So You Want To Make a Hit Song?
Reviewed by Jim Yoon
June 16, 2014
“So You Want to Make A Hit Song: Utilizing multiple regressions to determine predictor variables in music” by Paul Asercion tries to find out if there are factors that make certain songs more likely to become hits than others. He explains that due to the music industry becoming one of the biggest and most lucrative in the world, it is worthwhile to see if there are “predictor variables” inherent in a song that can increase its chance of becoming popular among the general population.
He goes on to note the various methods that have been used to this point to measure a song’s popularity. These include the callout method—when a record label or radio station sends out the song and receives feedback from a target audience—and Mscore—how often a listener will change the radio station when the specific song is played. However, these methods only evaluate a song after it has been made. Asercion proposes to use regressions based on data from Ellis & Engelhardt’s “Visualizing a Hit” project, the Billboard Top 50 Radio Airplay Charts, and Digital Sales Charts from 2010-2012. He constructs variables forenergy, “liveness”, “speechiness”, “acousticness”, valence, and “danceability”.
Asercion does note that there are possible errors within the regression and the resulting findings, such as not considering repetition of lyrics and disregarding influence from external variables, such as live shows, in dictating a song’s popularity. Although Asercion acknowledges that there is no singular magical formula for making a song a hit, he does suggest some guidelines for different genres for music—for example, for pop songs for the radio, he recommends making it the key of A#; for rock songs, one would be well-advised to steer clear of the key of C#, and to use common time.
To view the full article, click here!