Resumé

The notion that political structures and processes are important policy determinants
been questioned in a set of studies investigating relations between socioeconomic
politics, and policy. However, conclusions concerning the
position of politics are often restricted to statements dealing with the relative significance
socio-economic variables as compared to political variables, whereas
attempts at further characterizing the relations have been sparse. The aim of the
article is to characterize the relation between politics and policy by evaluating
the explanatory functions of politics when controlling for socio-economic conditions.
this end a typology of functions is developed, discriminating principally
between determining, intervening, and transforming functions. Some specific
ways of utilizing this apparatus are suggested as three areas of application, corresponding
three analytical levels, namely description, explanation, and interpretation,
briefly discussed.

In modern political science 'policy' has become a fashionable concept, turning up over and over again in various contexts. 'Fast auf jeder Seite der Anzeigen \m American Political Science Review kann man das Wort in einem neuen Buchtitel finden; jede Universität, die etwas auf sich halt, bietet meist gleich mehrere Lehrveranstaltungen an, in deren Titel das Wort auftaucht', Guenther Schaefer (1972, 262) notes in a discourse on American political science in the late 60's and early 70's. This forms a glaring contrast with the earlier state of affairs - Schaefer also notes that 'zwischen 1945 und 1965 war in der amerikanischen politikwissenschaftlichenLiteratur Wort 'policy' kaum zu finden'. The differenceseems great that one might feel justified to talk about a paradigmaticshift: then, would represent the new focus of research in political science. However, such a thought is only partially justified. The phrase that there is nothing; new under the sun clearly applies to 'policy' also. In earlier political science one was thus inclined to study the origins of laws and societal reforms, although one was not at the time aware of the

fact that this really coincides with the study of the coming into being of 'policies' (e.g. Hessler 1942). Also one was intensively occupied with the question how electoral systems affect party systems, although one was not at the time aware of the fact that this really coincides with the study of constitutional 'policy outcomes' (e.g. Teljo 1930, Duverger 1950).

On the other hand, it cannot be denied that the study of policies has brought new elements into the study of politics. The policy orientation has implied a deviation from predominant lines of thinking in three respects. Firstly, it has implied an orientation towards the content of politics and towards the impact of this content on society - for instance, it has encouraged posing of questions pertaining to the impact of allocations on the living conditions of various societal groupings. Secondly, it has involved a legitimization of normative approaches and recommendations based on such approaches: in policy studies one often estimates the correctness of decisions by relating these decisions to certain normatively defined goals. And thirdly, it has directed research interests towards macro-problems, involving efforts to estimate the importance of politics for policies or the impact of policy contents on political processes. In themselves these new features are only partly new. But they do represent aspects which were pushed in the background by the behavioral revolution, which, although in itself necessary for the scientific progression of political science, by its focus on processes and behavior, its sharp programmatic repudiation of normative elements, and its emphasis on micro-phenomena and precise quantitative methods caused a distortion of perspective and a reduction of the scope of the discipline. In its rejection of behavioralism the policy orientation has in its turn emphasized such aspects which were neglected by the behavioral school. But this emphasizing of certain aspects has involved a readiness to utilize valuable strains in the behavioral revolution. the policy orientation is at least partially to be seen as an expression of the way scientific epochs interact.

At any rate, three types of overarching problems can be distinguished in the field of policy studies as it is conceptualized today (see Figure 1 below for an illustration). There is firstly the question of policy determinants - the issue of how and why we get those policies we get. There is secondly the question of policy effects - the issue of how and why certain policies have a certain impact on the environment. And there is thirdly the question feedback - the issue of hov/ and why certain policy effects influence policy determinants in a certain v/ay. In this paper we shall dwell upon the first of the problem areas hinted at above, namely the issue concerning the importance of politics for policies.

To state that this issue has dominated the policy literature is probably an exaggeration. But the issue has certainly been among the central ones, and it is justified to say that it pretty well illustrates the new aspects implied by the policy orientation. Earlier, the conception that the structures and processes of politics shaped those items we now call 'policies' was no doubt both implicit and self-evident. When political scientists conducted research on electoral behavior, cabinet formation, political attitudes, party organization, pressure groups, party systems, etc., they were ultimately by the conception that these phenomena were important because they affected what David Easton (1965, 21) in his famous and disputed definition of politics called 'the authoritative allocation of values for a society', i.e. those actions and decisions emanating from the political system. When during the 1960's this very conception was questioned in a set of policy studies investigating relations between socioeconomic factors, and policy, political scientists were faced with the necessity of dealing with statements like: 'Economic development shapes both political systems and policy outcomes, and most of the association that occurs between system characteristics and policy outcomes can be attributed the influence of economic development' (Dye 1966, 293). Queer questions were asked: is it actually the case that politics, i.e. the object of political science, is no more than a rather unimportant intervening phenomenon? Is the appearance of the political system really an indifferent meaning that the same socioeconomic structure always produces same policies regardless of the appearance or the functioning of the political machinery?

The aim of this paper is not to provide empirical answers to such
questions, but rather to formulate a set of theoretically conceivable

answers. This attempt takes the form of constructing a typology for such
answers, a typology which offers elaborate ways of describing the impact
of politics in various situations.

1. A Typology

The results stressing the non-importance of politics for policy formation have of course been disputed, and during the last ten years the political science literature has been endowed with an abundance of books, articles, and reports questioning the correctness of such results. Sometimes these efforts have met with success: one has been able to demonstrate an impact of politics on certain policy sectors at certain positions in time and space. Sometimes - perhaps even more often - these efforts have met with less success: one has been able to find only a minor impact of politics. And sometimes these efforts have met with no success at all: one has not been able to detect any impact of politics on policies. The state of knowledge is thus to be described as motley (cf. Downs 1976,2-13), and one conclusion emanating from all these policy studies could be that it is a hard task to reach overarching and definite statements on the importance of politics as a determinant of policy.

The variation in results can perhaps be explained by referring to differencesin designs (Anckair 1978, 99-128). The dependent 'policy' variable has thus been operationalized in a variety of ways, some of which seem clearly more immune to politics than others: when this is the case, one should not be surprised to find varying results as to the impact of politics. Correspondingly, the independent 'politics' variable has been operationalized in a similar variety of ways, operating at different levels of political structures and political processes: when this is the case, one should not be surprised to find conflicting results. Various statistical techniques have been used, techniques which differ in their fitness for the task of grasping the impact of an independent variable on a dependent one: when this is the case one should not be surprised to face divergent conclusions. However, it is not our intention at this point to dwell upon matters of operationalization and validity. Our aim is more limited: we shall discuss only the usage applied in statements on the importance of politics. And we shall in this respect advocate a greater variation, not a lesser one. Some examples are useful. When Fry and Winters published their much-debated paper on 'The Politics of Redistribution', one of their main conclusions (1970, 521) was that'. .. the political variables included in our analysis are considerably more powerful than the socio-economic variables in explaining variance in state redistributive policies'. Phyllis M.

Frakt in a paper oriented towards the issue of explaining variations in the policy variable 'systems responsiveness' concluded (1977, 193) 'These findings overwhelmingly reveal the potential importance of economic development to an explanation of responsive government. . . The relative insignificance of democracy in the full sample can be demonstrated even if it enters the explanatory process first. . .'. When Michael S. Lewis-Beck in a methodological paper tried to estimate the fitness of various statistical techniques for comparing influences from independent variables, he presentedthis (1977, 566): 'When the effects coefficients for a common model of welfare policy are estimated in a data-based example, socioeconomic variables are found to be considerably more important than political variables'.

Now, our point is that conclusions concerning the position of politics between society (socioeconomic environment) and policy are often stated in terms of whether socioeconomic variables are more important than political variables in shaping policies. One hereby relates a set of socioeconomic variables and a set of political variables to a set of policy variables. On the basis of these relations conclusions are reached as to the relative importance of the sets of independent variables. The criterion underlying such conclusions is usually the amount of explained policy variance. However, it should be quite possible to problematize the relations society, politics, and policy in a more informative manner. For instance, the aim may be to understand what varying explanatory values can tell about the way politics function. The existence of politics is in itself essential: if we do not have a political machinery, then we cannot have policies, because policies are precisely decisions and actions emanating from such a machinery. But this very existence can have different shapings: in the flows from society to policies, politics can fulfill varying explanatory functions. One extreme case is when politics does not have any function at all: the appearance of politics does not affect policy. Another extreme case is when politics has a determining function: we have a certain set of policies because we have a certain set of politics. A third case is when an intervening function can be attributed to politics, etc. Thus, we obviously need a set of concepts or a typology describing such variation.

Our typology is based upon the following premises and simplifications:

(1) Three components are involved in the framing of the problem, namely the socio-economic environment (SE), politics (P), and policy (Po). There exists of course a whole network of relations between these components, but we are interested here only in the one-way relations SE

—>- P, SE -> Po, and P—>- Po. Hence, such admittedly envisable relations as
for instance Po —* SE or P—> SE remain outside our focus of attention.

(2) The P— ? Po relation is the central one, and we will try to characterize this relation in terms describing the position of politics between society and policy. This is done by evaluating the explanatory functions of politics in relation to policy. Our aim is to describe how politics influences policies when controlling for socio-economic conditions.

(3) The three relations mentioned above form the dimensions for our typology. Regarding these dimensions, we shall allow ourselves the use of a simplifying and dichotomizing approach, and assume that an explanatory either exists or does not exist. Variations in the socioeconomic environment explain variations in politics or they do not, variations in politics explain variations in policy or they do not, etc. Of course, we are not in a position to say exactly when such an explanatory power is at hand and when it is not: we are obviously unable to give cutting points defined in terms of explained variance. We only mean to say that such a power exists when a certain, and not altogether small, amount of explained variance is achieved.

When dichotomizing the relations in this way and then crossing them, a typology emerges, giving eight explanatory functions of politics (Figure 2). These functions then also stand for various explanatory models, which are given in Figure 3. These schedules are presented below by the aid of the following categorization:

(1) In one category we find functions which are of nature. Either policy variations are caused by variations in politics only, or they are caused by variations in SE also, in which case we label the function of politics as partly determining. This category is characterized by the presence an explanatory relation between politics and policy on the one hand, and by the absence of a corresponding relation between SE and politics on the other. The impact of politics is thus independent in nature, and not a function of SE.

(2) In a second category' we find functions which to a lesser degree express an independent impact of politics. These functions are either reinforcing (this is the case when SE explains politics as well as policy and when politics furthermore has an explanatory power of its own vis-a-vis policy) or intervening (this is the case when SE explains politics which then explains policy). This category is characterized by the presence of explanatory relations between SE and politics as well as between politics and policy. There is still an impact of politics at hand, but this impact is now partially a function of SE and the function of politics can therefore no longer be described as determining. The function is intervening in nature.

(3) In a third category we find functions which to a still lesser degree are of a determining nature. While, strictly speaking, they are not determining at all since there is no explanatory relation between politics and policy, a relation does exist between SE and policy. This relation can have two shapings: either SE explains politics as well as policy, or SE explains policy only. In the former case we label the function of politics as conditionallytransforming, the latter case we label the function as transforming.In category, then, politics performs transforming functions,

passing on influences from SE while adding nothing to these influences.

(4) In a fourth category we finally find functions which are really not functions at all. This category is characterized by the absence of explanatory between politics and policy as well as between SE and policy. Policy appears incomprehensible; in any case policy is not to be explained by social structure or by politics.

Then, we have made a list of some theoretically conceivable situations. The question could be raised whether these situations also are empirically conceivable, that is whether they are to be found in empirical reality as well. One answer is that there would be no great difficulty in finding studies and results in determinant research which exemplify the various situations. This holds true also for the situation where neither socioeconomic nor politics can explain variations in policy (e.g. Beckman 1977). Accordingly, if we leave determinant research to provide an answer, that answer would be that our typology covers a sphere which is empirically meaningful. Perhaps this answer is incorrect: it is of course to be supposed that empirical determinant research sometimes produces false results. But the very fact that such an answer can be given demonstrates the typology catches the various conceptions preserved in the research tradition.

2. Three Applications

The presentation of our typology naturally leads to the question of how it can be utilized, how it can be useful in determinant research. One way of answering this question is of course by generally referring to those codifying and predictive functions usually ascribed to typologies in science. it is our intention to point at some specific and heuristic ways of utilizing the apparatus; the remainder of this paper is devoted to this task. We suggest three fields of application, corresponding to three analytical levels, and, in consequence, to three leading questions. Levels and questions can be arranged in the following manner:

Description - What is the impact of politics on policy?

Explanation - Why is it that politics has a certain impact on policy?
Interpretation - What does it mean that politics has a certain impact on
policy?

2.1 Application I: towards Description

The first application concerns the description of reality, i.e. the establishingof
impact of politics as a determinant of policy. As we have already
mentioned, there is a considerable amount of research in this area, which

has produced varying results, the state of knowledge thus being rather motley. It therefore seems an important task to bring order and method to diversity. One way of working towards a greater clarity obviously is to form an idea of to what degree the diversity is dependent on method. Quoting a parallel we could refer to the criticism directed at power research,which been accused of choosing results as it chooses methods (Walker 1966). In the present context the question is whether such a criticism is also valid for determinant research. Thus, our point of departureis assumption that results concerning the impact of politics on policy are dependent on the way these two components have been conceptualizedin analyses as well as on the way they have been related to each other. Besides being theoretically plausible, this assumptionhas empirical support. A survey of a number of policy studies suggests that a higher explanatory value is ascribed to politics in studies utilizing political throughput variables than in studies utilizing political input variables, and that studies which are diachronically arranged emphasizemore than studies which are synchronically arranged the explanatory power of politics (Anckar 1978 a, 82-83).

Accordingly, the idea underlying this first application is that a sample of existing policy studies is classified in categories on the basis of the results of the studies. The categories are those indicated in the typology. This means that we have one category of studies suggesting an intervening function of politics, another category suggesting a determining function, etc. This categorization is then related to other categorizations which catch methodological attributes in the same studies, and we expect a covariation to appear. Due to lack of space we are not in a position to dwell upon exemplifications at any length; hinting at some possibilities must suffice.

One obvious possibility is to look for connections between research findings and essential opera tionalizations, especially with regard to such principal concepts as 'politics' and 'policy'. In Figure 4, we have reproducedexamples such considerations and hypotheses which come naturally to mind. On the one hand, we have made a distinction between politics as behavior and politics as structure; on the other we have distinguishedbetween and quantitative conceptions of policy. One hypothesis would be that the combination of politics as structure and policy as quantity is likely to produce results emphasizing a transforming function of politics. There are a number of reasons which support this hypothesis. One is that quantities, such as public expenditures, form a policy output in which the influence of politics is likely to be negligible. It

seems only reasonable to assume that socioeconomic factors determine policies as long as policy is comprehended in terms of level of public expenditures. Namely, this level must be dependent on the economic potential in the society: the economic resources fix a boundary which cannot be exceeded, and this holds true whatever features characterize the political system.

This remark is not a very original one. On the contrary, it has been made over and over again in the policy literature (e.g. Heiskanen & Martikainen 1974). Accordingly, attempts at turning to more discriminating policy measures have been made. However, they have not brought forth a general conviction that politics is a decisive factor. And this takes us to another remark, which concerns operationalizations of'polities', where a need for more refined and deliberated conceptualizations is likewise felt. For instance, in the literature it is pointed out that linear assumptions concerning the relationship between cause (politics) and effect (policy) should, whenever possible, be replaced by nonlinear assumptions (Wright 1975). For our part, we wish to call attention to the desirability of conceptualizing in terms of behavior, 'Politics' has for the most part been grasped in a rather noninformative manner. Operationalizations have focused on, for example, the representativeness of the electoral system, political resources, party competition, legislative professionalism, characteristics of the bureaucracy, etc. In a way this is understandable. The fundamental ambition has been to produce general statements on determinants of policy, and such an ambition brings about a choice of indicators that are as general (and thereby structural) as possible. Structural are, of course, not to be rejected simply because they are structural. There is much in the idea that they are in some respects at least to be regarded as summary variables for behavior. But, especially in research at subnational levels, focusing on structure might allure us into utilizing indicators which are so broad and general in nature that they are in fact empty, unfit for catching the interactive complexity of decisionmaking. think there is reason to believe that a choice of behavioral indicators which penetrate deeper into 'politics' lends itself to a more precise and more dynamic description, thus emphasizing the impact of politics.

There are of course other methodological attributes than those pertainingto For instance, it is only natural to ask to what degree research findings are dependent on the general research setting, and one reasonable hypothesis would be that studies which operate on a subnational level of analysis tend to produce results which underestimate

the explanatory power of politics. It is here rather important to remember that societal decision-making comprises chains of decisions which are made on different levels. This is of course fairly obvious, but the point should be made, since it has opened our eyes to the fact that politics can have an impact on policy even if this does not show in studies focusing on only one decision-system. Let us take into consideration a situation where state authorities impose various tasks on local authorities or distributeresourcesacross communes for certain purposes to be fulfilled. If studying local policies (the use of these resources on the local level), the analyst obviously comes to the conclusion that politics plays a minor explanatory role (ef. Sänkiaho 1973). Clearly, the result could be false. It overlooks the fact that there are interactions between decisional systems which are on different levels, and that these interactions reflect an impact from politics. By means of politics and political decisions the suprasystem imposes tasks and directives on subsystems which share some characteristics,such a certain socioeconomic structure. These systems then allocate their resources in the same way. But the similarity is not a result of a similarity in structure. It is a result of politics pursued at a higher systems level.

Still other possibilities come to mind. For instance, there is reason to point out that determinant research has to a considerable degree ignored an explicit temporal ordering of variables into causal sequences, meaning that dynamic inferences concerning the relationship between politics and policy have been drawn from a cross-sectional model of analysis. However, models might divert attention from important causal chains, thus concealing influences from politics (Tompkins 1975). Anyhow, we shall not pursue our discussion further. Our point is that the diversity in findings probably can be related to diversities in research characteristics in an instructive way, and that our typology might provide a helpful instrument in endeavours to this end.

2.2 Application II: towards Explanation

Our second attempt at illustrating an application of our typology is perhaps more ambitious than the first, and hence, it must be said, also more vulnerable. Nevertheless the typology seems conducive to attempts at explanation. Thus it ought to be possible to use it in pursuit of a theory that would help us to explain why there exist certain relations between politics and policy. We could then use our typology as the dependent variable in such a theory. The assumption would be that P and Po tend to show certain relations under specifiable conditions. As a matter of convenience, might conceive of politics as something that goes on within the formal or 'traditional' political institutions of a society. It is not, it would seem, too far-fetched to hold that the impact of that which goes on within these institutions or organizations is dependent on the context of politics. We should look for contextual determinants of the relationship between P and Po.

We cannot here, of course, construct a theory that we could substantiate careful testing. Let us, however, on a low level of ambition, point to two possible contextual factors that might have an impact on the relationship between P and Po. A society could, firstly, be characterized as being either fragmented orintegrated. These conditions could perhaps be considered as fairly synonymous with the conditions depicted by such concepts as pluralistic and corporativistic. A society could, secondly, in a somewhat related, but not identical way, be characterized as centralized or decentralized. If we combine these two dichotomizations, a fragmented society that is centralized would be a society in which we have several sectors that are fairly independent of each other (political, economic, cultural), but where each sector is fairly tightly organized (large European type of political parties, large labor and business unions, a strong elitism within the cultural sector). This is sufficient to show that it is indeed possible to separate from each other our two dimensions along which societies can be characterized, even if we expect them to show positive interaction in practice, i.e. we expect high integration to go hand in hand with high centralization, and possibly to bring that latter condition about as time goes by.

It now seems that the crosstabulation of the dimensions could produce a frame within which different expectations are created as to the relationshipbetween and Po (see Figure 5). Thus, if we have a highly integrated society in which politics and economics are highly overlapping (common institutions), and if that society is centralized (the institutions being few and pervasive), we would expect politics to play a determining role. The

institutional structure would mirror an elitist structure where the elite is precisely so strong that we would not necessarily expect SE to play an important part in determining policies. The corporativist elite would dominate and politics would prevail.

If, on the other hand, we have a society that is characterized as fragmented pluralistic, and where institutions are decentralized, we would expect politics to play only a minor part in explaining policies, i.e we would expect transforming or perhaps conditionally transforming functions. such a society politics and economics are provided fair autonomy. economy, however, produces the wealth. There are market mechanisms operating. Under such conditions, we would assume, the preferences of the people are channeled directly through the political system, thus, paradoxically, making for a correspondence between SE and Po. That is to say, politics does not explain policies exactly because it functions as it should - only as a channel for socio-economically induced preferences of the people. For another type of discussion around this point we must turn to our third application.

A society could also be characterized as fragmented, but at the same time as centralized. We have a few sectors that are relatively independent of each other, but within each sector we have a clear center. In such a situation we would expect traditional pressure politics to arise, and there will be a push toward integration between the power wielders within each sector. Politics will in these cases gain ever-increasing importance, and we can expect to find semi-explanatory functions, i.e politics can play a reinforcing or an intervening role in explaining policies.

Even in a rough formulation of the kind we have presented here, there
seem to be some appealing features to a theory of this type. It does seem to

preclude a research design that created sufficient .variance across the explanatory variables, and this, it seems, leads us to include socialist as well as non-socialist countries. That would be a fruitful and fairly rarely taken step within the policy research tradition that we have here taken as our point of departure. Also, once again, we find that subnational attempts at verifying relations between P and Po are hazardous indeed. It may, however, be possible to attain variance along our suggested dimensions if we turn to a longitudinal research design. The switching from a fairly liberal capitalistic society to a state with a socialis tic type of war economy would be a case in point. The possibility of nation-level longitudinal designs have all too seldom been used in order to penetrate the explanatorypower politics on policy.

Be that as it may, our aim has not, as we have said, been to construct a new theory. Rather, we have offered an example as an illustration of one ambition that it seems could be profitably attached to our typology of the explanatory functions of politics.

2.3 Application III: towards Interpretation

Our third application once more departs from the assumption that we have research findings suggesting a certain explanatory function of politics vis-å-vis policy. We further assume that these findings are expressed in terms of our typology. We now ask how these results can be interpreted from a more overarching perspective, what meaning they can be given in a broader context. For instance, the idea almost suggests itself that normative of political systems could be based on findings indicating that politics performs the one or the other function. Or stated differently: the typology offers a means for classifying political systems. If normative conceptions are linked to the categories of the typology, then the classifying systems equals the evaluation of systems.

An example is perhaps useful. As we all know, there are many conceptionsof constitutes a 'good' or a 'democratic' society. Let us, for the sake of reasoning, pick that which states that a correspondence between authoritative decisions and the preferences of the people is a prerequisite for democracy. Then, if this notion functions as a point of departure, opinions stating that politics functions in a 'good' or a 'bad' manner require as a basis that policies produced by the political system are contrasted with the preferences of the citizenry. If policies are in agreement with preferences, the system is democratic; if there is non-agreement, then the system is nondemocratic. It is quite easy to describe policies in this way, but it is less easy to describe the preference system. A conventional

shortcut in this respect is offered by the notion - in itself rather dubious (cf Kuklinski 1978) - that the socioeconomic structure of society is a valid indicator of preferences held by the citizenry. This notion is based on the assumption that the needs and wishes of people are shaped by the environmentand living conditions embedded in the environment (Godwin & Shepard 1976). If we accept this notion, we conclude that the SE —*¦ Po relation in fact denotes a sort of ideal state, as it seemingly satisfies the demand that policies correspond to preferences. Policy is explained by environmental factors, which, in turn, reflect the preferences of the citizenry.

There are obviously three kinds of research findings indicating threats
against this ideal state, namely:

a) those indicating a non-existence of the SE —> Po relation;
b) those indicating an existence of the P —? Po relation;
c) those indicating the validity of both a) and b).

Since intervening and determining functions imply a lack of an explanatory relation between SE and policy, research indicating an existence these functions unveils nonresponsiveness towards the preferences the people and thus motivates the judgment that the system in question is defective: it allows for a distorting effect due to politics. Further, since reinforcing and partly determining functions imply an explanatory relation between politics and policy, research indicating the existence of these functions likewise unveils a similar distorting effect. In contrast, research confirming the validity of transforming or conditionally transforming functions suggests an adequate functioning of the political system. Here politics only passes on influences from the environment, thus bringing the will of the people into force. Developmental trends from determining functions towards transforming functions are therefore to be regarded as advancements; if such trends are discernible, the system is moving in the right direction (see Figure 6, position A, for an illustration).

Other normative points of departure lead to other and perhaps opposite conclusions. For instance, one conceivable position would be that politics should come out independent from SE; the basis for this position would be the conviction that policies are to be shaped by the participation practised in political processes. Keeping this conviction in sight, and assuming, somewhat unrealistically (cf Ståhlberg & Helander 1975) that participation is equal to influence, it appears highly desirable that variables describing political participation are of decisive importance. Thus there should occur conformity between politics and policy without P directly following from SE. In this case, transforming and conditionally transforming functions

stand for a slate of affairs which is certainly unacceptable. Firstly, there is no explanatory relation between politics and policy, indicating the insignificanceof Secondly, there is a relation between SE and policy, indicating the significance of factors which are not necessarily related to participation. This last-mentioned feature discriminates also against reinforcing and partly determining functions, while on the other hand, intervening and especially determining functions reflect an ideal sort of state where policies are explained by participation, and the initial normative expectation is thus fulfilled. Developmental trends from determiningfunctions transforming functions are therefore to be regardedas if such trends are discernible, the system is moving in the wrong direction (Figure 6, position B).

3. Conclusion

Admittedly, our examples have been superficial and simplified throughout.But hope they have served the purpose of illustrating a noteworthy point, and this point is that the typology perhaps could play a role in separate research efforts aiming at contrasting ideology with reality and, accordingly, at closing the distance between research on ideas and researchon There are different steps in such an undertaking. A close scrutiny of normative theories should be followed by the formulation of a model describing a normative argumentation, which is then tested empirically,and conclusions drawn (Lewin 1973). In the present context, a step towards a bridge building would imply a bringing together of two research traditions, on the one hand the analytical tradition which penetrates into the content of ideologies and the ideas how policies are to be made in a society, and on the other the empirical tradition which orients

itself towards the output sector and output determinants. In other words, we suggest the possibility of a marriage between empirical policy analysis and analyses of political ideas concerning policy formulation. And in that marriage a conceptual apparatus such as the one we have suggested obviously has a function to fulfil. In the formulation of models, ways of systematizing criteria for policy evaluation in terms of political ideologies are needed. We have hinted here at some possibilities for working towards a systematization by means of criteria derived from the relation between politics and policy.

REFERENCES

Anckar, D. 1978 a. 'Om partivariabler i policystudier', Statsvetenskaplig Tidskrift 2, 79-91.

Anckar, D. 1978 b. 'Politikprocesser och politikinnehåll: till frågan om politik som orsak och
policy som verkan', Politiikka 2, 99—128.

Beckman, B. 1977. Regional förvaltning och regional planering. Statsvetenskapliga institutionen,
Universitet, Meddelande 1977:1.

Downs, G. W. 1976. Bureaucracy, Innovation, and Public Policy. Toronto: Lexington
Books.

Duverger, M. 1950. L'influence des systémes électoraux sur la vie politique. Paris.

Dye, T. 1966. Politics, Economics and the Public: Policy Outcomes in the American States.
Chicago: RandMcNally.

Easton, D. 1965. A Systems Analysis of Political Life. New York: John Wiley & Sons.

Frakt, P. M. 1977. 'Democracy, Political Activity, Economic Development, and Governmental
The Case of Labor Policy', Comparative Political Studies 10,
177-212.

Fry, B. R. & Winters, R. F. 1970. 'The Politics of Redistribution', American Political Science
Review 64, 508-522.

Godwin, R. K. & Shepard, W. B. 1976. 'Political Processes and Public Expenditures: A
Re-Examination Based on Theories of Representative Government', American Political
Science Review 70, 1127-1135.

Heiskanen, I. & Ma«l;ikainen, T. 1974. 'On Comparative Policy Analysis: Methodological
Problems, Theoretical Considerations, and Empirical Applications', Scandinavian Political
Vol. 9, Oslo: Universitetsforlaget, 9-22.

Hessler, C. A. 1942. Staten och konsten i Sverige. Stockholm.

Kuklinski, J. H. 1978. 'Representativeness and Elections: A Policy Analysis', American
Political Science Review 72, 165-177.

Lewin, L. 1973. Statskunskapen, ideologierna och den politiska verkligheten. Stockholm:
Rabén & Sjögren.

Lewis-Beck, M. S. 1977. 'The Relative Importance of Socioeconomic and Political Variables
for Public Policy', American Political Science Review 71, 559-566.

Schaefer, G. F. 1972. 'Policy Analysis und Politologie', Sozialwissenschaftliches Jahrbuch
fur Politik, Band 3, Munchen: Gunter Olzog Verlag, 261-280.

Ståhlberg, K. & Helander, V. 1975. 'Influence and Participation in Local Planning', Scandinavian
Studies, Vol. 10, Os;lo: Universitetsforlaget, 71-94.

Sänkiaho, R. 1973. Politiikan tuotoksen tutkimus: Keski-Suomen kuntien menoihin vaikuttavista
vuonna 1968. Jyväskylän Yliopisto, yhteiskuntatieteen laitos, julkaisuja
N:o 15.

Teljo, J. 1930. Yhdenmiehen vaalipiirit vaiko suhteelliset vaalit. Porvoo: WSOY.

Tompkins, G. L. 1975. 'A Causal Model of State Welfare Expenditures', Journal of Politics
37, 392-416.

Walker, J. L. 1966. 'A Critique of the Elitist Theory of Democracy', American Political
Science Review 60, 285-295.

Wright, G. C 1975. 'Interparty Competition and State Social Welfare Policy: When a
Difference Makes a Difference', Journal of Politics 37, 796-803.