This article focuses on bureaucratic roles, or more specifically, on the perception of bureaucratic norms and role enactment among civil servants in the Norwegian central administration.¹ Why is it interesting to analyze bureaucratic roles? Many studies emphasize that public bureaucracies and bureaucrats have increased their political influence in the postwar period (Olsen 1983a). The implications of this fact are of course dependent on how bureaucrats interpret and act in their roles, particularly with respect to political and professional norms.

A bureaucratic role is structurally and institutionally denned as a mixture of different elements, of which political and professional factors have been the most central ones historically (cf. Friedrich 1940; Finer 1941; Kaufman 1956; Jacobsen 1960). The need to focus upon both political control, hierarchy and rules on the one hand and the importance for public decisions of highly developed specialized knowledge on the other is in keeping with strong Weberian traditions. Of special interest has been how bureaucrats handle political and professional norms and values, and the relationship

* This study is part of a research project on the organization and decision-making behavior of the Norwegian central public administration. The project was sponsored by The Norwegian Research Council for Applied Social Science. This article is a revised version of a paper presented at IPSA Round Table, on "Modernization of the public sector; Dealing with problems of efficiency and legitimacy' held in Oslo. Norway. 24-26 August 1989. 1 express mv thanks to the discussant. Professor B. Ouv Peters, for valuable comments.

Resumé

Political leaders in many countries have experienced growing problems of capacity in the postwar period, a development leading to an increase in the delegation of authority to public administration and civil servants. This delegation of authority creates a significant potential for discretionary decision-making authority on the part of public bureaucrats. One way of studying how bureaucrats handle this situation is to focus upon bureaucratic roles. Bureaucratic roles traditionally contain both political and professional norms. This article discusses how these political and professional considerations can be defined and how these norms are balanced given the potential for conflict which is present. The discussion is illustrated by measuring the perception of bureaucratic norms and role enactment among civil servants in Norwegian ministries. Results indicate that civil servants appear to have few problems in attending to and balancing both political and professional role norms.

between these norms in discretionary decision-making. Such insight is of special importance for understanding modern political systems, since they generally are characterized by increasing freedom of discretion for civil servants in public decision-making processes.

There are in this connection several questions to be answered. First, what is really meant by political and professional norms and values in relation to the bureaucratic role? How are these concepts defined and applied in studies of bureaucratic roles? How can one measure the influence of these norms on role perception and enactment? Second, are political and professional considerations in conflict with one another, or are they integrated in the bureaucratic role? How do we measure the balancing of these norms?

The data on bureaucratic roles used in this article are relatively limited in scope, because they are a minor part of a larger survey focusing upon bureaucratic structure and behavior. These data, therefore, are primarily meant to have an illustrative function in relation to the main analytical questions discussed.

Studying Bureaucratic Roles

Studies of bureaucratic roles may be characterized along different dimensions. compare the roles of politicians and bureaucrats, while others analyze the role of civil servants in public organizations. The former studies analyze questions like:

Have they (bureaucrats and politicians) different priorities? Do they consider different criteria when making decisions? Do they regard public affairs and the process of policymaking Have they different world views? What do these differences, if any, imply for their relationships and for performance as policymakers? (Aberbach et al. 1981, 3)

The latter studies ask what the central norms and values in the bureaucratic
role are, including the relation between political and professional considerations
1973, 1977; Aberbach et al. 1988).

These two types of studies often discuss the same problems. The difference that the former studies are primarily comparative or inter-role oriented, while the latter include many of the same elements in an intrarole The analysis here is inspired by the latter type of inquiry.

There are, however, several ways of measuring political and professional norms and values in bureaucratic roles. One may, for instance, emphasize the opinions, beliefs and ideologies of civil servants concerning the public sector and the role of politicians and bureaucrats in public decision-making (Christoffersson et al. 1972; Putnam 1973, 1977; Mellbourn 1979; Anton 1980; Aberbach et al. 1988). The analysis reported here focuses upon

the decision premises and norms prevailing when civil servants make
discretionary decisions.2

The empirical analysis is based on two types of questions concerning bureaucratic roles.³ One type focuses upon actual decision problems the civil servants have experienced concerning political and professional elements in their role. The other type focuses upon the perception of decision-making criteria and problems connected with the handling of political and professional When operating with the latter type of question one may of course ask what is the relation between the civil servants' role perception and their actual decision-making behavior? Our answer is founded on Simon's theory of administrative behavior (Simon 1957; March & Simon 1958). Actors exercising discretionary decision-making behavior are forced, due to problems of cognition and attention constraints, to select and attend to a limited set of value and factual premises. These premises, reflected in their role perception, constitute their 'model of action', a model strongly influencing their actual decision-making behavior. The formal organizational structure (for instance formal goals, hierarchic position and formal role descriptions) and bureaucratic career considerations are supposed to shape the bureaucrats' model of action and decision behavior, working through mechanisms such as 'late' socialization and incentives (March & Olsen 1976; Scott 1981; Olsen 1983a; Egeberg 1987).

The present study deviates from most of the more recent studies of bureaucratic roles with respect to the theoretical significance of the empirical This implies new opportunities, but also creates some major limitations of comparison. In the work reported here, therefore, we build theoretically upon and discuss the major role concepts in these studies, but play down the comparison of empirical results.

Political Considerations Defined

In one view political leaders want bureaucrats to be politically loyal, to be 'instruments' for the existing government (Finer 1941; Jacobsen 1960, 232). This view is based on a classical Weberian distinction between politicians as policy-makers on the one hand and bureaucrats as administrators and implementors on the other (Putnam 1973; Aberbach et al. 1981, 4-6). According to this distinction, politicians rule and bureaucrats obey. In this context bureaucrats are mainly jurists with reponsibility for handling single cases on the basis of public laws and rules.

Another and more 'modern' version of political loyalty tells us that both politicians and bureaucrats participate in policy-making, but in different ways (Aberbach et al. 1981, 6-9). Politicians bring values and interests into the policy-making arena, while bureaucrats serve political leaders with

neutral expertise based on facts and knowledge, an obligation built into their formal roles and stressed through recruitment based on merit (Simon 1957; Jacobsen 1960, 233). The bureaucrat is an expert in the sense that he or she is means-end oriented. Politicians have to delegate public authority to civil servants because they have quantitative problems of load and qualitativeproblems cognition.

The principle of political loyalty must not be pushed too far, however. Demanding strong political loyalty from bureaucrats may eventually lead to problems with political legitimacy. Major changes and loss of stability in the civil service may occur when a new government comes into power, or conflicts created and independent expertise questioned because the bureaucrats have been too committed to certain political programs (Kaufman 1068).

A relaxation of political loyalty, on the one hand, may generate a more politically active type of bureaucrat. In this instance both politicians and civil servants are involved in politics, but from different perspectives (Aberbach 9-16). Politicians engage in articulating the broad, diffuse interests of unorganized groups based on ideological values and perspectives. by comparison, are more pragmatic, balancing special interests, trying to reach compromises, making incremental changes, and so forth. The decision-making behavior of both politicians and civil servants must in such circumstances be explained as negotiations between equal actors within institutional frames, connected to political segments. Reasons for such a development are the changing definition of political importance and possibilities of influence from bureaucrats in the implementation of policies (Olsen 1983a).

Putnam (1973, 260) summarizes this role in what he calls the 'political
bureaucrat':

. . . This sort of official operates with much more pluralistic conception of the public interest. He assumes that there can be legitimately differing interpretations of the public interest, even genuinely conflicting interests among different groups in society. He is therefore, both more aware of 'political realities' and more willing to treat political influences on policy-making as legitimate. He recognizes the need to bargain and compromise, yet at the same time he does not necessarily shrink from advocating and even fighting for his own preferred policies. . . the political bureaucrat sees the politician (instead) as a participant in a common game, one whose skills and immediate concerns may differ from his own, but whose ultimate values and objectives are similar. . .

Another kind of potential threat against political loyalty and control occurs when the distinction between politics and administration for all practical purposes disappears, the roles of politicians and bureaucrats strongly overlap, and there is exchange of personnel between the two groups (Aberbach et al. 1981, 16-20; Aberbach & Rockman 1988).

Political Considerations Measured

Measuring the political loyalty of civil servants, the influence of and problems to political norms, may be done in different ways. To begin with we may focus upon the attention structure among civil servants concerning various decision premises or criteria (March & Olsen 1976). How important are political decision criteria perceived to be among the bureaucrats compared to other criteria? Table 1 shows that nearly threequarters the respondents in the ministries emphasize that signals from political leaders are very important when making discretionary decisions. This set of decision premises is by far the most important one, so it is obvious that the political elements in the bureaucratic role really matter.

One thing is to stress the importance of political signals as such, another is to know or anticipate the content of these norms (Friedrich 1937). In many issues civil servants are delegated decision-making responsibility, but their freedom is limited and must be balanced against the needs of political leaders for relevant information and the making of politically important decisions. How problematic is anticipation of political signals for Norwegian bureaucrats?

Problems of political anticipation seem to be minor. Table 2 indicates that nearly 80 percent of the civil servants think it is very or relatively easy to anticipate which matters in their own issue area should be presented to political leaders. Minor problems with political anticipation may indicate a smooth functioning of mechanisms of socialization and discipline in public bureaucracies (Olsen 1983a). Civil servants have internalized the formal political norms and the incentive system supports the active use of these norms.

But do the political leaders also use more traditional hierarchical ways of controlling administrative behavior? Do they instruct and use their administrative leaders to control the subordinates in classical Weberian ways? Table 3 shows that one-half of the civil servants very or relatively seldom are corrected by their administrative superiors, whereas 41 percent say that it happens now and then. This mechanism of control seems therefore to be relatively weak. It is there to be used if necessary, however.

What are the relationships between structural and demographic variables and indicators of political norms in the bureaucratic role? Are there systematic variations in role perception and enactment? To answer these questions we will concentrate on a few independent variables supposed to cut across different countries and types of administrative systems, variables both theoretically discussed and empirically tested. These variables are hierarchical position, age, tenure and type of higher education.*

Aberbach et al. (1981, 225, 227) show that higher civil servants are more favorably inclined towards political aspects of the grey zone between politics and administration, and that frequent contact with political leaders seems to foster a more positive attitude towards the political elements of public administration. Table 4 shows a variation in a similar direction for Norwegian ⁵ People in leadership positions put more weight on political signals and think anticipation is easier than executive officers. The main reason for this seems to be a major difference in exposure to political signals, norms and practice.

Studies of the effects of age and tenure on bureaucratic role orientation are more divided in their results. According to cyclic thinking (life or organizational) older bureaucrats with many years of service may put more weight on political signals and have fewer problems of anticipation than younger bureaucrats coming directly from higher education. The presumption that this is a result of longer exposure and learning (Pfeffer 1983). On the other hand, younger civil servants seems to be characterized by attitudes connected to a political orientation (Putnam 1973, 279, 283, 285; Aberbach et al. 1981, 200-202). This is mainly believed to be a generational effect (Mayntz & Scharpf 1975, 60). Younger bureaucrats have been more exposed to democratic trends. It is also argued that age and tenure are positively correlated and that low tenure may foster less democratic cynicism than many years of service.

Table 4 gives generally few effects of age and tenure on the political elements of the bureaucratic role orientation. There is one exception to this: older civil servants have fewer problems of anticipation than younger ones, a result presumably due to exposure and learning effects.⁶

Type of higher education may also have an effect on bureaucratic role perception and enactment.⁷ Social scientists, and to some degree economists, in some countries to be more positive towards the political aspects of the bureaucratic role than natural scientists and jurists (Putnam 1977, 394-95; Aberbach et al. 1988, 5). Table 4 illustrates, however, that type of higher education is not connected to variations in role orientation in Norway.

What are the relationships between alternative measures of political
norms in the bureaucratic role? Findings displayed in Table 5 show first
that giving priority to political signals correlates positively with lack of

problems with political anticipation. Attention to political signals seems to make political anticipation easier, or, alternatively, minor problems with anticipation seem to foster more attention towards political signals. Second, attention and correction are not correlated. This means, for instance, that civil servants putting relatively less weight on political signals are not more frequently subjected to control by superiors than other bureaucrats.

Third, anticipation and correction are negatively correlated, i.e. civil servants having problems with political anticipation are more likely to be corrected by their superiors. This result reveals some important logic in the way political loyalty and control appear to function. If a civil servant is good at anticipating how to present issues to the political leaders, concerning both their number and content, he is given confidence and discretion by the leaders and is seldom corrected. On the other hand, problems with anticipation generate hierarchical control. This may also imply that less correction is a potential incentive for improving the civil servant's political anticipation.

Professional Norms Defined

Professional norms in the bureaucratic role may be defined in different ways.⁸ It was noted above that the civil servants formally must serve political leaders with their expertise, i.e. a situation in which professional norms are clearly subordinate to political considerations. But political leaders are also interested in civil servants having professional autonomy, thereby ensuring that public decisions are based on independent expertise (Jacobsen 1960). Professional autonomy as referred to here means that bureaucrats freely put forward their professional premises and opinions, even if these are contrary to the opinions and programs of political leaders.

Professional autonomy may in this sense be seen both as a central part
of or as a potential threat to and modification or relaxation of the principle
of political loyalty. Professional autonomy in the latter respect coincides

with the norm that bureaucrats also have to be apolitical or neutral to party politics, a potential instrument for any government (Jacobsen 1960, 233). From this perspective professional autonomy may safeguard the objectivity and permanence of the bureaucratic apparatus, facilitate the change of political leadership and continuous execution of public authority.

Professional autonomy may none the less imply some potential problems for political loyalty and control. Too much professional autonomy may, on the other hand, foster a technocratic attitude, i.e. create a social and political insensitivity among civil servants (Putnam 1973, 259, 289; Putnam 1977, 408; Aberbach et al. 1988, 2). On the other hand, the political type of bureaucrat mentioned above may also be seen as professionally autonomous, using his or her expertise for political purposes. It is important to underline that the 'political bureaucrat' may undermine both the principle political loyalty and the norms of professional autonomy. It is also conceivable, however, that a bureaucrat may want to be a more active policy-maker than stated in the formal role, and at the same time obey hierarchic political control.

Professional Norms Measured

For purposes of the analysis reported here, the importance of professional norms has been measured in three ways. First, the attention structure concerning professional norms was identified. Civil servants were asked about the importance of professional considerations as decision criteria. Fifty-seven percent of the bureaucrats say professional decision norms are very important, whereas 32 percent say that they are relatively important. Only political signals are more important when civil servants are making discretionary decisions, and demands from affected groups are far less important (see Table 1).

Second, an effort was made to measure how strongly the civil servants try to further their professionally based opinions within the hierarchic structure of the ministries. The logic of professional autonomy means that professional opinions and proposals shall float freely from bureaucrats to the political leaders. Low professional autonomy implies that political leaders limit this communication or that the experts exercise self-imposed restrictions concerning professional decision premises. Table 6 indicates that a clear majority of Norwegian civil servants would as a rule consider advancing a proposal, even if they knew it would evoke objections from their superior. It seems obvious from this that professional norms have an independent and strong position in the bureaucratic role.

Third, we tried to measure the impact of professional norms on policies.
A central aspect of professional autonomy must be that professional premisesactually

isesactuallyinfluence or are attended to in policies and programs. Table 7 shows that only 10 percent of the civil servants say that they very or relatively often prepare or implement policies they disagree with. Fifty-six percent on the other hand say that it happens now and then. These results may be interpreted as a confirmation of a relatively strong professional influence on public decision-making, but also as an indication of restricted professional influence in some issues and situations.

What are the effects of structural and demographic variables on the professional elements of the bureaucratic role? The basis for expectations here is not firm, especially because empirical results from different studies are somewhat ambiguous and often focus upon technocratic attitudes, and that is only one part of professional norms (Putnam 1977; Aberbach et al. 1988, 3, 13-15). The discussion here is therefore related to political norms.

Concerning hierarchical position, it is logical, based on a theory of exposure and influence, to expect that administrative leaders put less weight on professional norms in the bureaucratic role. This may imply that they focus less upon professional considerations, are more careful with controversialproposals less often prepare and implement policies they disagree with. Such expectations are not supported by findings displayed

in Table 8, however.9 There are no substantial differences between administratorsand
officers concerning professional norms.

The effect of age and tenure on professional norms may be connected to two alternative expectations. First, one can presuppose that younger bureaucrats with short tenure recruited directly from higher education will emphasize professional norms and values more than bureaucrats with long exposure, experience and influence. Alternatively, based on a generational or cyclic perspective, older bureaucrats with long tenure will put more weight on professional norms because they are more cynical concerning democratic political control and are more likely to trust their own professional As shown in Table 8, however, none of these alternatives are confirmed by the empirical results.

Does type of higher education have an effect on the perception of professional norms in the bureaucratic role? Do those with an education in natural science, because of the content of their education, place greater weight on professional values. Or, on the contrary, do social scientists and economists place less weight on professional norms? Again, according to Table 8 there are no such relationships between educational variables and role perceptions.

Again, one may also ask about the correlations between different measuresof norms. Findings in Table 9 show first that there is a positive correlation between placing weight on professional considerations and advancing controversial proposals. A professional orientation seems to encourage civil servants to emphasize their professional autonomy, or the other way round. Second, neither strong attention to professional norms, nor advancing controversial proposals are related to more conflict concerning the preparation and implementation of certain issues. There is.

in other words, no indication of negative sanctions from the political
leadership because civil servants emphasize their professional norms and
autonomy.

Balancing Political and Professional Considerations

It is quite common to define bureaucratic roles as a mixture of political and professional elements. Dogan (1975, 4) talks about top civil servants being Janus-faced; they are half-administrative and half-political. Campbell (1983, 299-300) defines the accountability of bureaucrats as a mixture of 'subjective' standards (of whom professional standards are important) and 'objective' standards (institutional obligations). Some studies of bureaucratic also emphasize that political and professional elements are in conflict, thus complicating decision-making behavior (Finer 1941). Others stress that bureaucratic reponsiveness is characterized by the integration of political responsiveness and professional expertise and values (Friedrich 1940; Jacobsen 1960).

One can also argue that the balance between political and professional considerations is changing over time. Historically, this balance is reflected in different ways of organizing public administration. In general, emphasis on political loyalty seems to foster professional administrative bodies close to the political leadership, while stressing professional autonomy appears to create a more independent professional administration (Christensen 1987). But political-administrative doctrines concerning these matters may also change over time. Kaufman (1956, 1059-1063, 1069) shows, for instance, that in American administrative history, neutral competence expressed institutionally through the merit system and independent administrative was supposed to threaten political control at one point in time, but somewhat later was regarded as safeguarding the political responsibility of bureaucrats.

Our perspective is that ambiguity and conflict concerning bureaucratic

norms and role enactment are both unavoidable and desirable (Jacobsen 1960, 243). Unavoidable, because the norms of political loyalty and expertiseare and institutionalized. The mixture of norms and values in the bureaucratic role is changing over time, partly as a result of conflict between political leaders and professional groups, partly as a result of a process of institutionalization. The design of a mixture of different norms and values is also conceivable (Peters 1987, 256), even in a version of consciously defined ambiguous roles (Hedberg et al. 1976; Suleiman 1984).

Conflicting norms and values, such as political loyalty and professional autonomy, must in some situations be ranked, but may for long periods live side-by-side. The ambiguities of norms and values in bureaucratic roles are not solely destructive, but may rather be acceptable and even desirable to those wanting both status quo or change (Egeberg 1987). More clear-cut roles may lead to consequences not intended by anyone. Our expectation, therefore, is that the potential and logical intra-role conflict between political and professional norms is moderately reflected in the bureaucrats role perception and enactment.

What in fact are the correlations between measures of political norms on the one hand and professional norms on the other? To begin with Table 10 shows that there is a weak positive and significant correlation between emphasizing political signals and professional considerations. But the underlying univariate distributions are very biased as shown in Table 1. so the important figure is that 81 percent of Norwegian civil servants answer that both political and professional signals are very or relatively important when they are making discretionary decisions. It seems natural to conclude that political and professional decision signals are perceived to be compatible. (1983, 304-305) shows much the same in analyzing the attitude to accountability among top civil servants in the USA. the UK and Canada. Individuals score high on both accountability towards political and administrative leaders on the one hand and professional standards on the other.

Other results in Table 10 also indicate that political and professional norms are compatible and that professional autonomy is an integrated part of the bureaucratic role. Paying attention to professional considerations and professional influence on policies correlates positively with the ability to anticipate political priorities. And civil servants advancing controversial proposals on a professional basis are less frequently corrected than bureaucrats a lower profile professionally. The strongest correlation (+0.41), in faet, is that between correction and professional influence on policies. Civil servants less frequently corrected reports that they seldom prepare or implement policies they disagree with. This means, on the other hånd, that there is a group of less integrated bureaucrats, experiencing rather often both corrections and conflicts. But this is a very small group, comprising less than 5 percent of all respondents.

Conclusion and Discussion

The purpose of this article has been twofold. First, different ways of defining political and professional norms in the bureaucratic role have been discussed. The different perspectives implied by the concepts of political loyalty and professional autonomy respectively have been summarized. Different perspectives regarding the balancing of these norms were also outlined, leading to the formulation of a perspective used in this analysis. In particular, it was expected that there would be an integration of these norms in the bureaucratic role and a moderate level of conflict in civil servants' perceptions of the potential intra-role conflict.

Second, an effort was made to measure civil servants' role perceptions and role enactment with respect to the influence of political and professional norms, and the balancing of these norms. In this instance data were used that are more directly connected to the bureaucrats discretionary decisionmaking than has been the case in former studies of bureaucratic roles, which are mainly preoccupied with more widely defined role elements concerning beliefs and ideology. Results obtained show a strong emphasis on both political and professional decision signals, and few signs of conflict between the political and professional norms in civil servants' role perception actual decision experience.

Norwegian civil servants seem to have a fine-tuned feeling for the norm of political loyalty or responsiveness, but do not hesitate to put forward issues that have a potential for conflict if they think there are good professional for doing so. Political leaders seem to accept professional autonomy and influence, but now and then remind the civil servants of their political 'duties' by using more direct hierarchical mechanisms of control.

The results also indicate that civil servants, in different ways, learn how to live with conflicting demands and a certain level of conflict and ambiguiity in their bureaucratic role. Bureaucrats seem to be partly 'pre-socialized' to bureaucratic norms through their education. A process of institutionalization the combination of the two main considerations they have to adjust to. Processes of socialization further convey central norms to the bureaucrats so that they can internalize the main values in the administrative milieu. Roles are formally defined through duties and rights, specifying political and professional constraints. And bureaucrats are taught how to distinguish personal opinions from formal roles through their administrative career and the existing incentive system.

Results presented here indicate a rather peaceful coexistence between politicians and civil servants. The growing influence of civil servants on public decision-making processes does not seem to have created much conflict, instability, crisis of legitimacy or problems with the execution of political control. We do not have survey data on politicians' role perceptions, an intensive study of the relationship between top politicians and top rank civil servants in Norway seems to confirm major results reported here (Eriksen 1988).

Third, the structural and demographic variables generally used for explaining variations in bureaucratic role perception and enactment seem to have moderate significance.¹⁰ General norms and values in the bureaucratic seem to be rather strong. This homogeneity is quite contrary to findings regarding the impact these independent variables appear to have on other types of individual, administrative behavior, such as innovative, turnover and budgetary behavior (Christensen & Egeberg 1989a; Christensen 1989b).

Are these findings valid for Norway or the Nordic countries only, or could they be generalized to other western political-administrative systems? This question is difficult to answer, because both the structural-institutional perspective and the 'narrower' data on bureaucratic roles used in this study differ from former studies of bureaucratic roles. There are, however, reasons to believe that some general structural and institutional mechanisms in public admininstration foster integration of political and professional considerations in the bureaucratic role. A further step in the direction indicated in this study would be to reveal more exactly the content of these mechanisms.

But one can also argue that results presented here reflect that the Norwegian and Nordic political systems more generally are rather unique in some respects. As noted in various studies, these countries are characterized a relatively moderate cleavage structure, mutual trust between politicians and bureaucrats and a consensus-oriented political culture (Olsen 1983b).

The present study gives a more general picture of the balancing of political and professional elements in the bureaucratic role in Norway. Further analyses, for instance through case studies, will help to establish how generalizable the findings reported are. Such studies will, among other things, aid in determining whether this balancing varies across different issue or policy areas and stages of the decision-making process.

---

NOTES

1. In keeping with Roos & Starke (1981), we make a distinction between role perception, i.e. how people interpret their roles, and role enactment, i.e. the actual role behavior.

2. Few studies explicitly focus upon both general opinions concerning politics and admininstration and more specific role attitudes towards political and professional considerations held by bureaucrats. But in one case Christofferson et al. (1972) did find a relatively weak positive correlation between positive attitudes of bureaucrats towards the state/public sector and attitudes stressing the importance of politicians when making administrative decisions.

3. The data presented are taken from a survey conducted within the Norwegian ministries in 1986. The survey is partly a follow-up to the one conducted by the Norwegian Power Study in 1976 (Lægreid & Olsen 1978; Olsen 1983a). The questionnaire was sent to all administrators (leaders) and executive officers in the central governmental ministries: 1185-72.4 percent- replied, 808 respondents are included in this analysis. These are civil servants claiming that they exercise administrative discretion, i.e. they are not guided by clear-cut rules in their decision-making. There are no substantial biases in the data concerning representativity.

4. A variable often used in such analyses is party affiliation (cf. Putnam 1973; Aberbach et al. 1988). Our survey did not contain this variable. If one compares the answers from respondents belonging to tråde unions and academic interest groups, however, there are small differences in role perception.

5. The independent variables in Table 4 are defined in the following way: Hierarchical position is dichotomized between administrators (leaders) - 47 percent - versus executive officers (lower level) - 53 percent. Age is divided into 10-year intervals ranging from "under 25 years" to "65 years and more". Tenure, i.e. years of service, is a continuous variable with values from 1 to 46 years. Type of higher education involves four categories depending in each case upon whether civil servants are educated in law, economics, social sciences and natural sciences or not. The dependent political variables are operationalized in the following way: Political signals is dichotomized respondents saying that political signals are very important (73 percent) versus all other answers (27 percent). Anticipation is dichotomized between respondents emphasizing that it is very easy to anticipate which matters to present to the political leadership (19 percent) versus all other answers (81 percent). Finally, correction is dichotomized between respondents reporting that it is very uncommon to get their proposals corrected by their superiors (19 percent) versus all other answers (81 percent).

6. Hierarchical position, age and tenure exhibit strong positive correlations and are therefore somewhat problematic to differentiate in such analyses.

7. One problem with this variable in a comparative perspective is that seemingly identical types of education may vary in their content between countrics.

8. A profession is normally defined by characteristics such as a high educational level, the monopoly of certain types of jobs, public certification, etc. By professions in public administration in Norway is traditionally meant civil servants having an expertise of a means-end character (Jacobsen 1977). This is an expertise that is traditionally connected to discretionary decision-making behavior. Jurists have an education not primarily connected to means-end thinking, but a lot of jurists in faet have jobs characterized by such thinking.

---

8. A profession is normally defined by characteristics such as a high educational level, the monopoly of certain types of jobs, public certification, etc. By professions in public administration in Norway is traditionally meant civil servants having an expertise of a means-end character (Jacobsen 1977). This is an expertise that is traditionally connected to discretionary decision-making behavior. Jurists have an education not primarily connected to means-end thinking, but a lot of jurists in faet have jobs characterized by such thinking.

9. Professional variables are defined in the following way: Professional considerations is dichotomized between respondents saying that professional considerations are very important as decision signals (58 percent) versus all other answers (42 percent). Controversial proposals is dichotomized between respondents that as a rule will consider advancing a controversial proposal to their superiors (64 percent) versus all other answers (36 percent). Finally, disagreement is dichotomized between respondents saying that they very seldom or never prepare or implement policies with which they disagree (34 percent) versus all other answers (66 percent).

10. The demography of the Norwegian ministries has changed substantially between 1975 and 1986, especially concerning type of higher education, age, tenure and sex distributions (Christensen & Egeberg 1989a). Results concerning role perceptions and enactment, however, are mainly the same. This suggests that changing personnel in public bureaucracies may have less impact on bureaucratic role behavior than structural redesign (Christensen 1987; Egeberg 1987).

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., Derhen, H.-U., Mayntz, R. & Rockman. B. 1988. 'American and West
German Federal Executives. Technocratic and Political Attitudes'. Paper presented at the
XlVth World Congress of IPSA, Washington DC.

Aberbach, J. D. & Rockman, B. A. 1988. 'Image IV Revisited: Executive and Political
Roles', Governance 1, 1-25.

Anton, J. 1980. Administered Politics. Elite Political Culture in Sweden. Boston: Nijhoff.

Christensen, T. 1987. 'How to Suceed in Reorganizing: The Case of the Norwegian Health
Administration', Scandinavian Political Studies 10, 61-77.

Christensen, T. 1989a. 'Causesof Innovative Behavior in Public Bureaucracies'. Oslo: Department
Political Science, University of Oslo. Mimeo.

Christensen, T. 1989b. "Causes of Turnover in Public Bureaucracies - Explanations and
Implications". Oslo: Department of Political Science, University of Oslo. Mimeo.

Christensen, T. & Egeberg, M. 1989a. "Ministerial Budget Making: Behavioral Characteristics and Organizational Constraints'. Paper prepared for the workshop 'The Politics of Finance: Financial Decision-Making in West-European Governments'. ECPR Joint Sessions, Paris 10-15 April 1989.

Christensen, T. & Egeberg, M. 1989b. 'Norske departementer 1976-86: Noen trekk ved
formell struktur, personeli og beslutningsatferd". in Egeberg, M., ed., Institusjonspolitikk
og forvaltningsutvikling. Bidrag til en anvendt statsvitenskap. Oslo: TANO.

Christoffersson. U., Molin, 8.. Månsson, L. & Strømberg, L. 1972. Byråkrati och politik.
Stockholm: Bonniers.

Dogan, M. 1975. 'The Political Power of the Western Mandarins: Introduction'. in Dogan,
M., ed.. The Mandarins of Western Europe: The Political Role of Top Civil Servants (see
Putnam 1977 ref). New York: Halsted Press.

Egeberg, M. 1987. 'Designing Public Organizations'. in Kooiman. J. & Eliassen. K. A.. eds.
Managing Public Organizations. London: Sage.

Eriksen, S. 1988. Herskap og tjenere. Oslo: TANO.

Finer, H. 1941. 'Administrative Responsibility in Democratic Government", Public Administration
1, 335-350.

Friedrich, C. J. 1937. Constitutionai Government and Politics. New York: Harper.

Hedberg, B. L. T., Nystrom, P. C. & Starbuck, W. H. 1976. 'Camping on Seesaws:
Prescriptions for a Self-designing Organization', Administrative Science Quarterly 21, 41-65.

Jacobsen, K. D. 1960. 'Lojalitet, nøytralitet og faglig uavhengighet i sentralforvaltningen',
Tidsskrift for samfunnsforskning 1, 231-248.

Jacobsen, K. D. 1977. Teknisk hjelp og politisk struktur. Oslo: Norwegian University Press.
Second edition.

Kaufman, H. 1956. 'Emerging Conflicts in the Doctrines of Public Administration', American
Political Science Review 50, 1057-1073.

Lægreid, P. & Olsen, J. P. 1978. Byråkrati og beslutninger. Bergen: Norwegian University
Press.

March, J. G. & Olsen, J. P. 1976. Ambiguity and Choice in Organizations. Bergen: Norwegian
University Press.

March, J. G. & Simon, H. A. 1958. Organizations. New York: Wiley.

Mayntz, R. & Scharpf, F. W. 1975. Policy-making in the German Federal Bureaucracy.
Amsterdam: Elsevier.

Mellbourn, A. 1979. Byråkratiens ansikten. Rolluppfattningar hos svenska høgre statstjånstemån.
Stockholm Studies in Politics, no. 13. Publica.

Olsen, J. P. 1983a. 'Top Civil Servants - Key Players on Different Teams', in Olsen, J. P.,
Organized Democracy. Bergen: Norwegian University Press.

Olsen, J. P. 1983b. 'Still Peaceful Coexistence or Revolution in Slow Motion?', in Olsen J.
P., Organized Democracy. Bergen: Norwegian University Press.

Peters, B. G. 1987. 'Politicians and Bureaucrats in the Politics of Policy-making', in Lane, J.
E., ed., Bureaucrats and Public Choice. London: Sage.

Pfeffer, J. 1983. 'Organizational Demography', Research in Organizational Behavior 5, 299-357.

Putnam, R. D. 1973. 'The Political Attitudes of Senior Civil Servants in Western Europe: a Preliminary Report'. British Journal of Political Science 3, 257-290. Reprinted in Dogan, M., ed., The Mandarins of Western Europe: The Political Role of Top Civil Servants. New York: Halsted Press, 1975.

Putnam, R. D. 1977. 'Elite Transformation in Advanced Industrial Societies: An Empirical
Assessment of the Theory of Technocracy', Comparative Political Studies 10, 383-412.

Roos, L. L. & Starke, F. A. 1981. 'Organizational Roles', in Nystrom, P. G. & Starbuck, W.
H., eds., Handbook of Organizational Design 1, 290-308. New York: Oxford University
Press.

Scott, W. R. 1981. Organizations. Rational, Natural and Open Systems. N.J: Prentice-Hall.

Suleiman, E. N. 1984. 'Introduction', in Suleiman, E. N., ed., Bureaucrats and Policy-making.
A Comparative Overview. New York: Holmes and Meier.