From the SelectedWorks of Riccardo Pelizzo. Party Direction. riccardo pelizzo. Available at:

Transcription

1 From the SelectedWorks of Riccardo Pelizzo 2010 Party Direction riccardo pelizzo Available at:

2 PARTY POLITICS VOL 16. No.1 pp Copyright 2010 SAGE Publications Los Angeles London New Delhi Singapore Washington DC PARTY DIRECTION The Italian Case in Comparative Perspective Riccardo Pelizzo ABSTRACT The purpose of this article is to further our understanding of the directional nature of left right scores. I suggest that a party s ability to modify its perceived position is conditional on whether parties adopt their manifestos to alter their perceived position and on whether voters are persuaded by parties attempts to relocate in political space. As voters knowledge of political parties is a major determinant of where parties are perceived to be located, new parties or parties with weak identities are more likely than old parties to modify their perceived positions, and for two reasons: they are neither willing nor able to adopt an identitybased platform and their freedom to move in political space is not constrained by what voters know about them. The results of statistical analyses show that while Italian voters modified their perception of party positions in the light of party manifestos, this was not the case in the other countries under study, where parties had longer histories than their Italian counterparts. KEY WORDS left right dimension party direction party manifesto data political space Introduction Party Manifesto Data (PMD) have been used for almost three decades in estimating party positions in the political space. Recent studies (Klingemann et al., 2006), however, point out that while left right scores estimated on the basis of party manifestos provide generally precise estimates of where parties are located in political space, there are some exceptions to the general trend. In respect of these anomalies, the literature has developed in different directions. Some scholars (Benoit and Laver, 2007; Laver et al., 2003) have developed new methodologies for estimating party positions [DOI: / ]

3 PARTY POLITICS 16(1) in political space. Others have suggested, as Budge (2006) also noted, that left right scores estimated on the basis of the party manifesto do not always indicate the party position, but rather party direction (Franzmann and Kaiser, 2006; Pelizzo, 2003). Left right scores reflect parties attempts to modify their image, to seek the median mandate (McDonald and Budge, 2005) and to maximize electoral returns. My purpose in the present article is to further our understanding of the directional nature of left right scores. Studies have shown that some parties use their manifestos to adjust to changes in the position of the median voter (Adams et al., 2004; Pelizzo, 2003). So we know from the literature that parties use the party manifesto to dissimulate where they really stand or to simulate where they stand. What the literature has not addressed so far is whether parties are actually successful in modifying voters perceptions of party positions, which is what I plan to investigate here, by analysing whether and under what conditions party perceived positions are affected by what parties say in the manifestos controlling for what parties do. In the course of the article, I proceed in the following way. In the first section, in addition to discussing the importance of spatial analyses and the various applications of the spatial analytic framework, I discuss some of the methodologies and data that have been employed to locate parties in political space. Building on this discussion, I argue that PMD that have been used successfully to estimate party positions in certain party systems can also be used to estimate party movement or direction in other settings. In the second section, I discuss the circumstances under which voters perceptions of a party s position are modified by PMD. In this respect, I suggest that a party s ability to modify its perceived position depends on two conditions; namely, whether the party wants to modify how it is perceived, and whether voters perceptions are constrained by their knowledge of party history. In the third section, I present the results of the analyses, finding that party perceived positions are strongly determined by party previously perceived positions; that the perceived position of old parties is more stable than the perceived position of new parties; and that while knowledge of a party s history is a major determinant of where it is perceived to be located in political space, the perception of party position is not fixed over time. Hence, I test whether change in perceived party position is explained by what parties say in their electoral manifestos controlling for their legislative behaviour. In the fourth and final sections of the article, I draw some conclusions as to the possible implications of this work. Spatial Analyses Issues and Methods The development of spatial analyses and the creation of new data and methodologies to estimate party positions in political space are important not just 52

4 PELIZZO: PARTY DIRECTION because they are technical improvements, but also, and more interestingly, because they allow a wide-ranging set of applications (Laver, 2001). Knowing where parties and voters are located in political space enables the analyst to assess important issues such as how responsive parties are to voters preferences (Miller et al., 1999) or what the polarization of the party system is (Sartori, 1976). 1 The importance of spatial analysis is not simply due to the fact that it can be applied when investigating a fairly wide range of political phenomena, nor is it simply that the spatial analytic framework provides an explanation for various political outcomes: it is above all due to the fact that it allows the scholar to make predictions. Knowing where parties are located enables the scholar to make fairly accurate predictions of which government coalition will be formed, which parties will be given which portfolios, and so on. Given the wide range of topics that can be investigated within the spatial analytical framework, it is not surprising that scholars have devoted considerable attention to estimating party (and voters ) positions in the policy space and to developing new techniques, methodologies and data for a more appropriate assessment of party (and voters ) positions. Policy positions have been measured a priori (Sartori, 1976; Taylor and Herman, 1971), but also on the basis of mass survey data (Sani and Sartori, 1983), elite survey data (Katz and Wessels, 1999; Miller et al., 1999), expert judgments (Benoit and Laver, 2007; Castles and Mair, 1984; Huber and Inglehart, 1995), parliamentary speeches (Ieraci, 2006), computerized word frequencies (Laver et al., 2003), roll-call voting (Poole and Rosenthal, 1997) and party manifestos (Budge et al., 2001; Klingemann et al., 2006). While PMD-based left right scores have generally provided reliable estimates of where parties are located, occasionally they have been implausible (Marks, 2007; Volkens, 2007). Given the apparent inability of PMD-based left right scores to estimate party positions, some scholars have devised new methodologies that could be used for analysing the content of political documents and estimating party positions. Others (Franzmann and Kaiser, 2006; Pelizzo, 2003; Volkens, 2007) have taken a different approach and underlined that the reason PMD-based left right scores did not provide seemingly plausible estimates of party positions in some party systems was simply that in those systems the party manifestos were used to indicate change in party position or direction. 2 Scholars have provided evidence supporting this directional interpretation of left right scores (Franzmann and Kaiser, 2006) and some studies have shown that party manifestos and PMD-based left right scores are used by political parties for strategic reasons. Studies have in fact shown that some parties use their manifestos to adjust to changes in the position of the median voter (Adams et al., 2004; Pelizzo, 2003). Studies supporting this directional interpretation have been able to show that parties use the manifestos for strategic reasons, as they hope to be able to alter voters perceptions of where they are actually located. Yet, previous studies have provided little evidence as to whether parties are actually 53

5 PARTY POLITICS 16(1) successful in their attempts to use the manifestos to modify their perceived position. Party Manifesto Data and Party Direction While other studies have focused on how the real positions of parties can best be estimated, the purpose of the present study is to address instead the factors that modify voters perceptions of party positions. While parties perceived positions to a large extent reflect long-term factors, such as voters knowledge of a party s history, the perception of party position can be modified in the light of parties legislative behaviour and of what parties say in their electoral manifestos. As I show in the next section, the perception of party positions is fairly stable over time, the reason being that parties have histories, identities, a past, and voters perceptions of where they stand reflects to some extent this historical knowledge (Adams and Somer-Topku, 2007; Erikson et al., 2002). This, along with historical memory of where parties come from, affects or, rather, stabilizes the way in which voters perceive party positions. If a party has traditionally been rightmost (or leftmost) on the political spectrum, when asked to locate this party voters will place it to the right of any other party. The Italian case provides several examples that can be used to illustrate this point. From its foundation in 1946 until its transformation into the National Alliance, the Neo-Fascist Italian Social Movement (MSI) was regarded as the rightmost party in the Italian party system. After its transformation into the National Alliance, its centripetal convergence, its adoption of more moderate stances, the National Alliance was still regarded as the rightmost party in 1994 and as the second rightmost party in the 1996 elections when a group of radical right-wing politicians, led by Rauti, broke away from the National Alliance to create the Social-Movement- Tricolor Flame. This is interesting because in 1996 the National Alliance was regarded as being located to the right of the Northern League (NL), which, in the 1996 election, had taken a much more radical stance than any other party. In that election, in addition to rejecting the logic of bipolar competition, to rejecting any alliance with either the centre left or the centre right coalition, to taking very harsh stances against illegal immigration, the NL advocated the secession of the Padania and yet was considered to be less radical or extreme than the National Alliance. 3 If voters had based their judgment of the NL and National Alliance positions exclusively on what these two parties were saying and/or doing, they should have placed the NL to the right of the AN, but they did not and this because they knew the histories of these two parties. They knew that, in spite of its centripetal convergence, the National Alliance was the party of the Neo-Fascists or that had been created from the ashes of the Neo-Fascist MSI, 4 while the Lega was a protest party that had been supported by many people who could be 54

6 PELIZZO: PARTY DIRECTION regarded as of the Centre or Left. 5 Hence, while experts (Benoit and Laver, 2007: 267) were assigning the AN and the LN the same position on the left right continuum, voters kept placing the AN to the right of the LN which testifies to the importance of historical factors in shaping voters perceptions of party positions. However, in spite of the role that historical knowledge plays in stabilizing the perception of party positions, these positions are not fixed over time and are thus of great importance for the purposes of the present analysis. In fact, it is precisely because there is some change in perceived party position that we can test whether, and under what circumstances, this change in party position is explained by parties legislative behaviour and manifestos (PMD). In the next section, I test the relationship between PMD, legislative behaviour and changes in perceived party positions and present my main findings. Before the empirical analyses, however, there is an additional question that has to be addressed, namely, under what circumstances do PMD induce a change in perceived party positions? According to McDonald and Budge (2005: 73), a party s real position can be estimated by measuring the average PMD-based left right score over the entire period of party history. This argument has an important implication, namely that the positional nature of the PMD is inherently tied to party history. If a party has a history, it has an identity, and its identity affects the content of the party electoral manifesto, which, in turn, provides a clear indication of the party s position. But there is a corollary for this line of reasoning, namely that if a party has no history or identity, as is the case of new parties, the party manifesto may contribute to shaping the party s future identity (position), but it cannot be based on it or reflect it, because a party, in its infancy, is still searching for its position in political space. In this second case, it is legitimate to postulate that PMD do not indicate party position, but reflect instead a party s search for an identity or, in spatial terms, for a position in political space a search that can be defined as party movement or direction. This point has a major implication, namely, that while old parties can use their manifestos to modify or to reinforce voters perceptions (by drafting identity-based PMD), only the second option is available to new parties. However, a party s ability to modify its perceived position on the basis of party manifestos does not depend exclusively on the strategic moves of parties, but also on whether voters are persuaded by parties attempts to relocate in political space. In this respect, the extent to which voters can modify their perception of party positions is constrained by voters knowledge of party history. Voters are less likely to change their mind about parties with which they have long been familiar, while the opposite is true of new parties about which they know very little. The comparison between the AN, heir of the former Neo- Fascist MSI, and the LN, a new protest party, is again a good case in point. In fact, while the AN was able to change its perceived position by only 6.1 percent in the period, the LN was able to modify its position by 55

7 PARTY POLITICS 16(1) 10.4 percent. These results sustain the claim that voters knowledge of a party s history has a major influence on the extent to which party perceived positions can be modified either in the light of what parties say in their electoral manifestos or of what they do in parliament. Therefore, we hypothesize that new parties have a greater ability to modify where they are perceived to be located in political space than old parties and that they can use their electoral manifestos in their perceived position. To test whether my argument is supported by empirical evidence, I test in the next section whether and to what extent changes in parties perceived positions are due to parties programmatic statements controlling for their legislative behaviour. In the course of the empirical analysis, I compare the American, Dutch, German and Italian cases. The German, the Dutch and the American cases were selected because they were the systems that differed most. The Dutch and the German cases differed from the American in terms of party system format, party system mechanics, electoral system and form of government. The only thing that these cases have in common is that their political parties are relatively old compared with the Italian parties of the so-called Second Republic; hence, if changes in party perceived position were not affected by PMD, this evidence would support our claim that a party s age is a key determinant of whether it is able to modify voters perceptions. The Italian case was also included in the analysis because it is the one most like the German and Dutch cases, but with one exception: Italian parties in the so-called Second Republic are relatively young. Therefore, if changes in party perceived positions were affected by PMD in the Italian case, but not in the other cases, this evidence would be consistent with the claim that parties ability to modify their perceived positions is a function of their age. In order to assess whether and to what extent short-term factors, in the form of programmatic statements, I run three models: PPP = a + b1 PPPP + b2 PMD + e (1) PPP = a + b1 PPPP + b2 LB + + b3 PMD + e (2) Shift = a + b1 PMD + e (3) Where PPP indicated party perceived position at the time of an election, PPPP indicates the party s previously perceived position; PMD refers to the left right scores estimated by Klingemann et al. (2006); LB refers to party positions estimated on the basis of the legislative behaviour of their members; while Shift refers to the difference between PPP and PPPP. Party perceived positions were estimated on the basis of data collected by the American National Election Studies in the period 1972 to 2004 by the Deutsche Nationale Wahlstudien conducted in Germany from 1976 to 1990, by the Dutch Parliamentary Surveys conducted from 1981 to 1994, and, for 56

8 PELIZZO: PARTY DIRECTION Italy, by the Istituto Cattaneo from 1994 to PMD were taken from Budge et al. (2001) and Klingemann et al. (2006). The LB data were collected for the USA and Italy. The LB variable for the American case is constructed on the basis of W-Nominate. This methodology, devised by Poole (1998), is a scaling technique that can be employed to estimate latent variables, such as ideological orientation, that underpin observable variables. This methodology was applied to roll-call votes (observable variables) to estimate the position of legislators in political space (latent variables). Party positions were then estimated by computing the average score for all the legislators belonging to a specific party in a given Congress. The LB variable for the Italian case refers to the party positions estimated by applying the Optimal Classification methodology (Poole, 2000, 2005) to the final votes in the Italian Chamber of Deputies in the Course of the XIII Legislature. 6 By performing this analysis, Landi and Pelizzo (2006) estimated the positions of all the members of the Italian parliament in a two-dimensional space. By computing the average positions of all the members belonging to a given party, they estimated the positions of all the parties on both dimensions. These variables and data are used here to test whether changes in perceived party positions are due to parties legislative behaviour and electoral manifestos. 7 Data, Methods and Empirical Results The first finding of my analyses is that the perception of party positions is stable over time. When correlating the positions of American parties (Democratic Party and Republican Party) as perceived by American voters in any election held in the period 1976 to 2004 with American parties perceived positions in the previous elections, the Pearson r is When correlating the positions of Dutch parties as perceived in the 1982, 1986, 1989 and 1994 elections with their perceived position in the previous elections, the correlation coefficient is Correlating the perceived position of German parties in the 1980, 1983, 1987 and 1990 elections with their perceived position in the previous elections yields a coefficient of The Italian case is no exception in this respect. When we correlate the perceived positions of Italian parties in 1996 with their perceived positions in the 1994 elections, or when we correlate the perceived position of Italian parties in the 2001 elections with their perceived position in the 1996 elections, the Pearson r is and 0.976, respectively. When we correlate party positions in 1996 and 2001 with a party s previously perceived position, the correlation coefficient is In other words, the perception of party positions is stable over time. Data are reported in Table 1. This evidence indicates that knowledge of the history and origins of parties affects the way in which voters perceive party positions and, therefore, that it is appropriate to regard voters historical knowledge as a key determinant of parties perceived positions. 57

9 PARTY POLITICS 16(1) Table 1. Correlation of perceived party positions (sig.) PPP Germany Italy The Netherlands USA PPPP (0.000) (0.000) (0.000) (0.000) N The fact that knowledge of a party s history may stabilize voters perception of party positions has a major implication, namely that the changes in party perceived positions should be smaller for old and well-known parties than they are for younger parties. This claim is supported by both withinsystems and between-systems analyses. Looking at the German case, for example, we can see that perceived party positions changed by 22.6 percent for the Greens, by 5.6 percent for the SPD, by 0.35 for the CDU and by 1.85 percent for the FDP. In other words, while the perceived position of old parties was fairly stable, that of the Greens, which was then Germany s youngest relevant party, was characterized by considerable fluctuation. A similar trend between party age and fluctuations in party perceived positions can be detected between countries. A quick comparison of the data presented in Tables 2 and 3 reveals that change in party perceived positions was greater for the Italian party system that had emerged in the wake of the First Republic than it was in the other countries under study. In fact, from 1996 to 2001 the perceived position of Italian parties changed on average Table 2. Changes in perceived party positions Percentage Average change change in in perceived perceived party position Country Party Period party position by country Germany Greens Germany SPD Germany FDP Germany CDU The Netherlands PPR The Netherlands PvdA The Netherlands D The Netherlands CDA The Netherlands VVD USA Democrats USA Republicans

10 PELIZZO: PARTY DIRECTION Table 3. Changes in perceived party positions Percentage change Average change in in perceived party perceived party Party Period position position FI LN DS DEM PPI CCD RI RC AN by almost 10 percent, which was remarkably higher than the values recorded in the other countries. The evidence presented so far makes clear that party perceived positions are fairly stable over time and that the stability of party perceived positions is affected by the age of the parties. The perceived position of old parties is more stable than that of young parties because voters knowledge of party history constrains the extent to which old parties can modify how they are perceived. This said, one should note that perception of party positions is not just a function of voters knowledge of party histories. Party positions are not fixed over time, but they change to a more or less significant extent. The question is: What factors are responsible for changes in perceived party positions? In order to answer this question, I run the three regression models presented in the previous section. Table 4. Determinants of perceived party positions Previously perceived PMD-based party Legislative left right Intercept position behaviour scores R-squared N Germany (0.367) (0.000) (0.204) Italy (0.020) (0.000) (0.059) The Netherlands (0.199) (0.000) (0.376) USA (0.723) (0.000) (0.955) 59

11 PARTY POLITICS 16(1) By running the first model, I find that when controlling for the effects of party previously perceived position, programmatic statements in the form of PMD do not induce any change in party perceived positions. The b- coefficients for the impact of PMD on change in party perceived positions are weak and statistically insignificant. The only exception to this trend is possibly represented by the Italian case, where the impact of the PMD on the perception of party position is relatively stronger and almost statistically significant. By running the second model with American and Italian data, it can be seen that in the USA the perception of party positions can be modified by parties legislative behaviour but not by parties programmatic statements and that parties legislative behaviour is a stronger and more significant determinant of where parties are perceived to be located than parties previously perceived position. The Italian case is substantially different from the American one. In fact, in the Italian case, even when we control for PPPP and LB, PMD exercise a strong, properly signed and substantially (though not statistically) significant influence of parties perceived position. Hence, while the perceived position of American parties is determined exclusively by what parties and legislators do in the legislative arena, the perceived position of Italian parties reflects long-term factors such as voters knowledge of party history and short-term factors such as legislative behaviour and programmatic statements. Why is there such a difference between the Italian and American cases? Part of the answer is to be found in the fact that programmatic statements always go hand in hand with legislative behaviour in the American case. The correlation between PMD data and previous legislative behaviour yields a correlation coefficient of 0.773, while the correlation between PMD and future behaviour is PMD reflect past behaviour and determine future behaviour, which is why, when controlling for the effects of parties legislative behaviour, the ability of PMD to alter party perceived positions is modest. The Italian case presents substantial differences. The correlation between Italian party manifestos in the 1996 elections and their behaviour in the XIII Table 5. Determinants of perceived party positions Previously perceived PMD-based party Legislative left right Intercept position behaviour scores R-squared N Italy (0.285) (0.002) (0.028) (0.077) USA (0.883) (0.407) (0.009) (668) 60

12 PELIZZO: PARTY DIRECTION Legislature ( ) is statistically insignificant (r = 0.404). The legislative behaviour of Italian MPs in the XIII Legislature correlates strongly (r = 0.830) and significantly with parties programmatic statements for the 2001 election. In other words, the fact that PMD in Italy reflect past behaviour is what makes them credible, but is not necessarily a major determinant of future behaviour. 9 This is why when LB and PMD are both used to predict change in party perceived positions, the coefficients for both variables are relatively strong and far from trivial. Models 1 and 2 indicate that changes in party perceived positions are affected by parties legislative behaviour in both the Italian and American cases, but that PMD are responsible for changes in party perceived positions only in the Italian case. Additional evidence in this respect is provided by running the third model described in the previous section. When testing whether changes in party perceived positions or shift are a function of parties programmatic statements, it can be seen that the impact of PMD on changes in party perceived position (Shift) is weak and statistically insignificant in Germany, The Netherlands and the USA, while it is strong and statistically significant in the Italian case. In the light of these findings, the question then becomes why is the Italian case an exception to the trend that I have detected in the other countries under study? The answer is provided by comparative analysis. In Germany, The Netherlands and the USA, parties had long histories; they had strong identities that had been consolidated over the years and their programmatic statements (as exemplified by the PMD) reflected parties identities. 10 In other words, in these three countries parties had the option of campaigning on their identity and voters perceptions of party positions were constrained by their knowledge of these parties. This was not at all the Italian case. In the wake of the 1993 electoral reform and the collapse of the so-called First Republic, the legislative behaviour of Italian parties was loosely based on their programmatic statements; the PMD did not reflect parties long history because, in the course of the transition, most Italian parties had very little history. Some Table 6. PMD and shifts in perceived party positions Shift in perceived party position Intercept PMD R-squared N Germany (0.998) (0.837) Italy (0.839) (0.018) The Netherlands 0.032) (0.900) (0.638) USA (0.964) (0.639) 61

13 PARTY POLITICS 16(1) parties, such as Forza Italia or the NL, were new, while others, such as the Democrats of the Left (Baccetti, 1997; Bellucci et al., 2000; Ignazi, 1992) or the National Alliance (Ignazi, 1994; Tarchi, 1997) were trying to dissociate themselves from their past. Under these circumstances, parties could use their manifestos to appeal to the electorate, although these were prospective rather than retrospective (Fiorina, 1981; Lewis-Beck, 1990) and were used not only to maximize electoral returns but also to provide parties with the identity that they were looking for. 11 In other words, the Italian parties, unlike their Dutch, German and American counterparts, could not use their electoral manifestos positionally and voters perceptions of party positions were not constrained by what voters knew about these parties. The evidence generated by this comparison sustains the claim that programmatic statements reshape so significantly voters perception of party positions only under very specific conditions (parties with no history, parties with weak to no identity, non-identity based manifestos, voters limited knowledge of the new parties). Conclusions Previous studies have noted that in some settings the left right scores estimated on the basis of the PMD do not adequately indicate where parties are located in political space. Some studies have suggested that the inability of left right scores to put parties in the right space is due to the fact that they indicate party direction rather than party position (Franzmann and Kaiser, 2006; Pelizzo, 2003). It has been shown that parties use their manifestos to modify their position. Parties have been shown to do so to adjust to changes in the position of the median voter (Adams et al., 2004) and to modify where they are perceived to be located. What previous parties have not shown is whether and to what extent they are successful in modifying their perceived position. The purpose of the present study was to address parties ability to modify their perceived position on the basis of their electoral manifestos. In the course of analysis, I found that voters perceptions of party positions are fairly stable over time, that the perceived positions of old parties are more stable than those of new parties, and that party age affects not only the stability of perceived party positions but also their ability to alter voters perceptions on the basis of their electoral manifestos. In this respect, I noted that a party s ability to change the way in which it is perceived depends on two conditions: the first is the party s willingness to use its electoral manifestos directionally, while the second is the voters willingness to change their views about party positions. I further argued that both conditions are related to party age. As older parties have longer histories and stronger identities than new parties, they may adopt identitybased manifestos that are not designed to alter voters perceptions of party 62

14 PELIZZO: PARTY DIRECTION positions. And even when older parties decide to use their electoral programmes directionally, they may fail to alter their perceived position because voters perceptions are anchored in their knowledge of party histories. The opposite is true for new parties. Because they are new and little known, new parties are bound to find a place in political space, and their ability to move in political space and alter their perceived positions is not constrained by voters knowledge. The results of our data analysis sustain the claim that parties ability to change their perceived position is related to party age. By doing so, this article has not only shown whether parties are successful in modifying how they are perceived but also the conditions (no party history, weak identity, limited voters knowledge) under which they are able to do so. The data at our disposal and the analyses that I have conducted in this article do not allow me to speculate as to whether my conclusions will be corroborated by further analyses. But, regardless of whether my conclusions will or will not be supported by further research, I have no doubt that the reason parties succeed in modifying their perceived positions is a question worth investigating as it may provide party scholars with a better understanding of the relationship between parties and voters. Notes 1 According to Sartori (1976: 135), there is ideological polarization when the parties located at the extreme ends of the ideological spectrum are literally two poles apart, and the distance between them covers a maximum spread of opinion (...) Briefly put, we have polarization when we have ideological distance (in contrast to ideological proximity). 2 In this respect, Franzmann and Kaiser (2006: 164) noted that the directional interpretation of the PM-based left right scores was placed in the context of a directional theory of voting, as developed by Rabinowitz and MacDonald (1989). While it is certainly possible and possibly interesting to use the PMD left right scores in light of the Rabinowitz and MacDonald directional theory, I have the impression that the directional interpretation of the PMD proposed by Pelizzo (2003) was actually consistent with traditional proximity theory. That article suggested the parties use their electoral manifestos to adjust their position or to take a position closer to that of the voter to maximize their chances of winning that voter s support. This is exactly what one would expect parties to do within the proximity theory framework, but not within the framework provided by directionality theory. Under the assumption of directionality theory, the utility that a voter attaches to a party is not a function of the distance between the position of the voters and that of the party, but is due instead to whether the party and the voters are on the same side of the issue and to how strongly the voter and the party feel on that issue. Under these assumptions, if there are two parties on the same side of an issue of the voter, their winning strategy is not to move closer to the voter but to take a more radical stance on the issue, even if this means moving further away from where the voter is located 63

15 PARTY POLITICS 16(1) on the issue. This, I believe, is very different from what is suggested by Pelizzo (2003). 3 Padania is the term used to refer to northern Italian regions such as Liguria, Piedmont, Lombardy, Veneto, Trentino-Alto Adige, Friuli, Emilia-Romagna and Tuscany. 4 Piero Ignazi, Il Polo Escluso (Bologna, il Mulino, 1989); Piero Ignazi, Postfascisti? (Bologna, Il Mulino, 1994); Marco Tarchi, Cinquant anni di nostalgia (Milano, Rizzoli, 1995); Marco Tarchi, Dal M.S.I. ad A.N. (Bologna, il Mulino, 1997). 5 Early studies on the Northern League indicated that it was in many respects a centre party; see Diamanti (1994, 1996); Mannheimer (1992); Leonardi and M. Kovacs (1993). More recent studies, however, have pointed to the Northern League no longer being regarded as a centre party and that in many respects it now resembles the parties of the new extreme right; see Betz (2002) and Tarchi (2003). The Northern League s transformation from a catch-all party to a populist party was noticed by Ilvo Diamanti (1997). 6 The OC methodology is also a scaling technique that can be used to detect latent variables, such as spatial positions, on the basis of observable variables. Specifically, this method, just like the nominate methodology, is used to estimate the positions of legislators on the basis of their voting record. The basic difference between nominate and OC is that while nominate maximizes the likelihood of legislators choices (Poole, 2000: 211), OC maximizes correct classification of the legislators choices (Poole, 2000: 211). Landi and Pelizzo (2006) interpreted the first dimension of the two-dimensional space that they had identified as the left right dimension. I provide external validation of the OC estimates by correlating with other estimates of party positions, i.e. I correlate both party positions on the first dimension of the space estimated with the OC methodology with party positions estimated on the basis of the expert surveys (Benoit and Laver, 2007) and with the party positions estimated on the basis of the analysis of the parliamentary speeches delivered in the course of the XIII Legislature for the investiture of the Prodi, D Alema and Amato II governments (Ieraci, 2006). By performing these correlations I find that the OC estimates are strongly and significantly correlated with all the other estimates of party positions (r = 0.898, r = 0.791, r = 0.595, r = 0.863). 7 The Liberal Conservative scale used in the United States survey is a 7-point scale, the Dutch survey adopted instead a 10-point scale, and the German surveys an 11-point scale. Parties perceived positions were indicated on a 5-point scale in the 1994 and 1996 Italian surveys, while on a 10-point scale in PMD range from 100 to OC data range from 1 to +1. Given this variation, variables were standardized before performing statistical analyses. 8 Both significant at the level. The N is 18 and 16, respectively. 9 Which, in turn, may erode the legitimacy of Italian parties. The inability or unwillingness of Italian parties to keep their promises is discussed by Panebianco as the major reason why Italian voters have little trust in Italian parties and that there is a second wave of anti-party sentiment in Italy (see Panebianco (2007)). The first wave of anti-party sentiment in Italy was examined by Bardi (1996). 10 The fact that PMD of the American parties are strongly related to party identities can be illustrated by PMD being strongly correlated with party previously perceived position (r = 0.835) and party perceived positions (r = 0.790). 64

16 PELIZZO: PARTY DIRECTION 11 As noted earlier, the legislative behaviour of Italian parties was inconsistent with their programmatic appeals, which is why, I believe, the parties that had emerged by the end of the first republic have failed to create and consolidate an identity that could be used to appeal to the electorate. References Adams, James, Michael Clark, Lawrence Ezrow and Garrett Glasgow (2004) Understanding Change and Stability in Party Ideologies: Do Parties Respond to Public Opinion or to Past Electoral Results?, British Journal of Political Science 34: Adams, James and Zeynep Somer-Topku (2007) Moderate Now an Win Votes Later. Paper presented at the Annual Meeting of the American Political Science Association, Chicago, IL, 31 August to 2 September. Baccetti, Carlo (1997) Il PDS, Bologna: il Mulino. Bardi, Luciano (1996) Anti-Party Sentiment and Party System Change in Italy, European Journal of Political Research 29: Bellucci, Paolo, Marco Maraffi and Paolo Segatti (2000) PCI, PDS, DS. Rome: Donzelli. Benoit, Kenneth and Michael Laver (2007) Party Policy in Modern Democracies. London: Routledge. Betz, Hans-Georg (2002) Contro la globalizzazione: xenofobia, politica dell identità e populismo escludente nell Europa occidentale, Trasgressioni, anno XVII: Budge, Ian (2006) Identifying Dimensions and Locating Parties: Methodological and Conceptual Problems, in Richard S. Katz and William Crotty (eds) Handbook of Party Politics, pp London: Sage. Budge, Ian, Hans-Dieter Klingemann, Andrea Volkens, Judith Bara and Eric Tanenbaum (2001) Mapping Policy Preferences. Oxford: Oxford University Press. Castles, Frances and Peter Mair (1984) Left Right Political Scales: Some Experts Judgements, European Journal of Political Research 12: Diamanti, Ilvo (1994) La Lega. Rome: Donzelli. Diamanti, Ilvo (1996) Il Male del Nord. Rome: Donzelli. Diamanti, Ilvo (1997) La Lega: dal federalismo alla secessione, in Roberto d Alimonte and David Nelken (eds) Politica in Italia. Edizione 97, pp Bologna: il Mulino. Erikson, Robert, Michael MacKuen and James Stimson (2002) The Macro Polity. Cambridge: Cambridge University Press. Huber, John and Ronald Inglehart (1995) Expert Interpretations of Party Space and Party Locations in 42 Societies, Party Politics 1: Fiorina, Morris (1981) Retrospective Voting in American Elections. New Haven, CT: Yale University Press. Franzmann, S. and A. Kaiser (2006) Locating Political Parties in Policy Space, Party Politics 12: Ieraci, Giuseppe (2006) Government, Policy Space and Party Positions in the Italian Parliament ( ): An Inductive Approach to Parliamentary Debate and Votes of Investiture, South European Society and Politics 11: Ignazi, Piero (1989) Il Polo Escluso. Bologna: il Mulino. 65

17 PARTY POLITICS 16(1) Ignazi, Piero (1992) Dal PCI al PDS. Bologna: il Mulino. Ignazi, Piero (1994) Postfascisti? Bologna: Il Mulino. Katz, Richard S. and B. Wessels (eds) (1999) The European Parliament, the National Parliaments and European Integration. Oxford: Oxford University Press. Klingemann, Hans-Dieter, Andrea Volkens, Judith Bara, Ian Budge and Michael McDonald (2006) Mapping Policy Preferences II. Oxford: Oxford University Press. Landi, Massimiliano and Riccardo Pelizzo (2006) A Spatial Analysis of the XIII Legislature. Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, April. Laver, Michael (ed.) (2001) Estimating Policy Position of Political Actors. London: Routledge. Laver, Michael, Kenneth Benoit and John Garry (2003) Estimating Policy Preferences Using Words as Data, American Political Science Review 97: Leonardi, Robert and M. Kovacs (1993) The Lega Nord: the Rise of a New Catch All Party, Italian Politics 8: Lewis-Beck, Michael (1990) Economics and Elections. The Major Western Democracies. Ann Arbor, MI: University of Michigan Press. Mannheimer, Renato (1992) The Electorate of the Northern League, in Gianfranco Pasquino and Patrick McCarthy (eds) The End of Post-War Politics in Italy. Boulder, CO: Westview Press. Marks, Gary (ed.) (2007) Comparing Measures of Party Positioning. Expert, Manifesto and Survey Data Compared, Special Issue, Electoral Studies 26. McDonald, Michael and Ian Budge (2005) Elections, Parties, Democracy. Conferring the Median Mandate. Oxford: Oxford University Press. Miller, Warren E., Roy Pierce, Jacques Thomassen, Richard Herrera, Soren Holmberg, Peter Esaiasson and Bernhard Wessels (1999) Policy Representation in Western Democracies. Oxford: Oxford University Press. Panebianco, Angelo (2007) La delusione dei moralisti, Corriere della Sera, September 22. Pelizzo, Riccardo (2003) Party Position or Party Direction? An Analysis of Party Manifesto Data, West European Politics 26: Poole, Keith (1998) Recovering a Basic Space from a Set of Issue Scales, American Journal of Political Science 42: Poole, Keith (2000) Non-Parametric Unfolding of Binary Choice Data, Political Analysis 8: Poole, Keith (2005) Spatial Models of Parliamentary Voting. New York: Cambridge University Press. Rabinowitz, George and Stuart E. MacDonald (1989) A Directional Theory of Issue Voting, American Political Science Review 83: Sani, Giacomo and Giovanni Sartori (1983) Polarisation, Fragmentation and Competition in Western Democracies, in Hans Daalder and Peter Mair (eds) Western European Party Systems, pp London: Sage. Sartori, Giovanni (1976) Parties and Party Systems. New York: Cambridge University Press. Taylor, Michael and Valentine Herman (1971) Party System and Government Stability, American Political Science Review 65: Tarchi, Marco (1995) Cinquant anni di nostalgia. Milan: Rizzoli. Tarchi, Marco (1997) Dal M.S.I. ad A.N. Bologna: il Mulino. 66

18 PELIZZO: PARTY DIRECTION Tarchi, Marco (2003) Radicalismo di destra e neofascismo nell Europa del dopoguerra, Trasgressioni, anno XVIII: Volkens, Andrea (2007) Strengths and Weaknesses of Approaches to Measuring Policy Positions of Parties, Electoral Studies 26: RICCARDO PELIZZO is a Research Fellow at the Centre for Governance and Public Policy of Griffith University, where he is also an academic member of the Griffith Asia Institute and an Associate Investigator of CEPS. Most of his early work, including an article appearing in 2007 in Party Politics, was on European parties and party systems. ADDRESS: Centre for Governance and Public Policy, Griffith University, 170 Kessels Road, Brisbane QLD 4000, Australia. [ r.pelizzo@griffith.edu.au] Paper submitted 8 July 2007; accepted for publication 23 November