THESE are days for political pollsters to be particularly visible in many places around the world, and for them to feel particularly beleaguered.
In France, which is to hold a referendum Sept. 20 on the Maastricht treaty, poll-takers are under an embargo on the publication of data on the vote - which is not to say that polls aren't circulating informally.
In the United States, with the presidential election campaign in its final phase, ordinary citizens are aware of how the polling results show President Bush doing against Governor Clinton.
And in Britain, the five biggest polling firms are still reeling from the embarrassment of having called the national elections in April completely wrong. All five projected the popular vote going for Neil Kinnock and the Labour Party over John Major and the Conservatives, and they were off by a composite average of 10 percentage points: Working from results of late polls, they projected a two-point margin for Labour, when the actual result was an eight-point margin for the Conservatives.
"It was the largest single discrepancy polling has seen from late polls in any major democracy" since modern polling began, as one American political scientist put it.
Since this fiasco, the opinion firms have gone back to the drawing boards to come up with some new methodologies. In recent days, one of the five, ICM, has announced that it will abandon face-to-face oral questioning in favor of a secret ballot. Two others, Gallup and MORI, are also said to be considering the option.
The main problem in April, ICM's research indicated, proved to be Conservative supporters' reluctance to say how they would vote.
Americans, who generally are polled by telephone, seem to exhibit no such reluctance. But ever since George Gallup started out in 1936, some of his organization's polling has been done in person, and each of these interviews has included a secret ballot marked by the person interviewed and deposited in a "ballot box" the poll-taker has brought along.
What's really going on, though, our political scientist friend notes, is that "polling has become part of the way we do politics." Just as the television cameras have affected the political process they originally intended simply to record, so polling has become a part of the process, too. The lab rats have learned the game.
The public being sampled knew that it cost them nothing to send John Major a signal, a sort of protest vote, even though they knew they would vote Tory. We can expect more of this over time. We can also expect that pollsters will have to work harder to get statistically meaningful samples as more and more would-be subjects feel freer to refuse to participate in surveys.
Does that mean that polling is useless, or somehow wrong? No. There are certainly times to greet polling data with skepticism: Like any professionals, pollsters are sometimes under pressure to cut corners and this leads to inadequate sampling and, ultimately, to less than useful data.
But a reporter's professional experience confirms that there are many subjects in the public discussion on which pollsters have the closest thing there is to hard evidence. To be taken beyond vague hunches and the things that "everyone knows" to a concrete assessment, complete with a sampling-error percentage, is a wonderful thing.
Some candidates use pollsters to help them shape their message, which is not necessarily a bad thing. A candidate who has to hire a pollster to find out what his own values are is in trouble, but sometimes a good message needs refining and simplifying. We may lament the noise and clutter of a political campaign, but it is not unfair to expect a candidate to be able to cut through it all with a clear explanation of why we should vote for him.
Fundamentally, one has to have confidence that the political process will be able to adjust to increasing voter sophistication on polls - just as it has adjusted to other potentially havoc-wreaking developments, such as universal suffrage and the repeal of poll taxes. Democracy is not at risk from the people with clipboards.