Double Jeopardy in digital entertainment: a preregistered movie-level replication with an exploratory franchise extension using MovieLens 32M

Abstract

Abstract This paper revisits the Double Jeopardy (DJ) regularity in digital entertainment using the public MovieLens 32M archive. DJ predicts that brands with higher penetration also have slightly higher loyalty. We first run a preregistered test at the movie level, treating each movie as a brand within its genre and measuring loyalty as the mean user-level share of category requirements (SCR). Across the five largest eligible genres—Drama, Comedy, Thriller, Romance, and Action—penetration is positively associated with mean SCR, with Spearman’s $$\rho $$ between 0.424 and 0.530 and every preregistered confidence interval excluding zero. But each user rates a given movie only once, so this measure captures how large a share of a viewer’s genre repertoire a movie fills, not repeat demand. We therefore read the preregistered result as a penetration–repertoire-share association and add an exploratory extension at the franchise level. Treating film franchises (identified from TMDb collections) as brands and their rated installments as repeat occasions, we again find a positive penetration–loyalty association and a modest, DJ-consistent asymmetry: larger franchises have higher mean SCR, and penetration varies more than loyalty. Overall, the results qualify rather than simply confirm DJ in digital entertainment: the clearest DJ-like evidence appears when the unit of analysis allows repeated category choice.