The Multidimensionality of Joining

By focusing on membership in voluntary associations in general, the question of whether correlates of affiliation vary by organizational type tends to have been neglected. This is a significant omission from the standpoint of describing the characteristics of “joiners.” Additionally, most of what we know about the reasons for belonging has been inferred from observations of who joins. In this study, nine frequently identified correlates of voluntary association membership were examined in relation to five different types of organizations. Only two, education and race, were found to be related to all types. Thus, results indicate that affiliation is not a unidimensional process and memberships in different types of associations are not interchangeable. The extensive literature on participation in voluntary associations provides considerable information about the characteristics of people who belong. This literature, in turn, has been used to develop rather elaborate theories regarding individual reasons for joining. An assumption underlying much of the empirical and theoretical work is that affiliation is unidimensional, i.e., that those social and demographic factors associated with belonging to one type of organization are equally relevant for other types. Indeed, one central proposition in the literature relates to the “cumulative effect” of social participation, suggesting that belonging to one organization increases the likelihood of belonging to another (cf. Smith and Freedman, 1972; Tomeh, 1973). Additionally, research has examined whether voluntary association membership is positively correlated with other forms of social participation, such as visiting friends and exposure to mass media (cf. Allardt et al., 1958; Babchuk and Thompson, 1962; Jacoby, 1966). A recent study by Edwards and White (1980) has extended this line of questioning by investigating whether selected demographic predictors of membership in voluntary associations also predict other types of social activity. Our search of the literature, however, has uncovered no studies which address this important issue with respect to membership in different types of voluntary associations. In other words, do the generally established social and demographic predictors of membership have equal relevance for all types of voluntary associations? Are they unidimensional or multidimensional? The answer to this question is especially important with respect to theories both about the characteristics of people who join voluntary associations and why they belong (cf. Smith, 1975). Three major methodological approaches have been used to study the correlates of voluntary group membership. The most frequently used strategy considers either who belongs 36 Williams & Ortega in NoNprofit aNd VoluNtary Sector Quarterly 15 (1986) to an organization (cf. Curtis, 1971; Hyman and Wright, 1971) or the number of organizations belonged to (cf. Knoke and Thomsen, 1977; Klobus-Edwards et al., 1978; Edwards and White, 1980), without reference to the kind of association it is. In this regard, McPherson (1981:708) comments that, “There are literally dozens of articles in the voluntary association literature which simply count the number of memberships, treating each as interchangeable with the other.” Consequently, it is unclear from this approach whether affiliation is a general propensity or if different factors lead to joining depending upon the particular nature of the organization. In contrast to the studies combining memberships, a few have dealt with belonging to a single organization or specific type of association (cf. Sills, 1957; Schmidt, 1980). It is clear, however, that knowing the characteristics of members in one group or type of group does not allow for a determination of whether these characteristics are those of joiners, regardless of the kind of organization, or of those choosing to affiliate with that specific kind of association. A third category includes those studies that have examined belonging to various types of groups, but that have focused only on specific substantive issues about who joins, such as whether blacks participate more than whites (cf. Williams and St. Peter, 1977; Guterbock and London, 1983), the elderly participate more than others (cf. Cutler, 1977; Babchuk et al., 1979), or if middle-class persons participate more than those in the working or lower class (cf. Booth et al., 1968). These studies have either relied on tabular analysis of one or two independent variables or have focused on the independent effects of a single variable without presenting the data necessary to assess how the control variables are related to affiliation (cf. Babchuk and Booth, 1969). For instance, Jacoby (1966) found that living arrangements had different effects depending on whether membership was in expressive or instrumental organizations. Babchuk and Booth (1969) found age, sex, and age by sex differences in types of organizational memberships. Similarly, Edwards et al. (1984) found sex differences in types of organizational affiliations, even after controlling for such variables as occupational prestige, marital status, and income. None of these studies, however, allows one to determine whether these variables, e.g., age, sex, race, have independent, differential effects by type or organization. Thus, these and similar studies provide useful information about whether specific factors are associated with joining more than one organization, but they do not answer the more general question of whether there are a set of characteristics associated with joining. Those few studies that more directly examine the unidimensionality/multidimensionality of formal voluntary association membership appear to challenge the widespread assumption of unidimensionality. Using factor analytic techniques, Houghland (1979) found that two dimensions underlie membership in different types of organizations. Houghland and Christenson (1982) provide evidence that values differentially affect belonging to various types of organizations, independent of covariates such as income, age, education, and community size. Thus, while limited to a North Carolina sample, their data clearly suggest a multidimensional process. Some theoretical work pertaining to membership also suggests the multidimensionality of joining. For example, from Lemon et al. (1972) it can be predicted that status characteristics will be differentially related to membership in that the emphasis is placed on the differing goals and membership requirements of groups rather than the characteristics of joiners. the multidimensiOnality Of JOining 37 In contrast to the many studies that have used community or regional samples, we use a representative national sample of 3,075 adult Americans to examine the independent and interactive effects of nine social and demographic variables, often reported to be important determinants of affiliation, on five of the most common forms of voluntary associations. Sample and Research Design The data for this study were collected by Response Analysis of Princeton, New Jersey in 1973. It is a probability sample of persons age eighteen and older residing in the contiguous United States. From the occupied and eligible households drawn for the sample, 3,540 successful interviews were completed yielding a response rate of 70 percent. The refusal rate is 17 percent with the other non-completions stemming primarily from designated respondents not being home after four visits. The samples of minority racial groups other than blacks were deemed too small for proper analysis. Thus, these cases, along with those for whom complete information is not available, were deleted, leaving 3,075 cases fur the present study. For further details about the sample, see Nunn et al. (1978). Voluntary association memberships were measured by showing respondents a card listing five different types of organizations. The types are church-related, job-related, recreational, fraternal/service, and civic/political. (An “other” category was included also, but is not used in the present study since the focus is on membership in specific types of associations.) The respondent was asked, “Which of these types of organizations, if any, do you belong to?” Recall was aided by providing the respondent with examples of different types of associations. For example, Lions, Masons, Eastern Star, and Rotary were mentioned in relation to fraternal/service organizations. The nine independent variables selected for analysis are: education, race, gender, age, marital status, whether the respondent is a household head, number of children under 18 years of age residing at home, size of the respondent’s community and region of residence. These variables are the ones figuring most prominently in the sociological literature dealing with participation in voluntary associations and thus the ones which should receive primary attention. Region provides a useful control in that many of the previous studies are based on samples drawn from a single area of the nation, leaving the generalizability of their findings unknown. The method of data analysis is the linear probability estimation technique developed by Grizzle et al. (1969). This procedure allows one to estimate the effects of independent variables on a dichotomous dependent variable, while simultaneously controlling for the effects of all other variables in the analysis. The model is similar to regression analysis and the interpretation of the resulting coefficients is nearly identical. One major difference, however, is the substitution of an appropriate Chi-Square test of significance for what would be a questionable use of an F-test (Swafford, 1980).