Jump to ContentJump to Main Navigation
Myth and Reality of the Legitimacy CrisisExplaining Trends and Cross-National Differences in Established Democracies$

Carolien van Ham, Jacques Thomassen, Kees Aarts, and Rudy Andeweg

Print publication date: 2017

Print ISBN-13: 9780198793717

Published to Oxford Scholarship Online: August 2017

DOI: 10.1093/oso/9780198793717.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 29 January 2020

Legitimacy Decline and Party Decline

Legitimacy Decline and Party Decline

Chapter:
(p.76) 5 Legitimacy Decline and Party Decline
Source:
Myth and Reality of the Legitimacy Crisis
Author(s):

Rudy B. Andeweg

David M. Farrell

Publisher:
Oxford University Press
DOI:10.1093/oso/9780198793717.003.0005

Abstract and Keywords

This chapter discusses the decline of political parties as a possible cause of the decline of legitimacy. Political parties constitute a link between the citizens and the political system, and therefore a loss of support could delegitimize the political system. However, the decline of political parties can only cause legitimacy decline if they are indeed in decline and if there is a causal relationship between citizens’ involvement in political parties and political support. The chapter argues that empirical evidence for party decline is limited, as parties may have undergone transformation rather than decline. Using ESS data from 2002 to 2010, the chapter finds only weak relations between political support and party membership and party closeness. However, being close to a particular party is more important than being a member of a political party, and is interpreted as a sign that the party system facilitates citizens in making meaningful political choices.

Keywords:   party decline, political parties, party membership, party closeness, political support, satisfaction with democracy, political trust

The perception that satisfaction with the functioning of democracy and trust in democratic institutions are declining is as widespread as the view that political parties are weakening, and the two diagnoses are usually seen as closely intertwined: “It is difficult therefore not to infer some link between recent party decline and the widespread feeling of cynicism, indifference and hostility towards politicians—especially in their partisan role—that has been revealed in a number of recent surveys” (Crewe et al. 1977: 188).

Any hypothesis relating a decline of political legitimacy to a decline of parties is based on the premise that political parties constitute a linkage mechanism between the citizens and the political system. When universal suffrage was introduced, many were skeptical about the feasibility of linking the enfranchised masses to political decision-making. After all, under the régime censitaire the electorate was still relatively small and homogeneous, restricted in terms of income and gender as it was. This small size and homogeneity at least had the advantage that they often enabled individual representatives to maintain some form of relationship with individual voters. With a much larger, and consequently also much more diverse electorate, individual political representation no longer seemed viable. It is not by accident that the interwar years saw a wave of enfranchisement in many countries, but also a wave of support for antidemocratic movements. Miraculously, whether by accident or by design, mass parties emerged or developed from preexisting cadre parties, as a new linkage mechanism between citizens and government. The importance of political parties for the functioning of mass representative democracy has been underlined by many authors since, and is epitomized in the well-known quotation from Schattschneider that “modern democracy is unthinkable save in terms of parties” (1942: 1). It is thus not surprising there should be fears that any erosion (p.77) of political parties could undermine representative democracy as such, and that a loss of support for parties could delegitimize the democratic regime.

The evidence presented in Chapter 2 of this volume, however, indicates that, at least in Western Europe since the 1970s, no secular decline of political support has taken place and that we are faced rather with variation across countries and fluctuation over time. If that is so, we are left with two possibilities concerning the relationship between support for democratic institutions and party strength: either political parties have also not declined, or the relationship between parties and political support is not as tight as it is sometimes thought to be. In this chapter1 we shall inquire into both possibilities. To determine whether there has been any decline of parties as elite–mass linkage mechanism we look at parties in the aggregate, at their strength in terms of members and supporters, and at their performance of the linkage functions that are attributed to them. And to find out whether party support is indeed related to political support, we look at the individual level, at citizens’ involvement in political parties, and the effect of that partisanship on their political trust and democratic satisfaction.

5.1 Party Decline Revisited

5.1.1 Evidence of Weakening Parties

In the opening lines of his posthumously published work Peter Mair notes that “[t]he age of party democracy has passed” (2013: 1). He is not alone in arguing that parties are in decline, nor is he the first, but his analysis provides a useful means of framing the main points usually made about the growing threat to parties.

Table 5.1 provides summaries of the types of evidence generally used to show how parties are becoming weaker in established democracies.2 Given that this argument on party decline has already been well rehearsed elsewhere (e.g. Dalton and Wattenberg 2000; Mair 2013) we can present the main points briefly here. The first commonly used indicator of change is decline in party membership: of all the indicators summarized in Table 5.1 this is the one that is most centered on political parties. The evidence is pretty incontrovertible: party membership is in decline to such an extent that the dues-paying party member is in danger of becoming a dying breed. In the most comprehensive study of cross-national trends Van Biezen and her colleagues contend that “party membership levels have now fallen to such a low level that membership itself no longer offers a meaningful indicator of party organisational capacity” (2012: 40). And this is a downward trend that continues unabated as shown by the most recent evidence reported in the international Political Party Database project (Poguntke et al. 2015).3

Table 5.1 Trend data on parties and elections

Country

Party membership as a percentage of the electorate

Turnout (VAP)

Mean aggregate electoral volatility

Party closeness (% not close to any party)

1980 (or as near as possible)

2008 (or as near as possible)

First two elections in 1970s

Last two elections in 2000s

1970s

2000–09

1978

2014

Austria

28.48

17.27

88.87

74.41

2.7

15.5

40.9

Belgium

8.97

5.52

87.16

85.32

5.3

14.5

37.3

25.8

Denmark

7.30

4.13

87.34

82.27

15.5

10.4

35.8

8.9

Finland

15.74

8.08

84.05

69.07

7.9

6.8

32.5

France

5.05

1.85

67.01

45.34

8.8

13.5

28.3

29.2

Germany

4.52

2.30

86.29

68.30

5.0

9.0

35.1

25.3a

Ireland

5.00

2.03

81.46

67.93

5.7

7.5

38.6

53.0

Italy

9.66

5.57

95.06

80.63

9.9

14.0

22.3

24.9

Netherlands

4.29

2.48

83.96

74.30

12.3

22.3

14.5

24.1

Norway

15.35

5.04

79.61

75.64

15.3

13.7

Sweden

8.41

3.87

86.56

81.61

6.3

14.9

15.6

Switzerland

10.66

4.76

44.32

38.53

6.0

7.9

UK

4.12

1.21

74.51

59.69

8.3

6.0

33.3b

49.2b

Notes:

(a) only the Western Länder;

(b) not including Northern Ireland.

Sources: Gallagher et al. 2011; van Biezen et al. 2012; <www.idea.int>; Eurobarometer 10 (1978); European Election Study 2014.

(p.78) The other three indicators summarized in Table 5.1 are more outward-facing, focused on the political behavior of individual citizens and how that might relate to political parties. First there is the apparent growing indifference of citizens toward electoral politics as shown by declining electoral turnout, which is seen as a phenomenon of the last few decades in particular (e.g. Blais et al. 2004; Franklin 2004). The summary comparisons provided in Table 5.1 are consistent with more detailed analyses, showing how in most of the established democracies turnout is in decline and in some cases quite sharply. But there are exceptions: most of the Nordic countries (though not Finland) tend to somewhat buck the trend, in part perhaps reflecting lower degrees of social inequality, as does Belgium most likely due to its compulsory voting regime.

The other feature that stands out is the variation in the extent of voter abstention across all the cases: turnout is distinctly (and notoriously) lowest in Switzerland and France (the regularity of elections and referendums in the former and the semi-presidential nature of the latter being partially to blame); but in five of the cases (Belgium, Denmark, Italy, Norway, and Sweden) more than three-quarters of voters are still turning out in recent elections, and in two others (Austria and the Netherlands) the proportion is averaging more than seventy-four percent. As Franklin has suggested: “Perhaps, in the light [of such evidence] the question that we should be asking is not why is turnout (p.79) declining, but why is turnout so stable?” (2004: 11). Nevertheless, having said that, the trend in voter turnout is downward and most especially so in recent decades, suggesting, as Mair puts it, a growing sense of indifference toward the electoral process.

An additional indicator of change relates to the extent to which those who still vote are becoming more fluid or inconsistent in their voting behavior, as shown by average trends in aggregate electoral volatility. As is well known, this is an imperfect measure—it doesn’t allow for individual-level switching in countervailing directions, nor does it allow for demographic changes among the electorate, and it is prone to dramatic shifts due to particular national circumstances—but as the summary trends in Table 5.1 show, aggregate electoral volatility is on the rise: in most of the countries it is in double-digit territory in the 2000s compared to just three instances in the 1970s. Again according to Mair (2013: 29), volatility points to citizen withdrawal: “Hand in hand with indifference goes inconsistency.”

Finally, there is the well-known individual-level measure of partisan attachment. In his detailed analysis of cross-national trends through to the end of the 1990s, Dalton finds striking uniformity in the downward trajectory of party identification. As he notes: “[s]eldom is the public opinion evidence from such a diverse group of nations so consistent in following a general trend” (2004: 33). Schmitt, however, who includes the early years of the new century for some countries, is somewhat more nuanced: “Taken together, [there is] compelling evidence of a decline of partisanship. However, this decline is neither uniform nor universal” (2009: 81). And Hooghe and Kern (2015: 9), looking at the most recent decade note that “current trends are stable.” It is hard to find data that are truly comparable over time and the summary data provided in Table 5.1 should be interpreted cautiously (we return to this issue later in this chapter). Nevertheless, of the eight countries for which we can compare over time, three show lower percentages of voters who do not feel close to any party in 2014 than in 1978 (i.e. where party attachment increased rather than declined). The erosion of party loyalty to which Dalton (2004) and others (e.g. Mair 2013) refer as additional evidence of a threat to parties is no longer a uniform and consistent phenomenon.

The trends summarized in Table 5.1 are not the only indicators used to make the case that parties are in decline. Other indicators commonly cited include: the emergence of alternative actors in the electoral process, such as interest groups, candidate-support organizations (e.g. US political action committees), or independent candidates with their own electoral machines (Farrell and Schmitt-Beck 2008), the rise of candidate-centered, presidentialized politics (e.g. Poguntke and Webb 2005; Wattenberg 1991), the impact of globalization and of non-majoritarian institutions on party control over the policy process (e.g. Majone 2001).

(p.80) In combination it seems the omens are not good: the future for parties does not seem too bright. Fewer of us are party members; fewer of us vote in elections; of those of us who do vote we are more inconsistent in our voting behavior—though curiously perhaps we are still inclined to show some loyalty in our attachment to particular parties.

5.1.2 Reinterpreting the Evidence

But before we write the obituary for political parties as a species, it is worth reflecting on a few things, the first of these being the fact that we’ve been here before. The refrain of political parties in decline is by no means a new one. Scholars have been warning about this for generations in the US, perhaps most notably in the influential “responsible parties” report issued by the American Political Science Association’s Committee on Political Parties published in the 1950s. By the early 1980s Crotty and Jacobson were warning that “a partyless era, with implications still uncertain, may be settling on us” (1980: 255; see also Kirkpatrick 1978; Wattenberg 1998). Outside the US the suggestion that parties may be “failing” is relatively more recent, but it still dates back over three decades. As Lawson and Merkl note in the introduction to their influential comparative study: “The phenomenon of major party decline, often remarked in the context of the American political system, is becoming increasingly apparent in other political systems as well” (1988: 3; also Reiter 1989).

One explanation for the long-term persistence of the “party decline” discourse is that this decline is not comparing the current empirical situation of parties with a historical empirical situation. We should not overestimate the degree to which the parties ever enjoyed the supposed Golden Age of “mass party” politics personified by Duverger’s classic account (1964). Scarrow’s careful analysis shows how Duverger’s model was more an ideal type than a widespread political reality. As she notes: “democratic parties with large individual memberships have been relatively rare” (2015: 67).

A second observation about the decline of parties and about the downward trends in party membership in particular, relates to the meaning of “party membership” in the modern age. As we have seen, most parties have experienced a drop in dues-paying members over the past two decades, but at least some of this could reflect a reconfiguring by citizens as well as by parties of what it means to be a party supporter. In her recent study, Scarrow refers to a growing tendency for parties to pursue “a multi-speed approach to party membership” (2015: 128) in which the traditional paid-up membership is supplemented by alternative ways of showing support such as by donating to parties without joining them or signing up to party’s e-newsletters, engaging in parties’ policy forums, and so on. Scarrow’s analysis indicates that many parties (p.81) have been experimenting with ways of making “it easier for their supporters to connect with them” (p. 151)—perhaps one factor that might help to explain the leveling off of party identification trends in recent years.

A third qualification of the party decline thesis is that the analyses tend to downplay other dimensions of party activity. While it may be argued—as we have suggested—that those who contend that parties are in decline perhaps over-fixate on features of party design that are of a bygone age, equally we might suggest that insufficient attention is paid to some of the classic functions of parties (e.g. Almond 1960)—functions that, in many respects parties are still fulfilling. To borrow Kay Lawson’s terminology (1980) we can refer to the “linkage” role of parties that distinguishes them from other organizations, marking them out as the primary representative agents between citizens and the state.

In many respects, parties still provide key linkage functions in our representative system of government.4 To this day parties continue to play a key role in the election campaign process. They control the selection of candidates for election (with few exceptions, the vast bulk of candidates are nominated by political parties; e.g. Hazan and Rahat 2010) and dominate the political discourse of campaigns (Farrell 2006). They also play a key role in turning out the vote at elections (e.g. Karp et al. 2008; Karp and Banducci 2011). Even in an age of declining voter turnout, most citizens still continue to vote in election after election, and studies provide clear evidence that parties play an important role in this regard.

Parties also help voters determine who to vote for in an election. Analysis reported in Dalton et al. (2011) shows that many voters still view their political preferences through the lens of the left–right policy continuum, even in a time when some differences between the major parties may be becoming less noticeable. Without denying the rise of new issues and parties that are less clearly linked to the left–right continuum, it is important to note that voters as a collective can still correctly identify most parties’ locations on that continuum. And having done so, for the most part the voters use this information to make informed choices between the parties. In short, the ideological or policy congruence between voters and parties (e.g. Huber and Powell 1994) continues to be an important determinant of voter choice.

Finally, there is substantial evidence that parties provide a meaningful policy linkage between citizens and the state, supporting the perspective that it “matters” which party or coalition of parties are in office (Castles 1982). Voters seek out parties for their policy goals and for the most part, parties implement those goals when they gain office (e.g. Klingemann et al. 1994; Benoit and Laver 2006). In short, the policy outputs of governments are broadly consistent with the ideological profiles of the parties that form them.

(p.82) In many respects, therefore, it could be argued that the classic notion of democracy as “party democracy” (Castles and Wildenmann 1986) still applies. In a typically forthright piece on the fate of parties, Philippe Schmitter (hardly a fan of the genre) opined that “parties are not what they once were” (2001). That may well be true, but does it inevitably lead to the conclusion that parties are in jeopardy as a consequence? As Michael Saward has observed: that parties may not be what they once were “does not necessarily mean that they are less than they once were” (2010: 133 emphasis in the original).

Finally, we should point out that even those aspects of party decline that are not in dispute need not necessarily indicate that “citizens are heading for the exits of the national political arena” (Mair 2013:43). Clearly the decline in party membership in recent decades is an important change in the dynamic of party politics, but whether blame for this should be attached to the parties is questionable. As Richard Katz has observed: “there are good grounds to believe that…the decline has been a by-product of social changes that neither can—nor in most cases should—be reversed” (2013: 63). Certainly, the party membership trends need to be considered in the wider context of societal change that has affected institutions beyond political parties: societal change has contributed to a breakdown of collective identities as citizens become increasingly individualized (Andeweg 2003; Putnam 2000).

Other indicators of party decline may even point to increased political engagement rather than disengagement. Electoral volatility, for example, unquestionably has made individual political parties more vulnerable, but at the same time it could be argued that it has made party democracy as a whole stronger. After all, liberated from their subcultural shackles, voters are finally beginning to choose (Rose and McAllister 1986), as they are expected to in a mature democracy.

There is also some evidence of ongoing adaptations to our democratic institutions to suit new styles of political participation by our citizens that may compensate for any party decline that has taken place. For instance, as Russell Dalton and his colleagues observe: “Although electoral participation is generally declining, participation is expanding into new forms of action” (2003: 1) as more of us engage in new, less conventional (sometimes even unconventional) forms of political action, as more of us become “good” (Dalton 2009) or even “critical” citizens (Norris 1999a), seeking a more active (less passive) role in the political system, prepared to challenge (and thereby engage with) existing systems and norms. In short, even if it may be the case that citizens are turning away from some forms of conventional politics (including party politics) that does not mean that they’re giving up on all forms of political activity.

(p.83) 5.2 Partisanship and Political Support

5.2.1 Party Membership and Party Closeness

These observations question the correlation between partisanship and political support at the individual level, and we now look more closely at that correlation. To the extent that involvement in a political party socializes citizens into politics, helps them identify with the political system, and assists them in making sense of politics, the weakening appeal of party involvement is likely to negatively affect political support. As such, the undisputed decline of formal party membership is an indicator of eroding partisanship, but the proportion of dues-paying party members in the electorate was never very high in most countries. Formal membership is likely to underestimate partisanship. For that reason, we include a less formal type of partisanship: party closeness. Here the longitudinal data are murky, but we cautiously concluded that the more recent measurements show a leveling off, or even a recovery, in many countries. We should state at the outset, however, that party closeness is not only less formal, but also more ambiguous in its meaning. It is often interpreted as a form of party attachment or even identification—a kind of psychological membership, but it can also be regarded as a different dimension of partisanship—of ideological proximity to a particular political party. We return to this point in our discussion of the data analysis later in the chapter.

To investigate the impact of these two variables, we make use of the first five rounds of the European Social Survey, covering the 2002–10 period, including all sixteen countries for which data are available from all five points in time.5 Rounds six and seven (2012, 2014) were not used because they do not contain a question about party membership. Although all analyses were done on all sixteen countries and on the thirteen Western European democracies alone, we present the results only for the latter as the focus of this volume is on well-established democracies. However, the differences between the analyses including and excluding Hungary, Poland, and Slovenia are negligible. All analyses were also done for each individual country, and for each individual ESS round, separately, and reported whenever noteworthy.

As indicators of legitimacy, we use the degree of satisfaction with the way national democracy works (political support at the level of the regime), and trust in parliament (political support for democratic institutions), both measured on an eleven-point scale. We did not use satisfaction with the national government as it is likely to be affected by one’s sympathy for a governing or opposition party. We also did not use trust in politicians and trust in political parties because they refer to whole categories of political actors of which some members are likely to be trusted more than others (Andeweg 2014: 184–5).

(p.84) Our two main independent variables are party membership and party closeness. Party closeness is measured by two survey questions: whether there is a particular political party to which the respondent feels closer than to any other party, and if so, how close the respondent feels to that party. We combined the answers to these two questions into a five-point scale ranging from “there is no party closer than any other party,” to “one party is very close to me.”

To mitigate the risk that we are measuring spurious relations we also include some of the “usual suspects” in explanations of individual variation in political support. Whether one is interested in politics or not (four-point scale) is such an important predictor at the attitudinal level. Economic satisfaction (i.e. satisfaction with the present state of the national economy, eleven-point scale) is also frequently mentioned as such (but see Chapter 9). Social background factors that may play a role are gender, age (in years), and level of education (five-point scale). Finally, we also include the disposition to trust or to distrust, measured by a question whether the respondent feels that most people can be trusted, or that you can’t be too careful in dealing with people—usually called “generalized trust” or “social trust” (eleven-point scale). There is evidence that a trustworthy government may not only generate political trust, but also facilitate the development of social trust, resulting in a spurious correlation between social and political trust (Levi and Stoker 2000: 493–5; Newton 2007; Zmerli and Newton 2011). The inclusion of generalized trust in the final model assumes that its role as an independent variable is dominant, but doubtful readers can simply ignore Model IV in our tables.

Tables 5.2a and 5.2b show the results of a multilevel regression analysis in which the clusters are added stepwise from Model 0 which only includes individual-level, country-level, and ESS round-level variation to Model IV which includes partisanship, political interest, and economic satisfaction, social background, and generalized trust.

Table 5.2a The effect of partisanship on satisfaction with the way national democracy works

Model 0

Model I

Model II

Model III

Model IV

Party membership (yes = 1)

0.079*

−0.035

−0.018

−0.022

(0.035)

(0.031)

(0.031)

(0.031)

Party closeness

0.188***

0.109***

0.116***

0.108***

(0.005)

(0.004)

(0.004)

(0.004)

Political interest

0.176***

0.175***

0.142***

(0.007)

(0.007)

(0.007)

Economic satisfaction

0.452***

0.451***

0.424***

(0.003)

(0.003)

(0.003)

Gender (male = 1)

0.015

0.021

(0.012)

(0.012)

Age

−0.003***

−0.003***

(0.000)

(0.000)

Education level

0.007***

0.004*

(0.002)

(0.002)

Generalized trust

0.128***

(0.003)

Intercept

5.729***

5.252***

2.780***

2.906***

2.413***

(0.269)

(0.260)

(0.144)

(0.145)

(0.133)

Individual-level variance

2.492

2.466

1.962

1.961

1.914

Country-level variance

0.903

0.837

0.200

0.200

0.161

ESS round-level variance

0.015

0.013

0.024

0.024

0.023

AIC1

670,007

667,877

635,942

635,764

633,260

N

124,563

118,555

117,078

116,371

116,202

Notes: The entries are parameter estimates and standard errors in parentheses of a multilevel linear regression. All models include random intercepts for individual respondents on the first level, thirteen Western European countries (Belgium, Denmark, Finland, France, Germany, Ireland, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the United Kingdom) on the second level, and five ESS rounds (2002, 2004, 2006, 2008, and 2010) on the third level. The data are weighted using ESS post-stratification and population size weights.

(1) To allow for comparison between the Akaike Information Criterion (AIC) of the various models, they have been calculated including only the 116,202 respondents for whom information on all variables is available.

Table 5.2b The effect of partisanship on trust in parliament

Model 0

Model I

Model II

Model III

Model IV

Party membership (yes = 1)

0.329***

0.133***

0.158***

0.154***

(0.035)

(0.032)

(0.032)

(0.032)

Party closeness

0.247***

0.125***

0.134***

0.122***

(0.005)

(0.005)

(0.005)

(0.004)

Political interest

0.408***

0.409***

0.360***

(0.007)

(0.007)

(0.007)

Economic satisfaction

0.396***

0.395***

0.357***

(0.003)

(0.003)

(0.003)

Gender (male = 1)

−0.045***

−0.038**

(0.012)

(0.019)

Age

−0.005***

−0.005***

(0.000)

(0.000)

Education level

0.010***

0.006***

(0.002)

(0.002)

Generalized trust

0.184***

(0.003)

Intercept

4.972***

4.338***

1.690***

1.909***

1.206***

(0.230)

(0.214)

(0.130)

(0.131)

(0.121)

Individual-level variance

2.335

2.285

1.883

1.877

1.796

Country-level variance

0.624

0.532

0.108

0.108

0.080

ESS round-level variance

0.024

0.022

0.040

0.040

0.038

AIC1

661,448

657,868

632,979

632,692

628,065

N

125,193

119,159

117,336

116,634

116,478

Notes: The entries are parameter estimates and standard errors in parentheses of a multilevel linear regression. All models include random intercepts for individual respondents on the first level, thirteen Western European countries (Belgium, Denmark, Finland, France, Germany, Ireland, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the United Kingdom) on the second level, and five ESS rounds (2002, 2004, 2006, 2008, and 2010) on the third level. The data are weighted using ESS post-stratification and population size weights.

The overall patterns are quite similar for both democratic satisfaction and trust in parliament. That nearly all coefficients are statistically significant need not surprise us given the huge number of respondents in the pooled data set of five rounds of surveys in thirteen countries. The decline of the AICs from one model to the next shows that adding variables does increase the models’ fit. However, comparing Model 0 to Model I—the most important step for our purposes—the contribution of party membership and party closeness to the fit can only be described as modest. It is definitely more substantial than the negligible contribution of social background to the fit, but the inclusion of political interest and economic satisfaction improve the model’s fit much more. It would seem that the assumption that the role of political parties as a linkage mechanism between individual citizens and the political system should translate into a correlation between partisanship and political support finds only modest support in the data.

(p.85) The separate analyses by country and by ESS round (not presented) confirm this finding. Although the R2s are slightly higher in 2010 than in 2002 there is not a linear trend between the first and last round of ESS. The patterns do not vary much across countries, but the variation that is explained by the models does: they perform worst in explaining trust in parliament and democratic satisfaction in Switzerland, and this seems largely due to a different role of partisanship there: Model I, for example, produces an R2 of only 0.005 for trust in parliament and 0.008 for satisfaction with the functioning of democracy in Switzerland. Factors most likely accounting for this finding include the fact that the Swiss Zauberformel for coalition formation weakens the impact of electoral choice on the government’s composition, and the regular use of the referendum which offers a non-partisan linkage between citizens and public policy. Model I performs best in Germany, with R2s of 0.027 and 0.038 for trust and satisfaction respectively. Adding variables does not change (p.86) the conclusion with regard to Switzerland, but Germany shows average results for Model IV. For that Model, R2 is highest in Hungary (0.347 for democratic satisfaction and 0.261 for trust in parliament).

5.2.2 The Meaning of Party Closeness

So far, we have not distinguished between party membership and party closeness. However, whereas the parameter estimates for party closeness are significant and positive in all models for both dependent variables, party membership is no longer significantly related to democratic satisfaction in three of the four models and the sign changes to negative, indicating that, if anything, party members are less rather than more satisfied with the way democracy works. The part correlation coefficients in a multivariate analysis confirm that of these two variables, it is party closeness, not party membership that has the stronger (positive) impact (see Table 5.3).

Table 5.3 Relative impact of party membership, party closeness, and other variables on democratic satisfaction and trust in parliament (part correlation coefficients)

Satisfaction with working democracy

Trust in parliament

Party membership

−0.001

0.014

Party closeness

0.069

0.097

Political interest

0.049

0.103

Economic satisfaction

0.418

0.327

Gender

0.002

−0.007

Age

−0.027

−0.034

Education level

−0.005

−0.004

Generalized trust

0.131

0.184

(p.87) The part correlation coefficients indicate how much each independent variable contributes to the overall explained variance. The importance of economic satisfaction and generalized trust is reaffirmed, as is the weak impact of the social background variables. Party closeness is the third- (for satisfaction) or fourth- (for trust) strongest independent variable, with party membership having but a marginal impact on the two measures of political support (and a negative effect on democratic satisfaction). The separate analyses by country and by ESS round show that this difference between the two party variables is present in all countries, and in each individual round of the European Social Survey. Hooghe and Kern (2015), using the same data but a different country selection, a differently constructed dependent variable, and a different selection of independent variables, have reached the same conclusion: the perception that one political party is clearly closer to you than the others is much more important for one’s political trust than being a member of a party.

This warrants a closer look at the variable of party closeness. We already mentioned that there are two interpretations possible. The most common interpretation sees survey questions about party closeness as measuring an informal party attachment or identification. As originally developed by the Michigan School, party identification refers to a strong psychological bond, with belonging to a party as part of an individual’s self-reported identity: “Generally speaking, do you usually think of yourself as a Republican, a Democrat, or what?” As such, party identification did not travel well beyond the US (and to some extent the UK) context and it was reconceptualized as a positive attitude toward a political party, more as a summary measure of ideological preferences and long-term experiences than as part of one’s identity (Bartle and Bellucci 2009). As a term, party identification continues to be used, but within this attitudinal approach we can see a convergence, at least in Europe, on a different measurement: whether there is a party that one feels close to (and how close). However, the question wording has not always been (p.88) identical. The Eurobarometer surveys, for example, asked the question “Do you consider yourself to be close to any political party” and if the answer was affirmative, the respondent was asked whether (s)he felt “very close, fairly close, or merely a sympathizer.” That last answering category refers to a feeling of attachment, rather than to perceived distance. This is even clearer in some of the translations that were used in other languages: in the French version “very/fairly close” was translated as “tres/assez attaché,” in German the word “verbunden” was used, and so on. In Table 5.1 we compared answers to the Eurobarometer question from the 1970s to European Election Studies from 2009 and 2014, because the Eurobarometer ceased asking about party closeness in 1994 and the European Election Study uses virtually identical questions and answering categories. The questions asked in the various rounds of the European Social Survey that we used in the other tables in this chapter have the advantage that they do not combine a question about perceived closeness with answering categories that indicate strength of attachment. The questions are exclusively about the degree of closeness, also in the translated versions of the questionnaire.

This is not the place for a lengthy methodological discussion of the pros and cons of different question wordings (for that see Sinnott 1998), but the substantive point to make is that there is considerable reason to doubt that questions about party closeness measure party identification or even party attachment, as is usually assumed. To the extent that the question wording also uses terms that refer to attachment, it may still be interpreted in that way albeit cautiously. But when the question wording is only about the (degree of) closeness to a party (“Is there any particular party you feel closer to than all other parties? Very close, quite close, not close, not at all close”) we prefer to take the question literally: as a question about perceived proximity in terms of ideological position of one party compared to other parties, rather than about party identification. In that sense it is similar to questions used in national election studies about the feeling that one’s views are represented by a particular party (see Holmberg 2014). This implies that party closeness refers to a different linkage mechanism than party membership. It is not a measurement of informal membership of a party, but of the perception that the supply of political parties collectively facilitates one in making a meaningful political choice (Wessels and Schmitt 2014). To the extent that all parties are seen as equidistant or very close to each other, the party system cannot perform its function to link a citizen to the political system. It is in this sense that also Hooghe and Kern (2015) interpret the question about party closeness. The fact that we did not find a marked decline of party closeness, and that the impact of party closeness on legitimacy is stronger than that of party membership, which has declined significantly, points to the importance of differentiating between the ways in which parties can link citizens to the state.

(p.89) 5.2.3 Populist Party Support and Political Support

Before we conclude that the very linkage function of political parties that shows fewest signs of erosion—offering citizens a meaningful choice—is also the linkage function that contributes most (in relative terms) to political support, we should note that we have not distinguished between political parties. We have assumed that feeling closer to any political party than to other parties provides the citizen with a linkage to the political system and contributes to that citizen’s belief that the political system is legitimate. A counterargument would be that this reasoning does not apply to antisystem parties. After all, such parties seek to convince citizens that the system is not legitimate and they do not aspire to act as a linkage between that system and the citizen. Citizens who are members of or voters for an antisystem party are unlikely to be satisfied with the way democracy works or to trust political institutions.

Traditional antisystem parties such as communist or (neo-)fascist parties have become rare while populist parties are in ascendance. Beyond the Manichean distinction between the “pure” people and the “corrupt” elite (Mudde 2004), populist parties do not have much in common, but most of them do not seem to be as opposed to the democratic political system as traditional antisystem parties once were. They prefer a different, more “populist,” and less “constitutional” democracy (Mény and Surel 2002), but they do not seek to abolish democracy. Nevertheless, they are generally opposed to the intermediary role between the people and the formulation of public policy that political parties are supposed to play—“populist democracy vs. party democracy” in the words of Peter Mair (2002). In that sense populist parties are sufficiently “anti” to expect their supporters to lack satisfaction with and trust in contemporary party democracy (but see Fieschi and Heywood 2004 for a more nuanced view).

The pluriformity of the populist party family has prevented a consensus on the classification of political parties as being populist parties or not. Here we follow one of the most recent attempts at identifying populist parties in European party systems (Van Kessel 2015: 33–73, 144–68), even though some of his classifications must remain open to discussion (such as the exclusion of the Socialist Party in the Netherlands).6 We confine the analysis to countries in which populist parties had significant electoral support, which means that we had to exclude Spain and Portugal from the selection of countries, leaving fourteen countries in the analysis. Again, the data presented in Table 5.4 refer only to the eleven established Western European democracies.

Table 5.4 The contribution of a populist vote to satisfaction with the way national democracy works and trust in parliament

Democratic Satisfaction

Democratic Satisfaction

Trust in Parliament

Trust in Parliament

Model IV

Model V

Model IV

Model V

Party membership (yes = 1)

−0.066*

0.059

0.190***

0.196***

(0.032)

(0.032)

(0.032)

(0.032)

Party closeness

0.114***

0.117***

0.136***

0.139***

(0.005)

(0.005)

(0.005)

(0.005)

Political interest

0.161***

0.162***

0.353***

0.355***

(0.008)

(0.008)

(0.008)

(0.008)

Economic satisfaction

0.431***

0.428***

0.349***

0.346***

(0.003)

(0.003)

(0.003)

(0.003)

Gender (male = 1)

0.035**

0.048***

−0.020

−0.007

(0.013)

(0.013)

(0.013)

(0.013)

Age

−0.004***

−0.004***

−0.007***

−0.007***

(0.000)

(0.000)

(0.000)

(0.000)

Education level

−0.006**

0.005**

0.007***

0.007***

(0.002)

(0.002)

(0.002)

(0.002)

Generalized trust

0.137***

0.134***

0.192***

0.189***

(0.003)

(0.003)

(0.003)

(0.003)

Voted for populist party (yes = 1)

−0.766***

−0.754***

(0.031)

(0.031)

Intercept

2.329***

2.378***

1.255***

1.304***

(0.129)

(0.134)

(0.127)

(0.128)

Individual-level variance

2.151

2.134

2.033

2.026

Country-level variance

0.127

0.141

0.080

0.082

ESS round-level variance

0.021

0.021

0.040

0.039

AIC

452,217

446,905

447,475

446,905

N

99,590

99,985

Notes: The entries are parameter estimates and standard errors in parentheses of a multilevel linear regression. All models include random intercepts for individual respondents on the first level, eleven Western European countries (Belgium, Denmark, Finland, France, Germany, Ireland, the Netherlands, Norway, Sweden, Switzerland, and the United Kingdom) on the second level, and five ESS rounds (2002, 2004, 2006, 2008, and 2010) on the third level. The data are weighted using ESS post-stratification and population size weights.

Table 5.4 first repeats Model IV from Tables 5.2a and 5.2b. Because the analysis is now based on eleven rather than thirteen countries, the coefficients are slightly different from those in Tables 5.2., but the patterns are quite similar. Models V in Table 5.4 add having voted for a populist party or not, and this variable indeed is strongly and negatively related to satisfaction with the (p.90) working of democracy and trust in parliament. However, the separate analyses for each country show considerable variation in this respect, with some countries showing an insignificant relation between a populist vote and democratic satisfaction or trust in parliament (e.g. Denmark and the UK). Apparently, the heterogeneity of populist parties also translates into a less than uniform relationship between voting for such a party and political support.

5.2.4 The Direction of Causality

What is also striking in Table 5.4 is that the inclusion of a populist vote does not improve the fit of the model by much. For trust in parliament in (p.91) particular, the decrease of the AIC for Model V compared to Model IV is insubstantial. This is also visible in each ESS round and in each individual country. The effect of a populist vote on political support appears to overlap substantially with the combined effect of the other independent variables in the model.

This raises questions about the causal relationship between populism and political support. Populism may lead to lower political support and directly or indirectly also reduce partisanship, political interest, economic satisfaction, and generalized trust. Or low partisanship, low interest, low economic satisfaction, and low generalized trust reduce political support, which in turn increases the probability of a populist vote. Both causal directions are plausible, but they are difficult to disentangle. Van der Brug, for example, explains the sudden emergence of the populist List Pim Fortuyn (LPF) in the Netherlands in 2002 by suggesting that:

preferences for the LPF are not caused by discontent, but the effect is reversed. It is plausible that someone acquires a preference for the LPF, because he/she agrees with the party on some of the main issues. The LPF consistently proclaimed that the political elite had lost touch with the feelings and ideas of the common man and that this elite could not be trusted. It seems a logical possibility therefore that this consistent message made the LPF supporters more cynical (Van der Brug 2003: 100).

Van der Zwan, on the other hand, argues that political cynicism and in particular political inefficacy (also seen to be affected by a vote for the LPF by Van der Brug) are basic attitudes that are unlikely to be endogenous to a change in opinion such as a decision to vote for a populist party (Van der Zwan 2004; rejoinder by Van der Brug 2004). Finally, Bélanger and Aarts (2006) use panel data covering but also predating the 2002 elections to show that voters who were already “discontented” (inefficacious and cynical) in 1998 were more likely to vote LPF in 2002. The level of cynicism and discontent did increase in 2002, but among all voters, not just among LPF supporters. This implies that it is discontent that is the cause and the populist vote is the consequence.

The debate on the rise of the LPF in the Netherlands as cause or consequence of low political trust or democratic satisfaction does not stand alone. Paskeviciute (2009) examines data from the United States, seven established European democracies, and five new European democracies, linking parties’ support for the constitutional order as expressed in these parties’ manifestos to the democratic satisfaction and external political efficacy of the parties’ voters, and calling this a party persuasion effect. Similarly, Anderson and Just (2013) use data covering fifteen countries and find that office-seeking parties express more positive attitudes toward the political system than policy-seeking (p.92) parties do, and that this difference corresponds to different levels of democratic satisfaction and external political efficacy among supporters of these parties. They conclude that political parties communicate their attitude toward the system to their supporters, and that the political support of these supporters is therefore endogenous. On the other hand, Bélanger (2004) uses data from three countries (Canada, the United Kingdom, and Australia) and finds that general antiparty feelings (as opposed to negative feelings towards a specific party) contribute to absenteeism unless a political party channels such feelings. If we may generalize from antiparty to antisystem feelings, political support is thus treated as exogenous to the existence of such third parties by Bélanger. Similarly, Denemark and Bowler (2002) compare the electoral appeal of populist and other minor parties in Australia and New Zealand. Finding correlations between a vote for minor parties and political disaffection, they conclude that such parties “represent,” “reflect,” or “tap into” disillusionment with the political system. In summary, we see that authors interpret similar correlations between voting for a populist party and expressing low political trust and satisfaction in opposite ways. Without the kind of individual before–after design that Bélanger and Aarts (2006) were able to use in the Dutch case, it will be impossible to settle this dispute in general.

This is unfortunate, because the direction of causality affects not only the relationship between populism and political support. It goes to the heart of the relationship between partisanship and political support in general. Throughout this chapter we have assumed that parties contribute to political support by acting as a linkage mechanism between individual citizens and the political system. Hence we have treated party membership and party closeness as independent variables, and satisfaction with the working of democracy and trust in parliament as dependent variables. But it could just as easily be viewed the other way around: citizens may no longer see the political system or its institutions as legitimate for other reasons (economic adversity, media malaise, etc.) discussed in other chapters in this volume, and that loss of confidence then leads them to resign their party membership and to see all political parties as Tweedledum–Tweedledee. This is the perspective taken by Dalton (2000: 34–5), when he discusses a “performance model” to explain party dealignment. He uses satisfaction with the working of democracy as an indicator of how well political parties are perceived to perform, and analyzes its impact on strength of partisanship on the basis of the 1976 and 1992 Eurobarometer surveys. The fact that democratic satisfaction is a rather distant proxy of satisfaction with the performance of parties, and the fact that Dalton found only a weak and inconsistent relationship at the time do not mean that the direction of causality cannot plausibly be seen in two ways.

(p.93) 5.3 Conclusion

“[W]hile Schattschneider’s proposition is usually taken to mean that the survival of democracy will guarantee the survival of parties…we can also read it the other way around, suggesting that the failure of parties might indeed imply the failure of democracy” (Mair 2013: 14–15).

As we discussed earlier, Peter Mair was not alone in arguing that political parties are in decline and with them representative democracy as we know it. But the evidence presented in this chapter suggests otherwise. In the first instance, there are doubts over whether political parties are actually in decline. There is certainly evidence of change (fewer members, fewer voters, more volatile voters), but not in all respects (in many countries large proportions of voters still vote; the supposed erosion of party loyalty is far from uniform), the trends are not unique to parties (erosion of memberships is widespread across society), the parties are proving to be quite adaptive (not least in how they treat the notion of “membership;” see Scarrow 2015), and they continue to play a central role as linkage agents for citizens in our representative system of government (Dalton et al. 2011).

Not only do we not find much evidence of a decline of parties in terms of their elite–mass linkage roles, there are also important questions to be raised over the degree to which support for parties is related to political support. What relationship there is seems pretty slight and relates more to measures of party closeness than to party membership. If our interpretation of party closeness as an indication of the degree to which the party system allows the voter to choose among the parties, rather than as a measure of commitment to an individual party, is correct, our findings suggest that political parties continue to contribute to support for the democratic system and its institutions not by individually engaging citizens in party activities, but by collectively offering citizens a meaningful choice.

It is not just the weakness of the relationship that is at issue; so, too, is the direction of causality. There is already discussion over whether the rise of populism is a cause or a consequence of declining political support, but this debate can be extended easily to the relationship between partisanship and political support in general. With the type of cross-sectional data that are available, it is not possible to decide whether a decline in partisanship leads to a weakening of political support or the other way around. But it is at least plausible that political distrust and dissatisfaction make people turn away from political parties, rather than that resigning one’s party membership causes a drop in trust and satisfaction. Contrary to Mair who wondered whether the failure of parties, that he believed was happening, might “imply the failure of democracy” (2013: 15), it would seem that, if anything, it is democracy’s failings that might, in time, detrimentally impact on the parties. In that sense, at least, Schattschneider’s “unthinkable” could yet come to pass.

Notes:

(1.) We are grateful to Tim Mickler for his assistance and to the anonymous reviewer for helpful feedback on an earlier draft. The usual disclaimer applies.

(2.) While Table 5.1 provides a useful summary of trends we are aware that the two time period presentation over-simplifies things and may mask important time-related or country-specific features.

(3.) See also <http://www.politicalpartydb.org>. In her recent book-length study of the subject, Kölln (2014) finds evidence of cases that buck the trends of membership decline, but on the whole her evidence also points to a general downward tendency in overall party membership.

(4.) The following paragraphs briefly summarize the key findings of Dalton et al. (2011).

(5.) Belgium, Denmark, Finland, France, Germany, Hungary, Ireland, the Netherlands, Norway, Poland, Portugal, Slovenia, Spain, Sweden, Switzerland, and the United Kingdom.

(6.) The following parties are classified as populist: Belgium (Vlaams Blok/Vlaams Belang; Front National; Lijst Dedecker); Switzerland (Schweizerische Volkspartei; Lega dei Ticinesi; Schweizer Demokraten; NB: the populist Mouvement Citoyens Genevois was not coded separately in the ESS Survey); Germany (Republikaner; Partei des Demokratischen Sozialismus; Die Linke; Nationaldemokratische Partei Deutschlands; Deutsche Volksunion); Denmark (Dansk Folkeparti; Fremskridtspartiet); Finland (Perussuomalaiset: “True Finns”); France (Front National); Hungary (“Justice and Life;” “Movement for a Better Hungary;” FIDESZ (after 2006)); United Kingdom (British National Party; UK Independence Party); Ireland (Sinn Féin); the Netherlands (Lijst Pim Fortuyn; Leefbaar Nederland; Partij voor de Vrijheid; Trots op Nederland); Norway (Fremskrittspartiet); Poland (“Self Defence;” “Law and Justice”); Slovenia (Slovenian National Party); and Sweden (Sverigedemokraterna).