August 30, 2006

Let's Go to The Videotape

All you need to know about the sad state of public education is contained in two graphs I made up (with the help of my buddy Excel) based on data I painstakingly collected from S&P's Schoolmatters for the state of Pennsylvania.

The first graph plots student achievement vs. instructional expenditures for the 501 school districts in Pennsylvania.

The correlation between instructional expenditures and student achievement at these funding levels is random (R2 = 0.0013). Below the median (about $6700) we have a scatterplot. Above the median we see the residual effects of SES sneaking in. Above the regression line we see a lot of affluent school districts who have fallen prey to their monopoly rent-seeking school districts. Below the regression line we see mostly big city districts with lots of state and federal funding being pumped in and having little affect on student achievement.

These effects become clearer when we go to the next graph, student achievement vs. median household income:

The correlation between household income and student achievement is 0.42. That's a pretty good effect size for the social sciences. The data is a bit noisy because the PSSA data (which I'm using for student achievement) is not a very good measure of student achievement and the household income data is community based data, which may not correspond exactly to the student's household data. Nonetheless, it's pretty clear that family income is a good predictor of academic success. An even better predictor would be parental education, which I'll scrounge up soon hopefully. The best predictor would be student IQ, but we don't have that data.

So when Affluent School DistrictTM claims that they educate your kids better, thus justifying the high tax rates, you can tell them that their value-added is minimal. Their performance is mostly the result of student ability. They've been dealt a better hand. Same is true for the supposedly bad schools at the lower end. Their performance is a product of the students they get.

The affluent schools don't do a much better job with their poor and minority kids and the bad schools would do almost as good a job with the higher ability kids. The fact of the matter is that in the end most schools teach the same hodge-podge of faddish crap. A few really good teachers doesn't make much of a difference, education is the by product of many teachers over the course of many years. It all evens out.

8 comments:

Mike in Texas said...

Nonetheless, it's pretty clear that family income is a good predictor of academic success.

Man, who ARE you and what have you done with KDeRosa?

I believe I've been trying to point out to you the correlation between academic achievement and household income for quite some time. The data on parental education should also be interesting.

KDeRosa said...

The main difference between the parental education data and the household income is that the correlation is higher (about 0.6 as I recall). This makes sense because parental education is a more accurate measure of parental IQ than is family income.

Where we differ is our conclusions.

Your view is that student achievement is a fait accompli among the low SES crows since schools are currently doing as good a job as can be expected.

My view is that schools and teaching can be improved such that most kids perform at grade level. The regression line will still exist, it'll just be shifted upward.

Mike in Texas said...

Actually, I do NOT feel that student achievement is fait accompli among the low SES crowd. I just disagree with you (greatly) on how to go about improving the student achievement levels of low-SES and all students. High stakes testing is not the way to do it, nor is a school system where politicians are making the major decisions. The way to improve schools is to put teachers in charge of the reform measures

Visit any low achieving school, ask the secretary who the best 3 teachers are, and ask those 3 teachers what the problems are. Not only will they ID the problems but they can supply you with the solutions.

The problem is, most politicians aren't listening.

KDeRosa said...

High stakes testing isn't the solution it only shows where the problems lie. Schools still have to do something to improve test scores, if they are unwilling or unable, improvement will not occur.

Politicians are only making some of the decisions, many of them bad. Most of the decisions are left up to the schools who make just as many bad decisions.

You're a teacher tell us what you'd do. But, stay within the realm of reality. Be specific. SHow support.

rightwingprof said...

I did a couple of very proto prelim analyses myself (with what data I had, which wasn't much), here and here.

rightwingprof said...

"The problem is, most politicians aren't listening."

I hate to say it, but this is, of course, the problem with putting any service in the hands of the public sector. Politics. But politics and politicians are the price we pay for public education.

KDeRosa said...

rightwingprof, I had read your analyses when you posted them. i also ran the ANOVA on the dataset with similar results.

Brett said...

What would classroom teachers do if they really held the reins?

It's possible that we don't have to speculate. I believe that there are stories of teachers who, fed up with "the system," banded together to open their own charter schools. If we can identify them, can't we look to see exactly what they've accomplished?