assessing party membership figures: the mapp dataset

  • emilie van haute
  • emilien paulis
  • vivien sierens
Datasets

Abstract

Assessments of party decline and decline of traditional forms of political participation often rely on the argument of party membership decline. Most studies analysing trends in party membership over time focus on aggregate country-level data at a few points in time. While they allow grasping general membership trends, they are not without shortcomings. This article presents the Members and Activists of Political Parties (MAPP) dataset related to the MAPP project. The dataset makes a large amount of data on party membership available to the larger public. The dataset provides 6,307 party membership data observations (M) covering 397 parties in 31 countries, mostly between 1945 and 2014. The article discusses the existing literature and data on party membership trends, how membership trends have been assessed so far, and the potential added value of the MAPP dataset.

Keywords

political parties party membership party organisations 

INTRODUCTION

The dominant form of party–citizens’ linkage has for long been party membership. However, memberless parties are conceivable today (Dalton and Wattenberg, 2000; Mazzoleni and Voerman, 2016), and parties have developed multi-speed forms of linkage or affiliation beyond membership (Scarrow, 2015). Parties are changing, and this makes the study of party membership and its mechanisms, processes, and evolutions crucial. Yet, party membership has for long remained a minor topic in the field (van Haute, 2009; van Haute and Gauja, 2015a, b).
‘...This article presents the MAPP dataset related to the MAPP project (Members and Activists of Political Parties). The dataset makes a large amount of data on party membership available to the larger public’.

Three articles have fundamentally contributed to the development of party (membership) research: the article by Katz et al in the European Journal of Political Research (1992), the update by Mair and van Biezen in Party Politics (2001), and the second update by van Biezen et al in the European Journal of Political Research (2012). What these articles have in common is that they provide a comparative longitudinal assessment of party membership figures using a collective dataset that has been updated over time. These articles were groundbreaking in the sense that, before their publication, early research on party membership (figures) consisted of case studies of single countries or parties with few systematic cross-national comparisons (with partial exceptions, such as Bartolini, 1983; von Beyme, 1985; Sundberg, 1987). Relying on ‘objective’ membership figures provided by political parties (Mair and van Biezen, 2001; van Haute and Gauja, 2015a, b), their conclusion point toward a deepening decline of party membership over time, to the point that it now concerns almost all democracies.

Given their empirical contribution, these articles rapidly became central in several sub-fields in the literature. Figure 1 represents the 408 most-cited articles on party membership in the post-1992 literature referenced in the Scopus dataset. Each circle represents one scientific publication, and the size of the circle is proportional to the article’s number of citations, with larger circles representing the most often cited publications in the literature. A link between two circles means that the publications have at least three references in common. Clusters mean that co-citations between these related articles are frequent. As Figure 1 shows, the three articles (respectively labelled in the figure as Katz et al,1992; Mair and van Biezen, 2001; van Biezen et al,2012) are located at the centre of the figure, and at crossroads of three major sub-fields.
Figure 1

Graphic representation of the post-1992 dominant literature on party membership.

Source: Dataset exported from the Scopus database, including 409 documents matching with 'party membership' as keywords and their 34,684 references; literature considered: only post-1992 articles (no books or conference/working papers). The reference list is available upon request.

Note: Representation via VOS viewer.

The first stream of research (Figure 1, top right) looks at party members and activists as a free resource for parties in electoral campaigns (see for example Fisher et al, 2006). The second stream of research (Figure 1, left) looks at party membership as one of the three faces of party organisations, the party on the ground (see for example Katz and Mair, 1995). The third stream of research (Figure 1 bottom right) investigates party membership as one form of political participation and looks at who joins and why (see for example Whiteley et al, 1994; Seyd and Whiteley, 2004). These three fields are relatively independent of each other (few co-citations), but they share a common reference to our three groundbreaking articles, which is illustrated by their centrality in Figure 1.

The conclusions of these groundbreaking articles have also become central in the literature. They have served as the foundation of party (membership) research. Party membership decline has become a straightforward indicator of declining organisational health of parties (Dalton and Wattenberg, 2000; Dalton et al, 2011). Today, it is often mobilised as the opening statement in many researches, and is rarely challenged (with some exceptions, see Delwit, 2011; Kölln, 2015).

However, these three studies mostly analyse membership figures at the aggregate (country) level, rely on few time points and, perhaps most importantly, are not paired with a publicly available dataset. The aim of this research note is to present a new dataset, the Members and Activists of Political Parties (MAPP) dataset,1 which tries to address these shortcomings and allows for a finer assessment of party membership.
‘...The dataset provides 6,307 party membership data observations (M) covering 397 parties in 31 countries, mostly between 1945 and 2014’.

PRESENTATION OF THE MAPP DATASET

One of the main objectives of the MAPP project was to conduct the largest data collection so far on party membership figures. In order to do so, it directly relies on the past efforts of scholars involved in the data collection for the three founding articles and tries to extend and complement it. Consequently, it relies on a convenience sample of thirty-one countries2 for which country expert(s) could be identified and have agreed to contribute to this collective effort. (A full list of the country experts involved in the data collection is available in Electronic Supplementary Material).

The unit of analysis in the dataset is political parties. Country experts were asked to be as exhaustive as possible and to report party membership figures for all parties that have been or are currently represented in the national parliament, as well as major regional parties. When available, additional data on smaller parties were included too. At this stage thus, the MAPP dataset contains at least one data point for 397 parties. Similarly, country experts were asked to provide data from WWII to 2014 (or since the democratisation of the country to present), but when available, membership data prior WWII were included too. The unit of observation corresponds to the number of party members of a given party in a given year (M). The dataset contains 6,307 observations.
‘...The article discusses the existing literature and data on party membership trends, how membership trends have been assessed so far, and the potential added value of the MAPP dataset’.

The dataset is available on the project’s website www.projectmapp.eu, either in the form of individual country datasets, or as an integrated comparative dataset.

For each unit of observation (M), country experts also provided information about the country (country name in English, country ISO codes) and the party (party acronym, full name in original language, full name in English, year of foundation, year of origin for parties with earlier roots, year of disappearance, and party family).3 The MAPP project team added three party ID variables, one that is specific to the MAPP project and, when available, the party ID used in the Political Party Database project (PPDB, http://www.politicalpartydb.org), and the Parliament and government composition project (ParlGov, www.parlgov.org) in order to facilitate the merge of the MAPP dataset with other large-scale comparative projects in the field of comparative politics.

The dataset also provides information regarding the source of the collected data. As mentioned in the introduction, the aim was to collect ‘objective’ party membership figures either directly provided by political parties themselves, or reported by institutions (in countries where parties are required to disclose their membership data), by the media, or in other academic publications. We are aware of the shortcomings of objective membership figures and the variance in the quality of the sources. As pointed by van Haute and Gauja (2015a, b), the quality of the data varies across countries, parties, and time. Quality improves when state institutions require parties to publicly disclose their membership figures. Quality also varies depending on the organisational capacity of the parties, as well as institutional or legitimacy factors that can push parties to inflate their membership numbers or make them reluctant to disclose them at all. Nevertheless, some of these hurdles are progressively removed. With the development of new communication technologies, parties have better tools today to maintain their membership databases. In parallel, growing party distrust may have affected the symbolic value of party membership and created incentives for parties to become more transparent and open to academic research, and more ready to disclose their membership figures. Furthermore, some of these measurement problems are relatively minor when looking at long-term trends (Scarrow, 2000). While some problems and difficulties remain, ‘there is little the analyst of party membership can do about this’ (Mair and van Biezen, 2001: 8). Besides, the only available alternative, the subjective measure based on individuals reporting their party membership in population surveys, is not without its own shortcomings (van Haute and Gauja, 2015a, b). Therefore, we believe that the dataset provides a unique, solid contribution to our empirical knowledge of party membership.

MAIN CONTRIBUTION AND OPPORTUNITIES OF THE MAPP DATASET

Compared to the existing data analyses on party membership, the MAPP dataset presents two main advantages. It contains more time points, and more parties. Combined, these two advantages make the MAPP dataset the largest available dataset in terms of data points (M) on party membership figures. In this section, we stress the empirical contribution of the MAPP project, and we sketch the opportunities and new avenues for research that the MAPP dataset opens.

The MAPP dataset offers more time points and covers more parties than the existing studies, which ultimately provides many more observations (M). As a matter of comparison, at the party level, Katz et al (1992) relied on around 510 observations of M/E ratios, Mair and van Biezen (2001) relied on 352 observations (M)4, and van Biezen et al (2012) added 307 new observations (M). Compiling the data from the appendix of the three articles together would allow to work on a time-series cross-sectional database of a little more than 1,000 observations, whereas the MAPP dataset relies on a total of 6,307 observations.

With fewer observations, prior studies had more limited options for their data analysis. They mostly discuss national aggregate membership data over all (available) parties as a proportion of the electorate (aggregate M/E per country) to run longitudinal cross-national comparisons (Katz et al, 1992; Mair and van Biezen, 2001; van Biezen et al, 2012). This leaves only two or three time points per decade per country. Even then, they had to use proxies for missing observations. Missing data have been estimated using either membership data for up to two years prior or after the time point as proxies, assuming temporal equivalence, or by computing averages from prior and post observations (the choice between the two strategies being not always clear).

By including more observations, the MAPP dataset allows for a finer depiction of party membership trends. Figure 2 illustrates this at the aggregate level. Using Finland as an example, it replicates van Biezen et al’s analysis of party membership trends from 1960 until 2010, and compares it to trends from the MAPP dataset from 1945 until 2010.
Figure 2

Trend in Aggregate Party Membership Ratio (M/E) in Finland, based on van Biezen et al (2012) and the MAPP dataset.

Source: Left: Katz and Mair, 1992; Mair and van Biezen, 2001; van Biezen et al, 2012 (appendix of articles); Right: van Haute et al, 2016.

Notes: The dots denote data points, the trend is the linear regression line.

van Biezen et al, 2012: Figure based on 69 direct observations and two proxies for party membership observations (M).

MAPP: Figure based on 272 direct observations (222 for the five main parties and 50 for the six smaller parties) and 97 proxies of party membership observations (M) (33 proxies used for the five main parties, 64 proxies used for the six smaller parties). Proxies for missing observations were calculated based on a linear interpolation between existing observations. For the five main parties: gap not exceeding two years in 90% of the cases, maximum gap of five years; for the six smaller parties: average gap of five years, maximum gap of fourteen years). Data on the electorate (E) from IDEA database; missing values between election years were calculated based on a linear interpolation.

The MAPP data confirm the downward trend between mid-1960 and 2010 at the aggregate level. However, the MAPP data show more fluctuation along the regression line. When expanding the data coverage from 1945 until 2010, as the MAPP data allow, the trend appears more curvilinear than linear, which confirms how crucial the starting point of the longitudinal study is (Norris, 2002; Scarrow, 2000). Besides, much of the trend was driven by the drop in membership of one single party, the Centre Party (KESK), which in the twenty-first century returned to its early 1950s membership levels after reporting historically high membership levels between the end of the 1960s and the end of the 1980s.

This brings us to the second way in which the MAPP dataset contributes to a finer depiction of membership trends: the possibility to disaggregate the analysis by party. This is a crucial contribution of the MAPP dataset. Although prior studies provided information at the party level, the more limited number of observations did not allow to fully test a set of party-level explanations for membership fluctuations. Indeed, the larger party in a given country often contributes to a sizeable extent to the aggregate country M/E ratio, which often conceals what is happening within the other parties in the national political system. A striking example of this was pointed by Kölln (2015: 466), who noted that in 1990 in Sweden, ‘the social democratic party alone contributed with around 72 per cent’ to the aggregate country M/E ratio. The MAPP dataset allows distinguishing actual party-level membership ratios (M/E) and their evolution over time. Figures 3 and 4 display the party-level M/E ratios for Norway, respectively from 1950 to 2010, and 1990–2010. It highlights that the Social Democrats dominate in terms of membership figures, but are clear outliers in the party system (Figure 3). Figure 4 shows that individual parties’ trends are more nuanced than the commonly accepted ‘general decline’ view.
Figure 3

Trend in Individual Party Membership Ratios (M/E) in Norway, 1950–2010.

Source: van Haute et al, 2016.

Figure 4

Trend in Individual Party Membership Ratios (M/E) in Norway, 1990–2013.

Source: van Haute et al, 2016.

Being able to disaggregate trends by parties is crucial as it allows testing for more solid explanatory models of party membership trends. Given the data limitations, most previous studies focused on aggregate country-level membership trends and have investigated the effect of macro-level explanatory variables. More specifically, two approaches have dominated in the literature. The first one -modernisation theories- is rooted in the supply side (Scarrow, 1996) and looks at broad societal changes linked to post-industralisation, which would have shifted participation repertoires to more individual modes of political action (Norris, 2002; Marien and Quintelier, 2011). The second one –the institutional approach– investigates the impact of the type of political regime (Bartolini, 1983; Tan, 2000), the size of the polity (Weldon, 2006), the electoral system (Norris, 2002), or party laws (Pedersen, 2003; Scarrow, 1996). If these models are fit to explain aggregate country-level longitudinal trends, they fail to explain why in certain polities, some parties experience party membership gains while other experience membership losses, as it is the case in Norway in recent years (Figure 4).

The MAPP dataset is the first large-scale database that allows testing for other, party-related factors.5 Our intention here is not to test these alternative explanations, but rather to propose a research agenda for the field. The pairing up of party ID used in other large-scale comparative datasets offers unprecedented opportunities.

First, the cross-party variations could be linked to the position that parties occupy in their respective party systems. This would link membership trends to electoral and governmental cycles, but also to the nature of the party system (numbers of parties and ideological competition). A quick first glance at the membership trend of the UK Green party (Figure 5) illustrates that there may be a link between electoral performances and membership trends that would deserve further research attention.
Figure 5

Trend in Party Membership Ratios (M/E), Green Party of England and Wales, 1970–2014.

Source: van Haute et al, 2016.

Second, cross-party variations could be linked to party characteristics such as party family, party organisational types, or party age (for a first test of these variables, see Kölln, 2015). The way parties organise and manage their affiliation rules may also matter (for a cross-sectional test of these variables, see Kosiara-Pedersen et al, 2017). A first hint of this is illustrated in Figure 6. It shows how a change in party affiliation rules within the Swedish Social Democrats (the abolition of corporate membership in 1990 – see Widfeldt, 1997) has generated what appears to be a huge drop in membership levels. As the Swedish Social Democrats weight so much on the aggregate membership ratio, one can start to question whether the downward trend discussed at the aggregate country level can really be attributed to macro-level societal changes.
Figure 6

Trend in Party Membership Ratio (M/E), Sweden Social Democrats (1970-2013).

Source: van Haute et al, 2016.

A second example is offered in Figure 7. It shows how, in parties using universal member suffrage for the selection of the party leadership, party membership can fluctuate drastically from year to year. As hypothesised elsewhere by Cross (2015), this phenomenon ‘results from members joining the party for purposes of voting in a personnel recruitment contest and subsequently allowing their membership to lapse the following year. Thus, membership levels spike when a party leadership contest is held and, to a lesser extent, during election years when candidate nomination contests occur’.
Figure 7

Trend in Party Membership Ratio (M/E), Likud (1990–2012).

Source: van Haute et al, 2016.

Finally, intra-party life and dynamics are expected to affect party membership ratios. Many single case studies have documented how internal dissatisfaction or intra-party conflicts may lead to collective exit (for a general discussion on the application of the exit-voice-loyalty model to party membership, see van Haute, 2015). Another avenue for research could therefore be linking factionalism, as expressed in the parliamentary party group or during leadership or candidate selection contests, and party membership figures.

The examples illustrate how the MAPP dataset offers unprecedented opportunities to test new, alternative explanations of party membership fluctuations beyond the general decline thesis.

Notes

  1. 1

    Further details on the MAPP dataset and access to the dataset are available in the MAPP project website: www.projectmapp.eu.

     
  2. 2

    The countries are: Australia, Austria, Belgium, Brazil, Canada, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Hungary, Iceland, Ireland, Israel, Italy, Lithuania, Mexico, the Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland, and the United Kingdom.

     
  3. 3

    The ‘party family’ variable was coded by country experts, using the same coding scheme as the Making Electoral Democracy Work project (MEDW) that distinguishes between nine categories: (Former) Communist, Christian Democratic/Religious, Conservative, Ecology, Ethnic and Regional, Liberal, National, Social Democratic, and Special Issue.

     
  4. 4

    In 20 per cent of the cases, the total membership levels stated in the 2001 article were collected for the same years as the M/E ratio computed in the annexes from the 1992 article.

     
  5. 5

    Kölln (2015) is the first attempt to test these on a sub-set of six countries.

     

Notes

Acknowledgements

We would like to thank all country experts who contributed to the data collection: Baras Montserrat, Barberà Oscar, Barrio Astrid, Bennie Lynn, Bolin Niklas, Christophorou Christophoros, Correa Patricia, Cross William, Delwit Pascal, Ennser-Jedenastik Laurenz, Enyedi Zsolt, Espirito Santo Paula, Gauja Anika, Heidar Knut, Holsteyn Joop van, Ikstens Jānis, Indridason Indridi, Ionascu Alexandra, Jastramskis Mažvydas, Jupskås Anders Ravik, Kenig Ofer, Kocijan Bojana, Koole Ruud, Kosiara-Pedersen Karina, Krasovec Alenka, Ladner Andreas, Linek Lukas, Paczesniak Anna, Pettai Vello, Rahat Gideon, Ribeiro Pedro Floriano, Ridder Josje den, Rodriguez Teruel Juan, Sandri Giulia, Soare Sorina, Spier Tim, Voerman Gerrit, Weeks Liam, Westinen Jussi, Zemanik Michal. The MAPP project is supported by the Belgian Fonds National de la Recherche Scientifique (FRS-FNRS) via a MIS grant (Mandat d’Impulsion Scientifique – Incentive Grant for Scientific Research).

Supplementary material

41304_2016_98_MOESM1_ESM.docx (18 kb)
Supplementary material 1 (DOCX 19 kb)
41304_2016_98_MOESM2_ESM.xlsx (806 kb)
Supplementary material 2 (XLSX 807 kb)

References

  1. Bartolini, S. (1983) ‘The Membership of Mass Parties: The Social Democratic Experience 1889–1978’, in H. Daalder and P. Mair (eds.) Western European Party System. Continuity and Change, Beverly Hills: Sage, pp. 177–220.Google Scholar
  2. Cross, W. (2015) ‘Party Membership in Canada’, in E. van Haute and A. Gauja (eds.) Party Members and Activists, London: Routledge, pp. 50–65.Google Scholar
  3. Dalton, R. and Wattenberg, M. (2000) Parties Without Partisans, Oxford: Oxford University Press.Google Scholar
  4. Dalton, R., Farrell, D. and McAllister, I. (2011) Political Parties and Democratic Linkage, Oxford: Oxford University Press.CrossRefGoogle Scholar
  5. Delwit, P. (2011) ‘Still in Decline? Party Membership in Europe’, in E. van Haute (ed.) Party Membership in Europe. Explorations into the Anthills of Party Politics, Brussels: Editions de l’Université de Bruxelles, pp. 107–128.Google Scholar
  6. Fisher, J., Denver, D. and Hands, G. (2006) ‘The relative electoral impact of central party co-ordination and size of party membership at constituency level’, Electoral Studies 25(4): 664–676.CrossRefGoogle Scholar
  7. Katz, R.S., Mair, P., Bardi, L., Bille, L., Deschouwer, K., Farrell, D., Koole, R., Morlino, L., Müller, W., Pierre, J., Poguntke, T., Sunberg, J., Svasand, L., van de Velde, H., Webb, P. and Widfeldt, A. (1992) ‘The membership of political parties in European democracies, 1960-1990’, European Journal of Political Research 22(3): 329–345.CrossRefGoogle Scholar
  8. Katz., R.S. and Mair, P. (1995) ‘Changing models of party organization and party democracy: The emergence of the cartel party’, Party Politics 1(1): 5–28.CrossRefGoogle Scholar
  9. Kölln, A.-K. (2015) ‘The effects of membership decline on party organisations in Europe’, European Journal of Political Research, advance online publication 20 August. doi:10.1111/1475-6765.12110.
  10. Kosiara-Pedersen, K., Scarrow, S. and van Haute, E. (forthcoming 2017) ‘Rules of Engagement? Party Membership Costs, New Forms of Party Affiliation, and Partisan Participation’, in S. Scarrow, P. Webb and T. Poguntke (eds.), Organizing Representation: Political Parties, Participation, and Power. Oxford: Oxford University Press.Google Scholar
  11. Mair, P. and van Biezen, I. (2001) ‘Party membership in twenty European democracies, 1980-2000’, Party Politics 7(1): 5–21.CrossRefGoogle Scholar
  12. Marien, S. and Quintelier, E. (2011), ‘Trends in Party Membership in Europe. Investigation Into the Reasons for Declining Party Membership’, in E. van Haute (ed.) Party Membership in Europe. Exploration into the Anthills of Party Politics, Brussels: Editions de l’Université de Bruxelles, pp. 34–60.Google Scholar
  13. Mazzoleni, O. and Voerman, G. (2016) ‘Memberless parties. Beyond the business-firm party model?’, Party Politics, advance online publication 25 January. doi:10.1177/1354068815627398.
  14. Norris, P. (2002) Democratic Phoenix: Reinventing Political Activism, New York: Cambridge University Press.CrossRefGoogle Scholar
  15. Pedersen, K. (2003) Party Membership Linkage: The Danish Case, København: Institut for Statskundskab, Københavns Universitet.Google Scholar
  16. Scarrow, S. (1996) Parties and Their Members, Oxford: Oxford University Press.CrossRefGoogle Scholar
  17. Scarrow, S. (2000) ‘Parties without Members?’, in R. Dalton and M. Wattenberg (eds.) Parties Without Partisans. Political Change in Advanced Industrial Democracies, Oxford: Oxford University Press, pp. 79–101.Google Scholar
  18. Scarrow, S. (2015) Beyond Party Members. Changing Approaches to Partisan Mobilization, Oxford: Oxford University Press.Google Scholar
  19. Seyd, P. and Whiteley, P. (2004) ‘British Party Members: An Overview’, Party Politics 10(4): 355–366.Google Scholar
  20. Sundberg, J. (1987) ‘Exploring the basis of declining party membership in Denmark: A Scandinavian comparison’, Scandinavian Political Studies 10(1): 17–38.CrossRefGoogle Scholar
  21. Tan, A. (2000) Members, Organization and Performance: An Empirical Analysis of the Impact of Party Membership Size, Aldershot: Ashgate.Google Scholar
  22. van Biezen, I., Mair, P. and Poguntke, T. (2012) ‘Going, going,…gone? The decline of party membership in contemporary Europe’, European Journal of Political Research 51(1): 24–56.CrossRefGoogle Scholar
  23. van Haute, E. (2009) Party Membership in Europe: Exploration into the Anthills of Party Politics. Brussels: Editions de l’Université de Bruxelles.Google Scholar
  24. van Haute, E. (2015) ‘Joining isn’t Everything: Exit, Voice, and Loyalty in Party Organizations’, in R. Johnston and C. Sharman (eds.) Parties and Party Systems: Structure and Context. Vancouver: UBC Press, pp. 184–201.Google Scholar
  25. van Haute, E. and Gauja, A. (2015a) ‘Introduction: Party Membership and Activism’, in E. van Haute and A. Gauja (eds.) Party Members and Activists. London: Routledge, pp. 1–16.Google Scholar
  26. van Haute, E. and Gauja, A. (2015b) Party Members and Activists, London: Routledge.Google Scholar
  27. van Haute, E. et al. (2016) MAPP Dataset, Zenodo. doi:10.5281/zenodo.61234.Google Scholar
  28. von Beyme, K. (1985) Political Parties in Western Democracies, York: St. Martin’s Press.Google Scholar
  29. Weldon, S. (2006) ‘Downsize my polity? The impact of size on party membership and member activism’, Party Politics 12(4): 467–481.CrossRefGoogle Scholar
  30. Whiteley, P., Seyd, P. and Richardson, J. (1994) True Blues. The Politics of Conservative Party Membership, Oxford: Clarendon Press.Google Scholar
  31. Widfeldt, A. (1997) Linking Parties with People? Party Membership in Sweden 1960-1997. London: Routledge.Google Scholar

Copyright information

© European Consortium for Political Research 2017

Authors and Affiliations

  • emilie van haute
    • 1
  • emilien paulis
    • 1
  • vivien sierens
    • 1
  1. 1.Centre d’étude de la vie politiqueUniversité libre de BruxellesBrusselsBelgium

Personalised recommendations