Interesting link.
The study doesn't account for the fact that many of these selective programs are interested in attracting the most able students, irrespective of economic conditions, and so offer partial or full scholarships to attract under-privileged children who have demonstrated outstanding talent.
Given that the study controls for student factors (among which the authors include economic conditions for the family), the results are certain to understate the influence of genetics due to the inclusion of outliers in the population for whom high cognitive profiles (and presumptive genetics) are present, but supportive environmental conditions are not.
Another consideration the study doesn't address is the functional form associated with access to supportive environmental factors. It seems the authors used some sort of OLS or GLS model, and environmental factors are probably entered in some log-linear format or as a combination of a functional specification and a vector of dummies. But experience teaches us that there are significant stepwise increments of environment that confer benefits, but for which intermediate adjustments yield no practical effect.
For example, accessing a museum once a year is likely insufficient to catalyze interest in a new subject, but monthly or bi-weekly visits would. If a dummy variable is entered into the model is unable to capture differences in practical significance of the predictor variables in the model. Likewise, a $5,000 increase in household income between $35,000 and $40,000 per year likely doesn't yield many discretionary educational opportunities for a gifted school-aged child, but that same $5,000 (on a tax-adjusted basis) between $110,000 and $115,000 would. If the vast majority of students studied have attributes clustered in ranges for which minor adjustments are not perceptible, it would be incorrect to state that the full effects of those variables had been properly controlled for.
For these reasons, and many others that I've chosen not to bore posters with, I take such research with a large grain of salt.
Last edited by aquinas; 03/23/18 08:38 AM.
