Candidate Selection and Parliamentary Activity in the EU’s Multi-Level System: Opening a Black-Box

Members of the European Parliament (MEPs) have a multitude of parliamentary duties and, accordingly, have to prioritize some parliamentary activities over others. So far, we know comparably little about this prioritization process. Based on principal–agent theory, we argue first, that MEPs’ parliamentary activities are systematically determined by the “visibility” and usefulness of parliamentary instruments for their key principal; second, we expect the exclusiveness of candidate selection procedures of an MEP’s national party—the nomination and the final list placement—to determine her/his key principal (i.e., elites or members of national parties). Combining multi-level mixed effects linear regression models and expert interviews, we show that MEPs who are nominated and whose final list placement is decided by an exclusive circle of national party elites prioritize speeches, whereas MEPs who are nominated or whose final list placement is decided by more inclusive procedures prioritize written questions and opinions or reports. In other words, speeches seem particularly useful to communicate with national party elites, while other activities are used to serve larger groups of party members. These findings open up the black-box of the “national party principal” and illustrate how a complex principal–agent relationship stimulates very specific parliamentary activity patterns in the EU’s multi-level system.


Introduction
Legislators have a large number of duties but only a limited amount of time. Therefore, they have to prioritize some parliamentary activities over others. This is a particularly severe problem for the Members of the European Parliament (MEPs). They act in a multi-level setting, have to bridge large geographical distances and to serve multiple principals on different political levels. MEPs mainly operate between their European Party Group (EPG) and their national party (e.g., Hix & Høyland, 2014). The na-tional party nominates candidates for (re-)election to the European Parliament (EP) and is also instrumental in elections for future domestic positions; the EPG controls a range of offices and benefits within the EP, including committee membership, chairmanships, positions in the party group hierarchy, rapporteurships and speaking time, and is therefore key to political success in parliament (Koop, Reh, & Bressanelli, 2018, p. 563;Kreppel, 2002). To disentangle this complex principal-agent relationship and its influence on MEPs' prioritization strategy in terms of parliamentary activities, we study the candidate selection procedure of national parties, which is known to affect parliamentary activity in national settings (Fernandes, Won, & Martins, 2019; for an overview see Hazan, 2014). In doing so, we distinguish between two steps-the nomination and the final decision on list placement-in the candidate selection procedure of a national party, and ask: Which parliamentary activities do MEPs prioritize and how does the specific candidate selection procedure (including nomination and final list placement) of national parties influence MEPs' prioritization strategy?
Based on the principal-agent theory, we argue that MEPs' prioritization strategy in terms of parliamentary activities depends on the exclusiveness of candidate selection procedures the MEPs' national party employs and, thus, on the key principal the MEP serves (i.e., party elites, party members). The main reason for this is that activities vary in their "visibility" and their utility to facilitate the interests of different principals (Klüver & Spoon, 2015). Some parliamentary activities are more suitable to "cultivate a personal vote" and thus, to communicate national or individual interests than others. Which specific interests an MEP wants to serve is determined by two components of the candidates selectin procedure: (1) the nomination of candidates; and (2) the decision over the final list placement of candidates. Thus, both components have an effect into the same direction, and their combination determines the overall effect size of candidate selection procedures on MEPs' prioritization strategy in terms of parliamentary activities.
We explore this novel argument based on a new and comprehensive dataset that includes four important but very different parliamentary activities (i.e., speeches, written questions, written opinions, and reports) of all MEPs, their personal characteristics and offices within the 7th legislative period of the EP (2009)(2010)(2011)(2012)(2013)(2014), as well as expert interviews.
Overall, we find considerable support for our theoretical expectations: The choice to prioritize some parliamentary activities over others varies significantly with the exclusiveness of candidate selection procedures of national parties and thus, with the key principal a MEP serves. In detail, if candidate selection processes are organized exclusively, MEPs tend to prioritize speeches; if they are organized rather inclusively, MEPs tend to engage more in written questions, opinions, or reports. Furthermore, we show that exclusiveness of candidate selection sometimes varies across the two stages of the selection process, namely nomination and decision on final list placement. Accordingly, it is essential to take both stages into account when evaluating the impact of candidates' selection procedures on MEPs' parliamentary activities.
In consequence, this study contributes to key questions in two larger literature steams: First, it opens up the black-box of the "national party principal" in the EP by exploring how variations along the inclusion-exclusion dimension in parties' candidate selection procedures influ-ence the type of activity MEPs engage. Thus, it clearly speaks to the increasing "politics turn" in EU studies (Braun, Gross, & Rittberger, 2020). Second, it speaks to a growing comparative literature analyzing the link between candidate selection procedures in political parties and legislative behavior. The focus of our research is, therefore, a relevant one for both EU studies and Comparative Politics. Finally, this is one of the first empirical studies to comparatively investigate the consequences of party recruitment procedures on the prioritization of parliamentary activities, namely across different political parties in EU member states. The EP offers an ideal testing ground, since elections are held simultaneously across the 28 EU member states but feature a broad variety of different electoral rules. More importantly, all MEPs face similar procedural rules with regard to each parliamentary activity, facilitating the comparison of parliamentary engagement.
The next section reviews the existing literature and elaborates the theoretical framework. Subsequently, we describe the measurement, data, and research design. Next, we provide a descriptive overview of the dataset and the most important relationships. Finally, we discuss the results and provide some conclusions.

Prioritization of Parliamentary Activities in a Multi-Level Setting: Exploring the Role of Political Parties' Selection Procedures
EU studies have extensively discussed the principalagent problems of MEPs. Many scholars who explore the dilemma of managing two or more principals with different interests focus on roll-call votes and, hence, on an instance in which MEPs explicitly have to take a position and where shirking this responsibility may lead to sanctions imposed by one or the other principal (Faas, 2003;Hix, 2004;Koop et al., 2018;Lindstaedt, Slapin, & van der Wielen, 2011;Meserve, Robbins, & Thames, 2017). These scholars illustrate that MEPs and especially EPGs behave in a more cohesive manner during recorded votes than their party manifestos or expert surveys would predict (Hix & Høyland, 2013, p. 181). However, when the EPG and the national party have different standpoints, MEPs are more likely to vote in line with the position of their national party, particularly if electoral rules promote a close relationship between the latter actors (Faas, 2003;Hix, 2004;Meserve et al., 2017).
Even though research on the effect of multiple principals on voting cohesion is extensive, we have little detailed information on how this dilemma of serving multiple principals and the pressure to prioritize some tasks over others affect MEPs' parliamentary activities in earlier stages of the legislative process. The few exceptions focus mainly on speeches and written questions (Font & Pérez Durán, 2016;Jensen, Proksch, & Slapin, 2013;Proksch & Slapin, 2011;Slapin & Proksch, 2010;Wonka & Rittberger, 2014), while other parliamentary activities are largely disregarded (e.g., reports, opinions, motions for resolution). Speeches and written questions are usually exercised on an individual basis and therefore, require a relatively limited effort of coordination and time compared to more substantial activities such as reports or opinions. The latter activities are not only more timeconsuming but also more difficult to prioritize on an individual basis as the allocation of reports (and opinions) follows a very complex procedure and requires strong inter-and intra-party coordination (cf. EP, 2014, Title V, Chapter 3-5). Beyond the comparably limited number of studies exploring the principal-agent dilemma of MEPs with regard to earlier stages of the legislative process, no study so far-at least to our knowledge-offers a broader overview of MEPs' parliamentary activities and hence, allows us to understand how this dilemma plays out in MEPs' prioritization strategies in terms of parliamentary tasks and especially how it is related with the candidates selection procedures of national parties.
The literature on legislative behavior in national settings is very inspiring in that regard, as a growing number of scholars analyze the link between the candidate selection processes of political parties and legislative behavior (for an overview see Hazan, 2014). Although this community tends to concentrate on roll-call votes, it devotes more attention toward individual parliamentary activities and thus, provides a more fine-tuned picture of the motives driving MPs' individual behavior (for written questions, see Martin, 2011;Rozenberg & Martin, 2011; for speeches see Bäck & Debus, 2016;Fernandes et al., 2019). A general finding is that the more inclusive the candidate selection process, the lower the level of party unity because legislators are not only liable to party leaders but also to a larger group of party members. Moreover, candidate selection procedures seem to predict legislative behavior more precisely than electoral rules (Hazan, 2014, p. 219).
These insights provide valuable theoretical ground for conceptualizing the effect of candidate selection procedures on different parliamentary activities in the multilevel system of the EU: our overarching argument is that the exclusiveness of candidate selection procedures of national parties determines MEPs' prioritization strategy in terms of parliamentary instruments. In detail, we first propose that some parliamentary activities are more suitable to "cultivate a personal vote" and thus to communicate national or individual interests than others. The main reason is that we assume parliamentary activities to vary in their "visibility" and utility to respond to the interests of different principals (Klüver & Spoon, 2015). The contact of national party actors to the European level is often ad hoc in nature (see Pittoors, 2020) and, therefore, these actors are in general less attentive towards European parliamentary activities than the EPG. We assume that this attentiveness varies among national party elites and national party members. National party elites might be much better informed over legislative processes in the EP due to the constant influx of EU legislation and the extensive formal and informal coordina-tion between the national party and its representatives on the EU level. Members of national parties, by contrast, may find it much more difficult to monitor EP legislation and might be more interested in very specific topics affecting their electoral districts. However, this does not mean that MEPs cannot directly serve national party members; written questions, for instance, are a useful tool to gather very specific but also very valuable information for a large group of party affiliates. Accordingly, we assume that not only the visibility but also the utility of parliamentary instruments to communicate with national principals vary systematically across elites and individual members of the national parties. This leads us to argue more generally that the exclusiveness of candidate selection procedures of a national party should affect MEPs' prioritization strategy in terms of parliamentary instruments.
Second, we argue that the exclusiveness of candidate selection procedures should be captured along two steps: (1) the nomination of candidates, and (2) the decision on list placement (cf. Fortin-Rittberger & Rittberger, 2015). Both steps are integral parts of the candidate selection process, might be organized more exclusive or inclusive in manner and thus, determine the overall identity of the principal and the power the national party exercises in its role as principal over MEPs. Hence, the effect of both stages goes into the same direction and the combination of both determines the overall effect size of the exclusiveness of candidates' selection procedures. For instance, MEPs could be nominated by the party executive and the party executive could also take the decision on the final list placement. Under these circumstances, the candidates' selection procedures are very exclusive and thus, the incentive for MEPs to serve elites of the national party is very strong. Alternatively, MEPs could be nominated by the party executive but the decision on the final list placement could be taken by many party delegates. In this case, the overall exclusiveness of candidates' selection procedures is intermediate in nature and thus, MEPs may try to communicate with both, party elites and party members (for more details on the coding of the respective variables, see Section 3).
In the following, we develop three specific expectations, theorizing in more detail on how the exclusiveness of the candidate selection procedures of national parties affects MEPs' prioritization of specific parliamentary activities (i.e., speeches, written questions, and reports/opinions).
First, we expect candidate selection procedures within national political parties to impact on the speechmaking activity of MEPs. MEPs use speeches not only to present policy plans in the plenary, but also to explain positions that deviate from either their EPG group or their national party (Slapin & Proksch, 2010). Slapin and Proksch established that speeches are a common instrument to demonstrate national loyalty because MEPs may use this parliamentary instrument to "explain their national party's position to other members of their EP political group, and to create a positive record for themselves in the eyes of the national party to serve their own re-election" (Slapin & Proksch, 2010, p. 333).
For creating such a positive record in the eyes of the national party, however, two questions are important: (1) Who is listening to these speeches? ("visibility"); and (2) who is able to substantially shape re-election within the party? ("closeness" to the principals). First, one could expect that political elites such as party leaders or the executive committee are much more informed about EU politics and especially about the activity of the MEPs of their own party through different coordinating committees (e.g., Verbindungsbüros) than ordinary party members. If these party elites have a strong say in the candidate nomination procedure, meaning that the process is rather exclusive, then it is highly likely that MEPs would attempt to send positive signals to their national principals through legislative speeches in order to increase their chances of re-nomination. If the candidate selection process is organized more inclusively, meaning that many individual or selected members are involved in re-nomination and decision making over candidate lists, then it is rather unlikely that the messages sent via speeches in the EP will reach the intended recipients. Accordingly, MEPs aiming to communicate with individuals or selected groups of members should be more likely to engage in other activities than speeches. Based on these reflections, we expect the following: Expectation 1: The more exclusive the candidate selection process for European elections within a national political party, the more likely it becomes that MEPs prioritize speeches over other parliamentary activities.
Furthermore, we argue that the candidate selection procedure also impacts MEPs' prioritization strategies in terms of written questions. Specifically, we expect that the more exclusive the candidate selection process, the less likely it becomes that MEPs systematically prioritize written questions over other parliamentary activities. Scholars detected that written questions serve national parties that are in the opposition by collecting valuable information from the European Commission (Wonka & Rittberger, 2014) or by alerting the European Commission to failures of national governments in implementing EU policy issues. Similarly, we argue that written questions are a valuable tool also for individual MEPs. Written questions can be used to raise important concerns or gather very specific information being of local interest or relevant for specific subnational groups. Hence, they are useful to serve in particular individual members of national parties or selected groups of party members. The targeted (local) principal does not need to be a fully attentive expert on EU politics when it comes to parliamentary questions. The MEP or her/his assistant can selectively disseminate the gathered information. Moreover, written questions are hardly controlled by EPG leadership or the leadership of the national party in the EP, unlike most other activities. Furthermore, they are particularly useful to capture individual prioritization strategies of MEPs as proposing such questions requires a certain amount of resources and time (e.g., a staffer must research the question, format it appropriately, submit it, await a reply, and communicate this accordingly; Martin, 2011, p. 263). Accordingly, we hypothesize: Expectation 2: The more exclusive the candidate selection process for European elections within a national political party, the less likely it becomes that MEPs prioritize written questions over other parliamentary activities.
Finally, we do not expect the candidate selection procedures of national parties to systematically affect MEPs' prioritization strategy in terms of opinions and reports. The allocation of opinions and reports follows a very complex procedure, influenced by the power of individual MEPs but also other factors such as expertise, party group size, or seniority (Hix & Høyland, 2013, 2014. Thus, the prioritization of reports and opinions is by no means an individual choice. Moreover, reports and opinions might be visible to both the EPG and the elites of the national party and, more importantly, it is very difficult to push through specific national or party interests, as both reports and opinions require the support of MEPs from other countries and national parties (except if the national party is very large in number; Mamadouh & Raunio, 2003). Similarly, obtaining a powerful position within the EP (e.g., committee chair) requires support from the EPG and thus, an engagement with European ideas instead of national party-specific interests. In others words, if we would expect any relationship between the candidate selection process of national parties and the time dedicated to reports or opinions, then this relationship should be negative because other instruments are less time consuming and may be more effective to serve national principals (i.e., elites of a national party or party members). However, because of the complex allocation procedure we do not expect any systematic relationship: Expectation 3: Candidate selection procedures for European elections within a national political party are not systematically related to MEPs' prioritization strategies in terms of opinions and reports.

Research Design, Data, and Methods
This article analyzes the parliamentary activities of the members of the 7th EP (2009-2014) and asks why MEPs prioritize some activities over others. Accordingly, we are interested in MEPs' prioritization strategy in terms of parliamentary activities and not in the parliamentary productivity of individual MEPs, nor do we compare the parliamentary productivity across MEPs. Instead, for all par-liamentary activities we measure the share of the respective activity of the total number of activities an individual MEP engages in: speeches in plenary, written questions, opinions, and drafted reports. We select these four activities because they represent two "extremes" of classical parliamentary instruments: (1) Speeches and written questions, for instance, encompass more "symbolic" engagement, while following different logics, they can be implemented on an individual basis, are less time consuming and hence, useful for cultivating an individual profile that links up to national principals (Slapin & Proksch, 2010); (2) opinions and reports, by contrast, represent more "substantive" activities (in terms of impact on the legislative outcome), requiring a large effort of coordination and time. Because of their political importance, the allocation of opinions and reports follows a highly complex and political procedure (cf. EP, 2014, rules 49-56). Hence, the activities in focus are very different in nature and require a different workload. However, we argue that MEPs behave rationally and weigh all activities against each other when they distribute their time (e.g., one report instead of many written questions). For this reason, the dependent variable calculates the share of each specific activity relative to the total amount of all four activities. As a robustness check for this specific selection, we run additional analyses using the activities' share of all major activities documented by the EP (this includes motions for resolutions and written declarations additionally to the four activities mentioned; see results in Table A7 in the Supplementary File). We obtained the activity measures for 692 MEPs of the 7th EP (2009-2014) from Vote Watch (2015).
In the following paragraphs, we summarize the central independent variable, provide information on the control variables and conclude with some descriptive statistics (see Table 1). Our central independent variable, "exclusive selection," captures the exclusiveness of the overall candidate selection process, including (1) candidate nomination (who nominates?), and (2) the decision over the final list placement of candidates (who decides?). The exclusiveness of both components of candidate selection is measured in ordinal variables using the following values: 1 (individual members); 2 (subset of members); 3 (committee); and 4 (executive). Next, we integrate both components, assuming that the identity of the selectorate taking the final decision on list placement is more important to the MEP than the nominating selectorate (to test the impact of this assumption, we check a different operationalization of the variable, results are in Table A8 in the Supplementary File). So, the key dependent variable, "exclusive selection," takes the value of: 1 if nomination and decision are inclusive; 2 if nomination is exclusive but the decision inclusive; 3 if only the decision is exclusive; and 4 if decision making is in both stages is exclusive. The data has been collected and shared by Fortin-Rittberger and Rittberger (2015). For an additional robustness check, we create dummy variables of both components of the candidate selection process: 1 if nominated/placed on the list by a committee or the party executive; 0 if nominated/placed on the list by individual or a subset of members. The results of this test are in Table A5 in the Supplementary File. Parties' candidate selection rules may distinguish between the formal right to propose a candidate-often the right of every party member-and the right to propose a certain order or list placement of the candidates-mostly done by a party committee. To control for potential biases that could arise from this, we leave out category 1 (individual members) in the nomination variable (see result of this robustness check in Table A6 in the Supplementary File).
To take the multi-level structure of our data into account, we run multi-level mixed effects linear regression models with random intercepts at the level of the party and of the EU member state. As our dependent variables (shares) are bounded between zero and 100, a linear regression model could therefore possibly suffer from nonnormal errors and heteroscedasticity because of potential out-of-sample predictions. As a robustness check, we run a fractional logit model (Papke & Wooldridge, 1996) that was developed for the analysis of percentages (for these results see Table A4 in the Supplementary File).
As the prioritization of some parliamentary activities over others is influenced by many more factors, we include a large number of control variables. One of the main control variables includes the electoral system of a country, which is often used to identify power of the national party over the MEP. Specifically, we distinguish between "closed electoral systems," where voters can only vote for an electoral (party) list (1) and other systems (0; see also Schwindt-Bayer & Mishler, 2005). We gathered this data from Pilet et al. (2009). National opposition parties might use their activities in the EP to control the domestic agenda (Jensen et al., 2013). Hence, we include a variable capturing how often the national party of the MEP was not part of the national government ("Share in National Opposition"), which is calculated by using the number of government memberships of a party from 2009-2014 divided by the total number of governments in the given time span (2009)(2010)(2011)(2012)(2013)(2014), and subtracted from one. Data on the composition of national governments in Europe originate from the ParlGov database (Döring & Manow, 2019). It has been shown that national parties value MEPs from more powerful committees in European elections (Frech, 2018). Therefore, we also control for "Committee Power." This variable captures whether a MEP was part of a more powerful legislative committee within the EP during the term (1) or not (0). The classification of parliamentary committees as more or less powerful is taken from Yordanova (2009; see Table A1 in the Supplementary File). Some legislative activities, speeches most importantly, are more likely to be held by MEPs with certain offices. Hence, we expect "Committee Chairs" and other "EP Leadership" to be more active in plenary sessions and therefore control for these offices. "Committee Chair" is a binary variable that takes the value of one if the MEP was chair of any EP committee at any time during the legislative term and "EP Leadership" takes the value 1 if the MEP was at least a member of the conference of presidents or the EPs Bureau during the term. Chairs of powerful committees are more active than chairs of less powerful committees and potentially write more reports. Therefore, we also include an interaction between committee power and the committee chair. The "Participation Rate" of an MEP in plenary tells us something about the ability of an MEP to engage in certain activities: A person who is absent during the plenary sessions on a regular basis has less time for speeches, but has potentially more time to write reports. When calculating the participation rate, MEPs who participated in less than 10% of all plenary sessions are omitted. The variable "EPG Left-Right Position" is an ordinal variable that captures the ideological position of the EPG from left (1) to right (7). The order is: European United Left-Nordic Green Left (GUE/NGL), Progressive alliance of Socialists and Democrats (S&D), Greens-European Free Alliance (Greens/EFA), Alliance of Liberals and Democrats for Europe (ALDE), European People's Party (EPP), European Conservatives and Reformists (ECR), and Europe of Freedom and Direct Democracy (EFDD). We control for the EPG's ideological position on the left-right spectrum to determine whether left-wing party groups are more active and whether their members engage in different kinds of activities. Further control variables are "Party Size," "Eurosceptic Party," "Female," "Age," and "Seniority." "Party Size" captures the total number of MEPs who are members of the same national party during the 7th legislative term. "Eurosceptic Party" is a dummy variable that captures whether or not the national party of an MEP is considered Eurosceptic (coded 1 if the party strongly opposes or opposes European Integration according to the Chapel Hill Expert Survey [Polk et al., 2017]; a list of Eurosceptic parties can be found in Table A3 in the Supplementary File). "Seniority" is measured by the number of legislative periods the MEP previously served in the EP. All information on personal characteristics of MEPs we obtained from Hoyland, Sircar, and Hix (2009).
Finally, we conducted nine semi-structured expert interviews to deepen our understanding of the underlying dynamics linking MEPs' prioritization strate-gies in terms of parliamentary activities and candidate selection processes in national parties (for details, see Table A2 in the Supplementary File). We interviewed MEPs from the Greens (very inclusive nomination) and from the Christian Democratic Party as well as the Social Democratic Party in Germany (more inclusive candidate selection procedures). We also interviewed members of the Socialist Party in France and the Liberal Party in Germany (rather exclusive selection/nomination procedures).

Empirical Analysis
Overall, we observe that MEPs nominated by the national party executive speak more (66 speeches on average) than MEPs nominated through a more inclusive process (MEPs nominated by selected party members or delegates speak 62 times on average). Also, the average number of written questions provides the first piece of evidence that supports our second expectation: MEPs nominated by the individual party members ask 33 questions on average, while MEPs nominated by the party executive only prepare 30 written questions. Furthermore, the descriptive analysis of the data shows that the prioritization of parliamentary activities varies across groups of MEPs: Female MEPs are more active in writing opinions (1.70 for male MEPs vs. 2.47 for female MEPs), and the members of the right-wing Eurosceptic EPG (EFD) on average ask the most written questions (44 in total) but are significantly less active in more substantial activities (about 0.34 opinions and 0.36 reports on average). While German MEPs for example draft more than four reports on average, Estonian MEPs write only 0.68 reports.
These descriptive insights are further substantiated and supported by the results of several multi-level mixed effects linear regression models that analyze the effect of the overall exclusiveness of the candidate selection procedure a party employs (see Table 2) and the effect of the two stages (i.e., nomination and decision on final placement decision) of the selection procedure separately (see Table 3). Models 1 and 5 focus on the prioritization of speeches relative to other parliamentary activities and show a positive and significant effect of the parties' candidate selection process as a whole as well Notes: Coefficients of multilevel linear regression models with random effects at the level of the national party and at the EU member state. Standard errors are in parentheses. leg.: legislative. Significance levels: + p < 0.10; * p < 0.05; ** p < 0.01; *** p < 0.001.
as of both stages. In detail, the more elite-dominated the candidate selection processes of MEPs within a national political party, the more time MEPs dedicate to speech-making activities. Figure 1 illustrates this pattern graphically by disentangling the relationship between the exclusiveness of candidate nomination procedures and speech-making activity. Our models predict 58 speeches for MEPs that are nominated by a large number of individual party members and a mean of almost 69 speeches for MEPs being nominated by a very exclusive cycle. These strong findings support our first theoretical expectation, proposing that speeches are particularly useful to communicate with the national party (Slapin & Proksch, 2010) and, specifically, more effective for a small group of party executives than a large number of party members. Our second expectation, proposing a systematic relationship between candidate selection procedure and MEPs engagement in written questions, also finds sup-port. Models 2 and 6 show that the more exclusive the candidate nomination procedures within national parties, the fewer written questions MEPs of these parties tend to propose. However, only the second stage of the candidate selection procedure, namely the exclusiveness of the decision on the final list placement of candidates, shows an effect that is statistically significant. These results are supported by evidence from the interviews. One interview partner, who is a member of the Social Democratic Party in Germany, whose candidate nomination and selection process is quite inclusive (a subset of members nominates and decides over nomination), explains: [I] submit written questions, if I am asked for by anybody; especially, local actors or groups. For instance, that was the case with Opel. As Opel wanted to relocate a factory [from Germany] to other European countries, I should prepare a written question to In other words, written questions seem to be an important instrument for MEPs who serve a larger group of selected members within their national party because written questions can be used to gather very specific but still highly important information for this specific group. The idea that written questions can be used to serve individual or a particular group of members is also graphically visible in Figure 1b: The predicted effects for the relative share of written questions MEPs propose is higher for more inclusive selectorates.
Models 3 and 7 as well as 4 and 8 display the effect of the exclusiveness of candidate selection procedures on the share of opinions and reports drafted by MEPs. Overall, highly exclusive candidate selection procedures in national parties seem to hamper the prioritization of opinions and reports, (see for a graphical overview Figures 1c, 1d). Looking at both stages of selection separately, we observe a more nuanced pattern. While nomination has a larger, significant effect on drafting opinions, the placement decision stage has a clear effect on reports. One of our interview partners explains the general phenomenon as follows: But look, you can secure your re-election in such a party, by profiling yourself almost only party politically; [you] can knit your network in a way you will definitely get higher in the next election, although you do not work so much content-wise [i.e., drafting reports or opinions]…Or you are strategically smart and just make one report, that brings a lot of public at-tention…Then you have to do something, but not so much. (Interview 01) In other words, if an MEP is elected in a national party that organizes its nomination and list placement process exclusively, the national party elite is the key principal for the MEP and, thus, a large number of reports and opinions might not be the most effective way to serve the key principal (cf. Interview 02). Instead, it seems wise to take over relevant tasks in Strasbourg, for instance, organizing the communication between national party elites and MEPs on different policy issues. The EPG is a central gatekeeper for report allocation (Hix & Høyland, 2013, p. 182) and, thus, another principal that might challenge the implementation of the specific interests of individual national parties. However, MEPs that are used to ac-commodate the preferences of diverse national party colleagues due to very inclusive candidate selection procedures are perhaps better prepared and may find more freedom in negotiating policy deals with colleagues in Strasbourg, which is key for the formulation of reports. The assistant of a Spanish MEP explained in that regard: It is like that "Ok, you will get the report, but please take into account what I am going to tell you in the coming month" or something like that. It is trying to find a way to work together and to get a good result. (Interview 09) In addition to the exclusiveness of candidate selection procedures in national parties, other control variables affect outcomes. For instance, committee power and committee chairing are highly important for the prioritization of reports. For the prioritization of written questions, the opposition status of the national party is key and thus, again, a variable that has been detected as relevant by other studies (Jensen et al., 2013;Wonka & Rittberger, 2014). Interestingly, members of the national opposition prioritize speeches less systematically than other MEPs, indicating that speeches in the EP are less useful to follow a blaming strategy directed at national governments (for more details on blaming strategies of opposition parties in the EU multi-level system, see in this issue Heinkelmann-Wild, Kriegmair, & Rittberger, 2020). Furthermore, individual speech-making activity is additionally stimulated by the power position and participation rates of MEPs. As MEPs can only speak or vote while sitting in plenary, it is not surprising that the participation rate in plenary is positively associated with the prioritization of speeches but hampers the engagement in other activities. Additionally, more senior MEPs do not necessarily follow other priority patterns than less experienced MEPs. However, our results reflect that seniority is important for the distribution of opinions and reports. Lastly, MEPs elected in countries with a closed-list system are likely to give a relatively large share of speeches but propose a comparably low share of written questions. Although the effects are not significant, the finding is in line with the basic argument of this article as well as the findings of previous studies using closed-list systems as a proxy for a close relationship between MEPs and national parties (Slapin & Proksch, 2010).
In sum, these results provide support for our general argument and for the two specific expectations in terms of speeches and written questions. Different than proposed in Expectation 3, however, candidates' selection procedures of national parties indeed seem to influence MEPs' prioritization strategies in terms of reports and opinions in a negative way. This finding supports the general idea that MEPs have to prioritize parliamentary tasks and if they face an exclusive selection procedure, they are more likely to shift their focus away from reports or opinions towards activities that are more visible and suitable to serve the national or party leadershipsí nterests. However, these findings might be influenced by many more factors, as the allocation of reports and opinions follows a complex process. Yet in sum, our general argument finds strong support: The candidate selection processes of national parties seem to determine the key principal MEPs serve and, thus, MEPs' prioritization strategy in terms of parliamentary instruments because some parliamentary activities are more "visible" and suitable to "cultivate a personal vote" and to communicate national or individual interests than others.

Conclusion
The question of how European legislators deal with multiple principals and a variety of parliamentary tasks is highly interesting because it concerns central standards of modern democracies, namely political representation and responsiveness. MEPs are the key actors in transmitting the interests of European citizens to the suprana-tional level and thus, may substantially increase the legitimacy of EU decisions. Hence, an analysis of MEPs' prioritization strategy of parliamentary activities and how this process is linked with candidate selection procedures of national parties is highly fruitful as it allows us to disentangle the complex principal-agent relationship in the EU's multi-level system and to understand how electoral rules in political parties shape political responsiveness. Moreover, these findings are essential for the current political plans of building up a transnational party system (cf. this issue Lefkofridi, 2020).
We discover that the more exclusive (i.e., elitecentered) candidate selection procedures of national parties, the more likely it becomes that MEPs prioritize speeches over other parliamentary activities; conversely, the more inclusive (i.e., member-centered) the candidate selection procedures, the more likely it is that MEPs prioritize written questions as well as opinions and reports. A main reason for these differences is the varying "visibility" of parliamentary activities to different principals (Klüver & Spoon, 2015) and thus, the varying utility of parliamentary activities to serve the different principals. Furthermore, our analysis discovers that it is worth distinguishing between two steps of candidate selection in national parties, namely nomination and decision on the final list placement because both steps may vary in their degree of exclusiveness and may therefore both affect the overall closeness of MEPs to their specific national party principals (i.e., party elites or party members).
Our insights complement existing studies in two innovative ways: (1) theoretically, by conceptualizing national parties not as "monolithic" principal but as conglomerations of groups of party members with different preferences; and (2) empirically, by focusing on the EP, which offers an ideal testing ground to compare candidate selection procedures and parliamentary activities, since elections are held simultaneously across the 28 EU member states but feature a broad variety of electoral rules and, more importantly, all MEPs face similar procedural rules with regard to parliamentary activity, facilitating any comparison.
Besides these innovative findings, our study also faces some weaknesses. We detect a systematic relationship between candidate selection rules and the prioritization of parliamentary activities of MEPs, and explain it with the visibility to and utility of certain actions for specific selectorates. Even though we have some evidence for this explanation from the expert interviews, a future article will have to prove the visibility of each legislative instrument to the different selectorates in more detail. Furthermore, we note a point related to the operationalization of the dependent variables, which measure the relative share of each type of activity of the total number of activities engaged in by MEPs: this is a strictly quantitative approach that does not consider the content nor the relative time required to prepare each activities. Moreover, it disregards activities of MEPs outside the parliamentary arena (i.e., organizing events at the state/local level for their constituencies or party members) which also might be used to serve elites or members of their national parties. Capturing this extraparliamentary activity in a systematic way would be highly promising to push this research agenda ahead. Finally, it would be fruitful to explore our argument in depth for Eurosceptic parties as their core-topics are European integration and European immigration, which might stimulate party members to be more attentive towards their representatives' activities in the EP and similarly, bias MEPs' prioritization strategies towards parliamentary activities that are most compatible with populist communication strategies.
Overall, however, this article makes an important contribution as first, it opens up the black-box of the "national party principal" in the EP by exploring how variations along the inclusion-exclusion dimension in party candidate selection procedures influence the type of activity MEPs engage. Second, it speaks to a growing comparative literature analyzing the link between party candidate selection procedures and legislative behavior. Besides this cross-cutting scientific relevance, this study is also of societal relevance as it illuminates the channels through which national and sub-national party interests could enter EP politics and thus, increase the legitimacy of EU policy-making.

Elena Frech is a Post-Doctoral Researcher at the Department of Political Science and International
Relations of the University of Geneva (Switzerland), in the "Parliamentary Careers in Comparison" project. She is interested in the European Union, in particular the European Parliament, and legislative and comparative politics. In particular she specializes in political parties, especially party behavior and decision making, public opinion, and political recruitment. She is furthermore working on international political economy topics and multi-level governance. Her work has been published in International Organization, European Politics, and other journals.