Resumé

It is argued that the role of the public bureaucracy must be analysed along at least two dimensions. One dimension concerns the relationship between politicians and bureaucrats, i.e. the possible conflict between political loyalty and autonomy. The other dimension concerns the relationship between citizens and bureaucracy, i.e. the possible conflict between openness towards citizens and bureaucratic closedness. With the aid of data from a survey of 407 agency heads in Norwegian municipalities, it was possible to group different bureaucrats along the two dimensions. Four categories emerged from the analysis; a political bureaucrat, an autonomous bureaucrat, the "classic" administrator and the "linking pin". An attempt to predict group membership was made using variables on the societal, organizational as well as individual levels. This attempt was only partially successful, indicating that there is a need for other independent variables to explain group membership.

Introduction

The role of the bureaucracy has always been of central interest for students of public policy. In modern societies public bureaucracy occupies a central position in the preparation and implementation of public policy. Public policy can be seen as policy carried out by large, formal organizations (Egeberg 1994).

This article departs from the notion that public bureaucrats will always exert a great deal of influence on public policy. Bureaucracy shapes public policy by preparing political decisions, thereby limiting the alternatives and consequences among which politicians can choose, and by implementing political decisions. Implementation is never politically neutral. Political decisions are often open-ended, vague and ambiguous; making it necessary for the bureaucrats to use discretion in order to be able to implement practical solutions based on political decisions (Olsen 1983).

If we accept that public managers/bureaucrats use discretion in both the preparation and implementation of public policy, we have to think of the bureaucrats as a distinct political force. Public policy must then be analysed from a political as well as an administrative perspective. Understanding what signals bureaucrats emphasize, with which groups they identify and which premises they find important or unimportant is as essential to understanding

public policy as are political decisions or the public "demand" for
public services and goods.

Studies on the role of the public bureaucracy have usually focused on the division of labour between politicians and bureaucrats. The possible conflict between neutrality and political loyalty and bureaucratic autonomy are recurrent topics in the contemporary literature on administrative behaviour (Aberbach, Putnam & Rockman 1981; Gruber 1987; Jacobsen 1960; Lægreid & Olsen 1978; Suleiman 1984; Svara 1985 & 1991; Weber 1922/ 1992; Wilson 1887/1992). During recent decades another dimension has been put on the agenda; the possible conflict between bureaucratic and political autonomy versus openness towards citizens (Blubaugh 1987; Denhardt 1989; Ingraham & Rosenbloom 1989, 1990; Long 1992; Miller 1987; Nalbadian 1989; Peters 1987). This dimension has been highlighted by the impact on the profession of what is known as "new public management", an ideological trend or school focusing on efficiency and effectiveness, responsiveness towards citizens, "closeness to the customer" and so forth (Downs & Larkey 1986; Metcalfe & Richards 1987; McKevitt & Lawton 1994; Willcocks & Harrow 1992).

In this article it is argued that the two dimensions can be characterized by three bureaucratic roles. The first role is that of the "classic bureaucrat" as outlined by Weber (1922/1992) in his essay on the bureaucratic structure of government and in Woodrow Wilson's (1887/1992) discussion on the organizing of American government. In these articles, the role of the bureaucrat is defined as politically loyal, emphasizing neutrality and cues given by elected politicians. The second role is to some degree the opposite of the loyal bureaucrat. Here the focus is on the autonomy of public bureaucrats, and this ideal type can be referred to as the autonomous bureaucrat. The third role emphasizes the link between the bureaucracy and the citizens, and raises the question to what extent bureaucrats respond to cues from citizens in the community. We will refer to this role as that of the citizens' advocate.

As already intimated, these three roles can be viewed as ideal types. The next question is whether these roles are mutually exclusive, or if it is possible to imagine all these roles being balanced by the bureaucrat. Previous empirical studies have tentatively concluded that there is no clearcut line between the three roles (Aberbach, Putnam & Rockman 1981; Christensen 1991; Svara 1985, 1991). Different role combinations are then discussed.

The next step is to view bureaucratic roles as dependent variables. The hypothesis is that different factors, associated with personal, organizational and societal attributes, can help us understand why some bureaucrats are more market-orientated, less politically loyal, etc.

Using data from a survey of public managers in Norwegian municipalities

we are trying to ascertain (a) to what extent the roles outlined above
actually exist, and (b) to what extent variation in these roles can be
explained.

Bureaucratic Roles - A Short Background

The role of the public bureaucracy in the making of public policy may be seen as an interaction between three central groups; elected politicians, employed bureaucrats and citizens. In this article the relations between citizens and politicians are not discussed. Of particular interest to us is the interaction between the bureaucracy on the one hand, and politicians and citizens on the other. The relation between politicians and bureaucrats is twofold: First, the politicians are dependent on having a bureaucratic apparatus capable of preparing and implementing political decisions. Second, the bureaucrats depend on having politicians capable of actually taking decisions and of monitoring and controlling the implementation of these very decisions. As far as the citizens are concerned the bureaucracy is one of the central service-providers, and the citizens constitute the clientele, user, consumers or what we can call the "market" for public services and goods.

This article sets out to determine the roles with which public bureaucrats identify. A role can be defined as "the perception of bureaucratic norms and role enactment" (Christensen 1991, 303). We are thus discussing the attitudes of the bureaucrats toward central norms.

Three central bureaucratic norms can be discerned. First, we have a role emphasizing loyalty towards politicians, and stressing the ideal of the neutral bureaucrat. The second role stresses bureaucratic autonomy and a certain scepticism towards politicians and citizens alike. Third, there is a role emphasizing the needs and wishes of the citizens.

The Loyal Bureaucrat - The Political Instrument

The thought that the bureaucracy should act as a neutral instrument for political leaders was elaborated to some extent by Woodrow Wilson (1887/ 1992), but is largely attributed to Max Weber in his essays on "the bureaucratic structure" (1922/1992). According to Weber, public administration should be bureaucratically organized, meaning specialization of tasks, extensive use of formalized rules and procedures, strong separation of organizational and individual control of resources, emphasis on formal education and lifelong careers. The "rulers" (i.e. politicians) are responsible for goal-setting, balancing different ideals and values and elaborating rules

and procedures that would secure a "technically efficient" implementation
of goals.

This view on bureaucrats is based on several rather unrealistic assumptions. One central assumption is that the bureaucrats do not pursue goals and preferences of their own. They will implement every decision loyally, even if they disagree with the content of the decision (Lægreid & Olsen 1978). If we relax this assumption, a "technical" view on the bureaucracy assumes that rules and procedures are clear-cut and easy to understand for all, making the use of discretion unnecessary or impossible. This assumes perfect rationality among the rulers, that they are capable of making rules that include every alternative, every consequence of every alternative and that all cases can be subordinated to a distinct rule (Simon 1957). Perfect rationality would mean perfect monitoring of bureaucrats.

Both these assumptions are clearly unrealistic. Is it then impossible to speak about a politically loyal bureaucrat? No, not if the intentions of the rulers are followed closely. In practice this would mean a close connection between bureaucrats and rulers, where the bureaucrats are trying to formulate and implement political decisions which are as close to the political intentions as possible.

The Autonomous Bureaucrat - The Bureaucrat as Policy-Maker

The problems connected with perfect rationality have led many theorists to conclude that public bureaucracy is unmanageable (see, for instance, Lipsky 1980). Increasing quantity and complexity in the area of public responsibility, increases the size of bureaucracy and the degree of professionalization which in turn makes political monitoring even more difficult. The information gap between politicians and bureaucrats has increased. The complexity in public policy has led to more extensive use of professionals in the public bureaucracy. As noted by Scott (1966, 269), the professionalization of the bureaucracy results in severe monitoring problems:

(1) the professional's resistance to bureaucratic rules, (2) the professional's rejection of
bureaucratic standards, (3) the professional's resistance to bureaucratic supervision; and
(4) the professional's conditional loyalty to the bureaucracy.

The professional bureaucrat will often emerge as an autonomous actor, trying to use professional instead of political premises when taking decisions. A professional will often be sceptical towards political decisions, arguing that they often are based on inadequate information and incomplete knowledge. They will also, to a certain extent, be sceptical towards citizens; arguing that citizens often do not know what is in their own best interests.

Professional autonomy becomes one of the most important aspects of the
public bureaucracy (Mintzberg 1979).

More modern economic theories about bureaucracy argue that the bureaucrats will try to become more autonomous so that they can maximize their own interests (Niskanen 1971). Incomplete information makes it possible for the bureaucrat to hide important pieces of information, and to use this hidden information to further personal interests. Economic theory refers to this as the problem of "moral hazard" (Milgrom & Roberts 1992). The more autonomous the bureaucrats become, i.e. the more they are sheltered from pressures from both politicians and citizens, the greater the likelihood that they can maximize own interests.

Whatever the motives of the bureaucrats might be, professional interest or more narrow personal interests, the consequence is the same - more and more autonomous bureaucratic actors. In practice we will assume that this kind of bureaucrat will try to buffer himself from politicians and users, and to put more emphasis on professional values and arguments.

The Citizens' Advocate — The Democratic Bureaucrat

Recent research on public policy questions the democratic nature of political as well as bureaucratic systems. Elected politicians often emerge as narrow elites, not representative of the public at large (Dahl 1961). Bureaucrats seem to be more interested in doing a professionally interesting job than in meeting the needs of the public (Gruber 1987).

But a totally different view on the bureaucracy can be applied, that of the bureaucrat acting in the interest of the citizen. When acknowledging that the bureaucracy is a political force because it always shapes political decisions in one way or another, it becomes important to ask to what extent the bureaucrats act as advocates for the citizens (Aberbach, Putnam & Rockman 1981). A third ideal role emerges.

This role emphasizes that bureaucrats are in frequent contact with the users of the services they provide and receive inputs from them in the form of decision premises, complaints, and so on. The citizens' advocate will try to meet the needs of the public, even if they run counter to political decisions or professional values and ideals.

Bureaucratic reforms, designed to make bureaucracy more democratic, have in fact been a constant fixture on the political agenda during the last few years. Several Western democracies have set out to create a more representative bureaucracy, for instance by recruiting more women, blacks, other professions, etc. (Ingraham & Rosenbloom 1989); and some have discussed the possibility of establishing a particular public culture, emphasizing democratic ideals and values (Denhardt 1989). Yet another school has adopted ideals from management in the private sector, arguing that

public bureaucracies should be "close to the customer" (Peters & Waterman 1982). These ideals play a role in most of the "new public management" theories. Public bureaucrats should get closer to the customer by increasing the degree of direct participation in the decision-making process (Harrow & Shaw 1992), by using market research to unravel citizens demands (Downs & Larkey 1986) and by focusing more on external results than on internal operations (Metcalfe & Richards 1987).

This new trend in public bureaucracy, by some called "consumerism" (Willcocks & Harrow 1992), implies that the bureaucrat tries to act in the interests of the citizen. Signals from the "market" will be more important than political or professional cues. A "citizens' advocate" may even dissociate himself from political and professional arenas on the assumption that politicians only represent a certain elite and that professionals are more interested in doing an interesting job rather than in providing the services that consumers need and demand.

The Ideal Roles - Mutually Exclusive or Complex Interplay?

The discussion in the last chapter would seem to suggest that the three ideal roles are at odds with one another. Empirically, this seems rather unlikely. (Aberbach, Putnam & Rockman 1981, Aberbach & Rockman 1988). Aberbach and his associates argue that these roles represent an evolution in the view on public bureaucracies. The strong Weberian distinction between politics and administration (what they call "Image I"), was replaced by a more complex view where both politicians and bureaucrats were seen to exert political influence (what they call "Image II"). Image II stresses the autonomy and administrative rationality of bureaucrats, as opposed to a more political rationality among politicians. The next step in the evolution stressed the close link between bureaucracy and "consumers" (especially interest organizations) on the assumption that bureaucrats also act as mediators between such interests (what they call "Image III"). The main point made by Aberbach et al. is that these three roles are intertwined in such a way as to produce a bureaucratic role which is both politically loyal, professionally autonomous and sensitive to citizens' needs and wishes. They call this the "pure hybrid", or "Image IV". The bureaucrat becomes a politician.

Other empirical studies have also concluded that the three ideal roles seem to interact rather than conflict. In a study of Norwegian central state bureaucrats, Christensen (1991) argues that there is little or no conflict between professional and political cues. Relying on several studies of American local and federal bureaucrats, Svara (1985, 1991) points out

that there is a duality between politicians and administrators, and that bureaucrats are both politically loyal and professionally autonomous. According to Peters (1987) politicians and bureaucrats eventually become socialized into the same belief systems, thereby eliminating the possibility of conflict between the two groups. The politicians become more and more "bureaucratized" and the bureaucrats more and more "politicized". Miller (1987), Kearney & Sinha (1988) and Streib (1992) conclude that there is no empirical evidence of conflict between professional, political and democratic values among bureaucrats. On the contrary, emphasis on professional values seems to increase the emphasis on democratic values.

These empirical findings imply that the ideal types may not be sufficient in order to account for the complex roles of public bureaucrats. We argue that the bureaucratic roles must be viewed along at least two separate dimensions. The first dimension concerns the relationship between bureaucrats and politicians. This dimension illustrates the possible conflict between political loyalty and neutrality on the one extreme, and the autonomous ideal on the other. The second dimension concerns the relationship between bureaucrats and citizens. On this dimension we run into the possible conflict between openness towards citizens vs. a more closed, classic bureaucratic position. Most empirical research focuses on one dimension alone: there has been less effort to try and combine the two.

Constant or Variable Roles?

So far we have discussed the roles as if they were constant. It is possible to apply a more dynamic perspective on bureaucratic roles. It can be argued that the role affiliation is to some degree determined by other variables; and in order to explain variance in role perception and role affiliation we have to identify at least some of the independent variables.

We start by identifying independent variables on three different levels. Variation in role perception and role affiliation can be explained by focusing on different societal bases for bureaucracies. Heady (1966) argues that differences in regime type (democratic vs. totalitarian) can account for variations in role affiliations, but the empirical evidence for this hypothesis is somewhat shaky (Heper, Kim & Pai 1980). Aberbach et al. (1981) claim that national culture and history can explain different bureaucratic roles - a claim partially corroborated by empirical evidence. This view is supported by Suleiman (1984). Narrowing the focus to more national variables, we hypothesize that the size (both physically and in terms of number of people) of the social unit under public authority will influence adherence to different bureaucratic roles. The smaller the social unit, the closer the contact with citizens, perhaps to the point of increasing the likelihood of an open

and responsive system, sensitive to cues from citizens and reducing the
likelihood of a closed system, where the bureaucrats are encouraged to
shelter themselves from the citizens.

The variance may also be accounted for by organizational differences. Different organizational sectors will be of different political interest. In sectors associated with strong political interest, the loyal bureaucrat will be more likely, while lack of political interest may open up prospects for more autonomous and citizen-oriented roles. Also, the longer a bureaucrat has been part of the organization, the more likely it is that he/she has become socialized into the existing political values. Agency heads recruited externally will probably be less socialized than heads recruited internally (Pfeffer 1994). Socialization will probably lead to a convergence between politicians and bureaucrats (Peters 1987), making the politically loyal role more likely. The greater the possibility to discipline the bureaucrats (by giving them prospects of higher wages, higher level jobs, and the fewer external possibilities they have), the better the politicians can manage the bureaucrats (Pfeffer & Salancik 1978). The sheer size of the organization can reduce the possibilities for political control and influence, thereby providing more room for an autonomous role. Size can also increase the distance between leaders and citizens. The degree of professionalization in different sectors (i.e. a homogeneous formal education as a criterion to getting a certain job) can also account for variation with respect to role affiliation.

The last group of independent variables is associated with individual differences among bureaucrats. Variables such as age, sex and education may be important (Christensen 1991). In addition, we are also interested in the bureaucrat's connection to the community. It is possible that the autonomous role is easier to choose if one does not have tight links with the community. A hypothesis could be that the closer the links between bureaucrat and community, the more likely he/she is to choose a role close to the "citizens' advocate".

The Survey

The focus of this survey is public, middle-level managers in Norwegian municipalities. Norway is divided into 448 municipalities, (ranging in size from a little more than 500 inhabitants to approximately 460,000), and 19 counties. Total municipal expenditures represent approximately 17 percent of total public expenditures, but account for 64 percent of public employment. The latter figure is larger than the corresponding figures for most other European countries, excluding the Scandinavian countries (Page & Goldsmith 1987). It is a by-product of the fact that the municipalities have

the main responsibility for the provision of labour-intensive public services. Certain services are obligatory and must by law be provided by the local authorities. This includes basic education (grades 1-9), health (excluding hospitals which are administered by the regional county level) and social services. Most local authorities also have a technical sector (waste and garbage collection, road maintenance, etc) and a cultural sector. In terms of expenditure, social services represent approximately 20 percent, education 19 percent, health 16 percent and culture/church 5 percent. The remaining 40 percent is accounted for by central administration, housing, business activities and "miscellaneous". This gives a slightly biased picture, because while the three sectors education, health and social services represent approximately 55 percent of total expenditures, they represent approximately 80 percent of municipality employment.

Traditionally, local authorities are administratively organized in four sectors - education, health/social services, technical services and culture. Each sector is headed by a "sector leader", or agency head. Political organization has largely followed the administrative division of labour. All local authorities are obliged to have an executive (kommunestyre), an "elite" political forum that presents cases to the council (formannskap) and a chief administrative officer (rådmann). Other political and administrative organs are largely voluntary, although there are some political fora specified in different, specific laws. There are some variations in the political and administrative organization of the municipalities, but the main picture is as outlined above. A new law concerning local authorities was passed in 1993 which paves the way for more political and administrative diversity, but it is still very much in an experimental stage.

The Sample

The selection of respondents was made in two steps. The population of municipalities was stratified by size (number of inhabitants). Almost 80 percent of the municipalities are smaller than 20,000 inhabitants. We then selected at random 130 municipalities, with a slight överrepresentation of larger municipalities (above 20,000 inhabitants), i.e. 9 percent in the population versus 20 percent in the sample. In each of these 130 municipalities we sent a questionnaire to all four agency heads. Data were collected between November 1993 and January 1994. The total population was 520 out of whom 407 responded, giving a response rate of 78 percent, which was deemed highly satisfactory. When checking for biases, we found no systematic biases regarding sector, municipality size, geographic region, sector or sex. The sample seems to be highly representative.

Some of the differences between the sectors are well worth commenting
upon at this stage. The educational and technical sectors have the oldest

agency heads with the longest tenure in the office as well as in the municipality. Both the educational and the technical sectors are dominated by semi-professional educations (3-4 years or more, i.e. teachers and engineers) and there is very little by way of variation, especially within the technical sector. The educational sector is strongly dominated by internal recruitment, which is the predominant form of recruitment in all four sectors. The between-sector variation is marginal as regards years of education, which clearly is not a relevant indicator of the degree of professionalization.

Data Analysis

In an attempt to tap the respondents' role perception we asked them to what extent they agreed or disagreed on a total of 19 statements. As a rule, the questions/statements were taken from other, empirical studies of bureaucratic roles (Streib 1992; Laegreid & Olsen 1978; Egeberg 1984; Christensen 1991; Edwards, Nalbadian & Wedel 1981; Svara 1991). The items are listed in Appendix 1. On theoretical grounds, seven of the items were expected to tap the dimension previously referred to as "openness towards citizens", while the remaining twelve items were seen as indicators of the dimension which we have identified as "political loyalty versus autonomy''. A stepwise factor analysis was carried out in order to check the validity of these assumptions. As a result eleven of the original items were excluded, because they were not valid and tapped dimensions other than the two of interest here.¹ The remaining eight items fell clearly into a two-factor solution explaining 41 percent of the total variance. Three items measure the attitude towards increased direct, citizen participation, an important component in the dimension "openness towards citizens". The other five items measure the dimension "political loyalty-autonomy".² The distribution on the eight items is presented in Table 1.

Interestingly enough, only two items show rather uniform results. Almost all the respondents agree with the propositions that "increased participation will make public policy better" and that one "should act as an administrator". The remaining items display a great deal of variance, indicating a high degree of heterogeneity among bureaucrats. The items measuring attitudes towards increased participation and attitudes towards loyalty/ neutrality are not significantly correlated. This may serve as an indication that there is nothing or little by way of perceived conflict between these different roles which would go towards corroborating the findings reported by Christensen (1991) and Streib (1992).

We then proceeded with an analysis designed to identify the different
bureaucratic roles. Assuming that it is possible for a respondent to score
high or low on both dimensions, or high on one dimension and low on the

other we opted for cluster analysis. With cluster analysis, respondents are grouped into different clusters by one or several variables. Basically, the technique produces groups of highly similar entities (Aldenderfer & Blashfield 1984). In this study we use the two dimensions ("attitudes towards citizen participation" and "political loyalty /autonomy") as a starting-point in an attempt to classify different groups of bureaucrats in the sample. The eight items (Table 1) were collapsed into two indices. Two straightforward additive indices were computed. The index measuring "attitude towards citizen participation" consists of the sum of items Y1?Y_1? Y₂, Y3Y₃ divided by 3 (the number of items), and the index measuring "political loyalty/autonomy" is defined as the sum of items Y₄, Y₅, Y₆, Y₇, Y₈ divided by 5.

A hierarchical cluster analysis suggested the presence of four different groups in the sample. Cluster analyses on various subsamples confirmed the existence of such a pattern.³ Table 2 demonstrates how the four groups score on the two dimensions.

There are significant differences between the four groups on the two dimensions (citizen participation: F= 221.0560, prob = 0.000; loyalty/ autonomy: F = 247.6811, prob = 0.000). We are, therefore, well advised to identify what the four groups/clusters stand for:

Cluster 1: "The political bureaucrat". This is a group of bureaucrats who are favourably disposed towards increased direct citizen participation, and who seem to strive for increased autonomy from politicians. This group of bureaucrats appears to be closely linked to the notion of a citizens' advocate (Denhardt 1989) responding to cues from citizens rather than to political signals. Thirty-four percent of the respondents belong to this group.

Cluster 2: "The autonomous bureaucrat", being sceptical towards increased citizen participation, and seeking autonomy from politicians. The bureaucrats within this category seem to shelter themselves from politicians and citizens alike. This category accounts for 22 percent of the respondents.

Cluster 3: "The classic administrator". The members of this group are sceptical towards increased participation, but seem to be more politically loyal than the total sample mean. This bureaucratic type is closely related to the Weberian ideal. It constitutes 29 percent of the total sample.

Cluster 4: "The linking pin". The members of this group are favourably
disposed towards increased participation and stand out as more
politically loyal than the total sample mean. This type of

bureaucrat can be hypothesized to function as a transmitter of signals from citizens to politicians (and vice versa), but without taking a personal stand in different situations. It accounts for 15 percent of the total sample.

But are these valid clusters? An indication is to use external variables to check for validity in the clustering (Hair et al. 1992). Using questions on who the respondents felt were important discussion partners and which groups they felt should be more important in the decision process we searched for correlation between group membership and these variables.⁴ The "autonomous bureaucrat" gave less emphasis on discussion with citizens than the other groups (cor. = —0.11, sig. LE 0.05), while the linking pin emphasized this more than the other groups (cor. = 0.10, sig. LE 0.05). The political bureaucrats are more inclined than the others to say that individual citizens should be more important in the decision process (cor. = 0.12, sig. LE 0.05). The autonomous bureaucrat would like the central government to be more important (cor. = 0.11, sig. LE 0.05), while the linking pin would assign more weight to organized interests in the municipality (cor. = 0.17, sig. LE 0.01). All these results are as predicted, indicating that the clusters are indeed valid groupings of respondents.

The data suggest that the political bureaucrat and the classic administrator
are the most common bureaucratic types. The linking pin seems to
a be a less common type.

The last step in the analysis is to explain group membership. Initially, we focused on three types of variables: (1) municipality size, (2) organizational variables, i.e. sector and potential for discipline and socialization,⁵ and (3) individual variables, i.e. sex, age, years and type of education and links to community (grown up in the municipality, years lived in the municipality, family in the municipality). The dependent variable is group membership. Group membership was transformed into four dummy variables where 1 denotes a specific group membership, while 0 denotes the other three groups. Since the dependent variable is a dichotomy, logistic regression was chosen (Aldrich & Nelson 1984). Four separate equations were computed, one for each dependent variable. The results are reported in Appendix 2.

The attempt to predict group membership gave mixed results. To a certain extent we were successful in predicting group membership in two groups: the autonomous bureaucrat and the "linking pin", but the —2 log likelihood ratio is not significant. Membership in the other two groups is not predicted very well at all. All attempts to explain group membership suffer from a rather low fit between the model and the actual results. If we look at some of the single, significant correlations we can, however, find some interesting results. The probability of finding a political bureaucrat

seems to be slightly lower in the technical sector than in other sectors, and in organizations with a high potential for discipline. This indicates that organizations can create more loyal bureaucrats by using different mechanisms for discipline (Lægreid & Olsen 1978). The autonomous bureaucrat seems to be slightly less tightly linked to the community. Bureaucrats who have grown up in the municipality they work in are less likely to adhere to an autonomous ideal. The probability of finding a classic administrator seems to be higher in organizations with a high potential for discipline. This ties in well with the findings reported for the political bureaucrat. Lastly, there is less likelihood of finding the linking pin among teachers and among bureaucrats tightly linked to the community. Teachers are the most sceptical towards both citizens and politicians, perhaps because of extensive experimentation with increased citizen participation (i.e. schools managed by parents) and political pressure towards reductions in the school sector brought out by the declining number of children of school age. Tight links to the community seem to make the position of the "neutral" linking pins difficult. Perhaps this role demands a more detached position from the community. Finally, we see that this role seems to thrive among highly socialized bureaucrats, bureaucrats who have learned to balance different demands and wishes from different groups (Christensen 1991).

Discussion and Conclusion

This article set out with two objectives. The first was to analyse empirically bureaucratic roles along two dimensions: political neutrality/loyalty vs. bureaucratic autonomy, and openness vs. autonomy from citizens. The second objective was to account for bureaucratic role perception by using a set of individual, organizational and societal variables.

The survey data at hand made it possible to divide the individual bureaucrats into four, not three as originally anticipated, distinct groups with different role perceptions and affiliations. The four groups can be placed along two dimensions, as shown in Fig. 1.

What we have called the "classic administrator" is part and parcel of the discourse about the Weberian bureaucratic ideal. This group combines political loyalty with autonomy from direct citizen influence. The "linking pin" could readily be associated with what we have referred to as the "citizens' advocate". Here we find a group of persons actively involved with politicians and citizens alike.

The "autonomous bureaucrat" calls for some modifications of the original assumptions. Empirically, we find two types of autonomy. The first, the "political bureaucrat", strives for autonomy from politicians, but not from citizens. This role could be rather close to a professional ideal if we accept

the notion that professionals are "(• • ) honoured servants of public need, conceiving of them as occupations especially distinguished from others by their orientation to serving the needs of the public (. . .)." (Freidson 1983, 19). Political autonomy may even be necessary in order for the bureaucrat to satisfy such citizens' needs. The professional must be able to adjust, sometimes rather quickly, to specific individual needs. This local adjustment is not easy to combine with political monitoring. The second kind of bureaucrat, the "autonomous bureaucrat", seems to be closely related to economic theories about bureaucracy. The members of this group try to protect themselves from politicians as well as citizens, perhaps because "they enjoy autonomy, because their lives are easier if they are their own masters, because they feel they know best" (Gruber 1987, 87). In this sense, they would seem to meet the criteria of a rather egoistic type of bureaucrat.

This study also underlines the importance of analysing bureaucratic roles along several dimensions. Nordic studies have generally focused on the loyalty-autonomy dimension (Lundquist 1993; Lægreid & Olsen 1978; Christensen 1991), while recent Anglo-American research tends to emphasize the citizen dimension (Aberbach, Putnam & Rockman 1981; Greene 1982; Streib 1992). This difference is possibly due to the fact that many Nordic studies have studied central government, while the Anglo-American

studies seem more preoccupied with the local and regional government. We argue that the closer a bureaucrat is to service provision and the closer he/she is to direct citizen influence, the more important the citizen dimension will become. Perhaps the citizen dimension will be more important in studies of local and regional government than in studies of central government.

This study indicates that it is both theoretically meaningful, and empirically possible to analyse bureaucratic roles along both dimensions. But it has, nevertheless, only scratched the surface of the complex reality of public bureaucrats. In real life, these bureaucrats have to balance expectations from other groups than politicians and citizens. Media, corporate interests, central government, professional organizations, all place some expectations on the bureaucrats, all want the bureaucrat to emphasize different roles. In future studies, it would be interesting to see how bureaucrats balance these role expectations with the two dimensions discussed in this article.

Another important conclusion can also be derived: There is not just one, perhaps not even a predominant, role in public bureaucracies. We will find several different roles, possibly existing side by side within the same organization. In this study, we have tried to explain role perceptions and role affiliation using a set of stable and structural variables. The hypothesis is that certain structural arrangements (either social, organizational or individual) will favour one of the roles over the others. This hypothesis gets very limited support from the data at hand. Other explanations also have to be considered. Two possibilities seem particularly interesting. First of all, we could include "personality" as a variable. Different personalities and capabilities can perhaps explain why some people choose one role over the other. This would pull in favour of more psychologically based studies, perhaps using more classic leadership theories (Strand 1987). Second, it is possible that one and the same bureaucrat can take on different roles in different situations. The study reported here is static, and records only what role the respondents felt affiliated with at the time they filled out the questionnaire. It has been argued that public bureaucrats have to live with several, and often conflicting roles (Jacobsen 1960). Later studies would benefit greatly from including information on how bureaucrats make priorities between different roles in specific situations.

---

NOTES

1. First, a general check for Measuring Sampling Adequacy (MSA) was conducted. Three items (Y₆, Y_g, Yl9)Y₁₉) had a MSA below 0.50, and were excluded (Kim & Mueller 1978). The subsequent factor analysis resulted in six factors. Two of these dimensions were of lesser interest to this study. One dimension measured the general perception of citizens (Y₃), the other the general perception of politicians (Y₉, YlO,Y₁₀, Y_n, Yl2).Y₁₂). Two other factors gave very unclear results (i.e. high and significant factor loadings on several factors). This led to the exclusion of Y₇, Y_n, Yl4.Y₁₄.

---

1. First, a general check for Measuring Sampling Adequacy (MSA) was conducted. Three items (Y₆, Y_g, Yl9)Y₁₉) had a MSA below 0.50, and were excluded (Kim & Mueller 1978). The subsequent factor analysis resulted in six factors. Two of these dimensions were of lesser interest to this study. One dimension measured the general perception of citizens (Y₃), the other the general perception of politicians (Y₉, YlO,Y₁₀, Y_n, Yl2).Y₁₂). Two other factors gave very unclear results (i.e. high and significant factor loadings on several factors). This led to the exclusion of Y₇, Y_n, Yl4.Y₁₄.

2. The factor scores for the seven items were: Factor 1 Factor 2 (Y,) -0.005 -0.751 (Y₂) 0.078 0.730 (Y₃) 0.012 0.666 (Y₄) 0.527 0.173 (Y₅) 0.511 0.113 (Y₆) 0.628 -0.032 (Y₇) 0.628 -0.054 (Y₈) 0.571 -0.045 Rotated factor solution (Varimax), Principal Component extraction. Kaiser-Meyer-Olkin measure of sampling adequacy = 0.62 Bartletts' test of sphericity = 199.74, sig. = 0.000

3. The hierarchical cluster analysis (squared Euclidian distance) resulted in the following "fusion coefficients" for the last seven clusters: Number of Fusion clusters coefficient change % change 7 2.01 6 2.14 0.13 7% 5 2.27 0.13 6% 4 2.61 0.34 15% 3 3.69 1.08 41% 2 4.43 0.74 20% 1 6.16 1.73 39% There is a significant leap from four to three clusters, indicating that at this point, two rather heterogeneous groups are fused (Hair et al. 1992). The sample was randomly split into two subsamples, and the same analysis was produced for both, giving almost identical results compared to the total sample. This indicates stability in the solution.

4. In this analysis group membership was recoded into four dichotomous variables, contrasting one group against all other respondents. The variables measuring importance of discussion partners have a range from 1 to 5, while the variables measuring which groups should be more important range from 1 to 4.

5. A variable indicating an organization's discipline potential was created, basedon three items tapping the respondents' perceptions of their chances of getting salary hikes or promotions in the near future in their current place of work and of their potential on the labour (easy or difficult to find a new job) as well as their commitment to their current place of work (whether they have any specific plans to quit). A high score denotes high discipline potential (both salary hikes and promotions are perceived possible; it is hard to find a new job and the respondent does not plan to quit). The variable ranges from 4 (low) to 12 (high). The organization's potential for socialization is measured by tenure in position and tenure in the municipality. The internal/external recruitment ratio is also an indication of the organization's potential for socialization (Pfeffer 1994).

REFERENCES

Aberbach, J. D., Putnam, R. D. & Rockman, B. A. 1981. Bureaucrats and Politicians in
Western Democracies. Cambridge, Mass.: Harvard University Press.

Aberbach, J. D. & Rockman, B. A. 1988. "Image IV Revisited: Executive and Political
Roles". Governance, Vol 1, No 1, 1-25.

Aldenderfer, M. S. & Blashfield, R. K. 1984. Cluster Analysis. Sage Series on Quantitative
Applications in the Social Sciences.

Aldrich, J. H. & Nelson, F. D. 1984. Linear Probability, Logitand ProbitModels. Sage Series
on Quantitative Applications in the Social Sciences.

Blubaugh, D. A. 1987. "The Changing Role of the Public Manager". Personnel Management,
Vol 69, 7-10.

Christensen, T. 1991. Bureaucratic Roles: Political Loyalty and Professional Autonomy.
Scandinavian Political Studies, Vol 14, no. 4.

Dahl, R. A. 1961. Who Governs? Democracy and Power in an American City. New Haven:
Yale University Press.

Denhardt, K. G. 1989. The Management of Ideals: A Political Perspective on Ethics. Public
Administration Review, March/April, 187-193.

Downs, G. W. & Larkey, P. D. 1986. The Search for Government Efficiency. From Hubris
to Helplessness. Philadelphia: Temple University Press.

Edwards, J. T., Nalbandian, J. & Wedel, K. R. 1981. "Individual Values and Professional
Education: Implications for Practice and Education". Administration & Society, vol 13:
123-143.

Egeberg, M. 1984. Organisasjonsutforming i offentlig virksomhet. Oslo: Tano.

Egeberg, M. 1994. "Bridging the gap between Theory and Practice: The Case of Administrative
Policy". Governance, Vol 7, no 1 (pp. 83-98).

Freidson, E. 1983. "The Theory of Professions: State of the Art". In Dingwall, R. & Lewis,
P. (eds): The Sociology of the Professions. Basingstoke: MacMillan Press.

Greene, K. 1982. "Municipal Officers' Receptivity to Citizen's and Elected Officials' contacts".
Public Administration Review, 42, 346-353.

Gruber, J. E. 1987. Controlling Bureaucracies: Dilemmas in Governance. Berkeley: University
of California Press.

Hair, J. F., Anderson, R. E., Tatham, R. L. & Black, W. C. 1992. Multivariate Data Analysis.
New York: Maxwell Macmillan International Editions.

Harrow, J. & Shaw, M. 1992. "The Manager Faces the Consumer". In Willcocks, L. &
Harrow, J. (eds), 1992: Rediscovering Public Services Management. London: McGraw-Hill.

Heady, F. 1966. Public Administration: A Comparative Perspective. Englewood Cliffs, N. J.:
Prentice Hall.

Heper, M., Kim, C. L. & Pai, S.-T. 1980. "The Role of Bureaucracy and Regime Types. A
Comparative Study of Turkish and South Korean Higher Civil Servants". Administration
& Society, Vol 12: 137-157.

Ingraham, P. W. & Rosenbloom, D. H. 1989. "The New Public Personnel and the New Public
Service". Public Administration Review, March/April 1989, 116-126.

Ingraham, P. W. & Rosenbloom, D. H. 1990. "Political Foundations of the American Federal
Service: Rebuilding a Crumbling Base". Public Administration Review, March/April.

Jacobsen, K. D. 1960/1992. "Lojalitet, nøytralitet og faglig uavhengighet i sentraladministrasjonen".
In Egeberg, M. & Christensen, T. (red): Forvaltningskunnskap. Oslo:
Tano.

Kearney, R. C. & Sinha, C. 1988. "Professionalism and Bureaucratic Responsiveness: Conflict
or Compability?" Public Administration Review, Jan/Feb, 571-579.

Kim, J.-O. & Mueller, C. W. 1978. Introduction to Factor Analysis. London: Sage Series on
Quantitative Applications in the Social Sciences.

Lipsky, M. 1980. Street Level Bureaucracy: The Dilemmas of the Individual in Public Services.
New York: Russel Sage Foundation.

Long, N. E. 1992. "Seeking the Polity's Bottom Line: A Conceptual Note", Administration
& Society, 24, 107-114.

Lundqvist, L. 1993. Ämbetsman eller direktör: Förvaltningschefen i demokratin. Stockholm:
Nordstedts.

Lægreid, P. & Olsen, J. P. 1978. Byråkrati og beslutninger. Oslo: Universitetsforlaget.

McKevitt, D. & Lawton, A. 1994. Public Sector Management: Theory, Critique and Practice.
London: Sage.

Metcalfe, L. & Richards, S. 1987. Improving Public Management. European Institute of
Public Administration. London: Sage.

Milgrom, P. & Roberts, J. 1992. Economics, Organization and Management. Englewood
Cliffs, N.J.: Prentice Hall.

Miller, C. M. 1987. State Administrator Perceptions of the Policy Influence of Other Actors:
Is Less Better? Public Administration Review, May/June, 239-245.

Mintzberg, H. 1979. The Structuring of Organizations. Englewood Cliffs, N. J.: Prentice Hall.

Nalbadian, J. 1989. "The Contemporary Role of City Managers", American Review of Public
Administration, Vol 19, Number 4.

Niskanen, W. A. Jr. 1971. Bureaucracy and Representative Government. Aldine-Atherton;
Chicago.

Olsen, J. P. 1983. Organized Democracy: Political Institutions in a Welfare State - the Case
of Norway. Oslo: Universitetsforlaget.

Page, E. & Goldsmith, M., eds., 1987. Territorial Politics in Western Europe. London: Sage
Publications.

Peters, B. G. 1987. "Politicians and Bureaucrats in the Politics of Policy-Making", in Lane,
J.-E., ed., Bureaucracy and Public Choice. Sage Modern Politics Series, Vol. 15, London.

Peters, T. J. & Waterman, R. H. 1982. In Search of Excellence. New York: Harper & Row.

Pfeffer, J. 1994. "Competitive Advantage Through People", California Management Review,
Winter 1994, Vol. 36, no. 2, 9-28.

Pfeffer, J. & Salancik, G. R. 1978. The External Control of Organizations: A Resource
Dependence Perspective. New York: Harper & Row.

Scott, W. R. 1966. "Professionals in Organizations - Areas of Conflict", in Vollmer, H. M.
& Mills, D. L., eds., Professionalization. London: Prentice Hall.

Simon, H. 1957. Administrative Behavior. New York: The Free Press.

Strand, T. 1987. "Bureaucrats or Contingent Actors?", in Kooiman, J. & Eliassen, K. A.,
eds., Managing Public Organizations. London: Sage.

Streib, G. 1992. "Professional Skill and Support for Democratic Principles: The case of Local
Government Department Heads in Northerns Illinois", Administration & Society, 24, 22-40.

Suleiman, E. N., ed., 1984. Bureaucrats and Policy Making. New York: Holmes & Meier.

Svara, J. H. 1985. "Dichotomy and Duality: Reconceptualizing the Relationship Between
Policy and Administration in Council-Manager Cities", Public Administration Review, Jan/
Feb, 221-232.

Svara, J. H. 1991. "Council and Administrator Perspectives on the City Manager's Role.
Conflict, Divergence or Congruence", Administration and Society 23, 227-246.

Weber, M. 1992/1922. "Bureaucracy", in Shafritz, J. M. & Hyde, A. C, eds., Classics of
Public Administration. Oak Park IL.: Moore Publishing Company.

Willcocks, L. & Harrow, J, eds., 1992. Rediscovering Public Services Management. London:
McGraw-Hill.

Wilson, W. 1887/1992. "The Study of Administration", in Shafritz, J. M. & Hyde, A. C,
eds., Classics of Public Administration. Oak Park IL.: Moore Publishing Company.