Parliamentary Activity, Re-Selection and the Personal Vote. Evidence from Flexible-List Systems

Parliamentary Activity, Re-Selection and the Personal Vote. Evidence from Flexible-List Systems Abstract In this article, we analyse how the degree of parliamentary activity affects both individual MPs’ performance in the candidate selection process within the party and their popularity with voters at the electoral stage. We expect that parliamentary work of MPs matters less for voters’ evaluations of MPs because of limited monitoring capacities and lower salience attached to this type of representation. The empirical analysis uses data from recent elections in the Czech Republic and Sweden. During the analysed period, these countries further personalised their flexible list electoral systems. Our results suggest that parties hold MPs accountable mainly through the threat of non-re-selection rather than by assigning them to a promising list position. While there is no evidence that voters consistently reward MPs’ effort, the case of the Czech elections in 2010 shows that they may do so if context draws attention to individual MPs’ work. Members of the parliament are agents of at least two principals: the candidate selectorate within the party and the voters. To make delegation work and induce the agent to work hard on behalf of the principals, the threat of non-re-election must be credible (Fearon, 1999; Ashworth, 2012). How do the two principals consider the degree of parliamentary activity when deciding about the re-election of MPs? This is a particular pressing question with regard to the voter side, since it is not clear if citizens know or care enough about MPs’ parliamentary work to reward the hard workers and send the others packing. Whether voters are able to hold MPs to account for their record has important implications for electoral system design. Shifting influence on intra-party seat allocation from party selectors to party voters by ‘personalising’ electoral rules is a common trend at least in Europe (Renwick and Pilet, 2016). Empowering voters, however, may be detrimental to the latters’ interests, if negative effects on representatives’ behaviour eat up benefits from increased choice. This article analyses how parliamentary activity—understood as the overall effort spent on individual forms of parliamentary behaviour—affects both an MP’s performance in the candidate selection process and her popularity with voters at the electoral stage. We expect that parliamentary work of MPs matters less for voters’ evaluations of MPs because of limited monitoring capacities and lower salience attached to this type of representation. The empirical analysis uses data from recent elections in the Czech Republic (2006, 2010 and 2013) and Sweden (2010 and 2014). These countries share three features that are beneficial for the purpose of our analysis: first, they use flexible-list electoral systems, under which both the list position and the personal vote are important determinants of re-election. Secondly, since voters may also cast their ballot for a party list, the votes cast for a candidate are ‘true’ personal votes for that individual politician. Thirdly, during the analysed period, these two countries further ‘personalised’ their electoral systems, allowing study of short-run consequences of these reforms. Our results suggest that parties hold MPs accountable mainly through the threat of non-re-selection. While there is no consistent relationship between parliamentary activities and the personal vote, the case of the Czech elections in 2010 suggests that this link can materialise when context draws citizens’ attention towards MPs’ personal record. 1. Parliamentary activity and re-election seeking We are interested in examining how party selectors and voters take into account parliamentary effort of incumbent MPs when deciding about their re-election. Almost all of the existing literature studies the evaluation of incumbents by only one of these two principals, or considers overall re-election as the combined result of the two processes (e.g. Navarro, 2010). Particularly regarding candidate selection, systematic research about the characteristics that selectors value is scarce. Only few studies focus on the question how candidates’ features and incumbents’ work record affect the outcomes of specific nomination processes. Candidates seem to benefit from previous experience as MPs (Gherghina and Chiru, 2010, Pemstein et al., 2015), but it is not clear if hard work as such pays off for incumbents. For instance, bill sponsorship does not affect incumbents’ subsequent list ranking in Slovakian national elections (Crisp et al., 2013). German Members of the European Parliament (MEPs) perform better at the candidate selection stage if they were members of influential committees, but do not seem to be rewarded for high attendance rates or drafting many reports (Frech, 2016). Hermansen (2016), on the other hand, finds a positive association between legislative reports and obtaining a safe seat when analysing the electoral career of MEPs from 11 Member States. Regarding voters’ reactions to the parliamentary work of incumbent MPs, a number of studies conclude that legislative activity contributes to individual electoral performance of candidates also in parliamentary systems (Bowler, 2010, Crisp et al., 2013, Kellermann, 2013, Loewen et al., 2014, Däubler et al., 2016).1 From these findings, the ones for preferential list proportional represenation (PR) systems are remarkable, since it is well known that candidates ranked at the top of the list receive a very large share of the preference vote (e.g. Marcinkiewicz and Stegmaier, 2015). One should, however, keep in mind that those results refer to one specific type of activity—initiating legislative bills, and in some cases certain types of legislative bills—rather than a measure of overall effort across different types of activity. There are two main principals which hold MPs accountable for their degree of parliamentary activity: party selectors and voters. We assume that party selectors and voters take into account the past behaviour of MPs to form an expectation about how they will behave in the future. If MPs anticipate to receive a reward for parliamentary work (or being sanctioned for a lack thereof), the pending verdict of the principals will also induce accountability by anticipation. We suggest that the principals differ concerning their ability to hold MPs accountable for two main reasons. First, party selectors can be expected to have a higher level of monitoring capacity (cp. Fearon, 1999; André et al., 2014) than voters. Party selectors should follow parliamentary proceedings and the individual activities of their MPs more closely than voters do. They should both be more aware of the importance of being represented well (or have higher personal stakes), and thus be more willing to pay the cost of monitoring from an instrumental perspective. At the same time, as actors in one way or another involved in politics, party selectors may also simply have a stronger intrinsic motivation to follow how MPs do their job. It is, however, more difficult to answer the question if voters know enough to reward or sanction MPs for their exerted effort levels. In this context it is important to note that there can be two different mechanisms for how effort improves reputation (Däubler et al., 2016). First, it is possible that principals base their decision directly and consciously on merit. They observe the effort by the agent, and credit her for the engagement as such. This direct form of assessment may also be facilitated by third parties providing fire alarm-type information—with quantitative data on parliamentary work being more easily available than in the past; media and websites may, for instance, directly report on the amount of activity.2 Secondly, learning about agent effort may take indirect routes, since work can create visibility and name recognition (e.g. Cain et al., 1987; Wilson et al., 2016). In addition—and independent of their knowledge about what MPs do—voters may also differ from party actors by preferring MPs who put less relative emphasis on parliamentary work as compared to other forms of representation (such as constituency work in a narrow sense). What MPs contribute in parliament to legislation or government oversight may not have an immediately visible relevance or benefit for people’s lives. This may lead citizens to prefer constituency service, which is more tangible in nature.3 In addition, voters get to ‘evaluate’ MPs only after the party has ranked the candidates on the list, which may further limit the scope for a positive association between parliamentary work and the personal vote. Our central expectation is, therefore, that the overall degree of parliamentary work has a stronger effect on the evaluation of MPs by party selectors than it has for that by voters. Note, however, that we refrain from making a prediction about the absolute level of the association we expect to find for voters. We empirically analyse the actual outcomes of both the nomination and electoral stage, using data from recent elections in the Czech Republic and Sweden. The flexible-list systems employed make the Czech Republic and Sweden particularly suitable cases for testing the arguments. Parties present voters with pre-ordered lists, and the order of candidates will be consequential for intra-party seat allocation unless all candidates clear the preference vote threshold. Since preference voting is not obligatory, the votes to candidates are ‘true’ personal votes, rather than party votes in disguise. In addition, the two countries under study reduced the preference vote threshold of their flexible-list system in a recent reform. This change makes it easier for candidates to be elected on the basis of personal votes rather than due to a good pre-electoral list position. So the incentives for incumbent MPs may change, and this again may also alter how party selectors and voters hold representatives to account. With stronger incentives to cater to voters directly, MPs may invest more resources in communicating with voters, and media may report more intensively on individual behaviour. Then, the reform may also make it easier for voters to hold individual representatives accountable for their activities. For each country, we study the last pre-reform election and all post-reform elections that had been passed at the time of writing. 2. Institutional context 2.1. Czech Republic The Czech Chamber of Deputies (Poslanecká sněmovna) consists of 200 MPs who are elected based on a flexible-list PR system. There are 14 constituencies mirroring the administrative division of the country, ranging in magnitude from 5 to 25 seats. Parties with at least 5% of the vote at the national level gain seats, which are distributed among parties based on their support within constituencies. Voters have the opportunity to cast preference votes for multiple candidates within one party list. The number of preference votes increased from two to four since the 2010 election. Candidates who have reached the preference vote threshold are moved to the top of the list and are ranked according to the number of personal votes received; for all other candidates pre-electoral list position applies. The threshold was lowered from 7% to 5% since the 2010 election (for more details about the preference vote rules, see Marcinkiewicz and Stegmaier (2015)). Together with the increase in the number of preference votes, this led to a considerable increase in preference voting, with the number of preference votes per ballot rising from 0.38 to 0.68.4 Candidate selection takes place at the local and regional level in most parties, and the influence of the national-level party organisations is generally limited. The process is remarkably similar in all major parties (ČSSD, KSČM, KDU-ČSL and ODS) covered in the present analysis (Outlý and Prouza, 2009). In general, party selectorates reward candidates with higher preferential votes in previous elections and put them on better list positions (André et al., 2017). In the Czech Republic, individual MPs enjoy extensive rights in the legislative process. The government has only weak control over the agenda which is decided mainly by majority vote in the Chamber (Zubek and Stecker, 2010). Moreover, not only the government, but also an individual MP or a group of them are allowed to introduce bills. While the number of private members’ bills has declined since the beginning of the post-communist transition period (Linek and Mansfeldová, 2007), their success rate is considerable. Most of the bills submitted by individual MPs (rather than a group), however, seek to put an issue onto the agenda rather than aiming at actually adjusting the law (Kolář et al., 2013, p. 238). In addition to that, an individual MP is allowed to propose amendments to the bills. Almost all decisions of the Chamber are done by registered electronic roll-call votes. During the four-year term, MPs usually take more than 10,000 votes and on average 50 votes per meeting day (Linek and Mansfeldová, 2007, pp. 30–32). Since 1996, there is an informal agreement among government and opposition parties to ‘pair’ an absenting MP so that the balance between government and opposition is not changed. This artificially leads to increase in abstentions, especially for the opposition MPs who are pairs for MPs who are also cabinet members. The Chamber uses oral and written questions, which are called interpellations. Any MP has the right to engage in both forms of interpellations. Oral interpellations take place once a week. They need to be submitted beforehand, their order on the agenda is drawn in the morning of the day they are covered. Written interpellations can be answered verbally or in writing. If an MP is not satisfied with an answer, he can ask the President of the Chamber to include the question on the agenda. MPs have the possibility to ask the government and ministers about any topic which is covered by their field of action (Kolář et al., 2013, p. 320). In interpellations, MPs are quite independent from the party leadership. In MP surveys, however, representatives state that interpellations are blunt instruments for influencing the government. Private members’ bills and amendments receive better assessments by MPs in this regard (Mansfeldová and Linek, 2009, pp. 88–93). 2.2. Sweden The Swedish national parliament, the Riksdag, consists of 349 seats, of which 310 are allocated to the 29 electoral constituencies, and 39 are levelling (second-tier compensation) seats. The Swedish preferential-list PR system is a flexible-list system, allowing voters to choose between casting a list vote and expressing a preference for a single candidate on a party list. If a candidate gets enough votes to pass the personal vote threshold, this candidate will receive the first seat allocated to the party. The threshold was lowered from 8% to 5% in the 2014 elections. Politicians widely agreed that the change was unlikely to have dramatic effects on various aspects of intra-party and MP–voter relationships (Berg and Oscarsson, 2015).5 If more candidates cross the threshold, the candidates are ranked according to number of personal votes they receive. Any remaining seats are distributed according to the order on the party list. The Swedish parliament is characterised by strong party discipline (Larsson and Bäck, 2008, p. 160). MPs, however, have several tools to pursue more individual representation. Individual MPs have the right to initiate private members’ bills (motioner). The so-called stand-alone bills, that is, proposals that do not seek to amend other proposals, may be initiated during a certain time period each year (from the start of the session until shortly after submission of the budget). Swedish MPs also have the right to ask questions to ministers and other government members. There is a weekly question time with spontaneous oral questions, interpellations (submitted in writing but answered by government members personally) and written questions that receive written answers. Roll-call votes occur frequently, and the threshold for a roll call to take place is low. Public attention to MPs’ activities is generally limited (Bergman, 2006, p. 607), and the activities seldom make the news (Larsson and Bäck, 2008). However, MPs find questions useful for creating media attention in local constituencies (Bergman, 2006, p. 604), and their usage has increased over time (Bergman and Bolin, 2013, pp. 269–270). Individual motions very seldom receive media coverage, since these are almost never passed. Candidate selection in Swedish parties usually takes place at the regional level (Aylott, 2013, p. 322). There is some disagreement, though, about the importance of local representation in the Riksdag. Esaiasson and Holmberg (1996) concluded from their analyses that local interest promotion is strong, but Bergman and Bolin (2013, p. 261) argue that, once elected, MPs perceive themselves rather as party/national than constituency representatives. Folke et al. (2016) also show that, on the local level, party selectors appear to use the personal vote to identify popular politicians and to promote them to positions of power within the party. Although only a minority of voters use the personal vote (23% in 2010 and 25% in 2014), its usage is distributed evenly across social groups (e.g. income, education, geography) and other voter characteristics (Berg and Oscarsson, 2015). According to data from the 2014 Swedish National Election Study, the most common motive for rewarding a candidate with a vote was individual candidate traits such as competence and reliability. Amongst the voters who did not use the personal vote, the most common reason was lacking information about candidates. 3. Data, measures and models To examine the arguments outlined above, we consider the link between parliamentary activities of incumbent MPs and their performance at the subsequent party nomination stage and their personal vote in the following elections. We analyse the consequences of parliamentary work undertaken during one pre-electoral reform and all post-reform legislative periods in the Czech Republic (2002–2006, 2006–2010 and 2010–2013) and Sweden (2006–2010 and 2010–2014). The two parliaments provide databases that contain information about activities and the parliamentary biography of politicians. We combined parliamentary data with information obtained from the election authorities, which provide details about the ballots/party lists used and the electoral performance of parties and individual politicians. For each legislative period, we analyse three dependent variables: whether MPs re-run for the same party, whether they obtain a promising position on the list (defined as being within the first N ranks, where N is the number of seats the party won in the previous election), and the intra-party vote share (for details on its transformation see below).6 The first type of analyses uses the sample of MPs who are in office at the end of the third year of the legislative term (this should approximately reflect the time point when selection processes start), and who have held their mandate for at least 200 days. We exclude any MPs who have been party leaders or government members during the term. The analyses of list positions and the personal vote covers those MPs who run for the same party in the same constituency as in the previous election, since we include a lagged dependent variable. As measures of the degree of parliamentary work, we consider two indicators.7 The first one is a summary measure of activities that are individual in a sense that they can be assumed to be fairly independent of party influence (and for which data are available). In the Czech Republic, we consider single-authored private members’ bills, oral interpellations and amendments submitted during the second reading of a bill.8 In the Swedish case, we look at single-authored private members’ bills (i.e. the stand-alone proposals from the respective yearly period), the number of oral interpellations and the number of written parliamentary questions.9 To create our summary measure, we divide each MP’s activity count by the duration of her mandate in that year, standardise within each activity–party-year, and take the mean over all the values (activities times years) for each MP (see Online Appendix A1 for more details and examples).10 Our second indicator is the share of missed roll-call votes among all votes during an MP’s mandate, standardised within parties.11 The measures, thus, capture how productive MPs are, but not the quality of their work. This approach has limitations, but what constitutes ‘good’ parliamentary work is hard to define and even harder to measure. When examining the effect of re-election-seeking activities on electoral performance, it is important to control for factors reflecting marginality, that is, the need to be active in the first place. We use lagged measures for this purpose (for indicators for list leader and for previous promising position, the logit-transformed personal vote and the ratio of the number of candidates and party seats as a measure of intra-party competition, see Crisp et al. (2007)). In addition, we also include variables for being a replacement MP, party group leadership (share of term serving as group or vice group leader), previous parliamentary experience (none, one term, two or more terms; counting spells with a minimum duration of 100 days), being the sole representative of a party-in-a-constituency, the share of term the party was in opposition, age and age squared and gender. In the Czech case, we also have data on involvement in subnational elections (obtaining a mandate at the municipality level or running in regional elections during the term as an MP). When analysing the personal vote, an indicator variable for the first list position and a continuous variable for list rank also enter the models, since the outcome of the nomination stage is of course an important predictor of the personal vote. Descriptive statistics are provided in Online Appendix A2. We use standard logit models for analysing the re-selection outcomes. For the personal vote models, we log-transform intra-party vote shares, from which we subtract the respective value of a reference category (ballots cast for the list and for candidates not in the sample), and then use ordinary least squares estimation. This model can be derived from a utility-maximisation framework, making certain simplifying assumptions (cp. Berry et al., 1995). To take into account unobserved heterogeneity at the party-list level, we use clustered standard errors.12 We note that the aim of our analysis is to examine whether parliamentary activity predicts (rather than causes) performance at the re-selection and re-election stages, and we acknowledge that the comparison between selectors and voters is merely an indirect test of their monitoring capacities. 4. Results Table 1 displays the results of the regression models for the re-selection stage, split by legislative term, for the Czech Republic. Models C1–C3 consider the binary indicator of running for the same party in the subsequent election as dependent variable, Models C4–C6 explain whether an MP received a promising list position. Shown are logit coefficients along with standard errors in parentheses. Table 1 Results from logistic regression of reselection outcomes in the Czech Republic Predictor Dep. var.: Re-running for same party Dep. var.: Good list position C1: 2006 C2: 2010 C3: 2013 C4: 2006 C5: 2010 C6: 2013 Parl. activity Years 1–3 (party-std.) 0.65 0.05 1.43** 0.41 −0.67 −0.44 (0.57) (0.62) (0.59) (0.69) (0.73) (0.61) Missed RCV Years 1–3 (party-std.) −0.61*** 0.34 −0.46** 0.10 0.30 0.10 (0.23) (0.25) (0.24) (0.30) (0.40) (0.33) Lag list leader 0.06 1.69 1.30 −0.01 1.23 −0.34 (0.70) (1.08) (0.88) (0.75) (0.77) (0.88) Lag good list pos. 1.20 −1.07 −1.47* −1.15 0.02 1.93** (2.06) (0.74) (0.76) (2.05) (1.30) (0.77) Lag logit pers. vote 0.51 0.47 0.75 0.73*** 0.47 1.42*** (0.35) (0.34) (0.58) (0.27) (0.32) (0.51) Lag intra-p. comp. −0.25** 0.04 −0.18* −0.07 −0.13* −0.03 (0.10) (0.10) (0.10) (0.17) (0.07) (0.09) Replacement MP 1.96 −2.30** −0.89 −0.34 1.69 (2.16) (1.04) (1.19) (2.07) (1.60) Sole representative −0.54 −3.74* 0.23 0.73 0.33 (1.73) (2.12) (1.12) (1.59) (1.29) PPG leadership −0.13 −0.60 0.87 1.41 −0.35 0.49 (0.74) (0.89) (0.71) (1.08) (0.92) (0.69) Seniority: 1 term exp. −1.14** 0.53 −1.17* 0.30 0.11 0.66 (0.57) (0.67) (0.60) (0.87) (0.66) (0.89) Seniority: ≥2 terms exp. −0.59 −1.57*** −0.58 −0.95 −1.12 −0.99 (0.50) (0.58) (0.71) (0.74) (0.80) (0.87) Opposition 1.49*** 0.32 2.82*** 0.64 0.66 0.69 (0.52) (0.49) (0.68) (0.48) (0.62) (0.67) Age 0.03 −0.15 0.26* −0.51 −0.25 0.14 (0.16) (0.22) (0.13) (0.50) (0.21) (0.18) Age squared/1000 −0.40 1.09 −2.88** 4.03 2.18 −1.89 (1.73) (2.21) (1.42) (4.69) (2.38) (2.01) Female 0.42 −0.68 −0.10 −0.62 −1.18 −0.09 (0.64) (0.45) (0.63) (0.60) (0.72) (0.63) Subnational politics 0.56 0.65 −0.85* 0.07 0.20 0.14 (0.48) (0.40) (0.48) (0.50) (0.41) (0.47) Constant 2.79 8.50 0.09 20.41 10.66** 1.65 (4.37) (5.29) (3.49) (12.71) (4.83) (4.95) N 182 171 161 141 120 114 Predictor Dep. var.: Re-running for same party Dep. var.: Good list position C1: 2006 C2: 2010 C3: 2013 C4: 2006 C5: 2010 C6: 2013 Parl. activity Years 1–3 (party-std.) 0.65 0.05 1.43** 0.41 −0.67 −0.44 (0.57) (0.62) (0.59) (0.69) (0.73) (0.61) Missed RCV Years 1–3 (party-std.) −0.61*** 0.34 −0.46** 0.10 0.30 0.10 (0.23) (0.25) (0.24) (0.30) (0.40) (0.33) Lag list leader 0.06 1.69 1.30 −0.01 1.23 −0.34 (0.70) (1.08) (0.88) (0.75) (0.77) (0.88) Lag good list pos. 1.20 −1.07 −1.47* −1.15 0.02 1.93** (2.06) (0.74) (0.76) (2.05) (1.30) (0.77) Lag logit pers. vote 0.51 0.47 0.75 0.73*** 0.47 1.42*** (0.35) (0.34) (0.58) (0.27) (0.32) (0.51) Lag intra-p. comp. −0.25** 0.04 −0.18* −0.07 −0.13* −0.03 (0.10) (0.10) (0.10) (0.17) (0.07) (0.09) Replacement MP 1.96 −2.30** −0.89 −0.34 1.69 (2.16) (1.04) (1.19) (2.07) (1.60) Sole representative −0.54 −3.74* 0.23 0.73 0.33 (1.73) (2.12) (1.12) (1.59) (1.29) PPG leadership −0.13 −0.60 0.87 1.41 −0.35 0.49 (0.74) (0.89) (0.71) (1.08) (0.92) (0.69) Seniority: 1 term exp. −1.14** 0.53 −1.17* 0.30 0.11 0.66 (0.57) (0.67) (0.60) (0.87) (0.66) (0.89) Seniority: ≥2 terms exp. −0.59 −1.57*** −0.58 −0.95 −1.12 −0.99 (0.50) (0.58) (0.71) (0.74) (0.80) (0.87) Opposition 1.49*** 0.32 2.82*** 0.64 0.66 0.69 (0.52) (0.49) (0.68) (0.48) (0.62) (0.67) Age 0.03 −0.15 0.26* −0.51 −0.25 0.14 (0.16) (0.22) (0.13) (0.50) (0.21) (0.18) Age squared/1000 −0.40 1.09 −2.88** 4.03 2.18 −1.89 (1.73) (2.21) (1.42) (4.69) (2.38) (2.01) Female 0.42 −0.68 −0.10 −0.62 −1.18 −0.09 (0.64) (0.45) (0.63) (0.60) (0.72) (0.63) Subnational politics 0.56 0.65 −0.85* 0.07 0.20 0.14 (0.48) (0.40) (0.48) (0.50) (0.41) (0.47) Constant 2.79 8.50 0.09 20.41 10.66** 1.65 (4.37) (5.29) (3.49) (12.71) (4.83) (4.95) N 182 171 161 141 120 114 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. In 2006, being the sole representative predicts obtaining a good list position perfectly, N = 2 cases were dropped. In 2010, being a replacement MP predicts obtaining a good list position perfectly, N = 3 cases were dropped. *p < 0.1; **p < 0.05; ***p < 0.01. Table 1 Results from logistic regression of reselection outcomes in the Czech Republic Predictor Dep. var.: Re-running for same party Dep. var.: Good list position C1: 2006 C2: 2010 C3: 2013 C4: 2006 C5: 2010 C6: 2013 Parl. activity Years 1–3 (party-std.) 0.65 0.05 1.43** 0.41 −0.67 −0.44 (0.57) (0.62) (0.59) (0.69) (0.73) (0.61) Missed RCV Years 1–3 (party-std.) −0.61*** 0.34 −0.46** 0.10 0.30 0.10 (0.23) (0.25) (0.24) (0.30) (0.40) (0.33) Lag list leader 0.06 1.69 1.30 −0.01 1.23 −0.34 (0.70) (1.08) (0.88) (0.75) (0.77) (0.88) Lag good list pos. 1.20 −1.07 −1.47* −1.15 0.02 1.93** (2.06) (0.74) (0.76) (2.05) (1.30) (0.77) Lag logit pers. vote 0.51 0.47 0.75 0.73*** 0.47 1.42*** (0.35) (0.34) (0.58) (0.27) (0.32) (0.51) Lag intra-p. comp. −0.25** 0.04 −0.18* −0.07 −0.13* −0.03 (0.10) (0.10) (0.10) (0.17) (0.07) (0.09) Replacement MP 1.96 −2.30** −0.89 −0.34 1.69 (2.16) (1.04) (1.19) (2.07) (1.60) Sole representative −0.54 −3.74* 0.23 0.73 0.33 (1.73) (2.12) (1.12) (1.59) (1.29) PPG leadership −0.13 −0.60 0.87 1.41 −0.35 0.49 (0.74) (0.89) (0.71) (1.08) (0.92) (0.69) Seniority: 1 term exp. −1.14** 0.53 −1.17* 0.30 0.11 0.66 (0.57) (0.67) (0.60) (0.87) (0.66) (0.89) Seniority: ≥2 terms exp. −0.59 −1.57*** −0.58 −0.95 −1.12 −0.99 (0.50) (0.58) (0.71) (0.74) (0.80) (0.87) Opposition 1.49*** 0.32 2.82*** 0.64 0.66 0.69 (0.52) (0.49) (0.68) (0.48) (0.62) (0.67) Age 0.03 −0.15 0.26* −0.51 −0.25 0.14 (0.16) (0.22) (0.13) (0.50) (0.21) (0.18) Age squared/1000 −0.40 1.09 −2.88** 4.03 2.18 −1.89 (1.73) (2.21) (1.42) (4.69) (2.38) (2.01) Female 0.42 −0.68 −0.10 −0.62 −1.18 −0.09 (0.64) (0.45) (0.63) (0.60) (0.72) (0.63) Subnational politics 0.56 0.65 −0.85* 0.07 0.20 0.14 (0.48) (0.40) (0.48) (0.50) (0.41) (0.47) Constant 2.79 8.50 0.09 20.41 10.66** 1.65 (4.37) (5.29) (3.49) (12.71) (4.83) (4.95) N 182 171 161 141 120 114 Predictor Dep. var.: Re-running for same party Dep. var.: Good list position C1: 2006 C2: 2010 C3: 2013 C4: 2006 C5: 2010 C6: 2013 Parl. activity Years 1–3 (party-std.) 0.65 0.05 1.43** 0.41 −0.67 −0.44 (0.57) (0.62) (0.59) (0.69) (0.73) (0.61) Missed RCV Years 1–3 (party-std.) −0.61*** 0.34 −0.46** 0.10 0.30 0.10 (0.23) (0.25) (0.24) (0.30) (0.40) (0.33) Lag list leader 0.06 1.69 1.30 −0.01 1.23 −0.34 (0.70) (1.08) (0.88) (0.75) (0.77) (0.88) Lag good list pos. 1.20 −1.07 −1.47* −1.15 0.02 1.93** (2.06) (0.74) (0.76) (2.05) (1.30) (0.77) Lag logit pers. vote 0.51 0.47 0.75 0.73*** 0.47 1.42*** (0.35) (0.34) (0.58) (0.27) (0.32) (0.51) Lag intra-p. comp. −0.25** 0.04 −0.18* −0.07 −0.13* −0.03 (0.10) (0.10) (0.10) (0.17) (0.07) (0.09) Replacement MP 1.96 −2.30** −0.89 −0.34 1.69 (2.16) (1.04) (1.19) (2.07) (1.60) Sole representative −0.54 −3.74* 0.23 0.73 0.33 (1.73) (2.12) (1.12) (1.59) (1.29) PPG leadership −0.13 −0.60 0.87 1.41 −0.35 0.49 (0.74) (0.89) (0.71) (1.08) (0.92) (0.69) Seniority: 1 term exp. −1.14** 0.53 −1.17* 0.30 0.11 0.66 (0.57) (0.67) (0.60) (0.87) (0.66) (0.89) Seniority: ≥2 terms exp. −0.59 −1.57*** −0.58 −0.95 −1.12 −0.99 (0.50) (0.58) (0.71) (0.74) (0.80) (0.87) Opposition 1.49*** 0.32 2.82*** 0.64 0.66 0.69 (0.52) (0.49) (0.68) (0.48) (0.62) (0.67) Age 0.03 −0.15 0.26* −0.51 −0.25 0.14 (0.16) (0.22) (0.13) (0.50) (0.21) (0.18) Age squared/1000 −0.40 1.09 −2.88** 4.03 2.18 −1.89 (1.73) (2.21) (1.42) (4.69) (2.38) (2.01) Female 0.42 −0.68 −0.10 −0.62 −1.18 −0.09 (0.64) (0.45) (0.63) (0.60) (0.72) (0.63) Subnational politics 0.56 0.65 −0.85* 0.07 0.20 0.14 (0.48) (0.40) (0.48) (0.50) (0.41) (0.47) Constant 2.79 8.50 0.09 20.41 10.66** 1.65 (4.37) (5.29) (3.49) (12.71) (4.83) (4.95) N 182 171 161 141 120 114 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. In 2006, being the sole representative predicts obtaining a good list position perfectly, N = 2 cases were dropped. In 2010, being a replacement MP predicts obtaining a good list position perfectly, N = 3 cases were dropped. *p < 0.1; **p < 0.05; ***p < 0.01. Effort and the relative frequency of missing roll-call votes show the expected associations with running again in some, but not all of the elections. The coefficient of parliamentary activity is of considerable size both in 2006 and 2013 (but statistically significant only in the latter case). Those MPs missing more roll-call votes were less likely to stand again in 2006 and 2013. The 2010 election, and we will see this repeatedly, shows a different pattern. The coefficient for the activity variable is much smaller, and the one for the roll-call vote measure even has the opposite sign. The relationships for 2006 and 2013 are of considerable size. If we take 0.75 as a baseline probability (approximately the mean of the 2006 and 2013 proportion of re-running MPs), a change of the size of the interquartile range of the respective variable implies expected probabilities of re-running of 0.84 (for activity in 2013), of 0.57 (missed roll calls in 2006) and of 0.62 (missed roll calls in 2013). These are substantive changes—the question about the direction of causality is justified, though. The findings do not necessarily provide evidence for party selectors sorting out the not so hard-working MPs. The pattern would also be consistent with those MPs voluntarily withdrawing from parliament exerting less effort. We cannot disentangle these processes in our setup, but we control for age and seniority, which may alleviate the reversed causality problem to some extent. Space constraints do generally not permit us to discuss findings for control variables, we therefore only occasionally point to particularly interesting findings. The fact that in 2006 and 2013 intra-party competition for seats is linked to a lower chance of running again also points into the direction that we are not only observing a last-period effect for the parliamentary effort-related variables. There is no evidence that the amount of parliamentary work or participation in votes helps re-running MPs to obtain a promising list position. None of the key variables reaches statistical significance in these models, and the signs are inconsistent (activity) or contrary to the expectation (roll calls). Overall, none of the explanatory variables shows consistent results across years in these models. Even the coefficient of the lagged personal vote is estimated with high uncertainty in 2010. Table 2 reports results of the parallel analysis conducted for the Swedish elections in 2010 and 2014. Also here, the more active MPs are more likely to run again (statistically significant at 10% in 2010 and at 5% in 2014). Taking the sample mean of 0.77 as a baseline, the expected probability changes to 0.81 (2014) when considering a difference in activity of the size of the interquartile range. The roll-call vote variables have a negative sign as expected, but are far from significant. The decision to run again strongly depends on seniority and age, with the latter showing an inverse U-shape pattern, with the maximum at approximately 39 and 42 years, respectively. Table 2 Results from logistic regression of re-selection outcomes in Sweden Predictor Dep. var.: Re-running for same party Dep. var.: Good list position S1: 2010 S2: 2014 S3: 2010 S4: 2014 Parl. activity Years 1–3 (party-std.) 0.59* 0.62** 0.21 −0.45 (0.36) (0.28) (0.33) (0.38) Missed RCV Years 1–3 (party-std.) −0.02 −0.17 −0.24 0.32 (0.11) (0.15) (0.35) (0.30) Lag list leader −0.14 0.34 1.30 0.43 (0.48) (0.40) (0.87) (0.96) Lag good list pos. 0.72 −2.59 0.26 3.68*** (2.05) (1.97) (1.13) (0.98) Lag logit pers. vote 0.14 0.02 0.09 1.04*** (0.19) (0.17) (0.22) (0.26) Lag intra-p. comp. 0.00 −0.01 −0.08* −0.07 (0.04) (0.03) (0.05) (0.05) Replacement MP 0.23 −2.38 −1.68 0.52 (2.25) (2.03) (1.15) (1.10) Sole representative −0.13 −0.64 0.77 −0.85 (0.80) (0.73) (1.26) (1.28) PPG leadership 0.52 1.53 0.17 2.03 (0.98) (1.34) (1.51) (1.53) Seniority: 1 term exp. −1.16** −1.02** 0.81 −0.85 (0.49) (0.51) (0.60) (0.65) Seniority: ≥2 terms exp. −2.42*** −1.78*** 0.58 −2.37*** (0.45) (0.53) (0.65) (0.82) Opposition −0.15 −0.14 0.51 0.41 (0.36) (0.32) (0.53) (0.68) Age 0.30 0.32** 0.15 0.08 (0.18) (0.12) (0.15) (0.19) Age squared/1000 −3.82* −3.78*** −2.44 −1.16 (1.96) (1.36) (1.58) (2.06) Female 0.06 −0.29 0.09 0.32 (0.29) (0.27) (0.36) (0.54) Constant −2.25 −0.96 0.47 4.03 (4.70) (2.70) (3.58) (4.66) N 331 316 253 243 Predictor Dep. var.: Re-running for same party Dep. var.: Good list position S1: 2010 S2: 2014 S3: 2010 S4: 2014 Parl. activity Years 1–3 (party-std.) 0.59* 0.62** 0.21 −0.45 (0.36) (0.28) (0.33) (0.38) Missed RCV Years 1–3 (party-std.) −0.02 −0.17 −0.24 0.32 (0.11) (0.15) (0.35) (0.30) Lag list leader −0.14 0.34 1.30 0.43 (0.48) (0.40) (0.87) (0.96) Lag good list pos. 0.72 −2.59 0.26 3.68*** (2.05) (1.97) (1.13) (0.98) Lag logit pers. vote 0.14 0.02 0.09 1.04*** (0.19) (0.17) (0.22) (0.26) Lag intra-p. comp. 0.00 −0.01 −0.08* −0.07 (0.04) (0.03) (0.05) (0.05) Replacement MP 0.23 −2.38 −1.68 0.52 (2.25) (2.03) (1.15) (1.10) Sole representative −0.13 −0.64 0.77 −0.85 (0.80) (0.73) (1.26) (1.28) PPG leadership 0.52 1.53 0.17 2.03 (0.98) (1.34) (1.51) (1.53) Seniority: 1 term exp. −1.16** −1.02** 0.81 −0.85 (0.49) (0.51) (0.60) (0.65) Seniority: ≥2 terms exp. −2.42*** −1.78*** 0.58 −2.37*** (0.45) (0.53) (0.65) (0.82) Opposition −0.15 −0.14 0.51 0.41 (0.36) (0.32) (0.53) (0.68) Age 0.30 0.32** 0.15 0.08 (0.18) (0.12) (0.15) (0.19) Age squared/1000 −3.82* −3.78*** −2.44 −1.16 (1.96) (1.36) (1.58) (2.06) Female 0.06 −0.29 0.09 0.32 (0.29) (0.27) (0.36) (0.54) Constant −2.25 −0.96 0.47 4.03 (4.70) (2.70) (3.58) (4.66) N 331 316 253 243 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. Table 2 Results from logistic regression of re-selection outcomes in Sweden Predictor Dep. var.: Re-running for same party Dep. var.: Good list position S1: 2010 S2: 2014 S3: 2010 S4: 2014 Parl. activity Years 1–3 (party-std.) 0.59* 0.62** 0.21 −0.45 (0.36) (0.28) (0.33) (0.38) Missed RCV Years 1–3 (party-std.) −0.02 −0.17 −0.24 0.32 (0.11) (0.15) (0.35) (0.30) Lag list leader −0.14 0.34 1.30 0.43 (0.48) (0.40) (0.87) (0.96) Lag good list pos. 0.72 −2.59 0.26 3.68*** (2.05) (1.97) (1.13) (0.98) Lag logit pers. vote 0.14 0.02 0.09 1.04*** (0.19) (0.17) (0.22) (0.26) Lag intra-p. comp. 0.00 −0.01 −0.08* −0.07 (0.04) (0.03) (0.05) (0.05) Replacement MP 0.23 −2.38 −1.68 0.52 (2.25) (2.03) (1.15) (1.10) Sole representative −0.13 −0.64 0.77 −0.85 (0.80) (0.73) (1.26) (1.28) PPG leadership 0.52 1.53 0.17 2.03 (0.98) (1.34) (1.51) (1.53) Seniority: 1 term exp. −1.16** −1.02** 0.81 −0.85 (0.49) (0.51) (0.60) (0.65) Seniority: ≥2 terms exp. −2.42*** −1.78*** 0.58 −2.37*** (0.45) (0.53) (0.65) (0.82) Opposition −0.15 −0.14 0.51 0.41 (0.36) (0.32) (0.53) (0.68) Age 0.30 0.32** 0.15 0.08 (0.18) (0.12) (0.15) (0.19) Age squared/1000 −3.82* −3.78*** −2.44 −1.16 (1.96) (1.36) (1.58) (2.06) Female 0.06 −0.29 0.09 0.32 (0.29) (0.27) (0.36) (0.54) Constant −2.25 −0.96 0.47 4.03 (4.70) (2.70) (3.58) (4.66) N 331 316 253 243 Predictor Dep. var.: Re-running for same party Dep. var.: Good list position S1: 2010 S2: 2014 S3: 2010 S4: 2014 Parl. activity Years 1–3 (party-std.) 0.59* 0.62** 0.21 −0.45 (0.36) (0.28) (0.33) (0.38) Missed RCV Years 1–3 (party-std.) −0.02 −0.17 −0.24 0.32 (0.11) (0.15) (0.35) (0.30) Lag list leader −0.14 0.34 1.30 0.43 (0.48) (0.40) (0.87) (0.96) Lag good list pos. 0.72 −2.59 0.26 3.68*** (2.05) (1.97) (1.13) (0.98) Lag logit pers. vote 0.14 0.02 0.09 1.04*** (0.19) (0.17) (0.22) (0.26) Lag intra-p. comp. 0.00 −0.01 −0.08* −0.07 (0.04) (0.03) (0.05) (0.05) Replacement MP 0.23 −2.38 −1.68 0.52 (2.25) (2.03) (1.15) (1.10) Sole representative −0.13 −0.64 0.77 −0.85 (0.80) (0.73) (1.26) (1.28) PPG leadership 0.52 1.53 0.17 2.03 (0.98) (1.34) (1.51) (1.53) Seniority: 1 term exp. −1.16** −1.02** 0.81 −0.85 (0.49) (0.51) (0.60) (0.65) Seniority: ≥2 terms exp. −2.42*** −1.78*** 0.58 −2.37*** (0.45) (0.53) (0.65) (0.82) Opposition −0.15 −0.14 0.51 0.41 (0.36) (0.32) (0.53) (0.68) Age 0.30 0.32** 0.15 0.08 (0.18) (0.12) (0.15) (0.19) Age squared/1000 −3.82* −3.78*** −2.44 −1.16 (1.96) (1.36) (1.58) (2.06) Female 0.06 −0.29 0.09 0.32 (0.29) (0.27) (0.36) (0.54) Constant −2.25 −0.96 0.47 4.03 (4.70) (2.70) (3.58) (4.66) N 331 316 253 243 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. As in the Czech case, the results from the models with receiving a promising list position as dependent variable remain inconclusive. The signs of the effort-related variables are as expected only for the 2010 election. The most interesting finding is that in 2014 both the lagged dependent variable and the lagged personal vote turn out as strong predictors, while this was not the case in 2010. With 2014 being the first election after lowering the preference vote threshold, this could be interpreted in a way that the selectorate in the political parties put more weight on candidate popularity before the elections to be held under ‘personalised’ electoral rules. Of course, it is impossible to infer a causal effect of the electoral reform from this simple change over time. Another noteworthy result is that in 2014 the most senior MPs were less likely to receive a good list position, whereas in 2010 this variable had a positive sign. One interpretation of this finding is that some longer-serving MPs lose the determination to stay in parliament, but are still included on the list, leaving the decision to the voters whether to move them up the list again. We can also speculate that party selectors’ incentives to retain well-known MPs on the list merely for their vote-attraction potential increase in anticipation of an election held under personalised electoral rules. We are now turning to the analysis of the personal vote results (see Table 3 for the Czech Republic).13 Again, the 2010 election stands out from the others. There is a highly significant association between the degree of parliamentary activity and the personal vote MPs achieved. Consider again the implied change when increasing parliamentary activity by as much as the interquartile range: it amounts to an approximate shift from 0.047 (which is the median of the difference in intra-party vote share relative to the reference category) to 0.056. While not exactly huge, we should keep in mind that the preference vote threshold was lowered from 7% to 5% of the party vote, so 0.9 percentage points can make a difference for an MP’s direct election. Attendance at parliamentary votes, on the other hand, does not predict MPs’ electoral fortunes in any of the years. Table 3 Results from binomial-logistic regression of preference vote share in the Czech Republic Predictor Dep. var.: Log of preference votes relative to reference category C7: 2006 C8: 2010 C9: 2013 Parl. activity Years 1–4 (party-std.) 0.15* 0.35*** 0.07 (0.09) (0.09) (0.09) Missed RCV Years 1–4 (party-std.) −0.02 0.02 −0.001 (0.06) (0.04) (0.05) List leader 0.14 −0.01 −0.03 (0.10) (0.09) (0.09) List rank (linear) −0.10*** −0.02** −0.03*** (0.02) (0.01) (0.01) Lag list leader 0.003 0.03 −0.02 (0.13) (0.09) (0.10) Lag good list pos. 0.45*** 0.03 −0.12 (0.13) (0.14) (0.13) Lag of depend. var. 0.65*** 0.67*** 0.72*** (0.05) (0.06) (0.09) Lag intra-p. comp. 0.06** −0.003 0.002 (0.03) (0.01) (0.02) Replacement MP 0.59*** 0.41*** 0.08 (0.12) (0.13) (0.22) Sole representative 0.01 0.20 0.05 (0.19) (0.19) (0.17) PPG leadership −0.07 0.12 0.04 (0.12) (0.12) (0.10) Seniority: 1 term 0.06 0.03 0.18* (0.15) (0.12) (0.10) Seniority: ≥2 terms 0.14 0.02 0.15 (0.12) (0.10) (0.11) Opposition −0.18* −0.05 −0.35*** (0.10) (0.13) (0.13) Age 0.003 −0.04 −0.03 (0.03) (0.03) (0.03) Age squared/1000 −0.11 0.35 0.20 (0.32) (0.33) (0.34) Female 0.34*** −0.03 0.02 (0.13) (0.10) (0.10) Subnational politics 0.01 −0.06 0.13 (0.10) (0.06) (0.10) Constant −1.61** 0.63 0.30 (0.74) (0.75) (0.83) Adjusted R2 0.83 0.74 0.60 N 142 122 114 Predictor Dep. var.: Log of preference votes relative to reference category C7: 2006 C8: 2010 C9: 2013 Parl. activity Years 1–4 (party-std.) 0.15* 0.35*** 0.07 (0.09) (0.09) (0.09) Missed RCV Years 1–4 (party-std.) −0.02 0.02 −0.001 (0.06) (0.04) (0.05) List leader 0.14 −0.01 −0.03 (0.10) (0.09) (0.09) List rank (linear) −0.10*** −0.02** −0.03*** (0.02) (0.01) (0.01) Lag list leader 0.003 0.03 −0.02 (0.13) (0.09) (0.10) Lag good list pos. 0.45*** 0.03 −0.12 (0.13) (0.14) (0.13) Lag of depend. var. 0.65*** 0.67*** 0.72*** (0.05) (0.06) (0.09) Lag intra-p. comp. 0.06** −0.003 0.002 (0.03) (0.01) (0.02) Replacement MP 0.59*** 0.41*** 0.08 (0.12) (0.13) (0.22) Sole representative 0.01 0.20 0.05 (0.19) (0.19) (0.17) PPG leadership −0.07 0.12 0.04 (0.12) (0.12) (0.10) Seniority: 1 term 0.06 0.03 0.18* (0.15) (0.12) (0.10) Seniority: ≥2 terms 0.14 0.02 0.15 (0.12) (0.10) (0.11) Opposition −0.18* −0.05 −0.35*** (0.10) (0.13) (0.13) Age 0.003 −0.04 −0.03 (0.03) (0.03) (0.03) Age squared/1000 −0.11 0.35 0.20 (0.32) (0.33) (0.34) Female 0.34*** −0.03 0.02 (0.13) (0.10) (0.10) Subnational politics 0.01 −0.06 0.13 (0.10) (0.06) (0.10) Constant −1.61** 0.63 0.30 (0.74) (0.75) (0.83) Adjusted R2 0.83 0.74 0.60 N 142 122 114 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. Table 3 Results from binomial-logistic regression of preference vote share in the Czech Republic Predictor Dep. var.: Log of preference votes relative to reference category C7: 2006 C8: 2010 C9: 2013 Parl. activity Years 1–4 (party-std.) 0.15* 0.35*** 0.07 (0.09) (0.09) (0.09) Missed RCV Years 1–4 (party-std.) −0.02 0.02 −0.001 (0.06) (0.04) (0.05) List leader 0.14 −0.01 −0.03 (0.10) (0.09) (0.09) List rank (linear) −0.10*** −0.02** −0.03*** (0.02) (0.01) (0.01) Lag list leader 0.003 0.03 −0.02 (0.13) (0.09) (0.10) Lag good list pos. 0.45*** 0.03 −0.12 (0.13) (0.14) (0.13) Lag of depend. var. 0.65*** 0.67*** 0.72*** (0.05) (0.06) (0.09) Lag intra-p. comp. 0.06** −0.003 0.002 (0.03) (0.01) (0.02) Replacement MP 0.59*** 0.41*** 0.08 (0.12) (0.13) (0.22) Sole representative 0.01 0.20 0.05 (0.19) (0.19) (0.17) PPG leadership −0.07 0.12 0.04 (0.12) (0.12) (0.10) Seniority: 1 term 0.06 0.03 0.18* (0.15) (0.12) (0.10) Seniority: ≥2 terms 0.14 0.02 0.15 (0.12) (0.10) (0.11) Opposition −0.18* −0.05 −0.35*** (0.10) (0.13) (0.13) Age 0.003 −0.04 −0.03 (0.03) (0.03) (0.03) Age squared/1000 −0.11 0.35 0.20 (0.32) (0.33) (0.34) Female 0.34*** −0.03 0.02 (0.13) (0.10) (0.10) Subnational politics 0.01 −0.06 0.13 (0.10) (0.06) (0.10) Constant −1.61** 0.63 0.30 (0.74) (0.75) (0.83) Adjusted R2 0.83 0.74 0.60 N 142 122 114 Predictor Dep. var.: Log of preference votes relative to reference category C7: 2006 C8: 2010 C9: 2013 Parl. activity Years 1–4 (party-std.) 0.15* 0.35*** 0.07 (0.09) (0.09) (0.09) Missed RCV Years 1–4 (party-std.) −0.02 0.02 −0.001 (0.06) (0.04) (0.05) List leader 0.14 −0.01 −0.03 (0.10) (0.09) (0.09) List rank (linear) −0.10*** −0.02** −0.03*** (0.02) (0.01) (0.01) Lag list leader 0.003 0.03 −0.02 (0.13) (0.09) (0.10) Lag good list pos. 0.45*** 0.03 −0.12 (0.13) (0.14) (0.13) Lag of depend. var. 0.65*** 0.67*** 0.72*** (0.05) (0.06) (0.09) Lag intra-p. comp. 0.06** −0.003 0.002 (0.03) (0.01) (0.02) Replacement MP 0.59*** 0.41*** 0.08 (0.12) (0.13) (0.22) Sole representative 0.01 0.20 0.05 (0.19) (0.19) (0.17) PPG leadership −0.07 0.12 0.04 (0.12) (0.12) (0.10) Seniority: 1 term 0.06 0.03 0.18* (0.15) (0.12) (0.10) Seniority: ≥2 terms 0.14 0.02 0.15 (0.12) (0.10) (0.11) Opposition −0.18* −0.05 −0.35*** (0.10) (0.13) (0.13) Age 0.003 −0.04 −0.03 (0.03) (0.03) (0.03) Age squared/1000 −0.11 0.35 0.20 (0.32) (0.33) (0.34) Female 0.34*** −0.03 0.02 (0.13) (0.10) (0.10) Subnational politics 0.01 −0.06 0.13 (0.10) (0.06) (0.10) Constant −1.61** 0.63 0.30 (0.74) (0.75) (0.83) Adjusted R2 0.83 0.74 0.60 N 142 122 114 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. Swedish voters do not appear to reward parliamentary activity with personal votes (compare Table 4); the coefficients for this variable are close to zero in both years. A somewhat curious finding is that missing more roll-call votes (relative to the standards of the own party) is associated with a larger personal vote (statistically significant at 10% in 2014). Applying the usual interquartile procedure implies a change in expected difference in vote share (relative to the reference category) from approximately 0.028 to 0.029, so the substantive size of the association is small. The obvious explanation may be that MPs who campaign (for themselves) in the constituency may be in parliament less frequently (Fukumoto and Matsuo, 2015). The measure used in Table 4 refers to the full four-year term, but the association is practically unchanged if we replace the variable with a version that considers only the first three years of the term (see Table A3c in Online Appendix A3). Perhaps this finding points to a more general pattern of MPs who care about their personal vote giving more priority to local at the expense of parliamentary work. Another possibility is that the able politicians are rewarded with important functions within the party, devoting more time to party service than parliamentary work. If these individuals are also better at attracting personal votes, this will induce a positive bias on the effect of missing roll-call votes.14 Table 4 Results from binomial-logistic regression of preference vote share in Sweden Predictor Dep. var.: Log of preference votes relative to reference category S5: 2010 S6: 2014 Parl. activity Years 1–4 (party-std.) 0.03 0.02 (0.04) (0.03) Missed RCV Years 1–4 (party-std.) 0.03 0.05* (0.04) (0.03) List leader 0.65*** 0.43*** (0.12) (0.06) List rank (linear) −0.06*** −0.07*** (0.02) (0.02) Lag list leader −0.46*** −0.35*** (0.09) (0.07) Lag good list pos. −0.58*** 0.03 (0.20) (0.09) Lag of depend. var. 0.82*** 0.79*** (0.06) (0.05) Lag intra-p. comp. −0.00 0.01 (0.01) (0.005) Replacement MP −0.29 0.15 (0.27) (0.11) Sole representative 0.03 0.001 (0.14) (0.12) PPG leadership −0.05 0.03 (0.16) (0.11) Seniority: 1 term −0.05 −0.13** (0.06) (0.06) Seniority: ≥2 terms −0.03 −0.11 (0.09) (0.07) Opposition 0.17** −0.00 (0.08) (0.06) Age −0.01 −0.02 (0.03) (0.02) Age squared/1000 0.04 0.06 (0.37) (0.22) Female 0.08 −0.02 (0.06) (0.04) Constant 0.21 −0.06 (0.53) (0.42) Adjusted R2 0.90 0.94 N 253 243 Predictor Dep. var.: Log of preference votes relative to reference category S5: 2010 S6: 2014 Parl. activity Years 1–4 (party-std.) 0.03 0.02 (0.04) (0.03) Missed RCV Years 1–4 (party-std.) 0.03 0.05* (0.04) (0.03) List leader 0.65*** 0.43*** (0.12) (0.06) List rank (linear) −0.06*** −0.07*** (0.02) (0.02) Lag list leader −0.46*** −0.35*** (0.09) (0.07) Lag good list pos. −0.58*** 0.03 (0.20) (0.09) Lag of depend. var. 0.82*** 0.79*** (0.06) (0.05) Lag intra-p. comp. −0.00 0.01 (0.01) (0.005) Replacement MP −0.29 0.15 (0.27) (0.11) Sole representative 0.03 0.001 (0.14) (0.12) PPG leadership −0.05 0.03 (0.16) (0.11) Seniority: 1 term −0.05 −0.13** (0.06) (0.06) Seniority: ≥2 terms −0.03 −0.11 (0.09) (0.07) Opposition 0.17** −0.00 (0.08) (0.06) Age −0.01 −0.02 (0.03) (0.02) Age squared/1000 0.04 0.06 (0.37) (0.22) Female 0.08 −0.02 (0.06) (0.04) Constant 0.21 −0.06 (0.53) (0.42) Adjusted R2 0.90 0.94 N 253 243 Notes: Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. Table 4 Results from binomial-logistic regression of preference vote share in Sweden Predictor Dep. var.: Log of preference votes relative to reference category S5: 2010 S6: 2014 Parl. activity Years 1–4 (party-std.) 0.03 0.02 (0.04) (0.03) Missed RCV Years 1–4 (party-std.) 0.03 0.05* (0.04) (0.03) List leader 0.65*** 0.43*** (0.12) (0.06) List rank (linear) −0.06*** −0.07*** (0.02) (0.02) Lag list leader −0.46*** −0.35*** (0.09) (0.07) Lag good list pos. −0.58*** 0.03 (0.20) (0.09) Lag of depend. var. 0.82*** 0.79*** (0.06) (0.05) Lag intra-p. comp. −0.00 0.01 (0.01) (0.005) Replacement MP −0.29 0.15 (0.27) (0.11) Sole representative 0.03 0.001 (0.14) (0.12) PPG leadership −0.05 0.03 (0.16) (0.11) Seniority: 1 term −0.05 −0.13** (0.06) (0.06) Seniority: ≥2 terms −0.03 −0.11 (0.09) (0.07) Opposition 0.17** −0.00 (0.08) (0.06) Age −0.01 −0.02 (0.03) (0.02) Age squared/1000 0.04 0.06 (0.37) (0.22) Female 0.08 −0.02 (0.06) (0.04) Constant 0.21 −0.06 (0.53) (0.42) Adjusted R2 0.90 0.94 N 253 243 Predictor Dep. var.: Log of preference votes relative to reference category S5: 2010 S6: 2014 Parl. activity Years 1–4 (party-std.) 0.03 0.02 (0.04) (0.03) Missed RCV Years 1–4 (party-std.) 0.03 0.05* (0.04) (0.03) List leader 0.65*** 0.43*** (0.12) (0.06) List rank (linear) −0.06*** −0.07*** (0.02) (0.02) Lag list leader −0.46*** −0.35*** (0.09) (0.07) Lag good list pos. −0.58*** 0.03 (0.20) (0.09) Lag of depend. var. 0.82*** 0.79*** (0.06) (0.05) Lag intra-p. comp. −0.00 0.01 (0.01) (0.005) Replacement MP −0.29 0.15 (0.27) (0.11) Sole representative 0.03 0.001 (0.14) (0.12) PPG leadership −0.05 0.03 (0.16) (0.11) Seniority: 1 term −0.05 −0.13** (0.06) (0.06) Seniority: ≥2 terms −0.03 −0.11 (0.09) (0.07) Opposition 0.17** −0.00 (0.08) (0.06) Age −0.01 −0.02 (0.03) (0.02) Age squared/1000 0.04 0.06 (0.37) (0.22) Female 0.08 −0.02 (0.06) (0.04) Constant 0.21 −0.06 (0.53) (0.42) Adjusted R2 0.90 0.94 N 253 243 Notes: Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. 5. Conclusion We have analysed the link between parliamentary effort and MPs’ performance at the re-nomination and re-election stages. Starting from the notion that the threat of non-re-election must be credible in order to induce MPs to take their job seriously, we distinguished between MPs’ retrospective evaluation by party selectors and voters. It is important to examine if the two principals differ with regard to rewarding or sanctioning MPs for their efforts, since there is a general trend to give voters more influence on intra-party seat allocation (Renwick and Pilet, 2016). In both the Czech Republic and Sweden, their flexible-list systems were even more personalised. We expected that parliamentary work of MPs matters less for voters’ decisions, because of limited monitoring capacities and the lower salience they may attach to work in parliament as compared to other forms of representation. This would be a cause of concern, since the ex ante control of parties through candidate selection is weakened, while voters may not be in a position to compensate through ex post control (cp. Bergman and Strøm, 2013). Overall, the evidence that party selectors and voters take into account MPs’ parliamentary work effort is mixed. The findings suggest that at the selection stage—as long as we assume that there is more than a last period effect—party selectors do sort out underperformers or signal them not to seek re-nomination at all. In contrast, parliamentary effort does not seem to affect the quality of the list position among the group of re-selected MPs. These patterns may also be interpreted as evidence that accountability works in practice: if MPs anticipate sanctioning, then all those interested in re-election will work, and relative differences between them can hardly predict the quality of their list rank. In addition, there are of course many other factors that influence list ranking, such as various balancing considerations, regarding gender, intra-constituency regional dispersion etc. Regarding the citizens, the case of the 2010 Czech election points out that, under certain conditions, the individual work of MPs can indeed make a difference to their personal vote. The election followed a major corruption scandal, and calls to vote in new politicians accompanied the introduction of the new preference voting rules. There was even an NGO whose main aim consisted in promoting to give preferences to ‘four [candidates] from the bottom’. Therefore, what may be required is a context that draws citizens’ attention to politicians’ personal record, which is facilitated by third-party actors like media and NGOs (compare also Stegmaier et al. (2014) on the success of female candidates in that election). This provides ground for optimism, suggesting that voters do not always know or care little about MPs’ work. Whether the reform of preferential voting before the 2010 election was a necessary condition for improving accountability is a question that we cannot answer. The association we found between lower attendance and a better personal vote performance in Sweden, on the other hand, allows for less positive interpretations. Pessimists may suspect that MPs shirk while mobilising voters through other, less valuable means (say social media entertainment). On the other hand, if the result is due to MPs simply substituting constituency service for parliamentary work, MPs may merely respond to (perceived) citizens’ expectations. Acknowledgements We would like to thank several MPs for participating in background interviews, Ertan Bat for research assistance as well as Václav Sklenář and the Czech/Swedish election authorities for providing additional data. All errors are ours. Funding This work was supported by German Research Foundation [grant number DA1692/1-1, to TD]; Czech Science Foundation [grant number GA16-04885S, to LL]. Conflicts of Interest No conflicts of interest. Footnotes 1 Compare also Martin (2010), a study using survey-based measures of constituency orientation and effort. The literature on participatory shirking is mainly concerned with the question whether legislators not running for re-election show lower attendance rates. Bernecker (2014) is an exception that also tests if participation in parliamentary votes improves election results; the study finds an effect for constituency candidates in Germany. 2 This is actually the case in both countries. For the Swedish case, see Dagens Industri, ‘They are most absent in parliament’ (25 June 2014) accessed at <http://www.di.se/artiklar/2014/6/25/de-skolkar-mest-i-riksdagen/> on 22 November 2017; in the Czech case, NGO Kohovolit.eu publishes regular reports on MPs’ activity which are used by almost all major media outlets. 3 For parliamentary systems, there seems to be little data on citizens’ preferences regarding constituency service per se (rather than more general representational focus). Vivyan and Wagner (2016) find that Britons prefer MPs to strike a moderate balance between constituency service and work on national policy (see also Cain et al. (1987, pp. 36–43) for data from 1979). 4 Own calculations based on http://www.volby.cz/pls/ps2006/ps111?xjazyk=CZ&xkraj=0&xstrana=0&xv=2&xt=1 and accessed at <http://www.volby.cz/pls/ps2010/ps111?xjazyk=CZ&xkraj=0&xstrana=0&xv=2&xt=1> on 22 November 2017. The sum of preference votes for a list was divided by the total number of ballots for that list. This was done for each list that won at least one seat, and then averaged across lists. 5 Yet, studying the first preference vote reform, Davidsson (2006) showed that electorally vulnerable MPs deviate more often from the party line in their private members’ bills and initiate a larger share of proposals with local focus. 6 This is in line with the notion that personal votes are intra-party votes, and that few voters choose a party just because of an individual candidate. 7 We choose two indicators, since different activities may have the character of substitutes, while this is not the case for voting. The absolute value of Pearson’s r between the two indicators does not exceed 0.15 in any of the subsamples. 8 While written questions exist in the Czech Republic as well, no systematic data on these are available (personal communication with parliamentary staff). For oral interpellations, we count all submitted interpellations save the ones that were withdrawn or not held due to absence of the asking MP. The amendment variable refers to the number of bills an MP amended. 9 Interpellations and questions withdrawn at a later point in time are not considered. 10 The term of the Swedish parliament is organized around official yearly sessions; for the Czech Republic we split the terms into parts with approximate length of one year. The legislative period starting in 2010 turned out to end after a little more than three years because of early elections. As a start point for the fourth part of this legislative period we choose the day after the resignation of Prime Minister Nečas. 11 In the Czech case, the data allow to separate ‘excused’ from unexcused missed votes. We ran all models also with the share of missed unexcused votes; the results are very similar. 12 The substantive results do not change when using regular standard errors. Random effects models for the personal vote also give very similar results (those for re-selection outcomes did not converge). 13 Tables A3a and A3b in the Online Appendix A3 report results of models using party indicator variables to account for variation in preference vote usage across parties. The conclusions for the key variables of interest remain unchanged. 14 We also examined how the results change when leaving out one type of activity at a time for creating the summary indicator of effort. These findings are presented in Online Appendix A3, Tables A3d–g. References André A. , Depauw S. , Shugart M. S. ( 2014 ) ‘The Effect of Electoral Institutions on Legislative Behavio’. In Martin S. , Saalfeld T. , Strøm K. (eds) The Oxford Handbook of Legislative Studies , Oxford , Oxford University Press , pp. 231 – 249 . André A. , Depauw S. , Shugart M. S. , Chytilek R. ( 2017 ) ‘ Party Nomination Strategies in Flexible-List Systems: Do Preference Votes Matter? ’, Party Politics , 27 , 589 – 600 . Google Scholar Crossref Search ADS Ashworth S. ( 2012 ) ‘ Electoral Accountability: Recent Theoretical and Empirical Work ’, Annual Review of Political Science , 15 , 183 – 201 . Google Scholar Crossref Search ADS Aylott N. ( 2013 ) ‘Parties and Party Systems in the North’. In Bergman T. , Strøm K. (eds) The Madisonian Turn: Political Parties and Parliamentary Democracy in Nordic Europe , Ann Arbor , University of Michigan Press , pp. 297 – 328 . Berg L. , Oscarsson H. ( 2015 ) 20 år med personval , Gothenburg , University of Gothenburg . Bergman T. ( 2006 ) ‘Sweden: From Separation of Power to Parliamentary Supremacy - and Back Again?’. In Strøm K. , Müller W. C. , Bergman T. (eds) Delegation and Accountability in Parliamentary Democracies , Oxford , Oxford University Press , pp. 594 – 619 . Bergman T. , Bolin N. ( 2013 ) ‘Swedish Democracy: Crumbling Political Parties, a Feeble Riksdag, and Technocratic Power Holders’. In Bergman T. , Strøm K. (eds) The Madisonian Turn: Political Parties and Parliamentary Democracy in Nordic Europe , Ann Arbor , University of Michigan Press , pp. 251 – 293 . Bergman T. , Strøm K. ( 2013 ) The Madisonian Turn: Political Parties and Parliamentary Democracy in Nordic Europe , Ann Arbor , University of Michigan Press . Bernecker A. ( 2014 ) ‘ Do Politicians Shirk When Reelection Is Certain? Evidence from the German Parliament ’, European Journal of Political Economy , 36 , 55 – 70 . Google Scholar Crossref Search ADS Berry S. , Levinsohn J. , Pakes A. ( 1995 ) ‘ Automobile Prices in Market Equilibrium ’, Econometrica , 63 , 841 – 890 . Google Scholar Crossref Search ADS Bowler S. ( 2010 ) ‘ Private Members' Bills in the UK Parliament: Is There an ‘Electoral Connection’? ’, The Journal of Legislative Studies , 16 , 476 – 494 . Google Scholar Crossref Search ADS Cain B. , Ferejohn J. , Fiorina M. ( 1987 ) The Personal Vote , Harvard , Harvard University Press . Crisp B. , Jensen K. M. , Shomer Y. ( 2007 ) ‘ Magnitude and Vote-Seeking ’, Electoral Studies , 26 , 727 – 734 . Google Scholar Crossref Search ADS Crisp B. , Olivella S. , Malecki M. , Sher M. ( 2013 ) ‘ Vote-Earning Strategies in Flexible List Systems: Seats at the Price of Unity ’, Electoral Studies , 32 , 658 – 669 . Google Scholar Crossref Search ADS Däubler T. , Bräuninger T. , Brunner M. ( 2016 ) ‘ Is Personal Vote‐Seeking Behavior Effective? ’, Legislative Studies Quarterly , 41 , 419 – 444 . Google Scholar Crossref Search ADS Davidsson L. ( 2006 ) I linje med partiet? Maktspel och lojalitet i den svenska riksdagen , Stockholm , SNS . Esaiasson P. , Holmberg S. ( 1996 ) Representation from Above: Members of Parliament and Representative Democracy in Sweden , Aldershot , Dartmouth . Fearon J. D. ( 1999 ) ‘Electoral Accountability and the Control of Politicians: Selecting Good Types versus Sanctioning Poor Performance’. In Przeworski A. , Stokes S. C. , Manin B. (eds) Democracy, Accountability, and Representation , Cambridge , Cambridge University Press , pp. 55 – 97 . Folke O. , Persson T. , Rickne J. ( 2016 ) ‘ The Primary Effect: Preference Votes and Political Promotions ’, American Political Science Review , 111 , 559 – 578 . Google Scholar Crossref Search ADS Frech E. ( 2016 ) ‘ Re-Electing MEPs: The Factors Determining Re-Election Probabilities ’, European Union Politics , 17 , 69 – 90 . Google Scholar Crossref Search ADS Fukumoto K. , Matsuo A. ( 2015 ) ‘ The Effects of Election Proximity on Participatory Shirking: The Staggered-Term Chamber as a Laboratory ’, Legislative Studies Quarterly , 40 , 599 – 625 . Google Scholar Crossref Search ADS Gherghina S. , Chiru M. ( 2010 ) ‘ Practice and Payment: Determinants of Candidate List Position in European Parliament Elections ’, European Union Politics , 11 , 533 – 552 . Google Scholar Crossref Search ADS Hermansen S. S. L. ( 2016 ) ‘ Striving for Influence: The Effect of Performance on Candidate Selection ’, Party Politics , Advance Access published on 13 September 2016, DOI: 10.1177/1354068816663036. Kellermann M. ( 2013 ) ‘ Sponsoring Early Day Motions in the British House of Commons as a Response to Electoral Vulnerability ’, Political Science Research and Methods , 1 , 263 – 280 . Google Scholar Crossref Search ADS Kolář P. , Kysela J. , Syllová J. , Georgiev J. , Pecháček Š. ( 2013 ) Parlament České republiky , Praha , Leges . Larsson T. , Bäck H. ( 2008 ) Governing and Governance in Sweden , Lund , Studentlitteratur . Linek L. , Mansfeldová Z. ( 2007 ) ‘ The Parliament of the Czech Republic, 1993–2004 ’, The Journal of Legislative Studies , 13 , 12 – 37 . Google Scholar Crossref Search ADS Loewen P. J. , Koop R. , Settle J. , Fowler J. H. ( 2014 ) ‘ A Natural Experiment in Proposal Power and Electoral Success ’, American Journal of Political Science , 58 , 189 – 196 . Google Scholar Crossref Search ADS Mansfeldová Z. , Linek L. ( 2009 ) Český parlament ve druhé dekádě demokratického vývoje , Prague , Institute of Sociology . Marcinkiewicz K. , Stegmaier M. ( 2015 ) ‘ Ballot Position Effects under Compulsory and Optional Preferential-List PR Electoral Systems ’, Political Behavior , 37 , 465 – 486 . Google Scholar Crossref Search ADS Martin S. ( 2010 ) ‘ Electoral Rewards for Personal Vote Cultivation under PR-STV ’, West European Politics , 33 , 369 – 380 . Google Scholar Crossref Search ADS Navarro J. ( 2010 ) ‘ Le travail parlementaire, un investissement payant? Les élections comme évaluation rétrospective du bilan des députés sortants ’, Revue internationale de politique comparée , 17 , 141 – 160 . Google Scholar Crossref Search ADS Outlý J. , Prouza J. ( 2009 ) Navrhování a výběr kandidátů: politické strany v ČR a ve střední Evropě , Olomouc , Periplum . Pemstein D. , Meserve S. A. , Bernhard W. T. ( 2015 ) ‘ Brussels Bound: Policy Experience and Candidate Selection in European Elections ’, Comparative Political Studies , 48 , 1421 – 1453 . Google Scholar Crossref Search ADS Renwick A. , Pilet J. -B. ( 2016 ) Faces on the Ballot. The Personalization of Electoral Systems in Europe , Oxford , Oxford University Press . Stegmaier M. , Tosun J. , Vlachová K. ( 2014 ) ‘ Women’s Parliamentary Representation in the Czech Republic: Does Preference Voting Matter? ’, East European Politics & Societies , 28 , 187 – 204 . Google Scholar Crossref Search ADS Vivyan N. , Wagner M. ( 2016 ) ‘ House or Home? Constituent Preferences over Legislator Effort Allocation ’, European Journal of Political Research , 55 , 81 – 99 . Google Scholar Crossref Search ADS Wilson S. L. , Ringe N. , van Thomme J. ( 2016 ) ‘ Policy Leadership and Re-Election in the European Parliament ’, Journal of European Public Policy , 23 , 1158 – 1179 . Google Scholar Crossref Search ADS Zubek R. , Stecker C. ( 2010 ) ‘ Legislatures and Policy Uncertainty: Evidence from East Central Europe ’, Journal of Public Policy , 30 , 63 – 80 . Google Scholar Crossref Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Parliamentary Affairs Oxford University Press

Parliamentary Activity, Re-Selection and the Personal Vote. Evidence from Flexible-List Systems

Loading next page...
 
/lp/ou_press/parliamentary-activity-re-selection-and-the-personal-vote-evidence-qZpnassa1w
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
ISSN
0031-2290
eISSN
1460-2482
D.O.I.
10.1093/pa/gsx048
Publisher site
See Article on Publisher Site

Abstract

Abstract In this article, we analyse how the degree of parliamentary activity affects both individual MPs’ performance in the candidate selection process within the party and their popularity with voters at the electoral stage. We expect that parliamentary work of MPs matters less for voters’ evaluations of MPs because of limited monitoring capacities and lower salience attached to this type of representation. The empirical analysis uses data from recent elections in the Czech Republic and Sweden. During the analysed period, these countries further personalised their flexible list electoral systems. Our results suggest that parties hold MPs accountable mainly through the threat of non-re-selection rather than by assigning them to a promising list position. While there is no evidence that voters consistently reward MPs’ effort, the case of the Czech elections in 2010 shows that they may do so if context draws attention to individual MPs’ work. Members of the parliament are agents of at least two principals: the candidate selectorate within the party and the voters. To make delegation work and induce the agent to work hard on behalf of the principals, the threat of non-re-election must be credible (Fearon, 1999; Ashworth, 2012). How do the two principals consider the degree of parliamentary activity when deciding about the re-election of MPs? This is a particular pressing question with regard to the voter side, since it is not clear if citizens know or care enough about MPs’ parliamentary work to reward the hard workers and send the others packing. Whether voters are able to hold MPs to account for their record has important implications for electoral system design. Shifting influence on intra-party seat allocation from party selectors to party voters by ‘personalising’ electoral rules is a common trend at least in Europe (Renwick and Pilet, 2016). Empowering voters, however, may be detrimental to the latters’ interests, if negative effects on representatives’ behaviour eat up benefits from increased choice. This article analyses how parliamentary activity—understood as the overall effort spent on individual forms of parliamentary behaviour—affects both an MP’s performance in the candidate selection process and her popularity with voters at the electoral stage. We expect that parliamentary work of MPs matters less for voters’ evaluations of MPs because of limited monitoring capacities and lower salience attached to this type of representation. The empirical analysis uses data from recent elections in the Czech Republic (2006, 2010 and 2013) and Sweden (2010 and 2014). These countries share three features that are beneficial for the purpose of our analysis: first, they use flexible-list electoral systems, under which both the list position and the personal vote are important determinants of re-election. Secondly, since voters may also cast their ballot for a party list, the votes cast for a candidate are ‘true’ personal votes for that individual politician. Thirdly, during the analysed period, these two countries further ‘personalised’ their electoral systems, allowing study of short-run consequences of these reforms. Our results suggest that parties hold MPs accountable mainly through the threat of non-re-selection. While there is no consistent relationship between parliamentary activities and the personal vote, the case of the Czech elections in 2010 suggests that this link can materialise when context draws citizens’ attention towards MPs’ personal record. 1. Parliamentary activity and re-election seeking We are interested in examining how party selectors and voters take into account parliamentary effort of incumbent MPs when deciding about their re-election. Almost all of the existing literature studies the evaluation of incumbents by only one of these two principals, or considers overall re-election as the combined result of the two processes (e.g. Navarro, 2010). Particularly regarding candidate selection, systematic research about the characteristics that selectors value is scarce. Only few studies focus on the question how candidates’ features and incumbents’ work record affect the outcomes of specific nomination processes. Candidates seem to benefit from previous experience as MPs (Gherghina and Chiru, 2010, Pemstein et al., 2015), but it is not clear if hard work as such pays off for incumbents. For instance, bill sponsorship does not affect incumbents’ subsequent list ranking in Slovakian national elections (Crisp et al., 2013). German Members of the European Parliament (MEPs) perform better at the candidate selection stage if they were members of influential committees, but do not seem to be rewarded for high attendance rates or drafting many reports (Frech, 2016). Hermansen (2016), on the other hand, finds a positive association between legislative reports and obtaining a safe seat when analysing the electoral career of MEPs from 11 Member States. Regarding voters’ reactions to the parliamentary work of incumbent MPs, a number of studies conclude that legislative activity contributes to individual electoral performance of candidates also in parliamentary systems (Bowler, 2010, Crisp et al., 2013, Kellermann, 2013, Loewen et al., 2014, Däubler et al., 2016).1 From these findings, the ones for preferential list proportional represenation (PR) systems are remarkable, since it is well known that candidates ranked at the top of the list receive a very large share of the preference vote (e.g. Marcinkiewicz and Stegmaier, 2015). One should, however, keep in mind that those results refer to one specific type of activity—initiating legislative bills, and in some cases certain types of legislative bills—rather than a measure of overall effort across different types of activity. There are two main principals which hold MPs accountable for their degree of parliamentary activity: party selectors and voters. We assume that party selectors and voters take into account the past behaviour of MPs to form an expectation about how they will behave in the future. If MPs anticipate to receive a reward for parliamentary work (or being sanctioned for a lack thereof), the pending verdict of the principals will also induce accountability by anticipation. We suggest that the principals differ concerning their ability to hold MPs accountable for two main reasons. First, party selectors can be expected to have a higher level of monitoring capacity (cp. Fearon, 1999; André et al., 2014) than voters. Party selectors should follow parliamentary proceedings and the individual activities of their MPs more closely than voters do. They should both be more aware of the importance of being represented well (or have higher personal stakes), and thus be more willing to pay the cost of monitoring from an instrumental perspective. At the same time, as actors in one way or another involved in politics, party selectors may also simply have a stronger intrinsic motivation to follow how MPs do their job. It is, however, more difficult to answer the question if voters know enough to reward or sanction MPs for their exerted effort levels. In this context it is important to note that there can be two different mechanisms for how effort improves reputation (Däubler et al., 2016). First, it is possible that principals base their decision directly and consciously on merit. They observe the effort by the agent, and credit her for the engagement as such. This direct form of assessment may also be facilitated by third parties providing fire alarm-type information—with quantitative data on parliamentary work being more easily available than in the past; media and websites may, for instance, directly report on the amount of activity.2 Secondly, learning about agent effort may take indirect routes, since work can create visibility and name recognition (e.g. Cain et al., 1987; Wilson et al., 2016). In addition—and independent of their knowledge about what MPs do—voters may also differ from party actors by preferring MPs who put less relative emphasis on parliamentary work as compared to other forms of representation (such as constituency work in a narrow sense). What MPs contribute in parliament to legislation or government oversight may not have an immediately visible relevance or benefit for people’s lives. This may lead citizens to prefer constituency service, which is more tangible in nature.3 In addition, voters get to ‘evaluate’ MPs only after the party has ranked the candidates on the list, which may further limit the scope for a positive association between parliamentary work and the personal vote. Our central expectation is, therefore, that the overall degree of parliamentary work has a stronger effect on the evaluation of MPs by party selectors than it has for that by voters. Note, however, that we refrain from making a prediction about the absolute level of the association we expect to find for voters. We empirically analyse the actual outcomes of both the nomination and electoral stage, using data from recent elections in the Czech Republic and Sweden. The flexible-list systems employed make the Czech Republic and Sweden particularly suitable cases for testing the arguments. Parties present voters with pre-ordered lists, and the order of candidates will be consequential for intra-party seat allocation unless all candidates clear the preference vote threshold. Since preference voting is not obligatory, the votes to candidates are ‘true’ personal votes, rather than party votes in disguise. In addition, the two countries under study reduced the preference vote threshold of their flexible-list system in a recent reform. This change makes it easier for candidates to be elected on the basis of personal votes rather than due to a good pre-electoral list position. So the incentives for incumbent MPs may change, and this again may also alter how party selectors and voters hold representatives to account. With stronger incentives to cater to voters directly, MPs may invest more resources in communicating with voters, and media may report more intensively on individual behaviour. Then, the reform may also make it easier for voters to hold individual representatives accountable for their activities. For each country, we study the last pre-reform election and all post-reform elections that had been passed at the time of writing. 2. Institutional context 2.1. Czech Republic The Czech Chamber of Deputies (Poslanecká sněmovna) consists of 200 MPs who are elected based on a flexible-list PR system. There are 14 constituencies mirroring the administrative division of the country, ranging in magnitude from 5 to 25 seats. Parties with at least 5% of the vote at the national level gain seats, which are distributed among parties based on their support within constituencies. Voters have the opportunity to cast preference votes for multiple candidates within one party list. The number of preference votes increased from two to four since the 2010 election. Candidates who have reached the preference vote threshold are moved to the top of the list and are ranked according to the number of personal votes received; for all other candidates pre-electoral list position applies. The threshold was lowered from 7% to 5% since the 2010 election (for more details about the preference vote rules, see Marcinkiewicz and Stegmaier (2015)). Together with the increase in the number of preference votes, this led to a considerable increase in preference voting, with the number of preference votes per ballot rising from 0.38 to 0.68.4 Candidate selection takes place at the local and regional level in most parties, and the influence of the national-level party organisations is generally limited. The process is remarkably similar in all major parties (ČSSD, KSČM, KDU-ČSL and ODS) covered in the present analysis (Outlý and Prouza, 2009). In general, party selectorates reward candidates with higher preferential votes in previous elections and put them on better list positions (André et al., 2017). In the Czech Republic, individual MPs enjoy extensive rights in the legislative process. The government has only weak control over the agenda which is decided mainly by majority vote in the Chamber (Zubek and Stecker, 2010). Moreover, not only the government, but also an individual MP or a group of them are allowed to introduce bills. While the number of private members’ bills has declined since the beginning of the post-communist transition period (Linek and Mansfeldová, 2007), their success rate is considerable. Most of the bills submitted by individual MPs (rather than a group), however, seek to put an issue onto the agenda rather than aiming at actually adjusting the law (Kolář et al., 2013, p. 238). In addition to that, an individual MP is allowed to propose amendments to the bills. Almost all decisions of the Chamber are done by registered electronic roll-call votes. During the four-year term, MPs usually take more than 10,000 votes and on average 50 votes per meeting day (Linek and Mansfeldová, 2007, pp. 30–32). Since 1996, there is an informal agreement among government and opposition parties to ‘pair’ an absenting MP so that the balance between government and opposition is not changed. This artificially leads to increase in abstentions, especially for the opposition MPs who are pairs for MPs who are also cabinet members. The Chamber uses oral and written questions, which are called interpellations. Any MP has the right to engage in both forms of interpellations. Oral interpellations take place once a week. They need to be submitted beforehand, their order on the agenda is drawn in the morning of the day they are covered. Written interpellations can be answered verbally or in writing. If an MP is not satisfied with an answer, he can ask the President of the Chamber to include the question on the agenda. MPs have the possibility to ask the government and ministers about any topic which is covered by their field of action (Kolář et al., 2013, p. 320). In interpellations, MPs are quite independent from the party leadership. In MP surveys, however, representatives state that interpellations are blunt instruments for influencing the government. Private members’ bills and amendments receive better assessments by MPs in this regard (Mansfeldová and Linek, 2009, pp. 88–93). 2.2. Sweden The Swedish national parliament, the Riksdag, consists of 349 seats, of which 310 are allocated to the 29 electoral constituencies, and 39 are levelling (second-tier compensation) seats. The Swedish preferential-list PR system is a flexible-list system, allowing voters to choose between casting a list vote and expressing a preference for a single candidate on a party list. If a candidate gets enough votes to pass the personal vote threshold, this candidate will receive the first seat allocated to the party. The threshold was lowered from 8% to 5% in the 2014 elections. Politicians widely agreed that the change was unlikely to have dramatic effects on various aspects of intra-party and MP–voter relationships (Berg and Oscarsson, 2015).5 If more candidates cross the threshold, the candidates are ranked according to number of personal votes they receive. Any remaining seats are distributed according to the order on the party list. The Swedish parliament is characterised by strong party discipline (Larsson and Bäck, 2008, p. 160). MPs, however, have several tools to pursue more individual representation. Individual MPs have the right to initiate private members’ bills (motioner). The so-called stand-alone bills, that is, proposals that do not seek to amend other proposals, may be initiated during a certain time period each year (from the start of the session until shortly after submission of the budget). Swedish MPs also have the right to ask questions to ministers and other government members. There is a weekly question time with spontaneous oral questions, interpellations (submitted in writing but answered by government members personally) and written questions that receive written answers. Roll-call votes occur frequently, and the threshold for a roll call to take place is low. Public attention to MPs’ activities is generally limited (Bergman, 2006, p. 607), and the activities seldom make the news (Larsson and Bäck, 2008). However, MPs find questions useful for creating media attention in local constituencies (Bergman, 2006, p. 604), and their usage has increased over time (Bergman and Bolin, 2013, pp. 269–270). Individual motions very seldom receive media coverage, since these are almost never passed. Candidate selection in Swedish parties usually takes place at the regional level (Aylott, 2013, p. 322). There is some disagreement, though, about the importance of local representation in the Riksdag. Esaiasson and Holmberg (1996) concluded from their analyses that local interest promotion is strong, but Bergman and Bolin (2013, p. 261) argue that, once elected, MPs perceive themselves rather as party/national than constituency representatives. Folke et al. (2016) also show that, on the local level, party selectors appear to use the personal vote to identify popular politicians and to promote them to positions of power within the party. Although only a minority of voters use the personal vote (23% in 2010 and 25% in 2014), its usage is distributed evenly across social groups (e.g. income, education, geography) and other voter characteristics (Berg and Oscarsson, 2015). According to data from the 2014 Swedish National Election Study, the most common motive for rewarding a candidate with a vote was individual candidate traits such as competence and reliability. Amongst the voters who did not use the personal vote, the most common reason was lacking information about candidates. 3. Data, measures and models To examine the arguments outlined above, we consider the link between parliamentary activities of incumbent MPs and their performance at the subsequent party nomination stage and their personal vote in the following elections. We analyse the consequences of parliamentary work undertaken during one pre-electoral reform and all post-reform legislative periods in the Czech Republic (2002–2006, 2006–2010 and 2010–2013) and Sweden (2006–2010 and 2010–2014). The two parliaments provide databases that contain information about activities and the parliamentary biography of politicians. We combined parliamentary data with information obtained from the election authorities, which provide details about the ballots/party lists used and the electoral performance of parties and individual politicians. For each legislative period, we analyse three dependent variables: whether MPs re-run for the same party, whether they obtain a promising position on the list (defined as being within the first N ranks, where N is the number of seats the party won in the previous election), and the intra-party vote share (for details on its transformation see below).6 The first type of analyses uses the sample of MPs who are in office at the end of the third year of the legislative term (this should approximately reflect the time point when selection processes start), and who have held their mandate for at least 200 days. We exclude any MPs who have been party leaders or government members during the term. The analyses of list positions and the personal vote covers those MPs who run for the same party in the same constituency as in the previous election, since we include a lagged dependent variable. As measures of the degree of parliamentary work, we consider two indicators.7 The first one is a summary measure of activities that are individual in a sense that they can be assumed to be fairly independent of party influence (and for which data are available). In the Czech Republic, we consider single-authored private members’ bills, oral interpellations and amendments submitted during the second reading of a bill.8 In the Swedish case, we look at single-authored private members’ bills (i.e. the stand-alone proposals from the respective yearly period), the number of oral interpellations and the number of written parliamentary questions.9 To create our summary measure, we divide each MP’s activity count by the duration of her mandate in that year, standardise within each activity–party-year, and take the mean over all the values (activities times years) for each MP (see Online Appendix A1 for more details and examples).10 Our second indicator is the share of missed roll-call votes among all votes during an MP’s mandate, standardised within parties.11 The measures, thus, capture how productive MPs are, but not the quality of their work. This approach has limitations, but what constitutes ‘good’ parliamentary work is hard to define and even harder to measure. When examining the effect of re-election-seeking activities on electoral performance, it is important to control for factors reflecting marginality, that is, the need to be active in the first place. We use lagged measures for this purpose (for indicators for list leader and for previous promising position, the logit-transformed personal vote and the ratio of the number of candidates and party seats as a measure of intra-party competition, see Crisp et al. (2007)). In addition, we also include variables for being a replacement MP, party group leadership (share of term serving as group or vice group leader), previous parliamentary experience (none, one term, two or more terms; counting spells with a minimum duration of 100 days), being the sole representative of a party-in-a-constituency, the share of term the party was in opposition, age and age squared and gender. In the Czech case, we also have data on involvement in subnational elections (obtaining a mandate at the municipality level or running in regional elections during the term as an MP). When analysing the personal vote, an indicator variable for the first list position and a continuous variable for list rank also enter the models, since the outcome of the nomination stage is of course an important predictor of the personal vote. Descriptive statistics are provided in Online Appendix A2. We use standard logit models for analysing the re-selection outcomes. For the personal vote models, we log-transform intra-party vote shares, from which we subtract the respective value of a reference category (ballots cast for the list and for candidates not in the sample), and then use ordinary least squares estimation. This model can be derived from a utility-maximisation framework, making certain simplifying assumptions (cp. Berry et al., 1995). To take into account unobserved heterogeneity at the party-list level, we use clustered standard errors.12 We note that the aim of our analysis is to examine whether parliamentary activity predicts (rather than causes) performance at the re-selection and re-election stages, and we acknowledge that the comparison between selectors and voters is merely an indirect test of their monitoring capacities. 4. Results Table 1 displays the results of the regression models for the re-selection stage, split by legislative term, for the Czech Republic. Models C1–C3 consider the binary indicator of running for the same party in the subsequent election as dependent variable, Models C4–C6 explain whether an MP received a promising list position. Shown are logit coefficients along with standard errors in parentheses. Table 1 Results from logistic regression of reselection outcomes in the Czech Republic Predictor Dep. var.: Re-running for same party Dep. var.: Good list position C1: 2006 C2: 2010 C3: 2013 C4: 2006 C5: 2010 C6: 2013 Parl. activity Years 1–3 (party-std.) 0.65 0.05 1.43** 0.41 −0.67 −0.44 (0.57) (0.62) (0.59) (0.69) (0.73) (0.61) Missed RCV Years 1–3 (party-std.) −0.61*** 0.34 −0.46** 0.10 0.30 0.10 (0.23) (0.25) (0.24) (0.30) (0.40) (0.33) Lag list leader 0.06 1.69 1.30 −0.01 1.23 −0.34 (0.70) (1.08) (0.88) (0.75) (0.77) (0.88) Lag good list pos. 1.20 −1.07 −1.47* −1.15 0.02 1.93** (2.06) (0.74) (0.76) (2.05) (1.30) (0.77) Lag logit pers. vote 0.51 0.47 0.75 0.73*** 0.47 1.42*** (0.35) (0.34) (0.58) (0.27) (0.32) (0.51) Lag intra-p. comp. −0.25** 0.04 −0.18* −0.07 −0.13* −0.03 (0.10) (0.10) (0.10) (0.17) (0.07) (0.09) Replacement MP 1.96 −2.30** −0.89 −0.34 1.69 (2.16) (1.04) (1.19) (2.07) (1.60) Sole representative −0.54 −3.74* 0.23 0.73 0.33 (1.73) (2.12) (1.12) (1.59) (1.29) PPG leadership −0.13 −0.60 0.87 1.41 −0.35 0.49 (0.74) (0.89) (0.71) (1.08) (0.92) (0.69) Seniority: 1 term exp. −1.14** 0.53 −1.17* 0.30 0.11 0.66 (0.57) (0.67) (0.60) (0.87) (0.66) (0.89) Seniority: ≥2 terms exp. −0.59 −1.57*** −0.58 −0.95 −1.12 −0.99 (0.50) (0.58) (0.71) (0.74) (0.80) (0.87) Opposition 1.49*** 0.32 2.82*** 0.64 0.66 0.69 (0.52) (0.49) (0.68) (0.48) (0.62) (0.67) Age 0.03 −0.15 0.26* −0.51 −0.25 0.14 (0.16) (0.22) (0.13) (0.50) (0.21) (0.18) Age squared/1000 −0.40 1.09 −2.88** 4.03 2.18 −1.89 (1.73) (2.21) (1.42) (4.69) (2.38) (2.01) Female 0.42 −0.68 −0.10 −0.62 −1.18 −0.09 (0.64) (0.45) (0.63) (0.60) (0.72) (0.63) Subnational politics 0.56 0.65 −0.85* 0.07 0.20 0.14 (0.48) (0.40) (0.48) (0.50) (0.41) (0.47) Constant 2.79 8.50 0.09 20.41 10.66** 1.65 (4.37) (5.29) (3.49) (12.71) (4.83) (4.95) N 182 171 161 141 120 114 Predictor Dep. var.: Re-running for same party Dep. var.: Good list position C1: 2006 C2: 2010 C3: 2013 C4: 2006 C5: 2010 C6: 2013 Parl. activity Years 1–3 (party-std.) 0.65 0.05 1.43** 0.41 −0.67 −0.44 (0.57) (0.62) (0.59) (0.69) (0.73) (0.61) Missed RCV Years 1–3 (party-std.) −0.61*** 0.34 −0.46** 0.10 0.30 0.10 (0.23) (0.25) (0.24) (0.30) (0.40) (0.33) Lag list leader 0.06 1.69 1.30 −0.01 1.23 −0.34 (0.70) (1.08) (0.88) (0.75) (0.77) (0.88) Lag good list pos. 1.20 −1.07 −1.47* −1.15 0.02 1.93** (2.06) (0.74) (0.76) (2.05) (1.30) (0.77) Lag logit pers. vote 0.51 0.47 0.75 0.73*** 0.47 1.42*** (0.35) (0.34) (0.58) (0.27) (0.32) (0.51) Lag intra-p. comp. −0.25** 0.04 −0.18* −0.07 −0.13* −0.03 (0.10) (0.10) (0.10) (0.17) (0.07) (0.09) Replacement MP 1.96 −2.30** −0.89 −0.34 1.69 (2.16) (1.04) (1.19) (2.07) (1.60) Sole representative −0.54 −3.74* 0.23 0.73 0.33 (1.73) (2.12) (1.12) (1.59) (1.29) PPG leadership −0.13 −0.60 0.87 1.41 −0.35 0.49 (0.74) (0.89) (0.71) (1.08) (0.92) (0.69) Seniority: 1 term exp. −1.14** 0.53 −1.17* 0.30 0.11 0.66 (0.57) (0.67) (0.60) (0.87) (0.66) (0.89) Seniority: ≥2 terms exp. −0.59 −1.57*** −0.58 −0.95 −1.12 −0.99 (0.50) (0.58) (0.71) (0.74) (0.80) (0.87) Opposition 1.49*** 0.32 2.82*** 0.64 0.66 0.69 (0.52) (0.49) (0.68) (0.48) (0.62) (0.67) Age 0.03 −0.15 0.26* −0.51 −0.25 0.14 (0.16) (0.22) (0.13) (0.50) (0.21) (0.18) Age squared/1000 −0.40 1.09 −2.88** 4.03 2.18 −1.89 (1.73) (2.21) (1.42) (4.69) (2.38) (2.01) Female 0.42 −0.68 −0.10 −0.62 −1.18 −0.09 (0.64) (0.45) (0.63) (0.60) (0.72) (0.63) Subnational politics 0.56 0.65 −0.85* 0.07 0.20 0.14 (0.48) (0.40) (0.48) (0.50) (0.41) (0.47) Constant 2.79 8.50 0.09 20.41 10.66** 1.65 (4.37) (5.29) (3.49) (12.71) (4.83) (4.95) N 182 171 161 141 120 114 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. In 2006, being the sole representative predicts obtaining a good list position perfectly, N = 2 cases were dropped. In 2010, being a replacement MP predicts obtaining a good list position perfectly, N = 3 cases were dropped. *p < 0.1; **p < 0.05; ***p < 0.01. Table 1 Results from logistic regression of reselection outcomes in the Czech Republic Predictor Dep. var.: Re-running for same party Dep. var.: Good list position C1: 2006 C2: 2010 C3: 2013 C4: 2006 C5: 2010 C6: 2013 Parl. activity Years 1–3 (party-std.) 0.65 0.05 1.43** 0.41 −0.67 −0.44 (0.57) (0.62) (0.59) (0.69) (0.73) (0.61) Missed RCV Years 1–3 (party-std.) −0.61*** 0.34 −0.46** 0.10 0.30 0.10 (0.23) (0.25) (0.24) (0.30) (0.40) (0.33) Lag list leader 0.06 1.69 1.30 −0.01 1.23 −0.34 (0.70) (1.08) (0.88) (0.75) (0.77) (0.88) Lag good list pos. 1.20 −1.07 −1.47* −1.15 0.02 1.93** (2.06) (0.74) (0.76) (2.05) (1.30) (0.77) Lag logit pers. vote 0.51 0.47 0.75 0.73*** 0.47 1.42*** (0.35) (0.34) (0.58) (0.27) (0.32) (0.51) Lag intra-p. comp. −0.25** 0.04 −0.18* −0.07 −0.13* −0.03 (0.10) (0.10) (0.10) (0.17) (0.07) (0.09) Replacement MP 1.96 −2.30** −0.89 −0.34 1.69 (2.16) (1.04) (1.19) (2.07) (1.60) Sole representative −0.54 −3.74* 0.23 0.73 0.33 (1.73) (2.12) (1.12) (1.59) (1.29) PPG leadership −0.13 −0.60 0.87 1.41 −0.35 0.49 (0.74) (0.89) (0.71) (1.08) (0.92) (0.69) Seniority: 1 term exp. −1.14** 0.53 −1.17* 0.30 0.11 0.66 (0.57) (0.67) (0.60) (0.87) (0.66) (0.89) Seniority: ≥2 terms exp. −0.59 −1.57*** −0.58 −0.95 −1.12 −0.99 (0.50) (0.58) (0.71) (0.74) (0.80) (0.87) Opposition 1.49*** 0.32 2.82*** 0.64 0.66 0.69 (0.52) (0.49) (0.68) (0.48) (0.62) (0.67) Age 0.03 −0.15 0.26* −0.51 −0.25 0.14 (0.16) (0.22) (0.13) (0.50) (0.21) (0.18) Age squared/1000 −0.40 1.09 −2.88** 4.03 2.18 −1.89 (1.73) (2.21) (1.42) (4.69) (2.38) (2.01) Female 0.42 −0.68 −0.10 −0.62 −1.18 −0.09 (0.64) (0.45) (0.63) (0.60) (0.72) (0.63) Subnational politics 0.56 0.65 −0.85* 0.07 0.20 0.14 (0.48) (0.40) (0.48) (0.50) (0.41) (0.47) Constant 2.79 8.50 0.09 20.41 10.66** 1.65 (4.37) (5.29) (3.49) (12.71) (4.83) (4.95) N 182 171 161 141 120 114 Predictor Dep. var.: Re-running for same party Dep. var.: Good list position C1: 2006 C2: 2010 C3: 2013 C4: 2006 C5: 2010 C6: 2013 Parl. activity Years 1–3 (party-std.) 0.65 0.05 1.43** 0.41 −0.67 −0.44 (0.57) (0.62) (0.59) (0.69) (0.73) (0.61) Missed RCV Years 1–3 (party-std.) −0.61*** 0.34 −0.46** 0.10 0.30 0.10 (0.23) (0.25) (0.24) (0.30) (0.40) (0.33) Lag list leader 0.06 1.69 1.30 −0.01 1.23 −0.34 (0.70) (1.08) (0.88) (0.75) (0.77) (0.88) Lag good list pos. 1.20 −1.07 −1.47* −1.15 0.02 1.93** (2.06) (0.74) (0.76) (2.05) (1.30) (0.77) Lag logit pers. vote 0.51 0.47 0.75 0.73*** 0.47 1.42*** (0.35) (0.34) (0.58) (0.27) (0.32) (0.51) Lag intra-p. comp. −0.25** 0.04 −0.18* −0.07 −0.13* −0.03 (0.10) (0.10) (0.10) (0.17) (0.07) (0.09) Replacement MP 1.96 −2.30** −0.89 −0.34 1.69 (2.16) (1.04) (1.19) (2.07) (1.60) Sole representative −0.54 −3.74* 0.23 0.73 0.33 (1.73) (2.12) (1.12) (1.59) (1.29) PPG leadership −0.13 −0.60 0.87 1.41 −0.35 0.49 (0.74) (0.89) (0.71) (1.08) (0.92) (0.69) Seniority: 1 term exp. −1.14** 0.53 −1.17* 0.30 0.11 0.66 (0.57) (0.67) (0.60) (0.87) (0.66) (0.89) Seniority: ≥2 terms exp. −0.59 −1.57*** −0.58 −0.95 −1.12 −0.99 (0.50) (0.58) (0.71) (0.74) (0.80) (0.87) Opposition 1.49*** 0.32 2.82*** 0.64 0.66 0.69 (0.52) (0.49) (0.68) (0.48) (0.62) (0.67) Age 0.03 −0.15 0.26* −0.51 −0.25 0.14 (0.16) (0.22) (0.13) (0.50) (0.21) (0.18) Age squared/1000 −0.40 1.09 −2.88** 4.03 2.18 −1.89 (1.73) (2.21) (1.42) (4.69) (2.38) (2.01) Female 0.42 −0.68 −0.10 −0.62 −1.18 −0.09 (0.64) (0.45) (0.63) (0.60) (0.72) (0.63) Subnational politics 0.56 0.65 −0.85* 0.07 0.20 0.14 (0.48) (0.40) (0.48) (0.50) (0.41) (0.47) Constant 2.79 8.50 0.09 20.41 10.66** 1.65 (4.37) (5.29) (3.49) (12.71) (4.83) (4.95) N 182 171 161 141 120 114 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. In 2006, being the sole representative predicts obtaining a good list position perfectly, N = 2 cases were dropped. In 2010, being a replacement MP predicts obtaining a good list position perfectly, N = 3 cases were dropped. *p < 0.1; **p < 0.05; ***p < 0.01. Effort and the relative frequency of missing roll-call votes show the expected associations with running again in some, but not all of the elections. The coefficient of parliamentary activity is of considerable size both in 2006 and 2013 (but statistically significant only in the latter case). Those MPs missing more roll-call votes were less likely to stand again in 2006 and 2013. The 2010 election, and we will see this repeatedly, shows a different pattern. The coefficient for the activity variable is much smaller, and the one for the roll-call vote measure even has the opposite sign. The relationships for 2006 and 2013 are of considerable size. If we take 0.75 as a baseline probability (approximately the mean of the 2006 and 2013 proportion of re-running MPs), a change of the size of the interquartile range of the respective variable implies expected probabilities of re-running of 0.84 (for activity in 2013), of 0.57 (missed roll calls in 2006) and of 0.62 (missed roll calls in 2013). These are substantive changes—the question about the direction of causality is justified, though. The findings do not necessarily provide evidence for party selectors sorting out the not so hard-working MPs. The pattern would also be consistent with those MPs voluntarily withdrawing from parliament exerting less effort. We cannot disentangle these processes in our setup, but we control for age and seniority, which may alleviate the reversed causality problem to some extent. Space constraints do generally not permit us to discuss findings for control variables, we therefore only occasionally point to particularly interesting findings. The fact that in 2006 and 2013 intra-party competition for seats is linked to a lower chance of running again also points into the direction that we are not only observing a last-period effect for the parliamentary effort-related variables. There is no evidence that the amount of parliamentary work or participation in votes helps re-running MPs to obtain a promising list position. None of the key variables reaches statistical significance in these models, and the signs are inconsistent (activity) or contrary to the expectation (roll calls). Overall, none of the explanatory variables shows consistent results across years in these models. Even the coefficient of the lagged personal vote is estimated with high uncertainty in 2010. Table 2 reports results of the parallel analysis conducted for the Swedish elections in 2010 and 2014. Also here, the more active MPs are more likely to run again (statistically significant at 10% in 2010 and at 5% in 2014). Taking the sample mean of 0.77 as a baseline, the expected probability changes to 0.81 (2014) when considering a difference in activity of the size of the interquartile range. The roll-call vote variables have a negative sign as expected, but are far from significant. The decision to run again strongly depends on seniority and age, with the latter showing an inverse U-shape pattern, with the maximum at approximately 39 and 42 years, respectively. Table 2 Results from logistic regression of re-selection outcomes in Sweden Predictor Dep. var.: Re-running for same party Dep. var.: Good list position S1: 2010 S2: 2014 S3: 2010 S4: 2014 Parl. activity Years 1–3 (party-std.) 0.59* 0.62** 0.21 −0.45 (0.36) (0.28) (0.33) (0.38) Missed RCV Years 1–3 (party-std.) −0.02 −0.17 −0.24 0.32 (0.11) (0.15) (0.35) (0.30) Lag list leader −0.14 0.34 1.30 0.43 (0.48) (0.40) (0.87) (0.96) Lag good list pos. 0.72 −2.59 0.26 3.68*** (2.05) (1.97) (1.13) (0.98) Lag logit pers. vote 0.14 0.02 0.09 1.04*** (0.19) (0.17) (0.22) (0.26) Lag intra-p. comp. 0.00 −0.01 −0.08* −0.07 (0.04) (0.03) (0.05) (0.05) Replacement MP 0.23 −2.38 −1.68 0.52 (2.25) (2.03) (1.15) (1.10) Sole representative −0.13 −0.64 0.77 −0.85 (0.80) (0.73) (1.26) (1.28) PPG leadership 0.52 1.53 0.17 2.03 (0.98) (1.34) (1.51) (1.53) Seniority: 1 term exp. −1.16** −1.02** 0.81 −0.85 (0.49) (0.51) (0.60) (0.65) Seniority: ≥2 terms exp. −2.42*** −1.78*** 0.58 −2.37*** (0.45) (0.53) (0.65) (0.82) Opposition −0.15 −0.14 0.51 0.41 (0.36) (0.32) (0.53) (0.68) Age 0.30 0.32** 0.15 0.08 (0.18) (0.12) (0.15) (0.19) Age squared/1000 −3.82* −3.78*** −2.44 −1.16 (1.96) (1.36) (1.58) (2.06) Female 0.06 −0.29 0.09 0.32 (0.29) (0.27) (0.36) (0.54) Constant −2.25 −0.96 0.47 4.03 (4.70) (2.70) (3.58) (4.66) N 331 316 253 243 Predictor Dep. var.: Re-running for same party Dep. var.: Good list position S1: 2010 S2: 2014 S3: 2010 S4: 2014 Parl. activity Years 1–3 (party-std.) 0.59* 0.62** 0.21 −0.45 (0.36) (0.28) (0.33) (0.38) Missed RCV Years 1–3 (party-std.) −0.02 −0.17 −0.24 0.32 (0.11) (0.15) (0.35) (0.30) Lag list leader −0.14 0.34 1.30 0.43 (0.48) (0.40) (0.87) (0.96) Lag good list pos. 0.72 −2.59 0.26 3.68*** (2.05) (1.97) (1.13) (0.98) Lag logit pers. vote 0.14 0.02 0.09 1.04*** (0.19) (0.17) (0.22) (0.26) Lag intra-p. comp. 0.00 −0.01 −0.08* −0.07 (0.04) (0.03) (0.05) (0.05) Replacement MP 0.23 −2.38 −1.68 0.52 (2.25) (2.03) (1.15) (1.10) Sole representative −0.13 −0.64 0.77 −0.85 (0.80) (0.73) (1.26) (1.28) PPG leadership 0.52 1.53 0.17 2.03 (0.98) (1.34) (1.51) (1.53) Seniority: 1 term exp. −1.16** −1.02** 0.81 −0.85 (0.49) (0.51) (0.60) (0.65) Seniority: ≥2 terms exp. −2.42*** −1.78*** 0.58 −2.37*** (0.45) (0.53) (0.65) (0.82) Opposition −0.15 −0.14 0.51 0.41 (0.36) (0.32) (0.53) (0.68) Age 0.30 0.32** 0.15 0.08 (0.18) (0.12) (0.15) (0.19) Age squared/1000 −3.82* −3.78*** −2.44 −1.16 (1.96) (1.36) (1.58) (2.06) Female 0.06 −0.29 0.09 0.32 (0.29) (0.27) (0.36) (0.54) Constant −2.25 −0.96 0.47 4.03 (4.70) (2.70) (3.58) (4.66) N 331 316 253 243 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. Table 2 Results from logistic regression of re-selection outcomes in Sweden Predictor Dep. var.: Re-running for same party Dep. var.: Good list position S1: 2010 S2: 2014 S3: 2010 S4: 2014 Parl. activity Years 1–3 (party-std.) 0.59* 0.62** 0.21 −0.45 (0.36) (0.28) (0.33) (0.38) Missed RCV Years 1–3 (party-std.) −0.02 −0.17 −0.24 0.32 (0.11) (0.15) (0.35) (0.30) Lag list leader −0.14 0.34 1.30 0.43 (0.48) (0.40) (0.87) (0.96) Lag good list pos. 0.72 −2.59 0.26 3.68*** (2.05) (1.97) (1.13) (0.98) Lag logit pers. vote 0.14 0.02 0.09 1.04*** (0.19) (0.17) (0.22) (0.26) Lag intra-p. comp. 0.00 −0.01 −0.08* −0.07 (0.04) (0.03) (0.05) (0.05) Replacement MP 0.23 −2.38 −1.68 0.52 (2.25) (2.03) (1.15) (1.10) Sole representative −0.13 −0.64 0.77 −0.85 (0.80) (0.73) (1.26) (1.28) PPG leadership 0.52 1.53 0.17 2.03 (0.98) (1.34) (1.51) (1.53) Seniority: 1 term exp. −1.16** −1.02** 0.81 −0.85 (0.49) (0.51) (0.60) (0.65) Seniority: ≥2 terms exp. −2.42*** −1.78*** 0.58 −2.37*** (0.45) (0.53) (0.65) (0.82) Opposition −0.15 −0.14 0.51 0.41 (0.36) (0.32) (0.53) (0.68) Age 0.30 0.32** 0.15 0.08 (0.18) (0.12) (0.15) (0.19) Age squared/1000 −3.82* −3.78*** −2.44 −1.16 (1.96) (1.36) (1.58) (2.06) Female 0.06 −0.29 0.09 0.32 (0.29) (0.27) (0.36) (0.54) Constant −2.25 −0.96 0.47 4.03 (4.70) (2.70) (3.58) (4.66) N 331 316 253 243 Predictor Dep. var.: Re-running for same party Dep. var.: Good list position S1: 2010 S2: 2014 S3: 2010 S4: 2014 Parl. activity Years 1–3 (party-std.) 0.59* 0.62** 0.21 −0.45 (0.36) (0.28) (0.33) (0.38) Missed RCV Years 1–3 (party-std.) −0.02 −0.17 −0.24 0.32 (0.11) (0.15) (0.35) (0.30) Lag list leader −0.14 0.34 1.30 0.43 (0.48) (0.40) (0.87) (0.96) Lag good list pos. 0.72 −2.59 0.26 3.68*** (2.05) (1.97) (1.13) (0.98) Lag logit pers. vote 0.14 0.02 0.09 1.04*** (0.19) (0.17) (0.22) (0.26) Lag intra-p. comp. 0.00 −0.01 −0.08* −0.07 (0.04) (0.03) (0.05) (0.05) Replacement MP 0.23 −2.38 −1.68 0.52 (2.25) (2.03) (1.15) (1.10) Sole representative −0.13 −0.64 0.77 −0.85 (0.80) (0.73) (1.26) (1.28) PPG leadership 0.52 1.53 0.17 2.03 (0.98) (1.34) (1.51) (1.53) Seniority: 1 term exp. −1.16** −1.02** 0.81 −0.85 (0.49) (0.51) (0.60) (0.65) Seniority: ≥2 terms exp. −2.42*** −1.78*** 0.58 −2.37*** (0.45) (0.53) (0.65) (0.82) Opposition −0.15 −0.14 0.51 0.41 (0.36) (0.32) (0.53) (0.68) Age 0.30 0.32** 0.15 0.08 (0.18) (0.12) (0.15) (0.19) Age squared/1000 −3.82* −3.78*** −2.44 −1.16 (1.96) (1.36) (1.58) (2.06) Female 0.06 −0.29 0.09 0.32 (0.29) (0.27) (0.36) (0.54) Constant −2.25 −0.96 0.47 4.03 (4.70) (2.70) (3.58) (4.66) N 331 316 253 243 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. As in the Czech case, the results from the models with receiving a promising list position as dependent variable remain inconclusive. The signs of the effort-related variables are as expected only for the 2010 election. The most interesting finding is that in 2014 both the lagged dependent variable and the lagged personal vote turn out as strong predictors, while this was not the case in 2010. With 2014 being the first election after lowering the preference vote threshold, this could be interpreted in a way that the selectorate in the political parties put more weight on candidate popularity before the elections to be held under ‘personalised’ electoral rules. Of course, it is impossible to infer a causal effect of the electoral reform from this simple change over time. Another noteworthy result is that in 2014 the most senior MPs were less likely to receive a good list position, whereas in 2010 this variable had a positive sign. One interpretation of this finding is that some longer-serving MPs lose the determination to stay in parliament, but are still included on the list, leaving the decision to the voters whether to move them up the list again. We can also speculate that party selectors’ incentives to retain well-known MPs on the list merely for their vote-attraction potential increase in anticipation of an election held under personalised electoral rules. We are now turning to the analysis of the personal vote results (see Table 3 for the Czech Republic).13 Again, the 2010 election stands out from the others. There is a highly significant association between the degree of parliamentary activity and the personal vote MPs achieved. Consider again the implied change when increasing parliamentary activity by as much as the interquartile range: it amounts to an approximate shift from 0.047 (which is the median of the difference in intra-party vote share relative to the reference category) to 0.056. While not exactly huge, we should keep in mind that the preference vote threshold was lowered from 7% to 5% of the party vote, so 0.9 percentage points can make a difference for an MP’s direct election. Attendance at parliamentary votes, on the other hand, does not predict MPs’ electoral fortunes in any of the years. Table 3 Results from binomial-logistic regression of preference vote share in the Czech Republic Predictor Dep. var.: Log of preference votes relative to reference category C7: 2006 C8: 2010 C9: 2013 Parl. activity Years 1–4 (party-std.) 0.15* 0.35*** 0.07 (0.09) (0.09) (0.09) Missed RCV Years 1–4 (party-std.) −0.02 0.02 −0.001 (0.06) (0.04) (0.05) List leader 0.14 −0.01 −0.03 (0.10) (0.09) (0.09) List rank (linear) −0.10*** −0.02** −0.03*** (0.02) (0.01) (0.01) Lag list leader 0.003 0.03 −0.02 (0.13) (0.09) (0.10) Lag good list pos. 0.45*** 0.03 −0.12 (0.13) (0.14) (0.13) Lag of depend. var. 0.65*** 0.67*** 0.72*** (0.05) (0.06) (0.09) Lag intra-p. comp. 0.06** −0.003 0.002 (0.03) (0.01) (0.02) Replacement MP 0.59*** 0.41*** 0.08 (0.12) (0.13) (0.22) Sole representative 0.01 0.20 0.05 (0.19) (0.19) (0.17) PPG leadership −0.07 0.12 0.04 (0.12) (0.12) (0.10) Seniority: 1 term 0.06 0.03 0.18* (0.15) (0.12) (0.10) Seniority: ≥2 terms 0.14 0.02 0.15 (0.12) (0.10) (0.11) Opposition −0.18* −0.05 −0.35*** (0.10) (0.13) (0.13) Age 0.003 −0.04 −0.03 (0.03) (0.03) (0.03) Age squared/1000 −0.11 0.35 0.20 (0.32) (0.33) (0.34) Female 0.34*** −0.03 0.02 (0.13) (0.10) (0.10) Subnational politics 0.01 −0.06 0.13 (0.10) (0.06) (0.10) Constant −1.61** 0.63 0.30 (0.74) (0.75) (0.83) Adjusted R2 0.83 0.74 0.60 N 142 122 114 Predictor Dep. var.: Log of preference votes relative to reference category C7: 2006 C8: 2010 C9: 2013 Parl. activity Years 1–4 (party-std.) 0.15* 0.35*** 0.07 (0.09) (0.09) (0.09) Missed RCV Years 1–4 (party-std.) −0.02 0.02 −0.001 (0.06) (0.04) (0.05) List leader 0.14 −0.01 −0.03 (0.10) (0.09) (0.09) List rank (linear) −0.10*** −0.02** −0.03*** (0.02) (0.01) (0.01) Lag list leader 0.003 0.03 −0.02 (0.13) (0.09) (0.10) Lag good list pos. 0.45*** 0.03 −0.12 (0.13) (0.14) (0.13) Lag of depend. var. 0.65*** 0.67*** 0.72*** (0.05) (0.06) (0.09) Lag intra-p. comp. 0.06** −0.003 0.002 (0.03) (0.01) (0.02) Replacement MP 0.59*** 0.41*** 0.08 (0.12) (0.13) (0.22) Sole representative 0.01 0.20 0.05 (0.19) (0.19) (0.17) PPG leadership −0.07 0.12 0.04 (0.12) (0.12) (0.10) Seniority: 1 term 0.06 0.03 0.18* (0.15) (0.12) (0.10) Seniority: ≥2 terms 0.14 0.02 0.15 (0.12) (0.10) (0.11) Opposition −0.18* −0.05 −0.35*** (0.10) (0.13) (0.13) Age 0.003 −0.04 −0.03 (0.03) (0.03) (0.03) Age squared/1000 −0.11 0.35 0.20 (0.32) (0.33) (0.34) Female 0.34*** −0.03 0.02 (0.13) (0.10) (0.10) Subnational politics 0.01 −0.06 0.13 (0.10) (0.06) (0.10) Constant −1.61** 0.63 0.30 (0.74) (0.75) (0.83) Adjusted R2 0.83 0.74 0.60 N 142 122 114 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. Table 3 Results from binomial-logistic regression of preference vote share in the Czech Republic Predictor Dep. var.: Log of preference votes relative to reference category C7: 2006 C8: 2010 C9: 2013 Parl. activity Years 1–4 (party-std.) 0.15* 0.35*** 0.07 (0.09) (0.09) (0.09) Missed RCV Years 1–4 (party-std.) −0.02 0.02 −0.001 (0.06) (0.04) (0.05) List leader 0.14 −0.01 −0.03 (0.10) (0.09) (0.09) List rank (linear) −0.10*** −0.02** −0.03*** (0.02) (0.01) (0.01) Lag list leader 0.003 0.03 −0.02 (0.13) (0.09) (0.10) Lag good list pos. 0.45*** 0.03 −0.12 (0.13) (0.14) (0.13) Lag of depend. var. 0.65*** 0.67*** 0.72*** (0.05) (0.06) (0.09) Lag intra-p. comp. 0.06** −0.003 0.002 (0.03) (0.01) (0.02) Replacement MP 0.59*** 0.41*** 0.08 (0.12) (0.13) (0.22) Sole representative 0.01 0.20 0.05 (0.19) (0.19) (0.17) PPG leadership −0.07 0.12 0.04 (0.12) (0.12) (0.10) Seniority: 1 term 0.06 0.03 0.18* (0.15) (0.12) (0.10) Seniority: ≥2 terms 0.14 0.02 0.15 (0.12) (0.10) (0.11) Opposition −0.18* −0.05 −0.35*** (0.10) (0.13) (0.13) Age 0.003 −0.04 −0.03 (0.03) (0.03) (0.03) Age squared/1000 −0.11 0.35 0.20 (0.32) (0.33) (0.34) Female 0.34*** −0.03 0.02 (0.13) (0.10) (0.10) Subnational politics 0.01 −0.06 0.13 (0.10) (0.06) (0.10) Constant −1.61** 0.63 0.30 (0.74) (0.75) (0.83) Adjusted R2 0.83 0.74 0.60 N 142 122 114 Predictor Dep. var.: Log of preference votes relative to reference category C7: 2006 C8: 2010 C9: 2013 Parl. activity Years 1–4 (party-std.) 0.15* 0.35*** 0.07 (0.09) (0.09) (0.09) Missed RCV Years 1–4 (party-std.) −0.02 0.02 −0.001 (0.06) (0.04) (0.05) List leader 0.14 −0.01 −0.03 (0.10) (0.09) (0.09) List rank (linear) −0.10*** −0.02** −0.03*** (0.02) (0.01) (0.01) Lag list leader 0.003 0.03 −0.02 (0.13) (0.09) (0.10) Lag good list pos. 0.45*** 0.03 −0.12 (0.13) (0.14) (0.13) Lag of depend. var. 0.65*** 0.67*** 0.72*** (0.05) (0.06) (0.09) Lag intra-p. comp. 0.06** −0.003 0.002 (0.03) (0.01) (0.02) Replacement MP 0.59*** 0.41*** 0.08 (0.12) (0.13) (0.22) Sole representative 0.01 0.20 0.05 (0.19) (0.19) (0.17) PPG leadership −0.07 0.12 0.04 (0.12) (0.12) (0.10) Seniority: 1 term 0.06 0.03 0.18* (0.15) (0.12) (0.10) Seniority: ≥2 terms 0.14 0.02 0.15 (0.12) (0.10) (0.11) Opposition −0.18* −0.05 −0.35*** (0.10) (0.13) (0.13) Age 0.003 −0.04 −0.03 (0.03) (0.03) (0.03) Age squared/1000 −0.11 0.35 0.20 (0.32) (0.33) (0.34) Female 0.34*** −0.03 0.02 (0.13) (0.10) (0.10) Subnational politics 0.01 −0.06 0.13 (0.10) (0.06) (0.10) Constant −1.61** 0.63 0.30 (0.74) (0.75) (0.83) Adjusted R2 0.83 0.74 0.60 N 142 122 114 Notes: RCV = roll call votes, PPG = parliamentary party group. Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. Swedish voters do not appear to reward parliamentary activity with personal votes (compare Table 4); the coefficients for this variable are close to zero in both years. A somewhat curious finding is that missing more roll-call votes (relative to the standards of the own party) is associated with a larger personal vote (statistically significant at 10% in 2014). Applying the usual interquartile procedure implies a change in expected difference in vote share (relative to the reference category) from approximately 0.028 to 0.029, so the substantive size of the association is small. The obvious explanation may be that MPs who campaign (for themselves) in the constituency may be in parliament less frequently (Fukumoto and Matsuo, 2015). The measure used in Table 4 refers to the full four-year term, but the association is practically unchanged if we replace the variable with a version that considers only the first three years of the term (see Table A3c in Online Appendix A3). Perhaps this finding points to a more general pattern of MPs who care about their personal vote giving more priority to local at the expense of parliamentary work. Another possibility is that the able politicians are rewarded with important functions within the party, devoting more time to party service than parliamentary work. If these individuals are also better at attracting personal votes, this will induce a positive bias on the effect of missing roll-call votes.14 Table 4 Results from binomial-logistic regression of preference vote share in Sweden Predictor Dep. var.: Log of preference votes relative to reference category S5: 2010 S6: 2014 Parl. activity Years 1–4 (party-std.) 0.03 0.02 (0.04) (0.03) Missed RCV Years 1–4 (party-std.) 0.03 0.05* (0.04) (0.03) List leader 0.65*** 0.43*** (0.12) (0.06) List rank (linear) −0.06*** −0.07*** (0.02) (0.02) Lag list leader −0.46*** −0.35*** (0.09) (0.07) Lag good list pos. −0.58*** 0.03 (0.20) (0.09) Lag of depend. var. 0.82*** 0.79*** (0.06) (0.05) Lag intra-p. comp. −0.00 0.01 (0.01) (0.005) Replacement MP −0.29 0.15 (0.27) (0.11) Sole representative 0.03 0.001 (0.14) (0.12) PPG leadership −0.05 0.03 (0.16) (0.11) Seniority: 1 term −0.05 −0.13** (0.06) (0.06) Seniority: ≥2 terms −0.03 −0.11 (0.09) (0.07) Opposition 0.17** −0.00 (0.08) (0.06) Age −0.01 −0.02 (0.03) (0.02) Age squared/1000 0.04 0.06 (0.37) (0.22) Female 0.08 −0.02 (0.06) (0.04) Constant 0.21 −0.06 (0.53) (0.42) Adjusted R2 0.90 0.94 N 253 243 Predictor Dep. var.: Log of preference votes relative to reference category S5: 2010 S6: 2014 Parl. activity Years 1–4 (party-std.) 0.03 0.02 (0.04) (0.03) Missed RCV Years 1–4 (party-std.) 0.03 0.05* (0.04) (0.03) List leader 0.65*** 0.43*** (0.12) (0.06) List rank (linear) −0.06*** −0.07*** (0.02) (0.02) Lag list leader −0.46*** −0.35*** (0.09) (0.07) Lag good list pos. −0.58*** 0.03 (0.20) (0.09) Lag of depend. var. 0.82*** 0.79*** (0.06) (0.05) Lag intra-p. comp. −0.00 0.01 (0.01) (0.005) Replacement MP −0.29 0.15 (0.27) (0.11) Sole representative 0.03 0.001 (0.14) (0.12) PPG leadership −0.05 0.03 (0.16) (0.11) Seniority: 1 term −0.05 −0.13** (0.06) (0.06) Seniority: ≥2 terms −0.03 −0.11 (0.09) (0.07) Opposition 0.17** −0.00 (0.08) (0.06) Age −0.01 −0.02 (0.03) (0.02) Age squared/1000 0.04 0.06 (0.37) (0.22) Female 0.08 −0.02 (0.06) (0.04) Constant 0.21 −0.06 (0.53) (0.42) Adjusted R2 0.90 0.94 N 253 243 Notes: Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. Table 4 Results from binomial-logistic regression of preference vote share in Sweden Predictor Dep. var.: Log of preference votes relative to reference category S5: 2010 S6: 2014 Parl. activity Years 1–4 (party-std.) 0.03 0.02 (0.04) (0.03) Missed RCV Years 1–4 (party-std.) 0.03 0.05* (0.04) (0.03) List leader 0.65*** 0.43*** (0.12) (0.06) List rank (linear) −0.06*** −0.07*** (0.02) (0.02) Lag list leader −0.46*** −0.35*** (0.09) (0.07) Lag good list pos. −0.58*** 0.03 (0.20) (0.09) Lag of depend. var. 0.82*** 0.79*** (0.06) (0.05) Lag intra-p. comp. −0.00 0.01 (0.01) (0.005) Replacement MP −0.29 0.15 (0.27) (0.11) Sole representative 0.03 0.001 (0.14) (0.12) PPG leadership −0.05 0.03 (0.16) (0.11) Seniority: 1 term −0.05 −0.13** (0.06) (0.06) Seniority: ≥2 terms −0.03 −0.11 (0.09) (0.07) Opposition 0.17** −0.00 (0.08) (0.06) Age −0.01 −0.02 (0.03) (0.02) Age squared/1000 0.04 0.06 (0.37) (0.22) Female 0.08 −0.02 (0.06) (0.04) Constant 0.21 −0.06 (0.53) (0.42) Adjusted R2 0.90 0.94 N 253 243 Predictor Dep. var.: Log of preference votes relative to reference category S5: 2010 S6: 2014 Parl. activity Years 1–4 (party-std.) 0.03 0.02 (0.04) (0.03) Missed RCV Years 1–4 (party-std.) 0.03 0.05* (0.04) (0.03) List leader 0.65*** 0.43*** (0.12) (0.06) List rank (linear) −0.06*** −0.07*** (0.02) (0.02) Lag list leader −0.46*** −0.35*** (0.09) (0.07) Lag good list pos. −0.58*** 0.03 (0.20) (0.09) Lag of depend. var. 0.82*** 0.79*** (0.06) (0.05) Lag intra-p. comp. −0.00 0.01 (0.01) (0.005) Replacement MP −0.29 0.15 (0.27) (0.11) Sole representative 0.03 0.001 (0.14) (0.12) PPG leadership −0.05 0.03 (0.16) (0.11) Seniority: 1 term −0.05 −0.13** (0.06) (0.06) Seniority: ≥2 terms −0.03 −0.11 (0.09) (0.07) Opposition 0.17** −0.00 (0.08) (0.06) Age −0.01 −0.02 (0.03) (0.02) Age squared/1000 0.04 0.06 (0.37) (0.22) Female 0.08 −0.02 (0.06) (0.04) Constant 0.21 −0.06 (0.53) (0.42) Adjusted R2 0.90 0.94 N 253 243 Notes: Entries are logistic regression coefficients, standard errors clustered by party-list in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01. 5. Conclusion We have analysed the link between parliamentary effort and MPs’ performance at the re-nomination and re-election stages. Starting from the notion that the threat of non-re-election must be credible in order to induce MPs to take their job seriously, we distinguished between MPs’ retrospective evaluation by party selectors and voters. It is important to examine if the two principals differ with regard to rewarding or sanctioning MPs for their efforts, since there is a general trend to give voters more influence on intra-party seat allocation (Renwick and Pilet, 2016). In both the Czech Republic and Sweden, their flexible-list systems were even more personalised. We expected that parliamentary work of MPs matters less for voters’ decisions, because of limited monitoring capacities and the lower salience they may attach to work in parliament as compared to other forms of representation. This would be a cause of concern, since the ex ante control of parties through candidate selection is weakened, while voters may not be in a position to compensate through ex post control (cp. Bergman and Strøm, 2013). Overall, the evidence that party selectors and voters take into account MPs’ parliamentary work effort is mixed. The findings suggest that at the selection stage—as long as we assume that there is more than a last period effect—party selectors do sort out underperformers or signal them not to seek re-nomination at all. In contrast, parliamentary effort does not seem to affect the quality of the list position among the group of re-selected MPs. These patterns may also be interpreted as evidence that accountability works in practice: if MPs anticipate sanctioning, then all those interested in re-election will work, and relative differences between them can hardly predict the quality of their list rank. In addition, there are of course many other factors that influence list ranking, such as various balancing considerations, regarding gender, intra-constituency regional dispersion etc. Regarding the citizens, the case of the 2010 Czech election points out that, under certain conditions, the individual work of MPs can indeed make a difference to their personal vote. The election followed a major corruption scandal, and calls to vote in new politicians accompanied the introduction of the new preference voting rules. There was even an NGO whose main aim consisted in promoting to give preferences to ‘four [candidates] from the bottom’. Therefore, what may be required is a context that draws citizens’ attention to politicians’ personal record, which is facilitated by third-party actors like media and NGOs (compare also Stegmaier et al. (2014) on the success of female candidates in that election). This provides ground for optimism, suggesting that voters do not always know or care little about MPs’ work. Whether the reform of preferential voting before the 2010 election was a necessary condition for improving accountability is a question that we cannot answer. The association we found between lower attendance and a better personal vote performance in Sweden, on the other hand, allows for less positive interpretations. Pessimists may suspect that MPs shirk while mobilising voters through other, less valuable means (say social media entertainment). On the other hand, if the result is due to MPs simply substituting constituency service for parliamentary work, MPs may merely respond to (perceived) citizens’ expectations. Acknowledgements We would like to thank several MPs for participating in background interviews, Ertan Bat for research assistance as well as Václav Sklenář and the Czech/Swedish election authorities for providing additional data. All errors are ours. Funding This work was supported by German Research Foundation [grant number DA1692/1-1, to TD]; Czech Science Foundation [grant number GA16-04885S, to LL]. Conflicts of Interest No conflicts of interest. Footnotes 1 Compare also Martin (2010), a study using survey-based measures of constituency orientation and effort. The literature on participatory shirking is mainly concerned with the question whether legislators not running for re-election show lower attendance rates. Bernecker (2014) is an exception that also tests if participation in parliamentary votes improves election results; the study finds an effect for constituency candidates in Germany. 2 This is actually the case in both countries. For the Swedish case, see Dagens Industri, ‘They are most absent in parliament’ (25 June 2014) accessed at <http://www.di.se/artiklar/2014/6/25/de-skolkar-mest-i-riksdagen/> on 22 November 2017; in the Czech case, NGO Kohovolit.eu publishes regular reports on MPs’ activity which are used by almost all major media outlets. 3 For parliamentary systems, there seems to be little data on citizens’ preferences regarding constituency service per se (rather than more general representational focus). Vivyan and Wagner (2016) find that Britons prefer MPs to strike a moderate balance between constituency service and work on national policy (see also Cain et al. (1987, pp. 36–43) for data from 1979). 4 Own calculations based on http://www.volby.cz/pls/ps2006/ps111?xjazyk=CZ&xkraj=0&xstrana=0&xv=2&xt=1 and accessed at <http://www.volby.cz/pls/ps2010/ps111?xjazyk=CZ&xkraj=0&xstrana=0&xv=2&xt=1> on 22 November 2017. The sum of preference votes for a list was divided by the total number of ballots for that list. This was done for each list that won at least one seat, and then averaged across lists. 5 Yet, studying the first preference vote reform, Davidsson (2006) showed that electorally vulnerable MPs deviate more often from the party line in their private members’ bills and initiate a larger share of proposals with local focus. 6 This is in line with the notion that personal votes are intra-party votes, and that few voters choose a party just because of an individual candidate. 7 We choose two indicators, since different activities may have the character of substitutes, while this is not the case for voting. The absolute value of Pearson’s r between the two indicators does not exceed 0.15 in any of the subsamples. 8 While written questions exist in the Czech Republic as well, no systematic data on these are available (personal communication with parliamentary staff). For oral interpellations, we count all submitted interpellations save the ones that were withdrawn or not held due to absence of the asking MP. The amendment variable refers to the number of bills an MP amended. 9 Interpellations and questions withdrawn at a later point in time are not considered. 10 The term of the Swedish parliament is organized around official yearly sessions; for the Czech Republic we split the terms into parts with approximate length of one year. The legislative period starting in 2010 turned out to end after a little more than three years because of early elections. As a start point for the fourth part of this legislative period we choose the day after the resignation of Prime Minister Nečas. 11 In the Czech case, the data allow to separate ‘excused’ from unexcused missed votes. We ran all models also with the share of missed unexcused votes; the results are very similar. 12 The substantive results do not change when using regular standard errors. Random effects models for the personal vote also give very similar results (those for re-selection outcomes did not converge). 13 Tables A3a and A3b in the Online Appendix A3 report results of models using party indicator variables to account for variation in preference vote usage across parties. The conclusions for the key variables of interest remain unchanged. 14 We also examined how the results change when leaving out one type of activity at a time for creating the summary indicator of effort. These findings are presented in Online Appendix A3, Tables A3d–g. References André A. , Depauw S. , Shugart M. S. ( 2014 ) ‘The Effect of Electoral Institutions on Legislative Behavio’. In Martin S. , Saalfeld T. , Strøm K. (eds) The Oxford Handbook of Legislative Studies , Oxford , Oxford University Press , pp. 231 – 249 . André A. , Depauw S. , Shugart M. S. , Chytilek R. ( 2017 ) ‘ Party Nomination Strategies in Flexible-List Systems: Do Preference Votes Matter? ’, Party Politics , 27 , 589 – 600 . Google Scholar Crossref Search ADS Ashworth S. ( 2012 ) ‘ Electoral Accountability: Recent Theoretical and Empirical Work ’, Annual Review of Political Science , 15 , 183 – 201 . Google Scholar Crossref Search ADS Aylott N. ( 2013 ) ‘Parties and Party Systems in the North’. In Bergman T. , Strøm K. (eds) The Madisonian Turn: Political Parties and Parliamentary Democracy in Nordic Europe , Ann Arbor , University of Michigan Press , pp. 297 – 328 . Berg L. , Oscarsson H. ( 2015 ) 20 år med personval , Gothenburg , University of Gothenburg . Bergman T. ( 2006 ) ‘Sweden: From Separation of Power to Parliamentary Supremacy - and Back Again?’. In Strøm K. , Müller W. C. , Bergman T. (eds) Delegation and Accountability in Parliamentary Democracies , Oxford , Oxford University Press , pp. 594 – 619 . Bergman T. , Bolin N. ( 2013 ) ‘Swedish Democracy: Crumbling Political Parties, a Feeble Riksdag, and Technocratic Power Holders’. In Bergman T. , Strøm K. (eds) The Madisonian Turn: Political Parties and Parliamentary Democracy in Nordic Europe , Ann Arbor , University of Michigan Press , pp. 251 – 293 . Bergman T. , Strøm K. ( 2013 ) The Madisonian Turn: Political Parties and Parliamentary Democracy in Nordic Europe , Ann Arbor , University of Michigan Press . Bernecker A. ( 2014 ) ‘ Do Politicians Shirk When Reelection Is Certain? Evidence from the German Parliament ’, European Journal of Political Economy , 36 , 55 – 70 . Google Scholar Crossref Search ADS Berry S. , Levinsohn J. , Pakes A. ( 1995 ) ‘ Automobile Prices in Market Equilibrium ’, Econometrica , 63 , 841 – 890 . Google Scholar Crossref Search ADS Bowler S. ( 2010 ) ‘ Private Members' Bills in the UK Parliament: Is There an ‘Electoral Connection’? ’, The Journal of Legislative Studies , 16 , 476 – 494 . Google Scholar Crossref Search ADS Cain B. , Ferejohn J. , Fiorina M. ( 1987 ) The Personal Vote , Harvard , Harvard University Press . Crisp B. , Jensen K. M. , Shomer Y. ( 2007 ) ‘ Magnitude and Vote-Seeking ’, Electoral Studies , 26 , 727 – 734 . Google Scholar Crossref Search ADS Crisp B. , Olivella S. , Malecki M. , Sher M. ( 2013 ) ‘ Vote-Earning Strategies in Flexible List Systems: Seats at the Price of Unity ’, Electoral Studies , 32 , 658 – 669 . Google Scholar Crossref Search ADS Däubler T. , Bräuninger T. , Brunner M. ( 2016 ) ‘ Is Personal Vote‐Seeking Behavior Effective? ’, Legislative Studies Quarterly , 41 , 419 – 444 . Google Scholar Crossref Search ADS Davidsson L. ( 2006 ) I linje med partiet? Maktspel och lojalitet i den svenska riksdagen , Stockholm , SNS . Esaiasson P. , Holmberg S. ( 1996 ) Representation from Above: Members of Parliament and Representative Democracy in Sweden , Aldershot , Dartmouth . Fearon J. D. ( 1999 ) ‘Electoral Accountability and the Control of Politicians: Selecting Good Types versus Sanctioning Poor Performance’. In Przeworski A. , Stokes S. C. , Manin B. (eds) Democracy, Accountability, and Representation , Cambridge , Cambridge University Press , pp. 55 – 97 . Folke O. , Persson T. , Rickne J. ( 2016 ) ‘ The Primary Effect: Preference Votes and Political Promotions ’, American Political Science Review , 111 , 559 – 578 . Google Scholar Crossref Search ADS Frech E. ( 2016 ) ‘ Re-Electing MEPs: The Factors Determining Re-Election Probabilities ’, European Union Politics , 17 , 69 – 90 . Google Scholar Crossref Search ADS Fukumoto K. , Matsuo A. ( 2015 ) ‘ The Effects of Election Proximity on Participatory Shirking: The Staggered-Term Chamber as a Laboratory ’, Legislative Studies Quarterly , 40 , 599 – 625 . Google Scholar Crossref Search ADS Gherghina S. , Chiru M. ( 2010 ) ‘ Practice and Payment: Determinants of Candidate List Position in European Parliament Elections ’, European Union Politics , 11 , 533 – 552 . Google Scholar Crossref Search ADS Hermansen S. S. L. ( 2016 ) ‘ Striving for Influence: The Effect of Performance on Candidate Selection ’, Party Politics , Advance Access published on 13 September 2016, DOI: 10.1177/1354068816663036. Kellermann M. ( 2013 ) ‘ Sponsoring Early Day Motions in the British House of Commons as a Response to Electoral Vulnerability ’, Political Science Research and Methods , 1 , 263 – 280 . Google Scholar Crossref Search ADS Kolář P. , Kysela J. , Syllová J. , Georgiev J. , Pecháček Š. ( 2013 ) Parlament České republiky , Praha , Leges . Larsson T. , Bäck H. ( 2008 ) Governing and Governance in Sweden , Lund , Studentlitteratur . Linek L. , Mansfeldová Z. ( 2007 ) ‘ The Parliament of the Czech Republic, 1993–2004 ’, The Journal of Legislative Studies , 13 , 12 – 37 . Google Scholar Crossref Search ADS Loewen P. J. , Koop R. , Settle J. , Fowler J. H. ( 2014 ) ‘ A Natural Experiment in Proposal Power and Electoral Success ’, American Journal of Political Science , 58 , 189 – 196 . Google Scholar Crossref Search ADS Mansfeldová Z. , Linek L. ( 2009 ) Český parlament ve druhé dekádě demokratického vývoje , Prague , Institute of Sociology . Marcinkiewicz K. , Stegmaier M. ( 2015 ) ‘ Ballot Position Effects under Compulsory and Optional Preferential-List PR Electoral Systems ’, Political Behavior , 37 , 465 – 486 . Google Scholar Crossref Search ADS Martin S. ( 2010 ) ‘ Electoral Rewards for Personal Vote Cultivation under PR-STV ’, West European Politics , 33 , 369 – 380 . Google Scholar Crossref Search ADS Navarro J. ( 2010 ) ‘ Le travail parlementaire, un investissement payant? Les élections comme évaluation rétrospective du bilan des députés sortants ’, Revue internationale de politique comparée , 17 , 141 – 160 . Google Scholar Crossref Search ADS Outlý J. , Prouza J. ( 2009 ) Navrhování a výběr kandidátů: politické strany v ČR a ve střední Evropě , Olomouc , Periplum . Pemstein D. , Meserve S. A. , Bernhard W. T. ( 2015 ) ‘ Brussels Bound: Policy Experience and Candidate Selection in European Elections ’, Comparative Political Studies , 48 , 1421 – 1453 . Google Scholar Crossref Search ADS Renwick A. , Pilet J. -B. ( 2016 ) Faces on the Ballot. The Personalization of Electoral Systems in Europe , Oxford , Oxford University Press . Stegmaier M. , Tosun J. , Vlachová K. ( 2014 ) ‘ Women’s Parliamentary Representation in the Czech Republic: Does Preference Voting Matter? ’, East European Politics & Societies , 28 , 187 – 204 . Google Scholar Crossref Search ADS Vivyan N. , Wagner M. ( 2016 ) ‘ House or Home? Constituent Preferences over Legislator Effort Allocation ’, European Journal of Political Research , 55 , 81 – 99 . Google Scholar Crossref Search ADS Wilson S. L. , Ringe N. , van Thomme J. ( 2016 ) ‘ Policy Leadership and Re-Election in the European Parliament ’, Journal of European Public Policy , 23 , 1158 – 1179 . Google Scholar Crossref Search ADS Zubek R. , Stecker C. ( 2010 ) ‘ Legislatures and Policy Uncertainty: Evidence from East Central Europe ’, Journal of Public Policy , 30 , 63 – 80 . Google Scholar Crossref Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Parliamentary AffairsOxford University Press

Published: Oct 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off