Masters level assignment




Traditionally, studies couched within the theory of representative bureaucracy have focused on examining outcomes for social minorities within a specific set of administrative practices or public policies. The concept of administrative discretion typically lies at the core of these studies, as public servants do indeed need to enjoy a certain level of administrative discretion in order to incorporate the values of a given group into their everyday decision-making. Yet, while the centrality of administrative discretion is seldom, if ever, questioned and it is most often implicitly assumed, its actual effects are rarely explicitly explored. This study, constructed on the theoretical insights drawn from representative bureaucracy and role theories, examines the relationship between administrative discretion and the assumption of roles by public servants, in particular, the representation role of steward of public interest. The empirical results confirm that administrative discretion has a significant impact on the types of roles assumed by public administrators.

596 PAQ WINTER 2015

One would be hard pressed to locate a concept that is more central to the evolution of public administration theory and practice than administrative discretion. From discussions on the democratic nature of the administrative constructs to the ones on the latest managerial technique, administrative discretion always seems to creep into the conversation. In fact, a number of the classical debates in public administration come down to questions of proper handling of administrative discretion. The Finer (1940) – Friedrich (1941) debate is a case in point. Finer (1941) asserted that administrative decision-making can be and should be clearly organized within the frame of well-designed legislative and organizational mandates. Accountability, according to Finer (1941), is a function of strong external structures. Friedrich (1940), in contrast, asserted that public administrators are more than capable of self-control and the boundaries imposed by social norms and professional values would be able to channel administrative discretion within appropriate democratic expectations. Taken together, most of the current public administration scholars tend to agree with Friedrich (1940) and have resigned to the idea that organizational constraints have significant limits in terms of imposing exacting controls on administrative behaviors. Discretionary power has been and remains a prevailing reality for many public servants and their agencies and continues to be an essential ingredient in conditioning administrative responsiveness and accountability.1 As Wilson (1966) has once argued, for administrators “large powers and unhampered discretion” seem to be “the indispensable conditions of responsibility” (p. 373).

Yet, while the centrality of administrative discretion for public administration theory and practice is never questioned, it is rarely explicitly analyzed and receives only a limited amount empirical attention. In particular, very little is known about the effects that administrative discretion has on the roles assumed by public servants. This study was designed with the specific purpose of examining the relationship between perceived access to administrative discretion and the assumption of administrative roles. The research explores whether a public administrator who

PAQ WINTER 2015 597

believes that he or she enjoys higher levels of administrative discretion is more likely to assume the representation role of steward of public interest. This study directly builds on the work of Selden, Brewer and Brudney (1999) and Sowa & Selden (2003). It extends their empirical conceptualizations by testing previously unexamined relationships.

From the start it should be noted that many of the studies carried under the umbrella of the theory of representative bureaucracy have a number of common limitations. First, the majority of the studies are, for obvious and necessary methodological and conceptual reasons, bounded to the organizational level (Bradbury & Kellough, 2011). Second, as a rule, the studies driven by the theory of representative bureaucracy often address policy specific outcomes; hence, their generalizability is seriously challenged by their careful selection of their eventual sampling frame. Kennedy (2013) noted that an overreliance on convenient sources of data, such as education data, is a particular weakness of research examining representation. While insightful, the findings developed in these convenient settings are limited in their generalizability and can rarely be transferred to other policy domains. Furthermore, few scholars venture into exploring representation outside of the demographic-behavior link (Kennedy, 2013). Within this context, studies that address representation on the part of administrators at the individual level and which would focus on administrative discretion, but without imposing policy specific constraints, become rather warranted. Finally, and perhaps most importantly for the purposes of this study, administrative discretion is regularly assumed, but rarely empirically tested. A similar condition can be identified when it comes to administrative roles. Some (Sowa & Selden, 2003, p. 707) have previously called for research that would examine the “interaction between the role accepted by the administrator and the degree of discretion perceived by that administrator.” The research presented here was in large part undertaken with the scope of answering such calls.

Notwithstanding the actual empirical results, there are a number of broad, conceivably significant, contributions that this

598 PAQ WINTER 2015

article makes. First, this study combines in a somewhat original manner insights provided by the theory of representative bureaucracy and role theory. In a sense, it adds an important twist to the traditional delimitation of representative behaviors by tracing them through administrative roles rather than through traditional demographic links. Second, this represents one of only a few studies that empirically explored representative bureaucracy questions at the individual level. Most studies examined the topic within the context of institutional level outcomes. Finally, this research was framed in a manner that does not make it policy domain dependent; meaning, that empirical results cut across policy issues. This provides the possibility to generalize part of the results to the larger context of public administration.


Based on his observation of the interaction during World War II between middle-class oriented British civil service and the ruling party, Kingsley (1944) argued that a bureaucracy that is more representative of the social classes would be more responsive to the population’s needs. Soon after, American scholars picked up Kingsley’s (1944) idea as a possible solution to reconciling the longstanding inherent tension between bureaucracy and democracy. Given that social class did not seamlessly translate into the American democratic narrative, American scholars moved away from social class to demographic characteristics as the primary variable of interest.

Within the American context, Levitan (1946) asserted that the actions of agencies whose workforce composition reflected the demographic makeup of the citizenry, would have an easier time being accepted and interpreted as legitimate. Referencing the ineffectiveness of external controls, Levitan (1946) also supported representativeness as an improved choice for enforcing bureaucratic accountability. The link between representativeness and legitimacy was later echoed in the works of Long (1952) and Van Ripper (1958). In fact, in his seminal

PAQ WINTER 2015 599

article on the topic, Long (1952) went as far as to suggest that bureaucracy might be more representative than Congress.

The work of Mosher (1968) further extended the theory and to some degree provided a turning point in its development. Mosher (1968) clearly delineated the dynamics and implications behind passive (descriptive) and active representation. Mosher (1968) endorsed passive representation as a fundamental factor for public administration in a democracy. He suggested that even in the cases when passive representation might not lead to obvious democratic improvements; the mere presence of such setup in the administration provides democratic value on its own. This perspective is currently widely accepted among scholars as normatively appropriate and is always linked to a number of positive outcomes. Active representation, on the other hand, was normatively undesirable for Mosher (1968).

It may be noted that active representativeness run rampant within a bureaucracy would constitute a major threat to orderly democratic government. The summing up of the multitude of special interests seeking effective representation does not constitute the public interest. (Mosher, 1968, p. 12)

Most scholars, as Lim (2006) has noted, have shied

away from joining Mosher (1968) in rejecting active representation as a desirable characteristics of bureaucracy. Many have been rather supportive, or at the very least cautiously optimistic, of the benefits of active representation.

Another important chapter in the evolution of the theory is provided by the work of Krislov (1974), who advanced the theory by constructing understandings regarding organizational socialization. For Krislov (1974), the demographic backgrounds of public administrators offered the framework within which certain values are created and espoused. In cases when the organizational demographic structures mimic the backgrounds of the population at large, the values that are shaped within organizational interactions will reflect the demands of the citizenry. Similar to Long (1952) and Van Riper (1958), Krislov (1974) supported the idea that bureaucracy, through it

600 PAQ WINTER 2015

representational capacity, complimented and perhaps even filled the association gaps between citizens and their political structures, which are often left unmet by political institutions such as Congress. Table 1 provides a summary of the presumed direct and indirect social benefits of bureaucratic representation. Table 1 Social Benefits of Bureaucratic Representation2

Partiality (advocacy) Bias in favor of a given social group and against other groups.

Shared values and beliefs

Discretionary behavior in accordance to one's own values and beliefs - intended or not - serves the interests of the social group with which one associates.

Empathic understanding

Due to a common social background, administrators not only share but also understand the values and norms of a specific social group.

Checking Public administrators who associate with a particular social group can check the excesses in behaviors of their colleagues.


Public administrators not associated with a specific social group might restrain their behaviors for fear of being disapproved of, exposed, or otherwise checked.


With time the mere presence of bureaucrats from minority social groups can lead to enhanced empathic understanding of their colleagues.

Demand inducement The presence of minority bureaucrats can stimulate more service demand on the part of minority clients.

Coproduction inducement

Minority bureaucrats might have greater success in stimulating behavioral changes on the part of minorities for purposes of improved program outcomes.

Adapted from Lim (2006) The most recent turn in the intellectual progress of the

theory is in large part hedged on the idea of role acceptance. A

PAQ WINTER 2015 601

number of scholars (Kennedy, 2013; Meier & Bohte, 2001; Kelly, 2013; Selden, 1997; Selden, Brudney, & Kellough, 1998; Sowa & Selden, 2003) have argued that role acceptance provides an enriched set of insights into understanding the representativeness of bureaucracy, which might under different conditions be missed by the traditional link between social minority and representation. Although, demographic background still remains a fundamental ingredient, scholars are now becoming open to the idea that it might not be the most critical in explaining representation; in fact, it might actually be secondary to role assumption, especially outside of policy specific questions (Kennedy, 2013). That is to say, outside of policies that specifically target women or social minorities, variables such as gender and race respectively, might be less useful in predicting active representation than administrators’ expectations regarding what constitutes their appropriate roles. “[T]he role accepted by administrators and the amount of discretion they perceive themselves to have exerts an impact on the outcomes serving minority interests above and beyond the individual characteristics of the administrators” (Sowa & Selden, 2003, p. 707).

All things considered, despite its wide acceptance and appeal, the theory does motivate a healthy dose of well-grounded criticism, some of which is coming directly from its supporters. First, scholars have argued (Frederickson & Smith, 2003) that the theory’s concept framework is fraught with ambiguity. One of the results of this nebulousness, particularly when it comes to the ambiguous conceptualization of core terms, is that there is surprisingly relatively little empirical research in the area. Furthermore, most of the empirical studies that are conditioned by the theory follow similar customary formats, use similar type of data and rarely venture into challenging or exploring some of the more vague aspects within the theory. Within the same vein, Lim (2006) was rather critical in his discussion of how partiality is conceptualized in empirical studies, arguing that what the instruments often capture are things other than partiality. Lim (2006) asserted that most of the assumptions behind the possible benefits of active representation are regularly confused with

602 PAQ WINTER 2015

those of passive representation and are frequently exaggerated, hence limiting the usefulness of the research based on those assumptions. Third, most studies take the relationship between an individual’s background and a supposed group as a given (Kennedy, 2013). This link, however, is often much weaker and less direct than typically believed. Krislov (1974), for instance, asserted that the most rigorous test for the representative capacity of bureaucracy would indeed be whether passive representation could seamlessly translate into active representation. Fourth, as Thompson (1976) argued, organizational socialization raises a significant challenge to the supposed link between passive representation and active representation. Finally, the mechanics linking passive representation and administrative outcomes are also highly ambiguous and not fully understood. For instance, does the presence of passive representation increase the probability of active representation? Or is passive representation necessary or sufficient for active representation? Wilkins and Keiser (2006) suggested that representation’s central question might not be whether there is a link between passive and active representation, but under what conditions such link exists.


There are many powerful implications that are bound to come with any active degree of administrative discretion and they warrant important levels of attention. In their efforts to act as representatives of citizens’ interests, through their discretion, in particular when it comes to interpreting public policy, public administrators regularly wield significant levels of political power (Mosher, 1968)) and are consequential policy makers in their own right (Kennedy, 2013; Lipsky, 1980). The legitimacy of this routine exercise of power, however, is often challenged and consistently triggers heated academic debates. As Lim (2006) has noted the line between legitimate active representation through administrative discretion and partiality exhibits dangerous levels of fuzziness. Many often believe that

PAQ WINTER 2015 603

they speak of the former, when in reality they are desperately attempting to justify the later.

Scholars have yet to reach a general consensus regarding the normative placement of administrative discretion within public service and it is hard to envision this happening any time soon. The theory of representative bureaucracy, at least up to this point, provides for one of the better approaches for reconciling the complexity associated with the attempts to properly operationalize administrative discretion. The theory posits that the actions and power that come with the regular access to discretion would be more representative of the public and its interests if the public servants would associate through their social and demographic characteristics to the citizens whom they serve (Denhardt & deLeon, 1995; Krislov, 1974; Meier, 1975; Mosher, 1968; Selden, 1997). The benefits of a representative administration, according to theory of representative bureaucracy, are not limited to a specific policy area or agency type. Any administrative area is set to gain from being more in tune with the demographic make-up of the population that is being served.

Administrative discretion is to be sure a fundamental concept within public administration and the theory of representative bureaucracy. It lies at its very core, especially in terms of active representation on the part of public servants. Scholars tend to agree that in order for administrators to engage in representation, they need to have access to administrative discretion first before they can decide to creatively make use of it (Denhardt & deLeon, 1995; Meier & Bohte, 2001; Meier & Stewart, 1992; Sowa & Selden, 2003). Depending upon the organizational and policy contexts, public servants have the discretion to shape the performance of their bureaus if they embrace a representative type role. Individual preferences, especially in terms of shaping policy interpretations, do matter and often carry significant power, sufficient to impact policy outcomes.

Meier and Bohte (2001) conceptualized administrative discretion as the ability of public administrators to command a sphere of influence to take actions that reflect the specific values

604 PAQ WINTER 2015

they might hold. Following the line of a similar argument, Sowa and Selden (2003) defined administrative discretion as an administrator’s perceived latitude in effecting administrative outcomes. For the purposes of this research, the perspective offered by Sowa and Selden (2003, p. 703) is embraced. Administrative discretion is conceptualized and measured at the individual level, specifically within one’s perception of one’s levels of discretion. Given that one simultaneously shapes and is shaped by one’s environment and that reality is often socially filtered and constructed through one’s values and experiences (Weick, 1995), one’s perceived level of discretion does perhaps perform better in capturing administrative discretion than other commonly used measures, which are based on organizational structure. In other words, a perception based measure is preferred because two administrators within similar organizational structures could, depending on their sense-making mechanics, differ significantly in their evaluation of how much discretion they enjoy; hence, the part played by the latter in conditioning their everyday behaviors. Measures of administrative discretion, which are constructed on organizational structures would miss or perhaps even distort this link between perceived administrative discretion and behaviors.

The uses and impacts of administrative discretion are believed to be especially important for organizations which are not legislatively designed to engage in active representation. In these contexts, public servants must be able to command respectable spheres of influence in order to translate the values they hold into policy outcomes (Meier & Bohte, 2001; Sowa & Selden, 2003). Here, it becomes crucial to note that the mere presence of administrative discretion does not guarantee representative outcomes nor representation per se; neither the literature nor the theory itself make claim to such associations. Although administrative discretion might be necessary it is not sufficient. Public servants still have to choose whether to exercise it in a manner that would fit within a representation pattern or to remain loyal to organizational doctrines.

PAQ WINTER 2015 605


Scholars (Dolan, 2000; Selden, 1997; Selden et al., 1998; Sowa & Selden, 2003; Zimmermann & Allen, 2009) have suggested tenure as a critical variable in explaining representative behaviors on the part of public administrators. Although there is a wide agreement in regards to tenure and representation being intimately linked, the direction of the relationship is less certain. On the one hand, with tenure, public administrators, due to their growing experience and organizational standing, are expected to be able to command increasing levels of power and discretion. This, in turn, will provide them with the necessary organizational influence and discretionary space to engage in representation should they decide to do so. Under this perspective, tenure is often linked within increased levels of policy involvement and representative type behaviors on the part of public administrators (see Svara 1990; Diggs & Roman, 2012; Roman, 2014; Roman, 2013a). On the other hand, with tenure employees can become increasingly indoctrinated with organizational values (Ewing, 1977; Ramos, 1981; Scott & Hart, 1979). As a result, unless representation is a specific value mandated at the organizational level, public administrators would be less likely to assume a representative type role (Dolan, 2000; Selden, 1997; Sowa & Selden, 2003).


Roles and Role Theory Role theory derives its core from the theatrical metaphor

(Biddle, 1986). The theory postulates that within a given social context every individual will have to assume a specific role, which will constrain the individual to following a given “script” and “enacting” expected behaviors. The framework focuses on individuals and their behaviors and it provides an accommodating perspective that bridges anthropology, sociology, social psychology and institutionalism. Roles are regularly applied to the study of consensus, social perception,

606 PAQ WINTER 2015

formal organizations and role conflict. It is assumed that individuals carry commonalities in their habitual behaviors, which are fairly predictable in cases when the context is known. Although there are a number of definition present in the literature (table 2), in grand lines, a role can be defined as patterned human behavior. Roles are formed by interactions with others and with institutional constraints. They are influenced by one’s self-expectations, the expectations of others as well as by contexts (Biddle, 1979, 1986).

Hickson (1966) asserted that, under one form or another, organizational theories, despite their apparently distinct terminologies, differ only in the degree of role specificity and legitimacy of discretion attributed to behavior. Hickson (1966) argued that the degree of role specificity ranges from complete legitimate discretion to fully specified role behavior. Frank (1964) suggested three types of roles depending of their specificity: underdefined, well-defined, and overdefined. Public administration scholars have associated role specificity with role responsibilities (e.g. Selden et al., 1999). Public servants with underdefined roles might enjoy more discretion than those with well-defined roles.

Table 2 Role Definitions

Definition Author An internally consistent series of conditioned responses. Cottrell (1942, p.

617) Includes the attitudes, values and behavior ascribed by the society to any and all persons occupying a specific status.

Linton (1945, p. 77),

Actions of individual occupants of a position. Newcomb (1950, p. 280)

Behavior of actors seen in the context of its functional significances.

Parsons (1951, p. 25)

Position differentiated in terms of a given social structure.

Levy (1952, p. 159)

Activities which in combination produce the organizational output.

Katz and Kahn (1966, p. 179)

An identity, a set of characteristic behaviors, or a set of expectations.

Biddle, (1979, p. 8)

PAQ WINTER 2015 607

Table 3 Conceptualizing Roles Core Propositions Certain behaviors are patterned and are characteristic of persons within contexts. Roles are often associated with sets of persons who share a common identity. Persons are often aware of roles, and to some extent roles are governed by the fact of their awareness. Roles persist, in part, because of their consequences (functions) and because they are often imbedded with larger social systems. Persons must be taught roles and may find either joy or sorrow in the performances thereof. Conceptualization of Roles Roles are behavioral (does not include: sex, race, national origin, attitudes, norms, values). Roles are performed by individuals. Roles are limited by context. Roles are constructed on behaviors that are characteristic of a set of persons and contexts. Adapted from Biddle (1979)

Administrative Roles

As a rule, agencies impose on its members specific work roles; each role is usually defined as a specific set of expectations and behaviors required from individuals within a certain job (Kahn et. al, 1964). Institutions themselves can be defined on the whole as symbolic and behavioral frameworks bound by rules, which motivate certain individual roles (DiMaggio & Powell, 1991). While roles are often associated with positions, within the context of current governance, roles need not be position specific, as individuals within diverse positions can engage in similar roles and vice versa. The contours of a role are transmitted and infused into the workforce through formal (e.g. job descriptions, meetings) and informal channels (socialization, organizational artifacts). Within every social context individuals might be simultaneously faced with the demand to enact several context dependent roles; generally, however, with time, one single role will become dominant (Merton, 1949; Biddle, 1979; Sieber, 1974). Once

608 PAQ WINTER 2015

institutionalized, roles can become so powerful that they provide a complete sense-making mechanism within the organizational milieu. In fact, scholars (DiMaggio & Powell, 1991; Meyer & Rowan, 1977) asserted that roles might often persist even after the institutional and rational motivations behind them might have long disappeared.

For Selden et al. (1999) “an administrative role is a cohesive set of job-related values and attitudes that provides the public administrator a stable set of expectations about his or her responsibilities” (p. 175). There are several important dimensions to this definition. First, the attitudes and values are job related. This means that one’s role will be defined by one’s job and work environment. Second, one’s expectations represent one’s interpretations of a given set of values. This implies that the actual set of values need not be objective or empirically tested, all that matters is one’s interpretation of that given value set. Finally, and most critically, the set of expectations regarding one’s role is assumed as being stable. The latter is highly significant, since it would imply that expectations do not change easily and a snapshot analysis would certainly provide an adequate evaluation of the overall nature of roles in public administration.

In order to truly understand the intricate dynamics evolved in administrative roles as conceptualized through theory of representative bureaucracy it becomes highly useful to carefully reexamine the passive-active distinction as framed by Mosher (1968). For Mosher (1968) descriptive representation is concerned with the social origins of the administrators (e.g. education, demographics, location within the social matrix) and the extent to which their collective profile reflects the society at large. According to Mosher (1968) passive representation can be rather easily measured and evaluated. Unlike passive representation, active representation is, per Mosher (1968), fully linked to expectations. Active representation encompasses directed efforts by administrators to press for the interests of those whom the administrators believe to represent. Active representation is conceptualized in its entirety through administrators’ guiding set of expectations. Such a formulation

PAQ WINTER 2015 609

would put into question the dominance of demographic backgrounds, passive representation in sum, as the driver behind active representation. Mosher (1968) argued that there are no realistic grounds to believe that a public servant with a given set of demographic characteristics will automatically nurture the necessary set of expectations for representing the interest of others with a similar makeup. In fact, Mosher (1968) suggested that administrators with a certain social background might “lean over backwards” in order not to lead others to believe that they engage in partiality. This view is also partially echoed by Saltzstein (1979), who asserted that the best way to represent and serve one’s social group in the long run is by being impartial.

Herbert (1974) suggested that by and large the challenges associated with role assumption will be present regardless of race. Organizations habitually generate a system of rules that demand certain behaviors as appropriate. As such, associations with social minorities can be often overruled by organizational expectations. Within the context of representative bureaucracy, then, administrative roles might often override administrators’ ethnic or gender characteristics. Selden et al. (1998) noted:

While minority employees can be expected to embrace this role [representative role] most often and most closely, nonminority administrators may also adopt it as a result of their background or socialization…In some circumstances, minority employees with discretionary authority may avoid the minority representative role. Regardless of race or ethnicity, moreover, the attachment of public administrators to the representative role will vary, so that they may not always make decisions and take actions responsive to the minority community…employee’s perception of their work role conditions the translation of demographic (and other) characteristics into policy outputs. (p. 720)

Although there are a number of frameworks and

typologies of administrative behaviors and roles (see de Graaf,

610 PAQ WINTER 2015

2011; Diggs & Roman, 2012; Kennedy, 2013; O’Kelly & Dubnick, 2005; Roman, 2013a; Selden, 1997), two models, somewhat more than others, have developed a significant standing in academe. Downs’ (1967) administrative role typology represents perhaps the most recognizable categorization currently present in the literature. Downs’ (1967) five roles - climber, conserver, zealot, advocate and statesman – are a common occurrence in public administration texts. Yet, somewhat surprisingly, given their popular appeal, the roles, as delineated by Downs (1967), have attracted only a limited amount of empirical attention.

The role conceptualization by Selden et al. (1999) represents the other relatively well established perspective on administrative roles. Table 4 provides the role descriptions. It also references the public administration literature where similar roles have been discussed. It is important to note here that, when compared to other roles identified by Selden et al. (1999), the idea of the public administrator as a steward of public interest has attracted on aggregate significantly more attention from public administration scholars.

PAQ WINTER 2015 611

Table 4. Administrative Roles

Role Role description Similar to concepts


Stewards of the Public


Public administrators "express a desire to participate in the formulation of good public policy - that is, policies that incorporate the needs and concerns of all citizens, disadvantaged groups…are more committed to social and political goals than to policy efficiency...see themselves serving the public and furthering the public interest, independent of the goals of elected officials or management" (Selden et. al, 1999, p. 185).

Abney & Lauth (1985), Chandler (1984), Box (1992), de Graaf (2011), Hassett & Watson (2002), Karnig & McClain (1980), Kennedy, 2013), Lovrich (1981), Sowa & Selden (2003), Wamsley et al. (1990)

Adapted Realists

Public administrators "balance equity and fairness with individual concerns. They express a commitment to both good management and equity….the expression of these values may be mitigated or influence by sources external to the individual, such as rules, regulation supervisors, and legislators....they reject the general value of neutrality...and hold the ideal social equity as important....they recognize that they must work within system constraints-rules and proper lines of authority-to survive in the bureaucracy" (Selden et al., p. 187).

Box (1992), de Graaf (2011), Kennedy (2013), Nalbandian (1990)

Businesslike Utilitarians

Public administrators "value efficiency as an organizational and individual goal…they are willing to reject what more senior agency officials tell them to do….will opt for the most efficient solution, ensuring the public interest is served…set limits on their quest for efficiency that prevent them from making exaggerated claims about a program for the sake of generating support...view efficiency as more important....they reject any politicization of their role and do not wish to advance the interests of less privileged or minority citizens...Although they feel political pressure from elected officials they are ambivalent about their relationship with these officials (Selden et al., p. 188).

612 PAQ WINTER 2015

Table 4, continued

Resigned Custodians

Public administrators "see themselves as neutral agents…who know their boundaries, which consist of established rules and regulations and expectations of supervisors and the chief elected officials…they distinguish the appropriate role of nonelected public employees from that of elected public officials...they feel no inclination to play a mediator role" (Selden et al., 1999, p. 189).

Box (1992), de Graaf (2011), Kennedy (2013), Nalbandian (1990)

Practical Idealists

Public administrators "see themselves as highly responsible and professional - working efficiently, quickly, and accurately, while implementing and advocating policy positions and legislation in the public interest…they do not believe they are agents of elected officials...committed to social equity...reject neutrality and the politicization of the public service" (Selden et al., 1999, p. 190).

de Graaf (2011), Kennedy (2013)

There are several reasons why the conceptualization

provided by Selden et al. (1999) was preferred. First, the study from which these roles were derived is among the more recent and complete discussions of administrative roles. It is also among the few works that attempted to bridge role theory and the theory of representative bureaucracy. Second, it is based on Denhardt and deLeon’s (1995) framework of administrative responsibility, which is typically believed to be one of the better ways of capturing the complexity of everyday decision-making challenges faced by public administrators (Selden et al., 1999; Sowa & Selden, 2003). Third, unlike other constructs suggested by scholars, for instance those by Downs (1967), these five constructs have been empirically derived and generated specifically from a public administration perspective with special sensitivity to public service ethos. In addition, perhaps even more importantly, the role constructs as formulated by Selden et al. (1999) are not constrained to a given policy domain nor are they position dependent. Their broad applicability to the universe of public administration allows for significant levels of generalizability that cut across policy sectors and professional associations. Finally, the role of steward of public interest is clearly much more representative in terms of underlining behaviors than any one of the remaining four. This permits for much cleaner and robust model. This means that although the

PAQ WINTER 2015 613

different roles overlap in terms of characteristic behaviors, there is sufficient variability to allow for empirical analysis.


Based on the insights drawn from representative bureaucracy and role theories, this study explores three main hypotheses. The hypotheses are framed within the context of administrative processes and agencies that are not legislatively designed to encourage and emphasize representation.3 H1. Public administrators who believe that they enjoy high levels of administrative discretion are more likely to assume the role of steward of the public interest, compared with public administrators who believe that they have low levels of administrative discretion. H2. Public administrators who believe that they should be involved in policy formulation are more likely to assume the role of steward of public interest, compared with public administrators who believe that they should not be involved in policy formulation. H3. Public administrators with higher levels of tenure are less likely to assume the role of steward of public interest, compared with public administrators with lower levels of tenure.


The empirical literature on representative bureaucracy behavior and outcomes (RBB&O) typically estimates the desired variables using the following regression model: RBB&O=βX+α+ε where X is a matrix of individual, or organizational characteristic, or both.

The model examined here has one categorical dependent variable, three independent variables and ten control variables. The five roles, as conceptualized by Selden et al. (1999), represent the five categories of the dependent variable: steward

614 PAQ WINTER 2015

of public interest, adapted realist, business like utilitarian, resigned custodian and practical idealist. It is hypothesized that an administrator’s level of discretion, expectation regarding involvement in policy formulation and tenure influence role assumption. In order to control for other possible explanations, ten additional variables are considered: perception of stakeholders’ expectations, position, age, gender, minority association, education, political ideology, the number years of government service, job satisfaction and the number of years in current position (see Appendix A for the rationale and detailed discussions of the model variables).


Sampling Frame The membership of the National Institute of

Governmental Purchasing (NIGP) was identified as an appropriate sampling frame for this research. There are several important reasons why this professional sub-group of public administrators was of particular interest and was chosen for this study. First, although NIGP’s membership is diverse in terms of education, gender, age and ethnicity – due to the professionalization efforts of the last decade, there are rather high levels of consensus regarding professional standards and expectations with the field (Emmett & Wright, 2012; Roman, 2013b; Thai, 2001, 2008). This fact is particularly important since this would mean that individuals will have clear and well- developed understandings about the overall professional expectations and their roles. Also, given their direct proximity to financial outlays, their actions and decisions are usually under important levels of scrutiny. The nature of their work contexts resembles a metaphorical “stage,” in which actors have to follow the provided “institutional script.” Consequently, it was expected that role behaviors would be easily identifiable across institutional levels and relatively homogeneous in character.

Second, the procurement professionals in the sampling frame came from a variety of organizations and they were employed by different levels of government that cut across

PAQ WINTER 2015 615

multiple policy areas. Public procurement, in general, as an area, routinely operates across policy domains. This allows for important levels of generalizability. Given that the sampling frame is not constrained to specific organizations or policy issues, the results, too, could conceivably be extended to the broader universe of public administrators. This realization is further strengthened by the fact that both the model and the instrument employed to collect the data did not make any major limiting assumptions about the nature of the administrative work of those being examined.

Third, public procurement specialists, as a professional sub-set, are rarely the target of systematic empirical research. Considering the latter, this research would be able to claim important levels of originality within an area that is relatively underexplored. Fourth, due to previous work in the area, this represented an accessible research environment for the researcher, for which entrance paths were previously established. In many ways, although not the primary motive, the selection of this specific sampling frame was driven by convenience considerations.

Finally, and perhaps most importantly, much of the body of extant scholarly literature would suggest that public procurement specialists are faithful enforcers of procurement ordinances (Thai, 2001, 2008). Any deviation from established procurement practices, which active representation would be, is normally discouraged and could be even penalized (Diggs & Roman, 2012). Administrative discretion, too, is rarely encouraged or supported. The rule-based professional rigidity offered by the public procurement context, makes for a very challenging environment for the enactment of representative behaviors, hence it provides a rigorous test for the proposed model. Confirming the hypotheses within this setting would allow the possibility of generalizing the results to environments which are more representation-friendly; to paraphrase some of Fred Ebb’s most famous music lyrics if the model makes it here it will make it anywhere.

616 PAQ WINTER 2015

Sampling Procedures The sampling procedures for this study aimed to exploit

the benefits of random sampling. A random sample of 2,000 contacts was drawn from the contact database of NIGP membership. In January of 2013 NIGP had a total of approximately 16,000 members. The 2,000 contacts were selected from the full list of members using a random sequence generator in Microsoft Office Excel. The contact info was obtained directly from NIGP’s research department. Given that typically response rates for NIGP internal survey range between 15% and 25%, a large contact sample was believed necessary in order to obtain the minimal work sample of approximately 3904 required for purposes of valid power and generalizability estimation. Given that sample representativeness is often considered to be more important than the sample size itself (Gliner, Morgan, & Leach, 2009), the main goal of the sampling procedures was securing a representative picture of NIGP membership.


The instrument employed by this study drew heavily on the conceptualization provided by Selden et al.’s (1999) research. Selden et al. (1999) have studied a diverse group of public administrators who had no apparent professional similarities. The instrument items proposed by the authors were broad in their formulation, as such, offering important levels of cross-sectional applicability. For the purpose of this study, survey items were transformed with an emphasis on action oriented statements and were modified as to remove any double- barreled questions. Given the design of the original study no reliability statistics for the role constructs were available. As such, an important part of this research was the provision of a reliability evaluation of survey instrument.

Each one of the five role constructs was measured using four Likert-type items. The average rating for the four items for each construct was used to identify the dominant role assumed by the public administrator and the responses were coded accordingly. The main reason for choosing average score rather

PAQ WINTER 2015 617

than total score lies within its ability to provide a decision even in cases with missing item responses. Hence, even when a respondent did not answer one of the items within the construct a decision can be made. In instances when a respondent would have a similar average across items for several different constructs, the number of answered items was used as the first tiebreaker. For example, if the respondent cumulated similar averages for two constructs the respondent was coded in the role within which he or she answered most items. Inter-item variance was used as the second tiebreaker. All being equal, in cases of equal averages the respondent was coded in the role that had the smallest inter-item variance. In cases when neither the average rating, number of completed items nor inter-item variance was useful in coding, the respondent’s self-identification was used to determine one’s dominant role.

A pre-final version of the instrument was self- administered and reviewed for accuracy and language appropriateness by ten tenured and experienced public procurement professionals. Their reviews led to a number of linguistic changes. The primary theme of their concerns appeared to be the overly “academic” language used in some of the questions. Appendix B provides the complete instrument employed for purposes of data collection.

Data Collection

The survey instrument has been administered using the SurveyMonkey platform.5 The on-screen presentation of the survey was broken down into five distinct pages. The first page included instructions only. The second page introduced the role measuring items. The respondents were given twenty statements and they were asked to indicate their level of agreement with each statement. The order in which each statement appeared was randomized for each respondent. In a similar manner the options for question three on page four were randomized. The randomization of the statements was introduced in order to reduce the response bias that might have been induced by the placement of the items. As a result, it should be expected that the answer to one specific statement was not systematically

618 PAQ WINTER 2015

influenced by its order of appearance or by the statements preceding it. Page four asked the respondents to self-identify within a given role based on the provided descriptions. The last on-screen page collected demographic and institutional level variables as they pertained to the respondents. The model and demographic items on page three and page five respectively, were not randomized. The order of pages and the items on pages four and five were identical for all respondents.

A total of 512 individuals replied to the invitation to participate. This represents 25.6 % response rate. Out of the total number of respondents, 493 completed and submitted the survey, while 19 had started, but only partially completed the survey – hence a completion rate of 96.28%. Although this response rate appears somewhat low, it is rather characteristic for large N Internet surveys and it falls within the upper limits for studies driven by NIGP database. The response rate also fits within the overall recent trends for Internet-based surveys observed in social sciences.

For any study based on data collected through self- administered instruments, nonresponse bias can be a serious concern. In order to check for any possible significant nonresponse bias independent sample t-tests for model variables among early and late respondents were conducted. The survey responses were ordered based on their time of receipt, with the first 10% labeled as early responses and last 10% labeled as late responses. The results for the independent samples t-tests raised no concerns regarding possible nonresponse bias. The only statistically significant difference in means (p<.05) between early respondents and late respondents was observed for role item 11 – “regardless of political pressure, I take the decision which is best for my organization.” Outside of item 11 no other statistically significant differences in means among early and late responses were identified among the model items.

Statistical Method

For purposes of hypotheses testing, multinomial logistic regression (MLR) was employed. MLR is an extension of binary logistic regression that allows for more than two categories of

PAQ WINTER 2015 619

dependent variables to be examined by means of several independent variables. MLR uses maximum likelihood estimation procedures in order to determine the probability of a subject being part of a specific group (Hair, Black, Babin, & Anderson, 2010; Kwak & Clayton-Matthews, 2002; Liao, 1999). The primary methodological appeal of MLR is that it is robust against deviations in normality, linearity or homoscedasticity. MLR makes fewer assumptions than discriminant analysis and linear regression analysis, and it is robust when those assumptions are not fully satisfied (Hair et al., 2010; Hosmer & Lemeshow, 2000; Klenbaum & Klein, 2000). Sample size and outliers represent critical considerations within MLR. While, there is no exacting criterion regarding the minimum sample size, scholars typically agree that the sample should be sufficiently large to provide a minimum of 10 cases per independent variable; larger ratios such as 30:1 provide improved confidence in the results (Hair et al., 2010).

Inspection of the data revealed that there were neither major issues regarding the assumptions being met nor regarding data quality.6 There were some minor concerns regarding the skewness in the distribution of two variables - number of years in current position and the number of years with current organization. To alleviate the concern, a logarithmic transformation was employed for both cases. Multicollinearity, too, did not appear to be an issue, as none of the correlation coefficients were above 0.85. Most importantly however, due to the fact that approximately 93% respondents indicated that they were either white or black – in terms of the empirical analysis, the race variable was substituted with a dichotomous variable of minority association (whether respondent thought of oneself as a minority). It was assumed that whether a person considered oneself as a minority was a more accurate and descriptive variable, and provides improved information for the purposes of this study than the actual race label that one carries.

620 PAQ WINTER 2015


Roles The five constructs used to code the respondents were

found to be adequate, with acceptable levels of internal consistency. All five roles constructs, “steward of public interest” (α=0.765), “adapted realist” (α=0.675), “businesslike utilitarian” (α=0.763), “resigned custodian” (α=0.733) and “practical idealist” (α=0.651) had Cronbach’s Alpha levels above 0.65. The coding results uncovered “practical idealist” and “adapted realist” as the two most commonly assumed roles by public administrators in the sample. The role of “businesslike utilitarian” was identified as being by a large margin the least assumed role (table 5).

Table 5 Sample Summary by Role Role Frequency Percent Practical Idealist 150 30.4% Adapted Realist 143 29.0% Steward of Public Interest 97 19.7% Resigned Custodian 68 13.8% Businesslike Utilitarian 35 7.1% Total 493 100% Hypotheses The results of the empirical analysis lend support to the hypothesized associations between presence of administrative discretion, individual expectations, tenure and the assumption administrative roles. Expectation about involvement in policy formulation (p < .05), perceived expectations of stakeholders (p<0.05), administrative discretion (p <.01), tenure (p <.05) and the number of years of public service (p <.01) are found to be significant in explaining the assumption of at least one role. No other variable including gender (p>.05), minority association (p>.05) nor position (p>.05) is found to be statistically significant. Similarly, job satisfaction (p>.05), education (p>.05) and political ideology (p>.05) fail to reach statistical

PAQ WINTER 2015 621

significance. Overall, the model classifies correctly 45.8% of the cases. This is significantly higher than the minimal criteria of 29.91% derived using marginal percentages. Table 6 provides the model fit statistics and the likelihood ratio tests. Complete parameter estimates are provided in Appendix C.

It is found that those who believe that they enjoy higher levels of discretion are more likely to assume the role of steward of public interest. Administrators were more likely to be categorized as a steward of public interest than adapted realist (- 1.034, p<.01), businesslike utilitarian (-1.126, p<.01), resigned custodian (-1.249, p <.01) or practical idealist (-1.125, p<.01) with increased levels of discretion.

622 PAQ WINTER 2015

Table 6 Model Fit Information and Likelihood Ratio Tests

-2 Log

Likelihood Chi-Square Sig.

Intercept Only 1403.665

Final 1167.762 235.903 .000

Model Fitting Criteria

Likelihood Ratio Tests

-2 Log Likelihood of Reduced


Chi- Square Sig.

Intercept 1167.762a .000 . Expectation of policy involvement

1178.419 10.658 .031

Stakeholders’ expectations

1178.238 10.476 .033

Administrative discretion

1248.669 80.907 .000

Log tenure 1179.959 12.197 .016 Years in public service 1186.585 18.823 .001 Job satisfaction

1171.389 3.627 .459


1172.202 4.440 .350 Years in current position (log)

1169.809 2.047 .727


1172.604 4.843 .304 Political ideology

1173.548 5.786 .216

Minority association

1171.973 4.212 .378 Gender

1170.620 2.858 .582


1169.428 1.666 .797 The chi-square statistic is the difference in -2 log-likelihoods between the final model and a reduced model. The reduced model is formed by omitting an effect from the final model. The null hypothesis is that all parameters of that effect are 0. a. This reduced model is equivalent to the final model because omitting the effect does not increase the degrees of freedom.

PAQ WINTER 2015 623

It is also determined that individual expectation regarding one’s involvement in policy formulation is a powerful predictor of role assumption. Public administrators were more likely to be categorized as steward of public interest than adapted realist (-.492, p<.01), businesslike utilitarian (-.509, p<.05), resigned custodian (-.547, p <.01) or practical idealist (-.514, p<.01) with increased expectations to be involved in policy formulation. Finally, it is established that one’s tenure is significant in predicting one’s assumed role. With tenure, there is a decreasing probability that an administrator will assume the role of steward of public interest. With increased tenure, administrators were more likely to be categorized as adapted realist (1.518, p<.01), businesslike utilitarian (1.896, p<.01), resigned custodian (1.580, p <.01) or practical idealist (1.465, p<.01) than steward of public interest. Indeed, organizational tenure appears to be inversely related to representative type behavior. Taken together, the empirical results confirm the significance of all three hypothesized links as well as the direction of their relationships. From the standpoint of representative bureaucracy theory and future research the most intriguing results might, however, be located outside the marginal effects of the independent variables and with the performance of the control variables. Neither gender nor race achieved statistical significance. The latter might be explained by at least two important factors. First, it is possible that individual expectations and administrative discretion exert an impact on role assumption above and beyond one’s demographic background. Second, this might also suggest that gender and race are perhaps important predictors of role assumption within policy domains that directly deal with women or minority issues, respectively; on every day basis, however, individual characteristics might be less reliable predictors of administrative behaviors.

624 PAQ WINTER 2015

Limitations As it is the case with any research endeavor of this scale, this study, through its partially-exploratory nature, is not perfect. There were several important tradeoffs that needed to be made in order to frame the research within its desired course. Although these aspects are not expected to seriously challenge the validity of the findings, they should, nevertheless, be noted and accounted for when making any generalizations beyond the context of the study. Several of the assumptions that come with this type of research, which although typical, are still significant and could strongly influence the outcomes if not fully upheld. It is assumed that the survey instrument can accurately capture, transmit and collect data necessary for answering the research questions in a manner that was envisioned in the research design. It is also presumed that, which was confirmed by the high levels of internal consistency and completion rates, the structure and delivery of the instrument did not induce important levels of confusion and was understood well by most, if not all, respondents.7 The most important assumption that was made, however, is that the responses provided by the public administrators who participated in this research accurately reflect their actions in practice. Strictly speaking, their stated behaviors are, or at least closely mirror, their actual behaviors.8 Waldo (1980) has noted that the study of public administration, social science research in general, is value-laden. The values of those being studied as well as the unrecognized values of the researcher unavoidably seep into the research through the choice of the study design or the wording of the survey instrument. In this sense, then, the captured responses might actually reveal administrators’ values more than they echo their administrative behaviors and decision-making. Possible concerns regarding the representativeness of the sample provides another limitation that should be considered. Although the data were collected using a random sample of a national association and given the nature of the research design there are sufficient reasons to believe that the results have a high degree of generalizability, it should be noted that procurement

PAQ WINTER 2015 625

specialists, as a professional sub-group of public administrators, might be much more homogeneous in terms of their everyday behaviors and understandings than most public servants. As a result, the reliability and validity of the findings should be considered uncertain when significantly extended beyond the context of this study to other sub-groups of practitioners that might not share similar levels of homogeneity in terms of professional expectations.


The dynamics involved in the relationship between administrative discretion and representation are perhaps too complex to allow themselves to be easily modeled. Nevertheless, while the research presented here might miss on some important nuances that surely warrant empirical attention, it does provide a crude attempt to isolate the impact of administrative discretion on the administrative roles. It does so by linking administrative roles, as constructed through the representative bureaucracy lens, to the behavioral insights provided by role theory. The findings show that public administrators who perceive to have higher levels of discretion are more likely to assume the representation role of steward of public interest as the dominant role that guides their habitual decision-making. Similar to Sowa and Selden (2003), this study demonstrates that the presence of discretion is not a mere variable that needs to enter in the equation, but it might be at the center of explaining representative behavior. The study extends on what has been done before by moving outside policy and organizational specific constraints. The research design employed here allowed representation to be studied without consideration for the organization context or policy setting. This could be interpreted both as strength and as a possible limitation.

To understand the complete contribution that this study makes, one would have to broaden one’s perspective on the possible implications of its overall findings. Of particular interest is the fact that roles are to be sure powerful explanatory mechanisms. Once assumed and institutionalized, they might

626 PAQ WINTER 2015

indeed provide a behavioral map for public administrators. Kennedy (2013) has suggested that examining questions of representation through the use of roles and role perceptions has the potential to overcome many of current weakness and critiques of research on representative bureaucracy. Furthermore, while gender and race, might be important triggers for representation within policy matters that are directly related to these specific associations, the key to understanding everyday policy-independent representative behavior might lie with administrative discretion. In fact, it would not be much of an exaggeration to claim that active representation is reflected within the realm of discretion; to be more precise, public administrators without discretion cannot represent. This realization falls in line with recent scholarship (see Brandbury & Kellough, 2008; Kelly, 1998; Kennedy, 2013; Sowa & Selden, 2003; Selden et al., 1998; Selden et al., 1999; Slack, 2001) that looked beyond passive representation and suggested that variables and individual characteristics other than race and gender might be just as important, if not more, in conditioning representation.

There is obviously much that remains to be done here. At the very least, future research should challenge what was found by this study. Are the findings just a simple byproduct of the sampling frame and research design, or can they be located in other domains as well? In addition, while there is relatively little hope to conclusively decide the normative debate regarding administrative discretion, future research should look into uncovering the accepted “normative intervals” of administrative discretion. At what point and under what conditions does legitimate active representation transition into partiality? Or is the former even a pre-condition for the latter? Can we delineate what represent “appropriate” levels of administrative discretion? These are just a few of numerous intriguing and exciting questions about administrative representation that remain to be addressed. As this study has shown, administrative discretion has a significant, traceable and direct effect on representative behavior. Within this context, then, there are few, if any grounds,

PAQ WINTER 2015 627

for future research on representation and administrative behaviors to avoid accounting for it.

REFERENCES Abney, G., & Lauth, T. P. (1985). Interest group influence in city

policy-making: The views of administrators. The Western Political Quarterly, 38(1), 148-161.

Bales, R. F. (1950). Interaction process analysis: A method for the study of small groups. Cambridge, MA: Addison- Wesley.

Biddle, B. J. (1979). Role theory: Expectations identities and behaviors. New York: Academic Press.

Biddle, B. J. (1986).Recent developments in role theory. Annual Review of Sociology, 12(1), 67-92.

Box, R. C. (1992). The administrator as trustee of the public interest. Administration & Society, 24(3), 323-345.

Bradbury, M. D., & Kellough, J. E. (2008). Representative bureaucracy: Exploring the potential for active representation in local government. Journal of Public Administration Research and Theory, 18(4), 697-714.

Bradbury, M., & Kellough, E. J. (2011). Representative bureaucracy: Assessing the evidence on active representation. The American Review of Public Administration, 41(2), 157-167.

Cottrell, L. S. (1942). The adjustment of the individual to his age and sex roles. American Sociological Review, 7(5), 617- 620.

De Graaf, G. (2011). The loyalties of top public administrators. Journal of Public Administration Research and Theory, 21(2), 285-306.

628 PAQ WINTER 2015

Denhardt, R. B., & deLeon, L. (1995). Great thinkers in personnel management. In J. Rabin, T. Vocino, W. B. Hildreth, & G. J. Miller (Eds.), Handbook of public personnel administration (pp. 21-41). New York: Marcel Dekker.

Diggs, S. N., & Roman, A. V. (2012). Understanding and tracing accountability in the public procurement process. Public Performance & Management Review, 36(2), 290-315.

DiMaggio, P., & Powell, W. (Eds.) (1991). The new institutionalism in organizational analysis. Chicago, IL: University of Chicago Press.

Dolan, J. (2000). The senior executive services: Gender, attitudes, and representative bureaucracy. Journal of Public Administration Research and Theory, 10(3), 513-529.

Downs, A. (1967). Inside bureaucracy. Boston, MA: Little, Brown and Company.

Emmett, S., & Wright, P. M. (2012). Excellence in public sector procurement: How to control costs and add value. Cambridge, UK: Cambridge Academic.

Ewing, D. (1977). Freedom inside the organization: Bringing civil liberties to the work place. New York, NY: Dutton.

Finer, H. (1941). Administrative responsibility in democratic government. Public Administration Review, 1(4), 335–350.

Friedrich, C.J. (1940). Public policy and the nature of administrative responsibility. In C. Friedrich & E.S. Mason (Eds.), Public policy (pp. 3–24). Cambridge: Harvard University Press.

Frederickson, G. H., & Smith. K. B. (2003). The public administration primer (1st ed.). Boulder, CO: Westview Press.

Fry, B. R., & Nigro, L. G. (1996). Max Weber and US public administration: the administrator as neutral servant. Journal of Management History, 2(1), 37-46.

Gliner, J. A., Morgan, G. A., & Leech, N. L. (2009). Research methods in applied settings: An integrated approach to design and analysis (2nd Ed.). New York, NY: Routledge.

PAQ WINTER 2015 629

Hair, J. F., Black, W, C., Babin, B.J., & Anderson, R. E. (2010). Multivariate data analysis (7th ed.). Upper Saddle River, NJ: Prentice Hall.

Hassett, W. L., & Watson, D. J. (2002). Long–serving city managers: Practical application of the academic literature. Public Administration Review, 62(5), 622-629.

Herbert, A. W. (1974). The minority administrator: Problems, prospects, and challenges. Public Administration Review, 34(6), 556-563.

Hickson, D. J. (1966). A convergence in organization theory. Administrative Science Quarterly, 11(2), 224-237.

Hindera, J. J. (1993). Representative bureaucracy: Further evidence of active representation in the EEOC district offices. Journal of Public Administration Research and Theory, 3(4), 415-429.

Hite, J. M., Williams, E. J., Sterling, C. H., & Baugh, S. C. (2006). The role of administrator characteristics on perceptions of innovativeness among public school administrators. Education and Urban Society, 38(2), 160- 187.

Hollander, E. P. (1985). Leadership and power. In C. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed., pp. 485-537). New York: Random House.

Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression. New York, NY: John Wiley and Sons.

Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, J., & Rosenthal, R. A. (1964). Organizational stress: Studies in role conflict and ambiguity. New York: Wiley.

Karnig, A. K., & McClain, P. D. (Eds.) (1988). Urban minority administrators: Politics, policy and style. New York: Greenwood Press.

Kathlene, L. (1995). Alternative views of crime: Legislative policymaking in gender terms. Journal of Politics, 67(3), 696-723.

Katz, D. & Kahn, R. L. (1966). The social psychology of organizations. New York: Wiley.

630 PAQ WINTER 2015

Kelly, R. M. (1998). An inclusive democratic polity, representative bureaucracies, and the new public management. Public Administration Review, 58(8), 201- 208.

Kennedy, B. A. (2013). Sorting through: The role of representation in bureaucracy. Journal of Public Administration Research and Theory, 23(4), 791-816.

Kingsley, J. D. (1944). Representative bureaucracy. Yellow Springs, OH: Antioch Press.

Kleinbaum, D. G., & Klein, M. (2002). Logistic regression: A self-learning text. New York, NY: Springer.

Krislov, S. (1974). Representative bureaucracy. Englewood Cliffs, NJ: Prentice Hall.

Kuhn, A. (1974). The logic of social systems. San Francisco, CA: Jossey-Bass.

Kwak, C., & Clayton-Matthews, A. (2002). Multinomial logistic regression. Nursing Research, 51(6), 404-410.

Levitan, D. M. (1946). The responsibility of administrative officials in a democratic society. Political Science Quarterly, 61(4), 562-598.

Levy, M. J. (1952). The structure of society. Princeton, N.J.: Princeton University Press.

Liao, T. F. (1999). Interpreting probability models: Logit, probit, and other generalized linear models. Thousand Oaks, CA: Sage Publications.

Lim, H. (2006). Representative bureaucracy: Rethinking substantive effects and active representation. Public Administration Review, 66(2), 193-204.

Linton, R. (1945). The cultural background of personality. New York: Appleton-Century.

Lipsky, M. (1980). Street-level bureaucracy: Dilemmas of the individual in public services. New York: Russell Sage Foundation.

Long, N. E. (1952). Bureaucracy and constitutionalism. American Political Science Review, 46(3), 808-818.

Loveridge, R. O. (1968). The city manager in legislative politics: A collision of role conceptions. Polity, 1(2), 213-236.

PAQ WINTER 2015 631

Meier, K. J. (1975). Representative bureaucracy: An empirical analysis. American Political Science Review, 69(2), 526- 542.

Meier, K. J. (1993a). Latinos and representative bureaucracy: Testing the Thompson and Henderson hypotheses. Journal of Public Administration Research and Theory, 3(4), 393- 414.

Meier, K.J. (1993b). Representative bureaucracy: A theoretical and empirical exposition. In J. L. Perry (Ed.), Research in public administration (pp. 1-35). Greenwich, CT: JAI.

Meier, K. J., & Bohte, J. (2001). Structure and discretion: Missing links in representative bureaucracy. Journal of Public Administration Research and Theory, 11(4), 455- 470.

Meier, K. J., & Stewart, J. Jr. (1992). The impact of representative bureaucracies: Educational systems and public policies. American Review of Public Administration, 22(3), 157-171.

Merton, R. K. (1949). Social theory and social structure: Toward the codification of theory and research. Glencoe: The Free Press.

Meyer, J.W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83(2), 340-363.

Mosher, F. C. (1968). Democracy and the public service. New York: Oxford University Press.

Murray, S., Terry, L.D., Washington, C. A., & Keller, L. F. (1994). The role demands and dilemmas of minority public administrators: The Herbert thesis revisited. Public Administration Review, 54(5), 409-417.

Nalbandian, J. (1989). The contemporary role of city managers. American Review of Public Administration, 19(4), 261-278.

Newcomb, T. M. (1950). Social psychology. New York: Dryden. O'Kelly, C., & Dubnick, M. J. (2006). Taking tough choices

seriously: Public administration and individual moral agency. Journal of Public Administration Research and Theory, 16(3), 393-415.

632 PAQ WINTER 2015

Palumbo, D. J. (1969). Power and role specificity in organization theory. Public Administration Review, 29(3), 237-248.

Parsons, T. (1951). The social system. Glencoe, IL: The Free Press.

Parsons, T. (1964). Essays in sociological theory (revised edition). New York, NY: Free Press.

Pitkin, H. F. (1967). The concept of representation. Berkeley, CA: University of California Press.

Pitts, D. W. (2007). Representative bureaucracy, ethnicity, and public schools: examining the link between representation and performance. Administration & Society, 39(4), 497-526.

Saltzstein, G. H. (1979). Representative bureaucracy and bureaucratic responsibility: Problems and prospects. Administration & Society, 10(4), 465-475.

Slack, J. (2000). Zero-sum politics, the Herbert thesis, and the Ryan White CARE Act: Lessons learned from the local side of AIDS. Journal of Health and Human Services Administration, 24(1), 80-102.

Svara, J. H. (1990). Official leadership in the city: Patterns of conflict and cooperation. New York, NY: Oxford University Press.

Ramos, A. G. (1981). The new science of organizations: A reconceptualization of the wealth of nations. Toronto: University of Toronto Press.

Roman, A. V. (2014). Counterbalancing perspectives on the current telos of American bureaucracy. Administration & Society, 46(7), 825-852

Roman, A. V. (2013a). The politics of bounded procurement: Purists, brokers and the politics-procurement dichotomy. Journal of Public Procurement, 13(4), 447-475.

Roman, A. V. (2013b). The “mental revolution” of the public procurement specialist: Achieving transformative impacts within the context of e-procurement. In N. Pomazalová (Ed.), Public sector transformation processes and internet public procurement: Decision support systems (pp. 1-16). Hershey, PA: IGI Global.

Rosenbloom, D. H., & Kinnard, D. (1977). Bureaucratic representation and bureaucratic behavior: An exploratory

PAQ WINTER 2015 633

analysis. Midwest Review of Public Administration, 11(1), 35-42.

Scott, P. G. (1997). Assessing determinants of bureaucratic discretion: An experiment in street-level decision making. Journal of Public Administration Research and Theory, 7(1), 35-57.

Sieber, S. D. (1974). Toward a theory of role accumulation. American Sociological Review, 39(4), 567-578.

Selden, S. C. (1997). The promise of a representative bureaucracy: Diversity and responsiveness in a government agency. Armonk, NY: M.E. Sharpe.

Selden, C. S., Brewer, G. A., & Brudney, J. L. (1999). Reconciling competing values in public administration: Understanding the administrative role concept. Administration & Society, 31(2), 171-204.

Selden, S. C., Brudney, J. L., & Kellough, E. J. (1998). Bureaucracy as a representative institution: Toward a reconciliation of bureaucratic government and democratic theory. American Journal of Political Science, 42(3), 717- 744.

Sowa, J. E., & Selden, C. S. (2003). Administrative discretion and active representation: An expansion of the theory of representative bureaucracy. Public Administration Review, 63(3), 700-710.

Strodbeck, F. F., James, R. M., & Hawkins, C. (1957). Social status in jury deliberations. American Sociological Review, 22(6), 713-719

Strodbeck, F. F., & Mann, R. D. (1956). Sex role differentiation in jury deliberations. Sociometry, 19(1), 3-11.

Thai, K. V. (2001). Public procurement re-examined. Journal of Public Procurement, 1(1), 9-50.

Thai, K. V. (Ed.). (2008). International handbook of public procurement. Boca Raton, FL: CRC Press.

Thomas, S. (1994). How women legislate. New York: Oxford University Press.

Thompson, F. J. (1976). Minority groups in public bureaucracies. Administration & Society, 8(2), 201-226.

634 PAQ WINTER 2015

Van Riper, P. P. (1958). History of the United States civil service. Evanston, IL: Row, Peterson & Company.

Vroom, V. H. (1969). Industrial social psychology. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology (2nd ed.), (pp. 196-268). Reading, MA: Addison-Wesley.

Waldo, D. (1980). The enterprise of public administration: A summary view. San Francisco, CA: Chandler & Sharp Publishers.

Weick, K. E. (1995). Sensemaking in organizations. Thousand Oaks, CA: Sage.

Wilkins, V. M., & Keiser, L. R. (2006). Linking passive and active representation by gender: The case of child support agencies. Journal of Public Administration Research and Theory, 16(1), 87-102.

Wilson, W. (1966). The papers of Woodrow Wilson Vol. 5. Princeton, NJ: Princeton University Press.

Zimmermann, J. A. M., & Allen, L. R. (2009). Public recreation administration: An examination of the perceived roles of local recreation administrators. Administration & Society, 41(4), 470-502.

PAQ WINTER 2015 635

NOTES 1. It should be noted here that this perspective clashes with the classic image of the “ value neutral” bureaucrat (de Graaf, 2011) which was often suggested by students of public administration as an appropriate mechanism for normatively dealing with administrative discretion (see Fry & Nigro, 1996). But given that even neutrality is a value, the choice to remain neutral is in itself an application of administrative discretion (Selden et. al., 1999). 2. “Representation” is an ambiguous term and idea in its own right. While representation can be defined in a number of ways, the definition provided by Pitkin (1967) appears to be preferred by a number of scholars. Pitkin (1967, p. 209) defines representation as “acting in the interests of the represented in a manner responsive to them.” 3. For instance, agencies such as the Internal Revenue Service (IRS) or the Transportation Security Administration (TSA) are unlikely to foster environments that encourage their employees to act as stewards of public interest. In other instances, such as the Small Business Administration (SBA) or Equal Employment Opportunity Commission (EEOC), representation might be sanctioned though the agency’s mission or legislative setup. Locating representative behaviors within the latter type of agencies and administrative processes would be of limited value and perhaps somewhat misleading. There is much more value in examining representation in administrative environments that are not envisioned to encourage it. 4. Assuming that the population size is finite and is known to be 16,000, following Yamane (1967, p. 886) the needed sample size for statistical inference is given by the following formula:

Where N is the population size, n is the sample size, and e is the level of precision. At a 95% confidence level, the size of the needed sample was determined to be approximately 390. 5. The research, including its design and instrument, was reviewed and found to be compliant with Institutional Review Board’s (IRB)

636 PAQ WINTER 2015

requirements; it received exception status A3. The survey’s digital invitation was sent out on February 12th, 2013. Reminders were sent on February 26th and March 11th, 2013. A final reminder was sent on March 18th, 2013. The survey included a cover message within the body of the invitation email and a unique individual link to the survey embedded into the text. The invitation message introduced the potential participants to the scope of the research and informed the recipients that their participation was voluntary. It was made clear that no individual data will be made available and the results will be presented in aggregate terms only. 6. Only minor refinements were undertaken. For instance, in certain cases respondents used letters rather than numerical characters when indicating their responses. In other cases, respondents used months rather than years to indicate their tenure within a certain position or with a given organization. When months were used, the specified number of months was transformed in their year equivalent. Finally, in terms of their positions, respondents at times filled in their actual position rather than choosing among the options provided to them. In the latter cases, their answers were recoded according to the original options. For example, if a respondent indicated that he or she was employed as a “buyer,” he or she would be recoded as a “non- manager.” Given the fact that very few cases outside of the partially completed responses had missing values, no cases were deleted from the final data set. The final sample included 493 survey responses. 7. It is also assumed that the late respondents are representative of non- respondents, hence the non-response bias examination is valid, which would indeed confirm that there isn’t a significant non-response bias concern. Although an examination of the actual non-respondents would have provided an improved estimation, the latter was not possible because of logistical reasons and due to the constraints imposed on access to the membership contact list. 8. As suggested by one of the anonymous reviewers, the instrument’s language, despite the efforts undertaken to simplify it, might still have induced some bias due to the presence of a number of buzz words. A particular concern might be associated with the statements provided for question three, which were used by respondents to self-select within a specific role. The survey fatigue that some of the respondents might have been feeling at that point might have led some to be less careful or detailed in the examination of the provided statements. Although the

PAQ WINTER 2015 637

statements’ order was automatically randomized for each respondent, this still remains something that needs to be considered as a possible methodological limitation. Appendix A. The Independent and Control Variables Independent

Variables Motivation/general explanations

Full or partial discussions found in:

Perception regarding one's levels of discretion

The perception that a public servant has regarding the levels of discretion he or she possesses influences one's behaviors and the type of role assumed.

Kennedy (2013), Selden (1997), Selden et al. (1998), Meier & Bohte (2001), Sowa & Selden (2003)

Years with current organization - tenure

Scholars often find representation and tenure to be intimately linked. On the one hand with increased tenure administrators, due to being indoctrinated with organizational values, might be less likely to engage in representation. On the other hand, with increased tenure public administrators are more likely to command significant levels of organizational power; hence they might be able to command the necessary discretion levels to engage in representation.

Selden (1997), Selden et al. (1998), Dolan (2000), Sowa & Selden (2003), Zimmermann & Allen (2009)

Expectations of others (regarding involvement in policy formulation)

“[R]oles are induced through the sharing of expectations for role behavior. Thus, those who exhibit the role are stimulated to do so because they learn what behaviors are expected of them, while others are stimulated through their own expectations to teach and enforce appropriate behaviors for those who are members for the position” (Biddle, 1987, p. 5). "Most versions of role theory presume that expectations are the major generators of roles, that expectations are learned through experience, and that persons are aware of the expectations they hold" (Biddle, 1986, p. 69). As such, it can be assumed that individuals who expect themselves to be active in policy formulation will be more likely to assume representation type roles.

Parsons (1964, 1951), Kahn et al. (1964), Loveridge (1968), Biddle (1986), Nalbandian (1989), Box (1992), Meier (1993b)

Self- Expectations (regarding involvement in policy formulation)

Parsons (1964, 1951), Loveridge (1968), Biddle (1986), Nalbandian (1989), Box (1992), Selden (1997), Selden et al. (1998)

Control Variables


One's position affects one's behavior. It should be noted that position and roles might overlap but they are not identical, "positions are classification of human

Kahn et al. (1964), Palumbo (1969), Kuhn (1974), Biddle (1979), Meier (1993), Dolan

638 PAQ WINTER 2015

beings; roles are classifications of behaviors" (Biddle, 1979, p. 93).

(2000), Hite et al. (2006), Pitts (2007)


Younger public administrators are often believed to behave somewhat different from more experienced ones.

Strodbeck & Mann (1956), Strodbeck et al. (1957), Rosenbloom & Kinnard (1977), Biddle (1979), Selden et al. (1998), Hite et al. (2006), Selden et al. (1999), Brudney et al. (2000)


A number of scholars have suggested gender as an important factor in determining one's assumed role. For instance, it is believed that women with access to policymaking are more likely to be active on feminist matters when ""policy solutions fail to account for women's unique needs" (Dolan, 2000, p. 514). Scholars have argued that women and men assume different leadership roles, attribute different rankings to issues and might undertake importantly different behaviors in similar contexts (Thomas, 1994; Kathlene, 1995; Dolan, 2000).

Strodbeck & Mann (1956), Strodbeck et al. (1957); Biddle (1979); Karnig & McClain (1988), Thomas (1994), Kathlene (1995), Selden et al. (1998), Brudney et al. (2000), Dolan (2000), Sowa & Selden (2003), Hite et al. (2006), Wilkins & Keiser (2006), Zimmermann & Allen (2009)


One's ethnic background is believed to shape one's administrative role. Minority administrators are often expected to be more involved in terms of policy advocacy and more likely to engage in representation.

Herbert (1974), Biddle (1979), Murray et al., (1994), Selden (1997), Selden et al. (1998), Selden et al. (1998), Selden et al. (1999), Brudney et al. (2000), Sowa & Selden (2003)


Scholars believe that education plays an important role in shaping identities and as a consequence one's consequent behaviors.

Loveridge (1968), Palumbo (1969), Rosenbloom & Kinnard (1977), Biddle (1979), Selden et al. (1998), Selden et al. (1999), Zimmermann & Allen (2009)

Political ideology/ party identification

It is suggested that in the public sector, one's political ideology might influence the pattern of behavior within a given role.

Selden et al. (1998), Selden et al. (1999), Wilkins & Keiser (2006)

Years in government service

The experience in the public sector is believed to influence one's outlook on one's role as well as one's typical behavior.

Selden et al. (1999)

PAQ WINTER 2015 639

Job satisfaction Scholars assume that there is an association between breadth of roles and job satisfaction.

Vroom (1969)

Experience/role specialization/ years in current position

As an individual is exposed to same context and expectation he or she will develop a role specialization, but focusing on one specific role in that context.

Bales (1950); Biddle (1979); Selden et al. (1999); Hite et al. (2006)

640 PAQ WINTER 2015

Appendix B. Survey Instrument PAGE – 1 Directions: In what follows, you are going to be provided with several statements. Please read each statement carefully and indicate whether the statement is representative of your everyday experience. Please indicate whether you agree or disagree with the statement on a scale of 1 to 7. where: 1-strongly disagree 2-disagree 3-somewhat disagree 4-neither disagree nor agree, 5-somewhat agree 6-agree 7-strongly agree 1. Please indicate whether you agree or disagree with the following

statements: I use organizational channels to advocate for policy positions that I find to be important. I seek opportunities to participate in the formulation of public policy issues that I find to

be important. I encourage procedures that support greater public access to programs and services. As an administrator, I encourage certain values over others. In my work, I try to balance fairness and efficiency concerns. I am committed to management objectives. In my work I attempt to reflect most current managerial perspectives. To survive in the organization, I follow the rules when strictly necessary. In my decisions, I give priority to efficiency over fairness. I believe efficiency is the most important goal in my work, even if my supervisors do not

agree. Regardless of political pressure, I take the decision which is best for my organization. If it is not the most efficient choice, I do not advance the interests of minority citizens. I behave according to the wishes of my superiors. I follow the rules as closely as possible. I do not assume a public leadership role in policy issues. In my work, I try to be as neutral as possible. I primarily implement policy, not formulate it. I keep politics out of my decision-making. I am committed to my professional standards. I attempt to be as responsive as possible. PAGE – 2 2. Please, indicate whether you agree or disagree with the following statements: If I would have to choose, I would choose to work in this organization again. I expect to be involved in policy formulation. I constantly receive feedback in terms of what is expected of me. I believe that my organization's stakeholders expect me to be involved in policy formulation. I believe that other professionals in positions and organizations similar to mine are expected to be involved in policy formulation.

PAQ WINTER 2015 641

I feel that I enjoy a high level of discretion in terms of my decision-making. I could face penalties if I get involved in policy formulation. I am satisfied with my job. PAGE – 3 3. Please read the five job role descriptions and select the one that accurately captures your current role: Please read the five job role descriptions and select the one that accurately captures your current role. -On my job, I balance equity and fairness with individual concerns. I am committed to both management and organization. I work within system constraints, rules and proper lines of authority. -On my job, I advocate for policy positions and participate in the formulation of policy. I serve the public and I attempt to further the public interest. I use my experience to delineate what is in the best interest of the public. -On my job, I am as efficient as possible. Sometimes I challenge what more senior agency officials tell me to do if it is not best for the organization. I keep my promises and I advance the interest of the minority citizens only if it is the most efficient thing to do. -On my job, I am a neutral agent. I work within the boundaries imposed by rules, regulations and expectations of those more senior than me and to elected officials. I don't play a mediator role. I complete the task set for me and I do not participate in policy formulation. -On my job, I am as professional as possible. I work efficiently, quickly and accurately. When possible, I advocate for policy positions and legislation. I support equity and I am not neutral, but I do not act as an agent of elected officials. PAGE – 4 4. Where do you work? Federal government State/provincial government County/regional government City / town government School system College / university Special authority / district Other (please specify) 5. What position category best describes you? Non-manager Manager Senior executive/director Elected official Other (please specify) 6. How many years have you been in your current position? 7. How many years have you been with your current organization? 8. How many years have you been employed in the public sector? 9. What is the population size of the community that you serve?

642 PAQ WINTER 2015

10. Please indicate your highest level of education Less than high school degree High school degree or equivalent (e.g., GED) Some college but no degree Associate degree Bachelor degree Graduate degree 11. Which of the following best describes the field in which you received your highest degree? Public Administration Business Administration Political Science Economics Mathematics Science Healthcare Medicine Computing Engineering Technology Other (please specify) 12. How would you describe your political ideology? Strongly liberal Liberal Moderate Conservative Strongly conservative 13. Do you consider yourself a minority? Yes No 14. What is your race? (mark one or more) White Black or African American Asian Native Hawaiian or Other Pacific Islander American Indian or Alaska Native Other 15. What is your gender? Female Male 16. What is your age?

Thank you so much for taking the survey! Your input is highly valuable.

PAQ WINTER 2015 643

Appendix C. Parameter Estimates

Rolea B

Std. Error

Sig. Exp(B)

Adapted Realist

Intercept 7.138 1.858 .000 Expectation of policy involvement

-.492 .175 .005 .611

Stakeholders’ expectations .248 .168 .140 1.282

Administrative discretion -1.034 .164 .000 .356

Current job satisfaction .178 .131 .172 1.195

Position .295 .259 .254 1.343 Log (years in current position)

-.640 .511 .211 .527

Log (tenure) 1.518 .493 .002 4.563

Years in public service -.066 .021 .001 .936

Education -.084 .151 .580 .920

Political ideology -.202 .182 .268 .817

Age .002 .021 .906 1.002

[Minority association =1] .094 .388 .809 1.099

[Minority association =2] 0 b . . .

[Gender =1] .325 .334 .331 1.384

[Gender =2] 0 b . . .

Businesslike Utilitarian

Intercept 10.394 2.500 .000 Expectation of policy involvement

-.509 .234 .029 .601

Stakeholders’ expectations .273 .229 .233 1.315

Administrative discretion -1.126 .196 .000 .324

Current job satisfaction .119 .183 .516 1.126

Position -.400 .433 .356 .670 Log (years in current position)

-.654 .768 .394 .520

Log (tenure) 1.896 .784 .016 6.662

Years in public service -.072 .034 .031 .930

Education -.076 .218 .726 .926

Political ideology -.548 .274 .046 .578

Age -.040 .030 .179 .961

[Minority association =1] .613 .519 .237 1.846

[Minority association =2] 0 b . . .

[Gender =1] -.114 .488 .815 .892

[Gender =2] 0 b . . .

Resigned Custodian

Intercept 12.447 2.181 .000 Expectation of policy involvement

-.547 .198 .006 .578

Stakeholders’ expectations -.192 .194 .321 .825

644 PAQ WINTER 2015

Administrative discretion -1.249 .178 .000 .287

Current job satisfaction .279 .154 .070 1.321

Position -.082 .359 .819 .921 Log (years in current position)

-.474 .655 .469 .622

Log (tenure) 1.580 .659 .017 4.857

Years in public service -.070 .027 .009 .933

Education -.344 .176 .051 .709

Political ideology -.444 .238 .062 .642

Age -.019 .026 .455 .981

[Minority association =1] -.163 .474 .730 .849

[Minority association =2] 0 b . . .

[Gender =1] .128 .420 .761 1.136

[Gender =2] 0 b . . .

Practical Idealist

Intercept 9.493 1.899 .000 Expectation of policy involvement

-.514 .175 .003 .598

Stakeholders’ expectations .073 .169 .666 1.076

Administrative discretion -1.125 .166 .000 .325

Current job satisfaction .143 .131 .273 1.154

Position .057 .275 .836 1.059 Log (years in current position)

-.262 .542 .628 .769

Log (tenure) 1.465 .529 .006 4.330

Years in public service -.089 .022 .000 .915

Education -.154 .154 .315 .857

Political ideology -.182 .193 .346 .834

Age .005 .021 .833 1.005

[Minority association =1] -.054 .403 .892 .947

[Minority association =2] 0 b . . .

[Gender =1] .115 .347 .741 1.122

[Gender =2] 0 b . . .

Copyright of Public Administration Quarterly is the property of Southern Public Administration Education Foundation and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.