Thursday, August 21, 2003

The Public Schools Study: Gallup and Phi Delta Kappa have published a survey being sold to the news under the headline "Most Americans Oppose Vouchers." (The publication is here, if you care to read it.) First, you should know that Phi Delta Kappa is a public school advocacy group. Now a look at the methodology.

I'll start with vouchers, since that's taking the headline. If you wanted a good sample of the public mood on vouchers, how would you ask the question? I bet you'd use the word "voucher," right? The survey asks:

"Do you favor or oppose allowing students and parents to choose a private school to attend at public expense?
That's one way to describe vouchers, but not the most complete or accurate way. Predictably, it only gets 38% support. (Note that the other side is excellent at this technique, too. Christian education boosters trumpet majority support for vouchers in America, though their questions tend to focus on constitutionality, not desirablility.) The follow-up question does use the word "voucher," and -- unsurprisingly -- support for the voucher concept rises, to 42%. (And when asked of only public-school parents, support goes up another four points.) But this number is unreliable anyway, since the interviewer already got the interviewee on record as opposing the concept. Psychologists can tell you all about priming the cognitive dissonance pump.

Briefly, there are flaws with other parts of the report. For example, the report concludes that Americans have "high regard for the public schools"; this is not true. Rather, Americans display the same prejudices they display in most surveys. They tended to grade their public school as "A" or "B" 48% of the time, while grading public schools nationally as "A" or "B" only 26% of the time. This fallacy is so common in polling as to be almost a gimme statistic. For example, poll victims usually say the same thing about their congressmen: "Mine's OK, but the rest are crooks."

Finally, there is the sampling methodology:

Within each contacted household, an interview was sought with the household member who had the most recent birthday. This frequently used method of respondent selection provides an excellent approximation of statistical randomness in that it gives all members of the household an opportunity to be selected.
Well, sure, if you're willing to live with the downside of statistical randomness. They may have got hold of Gramps, who thinks Old Lady Ferguson still teaches grades 2-5 in a single room down at the little red schoolhouse. (And when the sample age composition tells you only that 38% of those polled were "50 and over," that's not a stretch.)

So what's here? Nothing, really. It's a flawed poll with lax interpretation, sponsored and published by a group with a major stake in the answers. Even if you push aside the influence of PDK and rest on Gallup's reputation, you're really only left with a "popular" polling organization; a political candidate or a product placement manager wouldn't twitch a finger based on Gallup's results. Neither should you.

No comments: