Saturday, February 03, 2007

More evidence that white NFL players are discriminated against: In the fall of 2005, I did an analysis of NFL stats to assist Steve Sailer who wrote a piece for VDARE about possible discrimination against white players. Using data from Caste Football on the number of white starters and nonstarters, I found that over a three-year period, teams with more white benchwarmers were noticeably more likely to win games. The number of white starters didn't matter much, but each additional 2.2 to 5 white nonstarters (depending on the year) resulted in an extra win. Caste Football’s J.D. Cash has suggested that perhaps white utility players are more likely to master the playbooks for multiple positions, and Steve added that perhaps benching black athletes who often have large egos poisons the atmosphere and lowers team spirit.

In the scientific spirit of replication (and the fact that tomorrow is the Super Bowl) I ran the numbers again for the year that is now coming to an end. Looking at the regular season, we see the same pattern observed in previous years: the percent of white starters does not seem to matter (r=.06) but the correlation between nonstarters and wins is noteworthy: .20. An example based on estimating a regression model will make more intuitive sense: if Jacksonville, who had only 5 white non-starters, were like the Jets who had 15, they would have won an estimated 2 additional games. This may not sound like all that much, but as Sailer wrote, in a game of many complex factors, identifying a factor that might turn an average team into a playoff contending team is a big deal.


  1. Maybe the quarterback and the other white first string players will have more teammates to hang out with and will feel less isolated and depressed. Since those players are valuable, the team as a whole benefits.

    I can't think of any way to falsify this ... other than careful interviews with or psychological tests of the white players. Not too likely to happen.

  2. Well, ... if my theory is correct then all-black teams, or teams with only a few, second-string whites, should not have this problem and do well.

    Such teams would throw off the regression; so one would need a quadratic term.

    Maybe some historical examination of basketball teams could throw some light on this.


Meta-analysis of clinical trials: Eat walnuts

I am always looking for easy eating choices that are good for you. This new meta-analysis of 26 clinical trials looked to see if walnuts ma...