Comparing the Shot Length Distributions of Motion Pictures using Dominance Statistics

Abstract

Comparing the shot length distributions of motion pictures we are interested in answering three main questions: do two (or more) films have similar shot length distributions? If there is a difference between shot length distributions, how large is that difference? Where do the shot length distributions of two films overlap and where are they most different? This article demonstrates the use of dominance statistics—Cliff's d, the Hodges-Lehmann median difference, and the empirical cumulative distribution function—for comparing the shot length distributions of motion pictures that make better use of the available shot length data and provide an unambiguous description of the difference in style between films that corresponds more closely to questions we wish to ask of film style than using average shot lengths compared to the most commonly used method of testing μ X = μ Y, where μ is some measure of central tendency. R functions for calculating these statistics are easily available and details of relevant packages are included.