One of the things often heard is that someone is leaving the city for the burbs because the schools are better (I use the generic city, since, in my experience this attitude doesn’t appear to be limited to any particular city). But what if parents aren’t choosing better schools, but better student bodies? What if parents are paying exorbitant housing costs, not because the schools perform better, but because those high housing costs are able to exclude students who perform poorly?
A while ago, I wrote that, despite the best efforts of the doom-and-gloom educational assessment industry, some U.S. states actually fare extremely well when compared to other countries: Massachusetts, Minnesota, New Hampshire, and New Jersey, for instance, blow away every single European country by a large margin. What do the good states have in common? Well:
But it’s no accident that the states which do well typically have low childhood poverty rates, decent incomes, and lower divorce rates–all signs of stable families which are conducive to learning.
So I don’t know whether to be optimistic about this or not. Yglesias argues that other states can learn from MA and other high performing states. But I wonder if the real lesson to be learned is that the ‘extra-educational’ environment–what students walk into the school with–plays a really big role.
For the statistically minded, R2 = 0.48, which, for an uncontrolled experiment, is pretty good (R2 is an indication of the strength of the relationship and can range between 0 – 1). For purposes of this blog post (not journal article, not policy paper, but a fucking blog post), what I want to draw attention to is that we can estimate NAEP test scores based on the child poverty rate.
For example, a population with fifty percent of children below the poverty line (which tragically is the case in some areas) will have, a score of 209 with a 95% probability of being between 197 – 219*. A population with only two percent of its children in poverty should have a score of 254 with a 95% probability of being between 231 – 274.
This isn’t to say that individual students can’t over- or underperform. And populations (states or schools) can defy expectations: at the state level, Massachusetts scores twelve points higher than it should. Demography may not be destiny, but it is a heavy burden.
In the above example, there is virtually no way the poor population will ever appear as ‘good’ as the wealthy school, even if the poor school does a better job of educating its students–that is, getting them to achieve more than they should based on their socioeconomic status. Even if you compare a school with a ten percent poverty rate (half the national average) to a school with a two percent poverty rate, only a quarter of the time will the ‘poorer’ school perform better.
The point is when parents are choosing schools based on test scores, they are not necessarily assessing school quality, but child poverty. The educational system that they’re leaving might stink too, but there is a massive conflation going on here. Even if they don’t think they’re doing so, families who are moving in order to secure a better education are, to a considerable extent, fleeing ‘undesirable’ student bodies.
*We can estimate the variation in the slope as well as the Y-intercept. For purposes of this fucking blog post, I’m assuming the Y-intercept is constant. You get what you pay for.