As we saw with the Hewitt and Hoon affair, the question mark hanging around Gordon Brown's leadership is clinging to the Parliamentary Labour Party like a bad smell. But in the grand scheme of mainstream politics does leadership really matter? Does it make a difference to how a party is perceived and the number of votes it can expect in an election? The Blairites seem to think so - if only hapless Brown was swept away and an alternative leader was installed all would be rosy in the New Labour garden and it would romp home to an historic fourth term. But is this just wishful thinking on their part?
Political science has something to say about political leadership. The first notable study on this issue came in a contribution to the edited collection Labour's Last Chance? The 1992 Election and Beyond. In a paper called 'Did Major Win? Did Kinnock Lose? Leadership Effects in the 1992 Election' by Ivor Crewe and Anthony King, the authors asked if the respective leaderships of John Major, Neil Kinnock and Paddy Ashdown in some way determined the outcome of that election.
Crewe and King's starting premise was that political science has traditionally been very quiet on the question of political leadership. Their paper is dotted with anecdotal evidence of how political scientists, like any other interested observer of politics, will speculate on the effectiveness (or lack thereof) of particular political leaders in informal settings, but when it comes to formal research this topic has been neglected. For example in Anthony Heath's (1991) study of the 1979-87 period, Thatcher receives only three mentions and none of them address her leadership - despite the political dimensions her personality assumed over that period. To counter this lack, Crewe and King offered two hypotheses on the bearing leadership has on election results.
First a leader can make the policies, ideology, and identity of their party more attractive. Here the leader’s effect on the electorate is indirect and mediated by the party machine.
Second (and sometimes simultaneously) the leader can have a direct impact on the electorate independently of the party’s image. For example, Crewe and King argue Kinnock's leadership managed to enhance the standing of Labour by moving it towards the political centre and making it more of a serious electoral threat to the Tories while his leadership was held in low esteem by the electorate.
To demonstrate it more thoroughly, Crewe and King drew on contemporaneous Gallup data on the standing of Thatcher in her last 15 months of office, and Major’s first 15 months compared with Kinnock's status over this period. In 1989-90 polls put Labour’s lead over the Tories at an average of 13 points. Over 1990-2 the Major-led Tories transformed this into an average 2 point advantage over Labour. Kinnock’s approval rating averaged 10 points ahead of Thatcher's, whereas Major’s stood at 15 points over Kinnock.
But how does this translate into the study of leadership effects? How does one go about it?
In political science Crewe and King identified two possible (and complimentary) analytical strategies for determining leadership effects. The first is that of ‘improved prediction’. This assumes that by looking at the demographic background and ideological dispositions of individual voters, the ability to predict their judgements of a party leader is improved. The second involves performing ‘thought experiments’. This method tries to measure the appeal a party would have had if led by someone else. For example, many polls have been carried out to try and gauge whether Labour would now be any more electable if someone other than Brown was as the helm, and these often apportion results by previous voting preferences/party identification.
A similar Gallup poll from 1992 cited by Crewe and King shows up the kinds of results we can expect from a thought experiment. They asked how the Tories would have performed if led by Thatcher or Heseltine as opposed to Major vs. a Kinnock-led Labour party, and a Smith as opposed to Kinnock-led Labour party against Major’s Conservatives. It found that under Thatcher 10% Tory-identifiers would have been likely to have voted Labour, 34% less likely, and 56% no difference/don’t know. For Heseltine the respective figures were 7%, 19%, and 74% whereas for Smith it was 22%, 6%, and 72%. However Crewe and King were careful to emphasise that these figures can only be taken as suggestive: they note respondents are unlikely to accurately predict how they would have acted in situations they did not experience.
Crewe and King were quick to point out that party preferences do not automatically translate into leader preferences. The same data showed 54% of Conservatives rated Major most highly, 36% of Labour rate Kinnock highly, and for the LibDems 35% positively endorsed Ashdown, leaving significant gaps. Tories scoring Kinnock and Ashdown highly were 1% and 8% respectively, Labour supporters 4% Major and 9% Ashdown, and Major 9% and Kinnock 4% for LibDems. The Tories rating Major and Kinnock on an equal footing were 3%, 7% for Labour, and 2% LibDems. On Major and Ashdown’s equal rating, it was 30%, 4%, and 27% respectively, and on Kinnock and Ashdown the figures were 1%, 24%, and 10%. Rating all three equal the figures were 4%, 17%, and 13%. Overall 91% of Tories placed Major on a par with or above the other leaders, whereas the respective level of party support for Kinnock and Ashdown were 84% and 85%. The issue for Crewe and King then is whether the deviant leader preferences lead to “deviant” voting behaviour.
Using the same batch of figures the proportion of Tories who rated Major highly, but then gave their votes to another party was just 3%. Those who placed Major on a par with the other leaders saw deviant voting increase to 11%, and for those who rated Major lower than Kinnock and Ashdown saw the deviance rate hit 27%. For Labour identifiers the figures were 4%, 9%, and 33%, and for the LibDems 9%, 29%, and 47%. So the likelihood of voting deviantly increases as one moves away from strong support of the leader. But also the figures seem to suggest that Major had qualities the other leaders lacked, making Tory identifiers far less likely to switch their votes. Or it could be that the ideological baggage associated with long term adherence to the Conservatives demands stronger leader identification than either the Labour or Liberal traditions?
Crewe and King then perform an improved prediction strategy to gauge leadership effects. They looked at voters' feelings toward the parties and actual votes cast in the 1987 election, and compared those figures with their 1992 vote. They argue that their prediction of Conservative voters who were favourable toward their party, voted for it in 1987, and would therefore vote the same in 1992 was 87.9% accurate. For Labour the prediction rate using the same operation was 87.1%. However when leader preferences are factored into the calculations the prediction rate increases by 1.1% and 0.5% respectively, giving figures of 89% and 87.6%. Therefore while leader effects can be calculated, for Crewe and King they are rarely decisive, and would only be significant in marginal contests.
Therefore in the case of the 1992 election and despite the extensive media commentary on the issue, the outcome was not really a case of either Major winning or Kinnock losing. To back up this conclusion they cited poll evidence of Labour under Smith which shows no significant improvement on the personal approval ratings obtained during the Kinnock era, suggesting Labour’s then problems were more about politics than personality.
But does this situation still pertain today? Like Kinnock Brown is more unpopular with the wider electorate than his Tory opponent. If the findings from the (admittedly dated) study remain true then it won't be leadership that decides the outcome of the election. But then again the Blairite's whole electoral strategy is premised on retaining/winning marginal seats, which hold the key to who forms the next majority. In tight contests the perception of Brown's leadership might just edge it for the Tories. The hope then for Labour lies not in trying to prettify Brown with smiley YouTube appearances but with a credible policy alternative to the reckless slash and burn proposed by Dave and friends. Unfortunately for Labour, Brown's speech on Monday was a return to the empty 'aspirationalism' of the Blair years designed to appeal to the so-called Middle Englanders that will supposedly determine the election's outcome. By narrowing the policy difference between the government and the Conservatives, Brown is creating a contest where the marginal impact of leader effects could prove decisive. It raises the awful prospect that the Blairites might just be right.