Nicholas Kristof is at it again. He's written another misinformed op-ed about education in the NY Times.
Yes, I'm annoyed that he seems to imply that in-school factors are more important than non-school factors when we know the opposite is true. And, yes, I'm annoyed that he reverentially references Stephen Brill's hatchet job. And, yes, I'm annoyed that he cites anecdotal evidence to argue that the exception disproves the rule. And, yes, I'm annoyed that he assumes enacting certain reforms is moving in the right direction despite very little evidence that these reforms will improve things.
But what bothers me most is his statement that "A study found that if black students had four straight years of teachers from the top 25 percent of most effective teachers, the black-white testing gap would vanish in four years." Please -- everybody -- please stop saying this. It's, quite simply, not true
First of all, the study being referenced didn't "find" this -- it was speculation (or, in economist-speak, a back of the envelope calculation). Second of all, the speculation was based on an erroneous assumption -- that teacher effects were additive. In other words, that if a really good teacher could help kids close 1/4 of the achievement gap in one year, that four really good teachers could help kids completely close the gap in four years. But life doesn't work that way. For at least three reasons:
1.) The effect that a teacher has on students wanes over time as those kids go off and play over the summer and move on to different classes with different teachers.
2.) The large gain brought about by one teacher is, in part, due to the inferior teachers those students had in previous years. Now that they've had a world-beater, it will be harder for their teacher their next year to help them make as much progress.
3.) That large gain might not have even occurred. Measurement of teacher effects is notoriously imprecise. In a number of studies, the correlation between a teacher's effect one year and the next have been surprisingly inconsistent and have only low correlations.
note: Somehow an earlier, different, draft of this post ended up here after firefox crashed. This is the correct version.
update: really, seriously, his claim isn't true