John Bartle (right) says it was Blair 'wot won it' in six marginal seats
"It was Blair what won it,'' proclaimed The Mirror headline on May 2. "Labour's landslide victory ... is down to one extraordinary man ... Tony Blair."
Talk of elections is dominated by leaders. In the media the contest was transmogrified from a team game (Conservative v Labour) into a gladiatorial contest (Major v Blair).
Leaders' personalities are thought to swing voters from one party to another, possibly proving decisive in determining which party wins. In electoral politics, the theory is of a classical forum in which two or more orators vie for the audience's favour. Commentators and politicians alike toy with the idea of changing the party leader and speculate about its effect on party fortunes.
The notion that "leaders matter" is of tremendous political importance. It affects the authority and freedom of manoeuvre the parties afford their leaders. It affects how the parties campaign. It affects, not least, the position in government of the leader who happens to win. If the victorious leader is given personal credit for having achieved his party's victory, his authority as head of government is likely to be far greater. Ask Blair.
Can political scientists estimate the magnitude of any leadership effects? Together with my colleagues, Ivor Crewe and Anthony King, I constructed statistical models to address this question, using data gathered by Gallup in the final two days of the 1997 election campaign. Our models suggest that, once we know voters' class, general partisan inclinations and prospective evaluations of the parties' performance in office, the addition of further information about their evaluations of the party leaders does little to improve our prediction of how any individual will vote. Leadership evaluations do not loom as large in voters' minds as is often implied. Moreover, they had only a relatively small impact on the outcome on May 1.
Labour's lead over the Conservatives was 3.9 million votes. Our models suggest that favourable evaluations of Tony Blair, compared with John Major, swung 3,000 votes from Conservative to Labour, accounting for just 7 per cent of Labour's lead. If these swing voters were uniformly distributed across the country, Labour would have gained about 425 votes more per constituency from the Conservatives. The six Labour MPs who had a majority of fewer than 850 (Paul Stinchcombe in Wellingborough, Philip Sawford in Kettering, Brian White in Milton Keynes NE, Andy King in Rugby and Kenilworth, Anthony Clarke in Northampton S. and Eileen Gordon in Romford) might therefore owe their seats to favourable evaluations of Blair as a leader.
The lesson is clear. Leaders do have an effect on election outcomes, but it is what they do that influences voters. Only at a very close election, or when voters judge one of the leaders to be incompetent, is personality really likely to come into play. The 1964 general election provided one such example. That of 1997 did not.
John Bartle is a lecturer in government, University of Essex.