Figure 2. Left: Average trend changes in breed registration around the year of movie release, for trend changes measured over four different time spans.
Statistical significance and confidence intervals are based on two-tailed one-sample -tests. Sample sizes vary because, given availability of breed registration data for 1926–2005, we could perform 1-, 2-, 5- and 10-year trend analysis for movies released, respectively, in 1927–2004, 1928–2003, 1931–2000, and 1936–1995. Right: Correlation between trend change and movie release year. Negative correlations mean that later movies tend to be associated with smaller trend changes.