|
I agree it is likely selection bias but I don't think it is likely for the same reasons. These kinds of results often correlate strongly with parental income levels, which put another way "zip code". Yeah, the military isn't known for great salaries and you'd be right to point at plenty of rich counties, but how many rich counties are there to poor ones? We don't have the distributions and that's what makes this hard to read. Despite that, we do have some distributional information. Lucky for us, they included the demographics! Taking what we know above, we can actually back investigate to at least provide a "sniff test". Looking at the DoDEA scales, they are pretty low variance in comparison. Unless you think Asians are genetically smarter than whites, blacks, or hispanics then it needs to come down to other factors, which includes culture. The culture will probably be suppressed a bit in the military data, as military naturally creates a more homogeneous setting, but some variance will still exist for this part as well as some likely imbalances in incomes and other things. An important part of this rich correlation is that it ties very much into stable household. Certainly having active deployment will disrupt the household a bit, but some of that normalizes and well... let's be honest, there is a stable income and stable food situation at home. That's a major factor in a lot of households. So the real question would be "How do DoDEA schools compare to national schools when you exclude national schools that have a significant number of families that do not have a stable income?" I believe that would be a more fair comparison, though that would really just bring us to "apples and oranges" instead of "oranges and tomatoes". The claim is that the difference is due to some organizational influence, i.e. one that is actionable (like the way teachers teach or students are disciplined, etc), but frankly we just have so little data we can't rule out a million other things. |