The Change of Motives to Become and to be a Party Member. An Empirical Analysis of the German Party Membership Studies 1998, 2009 and 2017

In this paper, the thesis is put forward that selective outcome incentives for party membership gain relevance over time. Two possible mechanisms are identified as the cause of this increase in importance: a supply-side mechanism based on processes of societal change that took place through generational succession and a demand-side mechanism focusing on shifts in the self-image and organizational structure of political parties. The supply-side mechanism should lead to changes in the motives of potential members, whereas the demand-side mechanism alters the incentives potential and current members are exposed to. The empirical analyses are based on the German Party Membership Studies of 1998, 2009 and 2017. These three studies are nationwide representative surveys of the members of the following six parties: Christian Democratic Union (CDU), Christian Social Union (CSU), Social Democratic Party of Germany (SPD), Free Democratic Party (FDP), Alliance90/The Greens (Bündnis90/Die Grünen), and The Left (Die Linke). Within the German Party Membership Studies, both the motives for joining the party and the current membership motives are surveyed. Empirically, it is shown that there is indeed an increase in the importance of selective outcome membership motives over the period under study. The mechanisms behind this increase in importance are investigated using multivariate Age-Period-Cohort (APC) models based on the cumulated data of the German Party Membership Studies. These analyses are based on a total of nearly 30,000 cases. The results of the APC analyses are largely consistent with the supply-side explanation of the increased importance of selective outcome motives for party entry and membership. The demand-side explanation is only partially confirmed.


Introduction
In 1990, the parties then represented in the German Bundestag-Christian Democratic Union (CDU), Christian Social Union (CSU), Social Democratic Party of Germany (SPD), Free Democratic Party (FDP), Alliance90/The Greens (Bündnis90/Die Grünen), and Party of Democratic Socialism (PDS) (later renamed The Left (Die Linke))-had a total of 2.4 million members. By 2021, the number of members of these six parties had lowered to just 1.2 million-a decline of 50% over a period of just over three decades (Niedermayer 2022). Few parties defy this general trend of declining membership numbers. Bündnis90/Die Grünen and the FDP have recently increased their membership and the "Alternative für Deutschland" (AfD) has established a membership base since its foundation in 2013. However, these exceptions do not negate the fact that the Federal Republic of Germany is currently experiencing a dramatic downturn in the social anchoring of its party system.
The reasons for the declining membership of the German parties are seen in a series of processes of social and political changes that take place in the course of generational shift (see Wiesendahl 2006 as a prominent example). These processes are the individualization, the educational expansion, the resulting cognitive mobilization, the value change, the de-ideologization, and the disappearance of traditional social milieus. In the course of these processes, the individual's ties to large social groups, ideologies, and thus ultimately also to political parties diminish. At the same time, the individual's receptiveness to the expectations of the social environment decreases, whereas the orientation toward its own needs and desires increases.
If the outlined explanation of the decline in party membership is correct, then this decline ought to be accompanied by a change in the individual's motives for party membership. The members of upcoming generations should only join political parties if they expect specific personal benefits from a membership. However, as such advantages are generally associated with the need for active intra-party engagement, those people who join political parties now particularly intend to play an active role there. On the other hand, purely passive memberships, which are primarily intended to express support and affiliation, are becoming less relevant.
At the same time, changes can also be observed amongst the political parties, which are altering the incentives for joining a party and party membership. In the course of the frequently asserted development of mass integration parties, via membership of parties towards professionalized voter parties or cartel parties, the incentives for a purely passive membership should decline (Panebianco 1988;von Beyme 2000;Katz and Mair 1995). However, the change in the incentives for party entry, which are anchored in the parties' self-image and organizational structure, should not affect the motives of new members in a generation-specific way, but rather in dependency on when they joined the party.
Additionally, changes in the parties' self-image and organizational structure are likely to affect not only the motives for new members to join but also the motives for membership of those already in the party. This is because their incentive structure is also changing in the course of organizational change. Against this backdrop, it is advisable to survey and analyze not only the party members' motives for joining but also their current membership motives. This is also reasonable because, in general, the motives for joining can only be surveyed retrospectively, which might lead to distorted measurements. In comparison, the current membership motives should be surveyed with higher validity. Thus, the simultaneous analysis of entry and membership motives prevents substantive conclusions from being drawn solely based on potentially biased recalled entry motives.
Against the backdrop of the considerations outlined above, the aim of this paper is to empirically test the claimed gain in importance of selective outcome motives for joining a party and for being a member of a party. The analyses are based on the German Party Membership Studies of 1998, 2009 and 2017 (see for details Klein et al. 2019, pp. 81-84).
The following section of the paper uses Seyd and Whiteley's (1992) General Incentives Model to determine what incentives for party membership exist at all and why their relative weight should shift over time (Sect. 2). Then, the database of our empirical analyses is presented in more detail (Sect. 3). In the empirical part of the paper (Sect. 4), we first describe the development of membership motives over the period under study (Sect. 4.1). Subsequently, we examine the dynamics of change of each individual party entry and membership motive in detail using multivariate a(ge)-p(eriod)-c(ohort) analyses (Sect. 4.2). The paper ends with a summary of the most important findings and a few brief conclusions (Sect. 5).

Why and How Should Membership Motives Change?
The General Incentives Model by Patrick Seyd and Paul Whiteley is the standard model for explaining intra-party participation (Seyd and Whiteley 1992). It follows the tradition of the rational choice approach to the explanation of human behavior (Becker 1976) and is based on the assumption that people join a political party and become active within if the resulting benefits are greater than the associated costs. The General Incentives Model is essentially an attempt to systematically identify all conceivable positive and negative incentives for party membership and active involvement within a party. Once people are guided in their behavior by these incentives, corresponding individual motives for joining and membership emerge. In the further progress of this paper, the terms incentive and motive are therefore used synonymously.
A fundamental problem of the General Incentives Model is that in later publications Seyd and Whiteley Whiteley 2002, 2004;Seyd et al. 1996;Whiteley and Seyd 2002;Whiteley et al. 1994Whiteley et al. , 2006 presented and interpreted their own model in very different ways (see in detail Rohrbach 2013, pp. 36-55). As a consequence, other authors likewise did not interpret the model uniformly (see, for example, for Bürklin et al. 1997;Hallermann 2003;Müller and Traub 2004). Therefore, we will first explain how we have adapted the General Incentives Model for the purposes of the analysis presented here. Generally, it should be noted that the incentives described below are not mutually exclusive but can simultaneously determine people's behavior in different combinations and weightings.
When listing the positive incentives for party membership, the first thing to consider are the career opportunities associated with it. Party members potentially gain access to paid offices and mandates in politics, but also to professional functions in the party apparatus. In addition, they can potentially benefit from political patronage when filling positions in public administration and state-related sectors. Such advantages usually arise as a consequence of a long-term active engagement within the party and are exclusively granted to the member. In the General Incentives Model, they are therefore referred to as selective outcome incentives. Selective process incentives also exist. These are rewards that can arise directly from the member's activities within the party, such as the enjoyment of political work, the feeling of personal importance as well as social integration and conviviality.
The selective incentives just described can only become effective if a member is active within their party. In Germany, however, about half of the party members are completely inactive (Spier 2011, p. 111). What kind of incentives exist for such a passive membership? First of all, there are collective political incentives. These apply when a member believes that their membership contributes to the implementation of political goals that they consider desirable. It may come as a surprise that this incentive can also apply to passive membership. In fact, a member may be willing to believe that their membership and membership dues effectively strengthen their party's organizational power.
Additionally, the General Incentives Model includes four other types of incentives that can make passive membership in a party appear meaningful: Normative incentives are at work if the membership purposefully meets the expectations of the social environment. Altruistic incentives exist when a member believes that their membership contributes to the functioning of democracy. Ideological incentives occur when members want to commit themselves to certain values and ideological principles through their membership. Finally, there are expressive incentives, based on the opportunity of demonstrating support for the party and its politicians through a party membership. Active participation within the party may well further increase the member's benefit from these incentives, but it is not essential (Seyd and Whiteley 1992, p. 61).
For the sake of completeness, it should be mentioned that the General Incentives Model also takes into account negative incentives, i.e., the costs of party membership. These include the time spent on internal party activity, which is no longer available for alternative uses (opportunity costs). Second, active participation in political parties is not always just fun but can also cause disutility of labor. Finally, monetary costs exist at least in the form of compulsory membership fees. However, in research on the determinants of party membership, negative incentives of party membership are generally not systematically considered. This can be explained by the fact that only the fees are necessarily associated with membership. Their level, however, is "essentially below the threshold of saliency for practically everyone" (Seyd and Whiteley 1992, p. 78). Opportunity costs and disutility of labor, on the contrary, only arise when a member actively participates in party work. Consequently, they are also primarily relevant for explaining active intra-party participation.
Since its first publication, the General Incentives Model has been applied in a whole series of party membership studies and has proven itself well (see as a summary Rohrbach 2013, pp. 87-133;Power and Dommett 2020, pp. 506-507). Comparing the relative importance of the various types of incentives, it is evident that there are distinct differences between joining a party and being active within a party. In the past, selective outcome incentives played only a minor role in party entry, whereas collective political, altruistic, and expressive incentives were of particular importance. The results of those party membership studies that explicitly refer to the General Incentives Model show broad agreement with this. It applies regardless of whether simple frequency distributions of membership motives reported in membership surveys are analyzed (Bale et al. 2019;Bennie 2004;Heinrich et al. 2002;Klein 2006;Laux 2011;Rohrbach 2013;Seyd and Whiteley 1992;Whiteley et al. 1994Whiteley et al. , 2006 or multivariate statistical analyses based on population surveys are estimated (Hoffmann 2011;Klein 2006;Rohrbach 2013).
In contrast, selective outcome incentives are of paramount importance in explaining intra-party activity. Seyd and Whiteley already showed this in their pioneer study: "one of the most interesting findings is that one of the strongest motives for activism is political ambition, or a desire to build a career in politics at the local or national levels" (Seyd and Whiteley 1992, p. 8). Later membership studies, which explicitly referred to the General Incentives Model, were able to replicate this finding consistently (Gallagher and Marsh 2002;Heinrich et al. 2002;Klein 2006;Rohrbach 2013;Seyd and Whiteley 1992;Spier 2011;Whiteley et al. 1994Whiteley et al. , 2006Whiteley and Seyd 2002, p. 86).
Recently, however, it has been observed in Germany that selective outcome incentives are also gaining in importance for party membership. On the one hand, this can be seen from the increasingly frequent stating of such motives for joining and membership over time (Klein and Spier 2011, p. 36). On the other hand, an analysis of these motives differentiated by age groups also shows a clear pattern: The younger the party members, the more often they state selective incentives as a reason for joining the party and for being a member of the party (Klein 2006, p. 52;Klein and Spier 2011, p. 38; see also Seyd and Whiteley 1992, pp. 79-81, for Great Britain).
These age effects have been interpreted as generational effects (Klein 2006;Klein and Spier 2011). The idea behind this is that younger birth cohorts are increasingly guided by their own wishes and aspirations when deciding whether to join a political party (Gauja 2015, p. 91). In contrast, the behavioral expectations of society in general and their close social environment in particular would be of little importance to them (see for example Blais et al. 2004 for the erosion of the sense of duty to vote in younger generations). General trends of social and political change, such as the value change, the process of individualization, the educational expansion, the result-K ing cognitive mobilization, the de-ideologization, and the dissolution of traditional social milieus, are cited as causes of these changes (see Wiesendahl 2006). These processes have in common that their dynamics of change are essentially conveyed through generational succession (being most prominently discussed with regard to the value change, Klein and Pötschke 2004). Consequently, among the members of the younger cohorts, the probability of merely passive party membership, which at earlier points in time had promised emotional and social rewards, decreases. Their decision to join a political party should more strongly depend on the expected specific personal rewards from it. These are usually associated with the need for active internal party involvement.
However, the interpretation just described cannot be regarded as completely certain at the present time. The age dependency of party membership motives could also simply reflect life cycle effects. Accordingly, younger people have a higher interest in a career in political parties because they have not yet established another professional existence. With increasing age, interest in a political career should tend to decline as people manage to establish themselves professionally in other areas and moving up the ladder within the party becomes less likely (Bruter and Harrison 2009, pp. 1265-1266. In order to distinguish between a generational and a lifecycle interpretation of the observed age effects, the effect of cohort membership and the effect of age must be estimated simultaneously in multivariate statistical models. In this context, the age at the time of joining the party has to be taken into account in the motives for joining the party, and the age at the time of the survey should be taken into account for current membership motives. In order to clearly separate generation and life cycle effects, however, it is also necessary to control for period effects. In the case of the joining motives, this is the time point of joining the party; in the case of the membership motives, it is the time point of conducting the survey. As a result, complete Age-Period-Cohort (APC) models are estimated for both the joining and the membership motives.
However, life-cycle and generational dynamics of change cannot be separated based on cross-sectional surveys. Ideally, this requires long-term panel data or at least trend surveys with several sufficiently distant data collection points. However, the panel surveys conducted to date in party member research span far too short a time period Seyd and Whiteley 2002) and trend surveys of party members are extremely rare (see Kosiara-Pedersen 2015 for Denmark and Heidar 2015 for Norway).
A second obstacle to testing our propositions is that a majority of party membership studies ask respondents only about their main reasons for joining, rather than separately assessing the importance of a variety of possible motives on a scale. Given such a measurement, determinants of selective outcome incentives cannot be meaningfully analyzed, as such incentives are hardly ever reported as a member's main motivation. This might be due to many respondents' fear of being perceived as opportunistic when focusing on the personal benefits of being a party member (Garland 2016, p. 43). In contrast, for career-minded individuals, the separate assess-ment of motives provides the opportunity to positively evaluate selective outcome incentives while also favorably rating socially desirable motives. 1 In Germany, with the Party Membership Studies 1998 and 2009, there have been at least two comparable surveys at an interval of 11 years. With the third wave of the German Party Membership Study conducted in 2017, for the first time data are now available that allow for an analysis of the change in party membership motives within multivariate APC models. In contrast to the few other trend surveys of party members, the German data include separate ratings for the different incentives for joining a party and remaining a member. As described before, this measurement is a prerequisite for analyzing the change of selective outcome motives. In the empirical part of this paper, we will present such APC models utilizing these unique features of the dataset.
In addition to disentangling the life-cycle and generational dynamics, the estimation of multivariate APC analyses is also important because this is the only way to reliably identify the period effects on membership motives that have been alleged as a consequence of the organizational change of the parties. The background of these effects is the transformation of political parties from mass integration parties via membership parties to professionalized voter parties or cartel parties, which has been repeatedly asserted in the literature (Panebianco 1998;von Beyme 2000;Katz and Mair 1995). In the course of this process, the members become less important as a resource for the parties. For example, the unpaid work of party members in election campaigns would lose importance and increasingly be replaced by boughtin services from professional providers. In the course of an expanding state financing of political parties, the members would also become less important as a source of financing. Sometimes it is even argued that the membership develops into a "disturbing factor", which restricts party elites' possibilities of action and makes it more difficult for them to act professionally and maximize votes. As a result, the political parties would limit their efforts to attract a broad membership base and concentrate only on recruiting career-oriented activists who can potentially become part of the party elite. Incentives for inactive memberships of support or affiliation should therefore decline, whereas selective outcome incentives for party membership gain in importance (Lisi and Cancela 2019).
It could be objected that party members continue to have benefits for the parties (Scarrow 1994). For example, party members actively involved in the election campaign have a positive effect on the election result (André and Depauw 2015), and the financial contributions of members continue to be of enormous importance. This applies in particular to Germany, where the amount of state party funding is linked to the parties' self-generated financial resources. Ultimately, it must be clarified empirically whether the actual changes in the parties' self-image and internal organization go far enough to cause the alleged period effects on the motives for joining and membership.
In summary, it can be stated that we outlined two fundamental processes of change with regard to the membership motives of German party members, from which a number of hypotheses can be derived. First of all, in the course of social and political change conveyed through generational exchange, selective outcome motives should gain importance. This applies first and foremost to the motives for joining.
H 1 The younger the birth cohort, the more important were selective outcome motives for joining the party.
As the current membership motives remain decisively influenced by the original motives for joining, corresponding generational dynamics should also be reflected in the membership motives.
H 2 The younger the birth cohort, the more important are selective outcome motives for party membership.
In addition, we argue that, owing to the changing character of the parties, selective outcome incentives gain importance over time, independently of generational change. With regard to the motives for joining a party, we therefore state the following hypothesis: H 3 The more recent the year of party entry, the more important were selective outcome motives for joining the party.
When it comes to current membership motives, selective outcome incentives should be more important nowadays than in earlier years.
H 4 The more recent the time of data collection, the more important are selective outcome motives for party membership.
The question whether the claimed increase in importance of selective outcome incentives necessarily leads to a corresponding decrease in importance of all other incentives cannot be answered clearly. For instance, it cannot be ruled out a priori that career-oriented members may also claim to be interested in achieving political goals and strengthening democracy-even if only to conceal their essentially selfish motives. Non-selective incentives can therefore lose importance, but they do not necessarily have to. However, their importance should not increase in parallel with the selective incentives. This is because the relative weight of the different incentives would then remain the same, rendering an increase in importance of the selective outcome incentives meaningless.

Database: The German Party Membership Studies 1998, 2009, and 2017
The analyses in this paper are based on the German Party Membership Studies of 1998, 2009, and 2017 (see for Details Klein et al. 2019, pp. 81-84). These three studies are nationwide representative surveys of the members of the following six parties: CDU, CSU, SPD, FDP, Bündnis90/Die Grünen, and Die Linke (formerly named PDS). These are all parties that were represented in the German Bundestag at the time of data collection for at least one of the three studies. The members of the right-wing populist party "Alternative für Deutschland" (AfD) were not included because the AfD was only elected to the German Bundestag after the data collection in 2017. The random samples on which the German Party Membership Studies are based were drawn from the electronic membership registers by employees of the party headquarters on a uniform reference date. The procedure was specified in detail by the primary researchers and was the same for all parties and all three studies. The samples included 3000 persons for the parties operating nationwide and 2000 for the state party CSU, which is limited to Bavaria. In the case of parties not limited to one federal state, the samples were also disproportionately stratified: For the CDU, the SPD, Bündnis90/Die Grünen, and the FDP, the samples included 2000 persons from West Germany and another 1000 from East Germany. In the case of Die Linke, it was a little more complicated. Because its predecessor, the PDS, was de facto a regional party limited to the new federal states in 1998, the sample at the time included only 2000 people from East Germany. In the later party membership studies, an additional 1000 people from West Germany were included in the sample owing to the nationwide expansion of the party, now operating under the name Die Linke. In total, the samples of the German Party Membership Studies therefore included 16,000 persons in 1998 and 17,000 persons in 2009 and 2017.
The German Party Membership Studies were conducted as postal surveys at all three points in time. The main reason for this was that the membership registers maintained by the federal offices of the German parties did not include a valid telephone number or even an e-mail address for all members. Therefore, contact with the persons included in the sample had to be made by mail anyway. In addition, postal surveys carried out in accordance with Don Dillman's Total Design Method can generate comparatively high response rates (Dillman 1978). Indeed, the response rates of the German Party Membership Studies 1998Studies , 2009Studies , and 2017 are at pleasingly high levels of 66, 56, and 60%.
The questionnaires of the German Party Membership Studies were printed as brochures with 24 pages. For each individual party there was a specifically adapted version of the questionnaire. When designing the questionnaire, care was taken to ensure that the most important questions of the 1998 study were replicated identically in each of the two follow-up studies in order to enable an unbiased comparison over time.
The focus of the identical part of the three surveys was on the detailed operationalization of the General Incentives Model. The seven different incentives of the model were each measured with at least one item. In total, 14 items were used (see K  Table 1 for details). These items were surveyed with reference to the decision to join the party as well as with reference to the current membership. This is because membership surveys can only ask about the motives for joining the party retrospectively. As a result, respondents had to remember their party entry and to reconstruct the motives behind this decision. However, as the time point of the interviewees' party entry was on average 20 years in the past, distortions due to recall errors cannot be ruled out. In comparison, the survey of current membership motives should provide less distorted measurements. This is true even though a comparison of the frequency of missing values does not suggest that respondents had more difficulty remembering their motives for joining the party than indicating their current membership motives. For no single motive did the proportions of missing values differ by more than one percentage point. However, we had argued in the section Why and How Should Membership Motives Change? that the postulated effects of generation and period should be observable for the motives of party entry, as well as regarding the motives of current membership. Thus, by analyzing both types of motives, we subject our theoretical expectations to a very strict empirical test and avoid basing our substantive conclusions solely on the entry motives, which are potentially subject to measurement errors.

Empirical Analyses
The empirical analyses reported below are based on the cumulative data of the German Party Membership Studies of 1998Studies of , 2009Studies of , and 2017. In these studies, 10,373, 9413, and 9748 interviews were realized. Thus, a total of 29,534 cases are available for analysis. Our empirical analyses are based on weighted data. In a first step, the weighting variable corrects for the disproportional sampling. In addition, the members of the various parties are weighted relative to each other according to the number of members of their parties. Hence, the results of our analyses refer to the entirety of all members of the six parties we examined.  Table 1 shows the importance of the various motives for joining the party and the current membership in 1998, 2009, and 2017. For the sake of clarity, the percentage of respondents is shown that describe the respective motive as "important" or "very important." The column "delta" contains the percentage point difference between the corresponding shares for 2017 and 1998. This value should indicate any possible trend. Based on our theoretical considerations, we expect selective outcome incentives to gain importance over time. In fact, two of the three items measuring these incentives show an increase in the relevance attributed to them, both in terms of joining a party and party membership. This is true for the items "to gain public office" and "to gain a party office." An exception is the item "to gain job-related benefits"; its importance is stagnating at an extremely low level.
Two of the three items developed to capture selective process incentives also experience an increase in importance. These are "to be with nice people" and "doing party work for fun." The motive "to be better informed about politics", on the other hand, is of decreasing importance. Considering the dramatic increase in information on political topics on the Internet in recent decades, it seems perfectly plausible to us that this motive for party membership tends to lose importance.
Gains in significance can furthermore be observed in some nonselective motives. This applies primarily to altruistic, expressive, and collective political incentives. An increase in relevance can be seen here, particularly regarding the current membership. However, this is primarily due to the figures for 2017. The year was marked by a surge in political mobilization in response to the successes of the right-wing populist AfD, which was reflected in an increase in both turnout rates (Haußner and Leininger 2018) and the number of party members (Niedermayer 2022). This may have also led to a short-term upswing in some nonselective membership motives among party members.
Generally, it should be noted that the increase in importance of selective outcome incentives for joining a party and being a member of a party is not particularly strong. The reason for this is that in Table 1 Table 1.
In Table 2, the motives of the party members are therefore differentiated according to when they joined the party. Changes in motives should be observed more clearly here. The chronological classification is based on the survey dates of the German Party Membership Studies. For example, members who joined the party between 2010 and 2017 are only included in the Party Membership Study 2017. Members who joined the party between 1999 and 2009 are covered by the 2009 and 2017 studies. However, the figures in this column are based exclusively on the 2009 survey. In the case of the motives for joining, this should alleviate possible distortions caused by false or missing memories. For the aforementioned reasons, we only use data from the Party Membership Study 1998 for those members who joined their party between 1988 and 1998. For comparison, we also document the motives of the members who joined between 1931 and 1987, also determined based on the 1998 survey. The column "delta" shows the percentage point difference between the respective values of the members who joined before 1987 and those who joined between 2010 and 2017. Again, these numbers should indicate any possible trend.
The results from Table 2 are largely in line with our theoretical expectations: All three selective outcome incentives for joining a party become more important over time (8, 10, and 4 percentage points). The same applies for two of the three selective process incentives. In contrast, the expressive motive "because of impressive personalities at the top of the party" experiences a drastic decline in importance by 23 percentage points.
The results are quite similar for the current membership motives: Here, an increase in importance can be observed for all six selective incentives. There is also an  1931-1987 1988-1998 1999-2009 2010-2017 b 1931-1987 1988-1998 1999-2009 2010-2017 b

Selective outcome incentives
To gain public office   1931-1987 1988-1998 1999-2009 2010-2017 b 1931-1987 1988-1998 1999-2009 2010-2017 b Expressive incentives To      Linear regression models were estimated. Entries are the nonstandardized regression coefficients and the standard errors (in brackets) In addition to the independent variables documented in the table, gender, formal education, occupational status, region of residence (West/East), duration of membership, and party were also included in the model as control variables. However, for clarity and to save space, we do not report the coefficients of the control variables in the table. For the coding of these variables, see Table 5 Multicollinearity diagnostics are documented for the incentive "To fulfill my civic duties", as the lowest number of missing values was present here *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 increase in importance for the collective political, ideological, and altruistic motives. However, their increase is almost entirely attributable to members who joined in the last decade.

APC Models of the Determinants of Membership Motives
But what specific mechanisms lead to the changing importance of motives for joining over time? Theoretically, we had argued for generational effects (H 1) and effects of the date of party entry (H 3). But we also could not rule out the existence of lifecycle effects. Therefore, we will present a multivariate model of the determinants of the motives for joining a party. The effects of the variables, which are theoretically of interest, are estimated under mutual control. These variables include cohort (coded in birth decades), year of joining, and age when joining the party. In addition, we also include gender, formal education, occupational status, region of residence (West/ East), duration of membership, and the party as control variables in this model. The coding of these variables is documented in Table 5 in the appendix. However, for the sake of clarity and to save space, we do not report the coefficients of the control variables in the table.
Our multivariate model belongs to the category of APC models owing to consideration of the effects of age (A), period (P), and cohort (C). A challenge in the context of such models is the identification problem, which is caused by the "exact linear dependency among age, period, and cohort: period -age = cohort" (Yang 2008, p. 210). At first, this seems to make a separation of the effects of age, period, and cohort impossible. However, in our analyses, the identification problem does not exist owing to the use of cumulative cross-sectional surveys and because some of the variables are categorized. Although the age of joining, the year of joining, and the cohort membership are not independent of one another, none of these variables is perfectly predictable by using the other two. For example, if it is known that a person joined their party in 2000 and belongs to the cohort born between 1970 and 1979, then, at the time of joining the party, they may have belonged either to the group of under 24-year-olds or to the group of 25-to 34-year-olds. In this respect, multicollinearity exists, but there is no perfect predictability. To estimate the extent of multicollinearity, Table 3 (as well as Table 4) documents information on tolerance and variance inflation for each predictor.
The explanatory model described was estimated separately for each motive for joining. The reason for this is that the example of the incentive "to be better informed about politics" showed that even within one incentive category (in this case that of selective process incentives) the various incentives can develop very differently. We used the original five-point-scale as a dependent variable and regarded it as approximately metric. Therefore, multivariate analysis is performed using linear regression models.
The results of the multivariate analyses documented in Table 3 are highly informative. First of all, it can be seen that the younger the cohort, the more important are the three selective outcome incentives. The cohort effects we argued for can therefore be observed with great clarity. Hypothesis H 1 is thus confirmed. In addition to these K Table 4 An Age-Period-Cohort analysis of the motives for the current membership in a political party     Linear regression models were estimated. Entries are the nonstandardized regression coefficients and the standard errors (in brackets) In addition to the independent variables documented in the table, gender, formal education, occupational status, region of residence (West/East), duration of membership, and party were also included in the model as control variables. However, for clarity and to save space, we do not report the coefficients of the control variables in the table. For the coding of these variables, see Table 5 Multicollinearity diagnostics are documented for the incentive "To show my sympathy for the party", as the lowest number of missing values was present here *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 K cohort effects, life-cycle effects can also be observed. Selective outcome incentives tend to become less important for party entry as the age at joining increases.
Regarding the year of joining the party, a significant positive effect occurs for all three selective outcome incentives. The more recent the date of joining, the more important are these incentives. This is clearly in line with our theoretical expectations regarding the effects of the organizational transformation of the political parties. Hypothesis H 3 is thus also confirmed.
What remains is to consider the other motives besides the selective outcome motives. As explained in the theoretical part of this paper, it is not necessarily to be expected that the other motives will lose relevance in response to the increased importance of selective outcome motives. But they should at least not increase in importance in the same way, otherwise the importance of selective outcome motives would no longer increase in relative terms. And indeed, only the selective process motive "doing party work for fun" and the normative motive "owing to the influence of family and friends" show a clear increase in relevance in the cohort sequence. The former seems unproblematic to us, since it is also a selective incentive that is in principle included in our argumentation. The increase in importance of the normative incentive, on the other hand, is indeed unexpected. However, the majority of the other motives for joining do not gain in importance in the cohort sequence, which supports our general argument.
In a further step, the mechanisms behind the changing importance of the different motives for current membership are analyzed. The estimated model also belongs to the category of so-called APC models, as we consider age (A), time of survey (P) and cohort membership (C) as predictors. We again include gender, formal education, occupational status, region of residence (West/East), duration of membership and the party as control variables. As in the previous model, the identification problem does not occur in its pure form. For example, a party member of the birth cohort born between 1960 and 1969 interviewed in 2017 can either belong to the age group of 35-to 49-year-olds or 50-to 64-year-olds.
The results of the model just described are documented in Table 4. Again, the three selective outcome incentives show a clear pattern: the younger the cohort, the more important these incentives are to its members. This confirms the expectations presented in H 2. At the same time, a decreasing importance of selective outcome incentives can be observed over the life course, although these patterns are not continuous.
Regarding the selective outcome incentives, the period effects of the two survey years 2009 and 2017 do not meet our theoretical expectations. Although the aspiration for an office or a mandate is more important in 2009 and 2017 than in 1998, its importance in 2017 is not greater than in 2009. In the case of gaining job-related benefits, its importance in 2017 even falls back to the level of 1998. Hypothesis H 4 is therefore not confirmed. We assume that the specific effects of 2017 can be explained by specific historical circumstances. 2 Finally, we need to look at the other incentives. The main issue here is to ensure that the selective outcome incentives have indeed gained relative importance in the cohort sequence. And in fact, it turns out that none of the other incentives clearly became more important over the cohort sequence. The expressive motive "because of impressive personalities at the head of the party" even loses importance in the younger cohorts.

Conclusions
In this paper, the thesis was put forward that selective outcome motives for party entry and for party membership gain relevance over time. Two possible mechanisms were identified as the cause of this increase in importance: a supply-side mechanism based on processes of societal change that took place through generational succession and a demand-side mechanism focusing on shifts in the self-image and organizational structure of the political parties. The supply-side mechanism should lead to changes in the entry motives, whereas the demand-side mechanism should alter the incentives that potential and current members are exposed to.
To begin with, our comparison of the German Party Membership Studies of the years 1998, 2009, and 2017 has confirmed an increase in the importance of selective outcome motives for joining a party and for current membership. This is particularly evident when the analysis is confined to new members who joined their party shortly before the respective study was carried out.
Regarding the processes that might explain this trend, we have found empirical evidence for the alleged supply-side mechanism: the expected generational effects on selective outcome incentives for joining and for current membership are very clear. Members of younger cohorts tend to attach greater importance to these motives. The same applies to the selective process incentive "doing party work for fun" with respect to the motives for joining. At the same time, with a single exception, none of the other incentives for joining a party or for the current membership shows an analogous increase in importance based on cohort succession.
As far as the demand-side mechanism is concerned, our empirical results are contradictory. The positive effect of the year of joining on the importance of selective outcome incentives for party entry supports the existence of this mechanism. In contrast, the expected period effects could not be observed for the current membership motives. This may be due to the specific historical features of the survey dates. However, it could also reflect the fact that the organizational change of parties influences the joining motives of new members, but not the membership motives of those who are already in a party.
Scientific integrity requires us to clearly outline the limits of our empirical analyses: Although we were able to avoid the identification problem in our multivariate APC models, there is nonetheless substantial multicollinearity. In our judgement, the multicollinearity present does not make it impossible to estimate the relative effects of age, period, and cohort effects, but it may have led to slightly biased parameter estimates. Furthermore, the interviewees had to reconstruct their motives for joining. Some respondents joined the party a long time ago. This might have resulted in measurement errors due to distorted memories. In general, the period of analysis of just under 20 years is much shorter than one would wish for. Against this backdrop, it would be highly desirable to carry out a fourth wave of the German Party Membership Study in the near future.
Finally, a fundamental problem of our study is that the membership surveys we analyze by their very nature do not capture members who have already left their parties. Our analyses are therefore biased in that we can only examine the joining and membership motives of those individuals who have not revised their original decision to join a party. However, this is a problem that affects the entire body of international research.
Restricting the empirical analyses to actual party members may pose a problem in that there may have been members of older birth cohorts who joined their party with pronounced career ambitions, and then left the party when those career ambitions were disappointed. This would then also have led to the pattern we observed of a higher importance of selective outcome motives among party members from younger birth cohorts. With currently available data, however, a simultaneous analysis of current and former party members is not possible. Such an analysis remains a challenge for the future. Region of residence 0 = East Germany 1 = West Germany Education 1 = Low (up to "Realschulabschluss") 2 = Middle ("Realschulabschluss") 3 = High (at least "Fachhochschulreife") Occupational category 1 = Blue-collar worker 2 = Salaried employee 3 = Civil servant 4 = Self-employed 5 = Freelancer 6 = Have never been employed Membership duration Simple difference between the year of data collection and the year of party entry Party 1 = SPD 2 = CDU 3 = CSU 4 = FDP 5 = Bündnis 90/Die Grünen 6 = Die Linke