The Quicksand of Polling

Preelection political polls have their place. They inform voters on where the country stands on issues and candidates, while helping candidates adjust their campaigns to voter preferences. If polls indicate a close race, more voters might turn out to cast ballots.

But confusion can also result if voters face too many contradictory polls, such as the dozens of surveys relating to the Nov. 2 presidential election. Results appear daily, sometimes twice a day.

Though intended to reveal how Americans think about a candidate, political polling is actually more like an out-of-focus snapshot of a moving object than a firm prediction. Measuring a complex electorate can never be an exact science, even though polling technology certainly has improved. Pollster George Gallup took some two weeks to give Americans a sense of who won the 1960 Kennedy/Nixon debate. This year, ABC and CBS News were touting polling results just minutes after each of the presidential debates.

"Insta-polls" about candidates have less credibility than polls that plumb public thinking over time. Knee-jerk responses don't often predict the action of a person standing alone in a voting booth, contemplating the real responsibility of making a small difference in choosing a leader.

One new problem for pollsters is that more people, especially the young, are reachable only by cellphone and not a land line. Caller ID also impedes polling, with voters simply choosing not to respond to unknown callers. In fact, individual responses to pollsters have been in overall decline of late, leaving open the possibility that only highly motivated voters are likely to respond to them.

Troublingly, polls also can work as self-fulfilling prophecies. After the GOP convention, for instance, polls showed President Bush beginning to take a lead over Senator Kerry, thus dampening Democrats' optimism for Kerry (as measured by polls).

And sometimes, the undecided are drawn to a perceived "winner."

Some polls give extra weight to party identification in trying to determine likelihood of voting; some don't. Poll questions, the order in which they're asked, and the flow of news on any given day can also affect a poll's accuracy. Many pollsters identify themselves as Republican or Democrat. Predictably, their polls tend to favor their respective candidates more often than not.

Clearly, polling results should be taken by voters with a lump rather than a grain of salt. The most damaging effect of polls is during election day, when the media report exit polls from the Eastern time zones that can influence voters in the West.

And it should be noted that the rise in polling over the past few decades has coincided with a general decline in voter turnout.

Relying too much on polls for how to vote, or even whether to vote, runs the risk of voting becoming a tactical move relative to others' apparent choices rather than a civic duty to shape the direction of government.