Abstract The literature shows that, on average, electorally vulnerable Members of Parliament (MPs) are more active than their colleagues. This article investigates whether parties have an incentive to reselect active MPs in a county with a closed-list electoral system. Being active is usually a way to increase one’s visibility among the constituency. We argue that parties have incentives to reselect active MPs only when they believe that voters pay attention to the qualities of individual candidates. The Italian case is a perfect case to test this argument because the attention paid by voters to individual candidates varies significantly between the northern and southern regions. Our empirical analysis reveals that being active leads to a higher probability to be re-selected only in regions where the style of representation is personalised. Legislators are commonly assumed to be purposive actors with the basic goal of being reselected, and they are believed to strategically adapt their behaviour to the incentives set by the context (Strøm, 1997). In fact, it is surprising to note how much effort Members of Parliaments (MPs) devote to symbolic activities such as tabling questions (Martin and Rozenberg, 2012) or proposing legislation (Brunner, 2012) that have only a slight chance to influence policy outcomes. Recent studies have found that vulnerable MPs are likely to demonstrate more effort than their colleagues in more secure positions in legislative and extra-legislative arenas. Looking at the British House of Commons, Bowler (2010) found that MPs from more marginal seats are pushed to be more active. In a similar vein, Kellermann (2015) finds that electorally vulnerable MPs ask more questions than their colleagues. A study on bill introduction in Belgium also found that MPs are more active when they are reliant on personal votes (Bräuninger et al., 2012). MPs are also increasingly willing to communicate their activities to voters through new media, even in political systems where it is futile to cultivate a personal vote (Leston-Bandeira, 2012, p. 517). Understanding whether and to what extent these behaviours are consistent with the assumptions of rational institutionalists can advance our knowledge on how political representation operates. In the European context, political parties have a central place in the reselection of incumbents. Thus, how they evaluate whether parliamentarians deserve to be confirmed as candidates is particularly important. In countries with candidate-centred electoral systems, being active in parliament can be seen as a way to cultivate a personal vote (Carey and Shugart, 1995), but does this argument also apply in the many countries with party-centred systems? In those political systems, the relationship between MPs’ behaviour and re-election is completely mediated by political parties. The main goal of this article is to understand whether and under which conditions political parties reward MPs’ activity in the legislative arena. The analytical focus of this article is on party choices. Starting from the seminal study coordinated by Gallagher and Marsh (1987), a refined body of literature has assessed the procedures through which parties select candidates (Rahat and Hazan, 2001; Hazan and Rahat, 2010). However, to the best of our knowledge, only a few studies focus on how parties retrospectively evaluate the behaviour of incumbents when deciding whether to reselect them (Navarro, 2010). In this article, we assume that MPs wish to be reselected and strategically adapt their behaviour to achieve this goal. The second assumption holds that parties are rational and unitary actors with one main goal: winning elections. However, we recognise that parties may compete in several electoral arenas where voters follow different rules when deciding which party to vote for. In particular, while voters mainly rely on party labels in some contexts, their electoral choices are more influenced by the identity of the candidates in other contexts. The central hypothesis of this article is that parties evaluate incumbents based on their activity in the legislative arena only when they expect voters’ attention to be drawn to the personal characteristics of MPs. The Italian case has two unique features that makes it a good choice to test our model. Since 2005, Italy has adopted a closed-list proportional system that gives parties complete control over the fate of incumbents. It is the least-likely case to observe a direct appeal of MPs to voters because electoral lists are closed and long: the average magnitude of the districts is above 22. In addition, the candidate selection procedures adopted by Italian parties are rather centralised. In more recent years, some innovations towards democratisation (in terms of more decentralisation and broader inclusiveness) of candidate selection have been introduced in the statutes of (some) Italian parties.1 The specialised literature, however, is rather unanimous in arguing how, at least in the period covered by our analysis, ‘the push for more inclusiveness and more decentralisation, as emerged […] in the formal rules of some parties, clashed, then, with the current practices of candidate selection endorsed by the main parties turning out to be limited and centralised to a greater extent than formal rules’ (Cerruto et al., 2016, p. 877), being substantially, even comparatively speaking, an elite-driven process (Krouwel, 2012). To summarise this point, if parties typically have an important role in selecting candidates, Italy represents one country in which party centrality is virtually uncontested. Second, Italy is traditionally divided into several areas with different political cultures: in particular, political representation in the southern regions is much more personalised than in the rest of the country (Golden and Picci, 2008). From a methodological point of view, this difference provides an excellent opportunity to observe how parties react to different incentives set by the context while taking into account the institutional variables that complicate cross-country studies (Norris, 1997a). Our empirical findings make a strong case for the hypothesis according to which parties evaluate the behaviour of incumbents in ways that reflect the incentives set by the context. This article contributes to the literature on representation by showing that, even in closed-list proportional systems, electorally vulnerable MPs are more active than the average. This study also represents an original contribution to the literature on candidate selection because it shifts attention to one its rather neglected aspects, that is, the type of behaviour that parties value. Most studies on the relation between MPs’ behaviour and their career prospects have focused on the British House of Commons, whose electoral system is based on Single Member Districts (Norton and Wood, 1993; Searing, 1994; Norris, 1997b; Bowler, 2010). In that case, deselections are not common and re-election (which is often studied as a dependent variable of MPs’ behaviour) mainly depends on the level of popularity of the party and on the consensus enjoyed by incumbents. By analysing the fate of incumbents in a closed-list system, we can make inferences about party choices. In particular, we show that demonstrating effort in parliament by introducing Parliamentary Questions (but not Private Members’ Bills) increases the likelihood of reselection. However, the positive effect of parliamentary activity on reselection can be observed only in regions where the style of representation is personalised. This can be taken as important proof in favour of our argument. The next section develops our argument based on the assumption that both MPs and parties are purposive actors. We maintain that vulnerable MPs have more of an incentive to be more active than their colleagues. We also argue that parties are likely to reward this behaviour if they believe that it results in votes. The second section is devoted to explaining our research strategy, including the case selection, data and methods employed. The fourth section presents the results. Finally, in the conclusion, we assess the relevance of our results for the literature on the relationship between parliamentary activities and the chances of electoral reselection. 1. Theoretical framework and hypotheses To understand whether and under which conditions MPs’ activity is conducive to reselection, it is useful to divide the problem into its constituent parts. Parties, MPs and voters are the three relevant actors that play an important role in the reselection game. Each has objectives that can be pursued only when anticipating the behaviour of the other actors. Institutions set the context within which all actors can formulate their strategies. For the sake of simplicity, the following discussion will consider a political system where parties are unitary actors exercising fully centralised control over the candidate selection process. We are aware that in reality, parties may exhibit different degrees of factionalisation, which is likely to weaken the relationship between parliamentary activity and reselection. However, we prefer to begin with a more stylised model that, we believe, can be generalised to more complicated settings. For analytical purposes, we begin our exploration by first studying voters and then moving to MPs and parties. In the principal agent-framework of representation, voters delegate power to both parties and MPs (Andeweg and Thomassen, 2005). Although personal representation is often neglected in the studies of electoral systems, it is the necessary complement to party representation; electoral systems include procedures to assign seats to both parties and candidates (Colomer, 2011). Voting formulas can be classified according to the degree of party-centredness or candidate-centredness of electoral systems (Carey and Shugart, 1995); indeed, the opportunities given to voters to select their preferred candidate influence the strength of personal ties between MPs and voters. While some voting systems, such as the open-list PR or the Single Transferable Vote, are especially conducive to personal votes, some studies have shown that the quality of candidates have an impact on voting choices in a mixed system (Klingemann and Wessels, 2003) and even in closed-list systems (Riera, 2011). Despite the importance of electoral systems, other variables influence the propensity of constituents to develop personal ties with MPs. The Italian case is paradigmatic in this respect because under the same open-list PR that had been in force between 1948 and 1993, voters from different regions have always exhibited a variable propensity to cast preference votes. Preference votes had fundamentally been a southern phenomenon. Several explanations have been advanced to explain the differences in the propensity to cast preference votes between the southern and central/northern regions. Scholars have noted that the unusually strong personal links between constituents and MPs can be related to low levels of political consciousness (Furlong, 1977), a traditional and backward political culture (Parisi and Pasquino, 1979) with clientelism (Golden, 2003) or a combination of the three (Katz, 1985). A review of this literature is well beyond the scope of this article, but our main point here is that voters can have different propensities to vote for parties or for individual candidates, even in the same institutional setting, depending on important contextual factors. With regard to MPs, the basic assumption of this article that we share with rational choice scholars is that they are purposive actors; that is, they have a set of objectives and devise strategies to achieve them. When deciding how to employ their time and energy, which are two scarce resources, parliamentarians need to have a clear set of priorities in mind. Although MPs usually have more than one objective, re-election is generally seen as the basic goal shared by most (Schlesinger, 1966; Mayhew, 1974; Strøm, 1997). Research on U.S. Congress shows that incumbents usually adapt their behaviour to the expectations of their constituency. Representing the interests of the constituency (Cain et al., 1984) and working effectively within the legislature (Miquel and Snyder, 2006) prove to be useful strategies. In the European context, where parties play an important role in selecting candidates, the relationship between legislative performance and careers is much less certain. Studies on the British House of Commons, where MPs are elected in Single Member Districts, highlighted that those who focus their activities on the constituency do not improve their electoral prospects (Norton and Wood, 1993; Searing, 1994; Norris, 1997b). However, there is a growing literature showing that in several European legislative bodies, vulnerable MPs demonstrate more effort than their colleagues in the legislative arena. Looking at the British House of Commons, Bowler (2010) argues that introducing Private Members’ Bills can be used to enhance incumbents’ image among the constituency and found that MPs from more marginal seats are pushed to be more active. In a similar vein, Kellermann (2015) finds that electorally vulnerable MPs ask more questions than their colleagues. A study on bill introduction in Belgium also found that MPs are more active when they need to rely on personal votes (Bräuninger et al., 2012). The literature does not say much about the consequences of being active MPs. A study on the European Parliament (Navarro, 2010) shows that there is a small but not negligible relationship between the activity of MEPs and their chances of being re-elected. With regard to the UK, Bowler (2010) finds that presenting bills produced a small but significant increase in vote share. Regardless of the actual voting system, in virtually all European countries, parties have a crucial role in selecting and deselecting parliamentarians. Parties are MPs’ most proximate principals (Müller, 2000). Even in countries where MPs are traditionally considered representative of a given geographical constituency, such as the UK, national parties have still the power to deselect unpopular incumbents. The power of parties is greatest in closed-list proportional systems, where individual candidates cannot appeal directly to voters. Rather, the incumbents’ outcome is completely in the hands of their parties, which prepare the party lists for each district. In these cases, the relationship between incumbents’ activities and their electoral success is completely mediated by parties. Why and under which conditions should parties care about MP activity in parliament? The main objective for parties is winning elections and taking control of the government. Active MPs may help parties pursue this objective in at least two ways. Presenting Bills and PQs is often interpreted as a strategy to increase one’s reputation in the constituency. If a party believes that its electoral result in one district depends on the personal reputation of the candidates, it would have an obvious incentive to re-elect active MPs. However, there is also another reason for which active MPs might win re-election: a parliamentary group consisting of active and energetic parliamentarians may enhance the overall reputation of the party. These two arguments are not necessarily incompatible; however, they are based on different assumptions and produce rather different hypotheses. There is evidence that in consolidated democracies, parliaments are increasingly criticised for being unresponsive and distant from citizens (Leston-Bandeira, 2012). Whether this perception reflects an actual gap between parliaments and citizens or depends on broader political phenomena is well beyond the scope of this work. However, in this context, parties may benefit from organising a group of active parliamentarians who can form an enterprising party group. With the diffusion of information technologies and data journalism, it is becoming more frequent to read newspaper reports on the activity of MPs or ranking based on what MPs do in parliament. Believing that rewarding active incumbents can enhance the general reputation of the party produces the following hypothesis: H1: in general, more active MPs are more likely to be reselected by their parties.However, parties may also be motivated by a short-term consideration: visible politicians bring votes. The idea that parties select candidates for their popularity in the constituency is not new: for instance, German parliamentarians interviewed by Klingemann and Wessels (2003) believed that their capacity to win votes was one of the most important factors to secure reselection. If this is the case, parties would have a direct incentive to select active MPs. However, this argument only holds in case where voters are inclined to make their voting choices according to the identity of the candidates. On the contrary, when voting choices are independent from the composition of electoral lists, parties are indifferent to the visibility of their incumbents. In summary, if parties create electoral lists in an attempt to maximise their votes in the district, they will reward active MPs only in areas where politicians’ personal profiles matter to voters. H2: More active MPs are more likely to be reselected in contexts where the style of representation is more personalised.The two hypotheses are graphically represented in Figure 1. Figure 1 View largeDownload slide Effect of parliamentary activity on the probability of being reselected in Hypotheses 1 and 2. Figure 1 View largeDownload slide Effect of parliamentary activity on the probability of being reselected in Hypotheses 1 and 2. 2. Data and methods 2.1 Case selection We tested the hypotheses outlined above against data on parliamentary activity of the members of the Italian Chamber of Deputies during the 15th and the 16th Legislatures, which is between (May) 2008 and (March) 2013. Focusing on Italy and selecting these two legislative terms presents significant advantages from both a methodological and substantial point of view. The selection of the Italian case permits us to maximise the variation of basic attitudes towards a personalised style of representation. As stated above, Italy is traditionally divided into territories where people vote (almost) only for a party and zones where the importance of ‘personal vote’ is undoubtedly greater (see more on this point below). By analysing the 15th and the 16th legislative terms, we were able to hold the factor ‘electoral system’ constant. Even more importantly, the fact that the two legislatures were elected using a closed-list electoral system allowed us to check the validity of our hypotheses in the most (at least theoretically) party-controlled environment, focusing therefore (exclusively) on the choices by party selectorates. As far as individual cases are concerned, we started with the whole set of MPs elected to the Chamber of Deputies during the two legislative terms under consideration. We then excluded from our dataset MPs who were also members of executive bodies (as ministers or junior ministers) during their mandate and who, for this same reason, do not normally participate in the activities of the Chamber of Deputies. Deputies who did not serve for the entire duration of the respective legislatures (due, for instance, to resignation or premature death) are also excluded from the analysis. This leaves us with data on 1146 MPs. Note that each individual in our dataset is counted as one case each time he/she is elected to parliament. That is, if a single MP is elected in both the 15th and the 16th legislature, he/she has two entries (he/she is counted as two legislators) in the dataset. 2.2 The dependent variable: operationalisation and descriptive statistics of reselection As discussed above, being reselected to the party lists is the first basic objective of MPs seeking reselection. Our first aim is to see whether and under which conditions being active in parliament is useful for avoiding deselection. Operationalisation of the dependent variable is therefore rather straightforward. For this, we computed the variable reselection: a dichotomous variable that takes a value of 1 for legislators who are reselected by their respective party lists in the subsequent elections and 0 otherwise.2 Table 1 shows the percentage of MPs who were reselected and not reselected by the legislature. Overall, there is a relatively high quota of deputies (almost 69%) who were selected again as candidates for the following parliamentary elections.3 The data in Table 1, however, also show a non-negligible difference between the 15th (approximately 78% of reselected MPs) and the 16th legislatures (slightly more than 59%) due to a more general process of renovation of the Italian parliamentary class that has occurred with elections for the 17th legislature (Russo et al., 2014).4 Quite reasonably, this is also due to a basic difference between the two legislative terms under analysis. While the 16th legislature (almost) reached its natural end after 5 years (April 2008–March 2013), the 15th legislative term was closed early on March 2011 after the crisis of the Prodi II government and thus lasted only 2 years. The parties typically confirmed their incumbents because that experience was not counted as a full term and perhaps because they did not have time to recruit a new pool of candidates. Table 1 Rate of reselection in the 15th (2006–2008) and 16th (2008–2013) legislatures Legislature MPs not reselected (%) MPs reselected (%) Total (%) 15th (2006–2008) 127 (21.8) 456 (78.2) 583 (100) 16th (2008–2013) 230 (40.8) 333 (59.2) 563 (100) Total 357 (31.2) 789 (68.8) 1146 (100) Legislature MPs not reselected (%) MPs reselected (%) Total (%) 15th (2006–2008) 127 (21.8) 456 (78.2) 583 (100) 16th (2008–2013) 230 (40.8) 333 (59.2) 563 (100) Total 357 (31.2) 789 (68.8) 1146 (100) Table 1 Rate of reselection in the 15th (2006–2008) and 16th (2008–2013) legislatures Legislature MPs not reselected (%) MPs reselected (%) Total (%) 15th (2006–2008) 127 (21.8) 456 (78.2) 583 (100) 16th (2008–2013) 230 (40.8) 333 (59.2) 563 (100) Total 357 (31.2) 789 (68.8) 1146 (100) Legislature MPs not reselected (%) MPs reselected (%) Total (%) 15th (2006–2008) 127 (21.8) 456 (78.2) 583 (100) 16th (2008–2013) 230 (40.8) 333 (59.2) 563 (100) Total 357 (31.2) 789 (68.8) 1146 (100) 2.3 The explanatory factors The first hypothesis posits the likelihood of a legislator to be reselected as a function of his/her activity in parliament. We relied on a dataset containing information on bills (legislative activity) and parliamentary questions (non-legislative activity) presented by the members of the Chamber (Marangoni and Tronconi, 2011; Russo, 2011; Russo, 2013). To provide a reliable estimation of the level of activity of each MP, we computed two measures: the natural logarithm of the yearly average number of bills and the parliamentary questions presented as first signatory. Beyond their activities in the committee system, in the Chamber of Deputies, MPs may present bills, amendments, different types of questions and interpellations, motions and resolutions. MPs can also co-sponsor an act presented by one of his/her colleagues. We decided to consider only bills and ordinary questions (for written and oral answers) because these activities are less constrained by party discipline, yielding a reliable measure of MP activity. Only bills and questions presented as the first signatory were considered to exclude the frequent case where a parliamentary act is co-sponsored only to show support for a cause advocated by a colleague. Finally, a logarithmic form was chosen to assume that each additional bill or question has a decreasing impact on activity. In the Italian parliament, as elsewhere, some MPs present multiple bills and questions on the same issue to deliberately inflate their activity level. A log transformation attenuates this problem. To also include MPs who did not present any bills or questions in the analysis, we have increased each total by 0.5 before computing the yearly average. Legislative activity is the log of the yearly average number of bills presented as first signatories and non-legislative activity is the same measure based on parliamentary questions. The second hypothesis introduces a new explanatory factor in the picture. As argued, we expect that when compiling the electoral lists, parties are attentive to the different traditions of the various districts in terms of representation style and, therefore, are particularly likely to value more active and reputable incumbent MPs in those districts where the voters’ attitudes towards the ‘personal vote’ are traditionally stronger. It is well known that the propensity to cast preference votes (the basic indicator of a more or less personalised style of representation) has traditionally varied among Italian regions and that it has tended to grow from northern to southern Italian regions. A dichotomous variable indicating the macro-region (North or South) has therefore been traditionally used by scholars to distinguish different structures of opportunity in the electoral (and parliamentary) arena (for parties and individual MPs). We believe that this type of control is still useful (and necessary), even after the adoption of the closed-list proportional system in 2005. It is reasonable to expect that parties competing against one another in different territories with different traditions in terms of models of representation still operate within different structures of opportunity. At the same time, we believe that a simple north/south dichotomy might fail to capture the entire spectrum of territorial variation in this regard. We instead used the percentage of preference votes cast by voters from any given national electoral district in the closest regional councils’ elections (running with an open list system) as an indicator of the ‘latent’ attitude towards more or less personalised political representation of that district, and we assigned this value to the MPs elected in the same district. We refer to this variable as preferences. Other variables should naturally be introduced to control the hypothesised relationships. In particular, some traits in the political background and career profile of legislators might exert a significant impact on their chances of being reselected. Tenure, operationalised as the number of legislative terms in which each selected MP had served before the current term, is a first, natural factor to account for. On the one hand, there are sensible reasons to think that long-serving MPs have attained a prominent position within their party’s rank and file and are therefore more likely to be reselected. On the other hand, we might also expect a ‘natural’ turnover to involve more tenured MPs. Naturally enough, biographic and more personal factors may influence the likelihood of MPs’ reselection. A given MP, for instance, might forego re-election and retire. This is more likely to happen with the oldest MPs. To control for the potential impact of mechanisms of this type, we introduced the control variable AGE that, for each legislator in our dataset, is equal to the square of his/her age in years,5 and which we expect to have a negative impact on reselection. We also controlled for the parliamentary group to which the MPs belong (group) and whether they switched from one parliamentary group to another (group switching). A final variable we introduced measures the electoral vulnerability of individual MPs in previous elections. We used the measure that Andre et al. (2015) suggest for closed-list proportional electoral systems (the type of electoral institution we are examining by focusing on the 15th and 16th Italian legislatures), which is computed according to the following formula: vulnerability = P/S, where P is the list position occupied by each MP in previous elections and S is the total number of seats obtained in that district by the MP’s party. A legislator who was in 7th on the list of a party that won 10 seats in the district has a vulnerability of 0.7, while a legislator listed first has a vulnerability of 0.1. This variable is intended to complement the other control variables, giving a comprehensive albeit indirect measure of the prominence of MPs in their own party. In other words, being operationalised as just explained, the vulnerability of MPs also reflects their career profile and position within their respective party. We expect that more ‘prominent’ legislators were placed in less vulnerable district positions and were also relatively more likely to be reselected. Before moving to the empirical test, however, we want to devote some attention to the vulnerability-activity connection because of the interest that the literature has traditionally shown on this issue. Interestingly, the pattern we observe in Figure 2 is largely consistent with what other scholars have found when analysing the situation in other countries: more vulnerable MPs tend to be relatively more active than their safer colleagues. Figure 2 more specifically shows how the mean values of our two activity measures increase passing from relatively less vulnerable MPs (here, less than one standard deviation from the total mean of vulnerability) to relatively more vulnerable ones (more than one standard deviation from the mean). In particular, the increase in the level of legislative activity achieves the conventional test of statistical significance (the p-value of the ANOVA analysis is less than 0.001). Figure 2 View largeDownload slide Average values of Legislative and Non-legislative activity at different levels of MP vulnerability. Figure 2 View largeDownload slide Average values of Legislative and Non-legislative activity at different levels of MP vulnerability. Table 2 presents the main results of our analysis and presents the descriptive statistics of the independent and control variables that we have just introduced. Table 2 Explanatory factors of reselection: some descriptive statistics Explanatory factors Mean Standard deviation Min Max Non-legislative activity 0.971 1.264 −0.693 5.497 Legislative activity 0.423 0.921 −0.693 3.726 Preferences 49.949 24.867 23.300 90.800 Vulnerability 0.659 0.293 0.050 1.000 Age 51.492 9.320 26.000 82.000 Group switch (0–1) 0.171 — 0 1 Tenure 1.218 1.508 0.000 10.000 Explanatory factors Mean Standard deviation Min Max Non-legislative activity 0.971 1.264 −0.693 5.497 Legislative activity 0.423 0.921 −0.693 3.726 Preferences 49.949 24.867 23.300 90.800 Vulnerability 0.659 0.293 0.050 1.000 Age 51.492 9.320 26.000 82.000 Group switch (0–1) 0.171 — 0 1 Tenure 1.218 1.508 0.000 10.000 Table 2 Explanatory factors of reselection: some descriptive statistics Explanatory factors Mean Standard deviation Min Max Non-legislative activity 0.971 1.264 −0.693 5.497 Legislative activity 0.423 0.921 −0.693 3.726 Preferences 49.949 24.867 23.300 90.800 Vulnerability 0.659 0.293 0.050 1.000 Age 51.492 9.320 26.000 82.000 Group switch (0–1) 0.171 — 0 1 Tenure 1.218 1.508 0.000 10.000 Explanatory factors Mean Standard deviation Min Max Non-legislative activity 0.971 1.264 −0.693 5.497 Legislative activity 0.423 0.921 −0.693 3.726 Preferences 49.949 24.867 23.300 90.800 Vulnerability 0.659 0.293 0.050 1.000 Age 51.492 9.320 26.000 82.000 Group switch (0–1) 0.171 — 0 1 Tenure 1.218 1.508 0.000 10.000 3. Model and analysis Reselection has a typically dichotomous format: it can only assume values of 0 (not reselected) or 1 (reselected). Using a logit model, therefore, is the ‘natural’ solution we opted to use in the analysis to follow. It is noteworthy, once again, that any MP in our dataset is counted as one case any time he/she is elected to parliament (during the two legislatures under analysis), thus leading to non-independent observations. We therefore opted to cluster the standard errors on individual MPs to correct this possible bias (Rogers, 1993). Table 3 reports the two models we estimated. The difference between models 1 and 2 is that, consistently with H2, the latter model also estimates the effect of parliamentary questions and bill introduction in interactions with preferences. Table 3 Logit regressions with reselection (yes = 1) as a dependent variable Model 1 Model 2 Non-legislative activity (PQs) 0.035 −0.243 (0.071) (0.156) Legislative activity (Bills) 0.083 0.084 (0.097) (0.211) Preferences −0.012** (0.003) PQs * preferences 0.006* (0.003) Bills * preferences −0.000 (0.003) Vulnerability −0.849** −0.856** (0.277) (0.279) Age −0.051 −0.062 (0.094) (0.094) Age^2 −0.000 −0.000 (0.001) (0.001) Tenure −0.220*** −0.227*** (0.062) (0.063) Group switch (0–1) −0.395* −0.374 (0.200) (0.204) Group dummies Not shown Constant 5.188*** 6.040*** (2.491) (2.512) Log pseudolikelihood −601.471 −596.462 Pseudo-R2 0.160 0.168 N 1155 1155 Model 1 Model 2 Non-legislative activity (PQs) 0.035 −0.243 (0.071) (0.156) Legislative activity (Bills) 0.083 0.084 (0.097) (0.211) Preferences −0.012** (0.003) PQs * preferences 0.006* (0.003) Bills * preferences −0.000 (0.003) Vulnerability −0.849** −0.856** (0.277) (0.279) Age −0.051 −0.062 (0.094) (0.094) Age^2 −0.000 −0.000 (0.001) (0.001) Tenure −0.220*** −0.227*** (0.062) (0.063) Group switch (0–1) −0.395* −0.374 (0.200) (0.204) Group dummies Not shown Constant 5.188*** 6.040*** (2.491) (2.512) Log pseudolikelihood −601.471 −596.462 Pseudo-R2 0.160 0.168 N 1155 1155 Notes: Cell entries are logit coefficients. Clustered standard errors are in parentheses. * p < 0.05, **p < 0.01, ***p < 0.001. Table 3 Logit regressions with reselection (yes = 1) as a dependent variable Model 1 Model 2 Non-legislative activity (PQs) 0.035 −0.243 (0.071) (0.156) Legislative activity (Bills) 0.083 0.084 (0.097) (0.211) Preferences −0.012** (0.003) PQs * preferences 0.006* (0.003) Bills * preferences −0.000 (0.003) Vulnerability −0.849** −0.856** (0.277) (0.279) Age −0.051 −0.062 (0.094) (0.094) Age^2 −0.000 −0.000 (0.001) (0.001) Tenure −0.220*** −0.227*** (0.062) (0.063) Group switch (0–1) −0.395* −0.374 (0.200) (0.204) Group dummies Not shown Constant 5.188*** 6.040*** (2.491) (2.512) Log pseudolikelihood −601.471 −596.462 Pseudo-R2 0.160 0.168 N 1155 1155 Model 1 Model 2 Non-legislative activity (PQs) 0.035 −0.243 (0.071) (0.156) Legislative activity (Bills) 0.083 0.084 (0.097) (0.211) Preferences −0.012** (0.003) PQs * preferences 0.006* (0.003) Bills * preferences −0.000 (0.003) Vulnerability −0.849** −0.856** (0.277) (0.279) Age −0.051 −0.062 (0.094) (0.094) Age^2 −0.000 −0.000 (0.001) (0.001) Tenure −0.220*** −0.227*** (0.062) (0.063) Group switch (0–1) −0.395* −0.374 (0.200) (0.204) Group dummies Not shown Constant 5.188*** 6.040*** (2.491) (2.512) Log pseudolikelihood −601.471 −596.462 Pseudo-R2 0.160 0.168 N 1155 1155 Notes: Cell entries are logit coefficients. Clustered standard errors are in parentheses. * p < 0.05, **p < 0.01, ***p < 0.001. Several control variables have proven to be significant in both our models. Vulnerability in the previous election, an implicit measure of party standing, is a reliable predictor of reselections: those who had occupied the safest list position in the past are also more likely to be selected again. The quadratic term for age is also statistically significant in the expected direction: elderly MPs are likely to retire, perhaps spontaneously refusing to defend their seat. All else being equal, tenure appears to have a negative impact, confirming the expectation that it is progressively more difficult to obtain successive confirmation, at least for those who do not acquire a prominent position.6 Interestingly, switching party group does not appear to be a rewarding strategy for MPs.7 Coming to our hypotheses, we can conclude that, according to the first model, H1 finds no support: neither presenting parliamentary questions nor being the first signatory of multiple bills has an effect on the probability of being reselected. Being an active MP does not contribute to being named in the ballot list. By contrast, the second model is in line with our second hypothesis: the interaction between preferences and PQs is positive (as expected) and statistically significant.8 On the contrary, legislative activity has no impact. Perhaps introducing bills with almost no hope of becoming law is regarded as a rather futile activity by selectors.9 The interpretation of interactions in logistic regressions is not intuitive and can be better illustrated graphically. Figure 3 plots the average marginal effect of PQs at different levels of the variable by measuring the percentage of voters casting preference votes in different regions. The grey area around the solid line is the 95% confidence interval of the marginal effect. Whenever the confidence interval does not include the 0 axis, the marginal effect of the independent viable is significantly different from 0. Figure 3 View largeDownload slide Average marginal effect of PQs on the probability of being reselected and predicted probabilities of reselection in regions with different levels of preferences. Figure 3 View largeDownload slide Average marginal effect of PQs on the probability of being reselected and predicted probabilities of reselection in regions with different levels of preferences. The figure clearly shows that presenting parliamentary questions has a positive and significant impact only above a relatively high level of preferences (approximately 70%). This clearly confirms H2 and leads us to better specification of our findings. All else being equal, more active MPs have a higher likelihood of being reselected, but this applies only where the prevailing style of representation is personalised. To appreciate the substantive importance of this result, it should be noted that in five regions, all of which are located in Southern Italy, the variable of preference votes exceeds 70%: Puglia (70.8%), Calabria (84.1%), Basilicata (85.9%), Sicilia (88.7%) and Campania (90.8%). The other graphs in Figure 3 plot the probability of reselection as conditional on the (log) number of PQs asked per year at three different levels of variable preferences. In regions with low or average levels of preferential voting (10th and 50th percentile, respectively) asking PQs does not improve the probability of reselection. Indeed, in regions where preferences are virtually absent, being an active questioner appears even dangerous. By contrast, in a region where preferential voting is the norm rather than the exception, an incumbent asking only one PQ per year has an approximately 60% probability of being reselected, while a colleague asking 20 questions per year would improve that figure by 8% points. Finally, it would be interesting to see whether being an active questioner can be useful, not only for reselection but also to obtain a better position in the party list. In other words, the question is whether activism in parliament can decrease electoral vulnerability. To answer this question, we estimated a simple linear regression model with electoral vulnerability in the next election as dependent variable, with the same set of control variables already employed in the previous models. The results presented in Table 4 are straightforward. Electoral vulnerability in the following election is only explained by electoral vulnerability in the previous election, a measure of pre-existing party standing, and tenure, which tends to decrease vulnerability. Parliamentary questions are not helpful for securing a safer position in the list, even in regions where preferential voting is popular. In other words, being an active questioner can earn one a position on the party list but will not improve your party standing. In interpreting these results, we must consider that the vulnerability level of those who had been elected in the previous election was already low on average. Further studies, with a different research design, could test whether being active in parliament makes a difference, at least for those in the margins. Table 4 Linear regression with vulnerability at the next election as a dependent variable Model 3 Non-legislative activity (PQs) −0.013 (0.030) Preferences 0.001 (0.001) PQs * preferences 0.001 (0.000) Vulnerability (previous election) 0.595*** (0.059) Age 0.0163 (0.014) Age^2 −0.000 (0.000) Tenure −0.038* (0.018) Group_switch (0–1) −0.057 (0.109) Group dummies Not shown Constant −0.303 (0.367) R2 0.296 N 715 Model 3 Non-legislative activity (PQs) −0.013 (0.030) Preferences 0.001 (0.001) PQs * preferences 0.001 (0.000) Vulnerability (previous election) 0.595*** (0.059) Age 0.0163 (0.014) Age^2 −0.000 (0.000) Tenure −0.038* (0.018) Group_switch (0–1) −0.057 (0.109) Group dummies Not shown Constant −0.303 (0.367) R2 0.296 N 715 Note: Standard errors are in parentheses. * p < 0.05, **p < 0.01, ***p < 0.001. Table 4 Linear regression with vulnerability at the next election as a dependent variable Model 3 Non-legislative activity (PQs) −0.013 (0.030) Preferences 0.001 (0.001) PQs * preferences 0.001 (0.000) Vulnerability (previous election) 0.595*** (0.059) Age 0.0163 (0.014) Age^2 −0.000 (0.000) Tenure −0.038* (0.018) Group_switch (0–1) −0.057 (0.109) Group dummies Not shown Constant −0.303 (0.367) R2 0.296 N 715 Model 3 Non-legislative activity (PQs) −0.013 (0.030) Preferences 0.001 (0.001) PQs * preferences 0.001 (0.000) Vulnerability (previous election) 0.595*** (0.059) Age 0.0163 (0.014) Age^2 −0.000 (0.000) Tenure −0.038* (0.018) Group_switch (0–1) −0.057 (0.109) Group dummies Not shown Constant −0.303 (0.367) R2 0.296 N 715 Note: Standard errors are in parentheses. * p < 0.05, **p < 0.01, ***p < 0.001. 4. Conclusions Studies on candidate selection have traditionally been more interested in describing and comparing the procedures adopted by party organisations to prepare their electoral lists than explaining the preferences of selectors. The few comparative works that explicitly consider this aspect do not allow for any generalisations (Norris, 1997a). We attempted to go further in this direction by analysing the relationship between MP activity in parliament and their chances of being reselected by their own party’s selectorate. Building on the assumption that both parties and MPs are purposive actors, we hypothesised two reasons for which incumbents’ efforts can be rewarded with reselection: first, having an active parliamentary group may enhance the overall reputation of the party; second, active MPs, being more visible, are more capable of winning votes in the district. Following this line of reasoning, we have also hypothesised that parties tend to reward more active MPs, especially where voters show more positive attitudes towards ‘personal votes’. We focused on the Italian case because we believe it to be truly noteworthy in this regard for at least two reasons. Italy has an electoral system that is markedly party-centred (and is particularly so with the closed-list proportional system adopted during the 15th and 16th legislatures, as we have observed here). Second, Italian regions have traditionally exhibited great variation as far as the attitude towards personalised voting is concerned. We moved from the argument that this variation provides different structures for opportunities and incentives (for both MPs and parties) and that these different structures are also at work in closed-list electoral systems (where the individual attitude towards personal voting remains necessarily ‘latent’ because voters cannot indicate any personal preference). Methodologically speaking, therefore, the Italian case offered an opportunity to analyse how parties handle the different incentives set by the context while controlling for institutional factors that would necessarily complicate cross-country studies. The rules to select candidates indeed considerably vary in different countries, and the combination between these procedures and electoral systems creates a level of complexity that is difficult to disentangle (Shomer, 2014). This single (relevant) case study made this effort much easier and reached one main conclusion. We demonstrate that party selectorates appreciate MPs’ activity in parliament (i.e., reward more active MPs with reselection) almost exclusively in districts traditionally characterised by more personalised styles of representation: what this suggests is that the preferences and choices of parties in selecting their candidates are (also) conditional on the electoral context. We believe these results are important for legislative specialists who have analysed the factors explaining different MPs behaviour and ‘performances’ at length (to be active, in this sense, might be a ‘rational’ option for MPs aiming at increasing their own chances of being re-elected). Shedding light on the factors guiding the strategies of parties in a crucial act such as selecting individual representatives, in general, might significantly contribute to our knowledge of the functioning of democratic mechanisms of accountability and delegation. Conflicts of Interest The authors declare that there is no conflict of interest regarding the publication of this article. Funding This research received no specific grant from any funding agency. Footnotes 1 It is specifically the statute of the Democratic Party and of the Brothers of Italy-National Alliance that introduced open primaries to select some of the candidates since the 2013 parliamentary elections. The Five Stars Movement use online primaries that are open only to ‘certified members’ to select the candidates for national parliamentary elections. However, the Five Stars Movement’s deputies are not covered by our analysis, since they entered parliament only in the last legislative term. 2 In cases of MPs switching parties during the electoral term, we look at the electoral lists of their new parties. Obviously, party-switchers will not be re-elected by the parties they have left. 3 We consider an MP reselected if he/she is a re-candidate in the lists for the Senate. 4 The 2013 elections, in fact, have paved the way to a transformation of the Italian party system with the success of new emerging parties (the Five Stars Movement, above all) and the crisis (and transformation) of main political actors, such as the centre-right People of Freedom. 5 We use the square of the age because we assumed that the probability of an MP retiring was not a linear function of his/her age but tends to increase at a higher pace for older MPs (i.e., at highest values of age) than for younger MPs. 6 The alternative expectation in this regard (the one hypothesising that more tenured MPs are relatively more able to be re-elected) is therefore not supported by the data we analysed. 7 This is a rather remarkably result that falls outside the core interests of this piece of research. It nonetheless merits a future in-depth analysis given the magnitude that the phenomenon of group switching has recently assumed in Italy. A variable distinguishing the different ‘trajectories’ of group switching, for instance, might contribute to solving the puzzle. It could be the case that MPs abandoning their original group and forming (or joining) other smaller groups are de facto not re-elected because their new groups do not have the resources to form a new political movement and fail to present their own electoral list. 8 Estimating based on two different models, one for parliamentary questions and one for bills, yields almost identical results. 9 Future analyses should include other measures of legislative activism, such as proposing amendments or managing to find sufficient parliamentary support for one’s bills. References Andeweg R. B., Thomassen J. J. ( 2005) ‘Modes of Political Representation: Toward a New Typology’, Legislative Studies Quarterly , 30, 507– 528. Google Scholar CrossRef Search ADS Andre A., Depauw S., Martin S. ( 2015) ‘Electoral Systems and Legislators’ Constituency Effort: the Mediating Effect of Electoral Vulnerability’, Comparative Political Studies , 48, 464– 496. 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 Bräuninger T., Brunner M., Däubler T. ( 2012) ‘Personal Vote-Seeking in Flexible List Systems: How Electoral Incentives Shape Belgian MPs’ Bill Initiation Behaviour: Personal Vote-Seeking in Flexible List Systems’, European Journal of Political Research , 51, 607– 645. Google Scholar CrossRef Search ADS Brunner M. ( 2012) Parliaments and Legislative Activity? Motivations for bill introduction , Wiesbaden, Springer VS. Cain B. E., Ferejohn J. A., Fiorina M. P. ( 1984) ‘The Constituency Service Basis of the Personal Vote for US Representatives and British Members of Parliament’, American Political Science Review , 78, 110– 125. Google Scholar CrossRef Search ADS Carey J. M., Shugart M. S. ( 1995) ‘Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas’, Electoral Studies 14, 417– 439. Google Scholar CrossRef Search ADS Cerruto M., Facello C., Raniolo F. ( 2016) ‘How Has the Secret Garden of Politics Changed in Italy (1994–2015)?’, American Behavioral Scientist , 60, 869– 888. Google Scholar CrossRef Search ADS Colomer J. M. (ed.) ( 2011) Personal Representation: the Neglected Dimension of Electoral Systems . ECPR—Studies in European Political Science, Colchester, ECPR Press. Furlong P. F. ( 1977) ‘Il Voto di Preferenza e L’Elettorato Romano’, Rivista Italiana di Scienza Politica , 7, 393– 409. Gallagher M., Marsh M. ( 1987) Candidate Selection in Comparative Perspective: The Secret Garden of Politics , London/Newbury Park, CA, Sage. Golden M. A. ( 2003) ‘Electoral Connections: the Effects of the Personal Vote on Political Patronage, Bureaucracy and Legislation in Postwar Italy’, British Journal of Political Science , 33, 189– 212. Google Scholar CrossRef Search ADS Golden M. A., Picci L. ( 2008) ‘Pork-Barrel Politics in Postwar Italy, 1953–94’, American Journal of Political Science , 52, 268– 289. Google Scholar CrossRef Search ADS Hazan R. Y., Rahat G. ( 2010) Democracy Within Parties: Candidate Selection Methods and Their Political Consequences , Oxford, Oxford University Press. Google Scholar CrossRef Search ADS Katz R. S. ( 1985) ‘Preference Voting in Italy: Votes of Opinion, Belonging, or Exchange’, Comparative Political Studies , 18, 229– 249. Google Scholar CrossRef Search ADS Kellermann M. ( 2015) ‘Electoral Vulnerability, Constituency Focus, and Parliamentary Questions in the House of Commons: Electoral Vulnerability and PQs’, The British Journal of Politics & International Relations , 18, 90– 106. Google Scholar CrossRef Search ADS Klingemann H. D., Wessels B. ( 2003) ‘The Political Consequences of Germany’s Mixed‐Member System: Personalization at the Grass Roots?’. In Shugart M. S., Wattenberg M. P. (eds) Mixed-Member Electoral Systems , Oxford, Oxford University Press, pp. 279– 296. Google Scholar CrossRef Search ADS Krouwel A. ( 2012) Party Transformations in European Democracies , New York, University of New York Press. Leston-Bandeira C. ( 2012) ‘Parliaments’ Endless Pursuit of Trust: Re-focusing on Symbolic Representation’, The Journal of Legislative Studies , 18, 514– 526. Google Scholar CrossRef Search ADS Marangoni F., Tronconi F. ( 2011) ‘When Territory Matters: Parliamentary Profiles and Legislative Behaviour in Italy (1987–2008)’, The Journal of Legislative Studies , 17, 415– 434. Google Scholar CrossRef Search ADS Martin S., Rozenberg O. ( 2012) The Roles and Function of Parliamentary Questions . London, Routledge. Mayhew D. R. ( 1974) Congress: the Electoral Connection , New Haven, Yale University Press. Miquel G. P. I., Snyder J. M. ( 2006) ‘Legislative Effectiveness and Legislative Careers’, Legislative Studies Quarterly , 31, 347– 381. Google Scholar CrossRef Search ADS Müller W. C. ( 2000) ‘Political Parties in Parliamentary Democracies: Making Delegation and Accountability Work’, European Journal of Political Research , 37, 309– 333. Navarro J. ( 2010) ‘Le Travail Parlementaire, un Investissement Payant Ayantiélections Comme Évaluation Rétrospective du Bilan des Députés Sortants’, Revue Internationale de Politique Comparée , 17, 141– 160. Google Scholar CrossRef Search ADS Norris P. ( 1997a) Passages to Power: Legislative Recruitment in Advanced Democracies . Cambridge, Cambridge University Press. Norris P. ( 1997b) ‘The Puzzle of Constituency Service’, The Journal of Legislative Studies , 3, 29– 49. Google Scholar CrossRef Search ADS Norton P., Wood D. ( 1993) Back from Westminster: British Members of Parliament and Their Constituents , Lexington, University of Kentucky Press. Parisi A., Pasquino G. ( 1979) ‘Changes in Italian Electoral Behavior: the Relationship between Parties and Voters’ In Lange P., Tarrow S. (eds) Italy in Transition , London, Frank Cass, pp. 6– 30. Rahat G., Hazan R. Y. ( 2001) ‘Candidate Selection Methods: an Analytical Framework’, Party Politics , 7, 297– 322. Google Scholar CrossRef Search ADS Riera P. ( 2011) ‘Closed Party List’. In Colomer J. M. (ed) Personal Representation: the Neglected Dimension of Electoral Systems , ECPR—Studies in European Political Science, Colchester, ECPR Press, pp. 55– 80. Rogers W. ( 1993) ‘Regression Standard Errors in Clustered Samples’, Stata Technical Bulletin , 3, 19– 23. Russo F. ( 2011) ‘The Constituency as a Focus of Representation: Studying the Italian Case through the Analysis of Parliamentary Questions’, The Journal of Legislative Studies , 17, 290– 301. Google Scholar CrossRef Search ADS Russo F. ( 2013) Gli Onorevoli. Cosa Fanno e Come ci Rappresentano i Nostri Parlamentari , Bologna, Il Mulino. Russo F., Tronconi F., Verzichelli L. ( 2014) ‘Snipers and Switchers. The Difficulties of Parliamentary Representation in the Italian XVII Legislature’, Polis , 1, 85– 106. Schlesinger J. A. ( 1966) Ambition and Politics; Political Careers in the United States , Chicago, Rand McNally. Searing D. ( 1994) Westminster’s World: Understanding Political Roles , Cambridge, Harvard University Press. Shomer Y. ( 2014) ‘What Affects Candidate Selection Processes? A Cross-National Examination’, Party Politics , 20, 533– 546. Google Scholar CrossRef Search ADS Strøm K. ( 1997) ‘Rules, Reasons and Routines: Legislative Roles in Parliamentary Democracies’, The Journal of Legislative Studies , 3, 155– 174. 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: firstname.lastname@example.org
Parliamentary Affairs – Oxford University Press
Published: Jan 31, 2018
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
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
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.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera