|
The Education Department has posted the 2008-09 SOL scores for each school, each school division, and the state in a series of computerized reports on the Web. These scores should not reflect the "adjustments" that inflate some of the accreditation scores by as much as 18 points. Nonetheless the numbers in many cases have been fudged (see the data here and here and, especially, here).
For the 2009 English test, Richmond's 82% pass rate puts it in a four-way tie for eleventh from the bottom, 1.4 standard deviations below the mean Division score:
The entire list is here. The Richmond math score, 78, places Richmond in a five-way tie for tenth from the bottom, 1.6 standard deviations below the Division average.
A plot of score v. the fall Division ADM (educratese for enrollment) paints a more detailed picture of Richmond's sorry position (the gold square). The green diamonds are, from the left, Hanover, Henrico, and Chesterfield. The red diamonds are the peer cities, from the left, Hampton, Newport News, and Norfolk.
There is no particular reason to expect these data to fit a straight line but the computer is glad to fit one. In both cases, the line suggests increasing SOL scores with enrollment (0.5 points per 10,000 for the English test, 0.3 per 10,000 for the math test) but the R 2 in both cases tells us that Division SOL score and enrollment are essentially uncorrelated.The data also provide a picture of Richmond's progress, or lack of progress, toward improving its performance. Here are the numbers for Richmond, the State, the nearby suburbs, and Norfolk for the past six years: As you see, the changing state average (improving except for the bump from the obviously tougher math test in and after 2006) provides a moving target. We can remove the movement by looking at the scores relative to the state numbers: Here we can see Richmond failing to catch up with the State average in both English and math. As to the 2014 deadline under the No Child Left Behind Act, the extrapolated Virginia data on the English test almost reach the 100% requirement but the Richmond extrapolation falls well short.
The change in the math test in 2006 makes extrapolations problematic. Well, given that all extrapolations are problematic, let's say "even more problematic."
If we toss out the data from the years prior to the change, the extrapolation looks very good for Virginia and close to good enough for Richmond.
If leadership is an important factor in school (in this case division) performance, the math and English scores should correlate. This year the Virginia data show an R 2 of 66%:
Costly Failure (Last year's data: Stay tuned for current data) Turning to the cost of these results, we at last have the 2008 financial data from the Education Department.
Here are the 2008 disbursement data (transfers to and from reserves and disbursements for constructing and renovating facilities removed) vs. ADM for the Virginia school divisions. ("ADM" is Average Daily Membership, which is bureaucratese for the average number of kids). Richmond is the gold square; Norfolk is the red square; the green diamonds are, from the left, Hanover, Henrico, and Chesterfield. The huge enrollment out at the right is, of course, Fairfax.
Limiting the inquiry to Richmond, the suburbs, and several comparable, old cities, we see:
or, in terms of a graph,
Restricting the inquiry to Richmond, Norfolk, and the State average:
Multiply the $2,147 per kid excess spending vs. the state by the ADM to get the excess cost of the Richmond schools vs. the state average: $48 million. Vs. Norfolk, the difference is $56 million. If we juxtapose the English SOL scores with the expenditure data, we see:
The green diamonds from the top are Hanover, Chesterfield, and Henrico. The red diamond is Norfolk and the gold square is Richmond. Looking for a trend, the computer was glad to fit a straight line to the data. The negative slope suggests that, among the Virginia school divisions, higher per kid costs correlate with lower SOL scores. The R2, however, tells us that expenditure is essentially uncorrelated with the scores. I think that is because the quality of schools depends on the leadership, not the cost. Rearranginging the same data to measure output relative to $ input, we plot the scores and expenditures as differences from the state average, and reverse the expenditure axis so the high bang per buck divisions will be at upper right.
The blue diamond at upper right is Poquoson, which scored seven points better than the state average at a cost of $2,893 less than the state average per student. The red diamond at lower left is Alexandria, whose citizens paid $7,210 per student more than the state average and received an English score 7 points below average. The yellow diamond at the bottom right is Petersburg, of course. Richmond is the gold square in the high cost/low score quadrant with Alexandria. The English dataset is here. Here are the same graphs for the Math SOL:
The additional blue point on the math bang per buck graph is Scott County, which got one point more than Poquoson at a cost of $823 more per kid. The data are here.
Cheating the Kids to get Better Scores So, we have very high cost and lousy performance. The other Bad News is that Richmond has been inflating the SOL scores by getting rid of almost half of the kids who enter high school. The enrollment pattern by grade gives away the game:
Just looking at the raw enrollments, the State enrollment at Grade 12 is down 18% from Grade 9. Richmond, however, is down 37% and Norfolk, which has perfected this tactic, 56%:
In short, Richmond and Norfolk are culling the kids to improve their test scores. The people who pay the price for that, of course, are the kids who don't graduate:
More data on this issues here. These data suggest that the Mayor's 2005-06 truancy campaign (by the City [pdf], not by the School Board, of course) has had only a minor effect upon Richmond's outrageous practice of driving out the low-performing kids to improve the scores. |
