Accepting the Dare to Compare SATs Across States
So a number of folks opined that the list of SAT scores for all fifty states that I posted the other day is meaningless because the states vary with respect to participation rates (PDF). Many states’ students don’t even take the SATs unless they want to go to certain higher-end universities on the East Coast, so of course they’d be apt to score better. By contrast, some have hypothesized that states with high participation rates may be encouraging students who mightn’t otherwise bother to take the test, which would seem likely to drag scores down.
Cross-referencing the participation rates with the states’ average scores leaves little doubt about the former point; the highest-scoring states have only single-digit percentages of students actually taking the test. But what about the latter point? Does high participation correlate with lower scores? Well, not really (solid lines follow the left axis; dotted lines follow the right):
This chart includes the sixteen states in which more than 60% of graduating high school students take the SATs. Except for highlighting the fact that Rhode Island is on the wrong side of the chart to have such a low score, arranging the data this way doesn’t appear to tell us much. Participation doesn’t appear to correlate with SAT scores.
To see where Rhode Island stands by a different measure, I sorted the states by the point spread between public and private students, and beyond finding Rhode Island to be third worst, an interesting consequence of this arrangement emerges:
What’s interesting is that, although the scores drift apart as we move to the right of the chart, public and private school grades fluctuate in similar ways. The implication is that something irrespective of the school is playing a role, and figuring it out might salvage some utility from this line of inquiry.
So, let’s test the already-suggested child poverty explanation (data from 2007):
Although there does appear to be a downward drift of SAT scores as child poverty increases, it doesn’t appear to be a strong correlation. Note, especially, that, although Rhode Island is still near the bottom of the list, those states beneath us are entirely different from those beneath us on the point spread chart.
What if we switch to household income (data from 2007)?
Here, there doesn’t even appear to be a drift, and certainly no correlation. Look, however, at what happens if we re-sort by point spread:
It appears that median income helps to explain the fluctuations, but neither income nor child poverty tell us much about the increasing disparity between public and private schools. Unfortunately, I’ve been unable to devise a method of quantifying the strength and zealotry of teachers’ unions in each state, but it does not appear that public school teacher salary has much of an effect:
Of course, to some degree, income, private school attendance, public school teacher pay, and SAT scores all correlate, but Rhode Island stands out, in this regard: We’ve got an average median income, but the fifth highest public school teacher pay, the second highest private school student percentage, and the third lowest public school cumulative SAT score (despite the sixth highest private school cumulative SAT score). Take especial note of the coexistence of average wealth and high private school attendance.
We’re certainly in the realm of speculation, here, but my hypothesis stands (and is, I chance to say, slightly stronger for this analysis): In Rhode Island, at least, the crushingly strong teachers’ unions are draining the attractive qualities from the public schools, driving those families who can somehow manage to afford it — and who feel their children possess the potential to justify it — into the arms of the private schools.