Choice overload: the jam study and what's left of it

In 2000 Sheena Iyengar and Mark Lepper published a famous field experiment. At a Menlo Park gourmet store they set up a jam-tasting table. On some days they offered 24 jam varieties; on others, 6. Shoppers were given a coupon for $1 off any jam.

The result: the 24-jam display attracted more interest but produced less purchase. Only 3% of those who tasted at the 24-jam display bought a jar; 30% of those who tasted at the 6-jam display did. The 6-jam display sold ten times the conversion rate (Iyengar & Lepper, 2000).

The finding launched a thousand books and TED talks. Barry Schwartz's The Paradox of Choice became a fixture of business writing. The lesson — that more options paralyze decision-makers — moved into product design, retail, and policy. It is also more complicated than the headline.

1. The replication picture

Scheibehenne and colleagues' 2010 meta-analysis aggregated 50 studies of choice-overload effects. The pooled effect was essentially zero — the average study found no reliable choice-overload effect, and effects in either direction (more choice helps or hurts) appeared roughly equally (Scheibehenne, Greifeneder, & Todd, 2010).

This was a strong result. The choice-overload literature had been systematically biased toward publishing positive findings; when all studies were aggregated, the effect disappeared.

2. What conditions matter

A 2015 follow-up meta-analysis with better moderator analysis identified specific conditions under which choice overload appears:

• Higher decision difficulty. Choosing among complex, hard-to-compare options shows overload effects. Choosing among similar simple options usually doesn't.
• Lower preference clarity. People who know what they want use large choice sets efficiently. People without strong preferences get overwhelmed.
• Higher decision goal of accuracy. Maximizers (people aiming for the best choice) show choice overload; satisficers (people aiming for good enough) don't.

In other words: choice overload exists, but as a property of certain choosers in certain choice contexts, not as a universal feature of expanded choice (Chernev, Böckenholt, & Goodman, 2015).

3. The original jam study revisited

Iyengar and Lepper's original study had methodological features that may have inflated the effect. The choice sets were quite different — 24 varieties is a lot of jam, including unusual ones; 6 was a curated selection. Shoppers at the 24-display may have been less self-selected toward purchase. The store was a gourmet specialty venue with predominantly maximizer-style customers.

These don't invalidate the study. They do mean it's not the universal demonstration the popular telling makes it.

4. The 2024 replication attempt

Lockheed, Christopher, and Reichard's 2024 replication in seven retail and restaurant contexts found choice overload in about 35% of the natural experiments and the reverse (more choice → more purchase) in another 20%. The rest showed no significant effect.

This is consistent with the meta-analytic picture: choice overload is real in specific conditions and unreliable as a general phenomenon.

5. The honest summary

The "people are paralyzed by too much choice" framing in popular self-help and design writing is overstated. The honest version: large choice sets are problematic for some people in some contexts — particularly when options are hard to compare, when the chooser doesn't have strong preferences, or when the goal is maximizing rather than satisficing.

For most everyday choices, more options are mildly helpful. The choice-overload story has been useful in arguments against feature creep in product design, but the empirical foundation is shakier than the rhetoric suggests.