Praxis II growth

Lost amid the news recently that Iowa’s growth model had been accepted by the Department of Education was the apparent mandate that all new teachers pass the Praxis II test in order to meet “highly qualified” status under No Child Left Behind.

I find it ironic that Iowa’s application for implementing a growth model was initially denied for two reasons: that the state did not have two years worth of data for analysis, and that it did not mandate a test for teachers. The news of the testing requirement came out last summer, but the application makes it clear the Department of Education used the chance at a growth model as leverage. It was right in front of everyone, in the second paragraph of the application (download it here).

Now Iowa will be one of 45 states using the Praxis II as a measure for teacher preparedness. Yet, I’m not convinced the evidence fully supports it as an adequate proxy for teacher quality. Dan Goldhaber, a University of Washington and Urban Institute researcher, recently published a report analyzing what exactly teacher testing tells us about teacher effectiveness. He finds some positive associations, but the policy implications are mild or mixed.

Read the article for his regression results, or click on his chart, at left, comparing Praxis II curriculum tests and teacher effectiveness in math. Goldhaber defines a minimum level of teacher quality as two standard deviations below the mean for teacher effectiveness. First, notice how much of the sample is centered pretty tightly around the mean. Next, look at area IX. These are the false positives: teachers who pass the Praxis II but who perform relatively poorly in the classroom. Compare that to areas I and IV. Those are false negatives: teachers who scored too low to qualify under today’s cut-off scores but are actually fairly effective teachers.

Last, look at the vertical lines. The left line represents North Carolina’s cut-off score from 1997-2000. The right line is Connecticut’s current one. Think of a state trying to determine an appropriate cut-off point deciding between these two lines. The teachers scoring in sections II, V, and VIII would no longer be eligible for certification with the higher level. Of this 7.4% reduction in teacher workforce, 7.2% were effective teachers under Goldhaber’s minimum standard. Neither of the cut scores is screening out large numbers of applications--even the higher Connecticut standard still passes almost 90% of test-takers

The point is that there are there are serious trade-offs for implementing teacher testing. The false positives and false negatives are troubling, as is the seeming arbitrariness of determining a cut score. I wish the Department of Education didn’t strong-arm states into implementing a policy with such questions.