RESEARCH DOSSIER
Choice Set Complexity and Assortment Effects
Abstract. Choice-set complexity denotes the cognitive demands created by the composition and presentation of alternatives. Assortment effects encompass changes in attention, search, selection, deferral, confidence, regret, satisfaction, and consumption caused by properties of a choice environment. This monograph critically synthesizes foundational theory, behavioral experiments, field studies, meta-analyses, computational accounts, and recent studies of digital markets. Its central scientific questions concern the distinction between set size and comparative difficulty; the causal contributions of attribute conflict, preference uncertainty, decision goals, and information organization; the no-choice alternative; and generalization across populations and decision domains. The classic choice-overload results are analyzed against later meta-analytic heterogeneity and moderator evidence, assortment-reduction studies, neural choice-set valuation, and online recommender experiments. Evidence supports a conditional model: cardinality alone does not determine decision difficulty or welfare. Effects depend on comparative structure, access to preferred items, preference clarity, search strategy, how information is arranged, and the time horizon of observation. The synthesis identifies priorities for factorial experiments, replication, independent process measures, longitudinal outcomes, and transparent human–Artificial decision support during the Artificial Era.
Keywords: choice-set complexity; assortment size; assortment effects; choice overload; decision difficulty; consumer psychology; choice deferral; constructed preferences; preference uncertainty; perceived variety; attribute conflict; online recommendation systems; Human–Artificial decision partnership.
Scientific Overview: The Architecture of a Choice Set
A choice set is the collection of alternatives presented to an agent for consideration within a particular decision episode. Choice-set complexity refers to the demands that this collection imposes on discrimination, comparison, information integration, preference construction, and eventual commitment. Assortment effects are changes in perceived opportunity, decision process, selection, deferral, satisfaction, or subsequent behavior attributable to the number, composition, presentation, and organization of available alternatives. The basic scientific question is how the structure of an opportunity environment interacts with the cognitive work needed to select an option. An assortment of twelve nearly interchangeable products and an assortment of twelve products differentiated by conflicting attributes have the same cardinality and potentially very different psychological consequences. The construct therefore requires the separate observation of numerical size, attribute dimensionality, overlap, dominance relations, organization, and subjective difficulty.
The fundamental empirical result is conditional. Some enlargements of a choice set increase preference for the assortment, perceived freedom, the likelihood of a satisfactory match, and purchase opportunities. Others increase uncertainty, information processing, decision deferral, or regret. The classical experiments of Iyengar and Lepper (2000) Iyengar (2000) initiated a major research program on the adverse consequences of large assortments. A subsequent meta-analysis by Scheibehenne, Greifeneder, and Todd (2010) Scheibehenne (2010) reported a mean effect close to zero across heterogeneous studies, whereas the moderator-based synthesis by Chernev, Böckenholt, and Goodman (2015) Chernev (2015) identified the circumstances in which overload is more likely. A later retail assortment synthesis examined 177 studies drawn from 95 papers and placed benefits and costs within the same evidence framework (2022 review Sethuraman (2022)). These results establish the need to study the decision environment as a structured system rather than treat option count as a universal causal mechanism.
The object of this monograph is the relation between the measurable structure of alternatives and the psychological processes of choosing among them. Its research questions concern how assortments are mentally represented; which structures increase discrimination costs; how preferences, goals, expertise, time pressure, and affect moderate performance; when selection can be postponed; which outcomes qualify as choice overload; and how experimental, field, computational, and neural evidence can be reconciled. The synthesis is a critical, source-grounded narrative research monograph. It does not report new participants, a preregistered systematic search, a newly computed pooled effect, or a clinical recommendation.
Foundational Definitions and Construct Boundaries
Assortment size is the count of distinguishable alternatives encountered within a designated category and time window. Perceived assortment variety concerns the judged diversity and richness of that set. Choice-set complexity concerns the computational and psychological work needed to identify, compare, and commit among the alternatives. Decision-task difficulty also depends on instructions, consequences, time available, need to justify a choice, and whether abstention is permitted. Preference uncertainty concerns how clearly a decision maker knows which characteristics are most important. These variables are linked but not interchangeable. A high-count set with a dominant option may be easier to decide from than a low-count set whose alternatives involve difficult compensatory trade-offs.
The terminology also separates process and outcome. A person can experience high decision effort and still select well; can report satisfaction after deliberating extensively; or can decline to choose because available options are unacceptable rather than cognitively overwhelming. Choice overload is an outcome pattern in which an expansion of choice is associated with worse decision-related consequences under particular conditions, with the causal interpretation strengthened by random assignment of set size. Common operational measures include deferral, probability of choosing, time and effort, satisfaction or confidence, regret, and switching. Chernev et al. (2015) Chernev (2015) treated several such measures as alternative operationalizations in their synthesis; conceptual interchangeability should nevertheless be assessed rather than assumed in every experiment.
The scientific boundary is important. General utility theory, risk preference, and judgment research provide models of the chooser but are independently studied in the broader Psychology corpus. The present investigation centers on how the composition of a feasible set alters the processes and consequences of consumer or analogous decisions. Price anchoring and specific pricing mechanisms, persuasive message framing, and general interface accessibility are adjacent research objects. The assortment is analyzed as a relational structure: the identity and attractiveness of each item depend in part on which others appear beside it, and the task cost depends on the network of comparisons required to differentiate them. The definition remains operational: observable manipulations and responses, with theoretical mechanisms proposed and tested separately.
Historical Foundations: Rational Choice and Bounded Cognition
Traditional formal choice models ordinarily presume that an agent can compare alternatives under a stable preference relation and select an option that is best according to the relevant ordering. Under appropriate assumptions, adding an available alternative cannot make the former optimum unavailable. This formal property expresses an aspect of opportunity expansion and is important as a benchmark. Its psychological applicability depends on whether the preference ordering, attribute knowledge, and evaluation process remain stable during the decision. The science of assortment effects emerged by investigating these conditions instead of treating them as universally given.
Herbert Simon's 1955 account Simon (1955) of bounded rationality framed decisions in terms of limited computational resources and feasible procedures. Later work examined adaptive strategies, including selective information use, attribute-based elimination, and changes in the balance between effort and accuracy. Payne, Bettman, and Johnson's The Adaptive Decision Maker* (1993) Payne (1993) developed a systematic architecture in which task properties shape strategy selection. The historical contribution of this tradition is its shift toward the actual work performed by a decision maker: what must be searched, retained, transformed, and compared.
Context-dependent choice studies added a complementary historical foundation. Huber, Payne, and Puto (1982) Huber (1982) found that introducing an asymmetrically dominated option could change the relative appeal of existing alternatives. Tversky and Shafir (1992) Tversky (1992) demonstrated that conflict can increase delayed choice or reliance on a default even when a choice set is enlarged or improved. The lineage linking bounded rationality, context effects, and deferral shows why the number of offers is psychologically meaningful only within the comparative configuration those offers create. It also establishes that a large assortment may disrupt commitment through conflict without preventing the detection of desirable possibilities.
Preference Construction and Adaptive Decision Processes
A preference can be represented as a durable ordering, a provisional evaluation constructed during a task, or an expressed selection obtained under particular instructions. These are distinct observational targets. Bettman, Luce, and Payne (1998) Bettman (1998) argued that many consumer preferences are assembled through task-contingent strategies because processing resources are limited and relevant attributes differ in salience. Choice-set complexity therefore arises partly from decisions about which criteria to use and how to combine them. The same assortment can become easier after a person learns a priority rule, establishes a minimum acceptable threshold, or recognizes a clearly superior attribute.
Comparative processing can be alternative-wise, in which a chooser reviews the whole profile of each item, or attribute-wise, in which candidates are compared successively on features such as price, quality, and durability. Lexicographic processing concentrates on the most important attribute; compensatory integration permits advantages on one attribute to offset disadvantages on another. Strategies can change in the middle of a choice episode. A larger set may encourage initial screening using one or two attributes, followed by more deliberate inspection of a shortlist. Such multistage processes make total item count an imperfect measure of how many options were seriously evaluated.
The effort–accuracy framework provides a mechanistic interpretation. Individuals regulate cognitive expenditure in relation to expected decision quality, task stakes, and constraints. Under familiar and easily ranked attributes, a selective rule can yield an adequate option with low effort. Where attributes conflict and relative importance is uncertain, the same simplification may sacrifice valued features, thereby increasing anticipated regret or prompting further search. The claim that more options mechanically exhaust a fixed stock of decision energy is too coarse to explain these interactions. Empirical tests need to identify which strategy was invoked, what information was actually inspected, and whether longer deliberation improved the selected outcome.
The Early Experimental Choice-Overload Program
Iyengar and Lepper (2000) Iyengar (2000) reported three studies combining field and laboratory settings. Their assortment comparisons included a smaller set of six alternatives and larger sets of twenty-four or thirty in contexts involving gourmet foods and optional essay topics. Across those experiments, smaller assortments were associated with stronger engagement with selection and, in particular tasks, greater post-choice satisfaction or performance. The contribution was substantial: choice opportunity and selection motivation could diverge. The iconic jam display study should be read as one experimental context among several, with specific products, participants, and outcomes, rather than as proof that six is the universally optimal assortment.
The studies also raised unresolved identification questions. The treatment changed the number of options, which also could change the probability of preference conflict, perceived similarity, range of attributes, attention, and the meaning of selecting a product. Different stages of choice were not equivalent: attraction to a display is an entry behavior, whereas purchase is a commitment event. Field settings enhance ecological relevance but expose decisions to shopping motives, visibility, and incidental influences that may complicate process inference. Laboratory settings better control composition while constraining realistic search and market consequences.
Subsequent experiments examined the functional form of the number–choice relation. Shah and Wolford (2007) Shah (2007) studied purchase-related behavior across parametric changes in the number of options, further motivating designs that vary multiple set sizes rather than only two. The proper analytical question is whether the relationship is monotone, nonlinear, conditional on complexity, or specific to one outcome. Researchers must specify set composition, participant incentives, and the possibility of no choice before translating a single experimental contrast into a general design principle.
From Replication Disagreement to Moderator-Based Synthesis
The meta-analysis of Scheibehenne, Greifeneder, and Todd (2010) Scheibehenne (2010) aggregated 63 conditions from 50 published and unpublished experiments comprising 5,036 participants. Its average choice-overload effect was virtually zero, accompanied by considerable heterogeneity. This is a foundational finding because it contradicts the idea of a strong, uniform cost for all increases in assortment size. The near-zero average is compatible with positive effects in some settings, negative effects in others, and weak or negligible effects in many; the mean alone cannot adjudicate among these process explanations.
Chernev, Böckenholt, and Goodman (2015) Chernev (2015) reviewed 99 observations involving 7,202 participants and identified four moderators: choice-set complexity, task difficulty, preference uncertainty, and the chooser's decision goal. Their model predicts stronger overload when options are difficult to compare, the decision task is demanding, preferences are uncertain, and the chooser is inclined to minimize effort. These moderators convert an apparently contradictory literature into a conditional explanatory program. It remains essential to recognize that moderator estimation in an aggregate synthesis may draw on imperfectly comparable manipulations and may be sensitive to study selection, outcome coding, and correlations among moderators.
The two syntheses answer related but different inferential questions. A grand mean asks whether an expansion tends to have a common direction across sampled conditions. A moderator analysis asks how variation in context predicts variation in the response. The latter is useful for explanation and design, but claims of precise context-specific prediction require external validation in independently selected samples and experimentally isolated moderator manipulations. A contemporary synthesis should report both overall dispersion and theoretically justified conditional contrasts. Effect signs should be tied to clearly coded outcomes rather than to an undefined notion of better or worse choice.
Cardinality, Heterogeneity, and Perceived Variety
Counting items is a necessary description of an assortment but an incomplete account of its informational organization. A collection can consist of near duplicates, strongly differentiated products, or alternatives with overlapping feature bundles. Objective variety may be operationalized as the distribution of feature levels, the number of distinct attributes, category coverage, or entropy of the alternatives. Perceived variety is the consumer's judgment of the extent and diversity of available possibilities. The two can vary independently because display structure changes what is noticed, compared, and remembered.
Kahn and Wansink (2004) Kahn (2004) found across laboratory and field studies that assortment organization influenced perceived variety and, through it, consumption quantities. Mogilner, Rudnick, and Iyengar (2008) Mogilner (2008) found that categories can increase perceived variety and satisfaction among people unfamiliar with the category, even when categorical content itself is not the decisive factor. These experiments demonstrate that a choice set is experienced through its visual and conceptual partitioning. Category labels, grouping, and symmetry alter the immediate psychological structure without necessarily changing the stock of items.
A later experimental meta-analysis by Mejía (2025) Mejía (2025) differentiated perceived numerosity from perceived diversity and analyzed size, composition, arrangement, visual layout, and textual information. This distinction is especially useful for digital marketplaces: adding products can strongly increase the sense of many items while providing diminishing gains in perceived distinctiveness. Accordingly, the explanatory variable in assortment research should include a representation of perceived differentiation and the organization of information. An interface showing a vast collection of redundant results may convey extensive choice while offering relatively little additional coverage of consumer needs.
Alignability, Attribute Conflict, and Comparability
An alignable assortment varies primarily along a shared, compensatory dimension, allowing items to be ordered on a familiar scale. A nonalignable assortment varies across features that resist easy comparison or require exchanges between important values. Gourville and Soman (2005) Gourville (2005) identified assortment type as a moderator of assortment-size effects. In their work, nonalignable variety imposed cognitive effort and potential regret that could make greater variety counterproductive, while modifications to information display and decision reversibility reduced those consequences. The important mechanism is comparative structure rather than item count in isolation.
Attribute conflict is highest when alternatives offer different advantages of comparable importance. Consider two hypothetical devices, one superior in battery life and another in processing performance, for a buyer who has not decided which feature matters most. A larger array of variants can multiply the number of pairwise trade-offs and create additional opportunities to reverse a tentative ranking. Conversely, if one option dominates all others on the attributes a buyer values, an expanded set may still support rapid selection. The example illustrates a structural prediction and does not represent new experimental evidence.
Simulation research adds boundary conditions. Fasolo, McClelland, and Todd (2007) Fasolo (2007) showed in computational analyses that simple attribute-focused strategies can perform well even in large sets when attribute conflict is limited or priorities are highly unequal. Since simulation assumptions define the structure of attributes and payoffs, the results specify computational possibilities rather than universal observations of real shoppers. A strong research design must independently manipulate set size, attribute correlations, attribute weighting, and display comparability. It should determine whether the participant's actual strategy mediates the effect of experimentally induced complexity on choice quality.
Ideal Points, Expertise, and Preference Uncertainty
Preference uncertainty can be reduced when a chooser has an ideal point: an explicit target profile of desired features or acceptable ranges. Chernev (2003) Chernev (2003) examined how the availability of such an ideal point changes reactions to small and large assortments. Four experiments supported the prediction that clearer ideals can simplify large-set choice and alter the strength of preferences after selection. This evidence makes expertise, category knowledge, and prior goal formulation central to understanding why an assortment is manageable for one person and burdensome for another.
An ideal point serves as an internal filter. It converts an open-ended comparative problem into a targeted search in which nonmatching alternatives are excluded early. Yet an ideal can also be incomplete or unstable: consumers may know a budget but not the desired durability, or know a feature priority without knowing how product descriptions correspond to performance. Hence possession of a stated preference is not automatically evidence of decision competence. Investigators need to distinguish experimentally supplied targets, independently measured preferences, and preferences elicited after participants have already inspected the alternatives.
Theoretical analysis further distinguishes subjective uncertainty about tastes from uncertainty about facts. Someone may know which attributes matter but be unsure whether a product will deliver them; another person may know every product specification while remaining undecided about their desired trade-off. The first condition calls for better predictive information and trust calibration, while the second may benefit from reflecting on values or making acceptable trade-off rules explicit. Assortment research that combines these forms of uncertainty under one self-report item risks obscuring their different effects on confidence, regret, and choice deferral.
Information Search, Screening, and Cognitive Workload
A consumer rarely inspects every alternative with equal depth. Real searches involve scanning thumbnails, reading a subset of descriptions, comparing finalists, revisiting earlier options, and stopping when the expected value of further search is below its cost. These behaviors can be recorded through clicks, scrolls, gaze, dwell time, opening of details, and actual purchases. Search costs are influenced by the number of items visible at once, organization into pages or categories, stability of ranking, and the difficulty of remembering previously viewed alternatives. The effective set under consideration may be much smaller than the catalog.
An adaptive interpretation links expanded choice to shifts in information strategy. Search costs may produce shallower inspection per option, a stronger reliance on default rankings, or a transition from compensatory evaluation to screening heuristics. Such changes do not establish irrationality by themselves. Their consequences depend on whether the heuristic captures the consumer's important objectives. Payne, Bettman, and Johnson (1993) Payne (1993) provide the theoretical foundation for examining this effort–accuracy regulation. Time pressure, interruptions, and memory demands should be measured rather than presumed.
Crucially, observed time can have multiple meanings. A long decision may indicate conflict, motivated deliberation, enjoyment, or an elaborate but successful search. A short decision may reflect a clear favorite, disinterest, premature abandonment, or efficient filtering. Process tracing should therefore be linked to outcomes and participant goals. For example, a randomized design can compare how many options are examined, whether a shortlist is formed, and whether the selected product matches an independently measured goal. Interpreting time alone as a direct measure of overload is psychometrically fragile.
Choice Conflict and the No-Choice Alternative
The option to defer a purchase, retain a status quo, or resume search changes the decision problem. In an artificially forced choice task, every participant must select from the presented set, so uncertainty may appear as fluctuating relative shares or low confidence. In a real market, the same conflict may lead to abandonment, postponement, or selection of a familiar default. Tversky and Shafir (1992) Tversky (1992) showed that conflict can produce a preference for delaying a decision even when more alternatives are available. Deferral is consequently a central behavioral indicator of assortment effects, provided that it is interpreted alongside practical constraints and the attractiveness of the outside option.
Dhar (1997) Dhar (1997) examined the no-choice alternative and found that hesitation can depend on how clearly one option stands apart from its competitors. Dhar and Simonson (2003) Dhar (2003) showed that permitting no choice changes the pattern of context effects compared with forced-choice settings. These studies complicate direct extrapolation from multiple-choice experiments to checkout behavior: an observed selection distribution conditional on purchase does not identify the probability that an entire assortment will be rejected.
A complete outcome model distinguishes entry into the decision, evaluation of alternatives, initial commitment, completion of the transaction, and later satisfaction. Deferral may be adaptive when the consumer expects more useful information or considers the offered set poor. It may be maladaptive when decisions must be made promptly and further search yields little benefit. Causal attribution requires a defensible counterfactual: what would this decision maker have done under the same stakes and outside-option conditions if the choice set had been organized or sized differently? This question favors experiments that explicitly preserve the no-choice opportunity.
Context Effects, Dominance, and Relative Attractiveness
The set affects choice partly by changing the relative standing of its members. Huber, Payne, and Puto (1982) Huber (1982) demonstrated the attraction effect: adding an asymmetrically dominated alternative can increase selection of the item that dominates it. This is a departure from simple context-independent choice share predictions and reveals that preference is influenced by the structure of comparisons. It does not imply that every decoy will reliably produce the effect or that ordinary purchasers always calculate formal dominance relations.
A sufficiently attractive new item can also reduce choice difficulty by becoming a clear leader. Thus the consequence of adding one option depends on what the option contributes to the comparison network. If it clarifies a distinction, covers an unmet preference, or establishes a useful reference category, search may become easier. If it introduces a finely balanced trade-off or raises the salience of an unresolved attribute conflict, commitment may become harder. These propositions require comparison of sets holding constant baseline attractiveness and measurement of preferences prior to exposure where feasible.
Related context effects include compromise selection, similarity competition, and shifts induced by framing of attributes. Their relevance here is that comparative effects and overload can coexist, oppose each other, or vary at different stages of choosing. A decoy might simplify a difficult choice while also increasing the raw number of items. A crowded assortment might heighten use of a recommended option, producing higher selection of that option while lowering overall willingness to purchase. Assortment research must measure both total decision incidence and conditional choice shares to avoid treating these as identical outcomes.
Affective Conflict, Anticipated Regret, and Emotion Regulation
Choices involving important, noncompensable attributes can elicit emotional distress alongside cognitive effort. Luce, Bettman, and Payne (1997) Luce (1997) investigated processing in emotionally difficult decisions and developed an account in which coping goals interact with accuracy and effort considerations. Consumers may respond to emotionally painful trade-offs by examining more information, avoiding certain comparisons, seeking a justifiable reason, or delaying commitment. Emotional difficulty is therefore a distinct moderator of the response to a large assortment.
Anticipated regret can rise when many unchosen options remain salient. Every chosen item excludes alternatives that might later appear superior on a particular dimension. Such counterfactual comparisons are especially likely when outcomes are uncertain or irreversible. At the same time, a larger assortment can reduce anticipated regret when it contains a clearly suitable option or allows the chooser to feel that relevant alternatives were considered. The direction of the emotion effect depends on preference clarity, comparability, outcome feedback, and the reversibility of selection. Gourville and Soman (2005) Gourville (2005) explicitly connected certain nonalignable assortments to effort and regret.
Affective measurement needs temporal precision. Anticipated regret before choice, immediate post-choice confidence, and experienced regret after outcome feedback are different constructs. Consumers who defer may report relief instead of regret, whereas a quick choice can generate later dissatisfaction after new information appears. Physiological or neural indicators of arousal cannot by themselves establish the subjective meaning of that arousal. A strong experiment should separate psychological states, decision behavior, and objective consequences, while respecting the possibility that several causal paths operate simultaneously.
Decision Goals: Accuracy, Speed, and Justifiability
Decision goals are among the moderators supported by the 2015 Chernev et al. synthesis Chernev (2015). A person seeking the highest-quality option may welcome extra alternatives, because each new item could improve the best available match. Someone seeking rapid completion may experience the same expansion as an additional burden. A buyer seeking a publicly defensible choice may prefer familiar brands or products with simple rationales. The psychological meaning of a large assortment thus changes with the goal pursued during that episode.
Sela, Berger, and Liu (2009) Sela (2009) showed that assortment size can influence which kind of option is selected: greater difficulty in large assortments was associated in their studies with choices that were easier to justify, with circumstances capable of reversing that pattern. The result concerns the content of the selected option, not only the probability of choosing. It suggests that effort and justification can redirect preferences toward options for which reasons are readily available. These reasons may be socially learned, goal-dependent, or prompted by the display itself.
Goal manipulations offer an important causal test. Participants can be assigned accuracy-oriented, effort-minimizing, or justification-oriented instructions while encountering structurally matched assortments. The resulting interaction would test whether a proposed mechanism is conditional on goal. But instructions can themselves change demand characteristics and attention. Researchers should supplement self-reports of goals with consequential incentives, choices repeated across tasks, and behavioral traces. For practical applications, optimizing click-through or conversion is not equivalent to optimizing informed consumer outcomes, long-term satisfaction, or autonomy.
Satisfaction, Confidence, Regret, and Switching
Choice overload has often been operationalized through satisfaction, confidence, regret, deferral, and intended switching. These indicators reflect partly distinct psychological processes. Confidence is a judgment about the adequacy or correctness of a decision. Satisfaction can refer to the choice experience, the chosen item, or its realized performance. Regret involves comparison with forgone alternatives or imagined alternatives. Switching requires a later opportunity to reconsider, and is influenced by new information and market conditions. Treating these outcomes as a single latent dimension requires psychometric support in a specified population and decision domain.
Chernev et al. (2015) Chernev (2015) found comparable usefulness of several outcome families for detecting choice-overload patterns in their synthesis. That result supports their use as indicators in the reviewed literature; it does not prove that each is a perfect substitute in mechanistic explanation. A consumer can feel highly confident in selecting a dominated product because they lack knowledge, or can experience low confidence after an unusually conscientious assessment. Objective fit and subjective confidence can diverge. Accordingly, a rigorous outcome battery should include both experiential and performance measures.
Temporal ordering matters. A brief survey immediately after clicking a product assesses a different event from product retention or repeat purchasing months later. If an experiment compares large and small sets only on immediate satisfaction, it cannot establish effects on retention, welfare, or future choice skill. The same caveat applies to regret and switching intentions. Longitudinal designs can disentangle the experience of choosing, the utility of possessing the selected item, and subsequent market behavior. The predicted effect may change sign as the consumer learns or as more information becomes available.
Assortment Reduction, Missing Favorites, and Match Quality
Removing items can improve navigability while decreasing the chance that a particular consumer's preferred alternative remains available. This creates a central practical trade-off: a smaller set is easier to search, but it may provide poorer coverage of heterogeneous needs. Broniarczyk, Hoyer, and McAlister (1998) Broniarczyk (1998) studied consumer perceptions of grocery assortments under item reduction, emphasizing favorite-product availability and shelf-space cues. Their research underscores that the identity of removed items matters as much as the numerical amount of reduction.
Conversely, Oppewal and Koelemeijer (2005) Oppewal (2005) found in a choice experiment that assortment expansion improved evaluation across the conditions studied, with limited evidence for a generic favorite-available explanation. The coexistence of beneficial assortment enlargement and harmful overload effects is intelligible once perceived opportunity and commitment cost are modeled separately. Consumers may value access to numerous possibilities even when they have difficulty selecting one in a particular encounter.
A high-quality reduction policy therefore needs to preserve meaningful coverage while reducing redundant comparisons. It should attend to diversity of preferences across consumers, not just average popularity. Eliminating slow-selling variants can disproportionately burden people with uncommon requirements. Evaluations should report which items were removed, whether the remaining set spans key attribute combinations, how many shoppers lose access to a preferred option, and how navigational efficiency changes. An assortment is a distribution of opportunities across a population, with equity and accessibility consequences that aggregate conversion rates may conceal.
Variety Seeking, Consumption, and Repeated Choice
Choice sets may be used repeatedly rather than encountered in a single isolated decision. Consumers often seek variety over time to reduce satiation, explore new experiences, or accommodate changing occasions. Kahn (1995) Kahn (1995) integrated earlier research on variety seeking, highlighting the importance of motives and behavior across goods and services. A large assortment may be beneficial for a consumer with an established repertoire, even if selecting one item from a large unfamiliar menu is initially demanding.
Repeated exposure changes the psychological decision environment. Preferences become more defined, product attributes are learned, and previously costly comparisons can turn into simple recognition. Habit may reduce deliberation over recurring purchases, while evolving needs can restore exploration. The structure of an assortment can shape what consumers learn: category labels and systematic feature displays make comparison criteria discoverable; arbitrary rankings can foster reliance on salient items. Experiments that only involve first-time shoppers cannot establish whether initial overload persists over repeated decisions.
Consumption quantity is also a separate outcome. Kahn and Wansink (2004) Kahn (2004) reported effects of perceived variety on consumption, a mechanism that need not imply a difficult final product choice. Greater variety can increase engagement or consumption while increasing the effort of selecting a particular variant. This makes a multidimensional outcome framework essential in food, entertainment, learning, and service contexts. The appropriate welfare judgment depends on the specific goal and consequences of consumption, not merely on whether a larger display stimulates activity.
Age, Experience, Culture, and Individual Differences
The cognitive and motivational consequences of choice vary among individuals. Domain expertise, numeracy, working memory, processing preferences, prior knowledge, and decisional responsibility can alter how comparisons are performed. Some shoppers develop efficient category filters; others require explanatory assistance or prefer fewer simultaneous options. But demographic categories should not be treated as proxies for processing capability without direct measurement. Reed, Mikels, and Löckenhoff (2013) Reed (2013) investigated preference for choice across adulthood, making the age-related dimension a legitimate empirical question rather than a presumption.
Cultural and institutional contexts also shape the meaning of choosing. In some settings, a broad assortment signals freedom or market quality; in others, consumers may expect expert curation or place greater weight on collective responsibilities. Cross-national comparisons must establish equivalence of products, translated instructions, purchase constraints, and response measures. A category such as health care, education, or financial planning involves consequences and choice norms different from the selection of low-stakes consumer goods. Extrapolation from convenience samples of university students to all market participants is unwarranted without comparative evidence.
Individual differences are often measured using self-report scales for maximization, indecision, regret proneness, or need for cognition. Such measures can generate moderator hypotheses, but their associations with overload do not necessarily imply causal susceptibility. Effects may also reflect product familiarity, financial constraints, and task importance. The strongest designs estimate interactions between controlled assortment structure and directly measured capabilities or experimentally manipulated knowledge. Sampling must include people with different access needs and resource constraints if the results will guide universal product or public-service design.
The Neuroscience of Evaluating Choice Sets
Neural evidence offers a complementary perspective on the valuation of entire sets. Reutskaja, Lindner, Nagel, Andersen, and Camerer (2018) Reutskaja (2018) studied choice overload with behavior and functional brain imaging, reporting changes in signals associated with the value of choice sets in dorsal striatum and anterior cingulate cortex. The study is important because it examines possible neural correlates of the trade-off between the benefits of choice and the costs of searching among alternatives. It should be interpreted as evidence about an experimentally defined task and its associated neural signals.
The imaging approach does not create a simple biological cutoff after which all large assortments are harmful. Functional activation is influenced by task demands, expected value, conflict, learning, and other processes. Reverse inference from a region's activity to a single subjective state is unreliable without convergent behavioral manipulations and validated computational predictions. Neural measures can constrain models when linked to explicit hypotheses about set valuation, attention, or comparison cost, but they do not replace consumer outcomes.
A useful cross-level theory specifies how features of the external set alter information sampling and subjective set value, how these variables predict behavioral choice and deferral, and how neural signals covary with the latent computations. It should distinguish neural correlates of processing an assortment from neural mechanisms causing an overload response. Replication across decision domains and different measurement modalities is needed before treating a particular neural pattern as a general marker of assortment complexity.
Digital Commerce, Search Rankings, and Recommender Systems
Digital platforms alter the operational definition of an assortment. The database may contain thousands of products while a visitor sees only a small, dynamically ordered subset. Filters, recommendations, sponsored placements, infinite scrolling, and personalized ranking jointly determine what enters the effective choice set. The consumer's choice environment is therefore produced through an interaction between catalog structure and software-mediated exposure. Measuring database size without recording displayed positions, search terms, filtering, and impressions misidentifies the actual treatment experienced by the chooser.
A field experimental study by Long and colleagues (2024, journal issue 2025) Long (2025) examined the choice-overload effect in online recommender systems and highlighted search behavior as an important pathway. The evidence moves beyond laboratory product arrays to naturally occurring platform decisions, while the specific platform design and user population remain necessary conditions for interpretation. A related field experiment by Nobel (2024) Nobel (2024), involving 23,165 consumers at two online retailers, found no indication that a substantially modified recommendation order that prioritized a smaller group of highly attractive products worsened measured choice process or outcomes. The juxtaposition demonstrates that algorithmic assistance can matter without implying a universally optimal ranking rule.
A rigorous online experiment should identify the assignment unit, exposure, treatment uptake, rank position, user return frequency, and relevant outcomes. If ranking changes both what is displayed and when a visitor stops searching, the estimand is the total effect of that policy, not an isolated effect of item count. Instrumentation should preserve privacy and distinguish unique products from duplicated recommendations. Comparisons across systems must account for search history, recommendation quality, and the incentives of the platform. A personalized shortlist can reduce navigation cost while narrowing exposure to unfamiliar possibilities; both consequences deserve measurement.
Platform Assortments, Retention, and Longitudinal Trade-Offs
Online stores and services often optimize metrics that unfold over different timescales. A broader assortment may attract new customers seeking diverse possibilities while increasing the costs of making repeated selections. Natan (2024; volume issued 2025) Natan (2025) investigated large assortments on an online food-delivery platform and reported that expansion increased acquisition while reducing consumption frequency among continuing customers. The study used a model of costly attention and limited-information choice to interpret the observed pattern. The population and market studied matter, and modeled counterfactuals should be distinguished from direct experimental assignments.
The finding illustrates why a retailer's preferred outcome must be explicit. A change that raises browsing or initial acquisition can simultaneously reduce transaction frequency, long-term satisfaction, or the ability to find a suitable item quickly. The choice architecture may serve distinct populations differently: new users need discovery, returning users may value fast retrieval, and people with specific requirements may need full catalog access. Aggregate platform performance can conceal these divergent effects. Longitudinal estimation should report the composition of user cohorts and account for selective retention.
The scientific synthesis is that an assortment has both an option value and a processing cost over time. The option value increases when more alternatives match heterogeneous or changing preferences. The processing cost depends on how frequently a consumer must reidentify the relevant subset and resolve ambiguous comparisons. Effective personalization and saved preferences may reduce repeated costs, while algorithmic changes or unstable rankings may recreate them. These propositions can be evaluated with controlled rollouts, repeated-user experiments, and models that separately estimate discovery and search cost.
Contemporary Evidence Reviews and What They Establish
The field's evidence base has expanded beyond the early jam experiments. A 2022 meta-analytic review of retail assortment size Sethuraman (2022) assembled 177 studies from 95 papers covering research published from 1970 through 2021. Its scope included perceptions, choice, and sales-related outcomes, showing that assortment effects can be beneficial and detrimental through different pathways. The review explicitly examined moderating conditions and online–offline differences, making the construct more comprehensive than a binary choice-overload hypothesis.
Jacob, Thomas, and Joseph (2024) Jacob (2024) surveyed 92 articles from approximately two decades of the literature and mapped theories, contexts, methods, and areas needing additional work. Dar (2025) Dar (2025) published a separate systematic review of 53 peer-reviewed articles spanning 2000–2023, highlighting antecedents, outcomes, and moderators. These reviews are valuable for mapping the diversity of empirical approaches, but their count of studies does not represent a single pooled sample with a common outcome definition. Differences in inclusion criteria, coding schemes, and temporal coverage limit comparisons of review conclusions.
The 2025 Mejía synthesis Mejía (2025) analyzed perceived assortment variety using 135 experimental studies from 61 papers and a cumulative reported sample of 66,638. It distinguished perceived numerosity and diversity, which are not reducible to the emotional or behavioral consequences of choosing. Together, these reviews support a plural account: item count, variety perception, comparative complexity, and final choice outcomes are related constructs with partially distinct causal determinants. A strong contemporary theory needs to make that pluralism operational rather than declare one generic overload effect.
Experimental Identification and Factorial Designs
A standard causal experiment randomly assigns participants to encounter a smaller or larger assortment while holding the task, incentives, product category, and outside option as constant as possible. Such experiments estimate the effect of the specific assortment contrast they implement. If expanding from six to twenty-four items also changes the range of quality, number of categories, or availability of a favorite, the treatment is bundled. Randomization secures a causal estimate of that bundle, but it cannot identify which dimension produced the effect. Mechanism identification requires orthogonal manipulations or sufficiently informative process measures.
A stronger factorial design can cross set size with attribute conflict, category organization, preference articulation, and the presence of no choice. Item identities should be sampled or counterbalanced, because a single collection may have unusually attractive or unattractive candidates. The experiment should record initial familiarity, expected product importance, task goal, and prior experience, and use preregistered hypotheses for major interactions. Multiple set sizes support tests of a dose–response relation and permit comparison between simple monotonic and inverted-U specifications. Researchers should not infer a unique optimum from a two-level contrast.
Measurement extends beyond a post-task rating. Investigators can collect clickstreams, eye movements, task duration, decision completion, subjective difficulty, recall of attributes, and delayed satisfaction. These measures have different missingness and reactivity risks: eye tracking constrains participants, clickstreams capture only recorded interactions, and subjective ratings can be shaped by demand characteristics. A convergent design uses behavioral data alongside self-report while defining each outcome before analysis. Where a consumer's goal is known independently, alignment with that goal can provide an additional criterion for decision quality.
Meta-Analysis, Heterogeneity, and Replication
The meta-analytic history is unusually instructive because the apparent disagreement between Scheibehenne et al. (2010) Scheibehenne (2010) and Chernev et al. (2015) Chernev (2015) reflects both the heterogeneity of experiments and differences in the target of inference. One synthesis emphasized the virtually zero overall mean across a heterogeneous sample; the later review identified moderator patterns and an adjusted aggregate effect. Neither finding eliminates the need to scrutinize the studies' definitions, manipulations, and populations. Outcome direction can be reversed depending on whether researchers code purchase probability or deferral, confidence or regret.
The correct unit of evidence also requires attention. A publication may contribute multiple conditions or outcomes drawn from overlapping participants. Treating these as independent observations can understate uncertainty. Research synthesis should use dependency-aware models, prespecified coding of outcome families, and sensitivity analyses for imputed correlations and missing effect sizes where raw data permit. Publication selection and selective outcome reporting can affect the apparent evidence distribution. Moderator findings obtained mainly by comparing different studies may reflect correlated contextual differences rather than direct causal effects of the moderators themselves.
Replication should distinguish exact reconstruction of an experiment from independent tests of a theoretical mechanism. A narrow reproduction may be informative about a particular display, participant pool, and decision setting, while a conceptual replication can test whether comparative complexity predicts deferral across novel categories and interfaces. The strongest evidence emerges when the underlying structure is measured explicitly, experiments are preregistered, data and materials are available for inspection where feasible, and predictions generalize out of the discovery sample. The burden of explanation lies in specifying when a particular size increase should help, harm, or leave the outcome unchanged.
Measurement Validity and Statistical Interpretation
A major validity problem is using the phrase choice overload to name both an experimentally manipulated cause and a self-reported consequence. Set size, attribute conflict, and information presentation are features of the decision environment; regret, confidence, difficulty, and deferral are responses. Calling both groups overload makes causal mechanisms circular. A defensible research model measures complexity independently of the outcome and tests whether it predicts that outcome under controlled variation. It also distinguishes overload-related cognitive effort from simply having a large assortment.
For quantitative analysis, the selection of an outcome scale determines the relevant effect measure. Purchase and deferral can be represented as proportions or event probabilities, while confidence and satisfaction often use ordinal ratings. Decision time commonly has a skewed distribution, and repeated exposure generates within-person dependence. Comparisons should report effect sizes, uncertainty intervals, assignment balance, missing-data handling, and the exact set-size contrast. Small statistically significant effects on clicks may not correspond to meaningful improvements in welfare; large subjective ratings may not translate into observed market behavior.
A second challenge is construct transportability. An assay of chocolate choice, a questionnaire on medical options, and a recommender-system field experiment may all be described as tests of overload, but the choice stakes, time horizon, information completeness, and permitted deferral differ. Validation requires examining whether proposed moderators and mediating constructs behave consistently in each setting. Studies should report both common core variables and domain-specific measures. A research monograph can integrate such studies conceptually while preserving the boundaries of comparability.
Mechanistic Causal Models and Competing Explanations
Several mechanisms can generate a negative effect of assortment expansion. Expanded sets can raise comparison effort; create more difficult trade-offs; increase preference uncertainty; make forgone alternatives more salient; complicate search; or make the decision feel less justifiable. The mechanisms can also interact. A large nonalignable assortment may increase anticipated regret partly because the individual cannot establish a clear winner. In other settings, more options increase the chance of a close match and reduce conflict. These explanations make different predictions and should be discriminated experimentally.
One family of models emphasizes cognitive cost: the consumer decides when expected improvement from further search is outweighed by the processing burden. Another emphasizes preference construction: attributes and trade-offs acquire weights as the set is inspected. A third concerns affect and anticipated regret, and a fourth concerns information architecture and how alternative profiles become perceptually organized. Bettman et al. (1998) Bettman (1998), Gourville and Soman (2005) Gourville (2005), and Chernev et al. (2015) Chernev (2015) provide different empirical and theoretical anchors for these explanatory families.
Their claims can be separated through targeted intervention. Reducing the number of visible comparisons while keeping access to the same products tests search structure. Providing an ideal-point prompt tests preference specification. Making selection reversible targets commitment-related regret. Varying decision stakes or an explicit effort-minimization goal tests the objective pursued by the chooser. Mediator analysis is credible only if proposed mediator measurements precede outcomes and the causal assumptions are stated. A reduction in measured stress after a redesign is insufficient, on its own, to establish which structural mechanism operated.
Artificial Mediation and Human–Artificial Choice Relations
Artificial systems increasingly participate in the construction of the choice set itself. They can retrieve products, infer preferences, rank alternatives, generate explanations, suggest filters, and maintain user-specific decision histories. In such settings, the psychologically effective assortment is co-produced by the user, the platform, and an Artificial decision-support process. This expands the scientific problem from assortment display to the design of delegated attention and adaptive preference elicitation. The relevant empirical literature on recommender systems and online choice is a starting point, including Long et al. (2024) Long (2025) and Nobel (2024) Nobel (2024).
An assistant can reduce apparent complexity by presenting a shortlist, but the shortlist depends on the objective encoded by the system. Optimization for probability of purchase, advertising revenue, claimed similarity, long-term suitability, and informational diversity may produce different rankings. The user may experience lower cognitive effort while seeing a narrower set of opportunities. Scientific evaluation should therefore track exposure diversity, relevance to user-declared goals, explanations, reliance, correction, and subsequent satisfaction. The absence of visible effort is not by itself a sufficient measure of decision quality.
The Artificial Era perspective motivates testable propositions about human–Artificial cognitive partnership. A system that elicits user priorities before filtering should, in principle, be more likely to reduce preference uncertainty than one that simply hides alternatives; an interface that explains why options were excluded may improve calibration of trust; and reversible, inspectable recommendations may reduce anticipated regret. These are proposed hypotheses, not established universal findings. They can be tested by randomized comparisons of recommendation transparency, editable preferences, set size, and outcome tracking. The same framework can inform research on Artificial decision policies without making assumptions about subjective experience in the system.
Ethics, Accessibility, and Consumer Welfare
Assortment design affects agency by structuring what people can discover and how easily they can compare alternatives. Reducing visual complexity can support users facing limited time, impaired access to information, or high-stakes decisions; restricting the underlying set can also remove important possibilities for minority preferences. Ethical assessment must therefore distinguish making an assortment navigable from narrowing substantive opportunity. The appropriate goals include informed choice, fair access, intelligible comparison, and the ability to revise decisions, alongside transaction metrics.
Manipulative choice architecture can take advantage of relative comparison effects. By arranging dominated or conspicuous alternatives, a seller may influence conditional shares without improving customer fit. Research on context-dependent choice Huber (1982) and the no-choice option Dhar (2003) establishes that presentation is behaviorally consequential. Governance questions include whether ranking is sponsored, whether exclusions are disclosed, whether consumers can inspect the full set, and whether a recommendation can be challenged. Empirical evaluations should record distributional effects, not merely mean conversion.
Consumer welfare cannot be inferred directly from assortment size. More options may enhance access; fewer displayed options may reduce processing cost; a tailored shortlist may increase confidence while constraining exploration. Meaningful assessments can include whether selected products satisfy previously expressed constraints, whether a return or switch becomes necessary, and how consumers evaluate the process after receiving additional information. Studies in health, financial, and public-service settings require context-specific ethical and professional evaluation because decision error may carry consequences beyond ordinary consumer dissatisfaction.
Critical Scientific Discussion
The evidence supports a conditional account of choice overload. Large assortments are neither uniformly harmful nor uniformly beneficial, and a mean effect can conceal strong context dependencies. The classic experiments established that increased opportunity can coexist with reduced willingness to choose. Later meta-analytic research established heterogeneity and highlighted preference uncertainty, task difficulty, comparative complexity, and decision goals. Assortment structure, categories, attribute conflict, and the availability of an outside option further alter the direction and magnitude of the response. Recent digital field evidence adds search algorithms, platform retention, and ranking as consequential parts of the effective choice environment.
Several limits continue to qualify causal claims. Experimental set-size manipulations often bundle changes in option count with changes in attribute coverage and subjective attractiveness. Self-reported difficulty has imperfect correspondence with objective decision quality. Meta-analyses combine heterogeneous task structures and outcomes; coding and dependency assumptions can influence conclusions. Short-term purchases are an incomplete proxy for lasting satisfaction. And online algorithms alter both the alternatives displayed and the probability that a person inspects them. A convincing explanation must locate the relevant causal pathway rather than substitute a generic overload label.
The strongest current research direction is explicit comparison of multiple structural dimensions within the same study. A model should define the accessible universe of options, the displayed effective set, the chooser's initial preferences, the costs of inspecting attributes, the opportunity to defer, and the outcomes over time. It should permit positive, negative, and null effects as theoretically informative results. This preserves the value of competing traditions while identifying where their predictions converge and where targeted experiments can discriminate among them.
Scientific Synthesis and Open Research Questions
The research reviewed here supports an integrative structure with five analytically distinct components. First, objective availability describes how many items and which feature combinations exist. Second, presentation and ranking determine what becomes perceptually and cognitively accessible. Third, preference and goal state determine the value of potential matches and the acceptable processing cost. Fourth, comparison and search processes transform exposure into a possible commitment. Fifth, outcomes include purchase, deferral, experienced effort, confidence, regret, consumption, and longer-term welfare. This organization is a synthesis of published findings and methodological requirements, rather than a claim of a newly established empirical law.
The components interact. Increasing availability may raise match probability but also create search work. A category label may increase perceived variety while making navigation easier. An explicit ideal point may change a difficult high-dimensional comparison into a rapid elimination procedure. The option to decline a purchase may express rational recognition of missing fit or a difficulty in resolving conflicts. A recommender can reduce the set inspected while altering which alternatives the consumer even knows exist. These relationships generate testable hypotheses only when the variables are operationalized and their counterfactual contrasts identified.
Important questions remain unresolved. How can perceived complexity be measured comparably across tasks while preserving domain meaning? What constitutes a generalizable estimate of the benefit from one additional option? When does a shortlist improve welfare rather than merely conversion? How do repeated decisions change the trade-off between discovery and friction? Which moderator interactions replicate in older adults, across cultures, and under resource constraints? Can causal models separate algorithmic ranking effects from the effects of assortment composition? How can researchers evaluate long-term consequences when alternative products have unobserved qualities? Progress requires multi-site replication, behavioral trace data, preregistered factorial experiments, and transparent synthesis of null as well as positive findings.
Conclusion
Choice-set complexity and assortment effects describe a fundamental relationship between the organization of available alternatives and the psychological work of deciding. An assortment is experienced through its composition, comparative structure, presentation, and relevance to the chooser's goals. Its size has consequences because the addition or removal of options can change both the opportunity to find a suitable alternative and the effort required to identify that alternative. The research tradition developed from bounded rationality and context effects through experimental demonstrations of choice overload to contemporary moderator-based meta-analysis and digital field experiments.
The central conclusion is that number alone cannot serve as a sufficient scientific explanation of decision difficulty. Comparative conflict, preference uncertainty, task goals, information organization, decision reversibility, and the availability of no choice jointly shape behavioral and experiential outcomes. Evidence from traditional consumer experiments, retail research, and algorithmically mediated marketplaces converges on the need for mechanism-specific and context-sensitive models. Research should compare alternative explanations, preserve distinctions between behavior and welfare, and measure effects beyond immediate selection.
In the Artificial Era, research on assortment complexity also becomes research on the delegation and organization of attention. Human–Artificial decision partnerships can restructure the effective choice set while retaining the consumer's ability to inspect, revise, and authorize decisions. The enduring psychological problem remains the alignment of opportunities, cognitive processes, and consequential action. A fundamental science of assortments explains how that alignment is produced, where it breaks down, and how its success can be established empirically.
References
- Bettman, J. R., Luce, M. F., & Payne, J. W. (1998). Constructive consumer choice processes. Journal of Consumer Research, 25(3), 187–217. https://doi.org/10.1086/209535
- Broniarczyk, S. M., Hoyer, W. D., & McAlister, L. (1998). Consumers’ perceptions of the assortment offered in a grocery category: The impact of item reduction. Journal of Marketing Research, 35(2), 166–176. https://doi.org/10.1177/002224379803500203
- Chernev, A. (2003). When more is less and less is more: The role of ideal point availability and assortment in consumer choice. Journal of Consumer Research, 30(2), 170–183. https://doi.org/10.1086/376808
- Chernev, A., Böckenholt, U., & Goodman, J. (2015). Choice overload: A conceptual review and meta-analysis. Journal of Consumer Psychology, 25(2), 333–358. https://doi.org/10.1016/j.jcps.2014.08.002
- Dar, A. R., & Gul, M. (2025). The “less is better” paradox and consumer behaviour: A systematic review of choice overload and its marketing implications. Qualitative Market Research: An International Journal, 28(1), 122–145. https://doi.org/10.1108/QMR-01-2024-0006
- Dhar, R. (1997). Consumer preference for a no-choice option. Journal of Consumer Research, 24(2), 215–231. https://doi.org/10.1086/209506
- Dhar, R., & Simonson, I. (2003). The effect of forced choice on choice. Journal of Marketing Research, 40(2), 146–160. https://doi.org/10.1509/jmkr.40.2.146.19229
- Fasolo, B., McClelland, G. H., & Todd, P. M. (2007). Escaping the tyranny of choice: When fewer attributes make choice easier. Marketing Theory, 7(1), 13–26. https://doi.org/10.1177/1470593107073842
- Gourville, J. T., & Soman, D. (2005). Overchoice and assortment type: When and why variety backfires. Marketing Science, 24(3), 382–395. https://doi.org/10.1287/mksc.1040.0109
- Huber, J., Payne, J. W., & Puto, C. (1982). Adding asymmetrically dominated alternatives: Violations of regularity and the similarity hypothesis. Journal of Consumer Research, 9(1), 90–98. https://doi.org/10.1086/208899
- Iyengar, S. S., & Lepper, M. R. (2000). When choice is demotivating: Can one desire too much of a good thing? Journal of Personality and Social Psychology, 79(6), 995–1006. https://doi.org/10.1037/0022-3514.79.6.995
- Jacob, B. M., Thomas, S., & Joseph, J. (2024). Over two decades of research on choice overload: An overview and research agenda. International Journal of Consumer Studies, 48(2), Article e13029. https://doi.org/10.1111/ijcs.13029
- Kahn, B. E. (1995). Consumer variety-seeking among goods and services: An integrative review. Journal of Retailing and Consumer Services, 2(3), 139–148. https://doi.org/10.1016/0969-6989(95)00038-0
- Kahn, B. E., & Wansink, B. (2004). The influence of assortment structure on perceived variety and consumption quantities. Journal of Consumer Research, 30(4), 519–533. https://doi.org/10.1086/380286
- Long, X., Sun, J., Dai, H., Zhang, D., Zhang, J., Chen, Y., Hu, H., & Zhao, B. (2025). The choice overload effect in online recommender systems. Manufacturing & Service Operations Management, 27(1), 249–268. https://doi.org/10.1287/msom.2022.0659
- Luce, M. F., Bettman, J. R., & Payne, J. W. (1997). Choice processing in emotionally difficult decisions. Journal of Experimental Psychology: Learning, Memory, and Cognition, 23(2), 384–405. https://doi.org/10.1037/0278-7393.23.2.384
- Mejía, V. D. (2025). Assortment management strategies that people see: Insights from a meta-analysis of experimental research on perceived assortment variety. Journal of the Academy of Marketing Science, 53(4), 1230–1259. https://doi.org/10.1007/s11747-024-01075-0
- Mogilner, C., Rudnick, T., & Iyengar, S. S. (2008). The mere categorization effect: How the presence of categories increases choosers’ perceptions of assortment variety and outcome satisfaction. Journal of Consumer Research, 35(2), 202–215. https://doi.org/10.1086/588698
- Natan, O. R. (2025). Choice frictions in large assortments. Marketing Science, 44(3), 593–625. https://doi.org/10.1287/mksc.2023.0415
- Nobel, N. (2024). Recommender systems: Friend (of choice) or foe? A large-scale field experiment in online shopping platforms. Decision, 11(4), 450–467. https://doi.org/10.1037/dec0000236
- Oppewal, H., & Koelemeijer, K. (2005). More choice is better: Effects of assortment size and composition on assortment evaluation. International Journal of Research in Marketing, 22(1), 45–60. https://doi.org/10.1016/j.ijresmar.2004.03.002
- Payne, J. W., Bettman, J. R., & Johnson, E. J. (1993). The adaptive decision maker. Cambridge University Press. https://doi.org/10.1017/CBO9781139173933
- Reed, A. E., Mikels, J. A., & Löckenhoff, C. E. (2013). Preferences for choice across adulthood: Age trajectories and potential mechanisms. Psychology and Aging, 28(3), 625–632. https://doi.org/10.1037/a0031399
- Reutskaja, E., Lindner, A., Nagel, R., Andersen, R. A., & Camerer, C. F. (2018). Choice overload reduces neural signatures of choice set value in dorsal striatum and anterior cingulate cortex. Nature Human Behaviour, 2(12), 925–935. https://doi.org/10.1038/s41562-018-0440-2
- Scheibehenne, B., Greifeneder, R., & Todd, P. M. (2010). Can there ever be too many options? A meta-analytic review of choice overload. Journal of Consumer Research, 37(3), 409–425. https://doi.org/10.1086/651235
- Sela, A., Berger, J., & Liu, W. (2009). Variety, vice, and virtue: How assortment size influences option choice. Journal of Consumer Research, 35(6), 941–951. https://doi.org/10.1086/593692
- Sethuraman, R., Gázquez-Abad, J. C., & Martínez-López, F. J. (2022). The effect of retail assortment size on perceptions, choice, and sales: Review and research directions. Journal of Retailing, 98(1), 24–45. https://doi.org/10.1016/j.jretai.2022.01.001
- Shah, A. M., & Wolford, G. (2007). Buying behavior as a function of parametric variation of number of choices. Psychological Science, 18(5), 369–370. https://doi.org/10.1111/j.1467-9280.2007.01906.x
- Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of Economics, 69(1), 99–118. https://doi.org/10.2307/1884852
- Tversky, A., & Shafir, E. (1992). Choice under conflict: The dynamics of deferred decision. Psychological Science, 3(6), 358–361. https://doi.org/10.1111/j.1467-9280.1992.tb00047.x