- Top Features:
- War Room
- Glenn Greenwald
- Broadsheet
- Joan Walsh
- Camille Paglia
- Since You Asked
- Follow Salon:
Read other letters about this article
Good points made, but here's the real killer in the methods. The authors appear to have sampled on the dependent variable. That is, they found the group they were interested in and looked at characteristics its members shared, but didn't then compare to a different (earlier-dying) group for the frequency of those same characteristics. It's like surveying the most successful CEOs and seeing that they are driven, energetic, and male, and saying those were the predictors of CEO success. If you also considered unsuccessful CEOs, you'd find that most of them were also male (and maybe even driven and energetic). We can't be sure without some variance in the outcome (in this case, how long people live). Whoops!
