Boston University
Equal quality, unequal success? Social influence, identity, and dual cumulative advantage in a controlled music streaming experiment
Abstract
dc:description.abstractSignificant inequality in market success is a defining feature of cultural industries, yet scholars debate its origins—specifically, whether quality differences among artists are a prerequisite for such inequality, or whether social processes (e.g., through social influence) alone can generate comparable concentration even among artists of equal quality. Music streaming platforms, where listeners simultaneously encounter social signals (others' listening behavior) and identity cues (artist demographics)—and where evaluations of music or artist quality are inherently subjective, with no consensus on its measurement—offer an especially important window into these processes given their prevalence in people's lives and their potential to encode and amplify bias. This dissertation provides the first experimental test of this debate with quality directly controlled for, and the first application of a dual cumulative advantage framework—examining social influence and social identity simultaneously—to a cultural market context. Building an artificial music streaming platform, I employed a 2×2 factorial design crossing social influence (listener counts visible vs. hidden) with identity visibility (artist race and gender visible vs. hidden), yielding four conditions: Baseline, Social Influence, Identity, and Combined. Eight artists—one per race-gender combination (Asian, Black, Hispanic, and White; male and female)—were included, with easy-listening piano-driven instrumental music tracks randomly assigned to artist profiles across participants who were randomly assigned to one of the four conditions. Permutation-based macro-level factorial analyses showed that identity visibility produced more consistent effects on market inequality across outcomes than social influence. Pairwise condition comparisons showed that both mechanisms play a role in generating market concentration: the Combined condition—where both were present—consistently produced significantly higher inequality than the Baseline, with their combined effect approximating the sum of their individual contributions. Permutation-based micro-level analyses revealed that the White female artist gained share most consistently when both identity cues and listener counts were present, while Asian and Hispanic male artists experienced the consistent losses under identity visibility. Black artists gained share under identity visibility alone but lost that advantage when social influence was further added. Further mixed-effects model analyses of identity-based listening behaviors showed that participants exhibited race and gender homophily: same-race artists had significantly higher odds of being listened to and received longer listening times, while same-gender effects showed a similar but more limited pattern. Status analyses revealed that without controlling for homophily, the Hispanic male artist appeared disadvantaged relative to both White and Black artists—but after accounting for homophily, the disadvantage relative to White artists disappeared, indicating it reflected in-group preference rather than a true status difference, while the disadvantage relative to Black artists persisted. After controlling for homophily, Black artists consistently held an advantage over White artists—particularly the White male artist—for both listening probability and listening duration, underscoring the importance of distinguishing status effects from homophily. This dissertation makes four key contributions. First, it provides direct experimental evidence that social processes can generate statistically detectable market inequality independent of quality differences, supporting the position that quality differences are not a prerequisite for inequality in cultural markets to emerge and be statistically detectable. Second, by examining social influence and social identity simultaneously in a music streaming context, it demonstrates that both mechanisms play a role in market concentration, with factorial analyses showing that the main effect of social identity was more consistent than that of social influence in this experimental context. One possible explanation for this asymmetry is that identity-based preferences —operating through homophily and status beliefs—can activate immediately when demographic cues are visible, while social influence effects depend on the accumulation of listener counts that grow more randomly early in the experiment, particularly given that most participants explored five to eight of the eight available artists, making unequal count distributions slow to become evident. Third, by examining both macro-level inequality (Gini indices) and micro-level share dynamics across artists and demographic groups, it provides a layered account of how cumulative advantage processes generate and distribute inequality—revealing not just that concentration increased, but which specific demographic groups gained and lost. Fourth, by distinguishing homophily from status effects and employing an inclusive demographic design incorporating four racial groups crossed with gender, it reveals nuanced patterns of advantage and disadvantage that narrower designs would miss. More broadly, these findings suggest that design choices on music streaming platforms—what social signals to display and whose identities to reveal—shape not only individual experiences but also the distribution of visibility and success among creators. Platforms that aim to promote fairness might consider how displaying identity information and social signals jointly amplifies demographic concentration, and whether algorithmic curation could mitigate these effects by reducing the salience of demographic cues or actively diversifying recommendations.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy
- Level thesis:degree_level
- doctoral
- Discipline thesis:degree_discipline
- Sociology
- Grantor
- Boston University
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Zhou, Caoyifu
- Advisor dc:contributor.advisor
-
- Gondal, Neha
- ORCID dc:identifier.orcid
Subjects
dc:subject × 7Rights
dc:rights- Statement dc:rights
-
- Attribution-NonCommercial-NoDerivatives 4.0 International
- Licence dc:rights.uri
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/2144/53436