Those Political Polls - Sometimes Accurate, Sometimes Not

Election '96 will be remembered for the deluge of public opinion polls that confidently predicted President Clinton's reelection. It will also be remembered for some polls that grossly exaggerated the margin of his final victory.

On election eve, various pollsters said Mr. Clinton was leading by as little as seven points, or as much as 18. His final margin over Bob Dole was 8.4 percent.

For months, Mr. Dole was dogged by consistent prognostications of his defeat. He waved them off contemptuously, saying he did not believe in polls.

A late survey by The New York Times and CBS proved to be one of the most inaccurate. It gave Clinton 53 percent (he actually got 49.2 percent) and Dole just 35 percent (he ended with 40.8 percent).

One of the most accurate polls was Gallup/CNN/USA Today. It gave Clinton 48 and Dole 40 - both within a point or so of their final vote.

Pollsters also forecast the Republicans maintaining a lock on the Senate, where they picked up at least one seat. The only contests in which they were on shaky ground were those for the House, which some said might go Democratic. The GOP majority handily retained control of the 435-seat chamber, although with a thinner margin than it commanded in the 104th Congress.

This doesn't mean that polls are without flaws. Relying on a faulty election-day exit poll, the television networks projected that Sen. Bob Smith (R) of New Hampshire had lost his reelection bid. Senator Smith actually won.

Still, many experts say the findings of this year's public opinion surveys reaffirm that what began as an inexact art early in this century has been refined to a science. Some call polls an indispensable tool of electioneering and a permanent fixture on the political landscape.

"Polling is absolutely essential," says Christopher Arterton, dean of the Graduate School of Political Management at George Washington University in Washington. "From the president on down to state representative, most political candidates are relying on polling."

Voter surveys not only tell candidates where they stand in their campaigns, but also how to tailor their messages to public concerns. Polls also identify the groups of voters that are susceptible to particular messages, says Mr. Arterton. In addition, he contends, polls serve the public good by stressing to public officials the stands the majorities of their constituencies expect them to take on crucial issues.

The downside, he says, is that polls may have helped depress voter turnout to its lowest level - 49 percent - of any modern presidential election. With Clinton shown consistently ahead of Dole since April, "many people may have decided that the election was so boring that they decided not to vote."

Al Richman, an opinion survey analyst at the US Information Agency, says that while polls have become accurate, the way questions are formulated and other factors can produce different findings by different polls on the same subject.