4 ms·

You mean, they... controlled variables? Between those 3 cohorts, I see a phenomenon that cuts across a wide slice of the population.

• ivanbakel 6y ago
  If you're arguing for particular significance of a variable (grit), you can't go around controlling for all other variables with the same effect. It's not meaningful to see significance of a variable in a homogeneous cohort and translate that to the the variable being significant in the general population. It might make sense that West Point students differ largely by grit, especially if all other variables are controlled for. That does not prove that the success of people generally is significantly affected by their grit.
• JamesBarney 6y ago
  The problem is if I look at a population that has an IQ between 115 and 120 and scores 20-25% on conscientious I'll find that grit is far more predictive than IQ or conscientiousness. But this doesn't tell me anything about how important it is in the general population m But you'd get far different results in the general population.
• wutbrodo 6y ago
  It's ignoring the contribution of intelligence, grit, and other factors to _getting_ into this situations (whether casually or correlatively). As an easy example: if I independently studied kindergarteners, fifth graders, and twelfth graders, I will have might discover that age has no correlation with verbal proficiency. The problem is that by studying homogeneous groups separately, I've thrown away the possibility of studying the effects that may have induced the homogeneity.