How Expert Are The 'Experts'?

Sir Robert Worcester wrote this article for the April issue of Parliamentary Monitor before the postponement of the election from the expected May 3 was announced. Now we know that the Reuters 'Experts' were wrong even on the prediction of the election date, where he agreed with them, his remarks on the reliability of the pundits seem more relevant still.

"The general election will go ahead on May 3 despite the foot-and-mouth crisis and the government will win with a 97-seat majority."

Who says so? The latest 'Kalends' survey of 'leading political scientists and commentators', to be found on the Reuters' web site. Three in four of the 'talking heads' Reuters canvassed up to 21 March think May 3rd is still on, up from 51% a few weeks ago. Indeed, 83 percent of their 'expert respondents' said if the general election was to be delayed as a result of the foot-and-mouth crisis, it would have little or no effect on the outcome.

So what is the record of the Reuters' Panel of 'Experts', first invented for the 1997 election? Then they were first polled seven months before the 1 May election day, when the panel forecast a 36-seat win for Blair. By March however, the margin was up to 56, and by a month before the day, 77. So, at a comparable time to now, they underestimated the outcome by 102 seats.

And the day before the election four years ago, the 'Experts' anticipated a Labour majority of 92, which the next day turned out to be 179. Then Reuters' published the forecasts of each of their pundits: the highest forecast of all was 153, David Butler's, while the self-confident David Sanders of the University of Essex, the proponent of economic determinism, was second from the bottom, with 55, and the wooden spoon of the 1997 election went to Warwick University's David Carlton, who had been writing weekly in the Spectator of his confidence in the Tories doing really very well and that the best result that Labour could hope for was a hung Parliament, anticipating a dead heat between Labour and the Conservatives, so he was only 179 seats off, made on the day before the political deluge.

Who else opined with confidence then, whose pronouncements need to be taken with a grain of salt now? Rallings and Thrasher? They thought 'Labour is more likely to get a comfortable majority than a landslide', forecasting a 65-seat majority for Labour. Still, most of the Reuters' 'Experts' were closer than many pundits ^*.

• Stewart Steven (Mail on Sunday): "Mr Major is surely a safer bet that Mr Blair"
• Simon Heffer (Daily Mail): "One reason I have never subscribed to the Labour landslide school of thought is a firm belief that many people are lying to the opinion polls"
• David Cowling (New Statesman): "I think it will be a majority of 47"
• Hugo Young (Guardian): "There will be no electoral melt-down for the Conservatives on May 1. It's hard to imagine them emerging with fewer than 200 seats in the next Parliament"
• Bruce Anderson (Spectator): "A Tory victory is still possible, though unlikely. But an overall Labour majority seems equally unlikely."
• Woodrow Wyatt (The Times): "I believe that John Major, who has fought brilliantly, is on course for a majority of around 30-40", and...
• Andrew Cooper (Conservative Central Office): "If Sir Robert Worcester goes on predicting a landslide for Labour right up to polling day, he'll lose in spades."

So what did happen then, and what can we learn that can be applied now? Certainly there was massive abstention. The Conservative vote dropped from c. 14 million to 9.6 million. The Liberal Democrats' vote actually fell by 800,000 votes, while the nationalists increased by c. 200,000 votes, and Labour went up from 11.5 million to 13.6 million, up over two million. And the Referendum Party/UKIP had some 1.2 supporters collectively. While we don't know with precision who went where (or didn't), it would appear that some 2.5 million 1992 Tory voters didn't vote at all in 1997, knowing from the polls, if not some of the pundits, that Labour was in for a landslide, and so they didn't have to vote against their core values to get what they wanted, the Conservative Party out of office and John Major replaced.

Our research suggested that as many as a million and a half '92 Tories, roughly speaking, switched to Labour and a further million to the Liberal Democrats, while about 400,000 switched from the other parties to the Tories. Others have challenged this analysis, concluding against all logic that it was 1992 Labour voters who stayed at home. We think it was the disillusioned Tories who would have otherwise have swung to Labour who didn't turn out to vote in safe Labour seats, contributing to the lowest turnout since the war.

Our work to date suggests that an even lower turnout will occur on May 3rd. At the moment, I'd guess it will be a turnout of under 30 million, perhaps down a million from the 30.5 million that voted last time. There'll be a mixture: Old Labour, damned if they'll give Blair another landslide, and not, now, fearing that abstaining will risk something worse; One-world Tories, dismayed with the Europhobic slant to the current Tory policies, hoping that a sizeable Labour victory will result in moving the Tories leftward, leading to a replacement of William Hague; Liberal Democrats in the South and Southwest, followers of Paddy Ashdown unimpressed by Charles Kennedy; and finally New Labour idealists, disappointed by the failure, as they see it, of Labour's 1997 promises.

So, in this last, probably, Letter to our Readers until the election, my best guess of the outcome is a Labour victory of between 100-120 seats, with the Liberal Democrats losing one third of their Parliamentary weight, down from 46 (47 now), to 30-32.

And a final word: if there is not much change in the tracking polls over the election, don't think that public opinion is necessarily stable. Over the five elections I've done panel studies, recalling week after week on the same people, we've identified that some one voter in five (one in three in 1997) changes his or her mind over the course of the election campaign, many from whether or not they'll vote, as well as from one party to another.

31 days and counting.

Sir Robert Worcester

^* Sources for the quotations are listed in Explaining Labour's Landslide (Politico's, 1999). This letter is adapted from a column in the April 2001 edition of Parliamentary Monitor.