Why do young people join parties? An incentive based typology of young party members

Transcription

1 Why do young people join parties? An incentive based typology of young party members Regina Weber, Institute of Political Science, University of Duisburg Essen Youth political participation: The diverse roads to democracy McGill University, Montreal, June 15 th -17 th 2016 Abstract Party membership is usually explained either by resource based models such as the socio-economic standard model or seen as result of the incentives that a membership can provide. The first approach stresses the influence of skills, availability and social position. The latter explains the overcome of the paradox of participation by identifying individual as well as collective benefits of joining a party. What we lack so far is a combination of these explanations as well as an application that takes differences among the members into consideration. This paper focuses on young members. Due to their life stage, they can be expected to have different interests than elder ones. At the same time they are less likely to join a party out of traditional class or milieu belongings and could be truely ideologically convicted. Based on a data set from a survey among young party members in the German Social Democratic Party (N=4000) I argue that these young members differ greatly regarding the incentives they respond to. I will come forward with a typology of party members where a distinct group seeks all kind of incentives while another has strong ideological motivations for membership and a third group is rather reluctant to incentives as such. Further analysis will shed light on different resources of the three member types. Keywords: Party membership, latent class analysis, incentives, resources Introduction Why people join political organisations is maybe one of the most discussed question in political science. The future of political parties is seen as influenced by membership reductions in many parties all over the Western world (van Biezen et al. 2012) and the discussion of individual reasons for or against party membership is closely related to this. A widespread explanation is based on the assumption that those DRAFT please contact the author for an update before citing.

2 with more resources participate more often. The socio-economic standard model explains political participation by the individuals amount of time, money and civic skills (Brady et al. 1995). This strand of political participation explanations is helpful to understand what might influence decision making, but fails to explain why increasing levels of education over the past decades have not resulted in more participation and why people that are equipped with similar resources make different decisions on participation. A second explanatory strand focuses on the reasons for participation in organisations, based on the rational choice assumption that an individual calculates the costs and benefits of membership in an organisation before deciding whether it is worth joining (Olson 1965). Applying this to the membership in a political party, we distinguish between selective incentives that are beneficial for the individual and collective incentives where the benefit goes to a larger group of people or is more of a generic nature (Clark and Wilson 1961; Seyd and Whiteley 1992; Whiteley, Seyd, Richardson, and Bissell 1994; Bruter and Harrison 2009a). When a party member joins the party to have the opportunity to run for a public mandate, the membership provides him/her with a individual benefit. When a member joins because s/he wants to support the party in a policy case, this is an incentive of collective nature. Theoretical models of incentives are extensively elaborated, and the importance of different incentives is asked in all kinds of membership surveys. But only recently empirical investigation began to analyse the latent structures underlying these incentives. Bruter and Harrison (Bruter and Harrison 2009a) applied exploratory factor analysis to investigate the dimensions of party membership incentives among young members. They found a three dimensional model of party membership, consisting of a moral dimension, a social dimension and a professional dimension. But this approach has certain limitations. Even though we now see how the incentives are related and there might might be a dominant motivation, we do not understand much about similarities and differences between the members with regard to their motivation. Latent class analysis (LCA) is an actor centred approach that allows to focus on the relation between individual cases in a sample instead of the relation between variables. Latent class models allow to identify homogeneous groups of individuals within a population based on underlying factors of observed variables (Lazarsfeld et al. 1968). The unobserved categorical variable is most robust for summarizing data that stems from categorical responses. Further regression analysis is possible using covariates to understand better what predicts the belonging to a specific latent class (Huang and Bandeen-Roche 2004). Lately, this has been used in political science to analyse participation and citizenship norms, adding to our understanding of differences within and between groups of participants and with regard to citizenship norms (Oser et al. 2013; Hooghe et al. 2014; Weerts et al. 2014). Using LCA for analysing party membership incentives can uncover groups of party members in 2

3 terms of the importance they attribute to different benefits of their membership. In times of shrinking and ageing parties, it becomes more important to understand what makes young members join a party. Understanding differences in their membership reasons is crucial to discuss what kind of members parties are likely to have in the future. As parties lose members, discussions about future party models rise. Whether political parties will still be able to rely on a genuine supportive group of grass root members or rather function as elite parties that seek mainly periodical supporters during campaigns depends not only on the party leaderships. Also why members join the organisation and what they expect from their membership will influence how the future faces of parties look like (Seyd and Whiteley 2002). This article analyses the reasons for party membership based on the assumption that there are underlying factors influencing how members attribute importance to different incentives. It will be analysed whether there are distinct groups of party members within a party and how these groups are defined. The research question is: Are there identifiable groups of party members that can be distinguished according to the incentives they respond to and what are important resource related characteristics of these groups? The paper is structured as follows: First, I review the literature on incentives and resources for party membership and its implications for the specific life stage of young people, and then present the data and methods used. This is followed by the results of the latent class analysis and some bivariate descriptives of the classes. It will close with a discussion of the further implications of these results. 1 Reasons for young party membership Why do young people join parties? It is important to understand first, that young people engage not less in politics, but different. Younger cohorts are less involved in traditional politics of the representational democracy such as voting and party membership, but they advance other, non-electoral forms of political influence like signing petitions or joining demonstrations (Norris 2002; Dalton 2006; Inglehart and Welzel 2005). This pattern of young people s political activities is reflected in the decline of party membership. In some countries such as Germany and Great Britain, the percentage of senior party members, aged 60 or above, makes up more than 60% of all members, while the share of those under 30 years is only between 5 and 10% (Bruter and Harrison 2009a, p. 11; Niedermayer 2014). The withdrawal of young people from this traditional channel of representative democracy has implications for their interest aggregation and its transmission into official politics. We can expect that generational effects apply here, and the party distance of the youth remains once they grow older. This will have a significant impact on the future interest aggregation by political parties in general. It is very unlikely that those who refrain from being a party member at young age are suddenly joining these organizations 3

4 once they are older. Even if the different shape of young people s participation proves that we cannot talk about a disengaged youth, the pull-out from traditional channels of the political system is very likely to decrease the representation of young people in the power centres of politics throughout their lifetime (Martin 2012; Alwin et al. 1992). So as the general trend for young people is to avoid parties, it is even more necessary to understand what motivates those still joining these organizations. Given the fact that parties provide an almost exclusive access to government position and have heavy influence on the jobs in public administration, the remaining young members will probably be in influential positions in the future (Bruter and Harrison 2009b). When this group is shrinking, the opportunities for those that are still in would even increase, are parties could be able to provide even more inducement for the existing members. The main difference of research on that question is whether to put focus on individual resources (resource based explanations) that make it more or less likely to join versus emphasising the assumed benefits of membership (incentive based explanations). The most prominent resource based explanation (RBE) model for political participation is the socio-economic standard model: According to this model, political participation is closely related to the socio-economic situation of the individual. Those more privileged and well equipped are able to contribute time and effort into their political participation easier than those without the necessary means. The three decisive aspects are time, money and skills, often translated into job, income and education. On the individual level, this means the more time, money and civic skills someone has, the more likely s/he is to participate in politics (Verba and Nie 1976; Brady et al. 1995; Harding 1991). The resource based explanations are important to understand social differences in high level political participation, but they do not tell us much about the reasons some has to join a party and the reasons for becoming an active member. Becoming a party member is besides the influence that resources have the result of a very individual decision making. A different way to understand why someone joins a party is through incentive based explanations (IBE). This approach analyses the motives of individuals to join a party based on a cost-benefit-calculation of the gains and investments. The idea is that the individual expects certain costs of party membership, such as time and money that s/he has to spend. On the reverse side s/he enjoys some benefits. Traditional rational-choice approaches consider the chance to influence the organization in terms of own interests as main benefit, which is unlikely in a mass member organisation. Olson (Olson 1965) argues that any organisation pursuing collective goods, such as a party, has to provide members with incentives to overcome their hesitation of giving something to the organization. From a pure rational point of view membership in a party is a paradoxon. It has no obvious benefits for a single member in a big party as their individual vote (e.g. in a primary) is very unlikely to make a 4

5 difference. Newer explanations of political participation go beyond pure rational choice and include the explanatory power of altruism and norms from social psychology. The general incentives model (Seyd and Whiteley 1992) explains motivation for party membership based on rational considerations, altruistric activities and social norms. According to this model, we need to distinct between selective incentives, which bring personal benefit to the member and collective incentives that have a positive effect on a wider group, but not necessarily also for the individual (Clark and Wilson 1961; Wilson 1974). Not only individual advantages matter, but also supposed collective benefits are important to understand to the motives of party membership: Besides the opportunity to change political decision for the own benefit it is also important to be able to contribute to a certain ideological idea or to change the situation for a particular group as such (Seyd and Whiteley 1992; Whiteley, Seyd, and Richardson 1994). These incentives can also be distinguished between those related to the outcome of participation and those related to the process itself. Outcome incentives are given when the result of a decision making process is perceived as a benefit, e.g. when a legislation could be improved in the eyes of the individual. Process incentives are related to the experiences of participation, such as the positive experience of being part of a group that performs a certain action or getting to know people and thus extending own social networks (Seyd and Whiteley 1992; Whiteley, Seyd, Richardson, and Bissell 1994). This general incentive approach includes manifold different types of benefits of party membership that help to understand why people overcome the paradox of participation. Young people can be expected to have a specific set of motivations for party membership due to their age. They are in a certain stage of life where they still develop their political orientation and might just try and err and reconsider their choices quickly. They are also in a phase of life where they orient towards a profession and might look for potential support and orientation regarding their own future. Many individual incentives can have an effect on the personal career, it might be that party membership helps to access jobs in the public sector or the government. So the incentives for young party members are reasonably different from the average party members. Even though a great literature on the motives of membership and activism exist, the empirical evidence reasons for young people in today s parties is still lagging behind. Some explanations as the General Incentive model (Seyd and Whiteley 1992; Whiteley, Seyd, and Richardson 1994) were recently re-evaluated to match them with specific aspects that might apply young people, covering some life-cycle effects such as career prospects that are more prevalent for young members ahead of their transfer from education into the labor market (Bruter and Harrison 2009a). Also resource based explanations have been adopted to fit the special life stage of young people: The believe in political efficacy in the context of a party organization 5

6 as well as parental closeness to parties is important to understand why young people join parties today (Cross and Young 2008). Even though the theoretical distinction between the different types of incentives is elaborated, but empirical application is still developing. The model of general incentives has been further developed for young party members, taking into account the life stage of young people. Bruter and Harrison (2009) identify three groups of young party members with each of them having a predominant motivation for activism: The social-minded members are mainly looking for like-minded others in the party. They spent rather little time on activism and engage occasionally. The moral-minded members take party activism as their moral duty. They are motivated by the perspective to change society and have a high commitment to their ideology. Professional minded members emphasize their individual ambitions. They tend to invest a lot of time in their activity, but their ideological orientation is rather weak. The morally and socially driven members are the majority, but there is a significant minority of young members, that seeks mainly individual incentives in their party activity. They are the ones that most probably remain in the course of the organization and become future leaders (Bruter and Harrison 2009b). This application of the general incentives model on young people shows that there is good reason to believe that young people have similar reasons as everyone else to join a party, but some particularities to their life stage apply as well. This study will test whether the theoretical distinction between social, moralideological and professional types of incentives can be empirically confirmed as groups among young party members. I expect to be able to model reasons for party membership as a latent construct that is underlying certain measurable incentives. Build on that it can be expected that individual resources influence what type of party member a young person is likely to be. The main resources are time, money and skills. For time, it seems logical, that those that have less time will dedicate it more thoroughly. Applied to joining a party we expect that those that are probably more time restricted are more likely to see individual benefits in their membership. Regarding skills it is likely that those members that are more equipped with skills will be more aware of the efficacy benefits that party membership can provide them. Little has been done so far to empirically combine resource and incentive based explanations to party membership. This study will contribute to our understanding of young party members motives in two ways. First, it will be tested whether the theoretical distinction between individual and the collective types of incentives can be empirically confirmed among young party members. Second, it will investigate the influence of resources on incentives and analyse further covariates of an incentive based typology of party members. 6

7 2 Data and methods The study uses data from an online survey among young members under 36 years in the Social Democratic Party of Germany (SPD). The survey was available online from Mar 11th to Apr 14th A direct link was sent out via with an accompanying letter from the party central office communication department encouraging members to participate members were on the list. The net return rate is (N=4006). The gross return rate of finished surveys was (4129 cases). Basic data cleaning was done using degrade points for missing values. The sample has been tested against the full population and can be considered as representative regarding to gender and regional distribution (federal state level). Latent class analysis (LCA) models identify latent variables based on multiple empirical indicators. Unlike factor analysis, an LCA model does not perceive the latent variable as a continuum with different dimensions but allows to identify distinctive groups of people that are similar on a group of measured indicators. Therefore, LCA is suited well to analyse the question of whether theoretical groups such as ideologically motivated, socially motivated and professionally motivated party members exist based on a probabilistic model (Collins and Lanza 2013). The basic latent class model can be generalized by including covariates into the model. The covariates are used to predict individuals class membership in the latent classes. This methods uses means of bi- or multinominal logistic regression, but with a latent outcome variable rather than an observed one (Collins and Lanza 2013, p. 149). The coefficients in the latent class model with covariates are estimated simultaneously as part of the model to avoid biased coefficients. Model choice in LCA is informed by goodness-of-fit statistics like Bayesian Information Criteria (BIC) and a log likelihood ratio test. Once the model is identified, the likelihood of each latent class response to each indicator can be estimated in terms of conditional probabilities. Posterior probability of latent class membership is estimated to identify each respondents party membership type. Classes in the model need to be homogeneous and distinct to qualify as basis for a good model. Based on the LCA we can answer the question whether different types of members can be identified in the data (cf. Appendix, 4.3). The variables that measure incentives for party membership are based on a list of 13 items that are used in similar ways in different party member studies. They cover the three dimensions of incentives (Bruter and Harrison 2009a): moral-ideological 1, professional and social incentives. Each item could be considered as (very) important or not (at all) important. The four scale-items were recoded into binary scales to reflect the bipolar distribution of the items (Table 1, cf. Weerts et al. 2014, p. 149). The assumption of the latent class regression model is that the individual re- 1 Bruter and Harrison 2009a call it only moral, but I will consider also items that measure ideological conviction rather than pure moral considerations. 7

8 Item Abbreviation Important Moral-ideological incentives to change something in society Change 93% to do something meaningful Meaningful 84% to support the goals of the party Supportgoals 83% to influence the politics of the party Seekinfluence 74% to act as a responsible citizen Responsibility 72% a support a specific person in elections Supportperson 36% Professional incentives to get better information about politics Information 67% to have networks and contacts Networks 56% interest in party office or a mandate Office 35% to gain personal or professional benefits Benefit 23% Social incentives because I enjoy political activities Activities 83% to meet like-minded others Likeminded 74% to meet interesting people Interestingpeople 40% Table 1: Indicator variables for the Latent Class Model, Share of answers considering the incentive to be important. Source: own survey, N=3567 (includes only cases which had values for all incentives. sources influence the type of party membership. For example: If someone has already knowledge and skills in political processes due to previous activity, their motivation to choose a party might be driven by the specific influence opportunities that a party offers. Someone like that might be more strongly motivated to gain a mandate. In the following analysis, the covariates introduce individual resources into the model. A classical understanding of resources that influence political participation is the distinction between time, money, and skills (Brady et al. 1995). These tripartite structure of resources includes different types of structural resources and individual acquired or inherited capacities that are sometimes mutually dependent, e.g. given the strong relation between social status that implies a certain financial situation and educational access opportunities. Additionally, we have to take into consideration the tentative situation of young people regarding their educational process and their financial situation. This influences the choice of variables that can measure or stand as a proxy for a certain resource. Time: The time related variable measures a proxy for how much flexible time someone is able to invest for political activity. students: A dummy for being a (school) student. Those in education are typically more flexible with their time than those that are already working. It is expected that students (including school students and those in vocational training) have more available time for party activities and therefore a lower threshold to become a party member. Skills: The following set of variables measures different aspects of skills relevant 8

9 for joining a party. Education measures the length of participation in formal education. The joining age stands for the informal knowledge about politics that should increase among the age of 14-35, which is the spectrum of potential answers. The dummy for union membership differentiates between members that joined another political mass member organisation of a broader worker s movement and the dummy for those with previous political activity differentiated between activists that made a distinct choice for the party as a place to continue their activities. edu: Time of education in years. The duration of education usually gets along with an increase in knowledge about politics. Therefore can be expected that an increase in educational time has an influence on the motivation for joining. Because of the still ongoing educational processes of many of the young party members, this variable can measure only a snap-shot, but hints at differences among the young members. joinage: Measuring the age when the member joined the party. The age of joining can hint to the knowledge about possible implication of the party membership, e.g. how useful party membership can be for reaching a certain goal. union: Dummy for union members. Membership in a labour union is likely to provide skills about knowing the world of politics. In the social democratic party, being in a labor union is still a signal of connection to the labor movement. So it is expected that those that are a member of the labor union have a closer connection to the party. beforeact: A dummy for those that have been politically active in a group or nonparty organisation before they joined the party. Political activism before entering the party hints to a general interest in political activism. This activism could have provided knowledge about the field of politics and the opportunities a party has to offer, e.g. in providing access to mandates. Besides these independent variables, the latent class regression model will control for gender (dummy for women) and the duration of someone s membership in years. 2 3 Results The descriptives of the incentives for party membership, listed in Table 1 indicate that young party members consider certain incentives of party membership more important than others. Most important for them is to change society. But also doing something meaningful, supporting the goals of the party and enjoying political activities is important. The first three items relate to the moral dimension of incentives, while the last one is a social incentive. Least important is to gain personal and/or professional benefits. From the first descriptive view, moral and social incentives seem to be more important for young members than professional incentives. 2 An overview over the univariate distribution of the covariates can be found in the appendix. 9

10 The correlation table (Table 5, Appendix) shows medium correlation between different incentives, but there are some patterns of incentives. It is in line with what was expected about party membership incentive structures. There is a medium ( 0.25) correlation between seeking to influence the party and wanting to change society, as well as between wanting change and doing something meaningful. This would be a group of moral-ideological incentives. Another medium correlation is between enjoying activities and looking for like-minded others. These are social incentives. A third aspect concerns personal inducements: Gaining benefits, meeting interesting people and seeking networks correlate as well. The latter are professional incentives and career driven motivations. LCA is applied to explore these relationships further. 3.1 Latent class analysis of young party membership The implementation of the LCA consists of three parts. First, a basic latent class model is developed and the selection is discussed. Second, a three class basic latent class model is elaborated as the best basic model. Finally, covariates are introduced into the model to analyse the influence of individual resources for the type of party membership. The first step of the LCA is to define a model with a number of classes that best represents the data. Models with different numbers of classes are applied to the data, beginning with the null model consisting of one latent class up to a high number of classes that is beyond a maximum of theoretically useful classes. There is not only one best solution to decide on the number of classes, but several criteria have to be applied and discussed. The most widely used statistic for model identification is BIC. Additionally, the reduction of the likelihood ratio chi-square statistic is compared to a model with one less class and the null model (Bacher and Vermunt 2010; Magidson and Vermunt 2004). The comparison of AIC/BIC and a log-likelihood comparison between different models for different numbers of classes show three possible solutions for a latent class model (Table 2). The BIC of the five class model is the lowest. The log-likelihood ratio comparison shows that the relative improvement is highest from the empty model with one class to the two class model, and again relatively good from the two class model to the three class model. The three class basic model provides a relative improvement to the null model of 6%. This is beyond a reasonable minimum improvement of 5% for accepting a solution to be preferable to the null model (Bacher and Vermunt 2010).t This model seems to be the most appropriate one; it has more explanatory power than a two class model, and the five class model seems to be very complex from a theoretical point of view. 3 3 For robustness check, the five class model has been applied and analysed, but classes were not very distinct. It is common that a better BIC is reported while the log likelihood statistics shows 10

11 P χ 2 df AIC BIC RI0 RI1 LL LR Table 2: Statistics for models with 1-10 classes. P = Number of estimated parameters, χ 2 goodness of fit for the observed versus predicted cell counts statistics, LL = maximum log-likelihood, df= degrees of freedom (df = N - P - 1), RI0 = relative improvement of the model compared to the null model, RI1 = relative improvement of the model compared to the model with one class less, LR = log-likelihood reduction, all according to Bacher and Vermunt All data calculated using the R-package polca (Linzer and Lewis 2010) and own calculations. Based on the considerations about the number of classes the selected basic model for groups of young party members is a three class model based on 13 manifest binary variables. An LCA model estimates class population shares, provides conditional probabilities for giving a certain answer to each manifest item conditional on class membership and assigns class membership to each case in the sample. The basic latent class model of young party members therefore estimates conditional probabilities for considering a certain incentive of party membership as important given class membership in one of three groups. These three classes have an estimated population share of 0.48 (Class 1), (Class 2) and (Class 3), providing two large classes and one small class of party members. The quality of a latent class model is assessed by two main criteria: Homogeneity and distinctness. The classes should be homogeneous with respect to the items. This means that a class is characterised by conditional probabilities of giving a particular answer to an item being either very high (> 0.7) or very low (< 0.3) (Masyn 2013). Table 3 shows the amount of homogeneity in the model. The classes need to be distinct and clearly separated from one another. The posterior probability of class membership of the cases in the sample should be high for one class and low for all other classes. In this model, the mean probability for class membership is higher than 0.8 for all three classes, so the model can be considered to have distinct classes. Both criteria, homogeneity and distinctness show that the model is sufficient to describe distinct groups of young party members based on incentives as indicators. Figure 1 visualizes the conditional probabilities of members in the three classes to consider the respective incentives of party membership as important. Class one only little improvement. In this case, the more sparse model is the better option. See e.g. Oser et al

12 Class 1 Class 2 Class 3 Take all Ideologists Reluctants Estimated class population share 48% 39.1% 12.8% Mean prob. of class membership (sd) 0.88 (0.15) 0.82 (0.15) 0.84 (0.17) Moral-ideological incentives Change 0.97 (0.00) 0.99 (0.00) 0.62 (0.03) Meaningful 0.93 (0.01) 0.87 (0.01) 0.42 (0.03) Supportgoals 0.91 (0.01) 0.85 (0.01) 0.51 (0.03) Seekinfluence 0.83 (0.01) 0.78 (0.01) 0.37 (0.03) Responsibility 0.83 (0.01) 0.70 (0.01) 0.41 (0.03) Supportperson 0.46 (0.01) 0.25 (0.01) 0.28 (0.02) Professional incentives Information 0.83 (0.01) 0.58 (0.02) 0.37 (0.03) Networks 0.89 (0.01) 0.22 (0.02) 0.36 (0.03) Office 0.51 (0.01) 0.19 (0.01) 0.21 (0.02) Benefits 0.37 (0.01) 0.04 (0.01) 0.27 (0.03) Social incentives Activities 0.94 (0.01) 0.83 (0.01) 0.43 (0.03) Likeminded 0.89 (0.01) 0.69 (0.01) 0.33 (0.03) Interestingpeople 0.63 (0.01) 0.17 (0.01) 0.18 (0.02) Table 3: Basic model M1. Predicted probability of item response important conditional on class membership (standard error). Item probabilities < 0.3 and > 0.7 bolded to indicate high degree of class homogeneity. BIC of the model: is the biggest class, consisting of almost half of the sample. Members in this class have high probabilities of considering most of the given incentives as important, except for individual benefits and supporting a person. They will be called the Take all group, as they react to all types of incentives that a party has to offer. The second class will be called the Ideologists. Members in this class consider most social and moral-ideological incentives for membership as important, but they do not attribute importance to professional incentives. They make up 40% of the members. Members in class three have generally a low probability to consider any of the given incentives as important. They seem to be a distinct class of party members that does not consider the given incentives as important for their membership. They will be called the Reluctant members. This group is very small compared to the other groups (12%). The distribution of groups shows that the given incentives are generally considered important by young party members, but there is also small distinct group whose party membership is not motivated by these aspects. The dominance of members that do seek moral-ideological and social incentives is backed up by previous research (Bruter and Harrison 2009a). The ideologists and the take-all-group have some similarities. They share that they consider many ideological incentives of their party membership as important. But these groups are very different in how they respond to incentives that are important for potential future careers. While the 12

13 Figure 1: Membership types by incentives. Conditional probabilities of attributing importance to incentives. The y-axis plots the conditional probability that members of a latent class will consider the items on the x-axis to be important when they joined the party. The horizontal lines show the thresholds of 0.3 and 0.7 for class homogeneity. ideologist members consider it of little importance to build networks and to seek office or a mandate, this is important for the Take all-group as well. The groups fit differently in what we know so far about party membership incentives. The ideologists attribute importance to those incentives that can be considered as the classics of collective benefits (Seyd and Whiteley 1992) of party membership. They attribute importance both to moral-ideological incentives and to social incentives (Bruter and Harrison 2009b): They enjoy political activities, and want to meet like-minded others. These incentives are social benefits for the party members. On the moral-ideological part, they consider it important to do something meaningful, change society, seek influence and want to support the goals of the party. These incentives are related to the ideological aspect of party membership and provide them with moral benefits. They do not value the opportunity to support a person, and disagree with the professional incentives at large. I suggest that this is a distinct group of members that is driven by ideology and moral duty, while at the same time valuing the social benefits that the party can provide them with. The Take all-group also attributes importance to social and moral-ideological benefits of their party membership. But they are distinct on valuing also professional benefits, that might provide them with personal gains in the future. They want to improve their networks as well as getting information about politics. If we lower the constraint of having a conditional probability of > 0.7, they are interested in office or mandate and in meeting politicians. These benefits of party membership are clearly professional incentives (Bruter and Harrison 2009a). For them, all three 13

14 dimensions of incentives can be considered true. This class can be considered as those members that are basically interested in all types of incentives, but they are distinct in that they value professional benefits. The class of incentive reluctants includes those party members have a low probability of attributing importance to party membership incentives as such. Exceptions are changing society and supporting the goals of the party, but even there they have a lower probability than those members in the other latent classes and it is below the 0.7 threshold as well. This seems to be a small group of party members who tend not to expect inducements of their membership, at least not the ones that have been asked and are frequently used in membership surveys. They seem to have mainly an ideological connection to the party, as they share the goals of the party and see it as a place where they can contribute to societal change, but here is further investigation needed on what explains their membership. The data analysis shows that some ideological incentives are important for all young party members when joining a party. They all want to bring change and support the goals of the party. This is the baseline of party membership. The main divide is between those members that consider professional benefits as important and those that are mainly ideologically and socially motivated, these two groups together make 86% of the membership. But there is a small minority that is not at all interested in the benefits that a party provides them. As Seyd and Whiteley (1992) argue, there is more than rationality involved when membership reasons are evaluated. Social norms and altruism might also play a role. This might be the case for the reluctant members. The basic model showed that we can distinguish between three groups of party members based on the incentives they consider important. But this model assumes that each case in the sample has an equal chance of membership in each class. I argue that the individual resources, which have proven to influence the chances to participate in politics and becoming a party member as such should also influence what kind of motivation someone has for party membership. This can be included into the LCA using covariates which will be done in the following part. 3.2 Latent class regression model A latent class regression model includes a set of covariates that predict latent class membership, applying an ordinary logistic regression with the latent variable as outcome. The latent class regression model is based on a basic model which is expanded using the covariates. The base for the following regression model is the three class model estimated above. Model fit in the latent class regression model is based on a BIC comparison and a significance tests of the log-likelihood improvement compared to a more sparse model. The basic model with no covariates is the null model against which hypotheses are tested for significant improvement. Each covariate is introduced in the 14

15 Ideologists Reluctants b exp(b) b exp(b) (Intercept) (0.4) *** (0.51) *** women 0.03 (0.11) (0.14) * duration (0.02) *** (0.02) *** edu (0.02) *** (0.02) *** joinage (0.02) * (0.02) *** union 0.44 (0.11) *** (0.15) * student (0.12) * (0.13) beforeact (0.1) * (0.13) *** Table 4: Latent class regression model M2. Regression coefficients (standard error) and exponentiated regression coefficients, reference class: Take all. Significance levels: * 0.5, ** 0.01, *** (results of the Wald statistics). BIC of the model: model to test whether its effect is statistically significant. The significant covariates are combined in a model and the new model is tested again whether its increased complexity provides a significant improvement. The results of the significance test can be found in table 7. The best model includes the time-related covariate student, a dummy for students and the resource-related covariates edu, measuring education in years, joinage, measuring the age at joining the party, union, a dummy for union members and beforeact, a dummy for those that were political activists before they joined the party. It controls for gender ( dummy for women and the duration of membership. The reference class of the regression model is the Take-all group as the biggest class of the basic model (Table 4). The regression coefficients hint to different resources among the three groups, both regarding time resources and skills: Students: It was assumed that students have more flexible time available than non-students and that the availability of time is relevant for the motivation to become a party member. There is no significant difference between the prevalence of students in the take-all-group and the reluctants, but students have a higher prevalence in the groups of ideologists. The increased chances of being among the ideologist members as a student show that those with a potential for more time and flexibility emphasize moral-ideological and social reasons more than professional reasons when joining the party. Education: The influence of education on membership in the latent classes is statistically significant but very small. Regression coefficients are positive but close to zero. The duration of education therefore does not influence the chances of being in a certain latent class. Even though these results seem to be obvious, some restraints have to be taken into consideration. Since many of the party members are still within their educational process, the measurement of education in years is contingent and just a snapshot. The lack of influence can also be due to that fact rather than due to a lack of importance of education for a typology of party 15

16 members. Age at joining: The age at the joining time influences the prevalence in a certain group compared to the take-all group, but in opposing ways. Increasing the age by one year decreases the chances to be an ideologist against being in the take-all group and it increases the chances to be an incentive reluctant member. Membership in a union: Union membership influences the chances to be in the take-all-group compared to the other groups as well. Those young members that are also union members have a higher prevalence among the ideologists and a lower prevalence among the reluctants. Activities before party membership: Whether someone was already politically active before their party membership does distinguish only little between ideologist members and members that are in the take-all group. But those that have been active before have a lower prevalence to be in the reluctant group. Those that became a party member based on political activity experience before seem to consider the different incentives of a party membership generally as more important. The model controlled for the influence of gender and the duration of party membership. While there was no significant difference between men and women in their prevalence in the ideologist group, women are less likely to be among the reluctant members as compared to the take-all group. The duration of membership is significant for both classes compared to the reference class. An increasing duration of membership increases the chances of being in both classes, and this effect is a bit stronger for the class of reluctant members. It seems that a longer duration of membership decreases the importance of incentives. This does not necessarily mean that for those members incentives have never been important. There might also be a time effect: Those with a longer membership might have experienced that some of their motives did not materialise when they were members and as a result they become less important. Discussion This article contributes to the discussion about party membership incentives. The latent class analysis shows three distinct groups of young party members based on their incentives. One the largest group responds to all three types of membership incentives. The second groups considers only moral-ideological and social incentives as important and does not accept the professional incentives of a party membership to be relevant. A third much smaller group diverges from incentive based explanations as they do not consider any incentives to be important for their decision to join the party. The results of a case centred latent variable modelling of party membership add to our understanding of incentive based explanations as there is are distinct groups of party members with very different reactions to membership incentives. This cannot be explained by understanding different types of incentives 16

17 alone, but needs an analysis of distinct groups of members. What we can say so far is that there is no group in the sample that is mainly motivated by professional incentives. The take all-group is clearly also interested in incentives that might advance a political or professional career, but they are also motivated by moral-ideological and social reasons, which basically means that for those interested in a career the ideological family of the party plays a role as well and they are not political careerist no matter what. The introduction of resources into the latent class analysis improved the basic model significantly. The latent class regression analysis confirms that members are distinct in what set of incentives motivates them to become a party member and that this distinction is also influenced by their available time and their skills. The baseline category in the regression analysis is the class of members that considers all incentives as important. The difference between them and the the group of ideologists is that the latter do not consider professional incentives as important. Students are more likely to reject these professional incentives. It has been argued from a life cycle perspective that young people might value the professional inducements that a party has to offer more as they are still on their way of settling into a professional life (Bruter and Harrison 2009b). The results of this analysis show a different picture: Students seem to focus more on ideology and social aspects than on career aspects. A possible explanation might be the availability of time: As students might find it easier to have some spare time for political activities, they might choose a party membership even though they do not expect this to pay off directly for their future career. Being a union member also increases the chances to reject professional incentives of party membership. This might hint to union members being less interested in career planning and more motivated by the political and social space that a social democratic party has to offer. As we cannot track whether these young members became union members or party members first, but the analysis shows that being a member in both organisations seems to make the members more focus on moralideological and social motivations. The difference between the baseline category and the reluctant members is that the latter do not consider any incentives as important. This group of members seem to have other reasons that we have not yet captured fully. Reluctant members are more likely to be members for a long time, and that might be a result of disappointed hopes. We have to consider that the items are not able to measure the true reasons at the time of joining, but the ex-post rationalization of motivations. After some years and maybe the experience that not all incentives are realisable, incentives might depreciate. The skills that someone acquires through previous political activity and union membership reduce the chances to be reluctant against the incentives of party membership. It seems that those who show a broader political activity are also clear in 17

18 what they expect from becoming a party member. This latent class analysis is based on data from young members in one particular party. Despite some general implications, this limits the scope of the project. Comparative studies of young party members show that social incentives are more important in democracies where parties have historical roots in different political milieus or ideological worlds, which is the case for Germany and ideological incentives are more important in the social democratic and socialist party family (Bruter and Harrison 2009a). This might explain why the share of reluctant members is rather small and it might be different in other countries and other parties. References Alwin, Duane F., Ronald L. Cohen, and Theodore M. Newcomb (1992). Political Attitudes over the Life Span: The Bennington Women after Fifty Years. University of Wisconsin Press. 452 pp. Bacher, Johann and Jeroen K. Vermunt (2010). Analyse latenter Klassen. In: Handbuch der sozialwissenschaftlichen Datenanalyse. Ed. by Christof Wolf and Henning Best. VS Verlag für Sozialwissenschaften, pp Brady, Henry E., Sidney Verba, and Kay Lehman Schlozman (1995). Beyond Ses: A Resource Model of Political Participation. In: The American Political Science Review 89.2, pp JSTOR: Bruter, Michael and Sarah Harrison (2009a). The Future of our Democracies: Young Party Members in Europe. Basingstoke: Palgrave Macmillan. 272 pp. (2009b). Tomorrow s Leaders? In: Comparative Political Studies 42.10, pp Clark, Peter B. and James Q. Wilson (1961). Incentive Systems: A Theory of Organizations. In: Administrative Science Quarterly 6.2, pp JSTOR: Collins, Linda M and Stephanie T Lanza (2013). Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. Wiley. Cross, William and Lisa Young (2008). Factors Influencing the Decision of the Young Politically Engaged To Join a Political Party. In: Party Politics 14.3, pp Dalton, Russell J. (2006). Citizenship Norms and Political Participation in America : The Good News Is... the Bad News Is Wrong. In: The Center for Democracy and Civil Society Georgetown University, Occasional Paper Series 1. Harding, Russell (1991). Acting together, contributing together. In: Rationality and Society 3, pp Hooghe, Marc, Jennifer Oser, and Sofie Marien (2014). A comparative analysis of good citizenship : A latent class analysis of adolescents citizenship norms in 38 countries. In: International Political Science Review, pp