Maybe it’s just the people I know, but Jonathan Chait, arguing against Dean’s campaign, offers one of the smartest, most concise insights I’ve read in a while:
I once heard a Democratic pollster talk about how educated voters like to think of themselves as independents because they think it’s wrong to identify with a party–that doing so suggests blind partisanship rather than informed judgment. But such independents are actually more partisan in their voting behavior than blue-collar voters, who may identify themselves as Democrats but will cross party lines in the voting booth more frequently.
I wouldn’t be of the camp that statistics can tell you anything (they don’t), but we constantly have to unlearn false categories through which we understand statistics. The danger is that conventional wisdom tends to take categories at face value.
Appropriately, the Brookings Review has a special issue on opinion polling that brings up, though hardly resolves this tension. In fact, despite the editors insistance that they “bring to this magazine a straightforward bias in favor of polling”, the articles in it seem to point out the problems that opinion measurement brings. For instance, one author writes,
In mid-March, as diplomacy began breaking down, public support for war crept higher. The last Gallup poll before the invasion began showed 64 percent in favor.
This shift surprised commentators who had put stock in earlier polls showing that Americans were less likely to support the war if the UN refused to authorize it. The surprise reflected a misreading of what Americans were saying rather than an inconsistency in what they were thinking. Most Americans did not interpret questions about UN authorization as meaning that war could be legitimate only if the UN authorized it. Rather, for them it was a proxy for whether the United States should go it alone in Iraq or act with others. In the few instances in which pollsters asked people whether they would support attacking Iraq if the UN Security Council refused to authorize war but President Bush nonetheless assembled a coalition of the willing in support of U.S. policy, a majority of Americans supported war.
Again and again, the interpretation of polling data relies on application of a priori political analysis whose rigor leaves a lot to be desired… essentially it’s no more scientific a process than the kind of synthetic analysis marking, but not limited to, Marxist political science.
No comments have been added to this post yet.