Media’s coverage of opinion polls significant – but not infrequently misleading
The media’s coverage of voter opinion polls does not explain the statistical uncertainties, and the results can also influence other political reporting. This is seen in a thesis investigating Swedish news media’s use of opinion polls and the consequences for voters and politicians.
Media coverage of opinion polls often turns to the metaphors of sports journalism, depicting them as an ongoing competition with both winners and losers. However, the polls differ from reporting on sports and competitions in one important respect: their central role for democracy.
Media researcher Per Oleskog Tryggvason at the Department of Journalism, Media and Communication (JMG) has, among other things, studied how Swedish news media report on opinion poll results and how they behave as regards statistical uncertainties.
“Little respect for statistical uncertainties”
He has reviewed all headline reports in which the Swedish daily newspapers Dagens Nyheter, Svenska Dagbladet and Expressen presented their opinion polls in 2010/2011 and 2014/2015.
“The results show that journalists have very little respect for statistical uncertainties in their reporting. In half of the cases I’ve studied where the journalist provides an explanation as to why a party has risen or fallen in the opinion polls, such as due to the performance of the party leader in a debate, the change to be explained is so small that it could simply be a matter of coincidence. The media logic of not missing a potential story seems to trump other journalistic norms, such as maintaining a critical eye on the source material being used,” he says.
Can also influence other political reporting
Another result from the thesis is that media coverage of opinion poll results risks influencing the tone of other reporting on the concerned political parties.
“There seems to be a spillover effect. This means that the party which has climbed in the latest opinion polls is also favoured in other news items and events, receiving more positive coverage the following days.”
Polls can influence the opinion they attempt to measure
Opinion poll news can also have consequences for voters through the bandwagon effect, which means that people’s opinions and behaviours can be influenced by what they believe to be considered popular by others.
By following almost 1,900 voters in the months prior to the 2018 election campaign, Tryggvason has analysed whether voters’ views on how the parties are performing in the opinion polls is of significance to the parties’ internal evaluations and nominations.
“The results show that voters who consider a party to have climbed in the opinion polls have a more favourable view of that party, and in several cases are also more inclined to vote for them.”
Considerable importance for enthusiasm
Media coverage of opinion poll numbers is even of considerable importance to the parties themselves, according to Tryggvason’s questionnaire encompassing more than 2,400 politicians.
According to the politicians, the greatest significance of opinion polls is how they affect the media’s reporting, how they portray the parties. However, the news media’s opinion poll results coverage is also considered to have a strong influence internally, such as on the level of enthusiasm among the members and the standing held by the party leader.
“The media’s use of opinion polls can have a number of positive consequences for our democratic process, such as by raising the political issues the public currently considers most important. However, my results underline the importance of journalists being aware that these polls can have consequences. This means it’s important that both the actual polls and the journalism used to communicate the results are of good quality,” says Tryggvason.
Per Oleskog Tryggvason, email: firstname.lastname@example.org, phone: 070-992 3826
The thesis Under the Influence? Understanding Media’s Coverage of Opinion Polls and their Effects on Citizens and Politicians is comprised of four separate studies in which the author has used different types of media data, a voter panel and an extensive questionnaire encompassing more than 2,400 Swedish politicians.
http://Digital publication: http://hdl.handle.net/2077/67096