Bracketology: What the data says about why people love an underdog

Every year, at least a few underdogs win games; there is built-in difficulty in accurately predicting which teams and when. And while science can shed light on when and why people predict that an upset will occur, the one thing it still has not cracked is the code to filling out a perfect bracket.

People love an underdog for many reasons, even if the data tells them the underdog is unlikely to come out victorious. Why?

The answer to this question is somewhat baked into an annual rite of Spring undertaken by millions of people: filling out brackets for the NCAA Men’s Basketball tournament.

NCAA “Selection Sunday” is on St. Patrick’s Day this year, March 17. That’s when the NCAA will finalize the brackets for this year’s “March Madness” basketball tournament. The next morning, millions will download and print out those brackets and fill them out, all in the hopes of predicting the “Final Four,” and for some winning a few dollars.

The bracket is fairly simple; 64 men’s college basketball teams are divided into four geographical areas, and each region’s games are structured based on the team's rankings (No. 1 vs. No. 16, No. 2 vs. No. 15, etc.). Each game is an elimination game — the winner advances to the next round and the loser goes home. The tournament culminates with the “Final Four” this year in State Farm Stadium in Glendale, Ariz.

So, how do most people fill out their brackets?

The most simple approach, especially if you don’t follow the NCAA closely, is to consistently pick the better ranked team. Indeed, better ranked teams win more often, especially in the first round where big names like Duke and Gonzaga may play small schools that barely made the cut to get into the tournament. Some bracket websites like ESPN even give people the option to always pick the better ranked team by default.

But this is not what many people do. Instead, they pick the underdog to win, at least for some games.

This is partly due to the fact it can be a lot more exciting to take a risk and bet on a low probability upset. In fact, people have a general tendency to root for the underdog. Do you remember the first time you watched the motion picture “Rocky”? Didn’t you hope that Apollo Creed would lose to Rocky Balboa? The underdog has vicarious appeal.

Another reason people tend to root for the underdog is that it can be boring to always make the same choice. Decades of research  have shown that people change their decisions over time, even to less desirable outcomes, because they feel bored or tired of repetition. But they also do it for social reasons: to signal to themselves or to others that they are not the type of person who always does the same thing. Indeed, the desire for variety leads people to even change their decision strategies (for example, from choosing the cheapest product to a more expensive one) simply “for change’s sake.”

These are all strong psychological reasons to predict upsets at some point in a March Madness bracket. But when are people more likely to predict an improbable outcome, like picking the underdog to win?

We have conducted scientific research forthcoming in the INFORMS journal Management Science and found that people are more likely to do so in predictions they happen to make later, versus earlier. If you apply this to “March Madness,” our findings suggest that people are more likely to pick the lower ranked team (say, the No. 10 seed over the No. 7 seed) if predicting that game’s outcome third, for example, rather than first in their bracket.

This is called the “prediction order effect” which happens because of how people routinely think about probabilities. They tend to treat separate events as if they are related even when they aren’t. If you flip a coin three times and it lands on heads the first two times, you may have the instinct that the coin “should” come up tails the third time, since getting three heads in a row feels unlikely. But in truth, each coin flip is statistically independent, or unrelated to the last; the probability of heads (or tails) is the same for the third flip as for any other — 50 percent.

The same process can apply to “March Madness” brackets. When predicting the winner for the first game, people are very likely to behave logically and pick the favorite. But as they move on to making later predictions, they don’t ignore their previous choices. Rather, they consider all of the games’ outcomes together and think “it’s called March Madness for a reason — at some point there has to be an upset!” Since they’ve already picked the favorite a few times, people then feel inclined to predict the underdog will win for games they happen to decide later.

The NCAA Men’s Basketball tournament has exploded in popularity for just this reason. Every year, at least a few underdogs win games; there is built-in difficulty in accurately predicting which teams and when. And while science can shed light on when and why people predict that an upset will occur, the one thing it still has not cracked is the code to filling out a perfect bracket (and it probably never will). 

Jackie Silverman is an assistant professor of marketing at the University of Delaware. She researches consumer psychology, with a special focus on the impact of behavioral tracking technology on consumer motivation and behavior, as well as judgements and consequences of repeated behaviors (i.e. “streaks”). Uri Barnea is an assistant professor of marketing at Bocconi University (Milan, Italy). He researches how consumers process information and form judgments.