draft

158 Public Administration Review • March/April 2005, Vol. 65, No. 2

Judith R. Saidel Karyn Loscocco University at Albany, State University of New York

Agency Leaders, Gendered Institutions, and Representative Bureaucracy

This study examines three central questions: Do women state agency heads establish priorities that advance women’s interests more frequently than men agency heads? Among state agency heads with women-related top priorities, are there systematic differences between women and men in the influences on their priority choices? Do the organizational and political contexts in which agency leaders work explain variation in policy priorities? Analysis of data from a national survey of women and men department heads reveals that working in a redistributive agency affects whether a leader pursues a women-centered policy agenda, regardless of the leader’s gender, other per- sonal characteristics, or reported influences on priority choice. The authors conclude that the way representative bureaucracy actually plays out can be more fully understood if the tenets of social science theory on gendered institutions are incorporated into analyses of how representative bu- reaucracy works.

Scholars have wrestled with the theoretical complexi- ties of representative bureaucracy for decades. In the last 35 years, a number of public administration and political science researchers in the United States have invested sig- nificant effort in this enterprise (Mosher 1968; Krislov 1974; Meier and Nigro 1976; Thompson 1976; Rosen- bloom and Featherstonhaugh 1977; Saltzstein 1979; Reh- fuss 1986; Meier 1993; Riccucci and Saidel 1997).

One reason for the sustained interest in representative bureaucracy is that this notion is fundamental to prevail- ing ideas about how bureaucracy fits into a democratic polity (Kelly 1998; Naff and Crum 2000). Hale and Kelly articulate the core idea clearly: “[There is] widespread ac- ceptance of the notions that in a democracy public bureau- cracies ought to be representative in meaningful ways of the citizen-clients they serve, and that representation based on demographic characteristics can lead to meaningful rep- resentation” (1989, 9). Representative bureaucracy has become a “powerful symbol” (Meier and Nigro 1976, 467) of equal access, opportunity, and the inclusion of group interests in administrative decision making, despite some findings that challenge the central ideas of the construct (Meier and Nigro 1976; Rehfuss 1986).

A second reason for continuing research on representa- tive bureaucracy is the construct’s rich theoretical and nor-

mative content, which invites reexamination by investiga- tors with widely varying research interests. Like a com- pelling book or a work of art to which one returns again and again, the multidimensional set of ideas embedded in representative bureaucracy provokes continuous reengage- ment by researchers with new combinations of questions and approaches.

The study we describe here emerges from our interest in the intersection of gender, institutions, and represen- tative bureaucracy. Following Rosenthal (1997) and Kenney (1996), who exhort researchers to draw on inter- disciplinary scholarship in studies of political institutions, we draw on the research of sociologists, organizational theorists, and political scientists to inform our inquiry (Acker 1990; Martin 1992; Calas and Smircich 1992;

Judith R. Saidel is executive director of the Center for Women in Govern- ment and Civil Society and an associate professor of public administration and policy at the University at Albany, State University of New York. Her research interests include the demographics of executive branch leadership and representative bureaucracy, government and nonprofit sector relation- ships, and nonprofit governance. E-mail: [email protected].

Karyn Loscocco is an associate professor of sociology at the University at Albany, State University of New York. She is interested in how gender struc- tures social life. Much of her previous research focuses on how gender colors the meanings that people attach to paid work and the fit between work and family lives. She recently completed a study of women and men small busi- ness owners. E-mail: [email protected].

Agency Leaders, Gendered Institutions, and Representative Bureaucracy 159

Lorber 1994; Duerst-Lahti and Kelly 1995; Kenney 1996; Rosenthal 1997).

Introduction Scholarship on gender and institutions prompts us to

analyze gender not only as an individual variable, as tradi- tional research on representative bureaucracy has done, but also as an organizing principle of social institutions (Acker 1990; Lorber 1994). In fact, Glenn (1999) suggests the feminist emphasis on gender as a social construction leads us away from considering gender as solely—or even mostly—an individual attribute. Gender is embedded in jobs, departments, divisions, and organizational activity (Martin 1992). Thus, gender may be important to repre- sentative bureaucracy because it shapes the social and cul- tural contexts in which people do their work. Recent rep- resentative bureaucracy studies have incorporated such insights, leading to new theoretical emphases. For instance, Keiser et al. build a “neoinstitutional theory of representa- tive bureaucracy and gender” (2002, 557) that specifies links among gender, the institutional contexts in which bureaucrats function, and organizational outputs. They ar- gue that attention to gendered institutional contexts enables researchers to go beyond the question of whether gender as an ascribed characteristic matters in bureaucracies, to a deeper inquiry about the circumstances under which bu- reaucrats with certain demographic characteristics are more likely to act in ways that advance the interests of individu- als with the same characteristics. Kelly and Newman’s find- ings (2001), which support a similar conclusion, will be examined in greater detail in this article.

We begin our inquiry within a traditional framework that includes gender as an individual attribute, focused on the priorities of women and men department heads1 ap- pointed by governors in the 50 states, and ask whether gen- der, as a demographic characteristic of agency leaders, explains variation in their policy priorities. Specifically, we ask, do women state agency heads establish priorities that could advance women’s interests more frequently than men agency heads? Second, among state agency heads with women-related top priorities, are there systematic differ- ences between women and men in the influences on prior- ity choice?

As a preview to the extended analysis that follows, we offer this brief summary of our preliminary findings: Whereas 50 percent of women department heads indicated that one or more of the policy priorities at the top of their agendas was intended to help women, only 35 percent of men department heads did so (p < .05). At the same time, the strongest influence on priority choice for both women and men was identical—their own personal values, beliefs, and experiences.

Next, we extend the inquiry to the contexts of agency leader agenda setting and ask, do the organizational and political contexts in which agency leaders work explain variation in their policy priorities? We focus particularly on the state political culture, the political party of the ap- pointing governor, and agency type. Our finding? Work- ing in a redistributive agency affects whether a leader pur- sues a women-centered policy agenda, no matter what the leader’s gender, other personal characteristics, or reported influences. We conclude that the ways in which represen- tative bureaucracy actually plays out can be more fully understood if important tenets from recent social science theory on gendered institutions are incorporated into analy- ses of how representative bureaucracy works.

The article begins with an overview of selected studies, after which we explain the conceptual framework that struc- tures the study. In the research methodology section, we describe an exploratory pilot study conducted through on- site interviews in eight states, as well as a subsequent na- tional survey of all women department heads and a strati- fied matched sample of men department heads. The findings section presents our analysis of data on policy priorities, environmental influences on agenda setting, and the effects of contextual variables. The discussion and con- clusion draw out the implications of the findings, espe- cially with respect to representative bureaucracy, the link between representation and responsiveness, and the im- portance of institutional contexts characterized by gendered structures and processes. We end with suggestions for key issues in need of additional research attention.

Overview of Selected Studies Historically, representative bureaucracy researchers

have limited their attention to career bureaucrats, at least partly because of their interest in influences that might mitigate the rigidities of organizations in the hands of Weberian “cogs” in bureaucratic machines.2 Numerous studies have focused on “street-level bureaucrats” (Lipsky 1980), who often exercise significant discretion at the point of direct service delivery. Other studies examine the rep- resentativeness of mid-level managers and supervisors in public agencies.3

We argue here that, given the positional power of those at the head of complex or smaller but still hierarchically structured public agencies, it is especially important that appointed leaders be part of a representativeness calculus. Although it is certainly true that political appointees func- tion within a very different set of dynamics, which . Aberbach and Rockman (2000) describe as a “web of poli- tics,” that web nonetheless includes the competing pres- sures of bureaucratic processes. In an examination of the meanings of representative bureaucracy, Aberbach and

160 Public Administration Review • March/April 2005, Vol. 65, No. 2

Rockman challenge the accuracy of conclusions about the relative representativeness of different bureaucracies that do not take into account top-level decision makers:

In short, the most representative bureaucracies in the modern world—representative when one looks at a profile of all their employees—have a very dif- ferent appearance at the top. One can debate the relative impact of this difference, but at a minimum it does cast a pall over notions underlying ideas about the genuine representativeness of what may appear on the surface to be a fairly representative bureaucracy. (50)

Thus, the representative-bureaucracy construct should not be limited only to those within the bureaucratic arena by virtue of their career civil service status (Dometrius 1986; Riccucci and Saidel 1997). Explanations of both the sub- stantive and symbolic importance of representativeness do not lose meaning when applied to individuals working as chief executives of state government agencies. In fact, in the current climate of devolved responsibility for major re- source allocation and programmatic decisions, the repre- sentativeness of those at the top of public bureaucracies at the state government level assumes even more importance.

The differing impact of women and men political and managerial leaders on outcomes that could benefit women has been the subject of important research for many years (Carroll 1985, 1994; Hale and Kelly 1989; Saint-Germain 1989; Dodson and Carroll 1991; Tolleson-Rinehart 1991; Duerst-Lahti and Johnson 1992; Stivers 1993; Thomas 1994; Tamerius 1995). Dodson and Carroll’s (1991) study of the top priorities of women and men state legislators is especially relevant to our research questions because of its focus on individual agenda setting. They found the top leg- islative agenda items of women state lawmakers focused on women’s rights issues more frequently than did the pri- orities of men legislators. A large gender gap also existed with respect to the bills that each group of officials had worked on during the current legislative session.

A similar study by Saint-Germain (1989) focused on the initiation of policy proposals, a step in the policy pro- cess that is very close to priority setting. In a longitudinal study of bills in the Arizona state legislature between 1969 and 1986, Saint-Germain found that women legislators were more likely than men to initiate bills on issues that are traditionally associated with women’s concerns and promoting women’s equality. Reviewing the literature sev- eral years later, Carroll concludes, “Clearly, one of the most striking ways in which women officeholders make a dif- ference is through the attention they devote to women’s rights issues” (1994, 14).

By applying questions about systematic differences in policy priorities by gender to the leadership cohort of ap-

pointed state government department heads, we grapple again with the notion of active representative bureaucracy. Passive representative bureaucracy is the condition of con- gruence between the demographic characteristics of the bu- reaucracy and the general population and, therefore, the pre- sumed congruence between bureaucratic decisions and the policy preferences of the general population. Active repre- sentative bureaucracy suggests that government officials will act in ways that benefit those in the general public who share their gender or race group membership. Conventional wisdom and the research literature hold that, in general, political appointees bow to the priorities and values of the chief executive whom they serve. In a classic study of ex- ecutive branch politics, Heclo (1977) included evidence of the loyalties of political appointees to the elective official who appointed them.4 Interestingly, he also identified “di- rection-setting,” a term that overlaps with our study’s pri- ority-setting focus, as a central definitional element of po- litical leadership. Also writing about the federal level of government, Kingdon observed, “The appointee finds it prudent to bend with the presidential wind…” (1995, 29).

Past research suggests there are other important influ- ences on priority setting, including work socialization (Meier and Nigro 1976; Rehfuss 1986), personal values and expe- riences (Riccucci 1995), values and experiences rooted in professional training (Mosher 1982; Hebert and Wright 1982; Keiser et al. 2002), ties with outside interest groups (Carroll 1992), and other external pressures. Each of these influences, as well as loyalty to the chief executive, could either reinforce or mitigate active representative bureaucracy, in the sense that dominant agenda-setting influences could push the agency leader toward or away from priorities that favor groups with characteristics like his or her own.5

The research literature also provides valuable clues about contextual variables that might explain variation in the agenda-setting priorities of agency heads. Measures of dif- ference in state political cultures have figured in empirical investigations for many decades and have been reported as important predictors of variation in a range of outcomes (Elazar 1972, 1984; Erickson, Wright, and McIver 1993). Another traditional political measure of difference in policy choices is political party.

The third contextual variable in the multivariate phase of our study, agency type, derives from more recent re- search on gender and organizations. A discussion of key insights, relevant to the questions examined here, from pio- neering work on gendered institutions is included in the “Expanded Conceptual Framework” section.

Conceptual Framework Saltzstein’s theoretical model (figure 1) of the “mecha-

nisms of responsiveness” (1979, 466) guided the initial

Agency Leaders, Gendered Institutions, and Representative Bureaucracy 161

phase of our study.6 She integrates earlier studies and pos- its consecutive links between ascribed characteristics and values, values and behavior, and behavior and policy.

sion makers. For instance, Rehfuss argues, “the most im- portant decisions are made or influenced at top levels” (1986, 454). In a classic study of agenda setting and alter- native selection at the federal level, Kingdon (1995) finds that political appointees are part of the “visible cluster” that dominates the agenda-setting process. “Even when the political appointees do not originate an idea,” he reports, “they still play a large part in placing it on the agendas of important people, both within and outside of their agen- cies” (28). Writing about both political appointees and top- career executives, Aberbach and Rockman maintain, “Fi- nally, when looking at the representational qualities of a bureaucracy, it is important to distinguish the entire orga- nization from those at the top who have the major influ- ence over how policy is interpreted and implemented” (2000, 49). Along with their emphasis on positional power, Aberbach and Rockman contend that, with respect to ca- reer bureaucrats, “demographic or social class representa- tiveness is apparently a weak reed on which to hang hopes for a truly representative bureaucracy” (49).

While recognizing that players at all levels in public agencies may have an impact on policy outcomes, we ac- knowledge the dominant influence of positional leaders in hierarchically structured public agencies and, in the pre- liminary phase of this study, examine policy prioritizing by top-ranking gubernatorial political appointees. Because top decision makers likely face the fewest constraints, they are apt to show the closest correspondence among values, behavior, and policy outcomes. Thus, it is an important place to tackle the question of whether gender matters.

Expanded Conceptual Framework In the expanded conceptual framework that guided the

second phase of our study, we incorporate insights from social science theory on gendered institutions (figure 2).

By placing Saltzstein’s hypothesized links between as- cribed characteristics and policy within the context of

Figure 1 Saltzstein’s (1979) Mechanisms of Responsiveness

Ascribed characteristics → values → behavior → policy

Figure 2 Institutional Context and Representative Bureaucracy

Ascribed characteristics—values → behavior → policy

G en

de red

Institutional Context

Reinforcing the warnings of researchers such as Thomp- son (1976), Rehfuss (1986), Meier (1993), however, Saltzstein argues that each link is fraught with method- ological peril for those who attempt to operationalize parts or all of the model. Clearly, this is a heuristic device for disentangling the complex relationships that underpin rep- resentative bureaucracy theory. We recognize there are other variables that influence the links between each component of Saltzstein’s model.

The preliminary focus of our study is the link between an ascribed characteristic—gender—of executive branch state agency leaders and their top policy priorities. We use a department head’s self-identified priorities as a proxy for both the “values” and “behavior” elements of Saltzstein’s model. Findings reported later in the article substantiate the appropriateness of this proxy measure for values. For the people in our sample, there appears to be a close correspondence between values and behavior, per- haps because we are studying decision makers with some positional power at the top of their organizations. When we asked respondents how closely their policy work matches their preferred agenda, we found a high match between actual work and preferred agendas for almost three-quarters of the respondents. From the perspective of active representative bureaucracy, the particular agenda priorities we are most interested in are those that could benefit women inside and outside the agency.

As predicted by earlier representative bureaucracy re- searchers, a number of conceptual and methodological problems are immediately apparent. One of our major con- cerns is what we describe as the single-actor problem. Some authors have warned against overstating the influence of individuals in governmental bureaucracies on policy out- comes (Saltzstein 1979; Meier and Stewart 1992; Meier 1993; Moore 1995).7 We avoid this problem by focusing on the link between the individual at the top of a hierarchi- cally structured bureaucracy and that individual’s self-iden- tified policy priorities. We do not attempt to operationalize the complicated link between a department head’s policy priorities and formal agency policy. We think it is reason- able to assume that enacted agency policy is related to the agency chief executive’s priorities, but examination of this link is beyond the scope of this analysis.

A number of studies make the strong case that analyses of representative bureaucracy should target top-level deci-

162 Public Administration Review • March/April 2005, Vol. 65, No. 2

gendered institutions, we intend to suggest that studies of the way representative bureaucracy works or does not work must take into account the gendered nature of political in- stitutions such as the state agencies our respondents di- rect. We replace the arrow between ascribed characteris- tics and values with a dotted line to indicate the link between gender and values may be modified by the context in which women and men are working.

During the last two decades, sociologists and organiza- tional theorists have built a considerable body of literature about gendered institutions. According to Acker, a key theo- rist of this new paradigm, “The term ‘gendered institutions’ means that gender is present in the processes, practices, images and ideologies, and distributions of power in the various sectors of social life” (quoted in Kenney 1996, 446).

Insights that both foreshadow and contribute to the con- struct of gendered institutions are represented in classic and more recent studies of representative bureaucracy. Meier and Nigro (1976) found that agency affiliation had a stronger impact on attitudes than gender. Keiser et al. (2002) make the important point that institutions in which gender is embedded influence the context in which indi- viduals make decisions and act on those decisions. Based on a public education study that finds significant links be- tween institutional features and girls’ math scores, they argue that the compelling research task is “interpreting data in light of the larger institutional features that shape the circumstances in which bureaucrats exercise discretion and act to affect policy outcomes” (2002, 554).

Kelly and Newman (2001) also focus on institutional context and report that agency type is an institutional struc- ture in state government where gender is clearly manifested. Utilizing Lowi’s (1985) categorization of agencies as dis- tributive, redistributive, or regulatory,8 they examine data across agencies in Alabama, Arizona, and Florida on em- ployment patterns, pay inequalities, and sexual harassment. In each case, patterns of difference based on gender occur across agency types. For instance, the authors examine the proportions of women and men employed in the three kinds of agencies. Women and men held about an equal propor- tion of middle- and upper-level positions in regulatory agen- cies; in distributive agencies women held a substantially smaller proportion of such posts; and in redistributive agen- cies, the proportion of women was substantially higher than in either regulatory or distributive departments.

In addition to the analysis of employment patterns, Kelly and Newman’s examination of pay inequalities at the up- per levels of administration and women’s experience of sexual harassment in the three agency types leads them to conclude that “state bureaucracies are heavily gendered and … the policy type of the agency is an important factor in the gendering” (2001, 17–19). Based on evidence of variation in patterns of active representation of women

across agency type, the proposition they articulate is sup- ported: “Given the gendered nature of state bureaucracies across policy type, it is highly likely that more active rep- resentation of women will occur in redistributive than in either distributive or regulatory agencies” (20).

These studies recognize that passive and active repre- sentation are more likely to be linked when the context in which policy decisions are made supports that link. Un- derscoring this point, Meier (1993) suggests that organi- zations may make it difficult for policy makers to engage in active representation because of institutional policies or priorities that constrain identification with a group that shares one’s demographic characteristics. One pertinent study from the political science literature found that in some states, women legislators formed caucuses and constructed policy agendas focused on women’s interests. However, in other states women were so divided they did not form or- ganizational structures to facilitate joint work on agenda setting (Mueller 1984).

Building on the important research cited previously, the second phase of our study examines the context in which agency leaders set policy priorities. We recognize that gen- der may be embedded in the organizational contexts in which policy priorities are decided. We also acknowledge there may be constraints on the ability of agency leaders in state government to pursue an identity-based agenda, such as the appointing governor’s priorities or the prevailing political ideology of the governor’s party.

Data and Methods Exploratory Pilot Study

To lay the groundwork for a national survey of state agency leaders, we conducted a qualitative pilot study to explore whether and under what influences agency lead- ers promote internal and external policies that are benefi- cial to women’s interests.9 We drew primarily on the lit- erature on representative bureaucracy and women and politics to identify independent variables and to formulate hypotheses. Independent variables, to be described later, were at the individual, organizational, and contextual lev- els of analysis.

Between the spring and fall 1998, we pre-tested a semistructured interview protocol and then conducted in- depth, on-site interviews with a total of 24 women agency heads, including nine women of color, and 10 men agency executives in eight states.10 States were selected based on region, the presence of women in executive positions, the functional type of agency headed by women, the party of the governor, and political culture as measured by a com- posite index of state policies on a number of issues (Erickson, Wright, and McIver 1993). States in the study included California, Delaware, Georgia, Illinois, Indiana,

Agency Leaders, Gendered Institutions, and Representative Bureaucracy 163

Maryland, New York, and Vermont.11

Computer-assisted analysis of the 34 interviews from the pilot study yielded a richly textured picture of compet- ing pressures balanced by agency executives who run com- plex public bureaucracies around the country. The semistructured interview format gave respondents the chance to raise issues or suggest connections we had not anticipated. In addition, it uncovered the language and ways of thinking of the pilot-study interviewees, enabling us to build into the subsequent survey instrument the full range of choices relevant to respondents. In the spring of 2000, four former executive bureaucratic leaders who had recently moved to new career positions pre-tested the survey proto- col and made valuable suggestions about ways to clarify particular questions. These recommendations were subse- quently incorporated into the survey.

National Survey In June, we mailed surveys to all women department

heads in the country, to all men who head civil and human rights agencies, and to the same number of men and women who lead labor, human resources, health, and public wel- fare, and employment security agencies. These are the func- tional areas where women are most likely to head agen- cies. To complete the sample pool, we mailed surveys to a random sample of men, equal to the total number of women in agencies other than those specified previously. We used a stratified matched methodology for agency type to con- trol for gender bias in the distribution of women and men policy leaders across government functions (Saidel 1998).

To maximize the response rate, we followed up with a postcard to all nonrespondents about three weeks after the first mailing. In early August, another letter with a second survey enclosed was sent to all nonrespondents. In mid- September we contacted by telephone all women nonre- spondents and a random sample of men who had not re- sponded. These strategies yielded 215 usable surveys (129 returned by women, 86 returned by men) for a response rate of 42 percent for women and 28 percent for men.12 In 1999, 25.8 percent of the total number of department heads (n = 1,244) in the 50 states were women and 74.1 percent were men (Saidel 1999). Although women held about one- quarter of all agency head positions in 1999, they consti- tute two-thirds of our sample. In the multivariate results section, we explain how we corrected for the skewed pool of respondents in data analysis.

Variables Based on analysis of the pilot interview data, we identi-

fied 10 major influences on agenda setting among agency leaders. The appendix lists the 10 individual, organizational, and environmental influences. Individual-level variables may apply to the department head or to the governor who

appointed the executive; organizational-level variables apply to the executive’s agency; and environmental-level variables relate to sources of external pressure, including the legislature, interest groups, media, and statutory or regu- latory requirements.

In the survey, we first asked department heads the fol- lowing question: “Briefly describe one of your major policy priorities. Please provide a specific rather than a general priority.” We then asked respondents to indicate on a scale of 1 to 5 (1 = not at all important, 3 = somewhat important, and 5 = very important) the importance of 10 different in- fluences on the choice of the policy listed. The next ques- tion asked agency heads which influence had the strongest effect on his or her choice of policy priority. This sequence of questions was repeated for three different priorities, fol- lowed by the questions, “Among the priorities listed above, is one or more specifically intended to help white women or women and men of color? If yes, which one(s)?”

Consistent with the cautionary note of Meier (1993) and others about the difficulty of achieving conceptual preci- sion in empirical research on representative bureaucracy, we note some intended ambiguity in our model’s specifi- cation. We operationalized our dependent variable—policy priorities that could benefit women—in the phrase “spe- cifically intended to help white women or women and men of color.” Some priorities might benefit women of color (and men) but would not appear to benefit all women. However, we felt that respondents, especially women of color, should not have to make an artificial distinction be- tween gender-related and race- or ethnicity-related choices (Crenshaw 1993). It is entirely reasonable for women of color to identify policies that benefit people of color and to view those as women-related policies. Still, a conse- quence of this wording is that some priorities reported by respondents may reflect an intention to benefit people of color in general more than women in particular.

With respect to the model’s independent variables—in- fluences on priority choices—it is likely that, because of appointment screening processes, most appointees’ per- sonal values overlap with the governor’s values and policy preferences. As a result, the distinction between the strength of influence of these two independent variables may be difficult to determine. Similarly, it is possible that some respondents may include workplace socialization experi- ences in their interpretation of “your own personal values, beliefs, and experiences,” although we tried to head this off with brief examples excerpted from the exploratory interviews and included as examples under the response categories.13

Sample Characteristics Table 1 provides descriptive information about the de-

mographic characteristics, educational background, and

164 Public Administration Review • March/April 2005, Vol. 65, No. 2

regional distribution of agency leaders who completed us- able surveys.

Of the 129 women, 16 individuals (12 percent) were women of color; of the 86 men, 8 individuals (9 percent) were men of color. Of the 215 respondents, about 23 per- cent hold law degrees, 33 percent hold master’s degrees, 8 percent hold doctorates, and 25 percent reported a bach- elors as their highest degree. Outside of law, business and public administration were the two fields of degree study cited most frequently. The distribution of respondents across geographic regions ranges from 6.5 percent from the Rocky Mountain region (the only share under 10 per- cent) to 16.7 percent from the Southeast.

A comparison between the sample of respondents and the total population of agency heads with respect to cat- egories of agency function revealed that, in 10 of the 12 functional categories, the variation between percentages of agency heads in the sample and in the population did not exceed 3.7 percentage points. A weighting procedure used in the multivariate analysis, discussed later, is designed to correct for the mismatch between sample and popula- tion executives in the two functional agency categories that exceeded 3.7 percentage points.

Results Women-Related Priorities

In analyzing the data, the research team dis- covered an interesting anomaly. Of the 215 total respondents, 73 reported that one or more of their priorities were specifically intended to advance women’s interests. However, when the three researchers independently coded the priorities according to whether, in their judg- ment, the policies described by respondents would benefit women, the researchers agreed, after comparing coding decisions and recon- ciling a few differences between them, that the priorities listed by 94 respondents could ac- curately be described as benefiting women.

We suspect the anomaly can be explained by what we call a “claiming and framing” is- sue. In the political climate of the spring of 2000, when anti–affirmative action sentiment was strong, executive branch political appoin- tees in state government may have hesitated to claim that a policy they supported was in- tended to accomplish the ends framed in the question, that is, to benefit a particular group. In our subsequent analysis, we used the recoded variable as the basis for statistical

comparison involving women-related priorities. Some examples of policies either reported or coded as

advancing the interests of women are listed below, using the words of the respondents. In this article, we use the term “women-related policies” to refer to this group of pri- orities. • Advocacy for pay equity, nontraditional jobs for women,

and higher pay for traditionally female jobs • Enhancing diversity of the workforce • Fighting domestic abuse • Helping women attain greater economic self-sufficiency • Seeing more women and minority local firms do business

with us • Providing after-school care for middle-school children • Developing a project that addresses the issue of obesity

among state residents, specifically among women and children. A cross-tabulation between gender and policy priority

showed a significant difference between men and women agency leaders with respect to their top policy priorities. Whereas 49.6 percent of women department heads indi- cated that one or more of their top policy priorities was specifically intended to help women, only 34.9 percent of men department heads did so (p < .05). This finding seems to confirm a link between gender and women-related pri- orities among executive branch bureaucratic leaders and

Table 1 Descriptive Statistics on Respondents

Length of service Number Percentage in current position (percent)

Years 2 3–4 5–6 6+

Women 129 60.0 Men 86 40.0 Women of color 16 Men of color 8 Appointed by Republican governor 122 56.7 Appointed by Democratic governor 81 37.7 Appointed by Reform/Independent governor 12 5.6 Length of service 27.9 24.7 22.8 19.6

Educational background Highest degree Percentage Master’s 33 Bachelor’s 25 Law 23 PhD 8

Regional distribution Region Number Percentage Far West 29 13.5 Great Lakes 24 11.2 Mid Atlantic 24 11.2 New England 23 10.7 Plains 35 16.3 Rocky Mountain 14 6.5 Southeast 36 16.7 Southwest 30 14.0

Agency Leaders, Gendered Institutions, and Representative Bureaucracy 165

suggests a potential relationship between ascribed charac- teristics and policy outcomes.

Influences on Agenda Setting Given our choice to use agency leaders’ self-identified

priorities as a proxy for one element of how active repre- sentative bureaucracy might unfold, we were also inter- ested in examining the sources of influence on individual priority setting. The political science and public adminis- tration literatures, plus our own exploratory interview study, yielded the 10 possibilities listed in table 2. Table 2 in- cludes the mean strength-of-influence scores and standard deviations on a Likert scale of 1 (“not at all important”) to 5 (“very important”) for women and men for each of the 10 potential influences on choosing women-related policy priorities. The t-tests revealed no statistically significant differences between the mean scores of men and women for any of the 10 influences.

and policy priorities of the governors whom they serve. As we indicated earlier, however, substantial congruence be- tween the values of appointer and appointee is to be ex- pected. The influence of personal values, beliefs, and ex- periences is greater than influences emanating from the senior staff and culture of the agency they lead. Finally, it is greater than external pressures from other political ac- tors, the media, or legal mandates. The second highest means reported are for agency leaders’ own professional values and experiences.

Multivariate Results To explore these relationships further, we constructed

multivariate models and used logit regression to estimate them. To correct for the overrepresentation of women, redistributive agencies, and distributive agencies in our respondent pool, regression equations include a weights variable.14

Table 2 Comparison by Gender of Strength of Influences on Choosing a Women-Related Policy Priority

Men Women Mean Mean (s.d.) (s.d.)

Influence Governor’s general values 4.2 4.1 (n = 204) (.81) (.94) Governor’s explicit priority 3.9 3.6 (n = 203) (1.1) (1.2) Personal values/experiences 4.6 4.5 (n = 205) (.58) (.69) Professional values/experiences 4.4 4.4 (n = 205) (.81) (.62) Agency values and culture 4.0 4.2 (n = 204) (.90) (.90) Staff recommendations 3.6 3.7 (n = 205) (1.2) (1.1) Pressure from legislature 2.7 2.9 (n = 205) (1.1) (1.2) Pressure from interest groups 2.8 3.3 (n = 203) (.92) (1.2) Pressure from public/media 2.5 2.9 (n = 205) (1.0) (1.1) Statutory/regulatory requirement 2.8 2.7 (n = 204) (1.4) (1.3)

Both women and men department heads reported their own personal values, beliefs, and experiences were the strongest influence on their priority choice. We ran paired sample t-tests, comparing all possible pairs of influences, and found a statistically significant difference (p ≤ .01) between the strength of this influence and each of the other influences, except agency leaders’ own professional val- ues and experiences. For both women and men appointed to head public bureaucracies, the influence of the personal values, beliefs, and experiences they bring to bureaucratic leadership dominates their agenda-setting priorities. It is greater than the influence of the values, policy preferences,

Table 3 Logit Models of Individual and Contextual Influences on Agency Executives’ Choice of a Women- Related Top Policy Priority

Model 1 Model 2 N = 211 N = 145

Variable Odds ratio t-value Odds ratio t-value Gender (female) 1.92 2.08** 1.35 0.70 Race (person of color) 1.45 0.74 1.43 0.56 Years appointed 1.07 1.28 1.05 0.71 Graduate degree 1.37 0.76 0.71 –0.68 avggval 1.46 1.61* avgypro 1.18 0.41 avgyper 1.02 0.04 avgvcul 1.29 0.90 Political culture 1.26 0.91 Party (Democrat) 1.97 1.39 Redistributive agency 3.24 2.26** **p < 0.05 *p < 0.1

Table 3 presents the results of logistic regressions for two models. Model 1 contains variables that reflect possi- bly relevant characteristics of agency leaders: gender, race/ ethnicity, years in position, and graduate degree. Years in position is measured as a continuous variable. The other three variables are 0, 1 dummy variables, where women, persons of color, and individuals with graduate degrees are coded as 1. Model 2 adds the influence variables discussed previously15 and the measures of context: political culture, governor’s party, and agency type. The influence variables are measured on a Likert scale from 1 to 5. Political cul- ture is a continuous variable that is based on an index. Governor’s party and redistributive agency are both dummy variables with Democratic Party membership and redis- tributive agencies coded as 1. We present the coefficients in terms of odds ratios for ease of interpretation.

166 Public Administration Review • March/April 2005, Vol. 65, No. 2

The results for model 1 (table 3, column 1) show that even with other background characteristics controlled, women are 1.9 times more likely to have a women-cen- tered policy priority compared to their male counterparts. A more fully specified model (table 3, column 2) indicates, however, that a contextual variable best explains which agency heads are more likely to have women-friendly policy agendas. Working in a redistributive agency affects whether a leader pursues a women-centered policy agenda, no matter what the leader’s gender, other personal charac- teristics, or reported influences. People who work for re- distributive agencies are just over three times more likely to have a women-focused policy compared to people who work in other kinds of agencies. The influence of the ap- pointing governor’s values on an appointed executive’s choice of a women-related priority is also important, but much less so.

Discussion In Saltzstein’s (1979) mechanisms of responsiveness

model (figure 1), evidence of active representative bureau- cracy would be established by demonstrating that, com- pared to men agency leaders, the policy priorities set by women agency leaders are more frequently intended to advance the interests of women inside or outside the agency. In our preliminary analysis of the cross-tabulation data, we found such evidence: About 50 percent of women de- partment heads reported one or more policy priorities in- tended to help women compared to about 35 percent of their male counterparts.

Model 1 reinforces this finding, indicating that the odds of identifying as a top priority a policy that would advance women’s interests are 1.9 times greater for women than for men (p < .05). In the population of top-ranking politi- cal appointees who direct the affairs of state government agencies, model 1 offers evidence that a substantial pro- portion of women appointees are both “standing for” (pas- sive representative bureaucracy) and interested in “acting for” (active representative bureaucracy) women in the gen- eral population (Pitkin 1967). In this analysis, the gender of bureaucratic executives makes a difference for the agen- das they intend to pursue and may make a difference for agency policy outcomes.

What about the finding that women and men executives both assess their own personal values, beliefs, and experi- ences as the dominant influence on their choice of women- related policy priorities? We interpret this finding as a pow- erful reinforcement of the traditional notion of representa- tive bureaucracy and, more pointedly, of the first two elements of Saltzstein’s (1979) model. Each chief executive who occupies the number one position of influence in hier- archically structured organizations brings his or her unique

history to the agenda-setting task and establishes priorities on the basis of a mix of influences. Chief among these in- fluences, according to the research reported here, are the leader’s own values and experiences.

The governor’s values, preferences, and priorities are rated as the second strongest influence on priority setting by men agency heads and only one-tenth of a percentage point behind second place by women agency heads. Since only five of the 50 governors were women when we con- ducted this study, we surmise the aggregate set of values and experiences that inform the policy preferences of gov- ernors largely reflects men’s histories and, therefore, a powerful source of influence on agency leader priorities is less likely to push executive decision makers toward pri- orities that specifically advance women’s well-being.

Agency-related variables are rated as the second stron- gest influence on women department heads’ priority set- ting and the third strongest influence on agenda setting by their male counterparts. The strength of influence of these variables (“values and culture of the agency” and “recom- mendations from senior agency staff”) was above the mid- point of 3.0 for both women and men. Government agen- cies should not be discounted, therefore, as a source of agenda-setting cues for political appointees. We hypoth- esize that top-ranking appointees and career executive staff below the chief executive level are the agency advisors on whom the chief executive would rely most heavily, although not exclusively. Who are these influential advisors?

Another part of our survey helps to answer this ques- tion and begins to move the analysis toward the theory of gendered institutions. We asked, “Who can you rely on when you run into problems or need help with strategy? Who helps you turn your policy priorities into reality?” Analysis of the responses reveals that, of the almost 900 individuals identified as influential advisors, 66 percent work in the agency and 56 percent are men. Assuming sig- nificant overlap between agency advisors who assist ex- ecutives in setting priorities and those who help solve prob- lems of strategy and implementation, we conclude that men’s experiences, at least for more than half the agency leaders in this study, are also reflected in the thinking of another important source of influence on agency priority setting by chief executives.16

Model 2 substantiates the importance of taking into ac- count the gendered nature of political institutions when trying to understand how representative bureaucracy may actually work. When contextual variables are included, whether someone works for a redistributive agency—not gender—is the critical predictor of whether an executive chooses a women-related policy as a top priority. Kelly and Newman (2001) find that redistributive agencies are a particularly compelling example of a gendered institution, in that women hold a higher proportion of leadership posts

Agency Leaders, Gendered Institutions, and Representative Bureaucracy 167

and experience a higher incidence of sexual harassment than in other types of agencies.17

Our study integrates representative bureaucracy and gendered institutions theory by indicating that the way gender matters in executive priority setting is not as an ascribed characteristic of individuals, but as a social con- struction embedded in the institutional contexts in which agency executives make agenda-setting decisions. As Mar- tin contends, “Gender is such a constitutive part of formal organization that efforts to match worker and activity are often efforts to match gender and activity” (1992, 208). That the effect of the individual’s gender disappears in the multivariate model does not mean gender does not matter. Rather, we argue there are gendered institutional processes at play. In this case, women and some men executives ap- pear to be selected by—or indeed, seek out—the type of agency that is more likely to pursue an agenda that ad- vances women’s interests. The mission of redistributive agencies is inherently more oriented toward benefiting particular constituency groups, and, therefore, differences in policy priority choices between women and men execu- tives of such agencies will tend to be masked.

Conclusion We remain cautious in interpreting our results, for a

number of reasons. First, in the preliminary phase of the study reported here, 50 percent of women department heads did not identify a policy priority intended to help women. While women are more likely than men to pursue a women- friendly agenda, the fact that there is an equal number of women who do not make this choice suggests there are other important variables in play.18 Second, no consensus exists among representative bureaucracy researchers about a standard of evidence necessary to demonstrate the oc- currence of the phenomenon. Do all members, a majority of the members, or some other percentage of the members of a demographic group have to intend to act for their group’s members in the general population to confirm the occurrence of active representative bureaucracy? Third, the role that gender plays in determining policy outcomes is not yet understood. If women leaders have a difficult time turning their priorities into policy, for example, the fact that they are more likely to have a women-centered agenda is not as important.

Future research should continue to address an ambigu- ity embedded in the traditional notion of representative bu- reaucracy. The normative theory assumes, for instance, that members of a group with certain ascribed characteristics will “stand for” and “act for” people in the general popula- tion with the same characteristics, and that other people in government can not be expected to do so. With respect to gender, the assumption, stated in its most extreme form, is

that women will always pursue women-related agendas and men never will. Our finding challenges this assumption in that it is the contextual variable—agency type—rather than an ascribed individual characteristic that explains how rep- resentative bureaucracy works with respect to gendered in- stitutions. Both women and men in redistributive agencies are significantly more likely to choose as a top priority poli- cies that could advance women’s interests. Additional stud- ies that juxtapose the tenets of representative bureaucracy and insights from gendered institutions theory would be especially useful.

Despite these concerns, we are convinced that our ef- fort to wrestle with the complexity of representative bu- reaucracy contributes to long-standing empirical efforts to understand more completely whether and how active rep- resentative bureaucracy is realized.19 Our study has made progress in theory building by adapting and testing a model of how representation and responsiveness may be linked. Our examination of the policy priorities of women and men agency leaders and the contextual influences on those pri- orities suggests that continuing the long history of repre- sentative bureaucracy research is worthwhile. Systematic study of the relationships among the social construction of gender within institutions, policy priorities, and policy outcomes is needed to tell us more about how and when values translate into behavior and behavior results in en- acted policy.

Acknowledgments

We would like to acknowledge the important research con- tributions to this article of Julia Koschinsky, project director, Center for Women in Government and Civil Society, and the valuable comments of anonymous reviewers. We are also grate- ful for the research assistance of Kim Smith Cozzens, and for survey management by Alison C. Olin, Center for Women in Government and Civil Society. This study was supported by a grant from the Ford Foundation.

168 Public Administration Review • March/April 2005, Vol. 65, No. 2

Notes

11. Following widespread practice, we use the terms “women and men department heads” instead of “female and male department heads” (Thomas 1994; Welch and Studlar 1996; Meier and Wilkins 2002). When we use the terms “women” and “men,” we are using sex/gender categories that move away from traditional connotations of masculine and femi- nine, male and female.

12. Gerth and Mills (1946, 228).

13. Seldon’s The Promise of Representative Bureaucracy: Di- versity and Responsiveness in a Government Agency (1997) is a particularly noteworthy recent study.

14. He also offers examples of appointees who choose to assert their independence from the president who appointed them.

15. Studies that take a macro approach to the link between pas- sive and active representative bureaucracy propose a large number of influences on that link. After describing many possible influences, Thompson concludes, “In sum, then, there is no sense in denying that an impressive array of in- fluences can debilitate linkage between passive and active representation. Of equal importance, however, is the fact that other circumstances can help counteract these influ- ences” (1976, 217).

16. Kelly and Newman describe a similar model in their figure 1, “Typical Causal Model of Demographic Changes on Rep- resentativeness of Bureaucracies” (2001, 3).

17. Moore, for instance, observes, “I take it for granted that many factors other than managerial behavior contribute to the suc- cess of public enterprises. Indeed, it is plausible to me that the specific actions of managers may account for very little in the variance in results obtained” (1995, 3).

18. Like Kelly and Newman, we also do not include Lowi’s fourth category, constituent agencies, in the analysis because its defining characteristic is not substantive policy, but rather jurisdictional issues (Kelly and Newman 2001, 8).

19. Carroll (1992) provides a persuasive analysis of whether there is such a phenomenon as “women’s interests.”

10. The pilot study was conducted by coprincipal investigators Judith R. Saidel and Norma M. Riccucci, professor of pub- lic administration, Rutgers University. The interviews lasted about one hour and included 18 to 20 questions that asked agency leaders to identify three top policy priorities and the major influences that led to their choices. Demographic questions and questions about professional affiliations were also part of the interview.

11. We were able to identify states with relatively high numbers of women in executive appointee positions because gender, race, and ethnicity data on top-ranking gubernatorial ap- pointees in the 50 states from 1997 to the present are re- ported in publications of the Appointed Policy Makers in State Government project at the Center for Women in Gov- ernment and Civil Society. Unpublished data for the report, Women’s Leadership Profile, A Compendium Report, were used for sample selection. Identification of agencies in dif- ferent functional areas that are headed by women in each state is part of the publication series (Saidel 1998).

12. Given the low response rate among men (both absolutely and relative to the women in our sample), the men in our sample are clearly a select group. It seems likely that, all else equal, the male leaders most willing to respond were those who had an agenda friendly to women’s interests. They would be more interested in seeing the research question answered and more likely to find the study worthy of their time. Therefore, our findings probably overestimate the ex- tent to which men agency leaders have policy priorities that are friendly to women’s interests, attenuating the gender difference.

13. For instance, under “Governor’s general values, beliefs, and policy preferences” we included the following: “An example from prior research: I knew the governor cared about kids.”

14. The formula for the weighting variable is weight = (popula- tion proportion for gender/sample proportion for gender) * (population proportion for agency type)/(sample proportion for agency type). Six combinations are possible. For instance, (population proportion for women/sample proportion for women) * (population proportion for regulatory agencies)/ (sample proportion for regulatory agencies).

15. The strength-of-influence variables are as follows: average across priorities of the respondents’ assessment of the im- portance of governor’s values, beliefs, and preferences; av- erage across priorities of the importance of respondent’s own professional values and experiences; average across priori- ties of the importance of respondent’s own personal values and experiences; and average across priorities of the impor- tance of agency values and culture. These four strength-of- influence variables were included because they were rated at 4.0 or higher on a 1 to 5 Likert scale.

16. A government workforce study confirms the predominance of men in high-level career and appointed positions (Berheide and Saidel 2000). Based on EEO-4 gender, race, and ethnicity data collected by the U.S. Equal Employment Opportunity Commission, the study reported that in 1997, men held 66.6 percent and women held 33.4 percent of of- ficial and administrator positions in state and local govern- ments.

Agency Leaders, Gendered Institutions, and Representative Bureaucracy 169

17. Citing Guy (1994), the authors explain, “This pattern is consistent with the argument that once women are in a bu- reaucracy, sexual harassment often constitutes a barrier keep- ing them from remaining in a position or progressing to a higher position” (Kelley and Newman 2001, 17).

18. A preliminary analysis of all policy priorities identified by the respondents revealed that, of the total 643 priorities, over one-third dealt with improving agency operations (28.3 per- cent) or government operations (7.4 percent). This suggests

that, consistent with reform efforts in public management, agency leaders in 2000 focused substantial attention on ad- ministrative and management issues. Further analysis of all priorities identified by agency heads is beyond the scope of this article.

19. The 2003 volume, Representative Bureaucracy: Classic Readings and Continuing Controversies, edited by Dolan and Rosenbloom, includes a number of such studies.

References

Aberbach, Joel D., and Bert A. Rockman. 2000. In the Web of Politics: Three Decades of the U.S. Federal Executive. Wash- ington, DC: Brookings Institution.

Acker, Joan. 1990. Hierarchies, Jobs, Bodies: A Theory of Gendered Organizations. Gender and Society 4(2): 139–58.

Berheide, Catherine W., and Judith R. Saidel. 2000. The Chang- ing Government Workforce in States and Localities, 1990– 1997. Albany: University at Albany, State University of New York, Center for Women in Government and Civil Society.

Calas, Marta B., and Linda Smircich. 1992. Re-Writing Gender into Organization Theorizing: Directions from Feminist Per- spectives. In Re-Thinking Organization: New Directions in Organizational Research and Analysis, edited by Mike I. Reed and Michael D. Hughes, 227–58. London: Sage Publications.

Carroll, Susan J. 1985. Women as Candidates in American Poli- tics. Bloomington: Indiana University Press.

———. 1992. Women State Legislators, Women’s Organizations, and the Representation of Women’s Culture in the United States. In Women Transforming Politics, edited by Jill M. Bystydzienski, 24–40. Bloomington: Indiana University Press.

———. 1994. The Politics of Difference: Women Public Offi- cials as Agents of Change. Stanford Law and Policy Review 5(2): 11–20.

Crenshaw, Kimberle. 1993. Demarginalizing the Intersection of Race and Sex: A Black Feminist Critique of Antidiscrimina- tion Doctrine, Feminist Theory, and Antiracist Politics. In Feminist Legal Theory: Foundations, edited by D.K. Wiesberg, 57–80. Philadelphia: Temple University Press.

Dodson, Debra L., and Susan J. Carroll. 1991. Reshaping the Agenda: Women in State Legislatures. New Brunswick, NJ: Center for American Women and Politics.

Dolan, Julie, and David H. Rosenbloom, eds. 2003. Representa- tive Bureaucracy: Classic Readings and Continuing Contro- versies. Armonk, NY: M.E. Sharpe.

Dometrius, Nelson C. 1984. Minorities and Women among State Agency Leaders. Social Science Quarterly 65(1): 127–37.

Duerst-Lahti, Georgia, and Rita Mae Kelly, eds. 1995. Gender Power, Leadership, and Governance Ann Arbor: University of Michigan Press.

Duerst-Lahti, Georgia, and Cathy Marie Johnson. 1992. Man- agement Styles, Stereotypes, and Advantages. In Women and Men of the States: Public Administrators at the State Level, edited by Mary E. Guy, 125–56. New York: M.E. Sharpe.

Elazar, Daniel J. 1972. American Federalism: A View from the States. New York: Thomas Y. Crowell.

———. 1984. American Federalism: A View from the States. New York: Harper and Row.

Erickson, Robert S., Gerald C. Wright, and John P. McIver. 1993. Statehouse Democracy. New York: Cambridge University Press.

Gerth, H.H., and C. Wright Mills, eds. and trans. 1946. From Max Weber: Essays in Sociology. New York: Oxford Univer- sity Press.

Glenn, Evelyn N. 1999. The Social Construction and Institu- tionalization of Gender and Race. In Revisioning Gender, edited by Myra Marx Ferree, Judith Lorber, and Beth B. Hess, 3–43. Thousand Oaks, CA: Sage Publications.

Guy, Mary E. 1994. Organizational Architecture, Gender, and Women’s Careers. Review of Public Personnel Administra- tion 14(2): 77–90.

Hale, Mary M., and Rita M. Kelly, eds. 1989. Gender, Bureau- cracy, and Democracy. New York: Greenwood Press.

Hebert, F. Ted, and Deil S. Wright. 1982. State Administrators: How Representative? How Professional? State Government 55(1): 22–28.

Heclo, Hugh. 1977. A Government of Strangers. Washington, DC: Brookings Institution.

Keiser, Lael R., Vicky M. Wilkins, Kenneth J. Meier, and Catherine A. Holland. 2002. Lipstick and Logarithms: Gen- der, Institutional Context, and Representative Bureaucracy. American Political Science Review 96(3): 553–64.

Kelly, Rita M. 1998. An Inclusive Democratic Polity, Represen- tative Bureaucracies, and the New Public Management. Pub- lic Administration Review 58(8): 201–8.

Kelly, Rita M., and Meredith Newman. 2001. The Gendered Bureaucracy: Agency Mission, Equality of Opportunity, and Representative Bureaucracies. Women and Politics 22(3): 1– 32.

Kenney, Sally J. 1996. New Research on Gendered Political In- stitutions. Political Research Quarterly 49(2): 445–66.

Kingdon, John W. 1995. Agendas, Alternatives, and Public Poli- cies. 2nd ed. New York: HarperCollins.

Krislov, Samuel. 1974. Representative Bureaucracy. Englewood Cliffs, NJ: Prentice Hall.

Lipsky, Michael. 1980. Street-Level Bureaucracy. New York: Basic Books.

170 Public Administration Review • March/April 2005, Vol. 65, No. 2

Appendix Influences on Agency Leader Choice of Policy Priorities

Individual-level influences • Governor’s general values, beliefs, and policy preferences • Governor’s explicit priority • Your own personal values, beliefs, experiences • Your professional values and experiences

Organizational-level influences • Values and culture of the agency • Recommendation from senior agency staff

Environmental influences • Pressure from the legislature • Pressure from interest groups • Pressure from the general public and/or the media • Statutory or regulatory requirement

Lorber, Judith. 1994. Paradoxes of Gender. New Haven, CT: Yale University Press.

Martin, Patricia Y. 1992. Gender, Interaction, and Inequality in Organizations. In Gender, Interaction and Inequality, edited by Cecilia L. Ridgeway, 208–31. New York: Springer-Verlag.

Meier, Kenneth J. 1993. Representative Bureaucracy: A Theo- retical and Empirical Exposition. In Research in Public Ad- ministration, vol. 2, edited by James Perry, 1–35. Greenwich, CT: JAI Press.

Meier, Kenneth J., and Lloyd G. Nigro. 1976. Representative Bureaucracy and Policy Preferences: A Study in the Attitudes of Federal Executives. Public Administration Review 36(4): 458–69.

Meier, Kenneth J., and Joseph Stewart, Jr. 1992. The Impact of Representative Bureaucracies: Educational Systems and Pub- lic Policies. American Review of Public Administration 22(3): 157–71.

Meier, Kenneth J., and Vicky M. Wilkins. 2002. Gender Differ- ences in Agency Head Salaries: The Case of Public Educa- tion. Public Administration Review 62(4): 405–11.

Moore, Mark H. 1995. Creating Public Value, Strategic Man- agement in Government. Cambridge, MA: Harvard Univer- sity Press.

Mosher, Frederick C. 1968. Democracy and the Public Service. New York: Oxford University Press.

———. 1982. Democracy and the Public Service. 2nd ed. New York: Oxford University Press.

Mueller, Carol. 1984. Women’s Organizational Strategies in State Legislatures. In Political Women: Current Roles in State and Local Government, edited by Janet A. Flammang, 156–76. Thousand Oaks, CA: Sage Publications.

Naff, Katherine C., and John Crum. 2000. The President and Representative Bureaucracy: Rhetoric and Reality. Public Administration Review 60(2): 98–109.

Pitkin, Hannah. 1967. The Concept of Representation. Berke- ley: University of California Press.

Rehfuss, John A. 1986. A Representative Bureaucracy? Women and Minority Executives in the California Career Service. Public Administration Review 46(5): 454–60.

Riccucci, Norma M. 1995. Unsung Heroes: Federal Execucrats Making a Difference. Washington, DC: Georgetown Univer- sity Press.

Riccucci, Norma M., and Judith R. Saidel. 1997. The Represen- tativeness of State-Level Bureaucratic Leaders: A Missing Piece of the Representative Bureaucracy Puzzle. Public Ad- ministration Review 57(5): 423–30.

Rosenbloom, David H., and Jeannette C. Featherstonhaugh. 1977. Passive and Active Representation in the Federal Service: A Comparison of Blacks and Whites. Social Science Quarterly 57(4): 873–82.

Rosenthal, Cindy S. 1997. A View of Their Own: Women’s Com- mittee Leadership Styles and State Legislatures. Policy Stud- ies Journal 25(4): 585–601.

Saidel, Judith R. 1998. Women’s Leadership Profile: A Com- pendium Report. Albany: University at Albany, State Univer- sity of New York, Center for Women in Government and Civil Society.

———. 1999. Appointed Policy Makers in State Government, Trend Analysis 1997–1999. Albany: University at Albany, State University of New York, Center for Women in Govern- ment and Civil Society.

Saint-Germain, Michele A. 1989. Does Their Difference Make a Difference? Social Science Quarterly 70(4): 956–68.

Saltzstein, Grace H. 1979. Representative Bureaucracy and Bu- reaucratic Responsibility: Problems and Prospects. Adminis- tration and Society 10(4): 465–75.

Seldon, Sally C. 1997. The Promise of Representative Bureau- cracy: Diversity and Responsiveness in a Government Agency. Armonk, NY: M.E. Sharpe.

Stivers, Camilla. 1993. Gender Images in Public Administra- tion: Legitimacy and the Administrative State. Newbury Park, CA: Sage Publications.

Tamerius, Karin L. 1995. Sex, Gender, and Leadership in the Representation of Women. In Gender Power, Leadership, and Governance, edited by Georgia Duerst-Lahti and Rita Mae Kelly, 93–112. Ann Arbor: University of Michigan Press.

Thomas, Sue. 1994. How Women Legislate. New York: Oxford University Press.

Thompson, Frank J. 1976. Minority Groups in Public Bureau- cracies, Are Passive and Active Representation Linked? Ad- ministration and Society 8(2): 201–26.

Tolleson-Rinehart, Susan. 1991. Do Women Leaders Make A Difference? Substance, Style, and Perception. In Gender and Policymaking: Studies of Women in Office, edited by Debra Dodson, 93–102. New Brunswick, NJ: Eagleton Institute of Politics, Center for American Women and Politics.

Welch, Susan, and Donley T. Studlar. 1996. The Opportunity Structure for Women’s Candidacies and Electability in Brit- ain and the United States. Political Research Quarterly 49(4): 861–74.