment size on choice overload. Documenting the
convergence of different operationalizations not only validates
choice overload as a construct but also defines a set of
equivalent operationalizations that could be used interchange-
ably in future research.
Our analysis further documents the presence of a significant
main effect of assortment size on choice overload across studies
that test the main effect of choice overload without explicitly
controlling for moderating effects. Although not central to our
analysis, this finding is notable because it is counter to the
findings reported by prior research advocating the absence of
such an effect (discussed in more detail in the following
section). This discrepancy lends support to the notion that the
main effect of assortment size is vulnerable to a variety of
context effects and therefore is not a reliable measure of choice
overload (Chernev et al., 2010).
Prior meta-analytic research
A prior meta-analytic review found little evidence of choice
overload, concluding that the mean effect of assortment size on
choice overload is nonsignificant (Scheibehenne et al., 2010).
This analysis further questioned the existence of factors that can
systematically lead to choice overload, arguing that no sufficient
conditions could be identified that would lead to a reliable
occurrence of choice overload. The discrepancy between the
findings reported by prior research and the findings of our
meta-analysis raises the question of identifying the key factors
contributing to such disparate findings. We believe that this
discrepancy can be attributed to two key factors: differences in the
underlying data and differences in the model used to analyze the
data. We address these two factors in more detail below.
0
.1
.2
.3
.4
.5
-1 -.5 0 .5 1
Residual Value
Standard Error
Fig. 5. Funnel plot of publication bias. Note.The funnel plot with pseudo
95% confidence limits is approximately symmetrical, indicating that the effect
sizes do not appear to exhibit a systematic pattern. The slightly slanted solid line
represents the regression test for funnel-plot asymmetry proposed by Egger et
al. (1997). The symmetric distribution of the effect sizes suggests the absence of
a significant publication bias in the analyzed studies.
2-3
Assortment size
Frequency
35
5
10
15
20
25
30
Small assortment Large assortment
4-5 6 7-10 11-15 2416-23 25-30 31-50 51-70 70-300
Fig. 4. Distribution of the sizes of small and large assortments across individual
experiments. Note.Black bars indicate assortment sizes representing smaller
assortments and white bars indicate assortment sizes representing larger
assortments. The overlap between the two distributions suggests that some
assortment sizes have been used as both large and small assortments (in
different experiments)a finding consistent with the existence of significant
moderators that influence the impact of assortment size on choice overload.
352 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
In their meta-analysis, Scheibehenne et al. (2010) report 63
observations from 50 published and unpublished experiments
(N= 5036) conducted in the fields of psychology and marketing.
The data used in our meta-analysis are different in several respects.
First, our dataset is significantly larger, including 99 observations
from 53 published studies (N= 7202). Second, to ensure
consistent standards of experimentation, we focused only on
peer-reviewed papers published in academic journals. Overall, our
dataset included 78% of the Scheibehenne et al. (2010)
observations reported in published articles.
The model used by Scheibehenne et al. (2010) differed from
our model in several respects. First, these varied in the selection
of potential moderators of the impact of assortment choice: Our
model focused on four key conceptually derived factors,
whereas the model used by Scheibehenne et al. (2010) included
a mix of conceptual and descriptive moderators (i.e., whether
the study controlled for respondents' expertise or prior
preferences, whether the study was published or unpublished,
publication year, whether the study included a hypothetical or a
real choice, whether the study used satisfaction as a dependent
variable, whether the study used consumption quantity as a
dependent variable, the size of the large assortment, and
whether the study was conducted in or outside of the United
States). Second, our model distinguished among a variety of
different outcome measures used by the individual studies,
which enabled testing for their moderator-specific effects.
Finally, we used a three-stage hierarchical model that controls
for the fact that some of the observations are derived from the
same article and, therefore, are not independent. Accordingly,
our model accounts for 68% of the residual variances in the
underlying studiesa substantial improvement over that of the
model reported in Scheibehenne et al. (2010), which explains
only 36% of the variance in the underlying data.
4
To examine whether the discrepancy between our findings
and those reported by prior research can be attributed to the
differences in the data or the differences in the method, we
reanalyzed the studies used Scheibehenne et al. (2010) with our
conceptual model. The analysis produced results (reported in
the Appendix B) consistent with those reported in Table 2
earlier in this paper. The fact that we validated our model in the
context of the studies used by prior research suggests that the
differences in the findings reported by the two meta-analyses
cannot be attributed only to the differences in the underlying
studies, but that they also stem from differences in conceptu-
alizing the effects of assortment size on choice overload.
Future research
Despite its conceptual contributions, our research is only a
first step toward developing a comprehensive understanding of
how assortment size influences consumer choice and identifying
conditions in which larger assortments produce choice overload.
Indeed, the ability of meta-analytic research to uncover novel
factors and quantify their effect is bound by the underlying
empirical research. Therefore, further research might identify
factors in addition to the four outlined in our study that are likely
to influence choice overload. In this context, our analysis helps
pinpoint the areas that would benefit from further investigation.
One particular area in need of further investigation is the
impact of the decision maker's goals on choice overload.
Indeed, compared to the other moderators of choice overload
that have been investigated in multiple experiments, we were
able to identify only a handful of studies explicitly examining
how consumer goals influence choice overload. Accordingly,
further research might seek to identify whether and how other
goal-related factors such as decision accuracy, effort minimi-
zation, and purchase quantity influence consumer decision
processes (Kahn et al., 2014).
A related overlooked area is how consumers' affective states
influence the impact of assortment size on choice. Specifically,
future research might examine the effect of positive versus
negative emotions on the likelihood that consumers will
experience choice overload. Our meta-analysis is consistent
with the notion that the effect of assortment size on choice
overload is likely to be a function of the affective state of the
decision maker. Indeed, our data show that regret, as an
operationalization of individuals' decision goal, was a partic-
ularly strong driver of choice overloada finding consistent
with the fundamental role of regret in self-regulation (Gilovich
& Medvec, 1995; Simonson, 1992). In this context, future
research might examine whether other affective states are likely
to amplify or attenuate the impact of assortment size on choice
overload. Furthermore, in addition to studying emotions as
antecedents of choice overload, future research might examine
emotions as a consequence of choice overload, focusing on the
different types of affective outcomes associated with choosing
from large assortments (Inbar et al., 2011).
It is also notable that studies measuring choice among
assortments, while very similar in terms of their ability to capture
the effects of the four moderators, stand out in terms of their main
effect. Thus, the observations obtained from the three papers using
assortment choice as a dependent variable (Chernev, 2006;
Chernev & Hamilton, 2009; Goodman & Malkoc, 2012)indicate
a much weaker choice overload effect, often displaying the
opposite (more is better) effect. This result likely stems from the
fact that when choosing among assortments, consumers express
their expectations of choice overload rather than the actual
overload. This finding suggests that when choosing among
assortments, consumers are likely to underestimate the choice
overload they will experience when making their final choice from
the assortment selected. In this context, investigating consumer
accuracy in anticipating and measuring choice overload is an
important area for further research.
4
The 56% estimate reported by Scheibehenne et al. (2010) is calculated as
VarianceReducedModelVarianceFullModel
VarianceFullModel , which is an unconventional measure that we
believe does not accurately report the amount of explained variance and can
lead to outcomes whereby the amount of explained variance exceeds 100%. The
conventional and more meaningful measure VarianceReducedModel VarianceFullModel
VarianceReducedModel
suggests that the 2010 analysis accounts for only 36% of the variance in the
data.
353A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
Further research might also investigate the relationship
between the individual factors influencing choice overload. Our
analysis examines the impact of choice set complexity, decision
task difficulty, preference uncertainty, and decision goal in
isolation from one another. This is because a very small fraction
of the extant research has explicitly examined the simultaneous
impact of multiple moderating factors on choice overload.
Therefore, future empirical research is needed to examine the
way in which the four factors identified by our research interact
with one anotheran approach that involves using more
complex (e.g., three-factor) experimental designs.
The analyses of new empirical data collected to address
these issues will benefit from an overarching framework that
provides a basis for comparing the data reported by the
individual studies. Our model provides such a framework to
guide future research on choice overload. The data compiled by
future research will, in turn, serve as the basis for further
theory-guided analysis to enhance our understanding of the
decision processes that determine when and how assortment
size leads to choice overload.
Appendix A. Overview of the analyzed studies
The meta-analysis is based on 99 observations, 81 of which
capture the four experimental factors: choice set complexity,
decision task difficulty, preference uncertainty, and decision
goal.
5
Specifically, the impact of choice set complexity is
captured by 18 observations derived from nine studies reported
in four different articles by Chernev (2005);Gourville and
Soman (2005);Chernev and Hamilton (2009), and Sela et al.
(2009). The impact of decision task difficulty is captured in 17
observations from eight studies reported in six articles by
Gourville and Soman (2005);Haynes (2009);Scheibehenne et
al. (2009);Greifeneder et al. (2010);Inbar et al. (2011), and
Townsend and Kahn (2014). The impact of preference
uncertainty is captured in 22 observations from ten studies
reported in five articles by Chernev (2003a, 2003b);Mogilner
et al. (2008);Scheibehenne et al. (2009), and Morrin et al.
(2012). Finally, the impact of the decision goal is captured in
24 observations from eleven studies reported in four articles by
Oppewal and Koelemeijer (2005);Chernev (2006);Lin and Wu
(2006);Goodman and Malkoc (2012).
In addition to studies explicitly measuring the effect of the
four factors identified in our model, there are several studies
examining choice overload that did not include moderating
factors. Some of these studies aim simply to document the
choice overload phenomenon (Diehl & Poynor, 2010; Fasolo,
Carmeci, & Misuraca, 2009; Iyengar & Lepper, 2000; Shah &
Wolford, 2007), whereas others function as control conditions
in articles that contain other moderators (Inbar et al., 2011;
Scheibehenne et al., 2009; Sela et al., 2009). Collectively, these
data are represented in 18 observations from 15 studies reported
in seven articles.
Below, we offer a list of articles included in the meta-analysis,
outlining the studies and observations included from each article.
1. Research by Iyengar and Lepper (2000) examines the
impact of assortment size on choice overload. It consists of
three studies without moderators, each involving two
conditions (study 3 has an additional no choice condition;
for the meta-analysis, we consider only the small and large
assortment conditions). In all studies, the main dependent
variable is choice deferral, but the second and third studies
additionally report satisfaction. Therefore, this paper is
responsible for five total observations (coded as 15; one
from study 1 and two each from studies 2 and 3).
2. Research by Chernev (2003a) examines how the existence
of an ideal point moderates the impact of assortment size on
choice overload. In our meta-analysis, we use only the first
study (studies 2, 3, and 4 analyze decision processes, not
choice overload). This 2 × 2 study measures two separate
dependent variables: the percentage of participants choos-
ing from the larger assortment and participants' confidence
with their selection. This article is thus responsible for four
observations in our meta-analysis (coded as 69).
3. Research by Chernev (2003b) examines how preference
uncertainty moderates the impact of assortment size on
choice overload. Through four parallel studies, the depen-
dent variable is operationalized as the percentage of subjects
who switch their selection; in addition, the final study
measures respondents' confidence in their decisions. Each
study follows a 2 × 2 design in which one of the factors is
assortment size and the other indicates whether there is an
articulated ideal point. Consequently, these four studies are
recorded as 10 separate observations (coded as 1019; two
observations apiece for studies 13 and four observations
for study 4).
4. Research by Chernev (2005) examines how feature
complementarity moderates the impact of assortment
size on choice overload. The experimental task involves
choice from within an assortment, and the dependent
variable is operationalized as choice deferralthe
percentage of participants who postpone a selection. In
our meta-analysis, we use the second of three studies
(studies 1 and 3 do not use assortment size as a
moderator). The relevant study follows a 2 × 2 design
and is responsible for two observations (coded as
2021).
5. Research by Gourville and Soman (2005) presents three
diverse studies dealing with the impact of assortment size
on choice overload. Specifically, study 1 examines how
alignability of product attributes moderates the impact of
assortment on choice overload, operationalized as the
percentage of respondents choosing from the larger
assortment. The second study examines how the number
of attributes of the available options influences consumer
preferences for the larger assortment. The third study
examines the impact of accountability. Generalizing to
the conceptual factors, the first study examines the
impact of choice set complexity (operationalized in
5
Note that some studies/articles test multiple experimental factors and, as a
result, the sum of articles across factors is larger than the total number of
articles.
354 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
terms of alignability), while the latter two consider
decision task difficulty (operationalized in terms of the
number of attributes and decision accountability). Each
study follows a 2 × 2 design (study 1 has five
assortment-size conditions; we combine the two smallest
and the two largest assortments to construct small and
large sets). In total, this paper is responsible for six
observations in our meta-analysis (coded as 2227; two
observations per study).
6. Research by Oppewal and Koelemeijer (2005) exam-
ines the impact of assortment size on choice overload
when effort minimization is not a decision goal. From
the perspective of the meta-analysis, this paper follows
a 1 × 2 design in which all participants are variety
seeking (the study has eight separate assortment size
conditions; we combine the two smallest and the two
largest assortments to construct small and large sets).
Satisfaction with the assortment is the lone depen-
dent variable, so we record a single observation (coded
as 28).
7. Research by Chernev (2006) examines how decision focus
moderates the impact of assortment size on choice
overload. The experimental task involves choice among
assortments, and the key dependent variable is operation-
alized as the percentage of participants who select each
assortment. The third and fourth studies also measure
choice satisfaction as a secondary dependent variable. The
first three studies use a 2 × 2 design in which one of the
factors is assortment size and the other is decision goal
(choosing an assortment vs. choosing a specific option).
The fourth study uses a 2 × 2 × 2 design (the additional
moderator is the presence of a dominant option); for the
purposes of meta-analysis, we focus only on the control
condition of the third factor, which results in a 2 × 2
design. Overall, this paper is responsible for twelve total
observations (coded as 2940; two observations each
for studies 1 and 2; four observations each for studies 3
and 4).
8. Research by Lin and Wu (2006) examines how the need for
cognition moderates the impact of assortment size on
choice overload. The dependent variable is regret, which is
operationalized as the willingness to switch to another
option. In a single 2 × 2 study, subjects are split into
conditions by assortment sizeand need for cognition, which
determine whether or not they have an effort-minimization
goal. Even though this design lends itself to two
meta-analytic observations, the data reported in the
low-need-for-cognition condition had a much smaller
variance (more than 12 times smaller than the other
condition's reported variance) and was excluded from the
analysis. Accordingly, this paper accounts for one obser-
vation (high need for cognition) in our meta-analysis
(coded as 41).
9. Research by Shah and Wolford (2007) examines the
impact of assortment size on choice overload without
moderators. This is a single-factor (assortment size) study
that measures the percentage of participants who select
any option (we combined the eight assortment-size
conditions to construct small and large sets). According-
ly, we record a single observation in the meta-analysis
(coded as 42).
10. Research by Mogilner et al. (2008) examines how the
existence of an ideal point moderates the impact of
assortment size on choice overload. Satisfaction with the
chosen alternative operates as the dependent variable. In
the meta-analysis, we utilize a subset of the third study
(the others do not manipulate assortment size), which has
a 2 × 2 setup. One of the factors is assortment size and
the other indicates whether there is an articulated ideal
point. Accordingly, we record two observations from this
paper (coded as 4344).
11. Research by Chernev and Hamilton (2009) examines
how option attractiveness moderates the impact of
assortment size on choice overload. The experimental
task involves choice among assortments, and the
dependent variable is operationalized as the percentage
of participants who select each assortment. In our
meta-analysis, we use the first five studies (the sixth
study does not directly measure choice overload). All five
studies follow 2 × 2 designs in which one of the factors is
assortment size and the other is the attractiveness of the
available options (study 4 has three assortment size
conditions; for the meta-analysis, we consider only the
smallest and largest sets). Therefore, in our meta-analysis
these five studies are recorded as ten separate observa-
tions (coded as 4554; two observations per study).
12. Research by Fasolo et al. (2009) examines the impact of
assortment size on choice overload. The experimental
task involves choice from an assortment, and the
dependent variable is operationalized as satisfaction
with the chosen alternative. Study 1 utilizes a
single-factor design and study 2 has a 2 × 2 design,
which is treated as comprising two simple effects.
Consequently, this article is responsible for a total of
three observations (coded as 5557; one observation for
the first study and two for the second).
13. Research by Haynes (2009) examines how time pressure
moderates the impact of assortment size on choice
overload. In this paper, there are two dependent variables
as participants report their satisfaction and regret with the
chosen alternative. The article involves a single study with
a 2 × 2 design, which is responsible for a total of four
observations (coded as 5861; two per dependent variable).
14. Research by Scheibehenne et al. (2009) presents three
sets of studies dealing with the impact of assortment size
on choice overload. Whereas the first two sets measure
choice deferral as the dependent variable, the third
records choice satisfaction. The first study follows a
1 × 2 design without any moderators. In the second set of
studies, 2a utilizes expertise to moderate preference
uncertainty, 2b has no moderator, and 2c requires
subjects to justify their selections, thereby altering the
difficulty of the decision task. Study 2a follows a 2 × 2
design, while 2b and 2c have 1 × 2 setups. (Studies 2b
355A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
and 2c have three assortment size conditions; for the
meta-analysis, we consider only the smallest and largest
sets.) Finally, the third set of studies consists of two
equivalent 1 × 2 within-subject designs; all participants
are primed to gain expertise, which results in low
preference uncertainty. Accordingly, this paper accounts
for seven total observations (coded as 6268; two
observations for study 2a and one observation apiece
for studies 1, 2b, 2c, 3a and 3b).
15. Research by Sela et al. (2009) examines the impact of
product typeutilitarian or hedonicon choice overload
and selection. Throughout the dependent variable is
measured as the percentage of respondents choosing a
hedonic option (we focus on the assortments with an equal
number of utilitarian and hedonic goods). The paper
consists of six studies. Studies 1a, 1b, 2, and 3 have no
moderators, while 4 and 5 examine how choice set
complexity (operationalized in terms of ease of justification
that creates a dominant product or category) moderates the
impact of assortment on choice overload. Accordingly,
studies 13 follow 1 × 2 designs, whereas studies 4 and 5
have 2 × 2 set-ups. In our meta-analysis, these six studies
are recorded as eight separate observations (coded as
6976; one observation each for studies 1a, 1b, 2 and 3;
two observations each for studies 4 and 5).
16. Research by Diehl and Poynor (2010) examines the impact
of assortment size on choice overload. The experimental
task involves a choice within an assortment, and the
dependent variable is operationalized as choice satisfaction.
In our meta-analysis, we include studies 2 and 3 (study 1
does not directly manipulate assortment size). Both utilize
1 × 2 experimental designs, accounting for a total of two
observations (coded as 7778; one from each study).
17. Research by Greifeneder et al. (2010) examines how
decision task difficulty moderates the impact of assortment
size on choice overload. In two similar studies, the
dependent variable is recorded as satisfaction with the
chosen alternative. Both studies manipulate the number of
product attributes in a 2 × 2 design, and they are
collectively responsible for four observations (coded as
7982; two observations per study).
18. Research by Inbar et al. (2011) examines the impact of
assortment size on choice overload. The experimental
task involves choice from an assortment, and the
dependent variable is operationalized as regret with the
chosen alternative. In our meta-analysis, we use the first
two studies (the third study has a different focus). Study 1
follows a 1 × 2 design without any moderating variable;
study 2 follows a 2 × 2 design and examines how time
pressure (the feeling of being rushed) influences the
impact of assortment size on choice overload. Accord-
ingly, this article is responsible for three observations
(coded as 8385; one for study 1 and two for study 2).
19. Research by Goodman and Malkoc (2012) examines how
psychological distance moderates the impact of assortment
size on choice among assortments. The experimental task
involves choice among assortments, and the dependent
variable is operationalized as the percentage of participants
who select each assortment. In our meta-analysis, we use
the first five studies (the sixth study does not directly
measure choice among assortments). Studies 1A, 1B, 2, and
3 follow a 2 × 2 design, in which one of the factors is
assortment size and the other is the psychological distance.
Study 4 uses a 2 × 2 × 2 design (the additional moderator is
the prominence of the choice tradeoffs); for the purposes of
meta-analysis, we focus only on the control condition of the
third factor, thus resulting in a 2 × 2 design. Consequently,
this paper is responsible for ten observations (coded as
8695; two for each study).
20. Research by Morrin et al. (2012) examines how expertise
moderates the impact of assortment size on choice. The
experimental task involves choice among assortments,
and the dependent variable is operationalized as the
percent of participants who select an alternative. In the
meta-analysis, we utilize the first study (the other studies
do not manipulate assortment size). This study uses a
2 × 2 × 2 design (the third factor is the existence of a
default option [target fund]); for the purposes of
meta-analysis, we focus only on the control condition of
the third factor, thus resulting in a 2 × 2 design.
Accordingly, this paper is responsible for two observa-
tions in the meta-analysis (coded as 9697).
21. Research by Townsend and Kahn (2014) examines how
presentation format moderates the impact of assortment
size on choice overload. The experimental task involves
choice among assortments, and the dependent variable is
operationalized as the percentage of participants who
select each assortment. In our meta-analysis, we focus on
study 5 (studies 1 and 2 do not assign experimental
conditions randomly, and studies 3 and 4 do not directly
measure overload). Study 5 has a 2 (presentation format:
verbal vs. visual) × 2 (assortment size: high vs. low)
experimental design. Accordingly, this paper accounts
for a total of two observations (coded as 9899).
The data used in the meta-analysis were obtained directly
from the published papers with the exception of cases in which
the published version of the paper did not report all of the data
necessary to calculate effect sizes (Chernev, 2003a, 2006;
Mogilner et al., 2008; Oppewal & Koelemeijer, 2005; Lin &
Wu, 2006; Townsend & Kahn, 2014; Goodman & Malkoc,
2012; Morrin et al. (2012);Ketcham et al., 2012). In such cases,
additional data were obtained by contacting the authors. The
additional data for the studies by Greifeneder et al. (2010) and
Scheibehenne et al. (2009) were obtained from the dataset used
by Scheibehenne et al. (2010). We were unable to obtain the
necessary data from three papers (Huberman et al., 2007;
Ketcham et al., 2012; Reutskaja & Hogarth, 2009) and
accordingly did not include these papers in the meta-analysis.
Appendix B. Reanalyzing the data from prior meta-analytic
research
Supplementary data to this article can be found online at
http://dx.doi.org/10.1016/j.jcps.2014.08.002.
356 A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358
References
Arentze, Theo A., Oppewal, Harmen, & Timmermans, Harry J. P. (2005). A
multipurpose shopping trip model to assess retail agglomeration effects.
Journal of Marketing Research,42(February), 109115.
Babin, Barry J., Darden, WilliamR.,&Griffin,Mitch(1994).Work and/or fun:
Measuring hedonic and utilitarian shopping value. Journal of Consumer
Research,20(March), 644656.
Baumol, William J., & Ide, Edward A. (1956). Variety in retailing.
Management Science,3(October), 93101.
Becker, Betsy Jane (2001). Examining theoretical models through research
synthesis: The benefits of model-driven meta-analysis. Evaluation & the
Health Professions,24(June), 190217.
Berger, Jonah, Draganska, Michaela, & Simonson, Itamar (2007). The influence
of product variety on brand perception and choice. Marketing Science,
26(July), 460472.
Betancourt, Roger, & Gautschi, David (1990). Demand complementarities,
household production, and retail assortments. Marketing Science,9(Spring),
146161.
Bettman, James R., Luce, Mary Frances, & Payne, John W. (1998). Constructive
consumer choice processes. Journal of Consumer Research,25(December),
187217.
Boatwright, Peter, & Nunes, Joseph C. (2001). Reducing assortment: An
attribute-based approach. Journal of Marketing,65(July), 5063.
Botti, Simona, & Iyengar, Sheena S. (2004). The psychological pleasure and
pain of choosing: When people prefer choosing at the cost of subsequent
outcome satisfaction. Journal of Personality and Social Psychology,
87(September), 312326.
Broniarczyk, Susan (2008). Product assortment. Handbook of consumer
psychology (pp. 755779). New York, NY: Lawrence Erlbaum Associates.
Broniarczyk, Susan M., Hoyer, Wayne D., & McAlister, Leigh (1998). Consumers'
perceptions of the assortment offered in a grocery category: The impact of item
reduction. Journal of Marketing Research,35(May), 166176.
Carpenter, Gregory S., & Nakamoto, Kent (1989). Consumer preference
formation and pioneering advantage. Journal of Marketing Research,
26(August), 285298.
Chernev, Alexander (2003a). Product assortment and individual decision processes.
Journal of Personality and Social Psychology,85(July), 151162.
Chernev, Alexander (2003b). When more is less and less is more: The role of
ideal point availability and assortment in consumer choice. Journal of
Consumer Research,30(September), 170183.
Chernev, Alexander (2005). Feature complementarity and assortment in choice.
Journal of Consumer Research,31(March), 748759.
Chernev, Alexander (2006). Decision focus and consumer choice among
assortments. Journal of Consumer Research,33(June), 5059.
Chernev, Alexander (2012). Product assortment and consumer choice: An
interdisciplinary review. Foundations and Trends in Marketing,6(1), 161.
Chernev, Alexander, Böckenholt, Ulf, & Goodman, Joseph (2010). Choice
overload: Is there anything to it? Journal of Consumer Research,37(October),
426428.
Chernev, Alexander, & Hamilton, Ryan (2009). Assortment size and option
attractiveness in consumer choice among retailers. Journal of Marketing
Research,46(June), 410420.
Choi, Jinhee, & Fishbach, Ayelet (2011). Choice as an end versus a means.
Journal of Marketing Research,48(3), 544554.
Cohen, Jacob (1988). Statistical power analysis for the behavioral sciences
(2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.
Dhar, Ravi (1997). Consumer preference for a no-choice option. Journal of
Consumer Research,24(September), 215231. Dhar, Ravi, & Nowlis,
Stephan M. (1999). The effect of time pressure on consumer choice deferral.
Journal of Consumer Research,25(March), 369384.
Dhar, Ravi, & Simonson, Itamar (2003). The effect of forced choice on choice.
Journal of Marketing Research,40(May), 146160.
Diehl, Kristin (2005). When two rights make a wrong: Searching too much in
ordered environments. Journal of Marketing Research,42(August), 313322.
Diehl, Kristin, Kornish, Laura J., & Lynch, John G., Jr. (2003). Smart agents:
When lower search costs for quality information increase price sensitivity.
Journal of Consumer Research,30(June), 5671.
Diehl, Kristin, & Poynor, Cait (2010). Great expectations?! Assortment size,
expectations and satisfaction. Journal of Marketing Research,47(April),
312322.
Draganska, Michaela, & Jain, Dipak C. (2005). Product-line length as a
competitive tool. Journal of Economics & Management Strategy,
14(March), 128.
Fasolo, Barbara, Carmeci, Floriana A., & Misuraca, Raffaella (2009). The effect
of choice complexity on perception of time spent choosing: When choice
takes longer but feels shorter. Psychology & Marketing,26(3), 213228.
Gilovich, Thomas, & Medvec, Victoria Husted (1995). The experience of
regret: What, when, and why. Psychological Review,102(April),
379395.
Goodman, Joseph, & Malkoc, Selin (2012). Choosing here and now versus
there and later: The moderating role of psychological distance on assortment
size preferences. Journal of Consumer Research,39(December), 751768.
Gourville, John T., & Soman, Dilip (2005). Overchoice and assortment type: When
and why variety backfires. Marketing Science,24(Summer), 382395.
Greenleaf, Eric A., & Lehmann, Donald R. (1995). Reasons for substantial delay in
consumer decision making. Journal of Consumer Research,22(September),
186199.
Greifeneder, Rainer, Scheibehenne, Benjamin, & Kleber, Nina (2010). Less
may be more when choosing is difficult: Choice complexity and too much
choice. Acta Psychologica,133(January), 4550.
Griffin, Jill G., & Broniarczyk, Susan (2010). The slippery slope: The impact of
feature alignability on search and satisfaction. Journal of Marketing
Research,47(April), 323334.
Hamilton, Ryan, & Chernev, Alexander (2010). The impact of product line
extensions and consumer goals on the formation of price image. Journal of
Marketing Research,47(February), 5162.
Haynes,GraemeA.(2009).Testing the boundaries of the choice overload
phenomenon: The effect of number of options and time pressure on
decision difficulty and satisfaction. Psychology & Marketing,26,
204212.
Hoch, Stephen J., Bradlow, Eric T., & Wansink, Brian (1999). The variety of an
assortment. Marketing Science,18(4), 527546.
Hotelling, Harold (1929). Stability in competition. The Economic Journal,
39(March), 4157.
Huber, Joel, Payne, John W., & Puto, Christopher (1982). Adding asymmet-
rically dominated alternatives: Violations of regularity and the similarity
hypothesis. Journal of Consumer Research,9(June), 9098.
Huberman, Gur, Iyengar, Sheena, & Jiang, Wei (2007). Defined contribution
pension plans: Determinants of participation and contributions rates.
Journal of Financial Services Research,31(1), 132.
Huedo-Medina, Tania B., Sanchez-Meca, Julio, Marin-Martitnez, Fulgencio, &
Botella, Juan (2006). Assessing heterogeneity in meta-analysis: Q statistic
or I2 index? Psychological Methods,11(2), 193206.
Huffman, Cynthia, & Kahn, Barbara E. (1998). Variety for sale: Mass
customization or mass confusion? JournalofRetailing,74(Winter),
491513.
Inbar, Yoel, Botti, Simona, & Hanko, Karlene (2011). Decision speed and
choice regret: When haste feels like waste. Journal of Experimental Social
Psychology,47(May), 533540.
Inman, J. Jeffrey (2001). The role of sensory-specific satiety in attribute-level
variety seeking. Journal of Consumer Research,28(June), 105120.
Iyengar, Sheena (2010). The art of choosing. London, UK: Little Brown.
Iyengar, Sheena S., & Lepper, Mark R. (2000). When choice is demotivating:
Can one desire too much of a good thing? Journal of Personality and Social
Psychology,79(December), 9951006.
Iyengar, Sheena S., Wells, Rachael E., & Schwartz, Barry (2006). Doing better
but feeling worse. Psychological Science,17(February), 143150.
Kahn, Barbara E. (1995). Consumer variety-seeking among goods and services.
Journal of Retailing & Consumer Services,2(July), 139148.
Kahn, Barbara E. (1999). Introduction to the Special Issue: Assortment
Planning. Journal of Retailing,75(Fall), 289293.
Kahn, Barbara, Chernev, Alexander, Böckenholt, Ulf, Bundorf, Kate,
Draganska, Michaela, Hamilton, Ryan, et al. (2014). Consumer and
managerial goals in assortment choice and design. Marketing Letters.
http://dx.doi.org/10.1007/s11002-014-9307-0.
357A. Chernev et al. / Journal of Consumer Psychology 25, 2 (2015) 333358