Are opinion polls bad for democracy?

Ahead of today’s general election, the German broadcaster ZDF has broken with convention and published an opinion poll in the final few days before the vote. Previously, it had been agreed that surveys would not be released fewer than ten days before polling day for fear of influencing voters. This is a worrying development because it conflicts with three of the most prominent theories about why democracy is valuable: aggregative, epistemic and deliberative theories.

According to the aggregative theory, democracy is fundamentally a procedure for reconciling the different interests and preferences of citizens. On any given issue, citizens will come with ready-formed preferences, based on their personalities and circumstances. It does not matter why people want to criminalise cannabis or ban the burqa – the only salient question is whether more people are for or against these proposals. The more popular policies are the ones that should be enacted.

However, aggregative democracy depends on people accurately reporting their preferences. If we cannot rely on voters to vote for the policies they actually want, then the vote will not reveal which proposal is genuinely more popular, the one which satisfies the most preferences. Opinion polls make it more likely that people’s votes will not reflect their true preferences. If my favoured party or policy is far behind in the poll, I may believe it is pointless to vote for them, or even to bother to vote at all. Opinion polls make tactical voting more likely – I may cast a positive vote for a policy or party that I don’t like to stop a worse one succeeding. In other words, opinion polls help to mess up the clean aggregative interpretation of elections.

A second major theory of democracy is the epistemic view. Epistemic theories see democracy as trying to find the ‘correct’ or ‘true’ answers to political questions, such as ‘What is the best set of policies to encourage economic growth?’ Epistemic theories take their inspiration from ideas such as the wisdom of crowds – the notion that the more views are canvassed on an issue, the more likely they are to converge on a correct answer.

This basic principle is formalised in the Condorect Jury Theorem. However, one of the key assumptions of the theorem is that votes must be independent – person 1 voting for A should not make it any more likely that person 2 will vote for A. Imagine if you ask two separate people for directions, and they both advise you to go right. This would ordinarily give you reasonable confidence that this advice is sound. But if you knew the second person was only copying the first, you would be more likely to seek further verification.

Opinion polls may flout the independence criterion valued by epistemic theorists of democracy, because of the momentum they create. If my instinct is to vote Pirate, but everybody else appears to be voting CDU, I am bound to wonder if I have missed something incredible in the CDU platform. Conversely, the lack of support for the Pirates is likely to make me reconsider their competence. The Asch conformity experiments show that people are susceptible to peer pressure even on beliefs they have a great deal of confidence in – these effects are liable to be even more acute in politics.

A third perspective on democracy emphasises deliberation. According to this view, democracy is as much about forming opinions as aggregating them. Thus deliberative democracy emphasises free, open and rational debate, whereby citizens attempt to criticise and refine their own views, as well as understanding those of others, in the hope of achieving consensus. 

Deliberative democrats have reason to fear opinion polls because they distract from the process of debate. They distort media coverage and public attention by encouraging passive spectatorship of a competition, rather than engagement with ideas and policies. This creates a ‘horse race’ model of politics, where tactics, strategies and who’s winning draws more focus than substantive discussion of different platforms. Moreover, deliberative democrats are wary of excessive focus on voting rather than other elements of the political process (such as deliberation), a trend exacerbated by polling.

These arguments may give opinion polls too much credit – it could well be that they have much less influence on voting behaviour than suggested here. However, if opinion polls are in fact this powerful, they may indeed be damaging for democracy.


  1. Cool article, have there been many historical examples of opinion polls resulting in a major shift in results?
    I'm pretty sure your example of parties is of Swedish ones. Been getting into reading about local politics in Swedish media? 🙂

  2. Good question. There is experimental evidence to suggest such an effect does exist (

    However I suspect real life elections are much too complicated to differentiate between the multifarious influences on victory and defeat. Moreover the effect is likely to be very small, so you'd need a very close election for opinion polls to be decisive (Of course, the closer an election is, the weaker the influence of polls on voting behaviour)

    The example parties were actually German – Germany has had a relatively prominent Pirate Party too:

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: