Examining the Matthew Effect on YouTube Recommendation System

Abstract

YouTube has become the largest video sharing site and its recommendation system significantly affects people’s perceptions and behaviors. This study examined whether the Matthew effect exists on YouTube’s recommendation system, and tested how the two factors (views and subscriptions) affect the recommendation results. The data was collected from YouTube by Python program and analyzed by statistical methods. The statistical results confirmed the existence of the Matthew effect and revealed the significant influences of channel views and subscriptions on play counts. The implications for researchers, users, business units and channel owners are discussed.