Years ago I attended a talk given by the MORI pollster Bob Worcester, who told a story about a briefing he gave to the Labour shadow cabinet in the 1980s. Worcester explained to the assembled politicians that most of the time, they were wasting their time.
The majority of voters were already committed to one party or the other, and so it didn't make sense to target all voters with the party's messages. In fact, the UK's first-past-the-post system meant that only a few voters within a few constuencies really mattered - a number in the tens of thousands.
Worcester said that when he reached this part of his presentation, he heard someone around the table splutter. It was the shadow chancellor, John Smith. "If it's that few, can't we just bribe them?".
It's an axiom of modern politics that, at any one time, only a small number of voters are persuadable. Mitt Romney, speaking in private to an audience of, presumably, the already-persuaded, famously declared that 47% of the electorate would never vote for him. He also said that there were only 5-10% of voters who were open to moving across the partisan divide. Although candidates aren't supposed to speak this way, Romney wasn't actually saying much that a political strategist would find controversial.
But are voters really as locked into their views as modern political professionals assume? The political scientist Morris Fiorina has proposed that the polarisation of America's electorate is a myth. What's really happened, he argues, is that America's political parties have become more partisan and more extreme, and voters have had no choice but to choose between them, even though most retain a high degree of ambivalence and open-mindedness on the issues.
A recently published study from Sweden offers fascinating evidence that voters aren't as fixed in their attitudes as is commonly assumed. Like the U.S., the Swedish electorate is regarded as one of the most polarised in the world, albeit a step to the left overall. In 2010, when the Swedes (like the Brits) last held a general election, pollsters estimated that only about 10% of voters were undecided between the two coalitions competing for office.
In the run-up to the election, the researchers asked people in the street if they would fill in a questionnaire concerning their views on political issues. Participants were asked to indicate how certain they were of their political views, and their current voting intention, from extremely certain social democrat/green, to extremely certain conservatives.
They were then asked to indicate their positions on twelve salient political issues on which the coalitions held opposing positions. For example, "Gasoline taxes should be increased" or "Healthcare benefits should be time-extended". In collaboration with the participants, the researchers then tallied an aggregate score, indicating which political coalition the participant favoured, based on their responses to the policy issues. Finally, the participants were asked to indicate, once again, their voting intention for the upcoming election.
All reasonably straightforward. But here's the clever bit: one group of participants were tricked. While they were filling out the questionnaire, the researcher surreptitiously filled out another form with a pattern of responses that you would expect from someone of the opposite political affiliation, mirroring the skew of the respondent's answers but from the other direction. Using a sleight-of-hand, the researcher then attached this manipulated profile on top of the participant's original answers.
So when it came time to explain and justify their answers, the participants were doing so off a sheet that showed answers that were different to the ones they actually had given. If they previously thought the gasoline tax should be raised, they might be faced with explaining why they had indicated that it ought to be lowered.
The researchers' overall aim was to shift the participant's entire score into the opposite column, so that a left-wing coalition voter would end up with the profile of a right-wing coalition voter (imagine a confirmed Tory ending up with a political profile that indicated she was a strong Labour voter).
During the discussion stage, the participants were free to change their answers if they felt they didn't reflect their original opinion, and sometimes they did so (they would assume they had misread the question or marked the wrong end of the scale). If they did so consistently, they would nullify the the researcher's effort at shifting them to the other coalition.
Remarkably, however, the participants subject to this trickery didn't, for the most part, notice the apparent inconsistencies in their answers. Nearly half didn't correct any at all and most only corrected a few. Instead, many accepted and even justified, with apparent sincerity, opinions which were the opposite of the ones they originally held, or close to it.
Most strikingly, most participants ended up endorsing overall political profiles that put them in the opposite camp to the one they thought they were in, and many changed their voting intention. 19% went from expressing certain support for one coalition to becoming undecided. A further 10% moved across the full ideological span, from firmly right wing to firmly left wing. In total, almost half of these participants were willing to consider a shift from one coalition to the other (compared to the 5-10% that are usually thought to be persuadable). In a few minutes, the researchers had achieved what political leaders spend every day of every year trying to effect.
In one sense, that voters accepted a political profile putting them in the opposite camp wasn't surprising: after all, they thought they were looking at a summary of the answers they had just given. In another sense, it is extraordinary that a person's political identity can be so easily manipulated.
The American psychologist Jonathan Haidt has argued convincingly that political affiliation is primarily about tribalism. Voters align themselves with the parties to whom they feel an emotional attachment, and are adept at inventing policy-based reasons for that attachment after the fact.
The Swedish study is a clever way of circumventing that attachment, because it moves in the opposite direction: it starts by asking voters to reason about policies and builds from there towards party affiliation. That this method can create such radically different results tells us something important about how voters relate to politics.
Politics is a more fluid business than it can appear. As the researchers put it, "the polls can be spot on about what will happen at the vote, yet dead wrong about the potential for change." Voters hold more nuanced positions, and are more open to reason, than the polls suggest or than politicians and the media tend to believe. It's not that voters have their minds firmly made up on the issues that they think their party has got right. It's that they choose their party label and assume that the issues will take care of themselves. But when they are forced to actually the consider the issues, they can be persuaded to change sides. Even without a bribe.
You can read the full study here.