The science of electoral predictions - 18 October 1996

Everybody says that the presidential election is all over before the first polling booths open, which happens two weeks from next Tuesday. But it's hard to find a friend who, having declared the campaign to be the dullest in history, will then talk about anything else – except of course, those people who are very many, who are not quite certain who is the president of the United States, but can rattle off the batting and pitching statistics of everybody on the two teams who this week will do battle for the baseball championships: the so-called World Series.

The despair of the Republicans and the general joy of the Democrats is a reflection of our religious belief in the science of polling and in the present figure, which runs about 15 to 20 percentage points in Mr Clinton's favour.

Let me at once shatter any suspicion you may have that I am a Luddite who agrees with Winston Churchill. During a rough and gloomy period for him – and for Britain – during the Second World War, he was bitterly criticised in the House of Commons for not paying enough attention to the Gallup poll. The Churchillian reply was typically witty, pointed and defensive. He said:

"Nothing is more dangerous in wartime than to live in a temperamental atmosphere of Gallup polls.... A speaker at the weekend said that this was a time when leaders should keep their ears to the ground. All I can say is that the British nation will find it very hard to look up to leaders who are detected in that ungainly position".

Four years later, Mr Churchill went on paying no attention to the Gallup poll and he paid a heavy price for not keeping his ear to the ground as he grimly conceded when, as he wrote: "All our enemies having surrendered unconditionally or being about to do so, I was immediately dismissed by the British electorate from all further conduct of their affairs".

Churchill's contempt for polls still persists among people who can remember the astounding triumph of the despised Harry Truman in 1948 or the British pollsters' embarrassment in 1992 at John Major's easy victory. And I imagine there are still healthy sceptics among the generation that remembers the birth of polling in this country and its first famous howling clanger: 1936.

Something called The Literary Digest held a poll and predicted confidently that the Republican Governor of Kansas, one Alfred Landon, would handsomely defeat Franklin Roosevelt in his second go at the presidency. Roosevelt won forty-six of then forty-eight states. The Literary Digest soon went to a well earned grave. Its system had been to post questionnaires to people in the telephone directory, forgetting sixty years ago, that households with a telephone did not exactly represent a cross-section of the populous.

But there was a beginning pollster in Princeton, New Jersey, whom nobody paid any attention to. He gauged the 1936 result almost exactly. He was Dr George Gallup and he showed himself at once to be a creditable apprentice to the science of statistics. He it was, who established the statistical sample, scanning a great range of the population with a matching range of interests, income, political bias and so on. He was followed by others who went on refining the sample.

Though still, please note, nobody has ever been able to isolate for sure the ratio of people polled who mean to vote but don't and the ones who don't mean to vote but do. The science lags in this vital area by lumping all such tricky characters – along with people who haven't made up their minds – and calls them all "independents." Today there are between 20 and 25 per cent independents.

The dreadful object lesson in this country with the fallibility of polls came in 1948. The whole country it seemed was disgusted with President Harry Truman. On the day three and a half years before, when he succeeded the dead Roosevelt, he had the support of 70 per cent of the population. In the spring of the election year of 1948, it had sunk to 36 per cent. Wherever you turned to read or listen, the press, the radio – media had not been invented – it was the same sad story. There was television, but in those early days television was no more a source of news than was vaudeville. In fact, most television was revived vaudeville.

Truman looked at the polls. He admitted the accuracy of the two best, but it occurred to him that perhaps the polling was done among a large section of the population that took its opinions only from the mainly Republican owned press and radio.

He had a mystical, barmy hunch that there were a few millions out there who had never been tapped. "Out there" meant the whole vast landscape of the United States. He decided to tap it and them, not by doing what all candidates had done this century: to take a train and stop in great cities, get out, make a majestic speech. Truman, born in Missouri into a farming family knew that trains stop at small towns if there's anybody waving at the station. So he launched his much ridiculed whistle-stop tour. He went a hundred thousand miles. He stopped at cities, towns, tiny villages to crowds of a thousand or a dozen.

The local press everywhere went into shock, but also into print. Truman abandoned all scripts. He talked off the top of his head and hammered away in eighty little speeches at the "kept" press and the "bought" radio. His case seemed hopeless not only because of the stunning figure Governor Dewey was racking up in the polls, but because of the drag on him from having a third and a fourth party running against him. Well Mr Truman beat them all and the pollsters having been trying to give a satisfactory explanation ever since. Certainly the people at the whistle-stops who gave Truman the edge hadn't been polled.

So what reason have we now to trust them? The best of them concede a possible error of four per cent, which is very large. In a close race, that could miss by a mile. John F. Kennedy got in by just over one hundred thousand votes in sixty-eight million. Figure that minute percentage error. The pollsters at the end sensibly called it "too close to call".

Nixon swore that in winning Illinois, Kennedy had got the votes of the tombstones in and around Chicago. And to this charge of wholesale crookery Kennedy privately replied: "So? Look what they did to us in Ohio".

Today whereas the Democrats are delighted to agree with the independent polls that President Clinton is leading by about fifteen points, the Republicans have their own wretched pollster who says that the president leads by only nine points. The publication of this figure is meant to show how truly objective a party man can be. I can imagine scores of Republican managers hissing, why did we employ him if his job wasn't to cook the figures?

So until Wednesday evening, the second and last of the presidential debates, there really was a national consensus that it's all over. What the Republicans were beginning to fret about was whether they could retain their majorities in the House and the Senate.

I said until Wednesday evening because of a conflict inside the Dole campaign staff.

In the first debate many Dole people felt Mr Dole ought to have gone for the president's character and the allegedly low cronies who surround him in the White House. Mr Dole said afterwards, the American people didn't want to hear insults being traded, they wanted to hear about the issues.

Well that debate didn't make the figures on the polls wobble by a percentage point. So up spake Mr Jack Kemp, Mr Dole's running mate. Asked why Mr Dole had gone so easy on Mr Clinton's character, Mr Kemp said: "It is beneath Bob Dole to make personal comments on character." Unrecorded groans from half the Dole staff who began to scream in private, and say Mr Dole's only chance lay in his coming out frankly with the all but attested facts known about the character of the president and of much of his staff: the assistants who are being investigated or have gone to jail, the manager of his re-election campaign who had an affair with a prostitute and used to like to have a listen in on private calls to the president. He simply resigned, but his shame – if any – has been cushioned by a two million dollar advance on a book, a so-called respectable publisher rushed to offer him. The gallantry of the aide, a woman who went to jail rather than say what she knew about the Clintons' role in the Whitewater affair.

So Mr Dole's response? On Wednesday, Mr Dole just couldn't do it. No frontal attack. He did bring up the public character of this administration, mentioned no names, just types that were being investigated. He made these points in irritable, little asides, which had exactly the opposite effect to the one the Republicans had prayed for. It made him seem peevish and a little mean, the old image he spent two years trying to erase.

As for the president, he stood there pink, unblinking, not angry, and the moment it was his turn he told each questioner – all the questions came from a picked audience, a cross-section of Americans – he told them what he meant to do for them and their children by way of mortgages, children's tax credits, health benefits, education, taxes, thousands more policemen on the streets, making new jobs, expanding maternity leave. He was specific where Mr Dole was vague. But to Mr Dole's attempts at wrath, nothing but a soft answer. And when President Clinton grew sentimentally vague in his peroration, telling us for the umpteenth time he was building a bridge into the 21st century which we would all cross together, it was corny but at least it was a better peroration than Mr Dole's, who promised in his eloquent closing words, quote: "To deliver a new economy package." Thud! The polls, for a reason we all have to figure out for ourselves, remain substantially the same.

THIS TRANSCRIPT WAS TYPED FROM A RECORDING OF THE ORIGINAL BBC BROADCAST (© BBC) AND NOT COPIED FROM AN ORIGINAL SCRIPT. BECAUSE OF THE RISK OF MISHEARING, THE BBC CANNOT VOUCH FOR ITS COMPLETE ACCURACY.

Letter from America audio recordings of broadcasts ©BBC. Letter from America scripts © Cooke Americas, RLLP. All rights reserved.