Three Types of Voluntary Associations in Comparative Perspective: The Importance of Studying Associational Involvement through a Typology of Associations in 21 European Countries

Abstract

Although very different types of voluntary associations are often lumped together in (cross-national) analyses, we argue that scholars should distinguish between types of associations. In the first part of the paper, we develop a typology of voluntary associations based on theoretical considerations and previous empirical analyses of the primary concerns of these voluntary associations: leisure organizations fulfil recreational purposes, interest organizations aim to represent the interests of their members, and activist organizations advocate broad societal interests. We present a measure that is cross-nationally equivalent. The second part of the paper illustrates the validity and relevance of the typology for studies of associational involvement. First, the Mokken scale analysis shows that the repertoire of activities (membership, participation, volunteering, and donating money) that citizens employ differs across the three types. Second, cross-national analysis shows that the ratio between involvement rate and the share of volunteers in voluntary associations differs across the three types. Finally, the three types of associations are differently related to the (supposed) causes and consequences of associational involvement. We illustrate that many of these differences cancel each other out when we do not distinguish between leisure, interest and activist organizations. All in all, this article proves that the distinction between leisure, interest, and activist organizations has significant, substantial, and theoretically relevant outcomes.

Keywords:

• Voluntary associations
• volunteering
• schools of democracy
• associational involvement

Notes

Although this approach is strong in its theoretical argumentation, its weakness is the lack of empirical foundation.

This empirical approach starts from the questionable assumption that members of any organization are likely to become members of similar organizations (e.g., chess players would also embark on playing checkers) which seems doubtful (Van der Meer, 2009).

The typology of Maloney and Roßteutscher is more refined than the one we propose: they distinguish two types of leisure organizations (around family concerns and around leisure concerns), two types of political organizations (around ‘old’ politics and around ‘new’ politics), and three types of market organizations (around general welfare, around group-specific welfare and around traditional socio-economic interest representation).

The large number of measures in the ESS is not even extraordinary for recent cross-national survey research in this field. The unofficial record holder is the CID with four types of involvement in 27 (!) types of organization, leading to 108 measures.

From the 12 types of voluntary associations we leave out political parties and religious/church organizations. We consider participation in a political party as political rather than civic participation, as it falls under the label of ‘legal activities by private citizens that are more or less directly aimed at influencing the selection of governmental personnel and/or the actions they take’ (Verba, Nie & Kim, 1978). Political parties are professionalized campaigning parties, rather than the mass parties of before (Katz & Mair, 1994). Participation in religious and church organizations is left out as ‘church membership (…) may be somewhat less “voluntary” than other types of association involvement, even though most adults are formally free to change church memberships and sometimes do’ (Curtis, Grabb & Baer, 1992; see also Van Oorschot & Arts, 2005).

This typology is based on the primary concern of these voluntary associations. We do not support this form of data reduction with factor analysis (see paragraph 2 of this article).

Suppose we have two respondents. Respondent A is a member of no less than six sport clubs. Respondent B is a member of one sport club and one cultural club. Although respondent A has the most memberships, respondent B would score higher (i.e., ‘2’) than respondent A (i.e., ‘1’). Although it is sometimes used as such, the sum scores simply do not measure intensity but merely variety of civic participation (c.f. Morales & Geurts, 2007).

In this case with four items, there would be five correct orderings in the Mokken scale procedure. From the most common activity to the least common activity these correct orderings are: 0000, 1000, 1100, 1110, and 1111. A score of, for instance, 1010 would violate the rule that people who engage in a less common activity also engage in all more common activities. These violations are reflected in the H-coefficient, in which scores of 0.30 to 0.40 reflect a weak scale, and scores higher than 0.50 reflect strong scales.

Consequently, the score of 0 represents no involvement, 1 represents membership, 2 represents membership and active participation, 3 represents membership and active participation and volunteering, 4 represents membership and active participation and volunteering and donation of money.

Probably the respondents in Finland, Israel, and Italy classified themselves under the header that fits them best (f.i. either a member, or a volunteer), whereas the respondents in the other countries in the ESS data set reported for each type of activity whether they participated in that manner or not (f.i. both active as a member and as a volunteer). These differences may in turn be caused by translation, instruction or interpretation differences.

The three exceptions to this rule are activist organizations in Hungary, The Netherlands, and Slovenia.

A test on monotone homogeneity implies that if the value of the latent variable increases, the probability of a positive response to the item also increases. A stricter test is the one on double monotonicity, implying that the item response functions do not intersect. In case this assumption holds, it implies that the order of the manifest probabilities reflects an ordering of items according to their popularity that is uniform across sub-groups, i.e., countries in our case (Sijtsma & Meijer, 1992; Sijtsma & Molenaar, 2002).

Moreover, additional analyses show that these differences in effect sizes are stable across countries. The types of organizations do not have exactly the same effect on political action, but at least the same relative importance cross-nationally.