Why did they google 8000 screenplays and matched their lines, then used only 2000 for the analysis?
We're aware of sampling error and the potential for cherry-picking, but also struggled to figure out what was a representative sample.
If you go by box office success it seems to me you already introduce the bias of consumer preferences, not choices of the movie industry. Wouldn't it be better to go by production costs (and marketing budget, if that is not included in production costs)? Although over time one would hope the industry choices would reflect consumer preferences.