VI
Mark học AIThe Jam Study · Sheena Iyengar and Mark Lepper, "When choice is demotivating" (2000)
Is more choice always better? The jam study said: sometimes too much makes people walk away empty-handed. Today's evidence says: that does happen, but only under certain conditions, and on average it barely shows up.
One tasting booth, two Saturdays, and a discount coupon to count who actually bought.
Three settings, one direction. That's why the paper was persuasive. But all three came from one team, with moderate samples.
10 common claims about the experiment, checked against current research. Tap each row to see why.
Chernev, Böckenholt and Goodman (2015) pooled 99 observations, 7,202 people: once these four conditions are taken into account, the effect is significant. This is the best current reading.
In Scheibehenne et al.'s (2010) meta-analysis, expertise or prior preferences tilted results toward "more choice helps". The 2015 analysis likewise found preference uncertainty makes overload stronger.
These are the original figures (Iyengar and Lepper, 2000), but from one supermarket, two Saturdays, and very few buyers (31 vs 4). Scheibehenne (2008) repeated it in a German supermarket and found no effect.
One Stanford class, 193 students, a difference just reaching significance. Plausible but not widely replicated.
134 people in the lab: 48% of the 6-chocolate group took chocolates instead of cash, vs 12% of the 30-chocolate group. Greifeneder (2008, cited in the 2010 meta-analysis) repeated it with chocolates and found no difference.
Iyengar, Huberman and Jiang (2004) looked at data on nearly 800,000 employees: plans with a handful of funds had higher participation than plans with ten or more. This is observational data, not an experiment, so other explanations are possible.
Redelmeier and Shafir (1995): doctors offered 1 drug prescribed it 72% of the time; offered 2 drugs, only 53%. But a randomized trial with 402 doctors (Altinger et al., 2025) found the opposite: with 2 or more options, more doctors switched.
A meta-analysis of 63 conditions, 50 experiments, 5,036 people: average effect D = 0.02, virtually zero (Scheibehenne et al., 2010). Newer studies found the effect less often than older ones.
This goes beyond the data. The same meta-analysis shows that more choice often helps, especially for customers who know what they want. Cutting blindly may remove exactly what customers need.
Scheibehenne et al. (2010) found no clear relationship between the size of the large set and the size of the effect. The number 6 was simply one experiment's display.
A clever experiment with a striking result that doesn't replicate reliably. Keep the question "when does more choice hurt?", drop the slogan "less is always more".
In 2010, Scheibehenne's team pooled every experiment they could find, published or not, and took the average: close to zero. They also found that published studies reported the effect more often, and that the newer the study, the less often it appeared. That's a familiar sign of a first result that was too good.
In 2015, Chernev's team asked a different question: not "how big on average?" but "when?". Once the four conditions entered the model, the effect showed clearly. The two results don't contradict each other: the average is near zero because most situations lack those conditions.
For people at work, the 2015 version is more useful: it tells you when to trim options and when a long list is fine.
Work skill: presenting options to a boss or client so they can decide. These are Mark's suggestions built on the four conditions above, not results tested in offices.
| Situation | What to give | Why |
|---|---|---|
| Busy boss, must decide in the meeting | 2–3 options + a recommendation | hard task, wants it done |
| New client, needs unclear | ask first, then 2–3 packages | unsure what they want |
| Engineer picking familiar parts | full list + filters | experts overload less |
| Sending 10 options "so the boss can choose" | avoid | pushes filtering onto the busiest person |
Every number in this page, the Facebook carousel and the captions comes from the studies below.