PITY THE pollsters. Paul Perry, architect of Gallup Polls, said that to produce an accurate poll, “all one has to do is use a properly drawn sample of the electorate large enough to minimise random sampling error, get honest answers from everyone, do the questioning close enough to the time of voting to minimise changes in voting intentions, anticipate how the undecided will vote and, finally, distinguish between voters and non-voters in the electorate”.
Simple, right?
In fact, it is so complicated that political scientists have had to develop nuanced systems to refine raw poll data.
For example, a TNS-MRBI poll published in The Irish Timesbefore the last election predicted that Fianna Fáil would get 41 per cent of the vote. That was just 0.6 per cent under the actual result, which was pretty accurate.
The “pre-adjusted” figure however was 46 per cent. The adjustment was made because while those polled revealed whom they might vote for, they neglected to mention they might not vote at all. The accuracy of the poll’s final result was due to the skill of the pollster in knowing the difference between what interviewees said and how they would behave.
The challenge is to divine under which circumstances people will conceal their true intentions or fail to know their own minds. In the US, a well- known feature of polling difficulties is the Bradley effect.
In the California gubernatorial contest in 1982, Tom Bradley – an African-American – was beaten by white candidate George Deukmejian, even though Bradley was leading in all polls right up to election day. Analysts theorised that white voters fell victim to a “social desirability” effect and claimed they were voting for the black candidate because they did not want to be seen as racist.
The effect has diminished as black participation in politics has increased, but we had our own version of Bradley here.
A standard problem in Irish polls until 2007 had been an overstating of support for Fianna Fáil. Curiously, voters were more inclined to say they would vote Fianna Fáil in face-to-face polls, such as MRBI, and less so in phone polls like Red C. It is much easier to lie over the phone than face to face, so theoretically, it should have been the Red C polls that suffered more from the me-too phenomenon.
However, the overstatement in face-to-face polls makes sense if you see it in terms of “social desirability”. Those polled in person would need the approval of the pollster far more than those on the phone.
This leaves me looking at Fianna Fáil’s incredibly low polling numbers – the latest at just 14 per cent – and wondering if there is a reverse-Bradley effect operating. You have to think it’s now socially desirable to distance oneself from Fianna Fáil, but on the day will core voters secretly save their guy?
It might be wise to attribute a good proportion of “don’t knows” to those core voters who think they won’t vote for Fianna Fáil, but in the booth revert to type. Or is there a more malign force at work – the self-fulfilling nature of polls – that ensures Fianna Fáil’s vote will collapse because its meltdown in the polls renders it socially toxic?
Similarly, did the rise in Fine Gael’s numbers consolidate its position as people felt happier to associate themselves with a party traditionally underestimated in opinion polls? Bandwagon effects fit in with every theory of human behaviour. People like to back winners and avoid losers. Just as in quantum mechanics, it is impossible to separate the observer from that which is being observed.
I asked the pollster of pollsters, Prof Michael Marsh of Trinity College Dublin, about this and he assured me there was no evidence to support the contention that polls themselves influenced voter behaviour. On the other hand, there doesn’t appear to be much evidence against it either.
It’s practically impossible to come up with a test that shows how a person votes is affected by their knowledge of others’ intentions. Marsh argues that if bandwagon effects existed, Fianna Fáil in previous elections and Fine Gael today wouldn’t play down talk of overall majorities once they hit 38 per cent in the opinion polls.
It is possible there’s an Irish variation on the effect, whereby voters want to support a party that is successful, but get anxious at the prospect of too much success. Which leads you to wonder if Fine Gael will be denied an overall majority because people realised they were close to achieving an overall majority.
I can’t shake the conviction that polls affect behaviour and should be banned during campaigns. This entire campaign has been driven by the relentless publication of the latest figures. Policy hardly got a look-in. Partly that’s because of newsroom culture shock at Fine Gael’s success, partly it’s because polls are a core part of newspaper marketing.
The result is that debt restructuring has had to compete fiercely with opinion polls for space. Since the accuracy of those figures depends as much on art as science, this dominance in the debate can’t be good.
This alone increases the case for banning, or at least restricting, polls during campaigns. Maybe we should take a poll on that?
Listen to Michael Marsh’s podcast on irishtimes.com