Radical right and radical left party ideologies: a pan-European mapping of necessary attributes, differences and similarities

ABSTRACT

Radical parties, both to the right and to the left, have recently earned victories in a range of European countries. Previous efforts seeking to understand radical parties in general are, however, relatively rare. The present paper offers an examination of the possible necessary attributes of, as well as the differences and similarities between, radical right and radical left party ideologies in contemporary (Western, Central and Eastern) Europe. The expectation that nativism is a necessary attribute of radical right ideology while anti-capitalist socialism is a necessary attribute of radical left ideology receives partial support; although not present in all of the manifestos produced by these parties, these attributes are, from a broader perspective, necessary ingredients of the radical right and the radical left party ideology, respectively. The empirical examination also confirms that the two families clearly differ in their sociocultural and, to a lesser extent, socioeconomic views. Finally, it demonstrates that they take matching standpoints towards European integration and, more ambiguously, towards the political system.

KEYWORDS:

• Europe
• radical left
• radical right
• political parties
• ideologies

Notes

1 The rather scant research on radicalism in general is surprising, not least in view of a number of theoretical arguments suggesting that the radical right and the radical left families may have at least some features in common. One of the most well-known claims in this vein is provided by Bobbio (1996, pp. 21–22; see also Backes, 2010), who argues that the right and left outposts share ‘a rejection of democracy’ and ‘a catastrophic vision’ of history. ‘The extremes meet’, he contends, but ‘not because of their position[s] on the political spectrum, but because they occupy the two extreme points of that spectrum’. Other, more recent, theoretical contributions highlighting possible ideological similarities between the radical right and the radical left are, for example, Mudde (2004) on populist attitudes and Szczerbiak and Taggart (2008) on Eurosceptic stances.

2 In this paper, the left–right-dimension is understood as a ‘hopelessly multidimensional dimension’; as a ‘layman’s “index” of politics’ (Sartori, 1976, p. 79) that summarizes ‘citizens’ and parties’ positions on the issues of the day’ (Dalton, Farrell, & McAllister, 2011, p. 82). The distinction is, hence, not seen as a purely socioeconomic dimension but, rather, as a ‘superdimension’ that includes several prominent (e.g. socioeconomic and sociocultural) sub-dimensions (Dalton et al., 2011, ch. 4–5; see also Hooghe, Marks, & Wilson, 2002; Kriesi et al., 2006). Since the vast majority of the relevant parties are radical (i.e. they accept basic democratic norms but seek a root [lat. radix] and branch transformation of liberal democracy) rather than extreme (i.e. anti-democratic, oppose democracy per se), I use the terms ‘radical right’ and ‘radical left’ as generic labels (for a more detailed conceptual discussion, see e.g. Mudde, 2005, pp. 87–89).

3 The minimal definition disentangles the attributes that all non-idiosyncratic uses of the term have in common, while the maximal definition identifies all (non-idiosyncratic) characteristics that define the term in its most archetypical form. Minimal definitions thus have few attributes (low intension) but many referents (high extension), while maximal definitions tend to have many attributes (high intension) but few (if any) referents (low extension). For more details, see Gerring and Barresi (2003, pp. 205–209) and also Sartori (1970, 1984).

4 Whether nationalism is also a sufficient attribute is hard to determine outside a specific context, and Gerring and Barresi (2003) advise us to ignore sufficiency in min–max definitions. I note, however, that there is reason to suspect that sufficiency is violated in this case because there are probably parties that should not be classified as radical right although nativism is present. (Think, for example, of the Communist Party of the Russian Federation.) Various combinations of the attributes included in the maximal definition can, however, be considered as sufficient for a party to be classified as radical right.

5 The classification of ecologism as a sociocultural issue is certainly somewhat questionable; in reality, ecologism has both sociocultural and socioeconomic features (see e.g. Carter, 2007).

6 These are the two strategies that are at the core of the leading projects focusing on estimating party positions, namely the content-analytical Manifesto Research on Political Representation (MARPOR, see Volkens et al., 2016) and the Chapel Hill Expert Surveys (CHES, see Bakker et al., 2015).

7 Unfortunately, the CHES data does not provide any information about parties’ views on the political system. Hence, possible variances pertaining to the type of data cannot be hypothesized.

8 For a more detailed description of the categories and indicators used, see online appendix B1. The CHES data set also includes an indicator labelled Immigrate_policy. Since opposition to immigration is a non-salient issue in countries with little or no experience of immigration (see Mudde, 2007), I do not include this indicator in the analysis.

9 For a description of the categories and indicators used, see online appendix B1. Category per503, capturing (positive) mentions of social justice and equality, is not included because of its broad scope; it includes mentions of a broad range of issues of which some are more sociocultural than socioeconomic in nature (e.g. mentions of the need to end racial or sexual discrimination).

10 The socioeconomic scale is created by Laver and Garry (2000), based on categories identified by Laver and Budge (1992). The sociocultural scale, in turn, resembles a scale proposed by McDonald and Mendes (2001) but replaces the ambiguous categories per606, per705 and per706 with the more clear-cut categories per107, per109 and per502. The main problem with using per606 as an expression of sociocultural rightism is that included features such as public spiritedness and condemnations of anti-social attitudes may also be associated with the sociocultural left. The sociocultural left categories per705 and per706, in turn, are problematic because they include favourable references not only to groups typically associated with sociocultural leftism, such as homosexuals, immigrants, women and linguistic groups, but also to groups equally favoured by the sociocultural right, such as handicapped and old people. A more detailed presentation of the scales used is given in table B-III in online appendix B5.

11 While 0.75 is often used as a cutoff value for sufficiency relations, a score of at least 0.90 is sometimes recommended for necessity inclusion. As argued by Thiem (2016b), the inclusion cut-off for necessity, need, however, not be different from that used when testing for sufficiency because of the symmetry of sufficiency and necessity. The use of 0.75 as an (admittedly liberal) cut-off point is, hence, in line with best practice.

12 Calibrations are performed using the QCApro package for R (Thiem, 2016a). When analysing the Manifesto data, the ‘outcome’ of interest is party family membership. In the analysis of the necessity of nativism for the radical right, the outcome is 1 if the party is classified as ‘nationalist’. Since I find no theoretical reasons for excluding party families as irrelevant (i.e. families where nativism cannot occur), I code all non-radical right parties as 0 on the outcome (see the discussion in Goertz, 2006, ch. 7–8). Likewise, when analysing the necessity of anti-capitalist socialism for the radical left, the outcome is 1 if the party is classified as ‘socialist’ and 0 if it is not. The analysis of the CHES data follows the same procedure (radical right/radical left = 1; not radical right/radical left = 0). Finally, I note that the terms ‘condition’ and ‘outcome’ are well established in set-theoretic literature focusing on causal relationships. The relations of interest here are, however, ontological rather than causal; thus, I henceforth prefer to hold on to the term ‘attribute’ rather than the term ‘condition’, and the terms ‘(radical right/-left) family membership’ instead of ‘outcome’. I also note that the binary nature of the ‘outcome’ should be kept in mind when interpreting the results.

13 For details, see online appendix B5. For the sake of robustness, I also report ratio versions (see Kim & Fording, 1998) of the scales (see online appendix C). The ratio and logit scaling techniques are (after the missing values caused by division by zero have been re-coded as 0) closely related (with N = 1,976, r_P ranges between 0.88 and 0.96 for the four different scales), and the interpretation of the results remain largely the same.

14 The p-values reported in Figures 2–4 are from independent t-tests. The p-values in Figures 2 and 3 are one-tailed while the p-values given in Figure 4 are two-tailed. The p-values from comparisons of correlation coefficients reported in the text are all one-tailed.