<$BlogRSDUrl$>

My personal musings about anything that gets on my radar screen--heavily dominated by politics.

2007-08-02

A Note On Evaluating CSAP Scores 

NOTE: the numbers I used in this post are hypotheticals to make my point; for some reason, at the time I was writing this, I couldn't access the real numbers on the web. When I get real numbers, I'll use them (hopefully, they won't make my hypotheticals look overly-optimistic).

A common mistake people make when they're looking at CSAP scores is to compare last year's 3rd grade scores to this year's 3rd grade scores.

One can hardly be blamed for that--that's how the scores get reported by the notoriously uneducated press. But this is not the best way to look at the data.

If you really want to know how well your school is doing, you have to look at what educators call "cohort groups;" that is, look at how a class performs compared to how that same class performed last year.

Let me take an example: if you look at the 6th grade test scores in Jefferson County, you'll see that in Math, the 6th grade had 52% of the students score proficient or better. You might be tempted to think that this is a great thing, because last year only 48% of the 6th graders scored proficient or better.

But that would not be useful. This year's 6th graders are a completely different group than last years'--that improvement could be simply an unusual difference between classes. It's impossible to tell.

If, on the other hand, last year's 5th grade class had 55% of the students score proficient, than you must conclude that there's a problem. Why are 3% of the students not proficient this year when they were proficient last year?

You see how the data can be deceiving.

Want to really get depressed? Compare the 10th grade test scores with the 5th grade test scores from five years ago.

So I would caution you, as you are looking at the data, to not get too caught in the simple analysis. Work a little deeper and get a sense for how schools are doing by looking at how classes have performed in the past.

Weblog Commenting by HaloScan.com

This page is powered by Blogger. Isn't yours?