That rumbling you heard in recent weeks was a gathering stampede of election watchers declaring British prime minister Theresa May a woeful campaigner, uninspiring and incapable of projecting leadership. The Conservative manifesto was lacklustre, and May’s handling of Brexit left much to be desired.
And – credit where it’s due – Labour leader Jeremy Corbyn’s performance has pleasantly surprised. Who would have thought it possible that this divided opposition party, this accidental leader, a far-left backbench nobody, would be snapping at the heels of the incumbent?
These judgements are purely, totally, driven by the opinion polls, which over the campaign have narrowed markedly. The polls come first, followed by the reasons, always.
The collapse in the Tories’ lead has certainly been dramatic. When May announced the election in mid April, surveys generally put the Conservatives in the mid-to-high forties, around twenty points in front of Labour. Six weeks on, with a day to go, the latest surveys have the gap narrowing to anything from one point to eleven. (The differences between pollsters, which were always there but take on extra importance now the race has narrowed, mainly reflect varying turnout-by-age assumptions.)
The all-but-assured increase in the Conservative majority is now in doubt, and a hung parliament is a real possibility. If the scramble among pundits to shoehorn campaign perceptions into the poll numbers means contradicting their earlier assessments, then so be it. The show must go on.
But what does the poll tightening really mean?
It might surprise you to learn that this is actually a well-worn trajectory. When two parties enter an election campaign with one leader streets ahead of the other in personal ratings – in “satisfaction” and as preferred prime minister (or, in Australia, premier) – the less popular side will usually make substantial ground in voting intentions.
In Australia, probably the best example in terms of sheer numbers – and the most-remembered – is the federal election of 1984. Labor leader Bob Hawke entered his first campaign as prime minister with approval ratings in the sixties or seventies, around three times that of his opponent, Liberal leader Andrew Peacock, and with a two-party-preferred voting-intention lead of around 20 per cent. Little by little, the gap shrank, and on election day Labor’s margin was 3.6 per cent.
The accepted reasons for that closer-than-expected result are embedded in the received wisdom. The prime minister, emotionally hobbled by his daughter’s heroin addiction, was off his game; the long campaign was a mistake; the assets test on pensions bit harder than expected. And Peacock, of course, put in a stellar performance.
Yet this trend, though of lesser magnitudes (and off smaller gaps in leaders’ ratings), characterised all the elections Hawke contested as Labor leader (including from opposition in 1983). He always entered the campaign comfortably preferred to his opponent, and his party’s lead always dwindled as voting day approached.
And Hawke isn’t the only one:
• Campaigning for Australia’s 2001 “Tampa/September 11” election began in early October with a resurgent prime minister John Howard enjoying satisfaction rates in the sixties. Double the number of respondents preferred him to Kim Beazley as PM, and Coalition support hovered around 55 or 56 per cent after preferences. At the ballot box, the victory was a much more modest 51–49.
• In 2007, under Kevin Rudd – the most popular opposition leader in polling history – Labor’s lead of ten to twelve points was eventually whittled down to 5.4.
• In 2010, a honeymooning Julia Gillard, much preferred over her opponent Tony Abbott, succumbed to apparatchiks’ urgings to rush to the ballot box while the polls beckoned, only to see the figures collapse.
• And, back in Britain, Blair’s 2001 re-election can also be placed in this category: stratospheric approval of the prime minister and a huge party lead substantially narrowing.
Any counter-examples? Western Australia this year constitutes a mild one. Labor leader Mark McGowan began comfortably ahead of premier Colin Barnett in personal ratings, yet it was Labor that made a (slight) improvement by polling day.
Overall, though, the evidence is persuasive: a highly popular leader seems to artificially inflate voting intentions, and an unpopular one deflates them. The campaign begins, the polls narrow, and observers scramble for reasons – for something in the news that must be responsible.
Here’s a secret: political polls are useful pointers to results when an election is on the near horizon, but their usefulness declines the further out they go. Most of the time they can’t really tell you what would have happened “if an election were held today.”
Right now in Britain, like thirty-two and a half years ago in Australia, one of the crystallising reasons is that voters have balked at May’s cynicism in calling an early election. It is the inverse of another bit of received wisdom – that in 2007 voters marked Gordon Brown down when he “bottled it” by umming and ahhing and then not holding an election when the polls were friendly.
Damned if they do, and also if they don’t, it seems.
Anyway, from the pollsters’ point of view, there’s only one poll that matters – and that’s the final one before the vote. This is where reputations are made and broken. The British ones right now are all over the shop, reflecting attempts by each to adjust for what they believe caused them to err so much in favour of Labour in 2015.
Politicians and strategists won’t be the only ones biting their nails on Thursday evening. •
