Part of the commentary on the recent Chicago teacher strike pointed to evidence that students learn less when they don’t have teachers. Well, duh? Suppose that weren’t true–then why would we need teachers???
Mariesa Herrmann and Jonah Rockoff measure the effect of teacher absenteeism on student test scores (it’s large) using New York City data. First though, a little detour into scientific methodology. You might think that you could figure this out by looking directly at the correlation between student test scores and teacher absences. No economist would ever do this because of the likelihood that better teachers are also teachers who try harder to avoid missing work. If this is the case then the correlation between absences and results would be coincidental: better teachers have fewer absences and their students learn more but the former doesn’t cause the latter.
Herrmann and Rockoff employ several strategies to avoid such a fallacy, which is why their evidence is quite convincing. The simplest example is that they subtract off the average results for each individual teacher, using only unusually high or low absences for that teacher. So if a good teacher getting better student results only “counts” when that good teacher is absent even less than she typically is. (The authors use other techniques as well to control for the “spurious correlation” problem.)
On to the results…
- A teacher being absent matters a lot. Having a substitute is equivalent to replacing an average teacher with one at the 10th-20th percentile.
- Single day absences result in so little learning that the kids would be better off spending the day on the playground, at least they’d get some fresh air. (My words, not the authors.)
- Less is lost when newbie teachers are absent than when the teacher is experienced. This is consistent with the finding that new teachers aren’t quite as good as those with more experience. There’s less to lose when a teacher misses a day.
One further finding, this one with my interpretation rather than the authors’. Teacher absences close to test day have a larger effect than do absences earlier in the year. There’s little reason to think more real learning takes place close to the test, but there might well be more “teaching to the test.” This adds a bit to the evidence that standardized test scores measure testing skills and the like, rather than real education.