Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Soc Forces. Author manuscript; available in PMC 2012 Sep 17.
Published in final edited form as:
Soc Forces. 2011 Dec; 90(2): 453–473.
doi:  10.1093/sf/sor035
PMCID: PMC3444253

Understanding Activist Leadership Effort in the Movement Opposing Drinking and Driving


Why do some social movement leaders work harder than others? And, how does gender affect the patterns we uncover? Utilizing historical case study evidence of local chapters in the emerging movement opposing drinking and driving we are able to develop and test theoretical expectations about predictors of weekly effort among MADD and RID leaders. Taken together, our model explains 45 percent of the variation in leadership effort. We find bureaucratic complexity and victim support activities are more powerful predictors of effort than are individual leader characteristics, although all are important. Further analysis reveals that gender almost wholly conditions the strong effect of bureaucratic complexity on leadership effort so that increasingly complex chapter structures are associated with substantial increases in work hours for women but not men.

Why do some social movement leaders work harder than others? Although few would doubt the central importance of activist leadership in both creating the organizational vehicles for collective action as well as facilitating subsequent collective action by members (Morris and Staggenborg 2004; Oliver and Marwell 1992), little is known about why some leaders devote more time to leading their groups than others. A key reason for the gap in the literature is the lack of available data on social movement leaders in contrast to the more commonly studied rank-and-file members. Or perhaps the paucity of empirical work results from a lack of attention by social movement researchers to specify the mechanisms that may explain variation in leadership agency. Nevertheless, consistent with the observations of many recent analysts, “…leadership in social movements has yet to be adequately theorized… result[ing] from a failure to fully integrate agency and structure in theories of social movements.” (Morris and Staggenborg 2004:171; for an important exception, see Ganz 2009) In this article we aim to make a contribution to the understanding of one aspect of leadership agency – the amount of effort leaders expend in leading. And we do so by surveying chapter presidents of the movement opposing drinking and driving in the United States during its emergent phase (1985) more than two decades ago. Because that movement was framed, to an important extent, around grieving mothers (Weed 1990), and its leadership was and is composed predominantly of women, our theoretical approach necessarily makes gender a central factor in understanding leadership effort.

After first describing the activist leaders who are the focus of our attention here, we nest our first research question – How best to explain variation in the number of hours local leaders spend leading their group in a typical week? – in the theoretical discourse developed to account for activist participation in general. We then extend those explanations to the question of leadership effort, suggesting that theories applicable to the rank-and-file may be adapted to account for variation in levels of effort by leaders, especially by considering the importance of both structural and individual factors that may enlarge or diminish the demands placed on a leader. We conclude our theoretical arguments by bringing the question of gender to the foreground, suggesting that women may be expected to respond to organizational context in distinctive ways that contrast with men’s typical responses.

The Activist Leaders of MADD and RID

Beginning in the late 1970s and gaining momentum over the next decade, the social movement opposed to drinking and driving had spawned more than 450 local chapters by 1985, with the vast majority of groups affiliated with Mothers Against Drunk Driving. Fewer groups were associated with Remove Intoxicated Drivers, which had begun earlier than MADD but remained far more decentralized with its national offices providing fewer chapter services and less support than MADD (McCarthy and Wolfson 1996; McCarthy, Wolfson and Harvey 1987; Weed 1990).1

Our evidence was gathered from the presidents of local MADD and RID chapters and is an especially appropriate sample for investigating leadership effort because the estimates of hours worked each week vary quite dramatically among these leaders, with 25 percent of the presidents devoting less than five hours a week to their leadership activities while 43 percent devote more than 16 hours a week to being leaders, equivalent to holding a part-time job, and 19 percent report spending more than 30 hours a week on average, almost equivalent to holding a full-time job. Consistent with the public image of the movement and reinforced by the name of the dominant federation of groups, MADD, over three-quarters of the presidents were women and more than half had a school-aged child in the home. And in line with victims’ organizations more generally, the national face of this movement was dominated by self-identified “victims” such as Candy Lightner, the founder of MADD, who lost a daughter to a drunk driver. At the chapter level, however, little more than 50 percent of the presidents were victims as strictly defined by movement members as a person who had been (8%) injured or killed by a drunk driver or who had a close relative (44%) who was.

The local groups these leaders headed were, for the most part, rather modest enterprises. They reported an average membership of about 125 people (median = 35, with 80% reporting less than 100 members) and a mailing list with about 275 names. The median annual revenue of these groups was less than $1,500, and there was an average of about 50 aggregate hours of volunteer effort a month, and less than 20 members in attendance at typical chapter meetings. However, there was great variation among the groups, with some being relatively inactive while others showed signs of great vitality.

The chapters focused their efforts on three primary kinds of activities generally identified by members as public awareness (designed to bring the issue of drunk driving to wider public attention), legislative action and victim support services. Of these, victim support tended to be the most labor intensive. Although chapters varied widely in their services, most groups experienced great pressure to provide these resources because of the inevitable interactions between chapter leaders and each new cohort of local victims who sought relief from their personal grief resulting from drunken driving incidents. Groups developed distinctive profiles for dealing with their case loads such as informal counseling with new victims, taking on longer-term management of specific victim cases as they worked their way through the court system, as well as the development and management of victim support groups. Beyond the burden of managing the day-to-day activities of their local chapters, presidents found the provision of victim support activities to be an especially time consuming and emotionally demanding set of tasks.

The data we draw upon describe the activities of local anti-drunk driving groups in 1985. At that time the movement was thriving in several countries, including the United States, Canada and Great Britain (Rock 1988a; 1988b). In the United States the movement opposing drinking and driving had achieved wide national recognition as part of a broader victim’s rights movement that was integral in passing the Federal Victim and Witness Protection Act and in institutionalizing a “Victim’s bill of rights” in more than 30 states (Weed 1995). Furthermore, the group was tied to the successful passage of more than 130 state-level anti-drunk driving laws and in raising the federal minimum drinking age to 21.2 At the time of the survey the movement was at or near its peak and enjoyed tremendous political success and organizational legitimacy. At the local level, these groups were operated fairly independently from one another with only some oversight from the national associations. In this relatively laissez faire setting, chapter presidents had the opportunity to shape the local groups with wide discretion, resulting in chapters which varied tremendously in their size, organizational complexity and community outreach activities.

Theoretical Accounts of Activist Participation

Why do citizens participate in collective action? This question has motivated a substantial and diverse research effort since the rebirth of social movement scholarship three decades ago (Lofland 1996). Most of this work attempts to explain whether potential activists cross the threshold of participation, but rarely seeks to explain how active participants are after they become engaged. There are, of course, a few exceptions (Barkan, Cohn and Whitaker 1995; Cohn, Barkan and Halteman 2003; Passy and Giugni 2001; Wiltfang and McAdam 1991). What we want to explore in this research, however, is the level of effort expended by an activist leader as she orchestrates the many tasks that are embodied in her leadership role. Unfortunately, because little research has addressed this topic which might guide our expectations, we draw on past studies regarding movement initiation and the participation of rank-and-file members to help generate hypotheses that examine variation in the amount of effort leaders devote to leading.

Much of the prior work on SMO membership and participation tends to stress either the structural approach – how extensive and active the organization is and how well it turns bystanders into adherents and adherents into organizational constituents, or the individual attributes approach – the personal characteristics that motivate individuals to respond positively to organized attempts to engage them in social change efforts.3 Substantially more of this research has focused on individual characteristics than the contextual features of social movement organizations. And scholars consistently find weak associations between most individual-level indicators and initiation into a movement4 (Schussman and Soule 2005) and modest support for individual characteristics driving levels of effort after one joins (e.g., Barkan, Cohn and Whitaker 1995; Cohn, Barkan and Halteman 2003; Passy and Giugni 2001; Wiltfang and McAdam 1991). Among the latter studies, however, there are consistently strong findings indicating that micro-structural network links to other movement activists, a contextual feature of these environments, frequently have significant explanatory power. In line with these findings, we believe that the question of why some leaders work harder than others can be addressed by jointly assessing the structural and individual-level accounts of activism – what do citizens bring to the civic arena (individual) and what contextual forces (structure) shape the way these things determine the extent of participation?

Extending Structural and Individual Accounts to Explain the Effort of Leaders

Leading Structural Accounts

The key mechanisms for structural accounts of participation are the extent to which potential activists have social network connections to ongoing mobilization efforts and whether or not they are asked to participate. Activist leaders supply potential activists with opportunities to participate by contacting them (Schussman and Soule 2005), and they try to motivate existing activists to become more involved by directly requesting their help (Oliver and Marwell 1992). It is the historically specific contingencies that provide opportunities for activism that seem most important here. That is, particular movements emerge and their emissaries reach out to mobilize potential activists through a variety of channels putting a small subset of citizens at substantially higher risk of being asked to participate, which in turn predicts involvement in movement activities. But being a leader means you are the asker, not the one who is asked.

Is there a functional structural equivalent of leaders being asked to participate? We argue that organizational vitality, in this case tapped by general organizational complexity and the diversity of victim support services offered by a group, functions to create a series of pressures that expand or contract the amount of effort a leader is expected to devote to leadership activities. The more organizationally complex5 a group is in terms of recruiting and fielding many volunteers, engaging in wide contacts with other groups in the community, sustaining a large number of task committees, and meeting with other officers quite regularly, the more the demands on the leader’s time can be expected to rise. To the degree that a group takes on labor-intensive tasks related to its social change goals – in this case providing more or less diverse and extensive victim support services such as putting an emphasis of victim assistance, working with specific victim cases, and developing victim support groups – a leader’s responsibilities can escalate. To the extent that this logic is convincing, then, we would expect the necessities of leadership within the most active groups will require chapter presidents to devote more time to leading. This reasoning motivates our first hypothesis:

  • H1: The organizational vitality of the chapter (assessed as bureaucratic complexity and victim support services) is positively related to presidents’ weekly leadership effort.

Individual Level Accounts

In addition to our structural hypothesis, we believe there are individual-level characteristics which can also be useful in predicting leadership effort. Individual-level accounts are designed to explain who participates in social movements and often focus on the stable personal characteristics of potential activists, such as their gender, level of formal education, their feelings of efficacy, their ideological commitments and the extent to which they experience grievances central to the movement’s goals. The last three factors are considered components of cognitive availability for participation, but it is also important to consider whether people are biographically available to participate (McAdam 1986). This is usually assessed in terms of the opportunities and constraints associated with work and family life, such as the number of children a person has, marital status, job demands, and the amount of free time available, all of which are thought to increase or decrease the likelihood that potential activists who are personally motivated and asked to participate will actually do so. Prior research on participation rates among rank-and-file members indicates that an individual’s commitment to a movement’s concerns and grievances (cognitive availability) is often a better predictor of effort than biographical availability (e.g., Barkan, Cohn and Whitaker 1995; Cohn, Barkan and Halteman 2003; Passy and Giugni 2001; Wiltfang and McAdam 1991). We suspect the lack of strong findings may be linked to the truncated range of effort seen among these activists– since most devote only a few hours a week to their respective movements they may not be spending enough time away from work and home to create a conflict between their activism and their everyday life. We anticipate that by examining a wider range of effort that includes those who volunteer a larger number of hours each week we might see stronger results in this area. Consistent with this general line of argument, we anticipate that weekly effort will be influenced by a president’s cognitive availability, measured here as objective victim status, and biographical availability, measured as work and family obligations.

In terms of cognitive availability, we presume that victim leaders will, all other things being equal, devote more effort to their leadership activities than non-victims because they are more strongly motivated to advance movement goals (see Jennings 1999). For anti-drunk driving activists, being a victim means that one’s spouse, child or another close relative has been killed or seriously injured by a drunk driver, or that oneself has been seriously injured. Many social movement analysts suggest grievances that stem from personal victimization – known as objective grievances – are often motivating forces because “Felt injustice is at the root of any protest.”(Klandermans 1997:205) A key cognitive process underlying the transformation of objective grievance into action rests on the understanding that personal troubles are the result of public problems, and a corollary belief that these public problems can be solved through activism (see McCarthy 1994 relevant to this movement). Yet most of those who experience appropriate grievances do not get involved in social movements in the first place, so the role of grievance in explaining who crosses the threshold of any participation has received little empirical support. What is unusual about our approach, however, is the assessment of whether or not objective grievances – when they are widely understood as grievances – predict levels of effort among leaders already deeply immersed in a movement. This reasoning leads us to our second hypothesis:

  • H2: The cognitive availability of the leader (assessed as personal grievance stemming from drunken driving victimization) is positively related to presidents’ weekly leadership effort.

A second important component of individual-level factors is how a president’s biographical availability influences his or her weekly labor contribution. Our approach to this question relies on a widely-held economic perspective suggesting that people substitute time in one arena for time in another. According to this time-constraint approach, people who are married, have children or are employed outside the home have more demands placed on their time than those who are unmarried, have no children or are unemployed. The more time-constrained an individual, the less ability he or she will have to participate in other activities, including volunteer work (Hook 2004). This expectation flows from Becker’s (1981) time availability hypothesis, which suggests that because people are limited by family and work needs, individuals must act rationally to allocate their time in the most effective manner possible. When applied to social movement activism, a time-constraint approach can be used to think about how time is allocated to movement labor when it conflicts with actual or perceived work and family responsibilities. Because volunteer activities come in addition to the performance of work and family roles it is often found that as the number of responsibilities in the home or office increase the amount of movement participation decreases (McAdam 1986). Adopting this perspective leads us to expect that leadership effort will decline as family and work obligations rise regardless of any structural pressures for more extensive effort. Squeezing out an hour or two a week to devote to activism seems possible no matter how busy a person’s life, but finding an extra 15 to 20 hours a week is likely far more difficult. We build our fourth hypothesis on this time substitutability notion:

  • H3: The biographical availability of the leader (assessed as work and family obligations including employment, marital status and number of children) is negatively related to presidents’ weekly leadership effort.

A final individual factor affecting biographical and cognitive availability is the gender of a leader. While feminist scholars have paid extensive attention to the role of gender in feminist movements, the empirical work on gender as a main effect within non-feminist movements is less well understood, although there have been some notable exceptions (see for example Adams 2002; Einwohner 1999; Einwohner, Hollander and Olson 2000; McAdam 1992; Taylor 1999). The present study allows the opportunity to address women’s participation in an SMO led mostly by women, but one that is not typically thought to be a part of the feminist movement. Given the strong nurturing focus of this movement, the family friendly organizational form, as well as the gendered identifier of the lead federation as “mothers” we believe that, all else being equal, women leaders are likely to become more invested than men and will therefore spend more time each week tending to leadership activities. Therefore, we hypothesize that:

  • H4: The gender of the leader will have a direct impact on weekly leadership effort such that female presidents will devote more time each week to the organization than male presidents.

In addition to its possible direct effect on weekly effort, it is plausible that gender may interact with other factors that drive leaders’ levels of participation. Because gender embodies learned attitudes and behaviors that differ for men and women it is reasonable to assume that gendered realities may influence the relationship between participation and the other key variables such as work and family constraints, grievances and organizational complexity in unique ways. As noted by McAdam (1992:1215) in discussing gender differences in movement participation, “Given the male/female differences in socialization and the powerful ways that differences in gender-based behavioral expectations and opportunities get embedded in social structure, why should we presume that males and females experience any social phenomenon in the same way?”

While research on activism provides only a few hints about how the gender of an activist interacts with other factors that produce levels of effort, we can extend the time constraint approach to consider whether men’s participation will be as heavily affected by work and family obligations as women s. Because women perform more child-related duties than men (West and Blumberg 1990) and are less likely than men to view their housework as discretionary (Hook 2004), even when men and women have an equal number of responsibilities in the home women may be more likely to curb their weekly activist effort in order to fulfill familial obligations. Similarly, gender norms may lead men and women to view their activist obligations differently even when they work the same number of hours. For example, while men may be more likely to choose employment over activism as a key to fulfilling their role as provider, women may be willing to work more hours for the movement because it provides an opportunity above and beyond their employment to reinforce feminine goals of being nurturing and compassionate (see Musick and Wilson 2008 for findings that are consistent with such an interpretation in the volunteering literature).

In addition to the potential interaction of gender with individual attributes, it is equally plausible that gender will interact with structural characteristics of collective action environments. This is consistent with a research literature on leadership styles which finds that women are more sensitive to the pressures of leadership created by highly complex bureaucratic structures. In an extensive meta-analysis of 28 studies of gender and leadership style, Eagly and Johnson (1990) found consistent differences between men and women in whether they led their organizations in a democratic (allowing subordinates to participate in decision making) or autocratic (discouraging subordinates from participating in decision-making) manner Women were consistently more democratic than men, even though most of the studies were done in highly bureaucratic organizations. The authors reasoned that in such settings one might expect organizational role demands to dampen any gender differences stemming from gender role socialization. Even in those settings, however, gender differences remained strong. Their findings suggest that women, through their distinctive choke of leadership style, may work more hours than men as the bureaucratic complexity of their groups increases. Further given the female dominance of this movement, the men who become leaders may be different, and possibly less engaged, than men who become leaders of male-dominated movements. This leads us to our final hypothesis:

  • H5: The gender of the leader will have a moderating influence on weekly leadership effort such that group and individual context will be more influential in predicting women’s increased effort than men’s.

Data Collection and Measurement

The data used in our analyses come from a survey of anti-drunk driving activists who were the leaders of MADD and RID groups known to be in existence in 1985. The investigators used a self-administered survey (during 1986) which asked questions about each organization, its members and activities, the extent and scope of its community involvement, and the leaders themselves in 1985. The survey was mailed to the president of each group, as identified in the census of 458 groups. Follow-ups were made by telephone and mail. There were completed responses from 370 groups, comprising a 78 percent response rate (see McCarthy 1994; McCarthys Wolfson and Harvey 1987; McCarthy and Wolfson 1996). At the time of the survey 79 percent of the leaders were women with average tenure as president of a little less than two years (Table 1 provides a description of characteristics for all presidents, as well as by MADD and RID affiliation). Most presidents worked full time or parttime (64%), were married (86%), and had school-aged children in the home (54%). Leaders tended to have some college education (a mean of 14.6 years) and were on average about 44 years old. And, although the founders of anti-drunk driving groups are most often people with little or no previous experience in mounting citizen’s advocacy campaigns (McCarthy, Wolfson and Harvey 1987), most (75%) were involved at the time of the survey in an average of two other community organizations in addition to the chapter they oversaw, with the modal response of being very active (43%) in these groups.

Table 1
Characteristics of Chapter Presidents Affiliated with the Movement Against Drinking and Driving in 1985, Means Reported

Dependent Variable

The chapter presidents reported very high levels of weekly labor which ranged from 2 to 33 hours per week (mean 15 hours). Their reports serve as our dependent variable in the analyses and our measure of leadership effort. To put this level of effort in context we note that while 75 percent of these leaders reported spending more than three hours a week in their chapter activities, an Independent Sector Report for the same year estimated that only 20 percent of a national sample of U.S. adults reported spending that much time in any volunteer activities. In fact, most Americans donated much less time than this to volunteer work. Analysis from the 1985 American Time Use Survey suggests that on average, Americans spent about an hour a week in all organizational activities combined, including political union and/or social movements (Bianchi, Robinson and Milkie 2006). Of course the vast majority of Americans were not leaders of voluntary associations, and so we would expect them to devote much less time than the chapter presidents who are nor only activists, but oversaw other volunteers.

Independent Variables

Our key independent variables are measures of the structural characteristics of local groups and the individual attributes of leaders. Table 2 provides descriptive information on these measures as well as the bi-variate correlations between each item and presidents’ weekly effort. This table also reports the mean differences for male and female leaders, and whether the gender disparities are significant for each variable of interest. See Table 2 for more details.

Table 2
Descriptions and Correlations of Variables of interest with Presidents’ Weekly Effort in the Movement Against Drinking and Driving in 1985

Structural Characteristics

Bureaucratic Complexity is a standardized and summed scale of four variables that tap into the organization demands of each chapter, including the number of volunteer hours that must be supervised each month, the number of committees formed by the group, the frequency of chapter meetings, and a community outreach variable constructed as the proportion of contacts each group had with 34 existing community groups such as the State Department of Transportation, Alcohol Safety Action Program, the local police, the Mayor or County Executives’ office, etc.. The alpha reliability for this scale is .6010.6

Victim Support Services is a standardized and summed scale of three items regarding the amount of victim services provided to the community, including whether or not the chapter has an emphasis on victim assistance, whether the group works with specific victim cases, and whether the chapter has developed victim support groups. The response categories for these questions were 0 = none to 3 = a great deal. The alpha reliability for this scale is .8329. We also evaluated the adequacy of both measures by conducting individual unrotated principal-component exploratory factor analyses in STATA. For both scales, the factors patterned as predicted and the results exceeded standard cutoffs. Bureaucratic complexity had an eigenvalue of 1.83 and factor loadings that ranged from .58 to .73 while victim support services had an eigenvalue of 2.25 and factor loadings that ranged from .80 to .92.

Individual Attributes

Victim Status was the primary measure of cognitive availability and was assessed with a single question that asked whether or not the leader was a victim or a victim’s relative, as was the common nomenclature for the term “victim” among participants of this movement (victim = 1, non-victim = 0). For practical purposes, our measure of victim status is a proxy for cognitive grievance and will be evaluated as such in the final models. Biographical availability for participation was measured with four separate items, including: (1. a binary variables assessing employment status (1. = employed, 0 = not employed), (2. a binary measure of marital status (1 = married, 0 = not married), (3. a continuous measure of the number of school aged children living in the home (reported range from 0 to 3), and (4. gender (1 = female, 0 = male). In addition to these items we considered several other individual characteristics in our models, including race, age, education, the number of children of any age, the number of very young children, and broader constructions of marital status (including divorced, separated and widowed) and employment (none, parttime and fulltime). However, none of these measures improved model fit or explained any significant variance in leadership effort. Thus, they were not included in the final models.

Control Measures

Because our data is cross-sectional we cannot statistically establish the order of effects we have posited from structural and individual characteristics to weekly effort, and because two-thirds of the presidents are founders, there may be a legitimate question about whether founders are responsible for determining both the scope of a group’s operation (structure) as well as the number of hours leaders work per week. If this is the case, the associations between organizational structure and the dependent measure may be spurious. Consistent with this concern, we include in each model a control measure establishing whether the president was one of the founders of his or her local group (founder = 1, non-founder = 0). Supplemental analyses indicate the path between structural characteristics and leadership effort is not conditioned by founding status. While founders work more hours on average (15.4) than non-founders (13.6), these differences are not statistically significant in bi-variate models, conditional models that include all structural and individual level characteristics, or in two-group mean-comparison t-tests Furthermore, founding status is not a significant predictor of either bureaucratic complexity or victim services in bi-variate models and t-tests, nor does it moderate the relationship between the structural characteristics and weekly effort. Finally, given the mean differences between the MADD and RID leaders on weekly effort (differences in reported hours were significant in t-tests at the p < .001 level), we have controlled for group membership as a binary variable which indicates whether the president was a leader of a MADD chapter (0 = RID, 1 = MADD).

A second control is the concurrent level of involvement presidents reported with other community groups in addition to their chapter (e.g., participation in church, civic, fraternal, political and social groups). This is included in our models because involvement in other areas may constrain the amount of time available for leadership activities. Other activism is assessed here as a series of dummy variables (1 = condition applies, 0 = condition does not apply), including not at all active, not very active, somewhat active and very active in other community groups. The level of reported activity was quite high among chapter presidents with 43 percent of the sample reporting being very active in other groups.

Gender Interactions

We created six gender interaction terms to evaluate the possibility that gender may interact with structural and individual characteristics in predicting weekly effort. A gender value of 1 for women was multiplied by each of the following variable values to create gender interaction terms in the final models: bureaucratic complexity, victim support services, victim status, employment, marital status and number of children.


Table 3 displays the results from a series of hierarchical OLS regression analyses designed to answer the central question posed in this research, Why do some activist leaders work harder than others? In a stepwise fashion we evaluate the unique influence of structural characteristics (Model 1) and individual attributes (Model 2) on leadership effort within the movement opposing drinking and driving during its emergent phase in 1985. Then, we combine the elements of the two models to compare their relative influence (Model 3). Finally, we consider gender as a moderator of the link between measures of structural characteristics and individual attributes with weekly effort (Model 4). The non-italicized coefficients in Model 4 are the reported coefficients for the full model, which includes the gender*bureaucratic complexity interaction. The italicized coefficients are bi-variate associations between the specified interaction terms and the dependent variable, controlling for all other relevant structural and individual characteristics included in the full model. These italicized coefficients are presented in their bi-variate form because there is no significant interaction of gender with the effects of the independent variable on weekly effort.

Table 3
Hierarchical OLS Models of Organizational Structure, Individual Characteristics and Gender on Presidents' Weekly Effort in the Movement against Drinking and Driving in 1985

Because the metric for the dependent variable in all models is the number of hours worked per week by activist leaders, the intercept reflects the average effort, in hours, after controlling for each of the indicators in the model. In Model 1, for example, leaders work an average of 9.3 hours per week after accounting for the structural conditions of the chapter and the key controls used in these models, including: founding status, activity in other groups and affiliation with MADD. (As noted on Table 3, founding status, activity in other groups and affiliation with MADD are included in each model although the coefficients are not reported). As expected, both increases in bureaucratic complexity and victim support are associated with higher weekly effort. Specifically, a one-standard-deviation-unit increase in bureaucratic complexity is associated with a little more than a five-hour-a-week increase in volunteer effort, net of the influence of victim support and relevant controls. This finding is consistent with our first hypothesis that chapters with greater complexity tend to have presidents who work more hours per week. There is a similar positive and significant association between victim support services and presidents’ effort. For every one-standard-deviation-unit increase in victim support services there is an associated four-hour increase in weekly leadership effort, after controlling for bureaucratic complexity, founding status, activity in other groups and affiliation with MADD. The adjusted R squared for Model 1 indicates that the structural measures explain nearly 34 percent of the total variation in weekly leadership effort. And, if a leader presides over a group that is both more complex (+1sd) and provides more extensive victim services (+1sd) than average, we would expect them to volunteer almost 25 hours per week, which is equivalent to holding a part-time job. Comparing the t-ratios and standardized betas from Model 1 shows that victim support services and bureaucratic complexity are roughly equal in importance when trying to understand a president’s weekly labor, and both are about three times more influential than founding status, presidents other activism and MADD affiliation in predicting weekly effort. (All supplemental analyses, including standardized betas, are available by request.)

Model 2 assesses the influence of individual characteristics on weekly leadership effort and explains about 20 percent of the variation in the dependent measure. A comparison of the standardized beta coefficients indicates that victim status and presidents’ employment are the most influential individual characteristics predicting weekly effort followed by president’s gender and marital status. Overall, most of the individual-level coefficients are in the direction and magnitude anticipated from our theoretical discussions and previous research. For example, our second hypothesis predicted that personal grievance would be strongly and positively related to weekly effort; our findings support this notion with victims working, on average more than five additional hours per week than non-victims, even after controlling for all other leader characteristics. Also conforming to expectations are the findings related to our third hypothesis – when employment rime obligations are lower, presidents donate more hours to leading their chapters. In fact, chapter leaders who do not work for pay outside the home averaged an additional 5.3 hours of leadership effort each week compared to full- and part-time workers. Taken together, these results indicate that employed leaders spend about 46 percent fewer hours per week leading their local chapters than their counterparts.

As anticipated, many components of the biographical availability of the leader significantly predicted weekly work hours. Along with employment, marital status had a significant and negative association with work hours, such that married presidents worked about three hours less per week than non-married counterparts. These findings uphold the notion that people who work outside the home or are married will volunteer less time each week because they choose not to substitute finite work and family hours for volunteer hours. In contrast to these results, however, there was an unexpected finding regarding the number of children in the home; rather than having a significant negative association with work hours as suggested by Hypothesis 3, the presence of children in the home has no discernable impact on presidents’ effort, making this finding inconsistent with the substitutive approach. This may be due to countervailing influences of a family-friendly organizational form (which may be associated with increasing hours) competing with a rational allocation of rime between competing demands (which may be associated with decreasing hours). While we expected a significant negative association, the findings are not without precedent, given that most prior work on selection into membership and activist effort among the rank-and-file has found few family-level effects.

Consistent with our expectations regarding gender (Hypothesis 4), we find that women work an average of nearly 3½ hours more per week than their male counterparts, holding all other individual attributes constant. Given that the majority of the local groups are female dominated and mother-oriented, it may be that women are simply more comfortable working within this organizational structure. Nevertheless, this finding coincides with the widespread belief among many activist women that they do the bulk of the work in their organizations. This is in contrast with the few studies we cited earlier that fail to show any gender effect on the amount of effort rank-and-file activists report devoting to movement activities.

Model 3 compares the relative influence of structural characteristics and individual attributes on presidents’ weekly labor. Given that the variance explained by the structural-level model – in terms of the adjusted R2 – was nearly twice as large as that explained by the individual-level model, it is not surprising that the t-ratios and standardized beta coefficients for the combined analyses indicate that bureaucratic complexity and victim support services have the greatest impact on weekly effort, followed closely by presidential employment status and then trailed by grievance, gender and marital status. In the full model the number of children in the home remained an insignificant predictor of weekly effort white founding status, president’s other organizational activity, and MADD affiliation all significantly predicted the number of hours worked per week. Taken together, the structural and individual characteristics do an impressive job of explaining leadership effort for these chapter presidents. In fact, the adjusted R2 for Model 3 indicates that a sizeable 43 percent of the variation in leaders’ weekly labor can be explained by considering the simultaneous influence of structural and individual characteristics, a notable improvement over either model run separately.

The final model in Table 3 illustrates the importance of gender as both a direct and moderating influence on leadership effort.7 To test our fifth hypothesis we assessed the influence of gender interactions for each of our independent variables on weekly effort and found only one of them to be significant in the full model, the one between bureaucratic complexity and gender. The findings demonstrate that the association between bureaucratic complexity and work hours is almost wholly conditioned by the gender of the leader. That is, when bureaucratic complexity increases by one standard deviation for male presidents there is almost no increase in leadership effort. Conversely, when bureaucratic complexity increases by one standard deviation unit for female presidents there is a corresponding increase in weekly effort of about six hours. The difference in slopes between men and women is graphically displayed in Figure 1. When bureaucratic complexity is at its mean, women volunteer about three hours a week more than men, controlling for all other variables in the model. When bureaucratic complexity is low (around −l), men and women have nearly identical rates of weekly effort, just below 10 hours a week, controlling for all other variables. Most interestingly, when bureaucratic complexity is high (around +1), women work nearly 10 hours more than men each week. The comparison of men and women’s slopes in this interaction suggests that rising bureaucratic complexity has an exponential effect on increasing women’s work hours while only slightly bumping up men’s leadership effort

Figure 1
Presidents’ Weekly Effort and the Bureaucratic Complexity * Gender Interaction

Summary and Discussion

We were motivated in these analyses by one major question: why do some social movement leaders work harder than others? Through the examination of five hypotheses we have attempted to identify a number of theoretically plausible explanations for understanding variations in leadership effort and to explore our second question, the distinctive role of gender in complicating the answer to the central question. Our analyses have produced what we judge to be quite dramatic findings – explaining more than 45 percent of the total variation in reported leadership effort.

As anticipated by our hypotheses, we found that both structural and individual characteristics are important in understanding variations in leadership effort, although structural factors do a better job of explaining why some leaders work harder than others. In particular, bureaucratic complexity and victim support services (Hypothesis 1) were positively related to presidents’ weekly effort, and in combination these two scales explained nearly 34 percent of the total variation in work hours. When the local chapters were complex (e.g., focused on recruitment, managed many committees, had regular meetings and engaged with other community groups) the president worked more hours, on average, than when the chapter had fewer organizational demands. Likewise, groups that took on more labor intensive victim support services – provided victim assistance, managed victim case loads, and developed support groups – often had leaders who worked more hours than presidents from chapters that provided fewer services. Overall, these findings suggest that structural pressures appear to enlarge the amount of ongoing effort leaders are expected to contribute in order to successfully manage their groups. In fact, we believe that these structural pushes toward effort may act as the functional equivalents of being asked to participate, a significant predictor of basic participation among rank-and-file members.

In terms of individual-level accounts of participation, both cognitive availability, assessed as grievance stemming from victim status (Hypothesis 2), and biographical availability, measured as employment and marital status (Hypothesis 3) were significant predictors of president’s work hours, and explained roughly 20 percent of the total variation in weekly effort. Surprisingly, the number of children in the home was not significantly associated with effort. Of particular interest to us were the robust findings related to victim status. Although prior research on rank-and-file membership has demonstrated that grievance does not often do a good job explaining who joins social movement organizations (likely because the pool of potential recruits is so large), among the presidents in our sample who had already selected themselves into leadership positions, victim status was associated with three to five additional hours of work per week. In this context, objective grievance appears to provide an ongoing internal motivation to a greater commitment to working toward organizational goals.

Our final two hypotheses regarding gender (H4–5) were also confirmed, to some extent, with these data. In direct effect models we found that female leaders worked an average of 2½ to 3½ hours more per week than their male counterparts, net of other factors. Furthermore, in interactive models we see that gender almost wholly conditioned the effects of bureaucratic complexity on leadership effort, indicating that women are more likely to increase their hours per week as complexity increases while men show only modest increases in work hours under the same conditions. We anticipated this association based on findings from other literatures which suggest that women leaders tend to adopt democratic organizational structures–compared to men’s preference for more autocratic systems of governance – which are likely to require more day-to-day oversight and maintenance. In this way, gender maybe operating within the social movement organization similar to the structural variables we have analyzed – as a functional equivalent of ongoing requests for participation to rank and file sympathizers. To the extent that women work within organizational forms that require more interactional oversight, it is not surprising they tend to work more hours than men. However, this may not be the only explanation for these findings because the data do not allow a more nuanced glimpse of the details of leadership style. Additional research is needed to understand the complicated relationship between gender, organizational complexity and leadership effort, and we hope future studies will explore these issues in a wide range of social movement organizational and voluntary associational settings.

These findings are compelling, but it is unclear whether the strong differences we uncovered between male and female leaders (that the women work more hours and are more heavily influenced by context) would hold up had we surveyed leaders of a predominantly male-led movement. There is, however, a good reason why we might expect these findings to be more widely reflective of gender processes in SMOs as well as voluntary associations more generally. Specifically, systematic evidence indicates that female integrated or dominated groups are the norm among U.S. voluntary associations, with the majority of voluntary associations being either gender integrated such as those of the anti-drunk driving movement (36%) or female dominated, with at least 90 percent of their membership being women (43% of all groups) (Popielarz 1999). Consequently, even if our findings are only applicable to the most extreme cases (e.g., female-dominated groups), this probably includes a significant proportion of all social movement organizations and local voluntary associations. Further, parallel evidence from a recent survey of leaders of local Sierra Club groups around the United States also found female leaders report volunteering more hours a month than male leaders (Baggetta, Han and Andrews 2010) in a movement with more gender-integrated membership and leadership. As a result we have cautious confidence in the broader generality of our findings.


The data used in these analyses were collected by McCarthy with support from the National Sciences Foundation (“The Causes and Consequences of the Citizen’s Movement Against Drunk Driving” grant SES-8419767). Dorius was supported by a National Institute for Child Health and Human Development Training Grant from the Population Research Institute at Pennsylvania State University (T32.HD007514) and the Population Studies Center at the University of Michigan (T32.HD007339).


An earlier version of this article was presented at the 2006 annual meeting of the American Sociological Association in Chicago, IL.

1Although the two organizations shared common goals, they displayed important differences. In particular, MADD chapters were slightly larger and tended to have greater financial resources compared to RID chapters, and RID leaders worked only about 57% of weekly hours reported by MADD leaders. See Table 1 for more details.


3This distinction is also widely articulated as supply and demand accounts of social and activist participation for religious groups (e.g., Finke, Guest and Stark 1996) and social movement mobilization (e.g., Klandermans 2004).

4This is likely a result of the basic fact that very few potential activists ever actually participate in a social movement while the pool of those predisposed to do so by their personal characteristics is typically quite large. For example, Kilpatrick and colleagues (1990) estimated that in the late 1980s, 1.6% of the adult U.S. population had been a victim of drunken driving and that another 2.8% reported having a friend die in a drunken driving incident. Thus the pool of potential recruits for the movement was enormous while the actual number of activists was rather small.

5Increasing organizational complexity and greater diversity of activities in the trajectory of an SMO probably leads to higher levels of professionalization and formalization in the long run (McCarthy and Zald 1977; Staggenborg 1988), but few of the local groups we analyze here were led by professional or paid functionaries when we surveyed them. And, most were governed by similar organizational rules and regulations provided by their umbrella national organizations. Our concept of organizational vitality (composed of bureaucratic complexity and extent of victim support services) deployed here aims to capture merely the volume of demand on a leader to oversee and coordinate the activities of her/his chapter.

6Given the modest, though acceptable, level of reliability for this scale, the four items were also evaluated individually in the final models. The results were robust whether the constructed scale or individual items were included, with no changes in the pattern of association, size of coefficients or level of significance. However, the model fit was poorer, and the comparison with victim services rendered more cumbersome when separate measures were used, thus we retained the standardized scale.

7Given the small number of men in our sample, we performed statistical tests to confirm adequate sample size for multivariate models, and sensitivity analysis to verify gender results were not an artifact of the sample. First, DF beta tests were run to measure the influence of each observation on the entire model before and after excluding each case. After removing outliers, the models were rerun. The results were robust and suggest findings are not due to outliers. A second DF beta was run on the interaction term itself. As before, the constrained model was nearly identical to the original, suggesting the interaction terms were not overly influenced by outliers. Next, robust regression analyses were conducted to verify our belief that outliers were not influencing the coefficients themselves. Taken together, these three tests provided support for the conclusion that the data were adequate for the models assessed here.

Contributor Information

Cassandra R. Dorius, University of Michigan.

John D. McCarthy, Pennsylvania State University.


  • Adams Jacqueline. Gender and Social Movement Decline – Shantytown Women and the Prodemocracy Movement in Pinochet’s Chile. Journal of Contemporary Ethnography. 2002;31(3):285–322.
  • Baggetta Matthew, Han Henrie, Andrews Kenneth T. Harvard University Department of Sociology; 2010. Explaining Volunteer Leader Time Commitment: Civic Resources, Personal Motivations, and Organizational Characteristics. Working paper.
  • Barkan Steven E, Cohn Steven F, Whitaker William H. Beyond Recruitment: Predictors of Differential Participation in a National Antihunger Organization. Sociological Forum. 1995;10(1):113–134.
  • Bianchi Suzanne M, Robinson John P, Milkie Melissa A. Changing Rhythms of American Family Life. Russell Sage Foundation; 2006.
  • Becker Gary S. A Treatise on the Family. Harvard University Press; 1981.
  • Cohn Steven F, Barkan Steven E, Halteman William A. Dimensions of Participation in a Professional Social-Movement Organization. Sociological Inquiry. 2003;73(3):311–337.
  • Eagly Alice H, Johnson Blair T. Gender and Leadership Style: A Meta-Analysis. Psychological Bulletin. 1990;108(2):233–256.
  • Einwohner Rachel L. Gender, Class, and Social Movement Outcomes - Identity and Effectiveness in Two Animal Rights Campaigns. Gender and Society. 1999;13(1):56–76.
  • Einwohner Rachel L, Hollander Jocelyn A, Olson Toska. Engendering Social Movements: Cultural Images and Movement Dynamics. Gender and Society. 2000;14(5):679–699.
  • Finke Roger, Guest Avery, Stark Rodney. Mobilizing Religious Markets: Religious Pluralism in the Empire State, 1865. American Sociological Review. 1996;61(2):203–218.
  • Ganz Marshall. Why David Sometimes Wins: Leadership, Organization, and Strategy in the California Farm Worker Movement. Oxford University Press; 2009.
  • Hook Jennifer L. Reconsidering the Division of Household Labor: Incorporating Volunteer Work and Informal Support. Journal of Marriage and Family. 2004;66(1):101–117.
  • Independent Sector Report. Giving and Volunteering in the United States; Findings from a National Survey. Washington, DC: Independent Sector; 1986.
  • Jennings M Kent. Political Response to Pain and Loss: Presidential Address, American Political Science Association, 1998. American Political Science Review. 1999;93(1):1–13.
  • Kilpatrick Dean G, Amick Angelynne, Resnick Heidi S. The Impact of Homicide on Surviving Family Members. Charleston, SC: Crime Victims Research and Treatment Center, Medical University of South Carolina; 1990.
  • Klandermans Bert. The Social Psychology of Protest. Blackwell Publishing; 1997.
  • Klandermans Bert. The Supply and Demand of Participation: Social Psychological Correlates of Participation in Social Movements. In: Snow David A, Soule Sarah A, Kriesi Hanspeter., editors. The Blackwell Companion to Social Movements. Blackwell; 2004. pp. 360–379.
  • Lofland John. Social Movement Organizations: Guide to Research on Insurgent Realities. 1996 Transaction Publishers.
  • McAdam Doug. Recruitment to High-Risk Activism: The Case of Freedom Summer. American Journal of Sociology. 1986;92(1):64–90.
  • McAdam Doug. Gender as a Mediator of the Activist Experience: The Case of Freedom Summer. The American Journal of Sociology. 1992;97(5):1211–1240.
  • McCarthy John D. The Interaction of Grass-roots Activists and State Actors in the Production of an Anti-Drunken Driving Media Attention Cycle. In: Larana Enrique, Johnston Hank, Gusfield Joseph., editors. New Social Movements: From Ideology to Identity. Temple University Press; 1994. pp. 133–167.
  • McCarthy John D, Wolfson Mark. Resource Mobilization by Local Social Movement Organizations: The Role of Agency, Strategy and Structure. American Sociological Review. 1996;61(6):1070–1088.
  • McCarthy John D, Wolfson Mark, Harvey Debra S. Chapter Survey Report. Washington, DC: Life Cycle Institute, Catholic University of America; 1987.
  • McCarthy John D, Zald Mayer N. Resource Mobilization and Social Movements: A Partial Theory. The American Journal of Sociology. 1977;82(6):1212–1241.
  • Morris Aldon D, Staggenborg Suzanne. Leadership in Social Movements. In: Snow David A, Soule Sarah A, Kriesi Hanspeter., editors. The Blackwell Companion to Social Movements. Blackwell; 2004. pp. 171–196.
  • Musick Marc A, Wilson John. Volunteers: A Social Profile. Indiana University Press; 2008.
  • Oliver Pamela E, Marwell Gerald. Mobilizing Technologies for Collective Action. In: Morris Aldon D, Mueller Carol McClurg., editors. Frontiers in Social Movement Theory. Yale University Press; 1992. pp. 251–272.
  • Passy Florence, Giugni Marco. Social Networks and Individual Perceptions: Explaining Differential Participation in Social Movements. Sociological Forum. 2001;16(1):123–153.
  • Popielarz Pamela A. (In)Voluntary Association: A Multilevel Analyses of Gender Segregation in Voluntary Organizations. Gender and Society. 1999;13(2):234–250.
  • Rock Paul. Governments, Victims and Policies in Two Countries. British Journal of Criminology. 1988a;28(1):44–66.
  • Rock Paul. On the Birth of Organizations. Canadian Journal of Sociology. 1988b;13(4):359–384.
  • Schussman Alan, Soule Sarah A. Process and Protest: Accounting for Individual Protest Participation. Social Forces. 2005;84(2):1083–1108.
  • Staggenborg Suzanne. The Consequences of Professionalization and Formalization in the Pro-Choice Movement. American Sociological Review. 1988;53(4):585–605.
  • Taylor Verta. Gender and Social Movements: Gender Processes in Women’s Self-Help Movements. Gender and Society. 1999;13(1):8–33.
  • Weed Frank J. The Victim-Activist Role in the Anti-Drunk Driving Movement. Sociological Quarterly. 1990;31(3):459–473.
  • Weed Frank J. Certainty of Justice: Reform in the Crime Victim Movement. 1995 Transactions Publishers.
  • West Guida, Blumberg Rhoda Lois. Women and Social Protest. Oxford University Press; 1990.
  • Wiltfang Gregory L, McAdam Doug. The Costs and Risks of Social Activism: A Study of Sanctuary Movement Activism. Social Forces. 1991;69(4):987–1010.
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...


  • PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...