51¹ú²úÊÓƵ

Clinton's star waxes and the CIA's wanes but Blair has not yet eclipsed Thatcher in the American consciousness. Huw Richards savours the debates at the American Political Science Association Convention in San Francisco. Predicting an election even a couple of weeks ahead is a pretty risky business. So Helmut Norpoth of the State 51¹ú²úÊÓƵ of New York, Sunny Brook was sticking his neck out a fair way when he offered a prediction on the presidential election in 2000.

Speaking during a session on election forecasting at the American Political Science Association Convention in San Francisco last week, Professor Norpoth named current vice president Al Gore a clear frontrunner for the Democratic nomination next time round.

"I would predict a narrow Democratic victory," said Professor Norpoth, whose forecasting model incorporates a strong historical element. His leap into the future enlivened a session on the current election battle that operated under one serious handicap - everyone agreed. "Who is going to win?" may be the only real question in politics, but it loses much of its interest when everybody comes up with the same answer.

All six panelists had no doubt that President Clinton is an almost certain winner, even though they were speaking on the morning after senior aide Dick Morris was forced to resign after tabloid exposure of his association with a prostitute. "Clinton would have to have been in the room with Dick Morris and the girl for it to make enough difference," said Alan Abramovitz of Emory 51¹ú²úÊÓƵ, Atlanta, adding that a Clinton victory was "pretty damn certain".

The consensus was that President Clinton was likely to win between 55 and 57 per cent of the vote of the two main parties. Nobody mentioned third party candidate Ross Perot, who won 19 per cent in 1992, until questions were invited from the floor. Brad Lockerbie of the 51¹ú²úÊÓƵ of Georgia, who also attempted an estimate of the likely electoral college result, predicted that Clinton would win 446 of the 538 votes.

Much less predictable are the Congressional elections, where the Democrats will be aiming to wrest back the control they lost after 40 years in 1994. Professor Lockerbie predicts that they will gain 13 seats in the 435-member House of Representatives and one in the 100-strong Senate, leaving the Republicans in narrow control of both.

Professor Norpoth, one of the few experts willing this time last year to predict a comfortable Clinton win, notes the historical tendency for a party which regains the presidency after a period of exclusion - as the Democrats did in 1992 - to win comfortably the next time. An equilibrium effect sets in at the third election, which is almost invariably much closer. The only exception to this was Jimmy Carter's loss to Ronald Reagan in 1980, four years after ending a period of Republican rule.

A further historical pointer is that most eventual winners have led from the start, taking the first state primary election. "If you are strong early, stay ahead and win your nomination, you are clearly a strong candidate," said Professor Norpoth. He adds that nominees who did not win the first primary - like Republican candidate Bob Dole, defeated in New Hampshire by Pat Buchanan - have only a one in five chance of winning.

James Campbell of Louisiana State 51¹ú²úÊÓƵ, who has looked at Gallup poll findings since 1948, notes that early summer results have little predictive value. But by September, when the nominating conventions have taken place and the campaign proper begins, they have an excellent record - tipping the wrong winner only in 1948 and the near dead heat of 1960.

Professor Abramovitz noted the agreement of all six panellists that the factors affecting the outcome were largely settled before the start of the formal campaign - traditionally the Labor Day holiday at the start of September. "Dole and Clinton are going to spend the next nine weeks charging around the country and spending huge amounts of money. Perhaps they should take notice of this and stay home," he said.