Hillary Clinton made headlines earlier this week for rejecting the idea that teachers should be evaluated based in part on student test scores.
“I have for a very long time also been against the idea that you tie teacher evaluation and even teacher pay to test outcomes,” she said to an audience of New Hampshire teachers. “There’s no evidence. There’s no evidence. Now, there is some evidence that it can help with school performance. If everybody is on the same team, and they’re all working together, that’s a different issue, but that’s not the way it’s been presented...”
Clinton’s remarks are right in line with the stance that the two national teachers’ unions have taken on test scores in teacher evaluation—both reject it. And her remarks were widely viewed as a political statement distancing herself from the policies of the Obama administration. (To be fair, they are also consistent with her past positions on teacher pay.)
But is she correct?
There have been a number of empirical studies showing that value-added measures, which are based on test scores, do pick up on differences in teacher performance. Among the best known are two random-assignment studies that found that students taught by teachers with high value-added scores do tend to learn more.
But—and this is the crucial point—just because VAM contains some performance information does not necessarily mean it should be used formally to evaluate teachers. Using VAM for research purposes is an entirely different purpose from using it in a performance-evaluation system, where it could have unintended effects, such as warping instruction by encouraging test preparation. There are plenty of technical challenges, too.
It’s also fair to say that test scores don’t reflect all of what most people would consider a good, well-rounded education (for example, how well teachers engage students, help them develop a ‘growth mindset,’ shape their civic attitudes, or develop their creativity, for instance). They are used mainly because they’re available for almost every student in the nation.
And what about using test scores as a factor in teachers’ pay? Research studies paint a decidedly mixed portrait of whether pay programs based partly on test scores produce better learning. One recent study of a federal initiative showed a small effect in reading, but that stands in notable contrast to other studies that have found virtually no effects. Clinton seems somewhat more favorable of bonus programs that reward teams of teachers, but the evidence actually doesn’t seem to be any stronger for those approaches.
So where does that leave us? You could say that Clinton was partly correct, but she certainly glossed over a lot about what research says about teacher performance.