From a party system to a
Transcript
From a party system to a
From a party system to a ‘candidate system’? ‘Lords of Preferences’ and electoral support in Calabria Vincenzo Emanuele (LUISS “Guido Carli”) and Bruno Marino (Scuola Normale Superiore) 1. Introduction In this article we analyse patterns of personal vote in Calabria, a Southern Italian region, focusing on the role of major individual candidates in regional elections. We test candidates’ ability to shape the electoral competition and to predict electoral results, replacing the traditional role exerted by political parties and their system of alliances. Unlike previous research exploring personal vote in Southern Italy, mainly focused either on exclusively theoretical contributions or on purely descriptive data, this article is, to our knowledge, the first attempt to build an original empirical model to assess whether and to what extent personal voting shapes the electoral competition and predicts electoral results. Since the birth of the so-called Italian ‘Second Republic’, Calabrian regional elections have been characterised by an almost perfect alternation in power between a centre-left and a centreright coalition. Specifically, in the 2010-2014 electoral period in Calabria this alternation was followed by a noticeable level of instability affecting parties, coalitions and the party system as a whole. The landslide victory of centre-right in 2010 (57.8 per cent of ballots cast for Scopelliti against the centre-left candidate, Loiero, who obtained 32.3 per cent of votes) was followed by the return to power of centre-left in 2014 (61.4 per cent for the newly-elected President Oliverio against the 23.6 per cent of his conservative opponent Ferro)1. Moreover, out of the 16 lists contesting the 2010 regional election, only three (PD, UDC and Autonomia e Diritti) maintained their party label in 2014, while as many as 12 lists ran for the first time in a regional election2. This means that there has been a significant amount of party replacement (Birch 2003). At the same time, the level of electoral volatility (Pedersen 1979) has been as high as 51.33, while the share of votes obtained by non-national parties4 rose from 26.6 per cent in 2010 to 43.3 per cent in 2014, which means that both the supply side (parties) and the demand side (voters) have faced great fluidity in Calabria in the last years. This first empirical evidence prompts a serious reflection about the state of art of Calabria’s party system. Starting from Sartori’s definition of party system as ‘the system of interactions 1 resulting from inter-party competition’ (2005, p. 39), Casal Bértoa (2014, p. 17) defines a party system as ‘institutionalised’ when ‘the patterns of interactions between political parties become routine, predictable and stable over time’. This definition emphasises the predictability of interparty competition as the key element of party system institutionalisation. Conversely, as emphasised by Bardi and Mair (2008), when this system of interactions lacks, there is not a ‘system’ but only a ‘set’ of parties. Building on previously presented data regarding the dramatic level of party replacement and electoral volatility, we could argue that the Calabrian context can be considered as an unpredictable environment. Therefore, following Chiaramonte and Emanuele (2015), it can also be considered as a ‘de-institutionalised’ party system. Moreover, a certain degree of continuity in the main components of the system is conceived as a fundamental criterion to identify the presence of a party system (Mainwaring & Torcal 2006). On the contrary, when the identity of the top party vote-getters shows a great variability between elections, then the party universe is best described as an ‘inchoate’ party system or even as a ‘non-system’ (Sanchez 2009). While many recent contributions on party systems’ institutionalization and on their degree of systemness have focused on Latin American and on Central and Eastern European countries - that is, on polities in which the institutionalisation of the party systems is associated with the consolidation of democratic regimes - it could not be surprising to study the Calabrian case by using these tools of analysis, given the data we have just presented. In this context of extreme instability regarding party labels and voters’ preferences, and consequently of high unpredictability of electoral results, an important element of continuity of the system is secured by the extent to which Calabrian voters use the tool of preferential voting. In this sense, Calabria is similar to other contexts of Southern Europe and, in particular, of Southern Italy, where personal relationships between voters and candidates are an important element of grassroots politics. As we can note by looking at Figure 1, the level of preferential voting in Calabria has been particularly high since the early 2000s, ranging from 82 per cent in the 2000 regional election up to 87 per cent in 2005 and 2014 regional elections. The importance of preferential voting in Calabria is clearly visible by comparing its rates with those of Italy, which are significantly lower (between 26 and 34 points) for the whole 1970-2015 period. Conversely, Southern Italy shows similar patterns regarding the trend in preferential voting in regional elections, which is only slightly lower than the Calabrian ones (the discrepancy ranging between about 2 and 5 points over time). (Figure 1 about here) 2 Given the large extent to which the electoral behaviour in Calabria relies upon preferential voting, what we want to explore is whether personal vote is able to structure a system of interactions, acting at the level of individual candidates, and whether this candidate-based system replaces, either partially or totally, the system of interactions among parties, thus structuring a new form of political and electoral competition. In other words, we would argue that preferential voting in Calabria has not only the function of selecting representatives for public office within the classical party-based competition, but it has also started to shape a new form of systemness, replacing the declining systemness of the party system, which has increasingly become unpredictable. This new form of systemness is indeed based on the relations of cooperation and competition among the so-called ‘Lords of Preferences’, that is, politicians able to collect a large amount of votes thanks to personal networks, micro-distributive policies and so on. The intercandidate relations of cooperation and competition take primarily place in the electoral arena. They can involve the same candidate who changes party or coalition election after election, bringing his/her personal endowment of votes with him/her; an alliance among individual politicians who can switch from a centre-right to a centre-left national party and vice versa or even build new local electoral lists; endorsements given by politicians who have previously stood as candidates to freshly running candidates. This last point entails the transfer of blocks of personal votes from, say, candidate A to candidate B. This phenomenon takes place election after election and involves a number of candidates, regardless of their political origin. The structuring of this system of interactions at candidate level has important consequences in the shaping of the electoral supply, in the internal composition of the electoral alignments and in the final outcome of elections. The rationale behind this argument, which has to be empirically verified, is that parties are losing their historical functions of selecting candidates for public offices and competing for voters’ electoral support in favour of Lords of Preferences. These latter are endowed with a personal block of votes, concentrated in a specific territory. This block can be substantially retained even if the candidate changes party or coalition and regardless to the type of election and electoral system. Owning blocks of votes, Lords of Preferences exercise the lion’ share in pre-electoral bargaining with national or local parties, and their strength vis-à-vis party organizations (or what remains of them) is as higher as larger is their share of votes. Nonetheless, Lords’ power is not limited to pre-electoral arrangements with parties. Rather, moving from one coalition to another, or endorsing one candidate or another, they strongly influence electoral results. We argue that these considerations can find empirical support in several local contexts in Southern Italy (if not even in Southern Europe), in which the electoral environment presents features that are similar to the ones mentioned above. 3 In this article we will look at the specific case of Reggio Calabria province (and of its 97 municipalities). This choice has been driven by the availability of information that is needed to perform a reliable analysis of the territorial patterns of preferential voting and of the system of endorsements among candidates. In order to build such in-depth analysis of the local electoral context of the Reggio Calabria province (from now on RC), we have to rely on both qualitative and quantitative data. Specifically, we combine qualitative records of candidates’ endorsements, based on information retrieved in national and local media, with a quantitative analysis concerning the territorial patterns of support by Lords of Preferences and their ability to affect the electoral outcome in 2010 and 2014. 2. Personal Vote, Candidates’ Power and Electoral Accountability Calabria and more generally Southern Italy have had a specific development regarding both parties and their relationship with the electoral and social environment. Early works by Caciagli (1977), Belloni, Caciagli and Mattina (1979) and Chubb (1982) analysed the features of politics in Southern Italy, especially the wide diffusion of clientelism5. This phenomenon also affects several Southern European contexts (see for example Ignazi & Ysmal [1998]; Kopecky, Mair & Spirova [2012]). Many long-term explanations have been proposed to explain the inception and the persistence of such phenomenon, like long-term economic underdevelopment (Lemarchand 1977; Belloni, Caciagli & Mattina 1979), the late (absent) evolution of a politically organised working class, possibly in connection with economic backwardness (Lemarchand 1977; Chubb 1983) or the strong connection (if not overlapping) between patrons and the economically and socially dominant groups (Belloni, Caciagli & Mattina 1979). Regardless of the causes, a fundamental issue of clientelism is obviously its electoral aspect, that is, the exchange between various types of goods provided by the patron and the blocks of votes provided by the clients. How to detect such exchange? One of the most powerful tools is the analysis of personal vote and, regarding the Italian case, the analysis of preferential voting, which was an important tool of Italian national electoral law until 1993 and of current Italian regional and local electoral laws. Preferential voting has been widely used in Calabrian regional elections, as we have already shown in Figure 1, and several authors have written about the existence of ‘champions of preferences’ (De Luca 2001) or ‘grand’ electors (De Luca 2005), who are able to obtain a large number of preferential votes and also a noticeable percentage of their parties’ shares of votes (De Luca 2001 and 2005; Napoli 2005). Such ‘champions’ would be able, thanks to their patronage networks, to existing parties’ structures and, more generally, to the 4 connections with several local stakeholders (Napoli 2005; Fantozzi & Greco 2009), to influence the electoral competition, generally structured around a bipolar race between a centre-left and a centre-right coalition (Chiaramonte 2007). There is not a widespread consensus about the relationship among clientelism, Lords of Preferences and local parties. Earlier accounts (Caciagli 1977; Belloni, Caciagli & Mattina 1979) underlined the influence by the national party centre towards its Southern branches in which there are clientelistic relationships. Even more recent contributions do not totally rule out the role of parties in shaping Lords of Preferences’ interactions and clientelistic exchanges (De Luca 2005; Napoli 2005), even if the demise of mass party organisations has greatly weakened Southern Italian parties (Bova 2009). A quite interesting account of the role of parties in Southern Italy is given by Raniolo (2007), who - quoting von Beyme (1996) - writes about ‘omnibus parties’ in Sicily, that is, parties at disposal of politicians, without an integrative role with regards to the political system. We believe something similar, if not more pronounced, has been happening in Calabria. More specifically, we believe parties in Calabria resemble the model we could call parties as empty but communicating vessels. Building on the definition of ‘parties as empty vessels’ used in the literature – especially with regards to ‘American parties [that] are like two bottles on a shelf, one labelled [sic] ‘whiskey’ and the other labelled [sic] ‘milk’, but both empty’ (Katz & Kolodny 1994, p. 47) – we believe that the increasing fluidity of the electoral competition and the demise of party organisations have made parties not simply empty vessels, but also communicating ones. Politicians can decide to move from one vessel to another without particular problems, because of their share of personal vote that is substantially not related to the political label temporary attached to it. The changing relationship between parties and Lords of Preferences is not simply related to electoral politics. On the contrary, the declining role of political parties and the existence of patterns of interactions regarding Lords of Preferences have noticeable consequences for democracy at large. For instance, a first serious problem concerns accountability (as already underlined by Raniolo, [2009]). Parties, which normally represent cognitive shortcuts for lessinformed voters (Shively 1980), are not able to perform such task if they are just empty (albeit communicating) vessels, because they are strongly affected by Lords of Preferences’ behaviour and even their decisions are strongly influenced by such candidates. This means that voters may not use parties as a heuristic. This problem, which obviously is not only related to Calabria or to Southern Italy, must be seen in connection with the level of preferential voting in RC (and more generally in Calabria), in the sense that voters decide to support a candidate not to reward or punish his/her party, but for other reasons (including clientelistic ones). If parties are substantially weaker than before in the electoral competition, thanks to the reasons seen above, if 5 candidates, thanks to preferential voting, are very important in shaping the electoral competition and the electoral outcome and, finally, if candidates are able to move from one empty vessel to another one without suffering any serious electoral loss, it follows that electoral accountability is seriously blurred, if not missing. In fact, voters do not hold parties accountable for what they have done (or not done), because they are hardly recognizable and not able to carry any significant meaning beside their label; candidates are hardly accountable as well, since voters tend to follow Lords of Preferences regardless of their behaviour in office, apart from a possible retrospective evaluation based on the outcomes of micro-distributive policies. 3. Data Selection: the System of ‘Lords of Preferences’ How to empirically detect personal voting in Reggio Calabria? We have decided to focus on candidates for regional elections. Indeed, the electoral system for the regional elections in Calabria is based upon a PR system with a majority bonus granted to the coalition supporting the Presidential candidate who has obtained the highest number of votes. Specifically, the regional election in Calabria is based upon two separate arenas of competition: the first one is the majoritarian arena, where Presidential candidates compete for citizens’ vote; conversely, the second one is the proportional arena, where party lists compete for seats in the Regional Council. Therefore, the voter has at his/her disposal two votes: one to choose the Presidential candidate (plurality) and the other one for a party list. The voter can also express a preferential vote for one of the candidates running within the party list he/she has decided to support (open list system). By analysing preferential voting in regional elections, we are able to directly measure their electoral strength by considering the number and the distribution of preferences they obtain in such elections. Moreover, the diachronic element is fundamental to measure the resilience of candidates’ influence and the strength of their system of endorsements through time. Therefore, we have decided to focus on the regional elections of 2010 and 2014, that allow us to perform a valid and reliable empirical analysis, given the large amount of information provided by national and local media about candidates and their system of endorsements. We then empirically identify personal voting in the province of Reggio Calabria in the regional elections of 2010 and 2014. A first possible criterion to identify personal voting could simply be considering all preferences obtained by all candidates in 2010 and 2014. Yet, this would be misleading, since a candidate who obtains 10 preferences and a candidate who obtains 1,000 preferences are obviously different, in the sense that the latter is more able to mobilise a certain amount of support and to influence the electoral outcome. Indeed, our goal has been that of considering only relevant candidates, that is, those who can be defined as ‘Lords of Preferences’, 6 because of the alleged influence they can exert on the political system in RC. Following these considerations, we have selected candidates according to a quantitative criterion: having received at least 1 per cent of the total valid votes for the lists cast in RC in either the 2010 or the 2014 regional elections6. From a comparative viewpoint, the criterion we use is a very tight one, given that a threshold of 1 per cent of votes is usually applied to party systems when considering whether to include or not a party into a given analysis. The rationale behind this choice is that we want to select only those candidates able to collect such a relevant number of votes so that they could be considered as noticeable actors of the political system. From an empirical point of view, the 1 per cent threshold corresponds to 3,008 votes in 2010 and to 2,169 votes in 2014. We have selected 30 Lords of Preferences in 2010 and 29 Lords of Preferences in 2014, as shown in Table 1 and Table 2. (Table 1 about here) (Table 2 about here) Table 1 and 2 report the names of these Lords in 2010 and 2014, their party list and the votes they have received (absolute and percentage values). Looking at the table, some preliminary considerations can be drawn: as we can see, the large majority of those who stood as candidates in both 2010 and 2014 elections obtained 1 per cent of votes in both elections. The only exceptions are Giordano (IDV) and Nucera G. (PSI-Vendola) in 2010 and Mazza (Nuovo CDU) in 2014. Moreover, the total number of preferences obtained by the Lords of Preferences in 2010 is equal to 180,558 while in 2014 is equal to 145,589. This decrease is not fostered by a reduction in Lords’ ability to control the support of the electorate, but is mostly due to the relevant change in electoral turnout, which has fallen from 59.3 per cent in 2010 to 44.1 per cent in 2014 (and from 62.7 per cent to 44.6 per cent in RC province); indeed, by looking at the ratio between preferences and valid votes, in 2010, the 30 selected Lords collected 59.9 per cent of total votes cast in RC (with a mean share of 2 per cent), while in 2014 this percentage increases up to 66,9 per cent (with a mean share of 2.3 per cent). This means that the analysis of Table 1 and 2 could confirm our idea that Lords of Preferences increased their hold on the electorate between 2010 and 20147. This is supported not simply by the rate of preferential voting between 2010 and 2014, but also by the fact that a bigger percentage of personal vote is into the hands of a few candidates. For instance, in 2014 just the four most voted candidates (Romeo S., Irto, Battaglia M. and D’Agostino) received 19.7 per cent of the votes in RC. Indeed, in both elections between 60 and 70 per cent of the votes in RC were collected by a handful of politicians. 7 From this first empirical evidence, it is clear that personal vote has a large role in RC’s politics. But this is certainly not something new: as we have already stated, personal vote and clientelism have always been an important part of Southern Italian politics and from previous research we know that the electoral behaviour of Southern Italian voters is prominently ‘candidate-oriented’ (Fabrizio & Feltrin, 2007: 181). What could be a real novelty for the study of personal voting and party politics is to assess whether - and to what extent - candidates’ hold on the electorate and their interactions through time have replaced parties in a new candidate-based system. That is, we want to assess whether candidates’ territorial patterns of support in RC could represent an element of continuity within the political system which is able to explain and predict electoral results, contrasting with the increasing instability and unpredictability of the party system. In order to test this hypothesis, we analyse the patterns of territorial support of Lords of Preferences between 2010 and 2014 in RC. The empirical analyses are performed in the following sections and are based, for each Lord in each of the 97 municipalities of the RC province, on the ratio between his/her votes and the total electorate. Therefore, these municipalities constitute our unit of analysis. The choice to rely on the total electorate and not on the total number of valid votes has been done to avoid misleading results regarding Lords’ territorial patterns of support, since the level of electoral turnout between 2010 and 2014 is highly different even at municipalities’ level and, consequently, also the ratio between Lords’ votes and valid votes would have been biased by the shrinking of the denominator. On the contrary, the total electorate is a more stable measure that can be used to assess each Lord’s ability to mobilise his/her supporters over time. In order to detect the system of interactions among Lords of Preferences through time, it is necessary to consider, among the 59 Lords selected in Table 1 and 2, only those who stood as candidates in both 2010 and 2014 elections or those who stood as candidates in 2010 and endorsed other Lords in 2014. In particular, many different situations can emerge: the same Lord stands for election in 2010 and 2014, within the same party; the same Lord runs for election in 2010 and 2014 with different parties of the same coalition; the same Lord runs for elections with different coalitions; a Lord in 2010 runs in 2014 but does not become a Lord; a Lord in 2014 runs in 2010 but does not become a Lord; a Lord in 2010 endorses another Lord in 2014, belonging to the same party; a Lord in 2010 endorses another Lord in 2014, belonging to another party within the same coalition; a Lord in 2010 endorses another Lord in 2014, belonging to a different coalition. Figure 2 reports Lords’ re-candidacies and endorsements between 2010 and 2014, distinguishing among the possible situations we have just described. (Figure 2 about here) 8 As one can note, the 2010-2014 electoral cycle in RC offers a wide variability regarding Lords’ re-candidacies and endorsements. As regards re-candidacies, only in two cases (Nicolò Al. and Vilasi G.) the same Lord ran for election in the same party in 2010 and 2014 and in three cases (Minasi, Giordano and Nucera G.) the candidate changed party list but stayed within the same coalition. As many as four Lords changed their coalition between the two elections, moving from centre-right to centre (Fedele and Imbalzano) and from centre-right to centre-left (Mazza, Tripodi P.M.). Turning to the endorsements reported in Figure 2, based on national and local media’s information, we have found six verified endorsements8 between Lords in 2010 and Lords in 2014: once again, only one out of six has involved two Lords who run as candidates in the same party (Battaglia D. - Irto), while in the majority of cases9 the supported candidate belongs to a different party list with respect to the supporter’s one and even in one circumstance the endorsement is given to a candidate running with a different coalition, from centre-right (Bilardi, candidate in the Scopelliti Presidente) to centre (Arruzzolo, candidate in NCD). Overall, we have found 15 links, with nine re-candidacies and six endorsements. All the other Lords included in Table 1 and 2 have been excluded from the analysis, given that they did neither run in both elections nor endorse another candidate (or at least there is not reliable information about that). Do these links represent a ‘system’ - meant in terms of stability and predictability of interactions among Lords - that is alternative to the parties’ official one? Is this alternative candidate system able to explain the results of 2014 regional election in RC province? Moreover, can this alternative candidate system explain such results better than parties and their interactions? The next section will try to answer these questions. 4. Lords Versus Parties: Towards an Explanation We argue that in the RC context the combined effect of the growing unpredictability in the interactions among parties and voters (that is, high levels of party replacement between elections and volatile electoral environment) and the pervasiveness of candidate-oriented voting has changed the nature of the competition from a party-based system to a candidate-based system. In order to test this hypothesis we need to verify whether there is a certain degree of continuity in the patterns of territorial support held by Lords of Preferences over time. The presence of such continuity is tested through a series of correlations between the hypothesised links among Lords. 9 We expect that if the competition in RC is highly driven by personal vote, there will be a high level of association in the territorial patterns of support among Lords over time. This is because each Lord is able either to retain his/her block of preferential votes in subsequent elections or to transfer his/her endowment to another candidate. Conversely, in a party-driven politics - the ordinary model of competition in Western Europe - parties structure the electoral competition, establishing ties with voters and, consequently, they show high levels of territorial continuity of support, while individual candidates only play a minor role, which is highly dependent on parties’ performance. More specifically, in a traditional party driven politics, where the candidate is only a function of the party in which he/she runs, one would expect higher correlations over time for candidates who run at t and t+1 with the same party label and gradually lower correlations for candidates who run from time to time with a different party label, with a different coalition, and finally for candidates who endorse another candidate. Regarding endorsements, one would expect to find higher correlations among candidates running for the same party list and gradually lower for candidates running with different parties or even with different coalitions. (Figure 3 about here) The left part of Figure 3 shows the results of bivariate correlations (Pearson’s r coefficient) regarding the votes on electorate ratios, across the 97 municipalities of the RC province, among the 15 hypothesised links of Lords previously illustrated in Figure 2. We note that 11 out of 15 hypothesised correlations show a significant coefficient, most of the times at the best level of confidence (p<.001). Overall, the average level of correlation among Lords is .43, with no significant differences among the various subtypes we have previously identified: for instance, there is an impressive level of correlation among some endorsements, and in particular in the case of the Crinò brothers (.91), while on the contrary Vilasi G., despite running in both elections with the same party, shows a non-significant correlation between 2010 and 2014. Moreover, by enlarging the scope of the empirical analysis and correlating all the 59 Lords selected according to our criteria, the overall mean of correlation is almost zero (.04), with 870 observations, and this shows that the correlations among Lords resulting from the analysis shown in Figure 3 are not casual. This analysis shows two important findings. The first one is that personal vote for Lords of Preferences in RC shows a certain degree of continuity and predictability over time, with candidates who are able to maintain a similar territorial configuration of their support across subsequent elections. The second finding rejects the above-mentioned hypothesis regarding the decreasing level of continuity in the ability of Lords to maintain a similar pattern or support as far 10 as they change party, coalition or even transfer their votes to other candidates. Correlations show instead that the level of continuity among Lords’ territorial patterns of support is not driven by partisan links and also occur if the candidate stands again for election or if he/she decides to endorse other candidates. However, simply showing that personal vote exerts a large role in RC politics and holds a certain degree of continuity over time does not mean that party politics is without importance per se. It could be that personal vote is indeed important, but within a party system that holds a higher degree of continuity and predictability over time. Nonetheless, the right part of Figure 3 clearly rejects this scenario. This part of the Figure reports the results of the correlations among the ratios between votes and total electorate across the 97 municipalities of RC as regards the party lists for which we can hypothesise an organisational continuity between 2010 and 201410. There are nine couples of party lists which can be considered as having an organizational continuity: three parties which present the same list in both elections (Autonomia e Diritti, PD and UDC), a list that is a direct expression of PD (Oliverio Presidente), other three directly deriving from PDL (FI, NCD, FDI-AN), and other two couples of leftists lists (PSI-Vendola and La Sinistra; PRC-PDCI and L’Altra Calabria). Only four couples out of nine show a significant degree of correlation in their territorial patterns of support between 2010 and 2014 (PD-PD, PDL-FI, PDL-NCD and PRC-PDCI-L’Altra Calabria), while the other couples, including a national party like UDC, show completely independent patterns of support through time. Overall, the mean of the hypothesised correlations among party lists is barely .18, really lower than the mean coefficient found as regards Lords. Moreover, the difference between the nine hypothesised correlations among parties and the overall mean of correlations among all the party lists running in 2010 and 2014 (240 observations) is very low (.11). This means that the hypothesised correlations among parties are not significantly different from the correlations among the whole sample of parties, and this can be considered as a piece of evidence reinforcing the idea that the electoral competition in RC is driven by the Lords of Preferences rather than parties, whose patterns of support are not far from being random. Moreover, the analyses reported in Figure 3 constitute a first element supporting our idea that in RC there is an on-going shift from a party-based system to a candidate-based system. After having shown, through the analysis of the territorial patterns of support, that Lords’ recandidacies and endorsements grant a certain level of continuity and predictability which replaces the uncertainty of the system at the party level, we need to see whether such links among Lords are able to influence the electoral outcome. In other words, in order to go on with the comparison between Lords and parties, we have to assess whether the ‘system’ of Lords is able to predict the 2014 election results in RC better than traditional parties and their alliances. 11 The easiest way to perform such analysis is to build simple bivariate regression models in which the dependent variable is, each time, the electoral performance of the three main Presidential candidates in 2014 in RC (or, alternatively, the electoral performance of the coalition of parties supporting them) and the independent variables are the sum of votes on electorate ratio in 2010 of either the coalition of parties or the coalition of Lords supporting them in 2014. The rationale behind these models is to assess whether and to what extent Lords of Preferences are able to determine the success or the failure of a Presidential candidate (or of the coalition supporting him/her) and compare this result with the explanatory capacity of parties and their alliances. In a traditional party-based politics, we would expect that the electoral result of a Presidential candidate (or of his/her coalition) at time t is predicted, to a large extent, by the electoral results of parties supporting him/her at time t-1. Conversely, in a candidate-based politics, we would expect that Lords of Preferences are better predictors of electoral results than parties, which are conceived as empty (but communicating) vessels and therefore not able to significantly structure the electoral competition. (Table 3 about here) Table 3 reports the results of a series of OLS regressions where the votes for a Presidential candidate in 2014 in RC are predicted from different aggregations of coalition of parties and Lords in 2010 11 . Three models are tested, concerning the three most voted Presidential candidates in 2014 in RC: Oliverio (centre-left), Ferro (centre-right) and D’Ascola (centre). For each Presidential candidate, we test the impact of different independent variables. The first one is the sum of the votes on electorate ratios across the 97 municipalities of RC of the centre-left coalition in 2010 (in Oliverio’s model) and of the centre-right coalition in 2010 (in Ferro’s model)12. The second one is the sum of the votes on electorate ratios in 2010 of the parties supporting respectively Oliverio, Ferro and D’Ascola in 2014. To these two independent variables, which aim at testing the explanatory capacity of parties and their alliances from two different perspectives, we have added a third independent variable, which measures the impact of the ‘coalitions’ of Lords supporting the different Presidential candidates in 2014. In particular, the variable has been constructed by aggregating the votes on electorate ratios in 2010 across the 97 municipalities of RC of the Lords of Preferences supporting, respectively, Oliverio, Ferro and D’Ascola in 2014. Further specifications on the way these variables have been built can be found in the note reported below Table 3. The empirical results of the regression models are absolutely striking and confirm our expectations as regards the ability of Lords and their links over time to predict the electoral 12 results. Moreover, the model goes further in showing that the overall explanatory power of Lords is higher than that of parties and their alliances. As one can see by looking at the b coefficients, as well as at the R-squared coefficients, in each of the three models the explanatory capacity of the Lords is substantially higher than that of parties, both considering them in the 2010 coalitions or through the re-aggregation of these coalitions according to the 2014 structure of competition. As concerns Oliverio’s vote, the percentage of variance explained by the centre-left coalition in 2010 is barely .19 and that explained by the re-aggregation of this coalition according to 2014 is .16. On the contrary, the percentage of variance explained by the votes received in 2010 by the ‘coalition’ of Lords supporting Oliverio in 2014 is as high as .35. In the centre-right model, the difference between parties and Lords is somewhat smaller but still noticeable (respectively, .19, .21 and .26). Finally, in the D’Ascola model the partisan variable is totally non-significant, while the Lords’ one shows a coefficient that is significant at the highest level of confidence and it is able to explain, to a certain extent, a part of the variance of the dependent variable (.17). 5. Conclusions: Towards a Candidate System? In this article we have analysed the impact of personal voting in the Reggio Calabria province on the regional elections of 2010 and 2014. The main novelty of our research is having built an original empirical model concerning ‘Lords of Preferences’ in Reggio Calabria and having tested their ability to shape electoral competition and to predict electoral results better than parties and their alliances. Starting from an overview of the Calabrian electoral context, characterized by high levels of instability and unpredictability regarding both the changing electoral supply and the fluid links between voters and parties, we have focused on what it seems to be the only element of continuity within the Calabrian political system: personal vote expressed through preferential voting. We have identifies the so-called ‘Lords of Preferences’, that is, those candidates receiving at least 1 per cent of the total valid votes obtained by the lists in which they were running in 2010 or in 2014. Among these Lords of Preferences, we have selected those who either contested both elections or contested the 2010 election and endorsed another candidate in the 2014 election. This kind of data selection has allowed us to test the hypothesis of strong continuity in the territorial patterns of preferential voting between elections and to compare the results with those emerging from the analysis of the territorial patterns of support received by parties. The analysis has been performed through bivariate correlations of, respectively, Lords’ and parties’ support across the 97 municipalities of the Reggio Calabria province between 2010 and 2014 and has brought about three important findings. The first one is that personal vote for 13 Lords of Preferences in RC shows a certain degree of continuity and predictability over time, with candidates who are able to maintain a similar territorial configuration of their support over time. The second finding is that such continuity and predictability is found regardless of the specific partisan label and alignment in which candidates run time by time. The third finding shows that the continuity and predictability of personal vote retained by Lords of Preferences is by far higher than those concerning parties’ territorial patterns of support. The subsequent step has been that of verifying whether Lords’ system of re-candidacies and endorsements is able to predict the electoral results of 2014 and to do so better than parties and their alliances. We have tested these hypotheses through a series of bivariate regression models, where the dependent variable – the electoral performance of the three main Presidential candidates in 2014 in RC - is predicted, respectively, by the electoral performance of parties in the 2010 election, by the electoral performance of coalitions in the 2010 election and by the electoral performance in 2010 of the group of Lords of Preferences supporting each Presidential candidate in 2014. The results are impressive, confirming our idea about the ability of Lords and their links over time to predict the electoral results better than parties and their coalitions. Concluding, this analysis has revealed that in Calabria there has been a deep change involving the overall structure of competition: individual candidates, endowed with a personal support and with the ability to retain it over time or to transfer it to other candidates, have basically replaced parties and their alliances as the core element granting predictability to the system. While in traditional party-based politics parties have the function to recruit candidates for public offices and to compete for the electoral support, in Calabria it seems there has been a real overturning, bringing stronger candidates to decide what to do with their block of votes: running again or endorsing another candidate? Running in the same party of the last election or changing party or even coalition? It is by such decisions, taken by Lords of Preferences, that the electoral result is determined. The consequences of this system on the quality of democracy and the electoral accountability, in terms of voters’ ability to punish or reward those in power, can be noticeably negative. Further research is needed to explore whether such system concerning Lords of Preferences and their interactions is equally strong in other Southern Italian regions or in other Southern European contexts. Acknowledgments The authors thank Alessandro Chiaramonte, Lorenzo De Sio and Nicola Martocchia Diodati for useful feedbacks on earlier drafts of this article. 14 Notes 1 Detailed data about the electoral results of Calabrian regional elections of 2010 and 2014 can be found on the Italian Ministry of Interior’s webiste (http://elezionistorico.interno.it ). 2 The 16 lists contesting the 2010 regional election in Calabria were: for the centre-right coalition (supporting the Presidential candidate Scopelliti), Popolo della Libertà (PDL), Scopelliti Presidente (Scopelliti Pres.), Unione di Centro (UDC), Insieme per la Calabria (Insieme Calabria), Socialisti Uniti-PSI, Libertà e Autonomia Noi Sud (Noi Sud), Fiamma Tricolore-Destra Sociale; for the centre-left coalition (supporting the Presidential candidate Loiero), Partito Democratico (PD), Autonomia e Diritti (Aut. e Diritti), Rifondazione Comunista-Sinistra Europea-Comunisti Italiani (RC-PDCI), PSISinistra con Vendola (PSI-Vendola), Alleanza per la Calabria (Alleanza Calabria), Slega la Calabria (Slega Calabria); for the other leftist coalition (supporting the Presidential candidate Callipo), Italia dei Valori (IDV), Io Resto in Calabria con Callipo, Lista Pannella-Bonino. Conversely, the 15 lists contesting the regional election in Calabria in 2014 were: for the centre-left coalition (supporting the Presidential candidate Oliverio), Partito Democratico (PD), Oliverio Presidente (Oliverio Pres.), Democratici Progressisti (Dem. Progr.), Calabria in Rete-Campo Democratico (Calabria in Rete), La Sinistra con Speranza (La Sinistra), Autonomia e Diritti (Aut. e Diritti), Centro Democratico (Centro Dem.), Nuovo CDU; for the centre-right coalition (supporting the Presidential candidate Ferro), Forza Italia (FI), Casa delle Libertà (CDL), Fratelli d’Italia-Alleanza Nazionale (FDI-AN); for the centre coalition (supporting the Presidential candidate D’Ascola), Nuovo Centro Destra (NCD), Unione di Centro (UDC); finally, Movimento Cinque Stelle (M5S) and L’Altra Calabria ran separately with their own Presidential candidate. 3 The calculation of the index has followed Bartolini and Mair’s (1990) rules as regards the aggregation of splits and mergers. As a consequence, we have computed electoral volatility following the criterion of organizational continuity between party lists: PDL in 2010 has been compared with the aggregation of the three parties resulting from its split: FI, NCD and FDI-AN; PD in 2010 has been compared with the aggregation of PD and Oliverio Presidente (the list launched by the PD candidate for the Presidency Mario Oliverio); PSI-Sinistra con Vendola and IDV in 2010 formed the list La Sinistra con Speranza in 2014; RC-PDCI in 2010 formed Altra Calabria in 2014; Autonomia e Diritti and UDC run with the same label in both elections. All the other lists have been treated as new parties. 4 By non-national parties we have considered those party lists whose labels do not correspond to national parties’ ones. Specifically, there have been 6 non-national parties in 2010 (Scopelliti Presidente, Libertà e Autonomia Noi Sud, Autonomia e Diritti, Alleanza per la Calabria, Slega la Calabria, Io Resto in Calabria con Callipo) and 7 in 2014 (Oliverio Presidente, Democratici Progressisti, Calabria in Rete-Campo Democratico, La Sinistra con Speranza, Autonomia e Diritti, Nuovo CDU, Casa delle Libertà). 5 Several definitions of clientelism have been proposed (Lemarchand, 1977, p.15; Belloni, Caciagli & Mattina, 1979, p. 273; Piattoni 2001). There are many important elements emerging from such definitions, such as the unequal balance of power between the patron and the clients, the exchange of goods and services between the patron and clients and the difference between the selective incentives provided by patron and the electoral support given to him/her by clients. 6 Notice that also De Luca (2005) and Napoli (2005) use a quantitative criterion to select ‘champions of preferences’: having received a high number of preferential votes. Nonetheless, they do not specify the share of votes necessary for a candidate to be qualified as a ‘champion’. 7 Notice that between 2010 and 2014 the total number of seats to be allocated in the Calabria’s Regional Council decreased 50 from to 30. As a consequence, in 2014 the voter was faced with a smaller number of candidates in the 15 ballot paper (the total number of candidates in RC fell from 171 to 104) and this may partly explain the larger concentration of support for the most voted candidates in 2014. 8 Regarding Battaglia D. (2010) endorsing Irto (2014), see Cilione (2014), Corriere della Calabria (2014), Tripepi (2014b), Quotidiano del Sud (2014c). Since the articles connect Irto with many politicians, we have decided to rely on a conservative criterion, only considering Battaglia D. as endorser, being the person who always appears in the articles we have considered. Regarding Bova (2010) endorsing D’agostino (2014), see Galullo (2014) and Quotidiano del Sud (2014d). Regarding Crinò P. (2010) endorsing Crinò F. (2014), see Albanese (2014). Regarding Caridi (2010) endorsing Cannizzaro (2014) see Gemelli (2014), Cronache del Garantista (2014) and Quotidiano del Sud (2014b). Regarding De Gaetano (2010) endorsing Romeo S. (2014), see Tripodi (2014), Tripepi (2014a) and Quotidiano del Sud (2014e). Finally, on Bilardi (2010) endorsing Arruzzolo (2014), see Quotidiano del Sud (2014a). 9 As the reader can notice, one case involves Pietro Crinò, a Lord in 2010 who has decided not to run in the following regional election and has instead endorsed his brother Franco in 2014. 10 The choice regarding which party lists to correlate is consistent with the calculation of electoral volatility we have reported in the Introduction. 11 We have also performed an identical statistical analysis using as a dependent variable, each time, the electoral performance of the three coalitions of parties supporting the three main Presidential candidates in 2014 in RC. The results are extremely similar to those reported in Table 3, and this is not surprising, given that the electoral results of the Presidential arena and of the proportional arena have been almost identical. For this reason, we have decided not to report the results of these additional regression models. All the analyses have been performed by using STATA 12. 12 As regards D’Ascola, there was not a centre-leaning electoral coalition in 2010 and therefore this independent variable is absent in D’Ascola model. 16 References Albanese, P. (2014) ‘Un vero piccolo esercito di candidati dalla Locride’, Quotidiano del Sud, 26 October, p. 17. Bardi, L. & Mair, P., (2008) ‘The Parameters of Party Systems’, Party Politics, vol. 14, no. 2, pp. 147-166. Bartolini, S. & Mair, P. (1990) Identity, Competition and Electoral Availability: The Stabilisation of European Electorates 1885-1985, ECPR Press, Colchester. Belloni, F., Caciagli, M. & Mattina, L. (1979) ‘The Mass Clientelism Party: The Christian Democratic Party in Catania and in Southern Italy’ European Journal of Political Research, vol. 7, no. 2, pp. 253-275. Birch, S. (2003) Electoral Systems and Political Transformation in Post-communist Europe, Palgrave MacMillan, New York. Bova, V. (2009) ‘Il Mezzogiorno nelle trasformazioni del sistema politico nazionale’, in Legalità manipolazione democrazia, ed A. Costabile, Carocci, Roma, pp. 159-186. Caciagli, M. (1977) Democrazia Cristiana e Potere nel Mezzogiorno: il sistema democristiano a Catania, Guaraldi, Rimini Casal Bértoa, F. (2014) ‘Party systems and cleavage structures revisited: A sociological explanation of party system institutionalization in East Central Europe’, Party Politics, vol. 20, no. 1, pp. 16-36. Chiaramonte, A., (2007) ‘Il nuovo sistema politico italiano tra bipolarismo e frammentazione’, in Proporzionale ma non solo. Le elezioni politiche del 2006, eds R. D’Alimonte & A. Chiaramonte, il Mulino, Bologna, pp. 369-406. Chiaramonte, A. (2010) ‘Il voto di preferenza nelle elezioni regionali 2010: quanto, dove e chi’, paper presented at the 2010 Italian Political Science Association Conference, Venice, 16-18 September. Chiaramonte, A. & Emanuele, V. (2015) ‘Party System Volatility, Regeneration and DeInstitutionalization in Western Europe (1945-2015)’, Party Politics, Online First, pp. 1-13, DOI:10.1177/1354068815601330. Chubb, J. (1982) Patronage, Power and Poverty in Southern Italy: A Tale of Two Cities, Cambridge University Press, Cambridge. Cilione, G. (2014) ‘Irto incontra i cittadini jonici’, Il Quotidiano del Sud, 16 November, p. 24. Corriere della Calabria (2014) ‘Economia reggina, confronto imprenditori-Oliverio’, 20 November, available online at: 17 http://www.corrieredellacalabria.it/index.php/politics/item/27546-confronto-con-oliveriosui-temi-dell-economia-reggina . Cronache del Garantista (2014) ‘Si salvano solo Nicolò (FI) and Cannizzaro (CdL)’, 25 November, p. 19. De Luca, R. (2001) ‘Il ritorno dei «campioni delle preferenze» nelle elezioni regionali’, Polis, vol. 15, no. 2, pp. 227-245. De Luca, R. (2005) ‘La campagna elettorale permanente dei “campioni delle preferenze” – Le elezioni regionali del 2000 in Calabria’, in Elezioni e Personalizzazione della Politica, ed F. Venturino, Aracne, Roma, pp. 159-178. Fantozzi, P. & Greco, W. (2009) ‘Politica locale e ceto politico regionale nel Sud d’Italia’ in Legalità, manipolazione, democrazia, ed A. Costabile, Carocci, Roma, pp. 105-138. Fabrizio, D. & Feltrin, P. (2007) ‘L’uso del voto di preferenza: una crescita continua’, in Riforme istituzionali e rappresentanza politica nelle Regioni italiane, eds A. Chiaramonte, A. & G. Tarli Barbieri, Il Mulino, Bologna, pp. 175-199. Galullo, R. (2014) ‘Elezioni in Calabria: da Magarò a Belcastro decine di cambi di casacca dietro il blocco di potere a sostegno di Mario Oliverio (Pd)’, 21 November, available online at: http://robertogalullo.blog.ilsole24ore.com/2014/11/21/elezioni-in-calabria-da-magaro-abelcastro-decine-di-cambi-di-casacca-dietro-il-blocco-di-potere-a-sostegno-di-mario-oliveriopd/?refresh_ce=1 . Gemelli, B. (2014) ‘Alla Ferro mancano i piccoli’, Quotidiano del Sud, 26 October, p. 4. Huntington, S. (1968) Political Order in Changing Societies, Yale University Press, New Haven. Ignazi, P. & Ysmal, C. (1998) The Organization of Political Parties in Southern Europe, Praeger Publishers, Westport and London. Katz, R.S. & Kolodny, R. (1994) ‘Party Organization as an Empty Vessel: Parties in American Politics’, in How Parties Organize - Change and Adaptation in Party Organization in Western Democracies, eds R.S. Katz & P. Mair, Sage, London and Thousand Oaks, pp. 23-50. Kopecky, P., Mair, P. & Spirova, M. (2012) Party Patronage and Party Government in European Democracies, Oxford University Press, Oxford. Lemarchand, R. (1977) ‘Comparative Political Clientelism: Structure, Process and Optic’, in Political Clientelism, Patronage and Development, eds S.N. Eisenstadt & R. Lemarchand, Sage, Beverly Hills and London, pp. 7-32. Mainwaring, S. & Torcal, M. (2006) ‘Party System Institutionalization and Party System Theory after the Third Wave of Democratization’, in Handbook of Political Parties, eds R.S. Katz & W. Crotty, Sage, London, pp. 204–227. 18 Napoli, D. (2005) Dove le preferenze contano – Il caso della Calabria, Le Istituzioni del Federalismo vol. 6, pp. 1142-1182. Pedersen, M.N. (1979) ‘The Dynamics of European Party Systems: Changing Patterns of Electoral Volatility’, European Journal of Political Research, vol. 7, no. 1, pp. 1-26. Piattoni, S. (2001) Clientelism, Interests and Democratic Representation: The European Experience in Historical and Comparative Perspective, Cambridge University Press, Cambridge. Raniolo, F. (2007) ‘Sindaci forti, consiglieri mobili, partiti omnibus. Sul ciclo politico-elettorale siciliano (1988-1998)’, Quaderni di Sociologia, vol. 43, pp. 31-62. Raniolo, F. (2009) ‘Sui sovvertimenti della qualità democratica nel Mezzogiorno’, in Legalità manipolazione democrazia, ed A. Costabile, Carocci, Roma, pp. 187-224. Quotidiano del Sud (2014a) ‘Giovanni Arruzzolo’, 26 November, p. 10. Quotidiano del Sud (2014b) ‘Francesco Cannizzaro’, 26 November, p. 10. Quotidiano del Sud (2014c) ‘Nicola Irto’, 26 November, p. 11. Quotidiano del Sud (2014d) ‘Francesco D’Agostino’, 26 November, p. 11. Quotidiano del Sud (2014e) ‘Seby Romeo’, 26 November, p. 11. Sanchez, O. (2009) ‘Party Non-Systems - A Conceptual Innovation’, Party Politics, vol. 15, no. 4, pp. 487-520. Sartori, G. (2005) Parties and Party Systems, ECPR Press, Colchester. Shiveley, W.P. (1980) ‘The nature of party identification: a review of recent developments’, in The Electorate Reconsidered, eds J.C. Pierce & J.L. Sullivan, Sage, Beverly Hills and London, pp. 219236. Scaramozzino, P. (1990) ‘Il Voto di Preferenza Nelle Elezioni Regionali’, Il Politico, vol. 55, no. 2, pp. 293-311. Tripepi, R. (2014a) ‘Oliverio abbraccia De Gaetano e dà l’ok alla corsa di Romeo’, Cronache del Garantista, 29 October, p. 4. Tripepi. R. (2014b) ‘È Mimmetto Battaglia il campione di preferenze’, Cronache del Garantista, 25 November, p. 19. Tripodi, C. (2014) ‘Eclissi da regionali: caos nei partiti’, Quotidiano del Sud, 26 October, p. 17. Valbruzzi, M. & Vignati, R. (2014) ‘Un dato inaspettato: cala il tasso di preferenza’, Il Mulino, vol. 4, pp. 687-692. Von Beyme, K. (1996) Transition to Democracy in Eastern Europe, Palgrave MacMillan, London. 19 List of Tables and Figures Figure 1 Rate of Preferential Voting in Italian Regional Elections, 1970-2015 100 90 82.3 80 83.1 70 60 50 40 30 87.4 54.7 54.9 58.5 61.3 61.5 48.8 48.5 44 22.5 77.1 48.6 49.5 26 25.8 52.3 27.4 55.6 29.6 54.2 84.1 86.8 81.9 81.2 50.9 50.5 2010 2015 53.1 30.4 Calabria Southern Italy 20 Italy 10 0 1970 1975 1980 1985 1990 1995 2000 2005 Note: data for the 1970-1990 period come from Scaramozzino (1990); data for the 1995-2010 come from Chiaramonte (2010); data for the 2013-2014 come from Valbruzzi and Vignati (2014). Data on 2015 are authors’ elaboration on official electoral results. Notice that during the 1970-1990 period the rate of preferential voting could be slightly underestimated due to the possibility of casting two or three preferences for the voters in certain provinces. Further notice that the line representing Southern Italy includes Campania, Basilicata, Puglia and Calabria. Moreover, the 2015 points in the chart include the last election held between 2013 and 2015 in the different region considered. Finally, ‘Italy’ includes data concerning ‘ordinary statute’ regions. 20 Table 1 List of Lords of Preferences in the 2010 Regional Election (Province of RC): Names, Party List and Votes Received (Absolute and Percentage Values) Lords of Preferences in 2010 Name Party list Votes Alvaro PSI-Vendola 3237 Battaglia D. PD 9710 Biasi PDL 5438 Bilardi Scopelliti Pres. 8123 Bova PD 8770 Caridi PDL 11215 Cherubino PDL 6546 Crinò Scopelliti Pres. 4663 Cuzzola Aut. e Diritti 2994 De Gaetano PRC-PDCI 8765 Fedele PDL 7671 Giordano IDV 2279 Imbalzano Scopelliti Pres. 4780 Managò Noi Sud 3367 Mazza UDC 3104 Minasi PDL 6756 Naccari Carlizzi PD 7384 Nicolò Al. PDL 8082 Nicolò An. Scopelliti Pres. 3291 Nucera G. PSI-Vendola 1138 Nucera G. PDL 7717 Racco PD 3196 Rappoccio Insieme Calabria 3726 Raso Scopelliti Pres. 4305 Sarica PDL 4305 Tripodi M. PRC-PDCI 5465 Tripodi P. M. UDC 10393 Vilasi G. PDL 6867 Zappalà PDL 11078 Zito PDL 6193 Mean 6019 Total 180558 Valid share 1.1 3.2 1.8 2.7 2.9 3.7 2.2 1.5 1 2.9 2.5 0.8 1.6 1.1 1 2.2 2.5 2.7 1.1 0.4 2.6 1.1 1.2 1.4 1.4 1.8 3.5 2.3 3.7 2.1 2.0 59.9 21 Table 2 List of Lords of Preferences in the 2014 Regional Election (Province of RC): Names, Party List and Votes Received (Absolute and Percentage Values) Name Amodeo Arruzzolo Bagnato Battaglia M. Bombardieri Cannizzaro Creazzo Crinò D'Agostino Fedele Giannetta Giordano Imbalzano Irto Longo Mallamaci Mazza Minasi Nastasi Neri Nicolò Al. Nucera G. Pedà Pirrotta Romeo D. Romeo S. Sciarrone Tripodi P. M. Vilasi G. Mean Total Lords of Preferences in 2014 Party list Votes Centro Dem. 3404 NCD 5920 Aut. e Diritti 3357 PD 10450 PD 2252 CDL 6109 Democr. Progr. 3692 CDL 3492 Oliverio Pres. 7942 NCD 4840 FI 4084 PD 6428 NCD 3666 PD 12014 La Sinistra 2325 Oliverio Pres. 7158 Nuovo CDU 1792 CDL 4391 Democr. Progr. 2920 Democr. Progr. 5000 FI 7046 La Sinistra 3315 CDL 5009 FI 2289 CDL 3178 PD 12288 Calabria in Rete 2414 Centro Dem. 6120 FI 2694 5020 145589 Valid share 1.6 2.7 1.5 4.8 1 2.8 1.7 1.6 3.7 2.2 1.9 3 1.7 5.5 1.1 3.3 0.8 2 1.3 2.3 3.2 1.5 2.3 1.1 1.5 5.7 1.1 2.8 1.2 2.3 66.9 22 Figure 2 Lords’ Re-Candidacies and Endorsements Between 2010 and 2014 Lords in 2010 and 2014 Lords in 2010 who endorsed other Lords in 2014 Sam e p arty Nicolò Al. (PDL) Vilasi G. (PDL) Sam e p arty Battaglia D. (PD) (FI) (FI) Irto (PD) Dif f e re n t p arty b u t s am e c o alitio n Minasi (PDL) (CDL) Giordano (IDV) (PD) Nucera G. (PSI-Vendola) (La Sinistra) Dif f e re n t p arty b u t s am e c o alitio n Bova (PD) D'Agostino (Oliverio Pres.) Crinò P. (Scopelliti Pres.) Crinò F. (CDL) Caridi (PDL) Cannizzaro (CDL) De Gaetano (PRC-PDCI) Romeo S. (PD) Dif f e re n t c o alitio n Fedele (PDL) (NCD) Imbalzano (Scopelliti Pres.) (NCD) Mazza (UDC) (Nuovo CDU) Tripodi P. M. (UDC) (Centro Dem.) Dif f e re n t c o alitio n Bilardi (Scopelliti Pres.) Arruzzolo (NCD) Note: ‘Same party’ means same party list (e.g. PD) or a party list directly resulting from the former one (e.g., FI from PDL). Despite NCD is a splinter party from PDL, Fedele has been included in the subset ‘Different coalition’ and not in the subset ‘Same party’ since NCD did not support the center-right coalition in the 2014 regional election. 23 Figure 3 Lords and Parties’ Territorial Patterns of Support: Bivariate Correlations (2010-2014) Hypothesised correlations (2010-2014) Lords' candidacies and endorsements Pearson's r Party lists Nicolò Al. (PDL) Aut. e Diritti (FI) .35*** .14 PD PD (CDL) .59*** PD Oliverio Pres. (FI) Vilasi G. (PDL) Minasi (PDL) Aut. e Diritti .18 .41*** .10 .42 .17 PDL FI .34*** PDL NCD Fedele (PDL) (NCD) Imbalzano (Scopelliti Pres.) (NCD) Mazza (UDC) (Nuovo CDU) .69*** .10 .88*** PDL FDI-AN PSI-Vendola La Sinistra RC-PDCI Altra Calabria .31** .06 -.13 Tripodi P. M. (UDC) .24* UDC Giordano (IDV) (PD) Nucera G. (PSI-Vendola) Battaglia D. (PD) Bova (PD) Caridi (PDL) *** Pearson's r (La Sinistra) (Centro Dem.) Irto (PD) *** .56 D'Agostino (Oliverio) .44*** Cannizzaro (CDL) .54*** Crinò P. (Scopelliti Pres.) De Gaetano (PRC-PDCI) Bilardi (Scopelliti Pres.) UDC .22* .11 Crinò F. (CDL) .91*** Romeo S. (PD) .30** Arruzzolo (NCD) .11 Mean of hypothesised .18 correlations (N:9) Overall mean of correlations among Lords Overall mean of correlations (N: 870) .04 among parties (N: 240) .07 Note: bivariate correlations (candidates/parties' votes on electorate ratio) across the municipalities of RC (N:97); * p < .05, ** p < .01, *** p < .001. Mean of hypothesised correlations (N:16) .43 24 Table 3 Votes for Presidential Candidates in 2014 Predicted From Different Aggregations of Coalitions of Parties and Lords in 2010 Independent variable Centre-left coalition in 2010 Parties in 2010 supporting Oliverio in 2014 Lords in 2010 supporting Oliverio in 2014 Constant N R-squared Adj. R-squared F statistic Centre-right coalition in 2010 Parties in 2010 supporting Ferro in 2014 Lords in 2010 supporting Ferro in 2014 Constant N R-squared Adj. R-squared F statistic Parties in 2010 supporting D'Ascola in 2014 Lords in 2010 supporting D'Ascola in 2014 Constant N R-squared Adj. R-squared F statistic b 0.72*** 16.09*** 97 0.19 0.18 22.42*** b 0.30*** Model 1 (Oliverio) s.e. b s.e. b 0.15 0.69*** 0.16 0.88*** 2.10 17.47*** 2.01 18.46*** 97 97 0.16 0.35 0.15 0.34 18.09*** 50.83*** Model 2 (Ferro) s.e. b s.e. 0.06 0.38*** 0.08 1.24 1.93 1.93 97 97 0.19 0.21 0.19 0.20 *** 22.88 24.89*** b -0.07 3.34*** 97 0.01 0.00 1.17 b 0.57*** 1.71 6.37*** 97 0.26 0.25 32.43*** s.e. 0.12 1.18 s.e. 0.10 0.85 Model 3 (D'Ascola) s.e. b s.e. 0.06 0.48*** 0.11 0.46 1.45*** 0.42 97 0.17 0.16 19.74*** Note: OLS regression with unstandardized coefficients (b) and standard errors (s.e.); * p < .05, ** p < .01, *** p < .001. As regards the construction of the independent variables, PRC-PDCI has been excluded from the centre-left coalition in 2010, since the party supported another Presidential candidate in 2014. The same applies to UDC in 2010 for the centre-right coalition. ‘Parties in 2010 supporting Oliverio in 2014’ includes: PD, PSI-Vendola and Autonomia e Diritti. ‘Parties in 2010 supporting Ferro in 2014’ includes: PDL and Scopelliti Presidente. ‘Parties in 2010 supporting D’Ascola in 2014’ includes UDC. ‘Lords in 2010 supporting Oliverio in 2014’ includes: Mazza, Tripodi P.M., Giordano, Nucera G., Bova, De Gaetano and Battaglia D. ‘Lords in 2010 supporting Ferro in 2014’ includes: Minasi, Nicolò Al., Vilasi G., Caridi, and Crinò. ‘Lords in 2010 supporting D’Ascola in 2014’ includes: Fedele, Imbalzano and Bilardi. 25