My assumption is that they ran a correlation test, which checks how much of the variation in name-changing is explained just by Scrabble score. If that's high enough, there just isn't much room for other factors that aren't themselves correlated with Scrabble score.
