AS a newspaper editor, I used to spend a fair piece of the company's money on reader research - polling the readers to find out how they used the newspaper.
John Dollard, the Yale professor who did the research for us with his own polling company, had one unshakable belief about polling: The people you poll sometimes lie. Not maliciously. Not intentionally. Perhaps not even consciously. But for a variety of reasons, they sometimes give untruthful answers.
Sometimes they want to please the interviewer and give the answers they think the interviewer wants to get. Sometimes their memory of what they read in a newspaper (or which brand of cereal they bought, or which politician they voted for) is faulty. Sometimes their view of what they would like to read in a newspaper (or what cereal they would like to buy, or which politician they intend to vote for) is not in accord with what they will actually do.
Thus Dr. Dollard, the wise professor, was skeptical about research based on memories of the past or speculation about the future.
The only research of which you could be sure, he argued, was research done at the actual moment of commitment - when the reader had the newspaper in his hands, or made the choice of cereal at the supermarket, or pulled the lever in the polling booth.
I am sure there are those who would disagree with the Dollard thesis, but it is worth considering in these presidential election days when news organizations assail us daily with prognostications about how the electors will vote.
How accurate are the political polls? One week the Washington Post front-pages a story based on its own poll suggesting Gov. Bill Clinton has a commanding, double-digit lead over President Bush. Several other polls published elsewhere the same day suggest Governor Clinton has a much more modest, single-digit lead. The following week, the Post backtracks, in essence conceding its story was wrong and pinpointing the frailty of polls.
In a given week, a USA Today/CNN poll gives Clinton a 12-point lead, while other polls by equally reputable news organizations give Clinton an eight-point lead.
How to explain these very substantial variances and contradictions?
All the polls currently show Bush lagging, and that may be the reality of the situation. Maybe the presidency is already irrevocably lost to him.
One wonders about the volatility of the voters this election year. How seriously committed are they to the candidates they currently say they will support next month?
One wonders about the alienated voters. The Wall Street Journal said last week that it's becoming increasingly tricky for pollsters to take their political pulse. How to distinguish between the views of those who will readily respond to a poll and those who won't?
One wonders about those who will not vote, but who are telling the pollsters they will.
One wonders about those who are telling the pollsters they will vote for one candidate because they don't want to admit they're voting for the other.
One wonders about the voters afflicted by the bandwagon syndrome - who haven't really made up their minds but who tell pollsters they are voting for the candidate currently in the lead.
One wonders about the army of uncommitted - the voters who won't make up their minds whom to vote for until the ballot is in front of them on polling day.
If we are to believe the polls, the victory in Sunday's presidential debate went to Ross Perot, who with his folksy twang, his sharp one-liners, and his self-deprecating humor brought entertaining sparkle to an otherwise somber exchange.
But Mr. Perot is not going to be the next president, nor is he really likely to determine the outcome between the two other contenders.
That, despite what they now may say to the pollsters, will be determined by the voters when they enter the polling booths Nov. 3. That is the only true gauge of what the voters think.