1. Introduction
The European Union (EU) is under scrutiny by the European public and contemporary populist radical right parties (De Vries, Reference De Vries2018; Schneider, Reference Schneider2019). A gradual process coupled with exceptional crises events has contributed to the current situation. In the post-Maastricht period, the powers of European institutions have increased. However, the public responded with declining support, manifested by decreasing turnout in European Parliament elections and the rise of Eurosceptic parties (De Vries and Edwards, Reference De Vries and Edwards2009). Recent events, such as the European debt crisis in 2009, the migration crisis in 2015, and the COVID-19 pandemic in 2020, have brought new challenges. Although the European institutions have been able to respond to the crises, some of the proposed solutions (e.g., austerity measures, redistribution of refugees, and vaccination procurement) have led to a wave of popular discontent across Europe. This has not only put new pressure on the European project but also on domestic politicians, who are increasingly pressured to speak out and act against European policy-making. As a result, they have to ask themselves if, and how much, they want to engage in EU oversight.
Traditionally, Members of Parliament (MPs) have had little influence on European policy making (Auel, Reference Auel2007; Hefftler et al., Reference Hefftler, Neuhold, Rozenberg and Smith2015). Yet, specifically since the adoption of the Treaty of Lisbon—introducing the Early Warning System allowing MPs to directly engage in EU policy-making—MPs’ influence on European policy-making has been more powerful than ever. Also, national parliaments have established the European Affairs Committees to scrutinize EU decision-making and use debates and questions to engage in EU affairs (Rauh, Reference Rauh2015; Senninger, Reference Senninger2017; Winzen, Reference Winzen2017).
Figure 1 shows that MP involvement in European affairs is indeed ever more increasing, but we know little about how citizens evaluate this development. This is in stark contrast to the question of how to balance constituency service and national representation, which has received a great deal of attention in the political science literature (Butler et al., Reference Butler, Karpowitz and Pope2012; Vivyan and Wagner, Reference Vivyan and Wagner2016; Tromborg and Schwindt-Bayer, Reference Tromborg and Schwindt-Bayer2019). How much do voters actually want MPs to engage in EU oversight? Even more importantly, does this engagement matter when voters make a decision between candidates in domestic elections? To address these questions, we, first, discuss two contrasting expectations about citizen preferences over legislator effort allocation to European work, and then, we present our experimental design and findings. It is important to note that the existing literature considers MP involvement in EU affairs primarily as a means to control their own government's behavior at the EU level. Although this is an important part of the work of MPs, we also consider EU engagement that goes beyond this specific approach.
Development of MP involvement in European affairs. Note: The left panel shows the annual number of EU directives (legal acts that must be transposed by national parliaments) from 1970 to 2010. The right panel illustrates the average share of EU references per plenary statement in the German Bundestag from 1991 to 2010. Please refer to section ‘Data in Figure 1’ in the Supplementary materials for more information.
On the one hand, there are good reasons to believe that European citizens appreciate MP involvement in European affairs. During the recent EU crises, major policy decisions were made without consulting national electorates. This left many citizens with the feeling that European politics is dominated by non-elected bureaucrats. The involvement of national MPs in European politics has the potential to add legitimacy to EU decision-making (Thomassen, Reference Thomassen2009; Bellamy, Reference Bellamy2019). MPs are much closer to citizens than government representatives or European representatives and have the potential to enhance democratic accountability through visible debates about multilevel governance (Rauh and De Wilde, Reference Rauh and De Wilde2018). In addition, we know that citizens are not able to use European elections to hold EU decision-makers to account (Hobolt and Tilley, Reference Hobolt and Tilley2014). As a result, citizens might prefer more MP involvement in European affairs, which they can later punish or reward in national elections. This leads to the expectation that voters prefer MPs who become significantly involved in European politics.
On the other hand, there are reasons to believe that citizens reject MPs who spend too much effort on European politics. First, even though MPs have more rights to oversee EU politics than ever before, they still have little influence on EU decision-making. It is likely that citizens consider MP involvement in European politics a wasted effort and want them to focus on domestic affairs that they are perceived to have a more direct impact on. Second, citizens elect Members of the European Parliament (MEPs) who have the responsibility to represent citizens in EU decision-making. Even though turnout at these elections is relatively low, citizens are well aware of the European Parliament (European Commission, 2018a). The involvement of MPs in European affairs may be considered an unnecessary duplication of work given the role of MEPs. Third, European citizens have very strong feelings of belonging to their country and place of residence, but they feel relatively little attached to the EU (European Commission, 2018b). Citizens may feel that an MP who devotes a great deal of time and effort to European politics is out of touch with reality. In sum, the aforementioned arguments suggest that voters disadvantage MPs who become significantly involved in European politics.
Testing the two contrasting expectations is important because a mismatch between current MP effort allocation and citizen preferences may not only cause dissatisfaction with individual politicians but also the political system more broadly. In addition, knowing whether citizens actually put MPs who engage in European affairs at a disadvantage is relevant because it gives instructions to MPs who may wish to fulfill demands to legitimize EU decision-making. Our research thus directly contributes to recent debates calling attention to the actors that have the potential to increase public responsiveness in the EU (Meijers et al., Reference Meijers, Schneider and Zhelyazkova2019).
Although preferences of the average voter provide important information for MPs, they may also want to respond to specific segments of the electorate. Euroscepticism has become a common feature of European public opinion, and it may also impact on voters’ perception of the EU involvement of national parliamentarians. For example, Eurosceptic citizens might oppose MPs who spend time and effort on EU politics. To increase the practical relevance of our research note, we thus examine the above discussed expectations for different subgroups of the electorate. To carefully elucidate the causal relationship between MPs’ effort allocation to European politics and voter preferences, we designed a choice-based conjoint survey experiment to test our expectations.Footnote 1
2. Experimental design
Classical survey experimental designs are often seen as the “gold standard” to study causal effects in surveys. However, such designs come with the drawback of identifying one-dimensional treatments only (Hainmueller et al., Reference Hainmueller, Hopkins and Yamamoto2014). Employing such a design would limit us to a single attribute of MP effort allocation to European affairs. Yet, we do not only want to include attributes about MP effort allocation, but also additional attributes that are important to voters, such as gender and party label (Vivyan and Wagner, Reference Vivyan and Wagner2016). Conjoint experiments allow researchers to set up such a multidimensional context without the need to turn away from drawing causal conclusions.
We surveyed 993 German citizens on 6 March 2019, with respondents drawn from the Clickworker online crowd. Germany is a major player in European politics but also carries a large burden of costs caused by recent European crises. As a result, questions about legitimate decision-making and parliamentary involvement in European politics were particularly salient in Germany. Moreover, with the raise of the radical right party Alternative für Deutschland (AfD), the EU and its institutions experienced a surge in political relevance (Schmitt-Beck, Reference Schmitt-Beck2017). Given the resulting pressure for politicians, we believe that Germany constitutes an excellent case for testing our expectations. Furthermore, the German mixed electoral system has been shown to provide incentives for personal vote seeking by MPs (Gschwend and Zittel, Reference Gschwend and Zittel2015) and is therefore a valid case to study how MPs’ personal traits affect their electoral chances. We discuss the external validity of our study in the final section of this paper.
After an introductory screen, respondents were presented with five vignettes, each depicting a contest between two current MPs who run for re-election.Footnote 2 The two profiles randomly exhibit information about MP effort allocation to European affairs. The design of the EU attributes pays special attention to citizens’ limited knowledge of MP involvement in EU affairs.
In most existing research, MP involvement in EU affairs centers around questions of accountability and representation (Hefftler et al., Reference Hefftler, Neuhold, Rozenberg and Smith2015). From a scholarly perspective, it would, therefore, make sense to design attributes that show information about MPs’ debate and questioning activity and their committee membership to signal EU involvement. However, presenting such information would most likely demand too much of citizens who know very little about the daily work of MPs. As a result, we have designed attributes that represent MP involvement in EU affairs in a more intuitive sense and consider general consequences of issue involvement. First, we suggest that issue involvement will result in more expertise about the topic and show whether an MP has the most knowledge of “local,”“national,” or “European” problems. Second, we suggest that MPs who become involved in EU affairs will orient themselves toward the center of EU politics. We, therefore, present the most common reason for absent days in parliament to signal MPs’ prioritization. The three attribute levels are “meetings in the constituency,” “meetings in Berlin,” and “meetings at the EU level.” Even though voters might punish MPs with a very extreme number of absent days, we do not provide precise numbers. The reason for this is that the average MP is rarely absent, which makes the number of absent days an extraneous MP feature. Moreover, our aim is to experimentally manipulate MPs’ EU engagement. It is, therefore, less important to provide the total number of absent days and more important to provide the reasons for absent days to signal MPs’ prioritization toward different levels of politics. Third, politicians who become involved in a topic usually present ideas about its future development. Citizens are constantly presented with new policy proposals and ideas, especially during election campaigns. We suggest that MPs who are strongly involved in the EU want to influence the future development of EU policies and engage in EU policy reform. We, therefore, state the MP's effort to reform the Eurozone (“none,” “little,” “much”). The attribute does not differentiate between MP involvement in national and European policies (e.g., tax policy versus monetary policy), because such a comparison would come with the strong assumption that voters have information about the division of competences between the national and European levels and take this information into account when choosing MP profiles. Instead, we indicate EU policy engagement by showing three different levels of effort to reform the Eurozone, which constitutes a salient policy topic that is unambiguously identifiable as an EU policy topic. However, as the topic refers to just one of many possible EU policies, we address the question of generalizability in our discussion. In sum, our three attributes represent a candidate's knowledge about European politics, prioritization of European affairs, and EU policy engagement. By choosing the three attributes, we provide relevant multidimensional information about MPs’ effort allocation to domestic and European politics without presenting overly complex policy details. This is particularly important because the EU is usually not at the forefront of people's minds.
As mentioned before, the vignettes include additional MP features that have shown to be of high importance to voters. These include gender, party label, years in parliament, and motivation for candidacy. A candidate was either “male” or “female” and could be attached to either “SPD,”“CDU/CSU,” “Die Linke,” “FDP,” “AfD,” or “Die Grünen.” The years in parliament were “2,” “6,” “10,” and “14.”Footnote 3 The values for a candidate's motivation to run were “to serve the party,” “to represent ordinary people,” and “to impact personally on political decision-making.” We specifically included “to represent ordinary people” to reflect the success of the AfD, which has been understood to benefit from its populist tendencies by suggesting to represent “ordinary people.” The presentation of the various levels of these additional attributes also varied randomly.
Respondents were then asked to read the characteristics of both candidates carefully and state which candidate they would vote for. We use a factual manipulation check to assess respondent attentiveness to our manipulation.Footnote 4 Respondents were asked to answer several questions about national and European politics as well as their socio-economic and political background.Footnote 5 To address the potential trade-off between priming versus post-treatment biases (Montgomery et al., Reference Montgomery, Nyhan and Torres2018), half of the respondents were randomly presented to the vignettes along with the choice task after they had answered the additional questions. In the Supplementary materials, we show that the results of the two samples are substantially the same.
Our sample is not a nationally representative population-based sample.Footnote 6 We conducted robustness tests to rule out that sampling drives our findings. First, we re-weighted our sample on age groups, gender, level of school education, and EU preferences so that it matches the voter population in Germany (see “Sample description & re-weighting” in the Supplementary materials).Footnote 7 Second, we interacted our key treatments with covariates such as age, gender, and voting intention (see “Subgroup analysis” in the Supplementary materials). Our findings are robust to both specifications.
In sum, our study yields ten observations per respondent, one for each MP profile that respondents engage with across their five choice tasks, leading to a total of 9930 observations. Needless to say, our survey experimental design cannot replicate a real-world vote choice, but we made efforts to maintain validity by using a multidimensional and reasonably realistic choice setting.
3. Results
We, first, examine the average effect of MPs’ effort allocation to European work, and then, we investigate how this effect varies across relevant subgroups. Figure 2 presents average marginal component effect (AMCE) estimates for the levels of attributes that signal MPs engagement in European affairs as described above. The AMCE can be thought of as the degree to which a given value of a candidate feature increases or decreases respondents’ favorability toward a candidate profile relative to a baseline level, averaging across all respondents and all other profile features.Footnote 8 For each level, the dot shows the point estimate for the effect of that level relative to the baseline level. The baseline levels are “constituency” for the first attribute, “meetings in the constituency” for the second attribute, and “none” for the third attribute. The error bars represent 95 percent confidence intervals.
Average marginal component effects.
The results show that German voters do not base their voting decision on MPs’ awareness of problems, nor MPs’ most common reasons for absent days in parliament. We observe no significant differences between the baseline levels and the remaining levels of the two attributes. Importantly, this means that voters do not disadvantage MPs who allocate effort to European politics as they are not more likely to prefer candidates who know most about local and national problems over candidates who know most about European problems. Similarly, voters do not prefer MPs who are absent from parliament because of meetings at the local and national levels over MPs who are absent because of meetings at the EU level. Turning to the third attribute concerning an MP's effort to reform the Eurozone, the results show a clear aversion to MPs who show no policy engagement at all. Compared to an MP who shows no policy engagement, MPs who engage “little” or “much” are preferred by voters. These estimated effects are statistically significant and substantial in magnitude. Furthermore, the effect of policy engagement appears to be linear. An increase in MPs’ policy engagement from the baseline level “none” to “little” leads to a moderate increase in the probability that an MP profile is preferred between 1.1 and 6.0 percentage points. The point estimate shows an increase of 3.6 percentage points. However, an increase from the baseline level “none” to “much” leads to a substantial increase of between 9.6 and 14.5 percentage points. The point estimate shows an increase of 12.0 percentage points.
In addition to attributes indicating MP effort allocation to European work, respondents are presented with a range of attributes that have proven to be important in previous studies, including gender, party label, experience, and motivation for running for re-election. We find evidence that party label, experience, and a candidate's motivation matter to voters. For example, an increase in experience from the baseline level “two years” to “fourteen years” leads to an increase in the probability of choosing the candidate containing that attribute level of between 3.6 and 10.5 percentage points. The point estimate shows an increase of 7.1 percentage points. Furthermore, an MP profile that contains the motivation “to impact personally on decision-making” compared to the baseline “to serve the party” leads to an increase in the probability of choosing that profile of between 8.8 and 13.7 percentage points. The point estimate shows an increase of 11.2 percentage points. We now turn to the presentation of relevant subgroup differences.
In the upper panel of Figure 3, each dot and error bar represents average component interaction effects and 95 percent confidence intervals for citizens who perceive the EU negatively (combining the categories “very negative” and “negative”) and positively (combining the categories “very positive” and “fairly positive”). It shows that respondents belonging to the two groups differ in their probability of choosing MPs who know most about European problems. Although Eurosceptics dislike candidates who know most about European politics relative to the baseline level “constituency,” Europhiles prefer them. However, the results also highlight just how similarly the two groups react to the remaining attribute levels. Most importantly, both pro-EU and anti-EU voters prefer MPs who allocate much effort to EU policy-making over MPs who do not. This is interesting because it sheds new light on the preferences of Eurosceptics. Eurosceptics do not want politicians to build expertise on the EU; however, they still want them to become involved in EU policy. This suggests that Eurosceptics prefer MPs who observe EU policy developments and sound alarm if necessary. Crucially, Eurosceptics do not want their MPs to ignore EU policy completely.
Average component interaction effects (upper panel) and conditional marginal means (lower panel)—EU support. Note: A nested model comparison along the lines of Leeper et al. (Reference Leeper, Hobolt and Tillye2019) shows that subgroups do differ in their preferences F(14, 9759) = 2.1587, p < 0.01.
As described by Leeper et al. (Reference Leeper, Hobolt and Tillye2019), the comparison between AMCEs of different subgroups can be misleading because results are sensitive to the reference category used in the analysis. We, therefore, present conditional marginal means in the lower panel of Figure 3, which support the pattern presented in the left panel. Marginal means have a direct interpretation as probabilities. A marginal mean of 0 indicates that respondents select profiles with that attribute level with probability zero, while a marginal mean of one indicates that respondents select profiles with that attribute level with probability one.Footnote 9 In sum, we find evidence that MPs are not disadvantaged for allocating effort to European affairs as compared to local and national affairs. In addition, the level of MP reform effort in European policy-making matters to voters even when other important candidate attributes (e.g., gender and party label) are considered.
4. Discussion
Our study is the first to examine citizen preferences over MPs’ involvement in European affairs using an experimental design. We present two contrasting expectations of why citizens are likely to prefer/disadvantage MPs who become involved in European politics. Results from a pre-registered conjoint experiment in Germany show that MPs are not disadvantaged for allocating effort to European affairs as compared to local and national affairs. This holds true for MPs who know most about European problems as compared to local and national problems as well as for MPs who prioritize meetings at the EU level as compared to meetings at the national and constituency levels. This is important information for MPs who may want to spend more effort on European affairs but have not dared to do so because of the incompatibility of the EU and the left-right dimension and the resulting risk of being disadvantaged. In addition, we find that citizens generally prefer MPs who become involved in European policy-making over MPs who keep out of it.
Our finding relates to recent research showing that citizens fail to hold European representatives to account via European elections (Hobolt and Tilley, Reference Hobolt and Tilley2014) and demand responsiveness of their national representatives when they negotiate at the EU level via national elections (Schneider, Reference Schneider2020). Our research adds to these findings by suggesting that citizens welcome MP involvement in European politics to guarantee a high level of responsiveness in European decision-making. However, this does not mean that all voters want their MPs to become EU policy experts. Both Eurosceptic and Europhile voters want domestic politicians to become involved in the reform of the Eurozone, but Eurosceptics do not want MPs to build up extensive EU policy expertise at the expense of knowledge about local and national problems. Hence, Eurosceptic voters seem to dislike EU experts but appreciate MPs who engage in EU policy reform to scrutinize the policy-making process and represent voter interests.
Caution is warranted as our study has a focus on Germany. Although concerns about the legitimacy of EU decision-making are visible across Europe, it could be that preferences over MP involvement vary. Germany's image of being the leading power during the Eurocrisis could make MP involvement in the Eurozone policy area particularly desirable in the eyes of German voters. Because of party competition and spectacular court decisions, the Eurozone is a very politicized area of European policy in Germany. In other European countries, we might observe that citizens want MPs to engage in other policy areas that show similar high levels of politicization. The level of dependency on EU decision-making across policy areas could be one important explanatory factor. For example, in countries located at the external borders of the EU, citizens might want their MPs to focus on European migration policy. A final caveat of our design is that while two of our three attributes indicating MP effort allocation to European affairs distinguish between domestic and European politics, the third attribute evaluates differences between three levels of effort to reform the Eurozone. We cannot be entirely sure that our approach captures the supranational quality of the issue area or MP engagement more generally. Future research should examine whether citizens would accept a reduction of local and national policy work as a result of more European policy work.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/psrm.2021.54.
