Carnegie Mellon University
file.pdf (877.6 kB)
Download file

Predicting Box Office from the Screenplay: An Empirical Model

Download (877.6 kB)
journal contribution
posted on 2015-01-01, 00:00 authored by Starling HunterStarling Hunter, Susan Smith

Empirical studies of the determinants of box office revenues have mostly focused on postproduction factors, i.e. ones known after the film has been completed and/or released. Relatively few studies have considered pre-production factors, i.e. ones known before a decision has been made to greenlight a film project. The current study directly addresses this gap in the literature. Specifically, we develop and test a relatively parsimonious, pre-production model to predict the opening weekend box office of 170 US-produced, English-language, feature films released in the years 2010 and 2011. Chief among the pre-production factors that we consider are those derived from the textual and content analysis of the screenplays of these films. The most important of these is determined through the application of network text analysis—a method for rendering a text as a map or network of interconnected concepts. As predicted, we find that the size of the main component of a screenplay’s text network strongly predicts the completed film’s opening weekend box office.