Unanchored Voters Cause Wobbly Polls
BY post at the Roper Center for Public Opinion Research has brought me a flood of calls about the performance of the polls in the 1992 campaign. One especially insistent question: How do we explain how one survey can show Bill Clinton leading George Bush by 15 percentage points while another, just a day later, puts the Clinton lead at only 6 points?
Examining such divergent results, we soon find their main explanation not so much in the surveys as in the American electorate itself. Simply put, it's hard to measure an opinion which hasn't formed or a decision not yet reached.
People not anchored by clear partisan ties or philosophic commitments have always been "up for grabs" in presidential contests. In recent decades, their ranks have greatly expanded as party loyalties have weakened. Never before have so many American voters been unanchored.
Into this evolving electoral setting, conditions specific to the 1992 campaign give pollsters an added challenge. An unusually large segment of the electorate now feels tugged in opposite directions. After 12 straight years of Republican control of the presidency, and with the economy stuck in third gear, the impulse is strong to give the other side a chance.
Tugging many voters the other way, though, are continuing doubts about the Democratic standard bearer. In one form or another - and it cuts differently across various blocs of voters - the "character issue" dogs Mr. Clinton's campaign now just as it did early on. In addition, for all the concern over conditions in the country - which inevitably impacts negatively on the incumbent - Mr. Bush isn't now seen as a failed president, as Jimmy Carter was in 1980. Bush beats Clinton in leadership assessments - e .g., who is likely to show better judgment in a crisis - as well as on personal characteristics.
While the state of the economy is unquestionably Clinton's biggest opening and advantage, even that issue doesn't work neatly to his advantage. Though Bush takes a lot of heat on the economy's performance, he isn't seen as primarily responsible for it. The public doesn't give Clinton the edge he might expect. A Gallup poll for Newsweek, taken September 10-11, for example, found only a 9 point plurality of respondents (47 to 38 percent) saying Clinton would handle the economy better than Bush.
Given this very lightly anchored, highly ambivalent electorate, some variability among the polls isn't nearly as striking as their overall consistency. For the past several weeks, the composite of surveys nationally has put Clinton's margin at about 10 points, sometimes bouncing 5 points or so above or below this average. The polls have done their job well.
The real question for election handicappers is how to interpret the 10 point margin. If it's "real" - based largely, that is, on electoral judgments that have actually been made - Bush's re-election is in extreme jeopardy. But if it rests largely on a desire, 6 weeks or more out, to send the incumbent the message that "we're not happy with things - get going," the race looks different.
What happened in Britain's election last April is instructive here, though obviously not determinative. The composite of 5 national polls taken just before the balloting showed Labor up by a point or two; in fact, the Conservatives won by 8 points. In reviewing the results for "The Public Perspective" magazine, Robert Waller of the Harris Research Center in London wrote that "voters decided to look forward to the next five years rather than express a protest against the recent problems of the Conservativ e government."
This involves, Mr. Waller went on, something quite different from voters changing their minds at the last minute. "Had the election been held a week earlier, the same phenomenon - Labor ahead in the polls, the Conservatives winning the actual contest - might well have occurred. The act of deciding who will rule for the next five years ... is a different type of act than responding to a survey questionnaire."
We'll see what happens here in the US come Nov. 3.