About 3 months and a million news-years ago, the pollsters and election analysts were taking a kicking for not predicting the US elections accurately enough. Biden won the presidency by a smaller margin than predicted, and the Democrats lost seats in the House when they were projected to gain them, and lost gubernatorial and statehouse elections that they were projected to win. They also apparently lost the Senate, which they were thought likely to retake by at least a majority of 2 or 3.

The predictions look a bit better in retrospect than they did just after the 2020 election. As we know the Democrats have now just grasped the Senate majority by winning two run-off races, but this wasn't thought likely at the time. Biden did emerge as the winner, by a convincing popular vote margin.

In one of the FiveThirtyEight politics podcasts, the analyst Nate Silver defended election analysts, and by extension pollsters by saying:

  1. The polls and analyses correctly identified Biden as the winner. Hardly any (maybe none) of the polls showed Trump as the winner.
  2. Biden's vote lead over Trump was within the margin of error.
  3. It's not reasonable to expect polls months in advance to predict the result with great precision - respondents change their minds, they can't go and vote because of an emergency, and so on. The polls immediately before the election were the most accurate, and they are the ones which should be given most weight when deciding if the polls were accurate or not.

You can quibble with some of this - for instance a correct prediction months in advance of the event predicted is much more valuable than one only a few days before. But none of it establishes the case for polls per se, only the case that the polls we have are acceptable if we do want polls.

Silver also said that polls are valuable because they tell political parties what policies are popular. There's something to be said for this, but it's not a good justification for the existence of election polls, because

  1. This is issue polling, not election polling.

  2. You could argue that parties would still need election polling to see if their policies had the desired effect of increasing their support, but that's an argument for private polling. There wouldn't be any reason for news organisations to pay pollsters a lot of money to conduct election polls which anyone can read, which is what they do.

  3. Liberal democracies have existed for longer than scientific polling. There was no scientific polling until the 1936 election in the USA. There wasn't any polling until the 1945 election in the UK. Other polities, including France, Czechoslovakia, Canada, Australia, and others got on fine without polling for decades. Turnout in elections was higher back then than it is now, which casts doubt on the suggestion that polls help political parties to be more responsive to public opinion.

To be sure, all those countries had defects from a liberal democratic point of view. Britain and France had large de jure empires and significant property qualifications for some of the pre-ww2 period. The USA had a medium-sized de facto empire and restrictions on the voting rights of black people. France and Czechoslovakia were both overwhelmed by a foreign invasion.

However, the point is no-one would say those defects were caused by a want of opinion polls.

So why should we bother having them at all? Why not just wait until the election is over, when you'll know for certain who won?

  • 2
    This question seems to be based on the assumption that nobody questions that polls actually are beneficial for a liberal democracy. Would you accept a frame-challenge answer which argues that they are not beneficial, provides the reasons why and explains why they still exist?
    – Philipp
    Commented Jan 21, 2021 at 15:01
  • They might be beneficial as rough "sanity check"-type tests of the voting system. If they consistently yield the opposite of the election result, there is something systematically wrong. And as long as there are multiple pollsters, and they agree at least on the major trends, it would be difficult for a maleficiant actor to manipulate all of them.
    – Hulk
    Commented Jan 21, 2021 at 15:14
  • Are you asking why the are not banned? They exist because someone invented them and they caught on, i.e. there's a market for them now. Also, it's unclear if you contest both issue polling and pre-election (party/candidate) polling. Commented Jan 21, 2021 at 15:31
  • @Philipp Yes, that sounds interesting. Look forward to seeing it.
    – Ne Mo
    Commented Jan 21, 2021 at 15:46
  • 1
    @Fizz, no, I'm not asking why they are not banned. I'm asking what, if any, is the public-spirited reason why they ought to exist. Obviously media organisations buy them and report on them because they want to gain market share and revenue. However, if everyone only cared about making money, newspapers wouldn't exist - because they don't make money. Elections polls are often loss-making for the pollsters themselves too. They conduct them in order to gain prestige for their brand and custom for their marketing surveys, which do make money.
    – Ne Mo
    Commented Jan 21, 2021 at 15:48

2 Answers 2


One plausible argument is that information, and especially trusted authoritative neutral information, is helpful to political engagement.

On the most obvious, if you don't have that then you can't realistically fact check things. After all, if the only pollsters are directly commissioned by parties, one candidate can say "The data I've seen is that we're winning big all over the country" and that could have a suppressive effect on his discouraged opponents. The other party might counter, but really who's to say?

Moreover, having an informed expectation helps people prepare beyond just voting. Suppose that you're part of a charity that does work on a political hot potato. It's valuable to you to have some idea ahead of time which way the political winds are likely to be blowing. Let's say you're the media manager for an environmentalism charity 5 months ago. You're going to spend a few months making a film supporting your cause to float after the election. If you know it's much more likely that you'll be working under a Trump presidency, you'll perhaps have more of a defensive tone to your messaging, complaining about ecological damage to Alaska. If you know you'll likely be working under a Biden presidency, you'll take a more optimistic tone filled with hope for a future if we all pitch in.

Even for those who don't consider themselves politically engaged, the purpose of a liberal democracy is to be a society for living in. This also benefits from a chance to prepare. If you're a business owner on the US-Mexico border, whether you send your marketing guy to drum up interest among your Southern Neighbours may depend on whether you expect the next inhabitant of the Oval Office to stick a massive tariff on them. Again, it's helpful to have some idea who that next inhabitant will be.

Of course everyone should remember that the polls could be wrong, but they're better that flipping a coin.

  • I guess before Trump nobody gave much thought to this angle... Commented Jan 21, 2021 at 23:07
  • The first example is definitely quite a Trump-aware case. The other two could happen in any election.
    – Josiah
    Commented Jan 21, 2021 at 23:11
  • I think it's true that, as long as only picking the winner matters, polls are significantly better than chance even a few months prior to an election. Overall the 'informed expectation' argument makes sense wrt the utility of polls. It would be good to know if social scientists have ever produced any evidence to support that.
    – Ne Mo
    Commented Jan 25, 2021 at 14:02
  • Also occurred to me that, since elections are quite consequential, it might soften the blow a bit to know months in advance that your side isn't going to win. That might make you more prepared to accept the result and peacefully live your life, rather than trying to overthrow the other side with force. Immediately occurring rebuttal - Biden was correctly predicted to win for months, and Trump's supporters tried to take power by force. You could even argue that the adverse polls tipped them off that they might have to do that, or lose power.
    – Ne Mo
    Commented Jan 25, 2021 at 14:06

As a sort of summary on the main question, I think the political science view on pre-election polls is mildly negative (or at best mixed) in the sense that several studies observed that pre-election polls lower turnout in general, but there's also some evidence to the contrary that (published) polls can increase turnout when they announce a close election.

Also, pre-election polling also enables (better) strategic voting--and even not voting can count as such behavior in some circumstances. The question therefore devolves into a discussion whether strategic voting is good or bad--the kind of discussion that is more the realm of philosophers than political scientists.

Since you also criticized issue polling in your question, it is trivially more useful/scientific than the alternative of waiting for irate/loudest constituents to write to their representatives etc. Simply because of sampling rigor.

In some locations, including the EU, there's regular issue polling conducted by official institutions, e.g. Eurobaromter. US government grants that pay for similar things, e.g. the General Social Survey​ (GSS).

Now if you want some generic defense of polling, here's WAPORs:

Limiting the publication of opinion polls hurts everyone -- the public, the government, and even decision-makers -- because among other things, polls transmit citizens' goals, attitudes and desires, to governments and political parties. Polls give governments and parties better ability to represent and serve voters.

Somewhat "subtly" (not really), the don't quite make a distinction between issue polling and candidate/party polling in that statement...

As far as pre-election polls on candidate/party elections, what Philipp is probably trying to hint to is that they are banned in some countries within a certain time period of actual elections. According to a 2016 BBC article 38 countries have some kind of ban like that. The previously linked WAPOR paper flags four countries in which the ban is 30 days long before the election. And in ten more countries, the ban is for two weeks. In fact, a summary table summarizes this in more detail:

enter image description here

As for reasons why such bans exist, indeed it has been argued that they may demotivate voter turnout etc., so there may be some downsides to them. And there's some data to back this up:

using survey data and state-level election results from US presidential elections from 2000 to 2012 and find strong evidence that voters are influenced by polls—not in whom they vote for, but rather in whether they vote at all. [...]

Were polls wildly wrong—if they predicted a lopsided victory for a candidate whose “true” support were less than the opposing candidate— our data indicate that they could conceivably manufacture a victory for the “wrong” candidate, or at least a more competitive election than either that predicted or what would have occurred in the absence of polling.

(emphasis in original). Not measured in that paper, but conceivably:

in three-candidate races strategic voters might use predicted election results to vote strategically (cf., e.g., Cox 1997; McAllister and Studlar 1991).

Fivethrityeight actually has their own page on the topic, which highlights concurrence on that turnout issue with some French results/paper:

in 2013, researchers used a change in French law to get an idea of the potential impact of polls on voter turnout. Prior to 2005, citizens of France who lived in territories west of the country didn’t get to vote until after the mainland election had ended. Thus, they had the chance to see exit polls before they even went to cast their ballots. That changed after 2005, so researchers could compare several years worth of elections and see how knowledge of the presumed winner changed voter behavior. The result: After 2005, there was a nearly 12 percentage point increase in voter turnout. Far more people in those overseas territories voted when they didn’t already know who the winner was — a finding that has big implications for countries like the United States, where time zone differences mean voters in one part of the country can see the completed exit polls from earlier in the day.

Alas the rest of the page is somewhat less well written because it then "counters" that with the much more uncertain bandwagon effect (i.e. not just turnout, but whom those who do turn out vote for being influenced by polls). The US states paper didn't find such a bandwagon effect either, and a Dutch paper that 538 discusses also finds that even in hypothetical experiments a bandwagon effect from polls is hard to demonstrate, although possible with the "right narrative".

On the other hand, a Swiss study found a contrary effect, when polls predict a close race:

Closer elections are associated with greater turnout only when polls exist. Examining within-election variation in newspaper reporting on polls across cantons, we find that close polls increase turnout significantly more where newspapers report on them most.

Technically however, this study was on referenda and not regular elections.

Now a counter-argument to ban/blackouts is that polls may be hard to suppress nowadays:

Pollsters object that, in the event of a ban, privately commissioned polls would be commissioned and their findings published abroad and then reported in the [national] media. It would be virtually impossible to suppress them in this age of the internet.

ACE has some amusing examples of how some of this is/was done.

Since pre-election polling (better) enables strategic (aka tactical) voting (when there are 3 or more candidates), we now have to turn to that topic whether that is considered good or bad. Alas it gets pretty philosophical and not much researched in that regard (although there are tons of studies as to extent to which it happens), e.g.

it would be interesting to examine whether tactical voting can also have significant effects on citizens’ satisfaction with democracy. Although there is a vast empirical literature on the impact of electoral systems on overall satisfaction with democracy (Aarts and Thomassen 2008; Anderson and Guillory 1997; Dalton and Anderson 2011), there has been little systematic research on the microfoundations of this effect. Theoretically, tactical voters should be less satisfied than sincere voters, controlling for the winner/loser gap.

Actually, to make the matter a bit more complicated, even not voting can be considered strategic behavior in some circumstances (see The Many Faces of Strategic Voting, p. 6). So it's not even clear that absenteeism (triggered by pre-election polls) is always a bad thing, if the voter does this strategically, with the intention of affecting the election outcome that way.

And to see why strategic voting is ultimately a phislophical issue (if want to judge whether it's good or bad), I'll have to resort to such a discussion from a newspaper, as political scientists seem to eschew the issue:

Some people react strongly against this sort of consequence-based reasoning. Their stance is that voting in a general election is an opportunity to express your political viewpoint. You do that by casting your vote in a sincere way. Tactical voting is insincere and cynical since you don’t really support the candidate you’re voting for. Immanuel Kant would presumably have railed against tactical voting as it goes against his formulation of the categorical imperative: treat others as ends in themselves, never as means to an end. Instead, you use that candidate to get the desired result of a majority in parliament for your preferred party.

The insincere vote is also tantamount to a lie, since you don’t really want the chosen candidate to represent you yet your cross against their name implies that you do. What you want is for the chosen candidate to defeat the Tory candidate, and that’s not the same thing. Kant was notoriously absolutist on the ethics of lying: it’s wrong in every conceivable situation, even when a crazed axeman shows up at your house asking where your best friend is.

Against this, it could be argued that the voting procedure doesn’t require you to vote for the candidate whose party you want to win, only to put an X against the name of your chosen candidate, so tactical voting doesn’t involve a lie. You are free to choose the candidate for whatever reason you like, including to block another candidate. If you take that view, there is nothing insincere at all about tactical voting.

A utilitarian might argue that using your vote in this way is sincere, sophisticated and moral. For those who see probable outcomes rather than intentions as determining the morality of actions, someone who votes tactically wishing the best outcome for the country is taking the course of action most likely to maximise happiness – provided, of course, the pollsters are reasonably accurate about how the candidates in a constituency are faring against each other.

Yet there is the niggling worry that there might be something wrong with tactical voting. Is this just naive idealism about what political participation should be? Not necessarily. For some, the emotional cost of voting for a candidate or party they don’t sincerely believe in may be high even when they can see the logical arguments for doing so.

Although rationally justifiable, this will feel like a betrayal even if good comes from it. We can know what is ultimately for the best and still feel terrible doing it. These negative psychological factors would have to be included in any utilitarian analysis, weighing the small benefit of a single tactical vote to the outcome against the personal cost of emotional turmoil.

So even on a utilitarian analysis, tactical voting may not be the best course of action. It all depends on what kind of a person you are – and how you will feel about casting a vote for someone you don’t want elected.

  • This doesn't answer the question: why are publicly reported election polls beneficial to a liberal democracy?
    – Ne Mo
    Commented Jan 21, 2021 at 15:52
  • @NeMo: you spend some space in your long q disparaging issue polling (well, at least their accuracy). And since this was easier to address... I started with that. Commented Jan 21, 2021 at 15:54
  • It's true, I did. I'm not the downvoter, but if it's going to distract from the thing I really want to know, I'll take it out.
    – Ne Mo
    Commented Jan 21, 2021 at 16:01
  • I removed my downvote since you expanded the answer, but it still doesn't point to any benefits of election polls specifically. I suppose that could be the answer (there's no benefit, but they stick around because there's demand for them), but you don't say that explicitly.
    – divibisan
    Commented Jan 21, 2021 at 17:49
  • @divibisan: I'm honestly not sure if being able to vote strategically (which polls enable, but that needs 3+ candidates) is a plus or a minus... For now I've just offered the info. (The literature on strategic aka tactical voting is itself extensive. It seems political scientists avoid "is it good/bad" judgements on that.) Commented Jan 21, 2021 at 18:31

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .