Charting a Course for Success: A Statistical Analysis of Sociological Models in Popular Music Trends

Abstract

Literature about the success of popular music songs has traditionally focused on examining the relationship between song success and endogenous features such as sonic properties, record labels, and number of previously charting songs. However, in addition to their limited ability to explain chart performance, these studies also do not account for the popularity and evolution of certain music trends. As such, this thesis first compiles a novel dataset of endogenous song features and examines their relationship to chart performance on the Billboard Hot 100, demonstrating their limitations. Uniquely, our analysis highlights the increasing polarization of success: it has become more difficult for songs to climb the top of the chart, and those that do remain at the top positions for longer. Afterwards, we isolate two popular music trends--rap and boy band songs--and examine how sociological theory about trend popularity explains the growth and decline of these trends, using fitted Hawkes point processes models as evidence. We find significant clustering activity in both trends over time, but also that trends exhibited less clustering activity before their peaks in popularity compared to afterwards, driven by the prevalence of trend songs on the chart. Furthermore, we discover that highly successful songs induced stronger clustering effects in the form of additional highly successful songs in the near future, to more of an extent than moderately successful songs. Altogether, our findings present new perspectives regarding how record labels and artists strategize for chart success and more broadly on the evolution of trends in popular culture environments.