We propose the inclusion of two or three political knowledge items to the 2009 voter survey. As it will be briefly indicated below, considerable evidence exists about the multiplicity of the ways in which knowledge scales can contribute to a better understanding of voting behavior. Some evidence also exists about the ways they can improve our understanding of second-order elections as well as the level and determinants of support for and opposition to European integration. They can also be extremely useful in linking micro-data from the voter study to macro-data from the media content analysis, the manifesto study, candidate, expert and MEP surveys, and to macro-economic performance indicators. The administration of such items is standard practice in many election studies, do not pose special problems, requires very little interview time, and considerable previous experience is available about the reliability, scalability, and dimensionality of knowledge items. Their inclusion in a cross-national voter study opens the door to new ways of analyzing the impact and determinants of election outcomes.
1. The previous evidence about the multiplicity of the ways in which knowledge scales can contribute to a better understanding of voting behavior concern mostly interactions between citizens’ political information level and their reliance on various shortcuts to make electoral choice in complex situations. Prior research further suggests that the same people may vote rather differently if they became better informed (Bartels 1996; Fishkin and Luskin 1999; Lau and Redlawsk 1997; Sekhon 2004), and use different criteria for decision making (Sniderman et al. 1991). Better informed citizens are most likely to anchor their vote choices in their issue preferences, ideological orientation and performance evaluations (Andersen et al. 2005; Delli Carpini and Keeter 1996, 256-8; Gomez and Wilson 2001; Goren 1997; Hobolt 2007; Jacoby 2006; Lau and Redlawsk 2001, 2006, 2008; Lupia 1994; Luskin 2003; Sniderman et al. 1991; Sturgis and Tilley 2004). Hence it is plausible that greater political knowledge gives more influence over elected office-holders (but cf. Lupia 1994; Lupia and McCubbins 1998; McKelvey and Ordeshook 1990 for some counter-arguments). The balance of the evidence would suggest that properly specified models of voting behavior always need to take into account the heterogeneity of the electorate in terms of information level. However, previous EES voter studies – with the exception of the 1994 survey – included only indirect proxies of information level like level of education and interest in politics, which, according to previous studies, do not provide adequate substitutes of knowledge scales in reliably capturing individual variation in citizens’ political sophistication (cf. e.g. Luskin 1987, 1990, 2003; Smith 1989).
2. It seems particularly relevant to include information level in the European election study series. Sufficient prior evidence exists to suggest that the level and determinants of support for and opposition to European integration are also related to citizens’ political knowledge level (cf. Hobolt 2007; Gosselin and Henjak 2008). Whether European elections operate as second-order elections as a result of direct citizens’ responses to the fact that no executive office is directly seen at stake in these elections or to a less intense campaigning by political parties has important implications for what kind of institutional reforms can reduce the second-order character of European elections, and the two alternative explanation can be best tested by models involving citizens’ political knowledge level among the independent variables (Toka 2007).
3. Since citizens can be expected to respond differently to the flow of campaign information and to have different understandings of party positions depending on their knowledge level, measures of the latter can be extremely useful in linking micro-data from the voter study to macro-data from the media content analysis (cf. Zaller 1992), the manifesto study (van der Brug 1997), candidate, expert and MEP surveys, and to macro-economic performance indicators (Tilley et al. 2006). Controlling for variation in citizens’ knowledge is also an essential variable in testing different explanations of differences and similarities between the political attitudes of citizens and their representatives. Therefore the inclusion of a knowledge battery in the voter study is particularly commendable for an election study that collects data from all these various sources and attempts to link them in innovative and theoretically sound ways.
4. The administration of such items has become standard practice in many election studies, do not pose special difficulties, requires very little interview time, and considerable previous experience is available about the reliability, scalability, and dimensionality of knowledge items (cf. e.g. Delli Carpini and Keeter 1996; Zaller 1985). There is clear evidence that, in spite of previous concerns, relatively simple, quickly administered, pre-coded open-ended quiz questions that encourage the expression of ‘DK’ responses and are easily administered over the phone provide the best available measures (Sturgis et al. 2008; Luskin and Bullock 2006). While the CSES surveys do include such items, they are not sufficiently standardized across countries and elections (cf. Kroh 2002). The EES survey would be in a good position to include cross-nationally standardized items relying on the considerable body of evidence available on the Euro-wide variation in the functioning of different knowledge items in previous Eurobarometers as well as a 17-country survey conducted in the context of the IntUne FP6 project in March 2007.
5. The inclusion of a knowledge battery in a cross-national voter study opens the door to new ways of analyzing the impact and determinants of election outcomes. As Bartels (1996), Blais et al (forthcoming), Fournier (2006), Sekhon (2004), and Toka (2003) demonstrated, considerable variation exists across elections and countries in the extent to which otherwise identical citizens vote in different ways depending on their information level. As a result, some elections produce generally better informed outcomes than some others. Understanding the causes and impact of such variation is an important task for contemporary political science, with important normative implications for institutional design and the analysis of democratic government (cf. Lau and Redlawsk 2008; Toka 2008).
Previous research found that individual political information items often work in a highly interchangeable manner. Therefore the exact wording of the individual items included in the study is of no major relevance for the study of the above questions. The attached Excel file shows a sample of items previously used in Eurobarometer and EES surveys. We believe that pre-coded open-ended questions on, say, who president of the European Commission is, how many member states there are in the EU, and what percentage of the EU budget is spent on the common agricultural policy would probably be best suited for the nature of the EES and its specific analytical focus on European elections. The listed previous studies provide some evidence about the likely distribution of responses to such items.
Gabor Toka, Central European University, firstname.lastname@example.org
Sara Hobolt, University of Oxford, email@example.com
P.S. for the full list of references, please email Gabor Toka at firstname.lastname@example.org