How political opinion polls affect voter behaviour

There is little doubt politicians are influenced by opinion polls. Despite having said he is “not very interested at all in opinion polls”, Malcolm Turnbull used poll results in 2015 to declare his predecessor Tony Abbott’s reign as prime minister over:

We have lost 30 Newspolls in a row. It is clear that the people have made up their mind about Mr Abbott’s leadership.

Journalists and reporters also look at poll results. Phrases such as “knife-edge election” emerge when the election contest is seen as tight.

Many opinion polls are published during an election campaign. When the everyday punter sees or hears these estimates, many might believe that their fellow punters are indeed evenly split on their voting intentions.

The effect of political opinion polls on the individual voter is a complex matter, not least because the accuracy of polls themselves is now a matter of serious debate, given problems of access to the voters for comment.

The ‘bandwagon effect’

In 2013 ReachTEL cited results from a poll asking voters who they thought would win the federal election regardless of their own voting intention. The answer was 74.2% for the Coalition and 25.8% for Labor. Some 3% of Coalition voters thought Labor would win, and 43% of Labor voters thought the Coalition would win.

These figures are partly indicative of what is called “the bandwagon effect”, where voters who think a particular political party will win the election may end up voting for that party or candidate.

It could be argued that, in the ReachTEL example above, the Labor voters who thought the Coalition would win the election were not in fact going to vote for the Coalition.

If the social and media environment gives the impression that the election is already won, then the bandwagon effect is even greater.

Some countries impose an election silence that halts polling because of a belief that the bandwagon effect tilts the democratic process. For this reason, these countries ban the publication of polling results either outright or halt them several weeks before polling day.

India’s Election Commission put forward a proposal to ban exit polls ahead of Assembly elections held in May.

While Australia has election blackout laws pertaining to political advertising, there are no legal limits on conducting and reporting political opinion polls either before or during elections. Polling data can be presented to voters up until, and during, voting.

How does it play out?

Contagion and the bandwagon effect are not simple phenomena. There are a range of variables at play during an election and the assumption that voters are not particularly smart is not one of them. There are two types of effects: strategic voting and contagion.

Strategic voting is motivated by the intention of voters to affect which party wins the election. The move of voters to the perceived majority view is called “contagion”.

In an example set out below by economics researcher Chien-Yuan Sher, the difference between strategic voters and those backing a perceived winner shows how voting behaviour under a strategic consideration will be more complex than voting behaviour under a bandwagon effect.

For example, in a three-party election under the first-past-the-post method, Voter A and Voter B think their most-preferred candidate is going to get the fewest votes. Voter A is a strategic voter; Voter B is affected by the bandwagon effect.

Both Voter A and Voter B will tend to vote for their second-preferred candidate if their ballots are perceived to be crucial in their constituency, such as in the case where their second-preferred candidate is expected to win the election by a small margin.

If voters believe their ballots are trivial in their constituency, then Voter B will tend to vote for the candidate expected to win.

Although Voter A’s most-preferred candidate has no chance of winning, Voter A will still vote for their most-preferred candidate because a vote for their most-preferred candidate or for their second-preferred candidate cannot change the result.

A party or candidate who is trailing or doing badly in opinion polls may well be represented in news and commentary as the underdog. Interestingly, there does not appear to be an underdog effect (or “sympathy vote” effect) in elections.

No evidence of an underdog effect was found in all three UK elections between 1979 and 1987 – only the bandwagon or contagion effect was identified in the voting patterns.

Betting and polls work in similar ways

When the political opinion polls and the media climate appears to support one political party over another, there can be little doubt there will be a contagion effect.

Sports betting is very familiar with this type of effect. There is psychology in betting similar to psychology in elections. Sports betting, however, has expanded this psychology a step further by declaring election winners before the election is held.

In 2013, Sportsbet called the Australian federal election nine days before Australians voted, with banner headlines on its website declaring the win.

Political opinion polls are not always an accurate reflection of public opinion. This does not mean, though, that political opinion polls do not have an effect on voter intentions or on those trying to influence the outcome of elections, like Sportsbet.

In the current election, Labor needs to gain an additional 19 seats on a swing of 4% to win government. The contagion effect is less likely in the current environment where votes are now seen as pivotal and where the polls are predicting a shift of votes to microparties. If anything, there could be a major shift back to the major parties.