Extremeness Aversion

What it is

Simonson and Tversky stated the hypothesis this way: the attractiveness of an option is enhanced if it is intermediate in the choice set and diminished if it is extreme. It is the tendency behind the compromise effect, which Simonson had described in 1989.

Their camera study shows it at work. When 106 people chose between a Minolta X-370 at $169.99 and a Maxxum 3000i at $239.99, they split evenly. When 115 people also saw a Maxxum 7000i at $469.99, 57 per cent chose the 3000i, 22 per cent the cheapest model and 21 per cent the flagship. The 3000i had moved from one end of the range to the middle.

In effect

The most thorough test of the idea is a 2016 meta-analysis by Nico Neumann, Ulf Böckenholt and Ashish Sinha in the Journal of Consumer Psychology. It pooled 142 experimental observations and found that middle options were chosen significantly more often than the others. It also found the size of the effect varied by up to three times its average depending on how studies were designed: price-against-quality trade-offs, non-durable goods and comparisons of two-option with three-option sets tended to reduce it, while more trade-off dimensions, descriptions without numbers, high-quality extensions and utilitarian products tended to increase it.

Outside the laboratory, a 2012 working paper found an extremeness aversion bias in a sports betting market in which real money was at stake. A 2008 study of fast-food drink sizes found that removing the smallest cup or adding a larger one moved choices upwards.

What it does not say

It does not say people always avoid the ends of a range. A 2025 paper by Marissa Sharif, Elizabeth Webb and Sudeep Bhatia in the Journal of Consumer Research, with six studies and 9,377 participants, found that when people saw several extreme options, they became relatively more likely to choose an extreme one, because extremes then seemed more typical of the category.

Nor does it say the effect is as strong when money changes hands. A 2012 study in Marketing Letters found the related compromise effect in binding choices between real brands, but weaker than in hypothetical ones.

---