The School Report Cards on the VDOE Web site provide a lot of data but no way to compare schools or divisions except to copy the data out of the individual reports (Can you spell "pain"?). Recently, however, they put up a new front end to their SOL database that provides a much more convenient path to the data AND provides averages to two decimal places, in place of the rounded off numbers of the Report Cards. It is slow but, by golly, it works.
So I spent a Sunday afternoon poking around and obtained a more nuanced picture of the effect of Richmond's cheating on the VGLA.
You may recall that the tough new math test this year dropped the math SOL average and hit Richmond particularly hard. The new test provides an interesting experiment because it eliminates the VGLA. Here, from this new database, are the math scores for Richmond and the State average, by year, both for all students and for students with disabilities.
The lowered scores from the new, tougher math test are obvious.
A closer look reveals more:
First, the consistently rising scores for 2008-10 dropped in 2011. Is it a coincidence that, in 2010, HB304 and the Superintendent's newfound "concern" about the large VGLA population caused a sudden contraction in that population and a statewide drop in SOL scores when a number of students had to take the State-graded SOL rather than the locally-graded VGLA?
Next, Richmond has consistently underpeformed the State. Yet Richmond's students with disabilities have, until 2012, consistently scored higher than the statewide average for students with disabilities. But the new math test ended that and, for the first time, Richmond's students with disabilities underperformed the State average. Does this conform to the notion that the VGLA scoring was subject to local manipulation?
Last, the relative sizes of the score drops in 2012. The State average for all students dropped 18 points; the Richmond average dropped 28 points. The State average for students with disabilities dropped almost 26 points but the Richmond scores dropped almost 43 points.
Or, in terms of Richmond minus State scores by year:
The explanation that leaps to my mind is that depriving Richmond of the VGLA deprived it of a favorite tool for cheating on the SOL. Can you think of another explanation that better fits the data? Let's see what happens in 2013 when the new English tests eliminate the English VGLA.