April 25, 2010

Veritas? Caveat!

There's a new paper by Harvard researchers Matthew Chingos and Martin West on what ex-Florida teachers make, and relationships between post-teaching income and during-teaching value-added measures that I'm sure some will tout as proof that SB 6 and any like performance-pay plans are desperately needed. Err, no. Fortunately, Chingos and West do not make that argument, but they also don't tell the reader how many student scores were excluded from their rules that would tend to eliminate special education service recipients, nor how they justified combining value-added measures across multiple grade levels... nor why they used a linear measure of age in both the labor-participation propensity measure and income when the labor-participation (and thus income) effects of age for reproductive-aged women are not going to be linear.

Then there's R2 for the key non-public-school labor market equation: .06 (see the last column in Tables 4-6 on pp. 39-41 of the MS). This is an underwhelming amount of variance the models explain.

Unfortunately, the breathless reporting of this study by Joanne Jacobs does not pay attention to these details.

Listen to this article
Posted in Education policy on April 25, 2010 5:03 PM |