The Art of the Poll

How Americans are made to say the right thing.

AFTER A RECENT New York Times poll attesting to the growing disrepute of the Contract With America, the Speaker of the House howled of his betrayal. Not by the public, which, as the script goes, clamors unabatedly for tougher love and smaller government, but by its surveyors. It was a "nonsense left-wing poll," a "disgraceful example of disinformation." He obligingly suggested a more graceful example. Pollsters should have asked "whether respondents agree with the Republican plan to shrink government spending and balance the budget by enough to give a tax cut." In other words, the questions should have been posed in a manner more useful to the prerogatives of power. Generally this is what polls do best, but even the pollster's art has its sense of--well, not shame, exactly, but limits, beyond which even housewives from Wichita might slam down the phone or the newspaper.

Polling is "all part of an attempt to keep order. It defines politics and politicians to suit those who are already in power," former Democratic pulsetaker Pat Caddell told Christopher Hitchens in 1992, adding that "like many other technologies in politics, [it's] an instrument whereby the truth is obscured and the public will excluded and ignored." Just how this works is a matter of some nuance, but in the broad sense it involves asking questions that are (in the pollster's argot) useable, avoiding ones that are not, and fishing for opportune trend lines in the muddy stream of public sentiment.

To a great degree the results can be managed through the survey sample and the framing of the questions--"voters" or "likely voters" are by definition more invested in the status quo than nonvoters, and any person or program one wishes to cast in a bad light should be rhetorically linked with "big government"--but post-poll spin counts for plenty as well. Last year, for example, White House pollster Stanley Greenberg conducted a survey following the midterm elections. Its results were touted as proof that the New Democrat Clinton had been soundly cuffed by voters for slipping into Old Democrat liberalism and would have to track right in hope of becoming a still-newer Democrat.

The results did not bear out the gloss Greenberg and the Democratic Leadership Council subsequently put on them, but they didn't have to; hardly anyone checked. In fact, only 5 percent of those who actually voted in the '94 midterms said they were sending a message to "liberals," 15 percent to Clinton; this compared to 45 percent citing a shot across the bow of "politics as usual." As Ken Silverstein pointed out in the newsletter Counterpunch, the warrant for claiming a public love affair with the neo-liberal wing of the Democrats came down to a single loaded question: Did one prefer a "Traditional Democrat," defined to respondents as believing "government can solve problems and protect people from adversity," or a "New Democrat," defined by the credo that "government should help people equip themselves to solve their own problems"? Just try and guess which they picked.

The ensuing morality play was instructive: a chastened Clinton spouting acts of contrition based on a spurious reading of his own poll, and trudging mournfully to the precise spot he had occupied and meant to continue occupying as a member of the center-right governing coalition--albeit now with a clear mandate. It was a wonderfully cynical bit of political theater, throwing a sop to liberals even as it reassured the bipartisan right. The president may be a man of conscience and fellow feeling--deep, regretful sigh here--but he is forced by hard circumstance into hard choices, and he will struggle toward the will of the people.

The main fiction that contemporary opinion polls seek to promulgate is that of a public drift to the right commensurate with the one in Washington. The recent abundance of evidence that the public despises Gingrich and distrusts his Contract is no great obstacle to this conclusion, which the Star Tribune/WCCO Minnesota Poll reached again just last week. In the words of Steve Berg, who wrote the accompanying analysis piece, Minnesota "clearly is becoming more conservative." His evidence? I phoned Rob Daves, the director of the Minnesota Poll, and asked to see the complete survey. He informed me that the results were proprietary and that the Strib would release only those questions and answers referenced in Berg's story. It appears the basis for Berg's observation was twofold: A declining number of respondents are identifying themselves as liberal, and a rising proportion is citing crime and welfare as prime concerns.

The people have spoken. But what have they said? In the first instance respondents were asked simply to label themselves as liberal, moderate, or conservative; considering that "liberal" is strictly a pejorative term in the modern lexicon, more a charge to be fended off than a point on the political compass, one might argue the amazing thing is that even 21 percent laid claim to it. As to what they or their questioners meant by the term, there is no clue. Regarding the second point, Berg notes that "social issues" like crime, drugs, and welfare have swapped places with economic concerns in the public mind since 1992. This is mainly a tribute to the dialectics of polltaking. For the past three years the media have dispensed a steady diet of horror stories about crime and welfare and their linkages, steadily ratcheted upward by polls and focus groups showing that as coverage of these stories increased, so, incredibly, did public concern about them. QED, though it only proves what most pollsters will admit privately: Media coverage drives "social problems" numbers in Pavlovian fashion.

Next Page »