I'm A-Polled
I read the Crimson this morning with dread. No, not because I saw some new scandal related to the Summers presidency. No new Cornel West scandal, no new anti-Semitism speech scandal, no new firing of (fill in the blank) scandal, no new AIDS scandal, no new Andrei Shleifer scandal, no new fundraising scandal, no new women-in-science scandal.
I read the Crimson because the Crimson did a poll, and it is dumb, and because it says that students are pro-Summers by a ratio of 3:1, it is sure to be used as evidence of the faculty's foolishness by right-wingers everywhere who, without having a clue as to what is going on at Harvard, can now point to the Crimson's poll and say how stupid and wacko the faculty is.
And sure enough, here comes Andrew Sullivan, now blogging for Time, who writes: "The p.c. left on the faculty may despise Larry Summers, but a new poll shows that the students are fine with him."
(Sullivan, by the way, did his dissertation under Summers supporter Harvey Mansfield.)
I wonder what "p.c. left" Sullivan is referring to? It'd be nice if he could name a name. Except that—oh!—that might undermine his argument.
But that's what a bad poll will do. Let's look at why it's bad.
The Crimson e-mailed 840 students asking them various questions about Summers. The results? "Just 19 percent of undergraduates in the survey said that Summers should resign, while about 57 percent said he should not."
But consider the techniques of the poll. We don't actually know how these 840 students were chosen or whether they were generally representative of the student body. How many were men? How many were science majors?
Of the 840 people surveyed, about half—424 people—responded.
More than half of the respondents were freshmen.
Which means a) that half of the people who answered this poll have absolutely no idea what is going on, but answered it anyway. And b) as few as students from each class other than first-years answered this poll.
I'm not an expert on polling, but right away, that raises some issues. Freshmen, for one thing, are surely less likely to want Summers to resign, since, barely cognizant of Summers' history at Harvard, they wouldn't see much of a reason for him to.
I see on the message boards some posters have raised the issue of respondent bias—whether strong Summers supporters are more likely to be represented in the pool of respondents, if most students don't have particularly strong feelings about Summers (as might be expected, given their level of contact with/knowledge of him).
So...we don't know how the respondents were selected, we don't know the exact breakdown by class, we know that more men responded than women—who, given the women-in-science speech, are less likely than men to be strong Summers supporters—and we know that the poll is dominated by freshmen.
All told, sounds like a poll that probably shouldn't be trusted. I expect it's true that Summers has more support among students than among faculty. But I also expect that we're going to hear this 3:1 ratio tossed around a lot in the next few weeks.
I hear, for example, that the Boston Globe is working on a story about student opinion regarding Larry Summers....