political science, University of Gothenburg, May 7-8, Early draft.

Transcription

1 Costly Responsiveness: The Party as a Constraining Factor for Candidates Responsiveness to Citizen-Initiated Contacts Elin Naurin and Patrik Öhberg Department of Political Science, University of Gothenburg Abstract We investigate how individual politicians responsiveness to citizen-initiated contacts is regarded by their party colleagues: What type of responsive behaviour is accepted within the party and what circumstances increase and decrease acceptability for responsiveness? In a survey experiment 1600 Swedish politicians are presented to scenarios where citizens who disagree with the party make contact with a party colleague (to impact the shutting down of a school). Politicians answer if they approve of different communicative and adaptive responses that take place inside and outside of the party. We find that some responses (aimed at changing the party from within) are more approved of than other (aimed at changing the party via open external means). We also find that electoral and policy related incentives are more at play for internal informal responses than for other type of responses, although the main conclusion is that approval of responsiveness is not easily altered by electoral and policy incentives. One important implication of the paper is that previous studies seem to have focused on types of responsiveness that are seen as problematic within parties, and that dynamics of responsiveness in party-centered systems thereby might have been misunderstood. Paper to be presented at the 1 st Gothenburg-Barcelona workshop on experimental political science, University of Gothenburg, May 7-8, Early draft. Contact: elin.naurin@pol.gu.se, patrik.ohberg@pol.gu.se 1

2 INTRODUCTION The district Majorna-Linné in the central part of the Swedish city Gothenburg is a popular place to live among well-educated, politically conscious parents of young children. The rental apartments are small but cheap and allow the residents to make other priorities in life than to work. The caricature of how the political decisionmaking of the district takes place is that politicians are unable to handle their conscious constituents who are high on knowledge, money, time and incentive to engage politically. The district has cut down on most things during harsh economic times, but when one of the smaller libraries, the smallest schools, the open nursery or the kitchens at the smallest preschools are shut down, politicians stumble at the obstacle of the resourceful parents who firmly push decision-making ahead until libraries, schools, kitchens and open preschools open again, sometimes on behalf of other political areas. The link between what people want and what political decisions that are taken is central to representative democratic theory and practice. However, it is far from clear what decides when citizen input actually influences politics and when it does not. Most of the research on how politicians respond to direct citizen-initiated contacts comes from the United States (See review in Grose 2014; see also Butler and Broockman 2011; Broockman 2013; Butler, Karpowitz, and Pope 2012; Harden 2013; c.f. Richardson and John 2012; Richardson 2013). While this research brings important knowledge of how democratic responsiveness functions in candidatecentered systems, most political systems around the world have other constitutional arrangements for the relationship between voters and individual politicians compared to the U.S. Specifically, in most political systems, politicians need to balance contacting citizens views not only against their own views but also against their 2

3 party s position on the matter. This balancing of potentially conflicting wishes that takes place in the minds of politicians in party-centered systems is under-studied and we know fairly little about how and when the party actually constrains or encourages responsive behaviour from the side of the individual representative. Our aim with this paper is to contribute to the understanding of the complexities imbedded in the process of responsiveness in party-centered systems by focusing on what type of responsive behaviour that is accepted within the party. While existing work typically discusses political responsiveness from the perspective of the individuals giving or receiving the response, our approach is to give attention to politicians responsive behaviour from the perspective of colleagues in the political party. By investigating how representatives react to their fellow party colleagues responsive behaviour, we contribute to the understanding of how responsiveness is constrained by internal processes within political parties. More specifically, we focus on situations where fellow party colleagues want to represent their constituents view instead of the party s official position and we take the perspective of the party, investigating how responsive behaviour is perceived by the party colleagues of the responding politician. Also, where previous work has often concentrated on politicians formal and external responsiveness such as their roll call voting, participation in parliamentary debates, media appearances and direct responses to citizen requests, we identify the need to understand how politicians are responsive also in indirect and informal ways. Most importantly, a representative can work behind the scenes and be responsive indirectly by trying to influence the party, which is important to take into account when studying reactions to responsiveness from party colleagues. Apart from contributing with a theoretical focus on the interplay between the party, the individual 3

4 politician, and the contacting citizens, we thereby also contribute by defining policy responsiveness in a wider way than what is common in the empirical literature. More specifically, we measure politicians views of their colleagues adaptive and communicative responsiveness. Communicative responsiveness is defined as listening responses and explaining responses (see Esaiasson, Gilljam and Persson 2013). Adaptive responsiveness is measured as on the one hand responsiveness that aims at changing the party from within (which we call hidden adaptive responsiveness) and on the other hand responsiveness that takes place openly and independently of the party (open adaptive responsiveness). We are able to gain unique insights by conducting a large-scale survey experiment with 1600 Swedish politicians elected to local, regional, and national level political positions. We present scenarios to the politicians where citizens make contact with a party colleague to impact the shutting down of a school. School shutdowns are frequent in the Swedish political context (c.f. Uba 2015) and constitute situations where politicians, voters and parties can have different policy positions in different cases. By using a fictive school shutdown as a scenario, we can credibly vary whether or not the party s voters make the contact, whether or not the party risk s policy lost, and the opinion of respondent on the issue. We can also credibly create a situation where the party disagrees with both the contacting citizen and the party colleague. The design has been used previously successfully detecting relevant variation in Swedish politicians subjectively described behaviour (Öhberg and Naurin 2015). The paper proceeds as follows. First, we discuss how party representatives are affected by other politicians responsive behaviour: What theoretical and empirical expectations are reasonable to have when it comes to party representatives 4

5 encouragement or constrains on other individual politicians responsive behaviour? Thereafter we describe our interpretation of the notion of responsiveness and formulate our specific research questions. The study design then briefly describes the Swedish context, the survey the Panel of Politicians as well as the treatments used in the experiment. Results and concluding sections ends the paper. THEORY The party s view of individual politicians responsiveness to citizen contacts. What should we expect of party members when it comes to their views on their colleagues responsiveness towards direct citizen-initiated contacts? One common theoretical point of departure in the literature on political representation is that parties are coherent units with members who are motivated mainly by party victory (e.g., Adams, Merrill, and Grofman 2005; Budge 2001; Downs 1957; Müller 2000.). From such a perspective, party representatives acceptance of a party colleague s responsiveness to citizens depends on the party s position on the issue and on whether the colleague will increase or decrease the likelihood of getting the party elected. If party unity and party success is at risk through the behaviour of the colleague, acceptance will be lower. An alternative perspective is to emphasize that parties are built upon principles of division of labour (c.f. Andeweg and Thomassen 2011; Skjaeveland 2001). From such a perspective colleagues responsive behaviour will be accepted since they are trusted to do what is best for the party. The point of departure is then that the workload of parties is large, and that in order to attract the best talents to the party, members need to accept that each representative is passionate about and specialist on different issues. Also relevant, but less discussed in the literature, is that party 5

6 representatives are related to each other on a personal level, rather than only via the party. Party colleagues are co-workers, maybe even friends, and as such they likely sometimes simply trust each other. The acceptance of each other s behaviour then has to do with the fact that they share important goals and values and that they trust that their colleagues make good judgments based on this. Yet another theoretical approach is to study individual representatives independently of the party they represent (e.g., Mayhew 1974). Individual politicians are in this case single-minded seekers of personal re-election (c.f. Ashworth 2005; Entman 1983; Jones 1973), therefore their reaction to their colleagues responsiveness to public opinion is influenced less by the party s position or by the importance of party unity, and more by their own convictions and the likelihood that the citizens will vote specifically for them. If responsive behaviour leads to the decreasing of one candidate s possibilities to win in the next election, that candidate would then be less likely of approving of the responsive behaviour. However, the most reality-based perspective is probably that all these different factors play a role in how party representatives react to colleagues responsive behaviour (c.f. Öhberg and Naurin 2015). It is likely that the party s position, the own view and the contacting citizens all influence how responsiveness is regarded. In this study, we try to take into account the complexity of these complementary and/or contradictory incentives to encourage or discourage responsiveness of party colleagues. We lift forward and study three of the above-mentioned factors, all with basic relevance to representative democratic theory and practice. First, we consider the party s electoral incentives to approve of individual politicians responsive behaviour. If, by responding to the contacting citizens, the politician increases the party s future chances of getting electoral support, her colleagues are more likely to 6

7 approve of the responsive behaviour even though it goes against the position of the party. We operationalize this as contact being taken by likely voters of the party. Second, we consider our respondents personal policy incentives. If the politicians we study agree with their colleague (and the contacting citizens), they are more likely to approve of the responsive behaviour even though it goes against the party s position. Third, responsive behaviour can have different actual effect on policy. It is likely that responsive behaviour that goes against the party line is more approved of if there are no serious policy consequences that clash with the party line. Or put differently, responsive behaviour against the party can be approved of as long as it doesn t have any real consequences on policy decision, i.e. when policy vulnerability is low. The concept of responsiveness There are different ways for a politician to respond to citizen-initiated contacts. A classical interpretation of democratic responsiveness is that representatives adjust their political decisions to the views of the public. On the aggregated level, this means that public opinion in some sense should leave its mark on the output of the political system (e.g., Brettschneider 1996; Soroka and Wlezien 2010; Stimson, MacKuen, and Erikson 1995). On the individual level, in the relationship between the citizen and her representative, such adaptive responsiveness means that the representative in different ways acts to affect decisions in accordance with the view of the represented. The most often used measures of such individual actions are actions such as roll call votes (Miller and Stokes 1963; Butler and Nickerson 2011; Bafumi and Herron 2010; Kousser, Lewis and Masket 2007), parliamentary debates (Proksch and Slapin 2012). motion-writing (Bowler 2010) and participation in media (Arceneaux, Johnson, Lindstädt and Wielen 2015). Studies investigating these actions provide important 7

8 understanding of how politicians act openly in response to public opinion. However, they miss the action that takes place within parties, such as representatives efforts to influence their party colleagues to behave responsively. When representatives take the parliamentary floor, write a debate article or cast their vote in parliament, these are often the last acts in series of events. Much of the dynamics behind adaptive responsiveness might be missed if we do not consider responsiveness that takes place inside the political party. Moreover, the efforts to behave responsively risk being underestimated if we disregard one of the most important arena for political behaviour; the internal life of parties. Moving on, a part from contributing with enlarging the understanding of individual politicians adaptive responsiveness to also include adaptation that takes place inside parties, we also study communicative responsiveness. The reason for this is that the adaptive form of responsiveness is a narrow view of how representatives interact with citizens. Congruence and incongruence between citizens and representatives does not capture all aspects of how representatives respond to citizen requests. Nor does it comply with commonly applied normative theories that underline that the views of the represented do not always go hand in hand with the interests of the represented. Representatives are not normatively obliged to adapt to the views of contacting citizens. Instead, where there is a conflict, political representatives should provide a good explanation of why the wishes of the represented are not in accord with the interests of the same (Pitkin 1967, ; see also Manin, Przeworski, and Stokes 1999; Mansbridge 2003). This normative demand to communicate the reasons why an action is not in accordance with the wishes of the people highlights the concept of communicative responsiveness (Esaiasson, Gilljam, and Persson 2013; Öhberg and Naurin 2015). Communicative responsiveness can 8

9 mean that the contacted politician listens closely and tries to understand what the citizen is expressing before making a decision. It can also mean that the politician takes time to explain the views of the party or of herself/himself in order to ensure that the citizens understand the arguments behind the decision (c.f. Fenno 1978; Kingdon 1989). In line with this reasoning, Esaiasson, Gilljam and Persson (2013) propose a threefold theoretical approach to the notion of responsiveness that takes into account both adaptive and communicative responses: Politicians can be responsive by (1) listening and finding out more about the wishes of the public, (2) explaining and developing their own position to the public and (3) adapting and trying to meet the wishes of the public. Such a multifaceted understanding of responsiveness should be useful in all studies of the effects of citizen-initiated contacts, but maybe especially so in studies of party-centered systems. It seems likely that the complex relationship between the party, the politician and the citizen gives the politician incentives to be responsive in other ways than just adapting to views expressed when the contact is made. So far, most empirical work on communicative responsiveness has been done with a focus on citizens service requests (Butler and Broockman 2011; Broockman 2013). Less attention has been given to communicative responsiveness to policy requests, although the field seems to be moving quickly forward (c.f. Butler, Karpowitz, and Pope 2012; Harden 2013). In a general sense, it seems reasonable to understand communication about policy as a less costly response than adaptation to policy. Studies comparing adaptive policy responsiveness and service responsiveness find that representatives see the latter as easier than the former. Dealing with policies involves the risk of clinching with the party and upsetting other voter groups, while providing a service is a smoother way to show interest and engagement (Butler, 9

10 Karpowitz, and Pope 2012; Cain, Ferejohn, and Fiorina 1987; Harden 2013). Similarly, it should be less demanding to communicate with citizens than to adapt to their views, in terms of both the effort involved and the potential clinch with the party and personal opinions. RESEARCH QUESTIONS We investigate the extent to which and when party representatives approve of other politicians adaptive and communicative responses to citizen-initiated policy-focused contacts in a party-centered system and give special focus to situations in which the party believes one thing and the contacting citizens as well as the colleague believe another. We formulate two questions: Q1. What type of responsive behaviour from fellow party colleagues is acceptable when the party position is not in agreement with the contacting citizens policy request? Q2. What circumstances increase and decrease acceptability for responsive behaviour within the party when the policy request goes against the party s will? STUDY DESIGN Recent research on representation has applied an experimental design to the study of politicians responsiveness (See review in Grose 2014; see also Butler 2014; Butler and Broockman 2011; Broockman 2013; Butler, Karpowitz, and Pope 2012; Harden 2013; c.f. Richardson and John 2012; Richardson 2013; Öhberg and Naurin 2015). We took inspiration from these designs and conducted an online survey experiment in which we randomly assigned politicians to scenarios in which citizens make contact 10

11 with one of their colleagues. In this section of the paper, we explain the Swedish context in which the survey was conducted. We also describe the Panel of Politicians survey, which provides a unique sample of party representatives who are likely to have experienced, or will experience, similar situations to the ones we ask about. Thereafter, we clarify the design of the experiment and our stimuli (the respondents own opinion, whether or not it is a close call situation and contacting group). The section ends with a description of how our dependent variables communicative responsiveness as well as opened- and hidden adaptive responsiveness are operationalized. A few words about the context: Sweden is a fairly typical party-centered system in which it is likely that the party plays a significant role in the direct relationship between voters and representatives. Constitutionally, Sweden has strong and influential political parties that function in a parliamentary system with a low degree of separation of powers. Party unity is strong, and members of local and national assemblies tend to be loyal party representatives. That said, the personal relationship between the candidate and the voters is formally encouraged via a possibility to cast a personal vote for someone on the party list. Previous elite surveys of Swedish politicians support the claim that politicians have complementary loyalties to the party, their own opinions and the voters. Since 1985, the Swedish Parliamentary Study has regularly asked Swedish Members of Parliament (MPs) to describe who they personally think is important to represent. Judging from these investigations, with an average response rate of over 90 per cent, promote the policies of your own party, promote the interest of individual voters and promote views you personally consider important are all of great importance. On average, 79 per cent of the Swedish MPs testify that all three objectives are important. The 11

12 assessments have been stable for twenty-five years (the proportion has never fallen under 75 per cent) (Öhberg and Naurin 2015). Description of the survey: The Panel of Politicians The Panel of Politicians is an online panel of Swedish politicians from the local, regional and national levels conducted by the Laboratory of Opinion Research (LORE) at the University of Gothenburg (Öhberg 2014). The survey captures a unique diversity of politicians from the investigated context. Respondents were recruited via addresses found on local and regional parliament websites as well as via the surveys Comparative Candidates Survey (CCS) 2010 and the Municipal and Regional Parliament Survey 2012 (Kommun- och landstingsfullmäktigeundersökningen, KOLFU), where all respondents by the end of the surveys were asked if they wanted to sign up for the panel. The CCS was sent to all parliamentary candidates in the 2010 national election and the KOLFU was sent to all members of all local and regional parliaments during the fall of Panelists are also regularly asked if they are willing to recommend other colleagues to participate in the panel. When our study was performed, the panel consisted of around 3,200 politicians, which included as many as 22 per cent of the country s MPs. All individuals elected to local, regional and national parliamentary positions during had been contacted, as were all candidates for the 2010 national parliamentary election. The response rate for the survey used in this paper was 50 per cent. In total, politicians participated in the experiment. Taken together, our survey captures an unusual diversity of party representatives from the investigated context, which makes it possible to perform a survey experiment on an unusually large sample of representatives. 12

13 Experimental manipulations and questions The vignettes are formulated as follows. (Table 1 in the Appendix illustrates the eight groups that the design yields): It happens that different groups of citizens contact politicians with political propositions. Imagine a situation where a colleague of yours just received an from a group of citizens/a group of voters you know voted for your party in the last election. The citizens/the voters oppose a proposition from the civil servants to close a school in your municipality. Your colleague/you and your colleague agrees/agree with the proposition that they make. The proposal goes against your party s position on the issue and there is only a bare majority in your party s favour in the parliament/no mention of policy vulnerability. Given the scenario, to what extent would you approve of the following behaviour of your colleague? 1 The respondents rate on a 7-point scale (where 1 = Strongly disapprove and 7 = Strongly approve ) of each of the following eight types of responses: 1. Answer e- mail and develop his/her position on the matter, 2. Answer and explain your party s position on the matter, 3. Ask them to send additional information, 4. Suggest a personal meeting, 5. Try to change the party leadership informally via personal contacts, 6. Openly address the issue at party meetings to change the official party position. 7. Take the question further and try to get the media interested in the cause. 8. To vote according to the citizens wishes against the party position. 1 In Swedish: Det händer att olika grupper av medborgare hör av sig till politiker med politiska förslag. Föreställ dig en situation där en partikollega precis har fått ett mail från en grupp medborgare/en grupp väljare som du vet röstade på partiet i förra valet. Medborgarna/väljarna motsätter sig ett tjänstemannaförslag om att lägga med en skola i din kommun. Du och/din kollega håller verkligen med dem. Medborgarnas och din kollegas uppfattning går dock emot ditt partis hållning i frågan./och det finns endast en knapp majoritet för partiets förslag i kommunfullmäktige. Givet scenariot, i vilken utsträckning skulle du tycka det är acceptabelt att din kollega gör följande: 13

14 Hence, we use four measures of communicative responsiveness and four measures of adaptive responsiveness. Two of the communicative measures focus on explaining responses (explaining one s own opinion and explaining the party s position) and two focus on listening responses (asking for more information and suggesting a personal meeting). For the adaptive responsiveness, we measure adaptation by internally changing the party position in two ways (informally and formally) and adaptation outside of the party in two ways (go to the media, and vote against the party). Since there is little previous knowledge about how politicians react to responsiveness to policy requests of colleagues in the party, a number of potentially relevant moderating variables are included in the analyses. These are: 1) gender; Studies indicate that gender matters to how politicians perceive their role as representative, although there is no clear image of how exactly responsiveness is affected. Women have been found to put greater emphasis on the role of the party (Scully and Farrell 2003) while men have been shown to put greater weight to media appearances in party systems (Aalberg and Strömbäck 2011). These studies indicate that men are more interested in developing personal contacts with citizens while women put greater weight on their relation to their party. However, other studies indicate that women perceive it to be more important to represent minority groups such as immigrants, women, young etc, indicating that personal relationships are important for other reasons to women (Wängnerud 2009). 2) age; previous studies on party-centered systems indicate that young politicians give greater weight to the partisan role (Zittel 2012), motivating the inclusion of age in a model 3) sense of closeness to party; Studies show that degree of party identification varies within parties. Not all representatives feel equally close to their party (Önnudóttir 2014) 14

15 which likely influences how they regard responsiveness that goes against the party. We therefore include a question about how close the respondents feel they are to their party: How close do you feel to your political party?. The scale runs from 1. Not close at all 2. Not particularly close 3. Fairly close 4. Very close. 4) the size of politician s party. It has been shown that large parties generally put greater weight to keeping the party together (Desposato 2005; Sieberer 2006). The parties included in this paper are the Left Party, the Social Democratic Party, the Center Party, the Liberal Party, the Conservative Party, the Christian Democrats, the Green Party, the Swedish Democratic Party, and a category regrouping all other parties (often local small parties). In the analyses, these parties are divided into two groups consisting of big parties (The Conservative Party and the Social Democrats) and small parties (the others), 5) Level of education. It has been suggested that politicians with higher education are less loyal to their party and more focused on policy (Müller and Saalfeld 1997). In our case, education is included to account for similar effects. The variable has two values: having completed a university degree or not, 6) Position in the party; we know from previous work that position in the party matters for things like perception of the importance of party unity (Kam 2009) and knowledge of the party (Naurin 2014). In this paper, we measure whether our respondents are members of the municipal council or not. 15

16 RESULTS We begin with the first research question: What type of responsive behaviour from fellow party colleagues is acceptable when the party position is not in agreement with the contacting citizens policy request? Figure 1 describes party representatives views of what responsive behaviour toward contacting citizens that is acceptable when the party position is against the contacting group. The figure describes all eight types of responsiveness; explaining responses, listening responses, hidden adaptation and open adaptation. We use a 7- point scale where 1= strongly disagree and 7=strongly agree. The midpoint 4 is marked in the figure. Figure 1. Party representatives approval of responsive behaviour towards citizen-initiated contacts when the party and the citizens disagree (mean values) Own position Party position Ask for information Ask for personal meeting Internal informal meeting Inernal formal debates Media Vote against Explain Listen Adapt (hidden) Adapt (open) Note: The confidence intervals are at 0.05 level. Question wording is found in the text. 1= strongly disagree, 7= strongly agree. N=1595. Source: Panel of Politicians, February

17 If we begin with the explaining responses: It is clearly more acceptable for a colleague to explain the position of the party, than to explain that his/her own position is against the party. The fact that the mean value for answer and explain his/her own position comes close to the midpoint (mean value 4.2) indicates that a disagreeing politician who explains his/her position to a contacting group is in a grey zone of what is acceptable in the eyes of the party colleagues. Answer the and explain the party s position has a much higher mean value of 6.1 and is not problematic. Nor is it very complicated for a disagreeing colleague to ask for more information and to arrange a meeting (mean value 5.5 respectively 5.7). In other words, listening is an acceptable form of responsive behaviour for a politician who is contacted by a dissatisfied group of people and who is sympathetic to their cause, while explaining is more complicated and dependent on whose positions is about to be explained. If we then move to our distinction between hidden and open adaption; the respondents make a clear distinction between different types of adaptation. The least accepted type of responsive behaviour is to adapt open and against the party by going to the media or voting against the party (mean values 2.5 and 3.3 respectively) and the most accepted form of responsiveness is to adapt by following formal rules within the party (discuss the issue at internal party meeting) (mean value 6.2). Among the eight different ways a disagreeing colleague can be responsive towards a contacting group of citizen, formal debates within the party is the most acceptable form of behaviour. It is even more accepted than to respond by listening and explaining, i.e. via communicative responsiveness. However, there is a significant difference between working formally by attending party meetings and working informally to change the party leadership (mean value 5.1). To informally contact party leadership to change 17

18 the position of the party is less approved of than the listening responses, but it is more accepted than explaining a disagreeing position in an to contacting citizens. Taken together, it is obvious that the politicians have greater acceptances for some responses than for others. From Figure 1, we draw the conclusion that party colleagues generally prefer intra-party adaptation, compared to adaptation that goes against the party will. This means that the types of responses that are normally studied within the literature on responsiveness (adaptation in the form of voting, outspoken policy declarations as well as personal declarations in s) are the least approved of in our study. Moving on, we analyze whether it is possible change acceptance of con-party responsiveness by changing policy incentives and electoral incentives of the politicians. These analyses aim at answering our second research question: What circumstances increase and decrease acceptability for responsive behaviour within the party when the policy request goes against the party s will? To make a long story short (all details are found in the appendix); our survey experiment gives very few significant effect of electoral and policy incentives on approval of any of the types of responsiveness we measure. We consider two possible conclusions to draw from this at this early stage of the project - We either failed at creating relevant treatments in our survey experiment; or the types of incentives we study actually do not alter party representatives views of responsive behaviour in situations when the party and the citizens disagree. The first conclusion - that we failed - does not rime with our previous experiences of performing survey experiments on the Panel of Politicians on responsiveness. In a recent article, we show that both electoral and policy incentives matters for how the respondents say that they would respond had they themselves been in the situation when a group of 18

19 citizens contact them. That article finds that a party s disagreeing position diminishes both the willingness to adapt and communicate among the respondents, but that personal incentives mitigated some of the effect of the disagreeing party. These mitigating effects were quite substantial (Öhberg and Naurin 2015). The experiment done in that article shows that the design we use here (survey experiment with Swedish politicians on responsiveness) works if we want to detect variation in responsiveness. The difference compared to this study is that we here focus on the perspective of the party, and that we give more weight to studying internal procedures of the party. Taken together, there is therefore also reason to dig deeper into the second possible conclusion: What speaks to the conclusion that acceptance of responsiveness to disagreeing citizens are not affected by policy incentives and electoral incentives of fellow party members? Is it possible that electoral incentives and policy incentives matter very little to how responsiveness is regarded when the contacting citizens are against the party? To begin with, there are a few significant main effects in the data that deserve to be lifted forward. In fact, we manage to rock the boat, i.e. change the approval of responsiveness, for the response that takes place informally and inside the party. When the scenarios mention that the respondent agrees with the colleague and the contacting citizen, the mean value for acceptance of taking informal contacts with the party leadership is 5.2. When we do not mention the position of the respondent herself, the mean value is 5.0 (the difference is significant at level 0.02). Hence, if you agree with your colleague you are more likely to approve of informal hidden adaptive behaviour. There is also a significant difference between the mean values in the scenarios where we mention that there is only a bare majority of in favour of your party s position. The respondents are more accepting if there is a bare majority 19

20 (mean value, 5.0) than when there is no mention of policy vulnerability (mean value, 5.3, the difference is significant at level 0.02). So; when it is a close call politicians are more willing to accept that their colleague takes informal contacts with the party leadership. The fact that the treatments have significant effects on the least open of all activities (internal and informal responsiveness) indicates what was written above that research on responsive behaviour in party-centered systems should give more focus to what happens inside the party. Hidden and informal responsiveness seem to loosen up the party s readiness to take impression of electoral and policy incentives to react responsively. However, the overall impression is that different incentives do not affect politicians readiness to accept responsive behaviour of their colleagues. Our conclusion from this is that electoral and policy incentives collide with strong party norms which become visible when we like we do in our experiment remind politicians of their roles as party representatives. Moving on, this conclusion is strengthened when we analyze effects of moderating variables. OLS regressions for each of the responsive behaviours are reported in Table 1. As mentioned in the design section, the model includes gender, age, education, position, closeness to party and size of the party. The table orders the different responses according to what was found to be the most preferred behaviour from the side of the party representatives we ask (see Figure 1), starting with the least preferred response. Table 1 shows that four factors cause variation in approval of responsiveness (when using the highest significance level); gender, age, size of party and closeness to party. 20

21 Table 1. Explaining the acceptance of colleagues responsive behaviour (OLS regression) Go to media Vote against party Explain own position Internal informal meeting Ask for information Ask for personal meeting Explain party position Internal formal debates Agree (0.083) (0.095) (0.101) 0.251** (0.089) (0.083) (0.079) (0.064) (0.061) Close (0.083) (0.095) 0.172* (0.101) 0.261** (0.089) (0.083) (0.079) (0.065) (0.061) Voters (0.083) (0.095) (0.101) (0.089) (0.083) 0.146* (0.078) (0.064) * (0.061) Sex *** (0.087) (0.101) *** (0.108) *** (0.100) (0.090) (0.085) (0.068) (0.067) Year (0.004) (0.004) (0.004) 0.008** (0.004) 0.014*** (0.003) 0.017*** (0.003) 0.013*** (0.003) 0.014*** (0.002) Education (0.088) (0.101) (0.109) 0.222** (0.095) (0.090) 0.166* (0.085) (0.071) (0.067) Position (0.095) (0.114) ** (0.119) (0.105) (0.096) (0.092) (0.074) (0.070) Close to party *** (0.079) *** (0.082) *** (0.084) ** (0.070) 0.318*** (0.069) 0.135** (0.064) 0.228*** (0.058) (0.050) Big parties *** (0.086) *** (0.097) *** (0.102) ** (0.089) *** (0.084) *** (0.079) (0.065) *** (0.061) Constant (6.840) (7.797) (8.074) (7.078) *** (6.363) *** (6.291) *** (4.923) *** (4.724) R N Note: Age has values between 19 and 94. Sex; 0 = man, 1 = woman. Level of education; 0 = junior high school, senior high school or other post-high school education that is not university level, 1 = university. Big parties; 0 = small parties (Center Party, Christian Democrats, Green Party, Left Party, Liberal Party, Sweden Democrats and local parties), 1 = big parties party (Conservatives and Social Democrats). Political position; 0 = not a member in the municipal council, 1 = a member in the municipal council. Closeness to party; 1 = Not close at all 2 = Not particularly close 3 = Fairly close 4 = Very close. Source: Panel of Politicians. Robust standard errors in parentheses * p < 0.10, ** p < 0.05, *** p < 0.01 The coefficients in Table 1 that speaks to our above conclusion that policy incentives and electoral incentives clash with party norms is that closeness to the party yields interesting variation. We let Figure 2 illustrate the OLS coefficients for closeness to party for each of the responsive behaviours. 21

22 Figure 2. Closeness to party and party representatives approval of colleague s responsive behaviour (coefficients from OLS-regressions shown in Table 1) Vote against the party Media Explain own position Internal informal meeting Internal formal debates Ask for personal meeting Explain party position Ask for information In Figure 2 the OLS-coefficients on the left side of the line tell us that politicians who are not close to their party are more approving of disagreeing party colleagues openly giving attention and voice to disagreeing citizens. Meanwhile, politicians who are closer to the party sympathize with responsiveness that focuses on listening responses as well as responses that aim at explaining the position of the party. Following formal party rules and addressing the issue in internal debates is not controversial (in Figure 2, that coefficient ends up on the line). In other words, attachment to the party contributes a clear line between party loyal responsiveness and responsiveness that is disloyal and/or informal. This finding strengthens an interpretation that party unity is 22

23 not mainly about party leaders using carrots and sticks. Instead, when a politician is close to the party, the concern for party unity and the willingness to keep the conflicts inside the party is evident and thereby she to a lesser degree accepts when a disagreeing colleague brings conflicts and different opinions about the party s policy in open air. Table 1 also shows that the size of the party is an important factor for politicians acceptance of responsive behaviour. Politicians from the two biggest parties, the Social Democrats and the Conservatives are significantly more sceptical towards all kinds of responsive behaviour, with an exception for explaining party position where there are no significant differences. It is possible that large parties need stricter norms of unity in order to stay united. It is also possible that smaller parties simply need to allow for representatives to be more individualistic, because they have fewer candidates to choose from. Gender and age are two other factors worth lifting forward from Table 1. Women are less approving of trying to influence leaders informally when the party is against the position of the citizens. They are also significantly less approving of externally clarifying opposing views both in the form of contacts with the media and in direct s to the contacting citizens. Probably, different mechanisms are at play for these results. One way to interpret the result is that women generally risk more on informal rules (c.f. Sundström and Wängnerud 2014) and hence are less likely to approve of internal informal behaviour that aims to change the party position. The effect on the external responses speaks to the results found in other studies that women put greater weight to party unity than do men (Scully and Farrell 2003). Taken together the results indicate that gender is an under-studied dimension in research on responsiveness to direct citizen contact in party-centered systems. 23

24 Concerning the factor age; Age shows small but significant effects on acceptance of responsiveness. The younger the politicians are, the more accepting they are of the four most popular type of responses in our investigation (listening responses as well as explaining the party position and using formal channels like internal party meetings). Hence, the younger politicians follow the party line, but express more openness to the approved of responses to citizen-initiated contacts. CONCLUSIONS This study emphasizes the role of the political party in the relationship between contacting citizens and individual politicians. We find that the party approves of certain types of responsive behaviour more than other. We also find that electoral and policy related mechanisms are more at play for internal informal responses than for other type of responses, although the main conclusion is that approval of responsiveness is not easily altered by any of the incentives we include in our treatments. Moreover, the issue we investigate a potential school closure probably reminds the respondents of a pork bringing type of issue. It is possible that when citizen-initiated contacts concern more ideological left-right issues, the party values are even more important. One important implication of the paper is that previous studies seem to have focused on types of responsiveness that are seen as problematic within parties. This means that we likely miss important dynamics that take place due to the political party s constraints on responsiveness. Political responsiveness might take place without it necessarily being observable and we might be looking for the role of electoral and policy incentives in the wrong places. Important research on political responsiveness is performed measuring responsiveness for example in the form of 24

25 voting in parliament (Miller and Stokes 1963; Butler and Nickerson 2011; Bafumi and Herron 2010; Kousser, Lewis and Masket 2007), response via to contacting citizens (Butler and Broockman 2011; Broockman 2013) and how open speech in parliament adhere to public opinion (Proksch and Slapin 2012). However, all these types of responsive behaviour towards public opinion describe responses that are open and take place outside the party. They have adherence to a direct and rather personal relationship between the politician and her perception of public opinion. Our study shows that these types of responsiveness are not only the least likely to be approved of by the party, they are possibly also the least dynamic. The interesting variation in tendency to approve of responsive behaviour from the party s side is not found for the direct and personal relationship, but for the responses that takes place informally within the party. If we want to understand acceptance of responsive behaviour under different circumstances, it is probably adaption informally within the party that we should direct our attention towards. Lastly, this paper has focused on responsiveness from the perspective of the party. The results strengthen the idea that party unity is a collective good shared by party representatives, rather that just as a leadership driven phenomenon. However, the results tell us how the respondents look upon colleagues behaviour, which is obviously not the same as how they themselves would want to respond. Together with the vast research pointing out that parties are not, in fact, unitary actors, our results points to the dual role of the representative to keep the party loyalty high, while keeping the personal passion alive. 25

26 REFERENCES Aalberg, Toril, and Jesper Strömbäck. "Media-driven Men and Media-critical Women? An Empirical Study of Gender and MPs Relationships with the Media in Norway and Sweden." International Political Science Review 32.2 (2011): Adams, F. James, Samuel Merrill, III, and Bernard Grofman A Unified Theory of Party Competition. A Cross-National Analysis Integrating Spatial and Behavioral Factors. Cambridge: Cambridge University Press. Andeweg, Rudy B. and Jacques Thomassen (2011) Pathways to party unity: Sanctions, loyalty, homogeneity and division of labour in the Dutch parliament. Party Politics 17(5): Arceneaux, K., Johnson, M., Lindstädt, R., & Wielen, R. J. V. (2015). The Influence of News Media on Political Elites: Investigating Strategic Responsiveness in Congress. American Journal of Political Science. Ashworth, Scott Reputational Dynamics and Political Careers. Journal of Law, Economics, and Organization 21 (2): Bafumi, Joseph, and Michael Herron Leapfrog Representation and Extremism: A Study of American Voters and their Members in Congress. American Political Science Review 104 (3): Brettschneider, Frank Public Opinion and Parliamentary Action: Responsiveness of the German Bundestag in Comparative Perspective. International Journal of Public Opinion Research 8 (3): Bowler, Shaun. "Private Members' Bills in the UK Parliament: Is There an Electoral Connection?." The Journal of Legislative Studies 16.4 (2010): Broockman, David E Black Politicians are More Intrinsically Motivated to Advance Blacks Interests: A Field Experiment Manipulating Political Incentives. American Journal of Political Science 57 (3): Budge, Ian, ed Mapping Policy Preferences: Estimates for Parties, Electors, and Governments, , vol. 1. Oxford: Oxford University Press. Butler, Daniel M., and David E. Broockman Do Politicians Racially Discriminate Against Constituents? A Field Experiment on State Legislators. American Journal of Political Science 55 (3): Butler, Daniel M., and David W. Nickerson Can Learning Constituency Opinion Affect How Legislators Vote? Results from a Field Experiment. Quarterly Journal of Political Science 6 (1): Butler, Daniel M., Christopher F. Karpowitz, and Jeremy C. Pope A Field Experiment on Legislators Home Styles: Service Versus Policy. Journal of Politics 74 (2):

27 Cain, Bruce E., John A. Ferejohn, and Morris P. Fiorina The Personal Vote: Constituency Service and Electoral Independence. Cambridge, MA: Harvard University Press. Desposato, Scott W. "Correcting for small group inflation of roll-call cohesion scores." British Journal of Political Science 35.4 (2005): 731. Downs, Anthony An Economic Theory of Democracy. New York: Harper & Row. Entman, Robert M The Impact of Ideology on Legislative Behavior and Public Policy in the States. The Journal of Politics 45 (1): Esaiasson, Peter, Mikael Gilljam, and Mikael Persson Communicative Responsiveness and Other Central Concepts in Between-Election Democracy. In Between-Election Democracy. The Representative Relationship After Election Day, edited by Peter Esaiasson and Hanne Marthe Narud, Colchester: ECPR Press. Fenno, Richard F., Jr Home Style: House Members in Their Districts. Boston: Little, Brown. Grose, Christian R Field Experimental Work on Political Institutions. Annua Review of Political Science 17: Harden, Jeffrey J Multidimensional Responsiveness: The Determinants of Legislators Representational Priorities. Legislative Studies Quarterly 38 (2): Jones, Bryan D Competitiveness, Role Orientations, and Legislative Responsiveness. Journal of Politics 35 (4): Kam, Christopher J. Party discipline and parliamentary politics. Cambridge University Press, Kousser, Thad, Jeffrey B. Lewis, and Seth E. Masket. "Ideological adaptation? The survival instinct of threatened legislators." Journal of Politics 69.3 (2007): Kingdon, John Congressmen s Voting Decisions. Ann Arbor: University of Michigan Press. Manin, Bernhard, Adam Przeworski, and Susan C. Stokes Introduction. In Democracy, Accountability and Representation, edited by Adam Przeworski, Susan Stokes and Bernhard Manin, Cambridge: Cambridge University Press. Mansbridge, Jane. "Rethinking representation." American political science review (2003): Mayhew, David R Congress: The Electoral Connection. New Haven, CT: Yale University Press. 27