The Science of Hit Songs

Bjorn Carey

Date: 09 February 2006 Time: 09:01 AM ET

Photo taken by Karl-Erik Bennion. There are no usage restrictions for this photo.

When Ashlee Simpson tops the charts while a critically acclaimed ex-Beatle's album fails to crack the top 200, eyebrows go up in the marketing world.

So what makes a hit?

A new study reveals that we make our music purchases based partly on our perceived preferences of others.

Popularity contest

Researchers created an artificial "music market" of 14,341 participants drawn from a teen-interest Web site. Upon entering the study's Internet market, the participants were randomly, and unknowingly, assigned to either an "independent" group or a "social influence" group.

Participants could then browse through a collection of unknown songs by unknown bands.

In the independent condition, participants chose which songs to listen to based solely on the names of the bands and their songs. While listening to the song, they were asked to rate it from one star ("I hate it") to five stars ("I love it"). They were also given the option of downloading the song for keeps.

"This condition measured the quality of the songs and allowed us to see what outcome would result in the absence of social influence," said study co-author Matthew Salganik, a sociologist at Columbia University.

In the social influence group, participants were provided with the same song list, but could also see how many times each song had been downloaded.

Researchers found that popular songs were popular and unpopular songs were unpopular, regardless of their quality established by the other group. They also found that as a particular songs' popularity increased, participants selected it more often.

The upshot for markerters: social influence affects decision-making in a market.

This research is detailed in the Feb. 10 issue of the journal Science.

The Britney effect

The social-influence group was further divided into eight separate, non-interactive "worlds." Members of each world could not see the decisions of the other seven. The idea behind this was to observe multiple outcomes for the same songs and bands.

"If you look at Britney Spears, some people say she is really good. Others say she isn't good, she's just lucky," Salganik told LiveScience. "But by having just one argument, it's impossible to distinguish. However, if you have 10 worlds, and she's popular in all 10, then you can say she's actually good. But if she's only good in one, then you could say it was due to luck."

Although different songs were hits in each world, popularity was still the deciding factor, although the "best" songs never did very badly and the "worst" songs never did very well.

So what drives participants to choose low-quality songs over high-quality ones?

"People are faced with too many options, in this case 48 songs. Since you can't listen to all of them, a natural shortcut is to listen to what other people are listening to," Salganik said. "I think that's what happens in the real world where there's a tremendous overload of songs."

Alternatively, Salganik said, a desire for compatibility with others could drive the choice, since much of the pleasure from listening to music and reading books stems from discussing them with friends.

"If everybody is talking about 'Harry Potter,' you want to read it too," Salganik said.

The Columbia University Music Lab is here.