Why Can More Choice Sometimes Make Choosing Harder?

Geoffrey Chen

Evidence status: conditional support Large option sets can increase deferral, regret or dissatisfaction under some conditions, but meta-analysis does not support the claim that more choice is invariably worse. Task difficulty, option complexity, preference uncertainty and an effort-minimising goal are important moderators. This is a version 0.1 research draft.

How Everyday Psychology Takes Shape · Season One: Why Do We Think This Way? · Article 4

A streaming service offers more films than a video store once held. Thirty minutes later, six trailers have played and no film has begun. More colours of a familiar household product may be pleasant. The same number of retirement investment options can produce delay.

The “paradox of choice” has become a common account of modern life: more options create anxiety and leave us less satisfied. The claim has intuitive appeal and some classic experimental support. Treated as a universal law, it neglects an equally important fact. More options often increase the chance of finding a good match, and people voluntarily seek varied assortments.

The useful question is not how many options a human mind can tolerate. It is how a choice task is formed.

Six jams and twenty-four jams

Sheena Iyengar and Mark Lepper’s 2000 paper included the famous jam field study. Supermarket customers encountered a large or small display. The larger assortment attracted more people to stop, while customers exposed to the smaller assortment were more likely to purchase. Other studies in the paper used chocolates and optional coursework, reporting advantages of limited sets in motivation, satisfaction or performance.

The work genuinely challenged the assumption that more is always better. Popular retellings later turned a particular field condition into a fixed ratio, as though twenty-four options automatically cause paralysis. Purchasing percentages can depend on display, traffic, familiarity and how field observations are implemented. One study cannot be carried directly into medical consent, career choice or democratic rights.

Failures to reproduce the pattern and opposite findings accumulated. Some experiments found that larger assortments improved choice or satisfaction; others found no difference. The scientific task became explaining variation, not deciding which absolute slogan was true.

Why two meta-analyses seem to disagree

In 2010, Benjamin Scheibehenne, Rainer Greifeneder and Peter Todd synthesised 50 published and unpublished experiments, containing 63 conditions and 5,036 participants. The mean effect was virtually zero, with considerable variation between studies. Their analysis did not find a sufficient set of conditions that reliably predicted overload.

This does not establish that overload observed in every experiment was illusory. A near-zero average may contain negative effects in some circumstances and positive effects in others. What it undermines is an unconditional general effect of assortment size.

In 2015, Alexander Chernev, Ulf Böckenholt and Joseph Goodman analysed 99 observations involving 7,202 participants and organised moderators into a conceptual account. They identified four conditions facilitating overload: high choice-set complexity, difficult decision tasks, uncertain preferences, and a prominent goal of minimising effort. Once moderators were considered, an overall overload effect emerged.

The contrast is more informative than declaring that one paper proved the effect and the other disproved it. The first stresses a near-zero average and heterogeneity. The second attempts to explain that heterogeneity through task structure. Conditional support is the appropriate evidence label.

Option count is not task complexity

Imagine twenty pens differing only in colour. If blue is required, selection is easy. Now imagine six insurance policies varying in exclusions, waiting periods, excesses and provider networks. The smaller set creates the more demanding cognitive task.

Complexity depends on how many attributes matter, whether they can be compared, and whether one option clearly dominates. Preference formation is also critical. A camera expert may value dozens of models. A beginner does not yet know which properties deserve weight, so additional options add differences that first need to be learned.

Goals change the experience. A person exploring the market gains information from variety. Someone trying to complete a routine purchase quickly experiences the same assortment as friction. “Too much choice” often combines quantity, poor organisation and an absence of evaluative criteria.

Choice also has a time structure. A large unfiltered set at the first stage is different from a broad catalogue that can be narrowed using meaningful categories. Interface design can make the same numerical set manageable or exhausting.

Not choosing can be a competent decision

Studies use deferral, failure to purchase, lower satisfaction or greater regret as overload outcomes. These are not interchangeable in meaning. When a decision is consequential and information is insufficient, delay may be more rational than rapid commitment. For a trivial product, exit may conserve attention.

Satisfaction includes counterfactual comparison. A large set keeps unchosen alternatives visible, potentially increasing regret. It can also provide a better objective match. A satisfaction rating alone cannot establish whether the assortment improved welfare.

Designers who remove important options “for psychological ease” can restrict autonomy. In retirement, health and public services, simplification should preserve access to the full set and meaningful opt-out. The institution should not quietly decide what users are allowed to regard as relevant.

The cost of choice can also be uneven. People with time, expertise and language support can use a complex menu; others face the same formal options without equivalent capacity. Calling the menu free choice can conceal the resources required to navigate it.

Is recommendation help or substitution?

Ranking, filtering and defaults turn a large set into a manageable path. A recommender uses past behaviour to reduce candidates. It may also narrow exploration and embed commercial goals in the phrase “best for you”.

Good choice architecture does not pretend to be neutral. It explains the basis of ordering, lets users alter filters, and preserves access to the complete set and an easy way to reverse a default. For a high-stakes choice, the system should help someone understand attributes rather than merely output a recommended answer.

AI can compare options, but when a user has not articulated goals, the model fills the gap with default values. Efficient calculation does not mean a preference has already formed. Some choice difficulty comes from not yet knowing what trade-off one is willing to make.

An assistant can be most useful when it exposes the question behind the menu: which differences would actually change the decision? That is a form of preference clarification, not simply option reduction.

Reducing avoidable difficulty

Identify a few non-negotiable conditions before looking at a large assortment. This is usually more stable than inventing criteria while moving among options. Narrow in stages: remove ineligible options first, then compare a small set closely. For reversible, low-consequence choices, a time limit can prevent the search for perfection from becoming an unnecessary project.

High-consequence decisions should not be optimised only for speed. Staged information, qualified advice, a written comparison and a cooling-off period may be more suitable than “trust your intuition and pick”.

If a platform encourages endless scrolling, the problem may not be personal capacity. The set has no natural boundary, and commercial design has weakened stopping rules. Changing the environment can be more effective than demanding stronger willpower.

Simplification should remove repetition and obscure presentation before it removes substantive alternatives. Otherwise, a psychological finding becomes a reason to reduce rights.

Does freedom require unlimited options?

Option count and autonomy are not synonyms. A long menu of functionally identical alternatives provides little freedom. A set that people cannot understand or afford may not expand effective agency either. Autonomy requires relevant alternatives, intelligible information, time to form a preference, and space to decline.

The version 0.1 conclusion is that more choice sometimes makes choosing harder, particularly when alternatives are complex, the task is difficult, preferences are uncertain and the decision-maker wants to minimise effort. The evidence does not support one universal optimal number. Practical improvement should remove meaningless complexity, provide staged comparison and transparent defaults, while preserving differences that matter. What people usually need is neither the smallest set nor the largest menu, but a structure in which a preference can become clearer.

Primary research sources

Series navigation: How Everyday Psychology Takes Shape — Season One overview

Discover more from Geoffrey Chen

Subscribe to get the latest posts sent to your email.