Which events have long-term significance for party support in an election campaign? Oliver Strijbis outlines the findings of a research project into the German elections which attempts to address this issue by using prediction markets to assess party support. Using this method he identifies three major events which significantly altered German public opinion during the campaign: the creation of the anti-euro AfD party, the TV debate between Angela Merkel and Peer Steinbrück, and an as yet unexplained swing in support from the Greens to the Left Party during August.
According to data from polls and prediction markets the German election campaign has so far seen three major shifts in the party preferences of voters. These have been produced, in turn, by the foundation of the Eurosceptic Alternative for Germany party (AfD), and the television debate between Chancellor Angela Merkel and her challenger Peer Steinbrück. The reason for a third shift, with votes going from the Greens to the Left Party, however, remains open for speculation.
For the German election we are conducting a prediction market on politikprognosen.de. A prediction market is a market that is established with the goal of predicting certain events that have yet to take place. In design, the market is like a stock market – or more correctly, like a futures market. People are invited to buy and sell contracts (i.e. “shares”), which in an election context represent the parties or candidates that are up for election. In general, prediction markets have proved to be more accurate in predicting election outcomes than opinion polls have been.
More interesting than the accuracy of the predictions, however, is that prediction markets do not measure the same thing as polls do. As is well known, polls primarily measure the current political mood. That is to say, they measure the voters’ current preferences rather than the choices that they can be expected to make on Election Day. In contrast, prediction markets directly engage in forecasts.
Activation vs. persuasion
The comparison of survey and prediction market values allows conclusions about which events in a campaign are relevant for the election outcome. One can differentiate between two particular effects of campaign events: activation and persuasion. Here, activation means the translation of party sympathies into electoral choices. Most voters are more sympathetic to one party than to others, but this does not mean that these ‘sympathisers’ actually go to the polls. Hence, the daily business of parties during an electoral campaign is to mobilise “their” voters. Since all parties mobilise their voters during an election campaign, this activation can be anticipated. Because of the anticipation of the traders at the prediction markets, their forecasts do not react to the activation process. Polls, in contrast, do react to events that merely activate voters’ predispositions, since such events impact on the current political mood.
In some cases parties not only activate their own constituency, but are also able to attract voters from other parties or non-voters. To attract new voters through persuasion is much more difficult than to activate sympathisers. And it is much harder to foresee. This is why in cases where persuasion takes place both polls and the forecasts of prediction markets react.
Hence, the joint observation of polls and prediction markets allows us to understand if a campaign event has an activating or a persuading effect. If only the values of the polls change we are confronted with an activating event. A good example of an event in which a party merely activates its own constituency is a party convention. If, in contrast, both the values of polls and prediction markets move, we run across a persuading event. Typical cases of persuading campaign events are political scandals, natural disasters or economic downturns.
Two persuading events and a mysterious shift
In order to analyse whether persuading events have taken place in the German election campaign, I have compared for the last six months the weekly averages of the prediction market with that of the polls taken from the pollsters Forsa, Emnid and Infratest. Using this method, the first occasion in which voters were ‘persuaded’ was after the foundation of the Eurosceptic Alternative for Germany party (AfD). The AfD was set up in February and held its first large convention in March. A broad audience, however, took notice of the party when it officially celebrated its foundation on April 14. The values of the polls rose sharply thereafter. Over the following three weeks the share of voters that intended to vote for the AfD rose to about three per cent. On the prediction market the forecast for “other parties” – to which the AfD was a member – increased by about one per cent (for methodological reasons politikprognosen.de abstained from adding a prediction for the AfD). Hence, since both polls and the prediction market showed an increase in the vote share for the AfD, we can deduce that it persuaded a non-trivial amount of the electorate to switch their support.
An additional important campaign event was the TV debate between the CDU’s Chancellor Angela Merkel, and her challenger Peer Steinbrück from the Social Democrats (SPD). Although the polls that were conducted during and directly after the debate showed no clear winner, the TV duel helped Peer Steinbrück. The reason is that expectations in the front runner of the SPD were so low that his personal popularity could only profit from presenting himself to the public alongside Angela Merkel. The TV debate boosted the share of voters that intended to vote for the SPD by about 1.5 per cent and its vote share forecast on the prediction market by two per cent.
A third shift in the polling and prediction market data has taken place in mid-August. Both polls and prediction markets showed an increase for the Left Party of about two per cent, and a decrease of the Greens of about the same size. Observers remain puzzled about this shift since it can hardly be related to a specific event. One might speculate that the debate about Germany’s role in the case of a military intervention in Syria helped the pacifist hardliners from the Left Party.
Alternatively, it could be argued that sympathisers of the Left Party increasingly abstain from the idea of voting for strategic reasons for the Greens. The reason may be that with Election Day coming close, they are increasingly aware that a government formed by the Greens and the SPD is unrealistic, which means that another CDU-led government cannot be prevented through tactical voting. However, since neither of these two explanations are particularly convincing, German analysts remain in search for alternative interpretations.
Note: This article gives the views of the author, and not the position of EUROPP – European Politics and Policy, nor of the London School of Economics.
Shortened URL for this post: http://bit.ly/16bo1nu
Oliver Strijbis – University of Hamburg
Oliver Strijbis is a postdoctoral researcher at the University of Hamburg and initiator of politikprognosen.de. His research interests are elections, referenda, nationalism and migration. His research has been published in Sociological Methods & Research, Party Politics, and Parliamentary Affairs among others. Currently he is preparing a book on the 2009 German election campaign.