Making data-driven education reform decisions is great — unless the results come from flawed questions.
The results of an opinion poll will vary, and not by a little, as a function of how the questions are phrased. “Do you favor special preferences for minorities in the form of affirmative action?” will attract many fewer favorable responses than “Do you favor efforts to help minorities get ahead in order to make up for past discrimination?” And then, of course, there are “push polls,” which only pretend to sample people’s views while attempting to influence them: “Would you be more or less likely to vote for Congressman McDoodle if you knew he was a practicing Satanist?”
I find myself thinking about how much more—and less—there is to polling than meets the eye whenever I come across one of those surveys that school administrators like to distribute to parents. I have to assume these are not intended as the equivalent of push polls, that there’s a sincere desire to be responsive to the community and an honest pride in being able to cite “data” to judge the effectiveness, or at least the popularity, of school policies. (Data good.)