How democratic are pre-election polls?

In most Western democracies nowadays, pre-election periods are littered with polls. Some polls, conducted by polling organizations, are sophisticated and more likely to be challenged for their accuracy (as are the media houses that publish them). Other polls are simple. For instance, a news website may ask its readers who they would vote for if the election happened on that day. Polls represent a simple and cheap commodity for the commercial news media to offer to their audiences. As Jesper Strömbäck notes, polls generate fresh and often dramatic news items that are easy to analyze for journalists and easy to digest for audiences.

But how do polls, and particularly pre-election polls, fit into a normative vision of democracy? Do they enrich our democratic practices and institutions, or do they undercut democratic ideals? Despite being an epitome for divisive issues (49% of countries restrict the publishing of pre-election polls in some capacity, as Petersen notes), pre-election polls have attracted little interest of democratic theorists. Reaching a verdict on whether they are normatively compatible with democracy has been left almost entirely to political scientists and journalists.

On the face of it, pre-election polls have a positive case going for them:

• First, as long as their integrity is not in question, publishing pre-election polls in liberal democracies could be seen simply as what freedoms of the press and expression require.
• Second, pre-election polls provide voters with factual information that is seemingly relevant for reaching a decision on how to cast the vote.
• Third, the publishing of pre-election polls, assuming that they significantly deviate from electoral results, can help indicate whether an election has been tampered with.

Democratic theorists have their work cut out for them to discuss the strength of these three points. But assume, for the moment, that we grant their initial appeal. If these points spoke decisively in favor of pre-election polls, then deciding whether polls are democratic or not would hinge on how accurately they predict outcomes. It is only the publishing of accurate information that would be non-controversially protected under press freedom, contribute to informed voting, or reliably signal election tampering. And matters of polling accuracy are exactly the direction in which much of political science and journalistic takes on pre-election polls have steered the discussion. Insofar, our greatest fears about polls would be whether they are conducted poorly, or rigged.

On second thoughts, however, it doesn’t seem that polling accuracy should be the whole story of how democratic pre-election polls are. Here are two reasons why.

The first concerns the effects of publishing pre-election polls on voting behavior. How exactly voters will be affected remains inconclusive, despite nearly 50 years of research in the social sciences. For instance, it is uncertain whether a pre-election poll will bring about the bandwagon effect – causing voters to flock to the winning side – or the underdog effect – causing the exact opposite. Some authors believe these effects may in fact cancel each other out.

But not everything is uncertain about polling effects. Exposure to pre-election polls is sure to make voters more likely to “maximize the utility of their vote in producing a favorable election outcome”, according to Moy and Rinke. This means that if the poll informs the voter that her preferred option is unlikely to succeed, she will opt for a “second-best” or “third-best” option. Not everyone will succumb to this “strategic voting” effect, but those who do will often solidify two-party systems, since strategic voting is more likely to favor strong parties and candidates at the expense of the weak. If true, obstacles to entry become more significant for democratic newcomers, and voters, who have changed their voting preferences but not their political attitudes, may find themselves unrepresented. These effects, as well as voter indifference, which may be caused if polls project a significant margin between candidates, are all arguably undesirable occurrences in a democracy.

The second democratic worry about pre-election polls concerns the kind of information that these polls provide and how it affects voters. Many authors have linked the media coverage of polls with the culture of so-called “horse race journalism.” This kind of media content primarily emphasizes the standing of candidates in the race and their popularity compared to the competition. And since a democratic arena saturated with polls will direct most of the voter’s attention on the horse race, little attention remains to consider important political views and policy proposals. Along similar lines, Geoffrey Brennan and Philip Pettit once argued that polling effects, among other factors, represent “capricious influences” on voting behavior, and that we should come up with institutional pressures to downplay caprice and stimulate the democratic exchange of reasons that precedes voting.

Of course, pre-election polls are not the only driver of strategic voting, or the sole source of horse race journalism. But if they exacerbate these effects, then we should do more to determine whether and how they fit into a desirable vision of democracy. The institutional pressures against their “capricious influences” need not be strict bans, but may be inspired by numerous policy solutions not yet observed in the West.