The comfortable leads for Obama/Biden recorded by every national poll in the 2008 election in the past month could be giving a misleading view of the intentions of the voters, for a variety of reasons.
This has certainly happened in the past. In the UK in 1992, the Conservative Party comfortably won the General Election, with a 7.6 per cent lead in the popular vote, despite national polls all suggesting near equal support for both major parties or a slight lead for Labour. The reasons why this happened were investigated by a panel of academics and pollsters led by David Butler, appointed by the Market Research Society.
As well as the obvious culprits of late swings in a brief election campaign at a volatile period, poor sampling techniques and faulty demographic models, the panel identified a phenomenon in mass communications theory first formulated by Elisabeth Noelle-Neumann, known as the “spiral of silence“. According to this model of human behavior, people are less likely to publicly voice ideas that they believe are unpopular. Because polls and news reports had been showing the negative feeling about the Conservative government for some years, in 1992 the thinking goes that there was a general unwillingness to express support for that party. In such circumstances, the tendency is to refuse to take part in the poll, to refuse to respond to the question about voting intention, or to reply “don’t know” to that question. The result of all these actions is exclusion from the comparisons of the parties’ popularity. Analysis of the 1992 results suggested that a disproportionately high number of those who were omitted had previously voted Conservative, admired the Conservative leader John Major more than the Labour leader Neil Kinnock, or (during the 1992 recession) thought the Conservatives would manage the economy better. The phenomenon was dubbed the “Shy Conservative” (or Shy Tory) effect by some commentators.
Has this happened in the current election campaign? It’s hard to tell. On one hand, polls over the past month are relatively decisive in suggesting, for the most part, a clear Obama/Biden win, so it seems unlikely that a “shy conservative” effect alone could lead to a significant upset. But in conjunction with other effects, it could reversal of the result predicted at present (in late October).
One question that springs to mind is the effect of the saturation advertising of the Obama campaign. One reporter writes that even in solidly Republican South Carolina, which Obama surely cannot hope to carry, it is almost impossible to miss Obama’s advertising. The national media have also, I think it’s fair to say, carried a strong poll-driven message to the effect that Obama is expected to win decisively. In latter weeks, they have even begun to predict that several red states will turn blue for the first time in a generation. The effect of this advertising seems to have produced a positive feedback effect on responses to opinion pollsters. But it’s one thing to say you’re going to change the habits of a lifetime, quite another, in the privacy of the polling booth, to go through with it. It’s a secret ballot.
The McCain/Palin campaign has been widely criticised, not least in Republican circles, but in the last few days the focus will be on Obama/Biden, and those who have told pollsters in the last few weeks that they will vote for the Democratic team may find any number of reasons to think again. Perhaps the most convincing is that the next Congress is almost certain to be overwhelmingly Democrat, and traditional wisdom has it that American government works best when the executive and the legislature are in opposition. It is quite possible that, on this reason alone, enough voters in the “battleground” states may hold their noses, set their qualms to one side, and pull the lever for McCain/Palin.