A Machine Learning Approach to Preferential Attachment and Status Advantage in a Hip-Hop Collaboration Network

• In Event: 50220 - Section on Organizations, Occupations, and Work Roundtables
  In Refereed Roundtable (60 minute): Table 5: Collaborations and Networks

Abstract

Status remains difficult to measure because it accumulates from implicit acts of deference. Scholars tend to capture actors’ status based on their affiliation or external arbiters, where the sampling frame of deference is observable, and thereby infer individual status advantages attributed to individual performance. This analytical focus, however, left us less known about the collective processes in which actors with different status positions directly collaborate to climb up the status ladder. Using a “featuring” collaboration network longitudinally collected from more than 3,000 independent hip-hop artists in South Korea, we develop a machine-learning (ML) approach that computationally recovers, from a partially observed arbiter, the directed deference in collaborative songs and theorize status advantages attributed to preferential attachment in collaboration. The ML classifier performs better than naïve quality-based classification. Contrary to the usual expectation that artists benefit from piggybacking a high-status collaborator’s fame, we find a U-shaped effect of preferential attachment on performance.

Authors

• Jaemin Lee, The Chinese University of Hong Kong

• Yujie Li, The Chinese University of Hong Kong