Voters are generally assumed to support parties in elections on the basis of their policy positions, but what factors affect how accurately voters can judge these positions? Peter Grand and Guido Tiemann assess the influence of so called ‘projection effects’ in European Parliament elections, whereby voters who like a particular party for non-political reasons are more likely to assume that the party is closer to their political views than may be the case in reality (and vice versa). Using data from the 2009 European Parliament elections, they illustrate that this phenomenon can play a significant role in voters’ judgement of parties.
This article was originally published on our sister blog, EUROPP.
One prominent theoretical explanation for how voters cast their ballot stems from spatial voting theory: what is called proximity voting. Spatial voting theory assumes that we can position voters and candidates/parties in a political space, for instance along a line like the positioning of parties on a left-right dimension or a liberal-conservative dimension. Proximity voting assumes that voters know their own position in a political space and the position of the respective candidates or parties. A voter will cast their ballot for the party closest to their own position.
However, it is hard for human beings to be objective and honestly position parties. Or rather, vote choice is not only determined by the issue positions or policy proposals of political parties but also, for example, how voters like/dislike the leading candidate of a party. In other words, there are also non-spatial considerations influencing one’s vote choice. The effects of these considerations are often referred to as ‘projection effects’, with voters more likely to assume candidates they like have positions closer to their own (a so called ‘assimilation’ effect), while voters who dislike certain candidates are more likely to assume the candidate’s position is further away from their own (a ‘contrast’ effect). Projection effects are not only an interesting political phenomenon per se, but may also distort party positioning and therefore bias the results of spatial voting theories.
Explanations for projection effects build on consistency perspectives positing that individuals try to overcome and work against cognitive dissonance; predominant are dissonance theory, balance, and congruity theory. Individuals experience cognitive inconsistency if two cognitions do not correspond, and the perceived discomfort increases with the level of substantive importance attributed to these cognitions. Cognitive dissonance is also at work if, for instance, one buys a car while not sure if exactly this car has been the optimal or right decision. To overcome this cognitive dissonance those individuals often return to the respective advertisement brochure in order to confirm their decision.
Political scientists have collected valid and robust evidence for the presence and consequences of projection effects. Most studies have focused on US elections and used party identification. This is a binary indicator, with voters indicating if they like or dislike a political candidate/party. We deviate from this research by focusing on European Parliament elections, where most of the party systems of EU member states are multi-party systems compared to the US two-party system, which implies that European voters face a much more fragmented party system.
Further, we use a measurement called ‘propensity to vote’. This question has been asked in the European Election Study (EES) 2009 and indicates how likely it is that the respondent will ever vote for a specific party. This is not a binary measure, but instead ranges from 0 to 10. Further, the EES 2009 allows us to analyse projection effects not only on the left-right dimension, but also on a pro-anti-European integration dimension.
To assess projection effects we proceed in two steps. First, as a necessary condition, we look at the extent of assimilation and contrast effects according to the individual propensity to vote. Secondly, we analyse those effects dependent on characteristics of the political system of the respective EU member state. Thus, we have to assess the extent of projection effects, where we can use propensity to vote as a categorical or as a continuous variable. The results of both interpretations are shown in the Chart below.
Chart: Link between propensity to vote and projection effects
Note: Each chart shows the mean trajectories for projection bias relevant to two different dimensions (left-right, shown on the left chart; and European integration, shown in the right chart). The vertical axis shows a measure of the difference between voters’ perception of a party’s position and their actual position. The horizontal axis shows how likely they were to ‘ever vote’ for this party on a scale from 0 (never) to 10 (likely). So for instance the chart on the left shows that those who had a very low ‘propensity to vote’ for a party were likely to anticipate that this party was further away from their position on left-right issues than was actually the case. The charts each contain two different types of statistical display: (1) the thin solid line and the grey shaded area specify the predicted values and the 95 per cent confidence interval from the categorical specification; (2) the thick dashed line and the thin dashed lines represent the predicted values and confidence intervals from the continuous specification. The dashed horizontal line separates contrast (below) from assimilation effects (above).
The Chart illustrates the extent of projection effects dependent on an individual’s propensity to vote, clearly showing that the lower the propensity to vote the more voters ‘push’ parties away (i.e. anticipate that the party’s position on an issue is further away from their own position than it is in reality) and the higher the propensity to vote, the more voters ‘pull’ parties to their own position. However, as political scientists we cannot stop at describing a political phenomenon, but have also to explain it.
The EES 2009 comprises data about the then 27 EU member states and because it is likely plausible that projection effects are not only influenced by voter characteristics but also on idiosyncracies of the respective political systems, we have to control for country effects. Statistically, this is done using multilevel regression models where we can include effects at the individual, the party, and the party system level. Further, we distinguish between unsystematic and systematic projection effects, where only systematic projection effects would really distort spatial voting considerations. Consequently, we investigate on the voter level the effects of voter sophistication and voter extremism. On the party level we consider party extremism, party ambiguity, and incumbency. Finally, on the party level we analyse party system fragmentation and party system polarisation.
The results of our statistical analysis show that projection effects are more predominant on the left-right dimension compared to the pro/anti European integration dimension. Further, while the well-established patterns of assimilation and contrast play out on both dimensions, our novel empirical operationalisation, which builds on the fine-grained propensity to vote as the key independent variable, reveals that contrast appears to outweigh assimilation effects in multiparty electoral competition.
Our analysis ultimately aimed to explore the causal determinants of assimilation and contrast. We find much more evidence for the lack of than for the significance of causal and contextual effects. By and large projection effects appear to be unsystematic and context-independent rather than systematic and context-dependent. In the realm of practical politics, projection effects may easily affect an individual’s vote choice and facilitate a spill-over from individual non-policy to policy benefits.
Overall, the analysis identifies some empirical determinants of the mismatch between subjective and objective views of the positions of parties (i.e. the difference between the positions voters think parties hold, and the positions parties actually hold). In particular, low levels of voter sophistication, extremist voter ideologies, programmatic ambiguity and discordance within political parties contribute to errors in voters’ perception of parties’ positions. However, our analysis failed to find any specific interaction between these (and other) context effects on the association between this kind of error in judgement and an individual’s propensity to vote. Projection effects therefore appear to occur rather independently of socio-political and institutional contexts. In other words, assimilation and contrast are unsystematic rather than systematic effects.
Note: This article gives the views of the author, and not the position of the British Politics and Policy blog, nor of the London School of Economics. Please read our comments policy before posting.
Peter Grand – Vienna Institute for Advanced Studies
Peter Grand is a researcher at the Department of Political Science at the Vienna Institute for Advanced Studies. His research interests include European Integration, Electoral Behaviour, and Party Politics.
Guido Tiemann – Vienna Institute for Advanced Studies
Guido Tiemann is an Assistant Professor at the Department of Political Science, Institute for Advanced Studies in Vienna. Prior to joining the department, he was a Postdoctoral Fellow at the Max Planck Institute for the Study of Societies in Cologne and at the European University Viadrina, Frankfurt/ Oder). His fields of specialization are Political Methodology, Electoral Systems, Party Systems, Electoral Behaviour, and European Integration.