Party Rules, Party Resources, and the Politics of Parliamentary Democracies: How Parties Organize in the 21st Century 1

Transcription

1 Party Rules, Party Resources, and the Politics of Parliamentary Democracies: How Parties Organize in the 21st Century 1 Thomas Poguntke, Heinrich-Heine University, Germany Susan Scarrow, University of Houston, USA Paul D. Webb, Sussex University, UK Paper prepared for the ECPR General Conference, Université de Montréal, August 2015 ABSTRACT This paper introduces the first descriptive findings of the Political Party Database (PPDB) project, a major survey of party organizations in parliamentary and semi-presidential democracies. The survey takes in 122 parties in 19 countries, covering numerous variables. This paper describes the scope of the database, before presenting a number of initial findings relating to the resources, structures and internal processes of parties. Results are broken down by country and party family, and where possible, temporal comparisons with older datasets are made. Our analyses suggest a remarkable coexistence of uniformity and diversity. In terms of the major organizational resources on which parties can draw, such as members, staff and finance, the new evidence largely confirms the continuation of trends identified in pervious research, ie, declining membership, but enhanced financial resources and paid staff. Yet, substantial differences between party families and, more importantly, between counties persist. On the other hand, we find remarkable uniformity regarding the core architecture of party organizations. If they survive, parties adapt their organizational skeleton to what are apparently the functional exigencies of modern party politics. At the same time, however, we find substantial variation between countries and party families in terms of their processes, with particular regard to how internally democratic they are. 1

2 Introduction How do parties organize, and how much do parties organizational differences matter? The aim of the Political Party Database Project (PPDB) is to provide systematic answers to the first question so that we are better able to answer the second question, the crucial so what? question about party organizational variations. Other questions we seek to answer are to what extent, and why, do parties retain certain structural features despite changes in their competitive environments? For instance, are some traditional organizational features of parties in parliamentary democracies outmoded, such as party conferences and party membership, merely quaint relics and nostalgic remnants? To use Bagehot s terminology (1963/1867), have parties extra-parliamentary organizations become the dignified elements of party constitutions, with the real work of party politics being done by the efficient parts of the organization, be these the professionalized party staffs or the party officeholders? Bagehot wrote that the dignified elements were theatrical and often old elements which helped to excite and preserve the reverence of the population (61); similarly, some party practices might be remnants of earlier conditions which nevertheless contribute to the legitimacy of party government. Generating legitimacy is not a small thing, as Bagehot himself was aware, but is this all that extra-parliamentary party organizations contribute to contemporary politics? Another possibility is that parties retain these institutions because they continue to contribute in other ways. We do not expect to find simple or universal answers to any of these questions, but we do expect that we can gain traction in answering them by using systematic data to test posited relationships. This conviction has inspired the establishment of the PPDB. In the remainder of this article we introduce this database, and present some of our initial findings regarding the resources and structures of contemporary party organizations in 19 parliamentary and semipresidential democracies. I. The Long Tradition of Comparative Party Scholarship: Concepts, Categories and Data The comparative study of political parties extra-legislative organizations and activities is more than 150 years old, having arisen alongside the emergence of political parties as organizations that mobilize popular support. Many of the classics in the field were primarily descriptive and categorizing (for instance, Lowell 1897; Overacker 1932; Pollock 1932), but some offered more theoretical arguments about the origins and functioning of political parties (including Ostrogorski 1902 and Michels 1912). In the middle third of the 20 th century, the comparative study of political parties was stimulated and re-defined by authors who ambitiously constructed new categories and who posited causal theories to explain organizational differences between political parties in multiple democracies, and to explain changes over time (including Neumann 2

3 1954; Duverger 1954; Heidenheimer 1964; Lipset and Rokkan 1967; Kirchheimer 1966; Epstein 1968; Sartori 1976) More recent contributors have continued to develop this approach (cf Ware 1987; von Beyme 1988; Panebianco 1988; Schlesinger 1994). A common feature of these multicountry studies of party practices, particularly the older ones, is that they are largely nonquantitative, relying heavily on thick description to buttress their arguments. Their authors are polymaths who knew a great deal about the history and politics of multiple parties and countries, and who laboriously pieced together elusive evidence on party origins and organizations. Some of their most enduring contributions are now-familiar labels (e.g., mass, catch-all, electoral professional, etc.) Much of the theoretical speculation in these classic studies treats parties and party organizations as dependent variables, explaining how contemporary parties bear the marks of their origins, and how organizational differences reflect institutional contexts and ideological (party family) similarities. Some researchers in the 1950s and 1960s sought to construct theories of political parties that would apply across a wide range of regime types (including nondemocratic regimes) (e.g., the contributions in LaPalombara and Weiner 1969). For the most part, however, it is the mid-century comparative parties literature which focused on parliamentary democracies - the most similar systems approach - which has had the most lasting impact. Echoing more general trends in political science, recent decades have witnessed the rise of more systematically evidence-driven studies of political parties organizations and activities outside the legislative arena. Research of this type usually relies on party statutes and documents for evidence about party structures, and/or on expert judgments about how parties actually work. One notable investigation that combined both approaches was Kenneth Janda s pioneering study of party organization and practices in 53 countries (1980). This massive endeavor categorized and quantified information that was sub-divided into eleven broad aspects of party life (e.g., autonomy, centralization of power, etc). In his preface to the volume, Janda challenged others to follow him in constructing a more truly cross-national study of political parties, one that went beyond a most-similar-systems approach. Janda and his colleague Robert Harmel later proposed a different framework for collecting and interpreting data about party organizational change, one that was aimed more squarely at understanding practices in democratic regimes (1994). They collected only a limited amount of data using this framework, but other scholars continue to adopt their categories for their own data collection efforts. The 1980s also brought the start of another ambitious effort to gather cross-party and longitudinal data on party organizational development, what became the 12-country Party Organizations: a Data Handbook on Party Organizations in Western Democracies, (Katz and Mair 1992). This effort focused on what the editors dubbed the official story approach, primarily reporting published data and formal rules. Data collected in this Handbook have been used for a variety of studies, including ones that update parts of its data (for instance Caul 1999; Poguntke 2000; Bille 2001). 3

4 II. Introducing the PPDB 2 The PPDB project falls squarely within this tradition of evidence-driven approaches to the comparative study of political parties. It deliberately builds on and extends past efforts. In some cases it replicates questions that have been used in earlier studies, making it easier to use some of PPDB s snapshot data for longitudinal comparisons. In forming what was essentially a datagathering collaborative endeavour, members of this project agreed to pool our efforts and standardize variables in order to maximize the utility of our individual data gathering efforts. When the organizers assembled the initial team, we deliberately sought out members with varied theoretical and methodological approaches to the study of political parties, and who came from various academic generations. These include (but are not limited to) interests in political finance, gender and politics, political participation, electoral politics, executive-legislative relations, public opinion, regional politics, party-state relations, and political leadership. This diversity is reflected in the data that we chose to gather. (A full list of those involved in this data collection effort is included in Appendix 2.) Our choices about which data to gather, and what coding schemes to use, were driven in part by the diverse interests of group members. We decided early on to primarily follow an official story, in order to facilitate future replication; this decision also constrained our choice of variables. 3 At our initial organizational meeting, generously sponsored by the European Consortium for Political Research, we identified as one of our priorities the aim of gathering data that would be useful for studying parties and their resources as independent variables in other words, that would help us get better answers to the questions of why and how organizational variations matter. Another priority from the outset has been to facilitate the more general study of political parties by making the data easily accessible. Those who have collected the data obviously hope to use it in our own research projects, but we have worked to quickly make the data available to others with the aim of stimulating research in this field. We particularly hope that it will be of interest to those who study elections and legislatures, and who in the past may have ignored party agency and party organizational capacity because of the difficulty of finding good crossnational party data. The PPDB project is one of many contemporary efforts to facilitate access to cross-national political data. As much as possible we have sought to complement, not duplicate, other contemporary efforts to gather data on elections and representation. Thus, our data includes match keys that enable easy integration with several other major data sources. III. Scope and Coverage The PPDB Round 1 data provides information on 121 parties in 19 countries during the period. For some parties and some variables we have readings for more than one year; for most, however, we have just one data point for each party and variable so far. The four modules of the database include over 300 variables that collectively describe some of the most important aspects of party structures and practices. 4

5 The conceptual roadmap that guided our choice of indicators was the view that party organizations can usefully be described in terms of their structures, their resources, and their linkages. We further subdivided each of these dimensions with the aim of answering specific questions. For instance, a recurring question for scholars is the extent to which parties should be viewed as unitary actors. In order to better answer this question, we incorporated indicators derived from four structural sub-dimensions which illuminate the formal location of decisionmaking within the party, and at what level (if at all) these decisions are enforced (leadership autonomy, centralization, coordination, and territorial dispersion). Similarly, we sub-divided the resource dimension into three categories of resources (money, members and staff) in order to better identify dependency relationships and resource control (for more details see Scarrow and Webb forthcoming). We assume that these three aspects of party organizational development are related, and indeed, that some measures may have multiple meanings. For example, if party members provide funding or volunteer labour, party membership numbers may be interpreted as a measure of party resources. If party members act as ambassadors to their communities, membership numbers may be used as an indicator of one type of linkage. Thus, conceiving of parties in terms of structures, resources and linkages does not require analysts to assume that particular measures fit only a single category. It does, however, help to ensure that we have gathered sufficient data to test the predictions of the many theories of party organizational change, including those which posit that the locus of power shifts in parties as their resource bases shift. The data in the PPDB are divided into four modules. These divisions are made on both practical and thematic grounds. The Core Module and Module A focus on variables related to what might be described as the everyday life of extra-parliamentary parties. These modules include information on resources (party membership, party staff, party finance), and structures (including basic party units, party executive composition, and formal links to collateral organizations). Reflecting our hope that we or others will be able to regularly update some of these variables, the division of variables between the Core Module and Module A reflects our initial assessment of the cost and benefit of more frequent updates. The Core Module contains the basic organizational information that seems likely to change more often, and which should be relatively easy to gather. For some parties the PPDB already contains Core Module data for multiple years. For most countries we have Core and Module A data for all parties for one or more of the years 2011, 2012, or Variables in these modules reflect the state of affairs on December 31 of the collection year, unless otherwise designated. For instance, some parties publish party finance information with a lag of one or more years; for these variables, there is a supplemental question clarifying the year to which the information refers. Our other modules give information about rules and outcomes associated with more specific procedures and events. Module B variables cover candidate selection and elections. Module C variables cover party decision-making, including leadership selection, manifesto construction and approval, and intra-party referendums. We collected data on rules for candidate 5

6 selection and leadership selection only in the years in which these events occurred, to ensure that we describe the procedures that were used (which might differ from the procedures that were previously on the rule books). Because the data in Modules B and C refer to specific events, the reporting years necessarily vary by country (Module B) and/or by party (Module C). In almost all cases, the Module B or C data come from the period. The PPDB Round 1 release will present data on 121 parties in 19 countries. Our data collection has focused almost exclusively on parties which had have representation in their national legislatures at the point when the data was gathered. Due to resource constraints, in some cases our country teams were unable to gather information on all the parties with legislative representation; this was particularly true in countries with many small parties, and with parties that first entered or exited the legislature during the course of our study. Each country team made its own decisions on which parties to include. The full list of countries and parties can be found in Appendix 3. These countries are mostly parliamentary democracies (with the exception of semi-presidential France), and most are European, except for Australia, Canada, and Israel. Like our choice of variables, our choice of countries was guided by both theoretical and pragmatic considerations. We have deliberately included a large number of the countries included in the Katz/Mair Data Handbook to maximize the value of the data; we assume that scholars (including ourselves) will combine data from both collections to make longitudinal comparisons. Because of this overlap, our snapshot of party organizational data enables researchers to continue the Data Handbook s endeavour to see how, if at all, party organizations were changing, and to understand how they were adapting to the challenges posed by the new institutional, cultural, social, and political environments within which they operated (1992: 9). Yet the PPDB does more than merely update some of the Data Handbook indicators, even though this would be a useful project in itself. In addition to the updates (primarily about rules and practices related to party decision-making), for theoretical reasons we wanted to go beyond the universe of the earlier project to include both newer democracies, and non-european democracies. On pragmatic grounds we limited ourselves to 19 cases for this initial round of data collection: this admittedly arbitrary but pragmatic figure represents our estimate of the number of participants who could be funded to attend the necessary coordinating meetings, and who could physically sit in a room together. We expect to add a limited number of new cases in a Round 1a release. Our hope is to eventually include some presidential democracies in our sample. However, ensuring that our variables make sense across multiple regime types is a formidable challenge, and it is one that we are pursuing as a separate step. Even without this regime type variation, our countries display a wide range of institutional differences. For instance, they have different electoral systems, different electoral thresholds, a mixture of federal and unitary structures, varied lengths of democratic experience, population size differences, and different levels of state funding for political parties. In short, this collection offers multiple opportunities to test questions about how institutional settings can affect the ways that parties organize, and about when and how this matters. Taken as a whole, 6

7 this collection provides an extraordinarily detailed snapshot of how extra-parliamentary parties function today in both established and newer democracies. IV. Extra-Parliamentary Parties in Contemporary Democracies: Structural Similarities, Resource Differences? In other publications we will use PPDB data to study the impact of party organizational differences, and to study the development of party organizations over time. As such, it will entail the testing of hypotheses. However, our aim in the current paper is a more straightforwardly descriptive one: we want to highlight some important similarities between - and key differences across - party organizations in established party democracies. In what follows, we will investigate differences across countries, as well as across party families. Parties are moulded by their social and institutional environments (Harmel 2002, Janda Harmel/Janda 1994). As a matter of fact, all major party typologies refer to these factors as an explanation of party organizational decisions and party change. While the major contributions towards the literature on party types identify a developmental trajectory leading from cadre to mass to catch-all to cartel parties as the dominant pattern (Duverger 1964; Katz/Mair 1995; Kirchheimer 1966; Neumann 1956), and to emphasise the impact of similarities across ideological tendencies (party families), they tend to underestimate the impact of country-specific institutional and social factors. In a way, this is not surprising as the main focus of these contributions is on the identification of parallel change across countries and this necessarily leads to a lack of attention to variation between countries. In the following investigation we consider whether there continues to be strong evidence of party family organizational tendencies. With our comprehensive cross-national data, we are now in a position to also focus on the impact of national factors on the way parties organize. Thus, we also look for cross-national similarities and differences, expecting that we will find some similarities, but not assuming that we are likely to find evidence of a single dominant type. In what follows we present a few of the key findings from the PPDB, pointing out important patterns of practice in terms of resources, structures, and linkages. IVa - Resources: Money, Staff, Members In this section, we seek to assess the strength of contemporary party organizations across the democratic world through analysis of three different types of resource: money, staff and members. This is not to overlook the obvious fact that party members might also be considered a form of linkage between parties and society, but here they will be examined from the perspective of organizational resources. Discussion of party structures and linkages follows in subsequent sections. The PPDB dataset takes in 121 parties from 19 different countries for which we have data (although inevitably there are missing cases on each of the resources), and describes the situation across nations and party families. Variables are reported in terms of absolute values, and relative to the size of national electorates or other appropriate benchmarks. 7

8 Money The first and perhaps most important - resource on which parties rely is of course money. In this section we review the data on national party head office income. Table 1 reports three separate pieces of information, by country: average overall national party income, the size of the average party income relative to that of the median household in each country, and the average income per registered member of the electorate. The first of these indicators tells us which countries have the richest and poorest parties in absolute terms; inevitably, however, these things can be expected to reflect to a considerable extent the relative size and wealth of each country, which is why it is also interesting to see how rich parties are relative to the median household and the number of electors. Country Table 1: Mean national party income , by country Mean income of national party head offices 8 Mean party income per national median household Income per registered elector Australia (n=8) 19,223, Austria (n=5) 12,521, Belgium (n=24) 6,919, Canada (n=5) 15,564, Czech Republic (n=5) 8,426, Denmark (n=8) 3,509, France (n=2) 60,888, Germany (n=14) 60,701, Hungary (n=4) 2,360, Ireland (n=4) 3,178, Israel (n=10) 2,456, Italy (n=5) 28,827, Netherlands (n=21) 4,727, Norway (n=7) 11,153, Poland (n=5) 5,542, Portugal (n=6) 7,102, Spain (n=7) 60,224, Sweden (n=7) 10,418, UK (n=12) 14,389, Mean (n=159) 16,401, Eta squared.460**.467**.389** Note: All amounts are expressed in Euros (using historical exchange rates as quoted for December each year in N refers to the number of data points we have per country rather than the number of parties; while these are usually one and the same thing, for a some countries (Belgium, Germany, Netherlands and the UK) we have more than one datum per party. Income per median household income refers to the number of times more than the annual median disposable household income that the mean party income is for

9 each country. 4 Eta squared refers to the between-groups variance explained by country differences. ** = p<.01, * =p<.05. In terms of absolute levels of income, it is plain from Table 1 that the German, French and Spanish parties are much wealthier than those of any other country on average, while the Italians also receive well above the overall average of 16.4 million euros per year. The Israeli, Hungarian, Irish and Danish parties feature among the poorest in these terms. When we control for the size and distribution of the national economy by comparing party funds to that of the median household, we see that this same pattern is confirmed: on average, the Spanish, French and German parties all receive an income that is around times greater than that of the median household in their respective countries, while the Italian parties also enjoy an average income that is approximately 1700 times more the median household in their country. Relative to the income of their median national households, parties in Belgium, Denmark, Ireland, Israel and Netherlands fare rather poorly that is, well below the sample mean. And when we correct for the number of registered electors the size of the active citizenry or body politic, as it were we find that the Poles, British and Dutch are the most impecunious, their parties only attracting 18, 33, and 37 cents per registered elector, respectively; at the other end of the scale, the Norwegians and Austrians stand out as being in a league of their own, the former country's parties earning a little over 3 euros and the latter s slightly under 2 euros per elector. Even so, we might reasonably reflect that even two or three euros per elector is not such a high price to pay for one s democracy: arguably, the world s parliamentary democracies get their party politics on the cheap. It is not easy to get a clear sense of what these figures might represent about change over time given the limited amount of historical data, but we can draw on Katz-Mair data for a few countries, and compare the changing relationship between average national head office income and median household income in the late-1980s and the contemporary PPDB years. The six countries for which we can do this are Denmark, (West) Germany, the Netherlands, Norway, Sweden and Britain. Analysis suggests that average party head office income has increased relative to median household income since the late 1980s in four of these cases, and declined in just two (Germany and the Netherlands). One cannot really make confident generalisations on the basis of such a limited sample of cases, of course, but they do at least constitute a prime facie suggestion that party income is unlikely to be declining in real terms in the established democracies, and is more likely to be increasing overall. 5 What of the different party families? Table 2 reveals a straightforward and not particularly surprising story when the data are broken down by party family; the two most wellestablished party families that have dominated much of Europe s post-war history as governing parties, the social democrats and the Christian democrats, are clearly the wealthiest, while all other party families have lower, but relatively similar, average income levels. The big two are 9

10 well above the overall mean income of 16.8 million euros per year (indeed, virtually twice that level in the case of the social democrats), while the remainder are considerably below it. This pattern remains broadly true, no matter how you look at it in raw currency values, relative to median household income, or per elector. The social democrats do best in each of these regards, while the regionalist parties fare poorest. Analysis of variance suggests that differences between countries explain more of the variance in party income than differences between party families, in so far as eta-squared is higher for the inter-country variation in Table 1 than for the interfamily variations in Table 2. This is, of course, only preliminary evidence: multivariate modelling would be required to draw more definitive conclusions. Nevertheless, it points to the likelihood that patterns of party funding converge around national models more than they do around typical party family models. Table 2: Mean national party income , by party family Party Family Mean income of national parties Party income per median household Income per elector Christian Democrats/Conservatives (n=41) 23,794, Social Democrats (n=27) 32,800, Liberals (n=25) 8,952, Greens (n=18) 7,141, Left socialists (n=13) 8,692, Right-wing populists (n=7) 6,200, Extreme right (n=6) 8,147, Regionalist (n=10) 5,043, Overall mean (n=147) 16,797, Eta squared.131**.136**.205** Note: Eta squared refers to the between-groups variance explained by party family differences. ** = p<.01, * =p<.05. In summary, the German, French and Spanish parties seem to be the richest in terms of absolute levels of funding, even when we correct for the size and distribution of national economies. When we allow for the size of national electorates, we find that parties in several other countries are as strong or indeed stronger (Austria and Norway being good examples). The biggest party families the social democrats, Christian democrats and conservatives - fare prominently in financial terms, but we suspect that nationality counts for more than party family in explaining patterns of funding. 10

11 Staff One of the most under-researched fields in the study of political parties is that of party employees. This is curious given how much we now know about most other significant aspects of party life, including developments in party ideology and policy; the role, powers and social background of party members and leaders; the recruitment and sociology of legislators and candidates; and the marketing of parties. By contrast, relatively little is known of the men and women on the organizational payroll who run the day-to-day operations of parties. This is a significant oversight, which leaves us with a deficient understanding of an important aspect of party organizational development. This is particularly so since it seems likely that the importance of payroll staff may well be greater now than ever before. In part this is because it is clear that the modern age of election campaigning and political marketing makes certain types of professional expertise all the more pertinent. In addition (and relatedly), it is likely that parties have come to rely increasingly on paid professionals in the context of the overwhelming evidence of party membership decline which we have revisited in this paper and de-energization (Seyd and Whiteley, 1992; Whiteley et al, 1994) around the democratic world. As the voluntary wings of party organizations have gone into decline, it would seem likely that the resources available through paid professionals have increased. What evidence does our database provide that this supposed substitution of paid for voluntary labour might indeed be the case? It is important to note that, in investigating this issue, we immediately become aware of one of the main reasons for the relative lack of research into party employees: the sheer difficulty of getting the relevant data. For whatever reason, many parties tend to be reluctant to provide data on the number of payroll employees that they have. That said, we believe that we have sufficient information to generate meaningful evidence. We have central party staffing data for 15 countries, and legislative party staffing data for 12 countries. This includes 10 Belgian parties for which we have two years of central staffing data (for 2011 and 2012), and 7 British parties for which we have three years of central staffing data ( ), which gives us a total of 88 data points from 62 parties (ie, more than half the parties in the dataset). For legislative parties we have two years of staffing data for 8 Belgian parties, and 7 German parties, giving us 82 data points in all (from 66 parties). 6 There is also a risk that any snapshot of party payroll figures across countries could be misleading because the number will usually vary systematically according to whether it is an election year or not. For most parties, it is safe to assume that they will employ more staff in election than non-election years, for obvious reasons. As it happens, in only three cases in the database does the staffing information that we have come from parties contesting a national election in the year to which that information pertains Denmark, Ireland (for Fine Gael and Fianna Fail) and Portugal. This means that the particular snapshot we have can be regarded as largely representative of parties' 'normal' mode of operation in non-election years. 11

12 Country Table 3: Number of full-time party staff, by country Number of fulltime paid staff in head office Head office staff as percentage of party members Number of fulltime paid staff in legislative party Number of legislative staff per MP Australia (n=3).04 (n=3) - - Belgium (n=20).12 (n=20) (n=16).98 (n=16) Czech Republic (n=4).17 (n=4) 2.75 (n=4).09 (n=3) Denmark 9.00 (n=4).08 (n=4) (n=4) 1.62 (n=4) Germany (n=1).11 (n=1) (n=14) 6.99 (n=12) Hungary (n=1) 1.50 (n=1) (n=4) 1.11 (n=4) Ireland (n=3).15 (n=2) (n=5) 1.88 (n=4) Israel (n=2) (n=3) 2.28 (n=3) Italy (n=4).01 (n=4) - - Netherlands (n=2).12 (n=2) - - Norway (n=7).10 (n=7) (n=7) 1.66 (n=7) Portugal 1.00 (n=1).02 (n=1) (n=6) 2.30 (n=6) Spain (n=7).08 (n=7) (n=6).74 (n=6) Sweden (n=8).12 (n=8) (n=6) 1.10 (n=6) United Kingdom (n=21).14 (n=11) 3.29 (n=7) 1.05 (n=6) Total (n=88).12 (n=75) (n=82) 2.18 (n=77) Eta squared.264*.868**.612**.862** Note: Eta squared refers to the between-groups variance explained by country differences. ** = p<.01, * =p<.05. What do we find, then? In absolute terms, Table 3 shows that the Spanish, British and probably the German parties (although we only have data for one German party) generally appear to be strongest in terms of central staffing. We should perhaps be wary of taking some of the very low national averages too literally, because they are either based on very few cases (eg, Portugal, Hungary, Israel) or key data are missing for large parties (eg, the Danish Social Democrats). The figures on legislative party staff are distorted by an obvious outlier Germany, whose parties appear to employ quite extraordinarily high numbers of parliamentary staff. The German parliamentary party staff are in fact formally employees of the state; however, they have a 12

13 number of functions, some of which are party-related, so we think that it is justified to regard them as a party resource. 7 If the German parties are excluded, then the average number of legislative party employees is just 25.76, which is perhaps a more generally representative figure of the database countries as a whole. From this perspective, parties in countries such as Hungary, Spain, Portugal, and Ireland can be regarded as placing their human resources more in Parliament than in the national headquarters, while parties in countries such as Britain tend to opt for the opposite approach. Of course, the number of staff that parties employ to service their MPs might reasonably be expected to reflect the number of legislators that they return to Parliament, so it is also useful to control for the size of parliamentary parties in assessing staffing establishments. Hence, Table 3 also reports the mean number of legislative employees per MP that parties maintain in each country. Overall, this produces a rather modest figure: the German parties are of course substantially higher than any others, being able to call on the support of nearly 7 support staff for each MP, but for all other the countries the norm is only about 1 or 2. It is difficult to say anything authoritative about over time trends given the dearth of earlier studies on party staff, but once again some data are available from the Katz-Mair project on party organization (Katz & Mair 1992), with which we can offer limited comparisons. There are eight countries for which we have data on head office staffing levels in both the Katz-Mair study and our own party organization database: Denmark, Germany, Ireland, Italy, Netherlands, Norway, Sweden and Britain. Given the fact that we only have head office staffing data for one German party in the PPDB (and the relatively small Left Party at that) the inclusion of Germany in a longitudinal comparison is questionable. However, bearing that caveat in mind we can see that the average number of head office staff in these eight countries has risen from in the late 1980s 8 to 46.08, and if we exclude the German case for being unrepresentative of the general picture in the country, then the average number of staff has actually increased more markedly, from to The average number of central party employees rose in 6 of the 8 countries, and was stable in one other: only in Germany does there appear to have been a decline, but this is likely to be a misleading inference given that we would be comparing the two major parties of the 1960s with one of the minor parties of the present day. All in all, it is reasonable to conclude, if a little tentatively from this limited number of cases, that central party staffing levels have generally been augmented since the 1960s and 1970s. Earlier, we posed the question of whether or not paid staff might be substituting for voluntary labour. We can get a handle on this by examining their changing numbers relative to each other. Table 3 (column 2) shows that on average the parties in our sample only employed 0.12 members of staff in their head offices for every hundred individual party members they have which hardly strikes one as an imposing number. The variation is not great around this mean: Hungary produces the highest figure, at 1.5 employees per hundred party members, but even this is based on only 1 party, so it is hardly a robust measure. Nevertheless, if we go back to the countries for which we have earlier data from the Katz-Mair project, we can observe an interesting pattern: between the mid-1970s and the late 1980s, the Katz-Mair data show a 13

14 consistent growth in the central party staff/membership ratio: in all seven cases for which it is possible to calculate this, the ratio increased (such that the overall mean grew from.028% to.053%) in this period. This trend seems to have continued since then, with the mean percentage having further increased in six out of the seven cases; only in Ireland has it dropped. 9 What of trends by party family? Table 4 shows a pattern that is broadly familiar from the analysis of financial data. Once again, the major parties of the Christian democratic/conservative and social democratic families predominate. The right-wing populist category appears to score highly on the score for total staffing, but this is based on just one party (the Belgian New Flemish Alliance). The figures in this table deliberately exclude German parties, which are such outliers on legislative party staffing that they tend to distort general patterns that would otherwise be apparent; if they are added, it drives up the overall mean for legislative party staffing to Table 4: Number of full-time party staff, by party family Party Family Number of fulltime paid staff in head office Number of fulltime paid staff in legislative party Christian Democrats/Conservatives (n=21) (n=14) Social Democrats (n=17) (n=11) Liberals (n=15) (n=13) Greens (n=12) (n=10) Left socialists (n=4) (n=8) Right-wing populists (n=1) (n=1) Extreme right (n=6) (n=6) Regionalist (n=11) 8.43 (n=7) Overall mean (n=86) (n=70) Eta squared.304**.273** NB: This table deliberately excludes German parties as outliers that distort the general pattern. See text for further details. Eta squared refers to the between-groups variance explained by party family differences. ** = p<.01, * =p<.05. Parties and members The literature on party members has grown considerably over the past two decades, seemingly in inverse relationship to the numbers of the subject under investigation (Heidar 1994; Katz, Mair et al 1992; Mair & Van Biezen 2001; Scarrow 1994; Scarrow 1996; Scarrow 2000; Seyd & Whiteley 1992; Van Haute 2011; Van Haute & Carty 2012; Van Biezen, Mair & Poguntke 2011; Weldon 2006; Whiteley 2011; Whiteley, Seyd & Richardson 1994; Widfeldt 1999). The evidence on the decline of party membership numbers across the democratic world is 14

15 overwhelming (Katz, Mair et al 1992; Mair and Van Biezen 2001; Van Biezen et al 2011). In Table 5, we update the story of individual party membership trends by reporting a number of things: the aggregate membership across all parties for each country is noted, along with the size of the registered national electorate at the nearest national election, and the consequent membership/electorate ratio. The downward trend which has so often been observed previously remains apparent in our data. The mean aggregate membership figure for the 15 countries for which we have longitudinal data since the 1980s was 886,850 per country at the start of the time-series; by the mid-to-late 2000s when van Biezen et al (2011) reported their figures the average had fallen to 633,425 for the same countries; and in the PPDB data round for the years , it has dropped to 549,360. Indeed, if we include the three further countries that are part of the PPDB but were not in the van Biezen et al study (Australia, Canada and Israel), the national average falls to just 501,337. Not surprisingly, the picture is similar even after controlling for the size of electorates; the average membership/electorate ratio (ME) for the original 15 countries was 7.50 in the early 1980s (or 1990 in the case of Hungary), but had declined to 4.14 by the mid-2000s. The PPDB shows that it now stands at 3.53 (or 3.45 if you include Australia, Canada and Israel). The only country in which the ME ratio has not declined in recent years is Ireland, which appears to have experienced a modest increase from 2.03 to 2.15 in the 5 years following Table 5: Party membership trends, by country Country Electorate Total party membership Australia ,722, , * Austria Belgium Canada 5,186,735 (1979) 5,628,099 5,838,373 6,107,892 (2006) 6,333,109 6,878,141 (1981) 7,039,250 (1987) 7,343,464 7,720,796 (2007) 8,008, ,477,261 1,334,544 1,031,052 1,054, , , , , , ,729 Total membership as % of national electorate *

16 ,257, , Czech Republic Denmark Germany 1980 (West) 1989 (West) Hungary Ireland ,738,981 (1992) 8,116,836 8,333,305 (2006) 8,415,892 3,776,333 (1981) 3,941,499 (1990) 3,993,099 4,022,920 (2007) 4,079,910 43,231, ,251 60,762,751 61,870,711 (2005) 61,946,900 (2013) 7,824,118 8,062,708 (1998) 8,043,961 (2006) 8,034,394 2,275,450 (1981) 2,471,308 (1989) 2,741,262 (1997) 3,110,914 (2007) 3,245,348 (2014) , , , , , , , , ,640 1,955,140 1,873,053 1,780,173 1,423,284 1,317, , , ,932 86, , ,228 86,000 63,000 69, * Israel ,278, , * Italy Netherlands ,181,664 (1979) 45,583,499 (1987) 48,744,846 (1996) 47,098,181 (2006) 47,126,326 10,040,121 (1981) 11,112,189 11,755,132 (1998) 12,264,503 (2006) 12,689,810 4,073,927 4,150,071 1,974,040 2,623,304 2,106, , , , , , *

17 Norway Poland Portugal Spain Sweden UK Mean for earliest reading Mean for penultimate reading Mean for most recent reading a Mean for most recent reading b 3,003,093 (1981) 3,190,311 (1989) 3,311,190 3,421,741 (2005) 3,641,753 28,409,054 (1997) 30,615,471 (2007)) 30,762,931 6,925,243 8,222,654 8,673,222 (1999) 8,944,508 (2005) 9,624,425 26,836,500 (1979) 29,603,700 (1989) 33,045,318 35,073,179 35,779,491 6,040,461 (1979) 6,330,023 (1988) 6,601,766 6,892,009 (2006) 7,123,651 (2010) 41,095,490 (1979) 43,180,573 (1987) 43,818,324 (1997) 44,245,939 (2005) 46,502,545 (2015) 16,096,207 (n=15) 19,184,399 (n=15) 19,554,355 (n=15) 18,754,147 (n=18) 460, , , , , , , , , , , , , , ,998 1,131,250 1,530,803 1,494, , , , , ,632 1,693,156 1,136, , , , , , , , * * * * Notes: Data for all but the most recent readings are from Van Biezen et al (2011).*Indicates that period of comparison is approximate because data come from various years (eg, data for some parties in a given country are for 2011, while for others they might be for 2010 or 2012). No data from earlier years are available for Australia, Canada or Israel. France has been excluded because membership data are only available for two parties, which is too 17

18 few from which to calculate a meaningful national total or M/E ratio. Mean for most recent reading a' is based only on the 15 countries for which we have consistent time series data across all periods; mean for most recent reading b is based on all 18 countries for which we have PPDB data. What is the picture if we break the analysis down by party family? Table 6 sheds some light on this question. The pattern revealed is familiar: as usual, the social democrats and Christian democrats have the largest average memberships of any party family, and the highest average ME ratios. Some of the smaller parties (right-wing populists and left socialists) have surprisingly high ME ratios where they are successful, but this is only in a limited number of countries. In summary, then, ME ratios continue to fall in almost all democratic countries for which we have data, such that only 3% or so of voters now join political parties overall. The Christian democrats, socialists and conservatives continue to have the highest ratios of members to electors in their countries. Table 6: Average individual membership by party family Party Family Mean party membership Mean ME ratio per party Christian Democrats/Conservatives (n=44) 148, Social Democrats (n=31) 157, Liberals (n=27) 38, Greens (n=21) 16, Left socialists (n=14) 37, Right-wing populists (n=8) 29, Extreme right (n=8) 31, Regionalist (n=12) 24, Overall mean (n=165) 85, To summarise the major findings from our survey of the organizational resources at the disposal of the 121 parties in our database, we have found that ME ratios continue to fall in almost all democratic countries for which we have data, such that little more than 3% voters now join political parties across the democratic world; that German, Spanish and French parties seem to be the richest in terms of funding and staff; moreover, parties have very likely become wealthier over the past quarter-century, at least in the established democracies of Western Europe. Even so, party staffing levels are relatively modest in most countries, but extraordinarily high in Germany. In all, this analysis tends to confirm the perception that, while the resource of voluntary labour through the membership might be shrinking, parties are sustaining themselves organizationally through access to funds with which, among other things, they employ staff on their payrolls. For many parties the focus of their paid human resources is the national party head office, although for some it is more likely to be the party in parliament. However, it has also 18

19 become apparent that general trends have not wiped out considerable differences between countries and ideological families. It seems likely to us that those differences are usually better explained by country rather than party family. IVb. Structures: Surprising Uniformity? Extra-parliamentary organizations first developed in late 19 th and early 20 th centuries, stimulated by the organizational efforts of opposition parties (socialists and workers in Germany, Austria and the UK, Liberals in the UK, farmers parties in Scandinavia, religious parties in Belgium and the Netherlands). These parties had widely differing aims, but many of them adopted very similar organizational structures: they operated as clubs with statutes, membership procedures and annual dues, local branches, annual or biennial national congresses as the nominally highest party organ, and a smaller executive committee that held broad decision-making authority between meetings of the national conference. This subscriber democracy model was particularly well-suited to parties which began as extra-parliamentary organizations, or which had small legislative delegations; in such parties there was less chance for conflict between a party s legislative delegation and the party congress as the top authority (Morris 2000; Scarrow 2014: ch. 2.). By the middle of the twentieth century, parties in most parliamentary democracies had adopted some variant of the subscriber democracy model of party organization. Of course, similarity in structures can accommodate multiple practices, and in political parties (as in many other organizations) informal channels can be at least as important as the formal decision-making process. Nevertheless, the adoption and spread of the individual member/congress model may be politically consequential. It use signals acceptance of the norm of parties as micro-polities: parties and their leaders gain legitimacy from their relationships with a self-defined polis, in addition to their relationship with the wider electorate. It also signals recognition of the utility of permanent organization for policy and mobilization, tasks that might otherwise be done solely by elected representatives and ad hoc campaign organizations. The extent to which extra-parliamentary organizations contribute to legitimacy, or help electoral mobilization, are empirical questions. In both areas we would expect that some types of arrangement are more effective than others, and that their impact may vary by circumstances, and according to fashion. As a result, even if many parties adhere to a basically similar model, we would expect to find cross-party organizational variation, and experimentation within single parties over time. After all, we know that a string of parties have made headlines in recent years by claiming that they are going to do politics in a new way, and therefore will have different kinds of party structures and organizational practices. (These are sometimes given poetic names, like the liquid democracy of the German Pirates Party, or the Operating System software of the Italian Five Star Movement.) If novel party organization can increase a party s election prospects, we would expect organizational experimentation to flourish, as parties compete for marginal advantages. 19

20 Partly confounding this prediction is one striking finding from our survey of contemporary party organizations in parliamentary democracies: the sheer uniformity in basic organizational structures and rules. Old parties and new continue to adhere to a subscriber democracy organizational model in which dues-paying members are the polis for most or all important decisions, and in which the party conference is (formally) the party s highest organ. Thus, not only do all but one of the PPDB parties seek to enrol dues paying members (with the exception being the Dutch right populist Freedom Party, led by Geert Wilders)., but almost all of these parties (95%) keep national membership registers, meaning that national parties are able to contact members directly. This is a big change from practices in pre-computer days, when membership enrolment and communication were generally handled by local parties. In keeping with the subscriber democracy model, almost all of these parties have statutes which establish representative governance structures, with the party congress at the formal apex of party decision-making. The following section will say more about the actual distribution of power among party levels; for now, what we want to emphasize is that the member/congress template still plays a prominent role in party claims to be internally democratic. Most party statutes stipulate that the party congresses will meet regularly, with 75% of parties requiring these assemblies to be held more than once every three years. Across party families there is a modest amount of variation in the frequency with which these need to be held. Most notably, three quarters of Green Parties require their congress to meet at least once a year. In contrast, the old left Left Socialists are most likely to set loose requirements, with 40% of them stipulating that party congresses must be held at least once every 4 or 5 years (see Table 7). Table 7: Party congress frequency (per party statutes) At least once per year Between 1 to 3 years Every 3 years Every 4-5 years % 34% 9% 15% N= 108 Most parties have some kind of smaller Executive Committee at the top of their extraparliamentary organizations. Because these bodies have different names across parties, we asked our respondents to tell us about the highest executive body that is recognized in the party statutes. In other words, we are not interested in cabinet meetings or informal meetings between a party leader and her trusted advisors. As a rough rule of thumb, we suggest that the smaller these bodies are, the more likely it is that they are conducting some of the real business of leading the party. About half the parties have executive committees with 20 or fewer members; these are small enough to be effective governing bodies. When we compare this to analyses based on the data documented in the Katz/Mair Handbook, we see a remarkable stability in the 20

21 configuration of the basic intra-party bodies. In other words, organizational innovation has been very limited over time (Poguntke 2000: ch 6). In the majority of parties (56%), these executive committees report directly to the party congress. In most of the rest, there is an additional medium-sized committee between the party congress and the executive. As Table 8 shows, the incidence of such intermediate-level committees is inversely related to the frequency of the required meetings of party congresses: the more committee layers, the greater the time span between required meetings of the party congress (see Table 8). 10 Table 8: Congress (In)frequency and Layers from Congress to Executive: Bivariate Relationship Correlation 2-tailed significance N In terms of the relation between different faces of the party within the party organs: only 5% of the parties give the party leader a statutory right to appoint one or more members of the party executive. 11 On the other hand, this does not mean that the party executives solely represent the extra-parliamentary parties. In half the parties, at least 20% of the members of the party s executive committee are also members of the national legislature. Despite the widespread adherence to the subscriber-democracy organizational model, party statutes vary widely in the powers and responsibilities they grant to their party leaders. These differences affect both the extent to which these roles are explicitly specified, and the specified relationship between the party leader and the extra-parliamentary party. For instance, as Table 9 reveals, while 28% of the party statutes make the party leader formally accountable to the party congress, 21% of the party statutes give the party leader the right to summon the party congress % of the statutes explicitly mention that the party leader could or should attend the party congress; 31% give the leader the right to summon party officials; 40% of the statutes specify that the party leader is the party s external representative. Table 9: Party leader position (statutes and practice) Party Political Leader: % yes May summon party officials 31.9 May summon party congress 21.3 Is designated as the party s external representative 59.6 Is accountable to the party congress (negative coding 27.7 in index) Party leader appoints 1 or more members of party 7.0 executive* N=94, unless otherwise stated. *N=114 21

22 These items can be combined to produce a 5-point additive index of the autonomy of party leaders. As Table 10 shows, parties are widely dispersed on this index, but no party earns the top possible score for leadership autonomy. Table 10: Index of leadership strength Index Frequency Percent N = 90 Note: This is a composite index constructed by adding together the total number of leadership rights referred to in Table 8. What, if anything, do these statutes tell us about how parties distribute decision-making authority between the party leader, the extra-parliamentary party, and the parliamentary party? Although structures seldom or never provide the complete story about who holds power and influence within a party, we do expect that intra-party conflicts will be strongly shaped by the formal rules, and by the norms these rules embody. For instance, we might assume that responsibilities would be tipped more towards the extra-parliamentary party in those parties which require frequent meetings of the party congress. The statutes seem to provide some evidence of this in the form of a negative correlation between leader power index and party congress frequency: the greater the maximum allowable timespan between meetings of the party congress, the greater the enumerated powers of the party leader (see Table11). Similarly, the larger the party executive, the greater the powers that are ascribed to the leader. Yet this also holds for the size of the party membership (as the cartel party model seems to predict): parties with more members tend to have leaders with more enumerated powers. Table 11: Leadership strength and other party organs, bivariate relationships Correlation 2-tailed significance N Executive Size Congress (in)frequency Number of Party Members % of seats in national legislature To the extent that party statutes contain an element of rhetoric, we would expect that ideological (as in party family) preferences would affect the arrangements outlined by statutes, 22

23 including the roles ascribed to leaders. Our data show only faint traces of this (Table 12). As we would expect, Green and Left Socialist parties are mostly found in the lowest two categories, and the Far Right parties are mostly found in the uppermost cells that are populated (i.e., categories 3 and 4). Yet this pattern was by no means overwhelming: one of the Left Socialist parties scored a 3, and one of the Far Right parties scored a 1. In addition, 70% of all parties grouped in cells 1 or 2. In other words, although party family seems to play a slight role in determining these arrangements, the impact is far from overwhelming. Table 12: Mean: Party Leader Strength Index by Party Family Mean Score Christian Democrats/Conservatives Social Democrats Liberals Greens Left Socialists Radical right N To conclude: the predominant finding of this section is that there is striking similarity in what might be termed the organizational skeletons of the parties. Whereas the previous section showed considerable cross-national variation in the distribution of resources, the descriptions in this section show the continued dominance of the subscriber democracy model across established and newer democracies, and across all party families. This enduring similarity is seldom remarked upon, but we find it notable, not least because it has survived several waves of populist challenges over the past four decades. Parties that proudly deviate from this basic model, and which claim to pursue a new brand of democracy, tend to receive a great deal of attention from the media and scholars alike. In fact, however, few of those parties have gained enough traction to join and stay in a legislature for more than one or two terms. Many of those that have done so have changed their organizations in ways that make them more similar to the organizations of their older peers. Such organizational convergence is undoubtedly encouraged by national regulations and statutes that dictate some of the fundamental organizational options for parties and/or voluntary organizations. Yet that is not the whole explanation, because in some cases party structures pre-date the laws, and in any case parties themselves are in a position to alter the regulations if they wished to do so. If the organizational convergence is not driven by ideology, perhaps it has been driven by the utility of the model (cf. Poguntke 1998), and/or by its perceived legitimacy. 23

24 IVc Parties as Democratic Linkage Measuring Intra-Party Democracy We have seen in the previous section that political parties largely resemble each other when it comes to the configuration of their core party bodies. Yet, when we take a closer look at how their organizations provide for linkage to the citizenry, we find remarkable variation which put the major typologies in perspective. The membership organization of political parties is one of their principal ways of generating linkage to society (Poguntke 2000). While adherents of a Schumpeterian view of democracy would argue that democracy does not necessarily require democratic linkage through parties, others maintain that it is virtually unthinkable save in terms of democratic linkage through party. Obviously, we cannot decide this debate here. However, our data allows us to investigate the empirical realities irrespective of normative desirability. We have collected data on a considerable number of variables that are related to the democratic quality of political parties internal politics. While many of them are not very interesting individually, they can be combined in a meaningful way to create valid measures of intra-party democracy (IPD). 13 The benchmark of IPD is that it should facilitate the involvement of as many party members as possible in the decisions that are central to a party s political life, namely, programme writing, personnel selection and general intra-organizational decision-making. From this perspective, it seems plausible to argue that the degree of organizational decentralization represents an independent component of IPD that should be measured independently of general inclusiveness, a point we have made elsewhere (Scarrow 2005: 6; von dem Berge et al. 2013). Upon reflection and closer inspection of our data, however, we have come to the conclusion that it is preferable to focus solely on the degree of inclusiveness as a measure on IPD because there is very little gained by also constructing a measure of decentralization. As a matter of fact, in one sense, a higher degree of decentralization automatically leads to a higher degree of inclusiveness because more party bodies (like the German Land parties or the British constituency parties) play a role in certain crucial decisions. We have, however, constructed theoretically distinct IPD indices, for discursive and plebiscitary variants of IPD, each of which represents a different approach to realizing inclusiveness. Two Variants of Intra-Party Democracy Drawing on the insights of democratic theory, we distinguish two variants of IPD, namely discursive and plebiscitary intra-party democracy: The comprehensive Discursive IPD index (DIPD) measures the democratic inclusiveness of intra-party decision-making for the three essential components of intra-party democracy, namely programme writing, personnel selection (leaders and candidates) and organizational structure. In a nutshell, our discursive IPD-index measures the inclusiveness of 24

25 decision-making inside parties that is based on discussions within party bodies and assemblies. In other words, it combines two aspects, namely the maintenance of organizational boundaries and decision-making based on the exchange of arguments within designated party bodies. To be sure, this includes also assemblies of all members (e.g., at the constituency level). Hence it is not confined to representative decision-making, but it requires the temporal coincidence of discussion and decision. A higher index score indicates that a more inclusive party body has the final say over a certain aspect. 2. A Plebiscitary IPD Index (PIPD) measures the degree to which parties allow for decision-making based on the plebiscitary logic of one member one vote. This means that here we are interested in decisions which are not at the end of a discussion within representative or membership assemblies but are eventually made by the lone party member at his or her desk or computer screen. We contend that this is a fundamentally different logic as there is no way to deliberate and reach compromise (frequently through repeated rounds of voting). It is the politics of either/or. It is dichotomous politics which arguably gives a lot of power to the leaders (Katz/Mair 1995: 21). It may, however, also be used as a leadership-challenging device. This is an empirical question. What counts for us now is that it follows an inherently different logic (Cross/Katz 2013). This index includes all parties which envisage such procedures in their rules. These procedures may be optional or only compulsory in certain situations. This is logically necessary as it is virtually impossible to envisage a party organization which is exclusively based on plebiscitary decision-making (even though the Italian Five Star Movement may be a case in point). We found a surprisingly high number of parties which score on this scale: Only 43.6 per cent of the parties included in our study have no provision for plebiscitary decision-making. Conceptually, our one-dimensional DIPD indices represents a formative index (Diamantopoulos et al. 2008; Diamantopoulos/Winklhofer 2001) consisting of three components: programme writing, personnel selection and organizational structure. This would require deleting all cases where we have no valid data for all three components. As this would lead to a substantial reduction of the number of cases, we have tested different variants of calculating the index and have found that they are so highly correlated that we have decided to include all cases where we have valid data for at least 2 of our 3 components. 15 Our indices are mainly based on items covering the official story complemented by few items reporting real story of IPD. 16 Our calculations are based on data for the years 2011 to 2014 using the most recent measurement point for which data was available. We will now analyse whether our conceptual distinction between plebiscitary and discursive intra-party democracy holds empirically. First, the correlation coefficient of 0.37 (Pearson) indicates that both indices are related yet most likely measure separate dimensions. Also, Figure 1 plots both index values of all parties in our study and it is evident that both variants exist empirically. 25

26 Figure 1: The relationship between discursive and plebiscitary IPD Patterns of Intra-party Democracy: Divergence rather than Uniformity When we turn to simple descriptive statistics, we also see substantial differences between our two measures. Both indices have a theoretical minimum of 0 and a maximum of 1. The midpoint of 0.5 indicates a neutral position vis-à-vis our two IPD measures in that such a party is neither particularly inclusive nor elitist in its intra-party politics (see Appendix I for details). The results for the discursive IPD index show that all but one of the 122 parties included in our study have internal structures that satisfy a minimum level of internal democracy. The exception is the one-man Dutch Freedom Party of Geert Wilders which has no party members and hence no internal structure to speak of. It has therefore been coded missing for our IPD indices. 26

27 Table 13: Descriptive Statistics of IPD-index-scores N Range Min. Max. Mean DIPD (discursive IPD) PIPD (plebiscitary IPD Source: PPDB Data Set I Our data provide a strong support for our contention that the DIPD index, which is designed to measure what we conceptualize as intra-party democracy based on meetings and exchange of arguments within party bodies, represents the essential core of intra-party democracy: We have a valid measurement for all parties, and none of the parties comes close to the minimum value of our index (the lowest value is.26) while some parties go all the way towards almost perfectly democratic internal procedures (see Table 13 and figure 2). The picture changes substantially for the plebiscitary variant of IPD. While more than 40 per cent of the parties in our study have not institutionalized any plebiscitary mechanisms, some parties reach our maximum value of 1.0. For both indices, we find substantial variation between parties. Figure 2: Frequency Distributions of DIPD and PIPD indices Source: PPDB Data Set I Patterns of IPD by Party Family This section examines the degree to which the two variants of IPD can be found in different party families and countries. It is easy to imagine how plebiscitary and discursive democracy can be combined in various ways. Think, for example, of a populist party which is likely to use 27