Request for Statistical Information on Abuse

In the video Teamwork OP from a long while ago, Riot lists a statistic that sportsman-like players win over 1.7 million more games than average players. They also say that there is a 54% average win rate with zero ragers, which decreases the more there are. Furthermore, they claim a correlation between behavior and gold earned.

This certainly sounds like good evidence to back up Riot's position on anti-abuse, but as someone who's done a bit of social science work, it has me questioning the validity of Riot's statistics. In fact, I'm more prone to believe that the correlation flows the other way. I think this is the same for everyone, but on teams where the game is going well and you're winning, players are far less likely to rage at each other. I get the feeling Riot may be misinterpreting statistics in order to make a claim that may not be true.

Why does this request for statistical information matter?

Lyte's crackdown on negative behavior isn't uncalled for on moral reasons. I do, however, think that most players, when online and behind the screen of anonymity, have little care about their fellow player. Having statistics actually prove the correlation flows this way gives more proof to positive behavior, even on selfish grounds.

But taking this a step further to what might not be PC territory, if it shows that the correlation flows in reverse, and it's actually good players in good games that promote positivity, this should be impetus for Riot to look into ways to better balance games and give options to players that allow them to practice outside of games. In fact, it could lend support for the sandbox mode that some people have pushed for.

I certainly hope to see some response on this in terms of someone who's done some hard statistical analysis. In the meantime, good luck fellow summoners!