Cognitive biases
The bandwagon effect: why we choose what others have already chosen
Tom is putting together music for a road trip. His app has a section called “New Bands”: songs he has never heard. Under each one it says how many people have already picked it. One song has 1,240, the one next to it has 12. Tom glances at the descriptions and, almost without thinking, plays the one with 1,240. If that many people picked it, it must be good.
This is the bandwagon effect: we more often choose what many others have already chosen, and we choose it precisely because it is popular. Sometimes that is sensible: popularity can be a useful clue. But it can also mislead, because “popular” does not always mean “best.”
Below, we look at what scientists’ experiments have shown, why we are drawn to the majority, and when it is worth choosing for yourself. In the middle of the article you will take Tom’s place and put together your own playlists.
Contents
In short: how the bandwagon effect works
What the bandwagon effect is
The name comes from the word “bandwagon”: a wagon carrying a band in a parade. Later the word came to mean any movement that many people jump on. In a 1950 paper, the economist Harvey Leibenstein described this effect in buying: people buy a product more because others are buying it. Today the term covers any shift of choice toward a popular option when the cause is the popularity itself. Scientists are still working out what exactly should count as this effect and what counts as similar phenomena.
The most striking experiment was run by the researchers Matthew Salganik, Peter Dodds, and Duncan Watts. In a 2006 study they built a music website. It had songs by unknown bands, and participants chose what to listen to and download. Some saw only the names of bands and songs. Others also saw how many times each song had already been downloaded before them. The second group was split into eight separate “worlds.” Each world had its own participants and its own counters, but the songs were the same. A person could take part only once, and participants could not share opinions with each other. So the only way to learn about the choices of others was the numbers on the screen.
The result: where participants saw the counters, popular songs became even more popular, and unpopular ones became even less noticeable. And in different worlds, different songs became hits. When the songs were also arranged in a list from most popular down, the skew got stronger. Quality mattered too: the best songs rarely flopped, and the worst rarely made it to the top. But everything in between was possible.
In a follow-up study the scientists went further and flipped the counters: songs from the bottom of the list were shown as the most popular, and the other way around. For most songs, the false popularity became real over time. But the very best songs eventually won their place back.
Try it yourself: Tom’s playlists
Now you are in Tom’s place. Here is what will happen.
Step 1. You will put together five small playlists: for the morning, for the road, for work, for workouts, and for the evening. Each has four unfamiliar songs with a short description. Under each song is the number “Already picked.” You choose the one you would play first. We don’t play any music: as in the scientists’ experiment, you have to choose by title and description.
Step 2. You use a slider to mark how much, in your opinion, the numbers influenced your choice.
Step 3. We will show what another player would have seen, and count how often you picked the song with the biggest number.
The songs are the same for all players, but the numbers under them are different for each: the same song may show different numbers to different players. We tell you this in advance to be fair, though it may make the numbers pull less. Where the numbers come from will become clear at the end.
What usually happens. If the numbers had no influence, the song with the biggest number would end up in your picks in about one playlist out of four. Over five playlists, that is about once. The scientists’ experiments show something different: when people see what others have chosen, the popular becomes even more popular.
What your result can and cannot show. Five playlists is not many. If you picked the song with the biggest number three times or more, that is a good hint: by chance this happens to about one person in ten. But it is not a test. Scientists see the effect when they compare thousands of people who were shown different numbers.
When it is not a mistake. Popularity often really does mean something. If a café has a line and the one next door is empty, the line is a useful clue. The mistake begins where the number is the only thing you know about an option, and you don’t know where it came from. In the game, the numbers say nothing about the songs. That is the whole trick.
Why we are drawn to the majority
Picture a crowd at a shop window. You can’t see what is in it, but you step up to look too. This is an image, but it conveys the main point well: other people’s choices work like an arrow saying “this way.”
Others probably know something. When you have little information of your own, it is sensible to lean on other people’s choices. Economists described how this turns into a chain. Each next person looks at those who chose before them and repeats their choice, even if they themselves thought differently. For an individual, this can be the right decision. But such a shared craze is fragile: it rests not on everyone’s knowledge but on everyone watching each other. So it can pass just as quickly.
We want to be with everyone else. A song everyone listens to is one you can talk about with friends. A movie everyone is discussing is scary to miss. Popularity attracts not only as a hint that “this is good,” but also as a chance to be part of the shared conversation.
We don’t notice it. In one study, people were told why to save energy, and were given different arguments. The argument “other people do this” changed behavior the most. Yet the participants themselves rated it the least convincing. The influence of the majority is easy to underestimate: from the inside, it can feel like your own taste.
Where it shows up
Likes and ratings. On one news website where users vote on comments, scientists ran a large experiment. Some new comments were randomly given an upvote right at the start, and others a downvote. After a gifted upvote, people were also more likely to upvote themselves. As a result, the rating of such comments was on average about a quarter higher than that of similar comments without the gift. A random downvote, on the other hand, people often corrected. A gifted plus snowballed into success; a gifted minus did not.
Shopping. Labels such as “bestseller,” “customers’ choice,” and “2,000 people have already bought this” work like the counters in the song experiment. Sometimes real buyer experience stands behind them. Sometimes it is just that the product was the first to land in a prominent spot.
Elections. People often say that polls before an election pull votes toward the leader. But scientists still disagree even about what exactly should count as such an effect in politics. So phrases like “polls decide elections” should be treated with caution.
How to choose for yourself
The bandwagon effect is one of the biases on the cognitive bias map, in the section on decisions. Next to it is the anchoring effect: there too, a random number on the screen quietly shifts our answer.
Frequently asked questions
Is this the same as conformity?
Close, but not the same. Conformity usually means giving in to group pressure. The classic example is the experiments of the psychologist Solomon Asch. In a simple task about the lengths of lines, one person found themselves up against a unanimous majority that answered differently. Such experiments were later repeated many times in different countries. With the bandwagon effect there may be no pressure at all. Nobody is watching you, nobody is arguing. You just see a number and choose in the same direction.
Do polls before an election pull votes toward the leader?
People often say so. But scientists still disagree even about what exactly should count as such an effect in politics. So phrases like “polls decide elections” should be treated with caution.
