Resumé

This article examines whether and to what extent opinions on class-related issues coincide between MP candidates and party supporters in Finland. This strategy is used to approach the mediation of interests from grassroots level to the level of political decision making. In theory, adverse selection and moral hazard problems are identified to have serious implications for the mediation process. Furthermore, the representation role adopted by members of parliament is assumed to be significant in the mediation of interests. Attitudes of the candidates in the 1995 parliamentary elections in Finland and those of citizens are compared empirically. The results indicate that the attitudes of the citizens are not necessarily the same as those of their representatives, i.e. the mediation of interests is not self-evident. A more detailed analysis shows that the candidate's gender, education and adopted representation role can be important as regards the conformity of the candidate's attitudes with those of citizens. Findings also reveal that class struggle is tougher at the level of candidates than at the level of party supporters. Differences in opinions are wider among candidates of the political parties than among party supporters.

Introduction

One of the biggest issues in political science and political sociology is whether, and to what extent, ordinary citizens influence political decision making. Do citizens have a say on issues on the political agenda or do the elite groups dominate decision making? Although the question is fundamental, not very much academic energy is consumed studying it. For example, the power resources approach (see Stephens 1979; Korpi 1984; Esping-Andersen & Korpi 1987), which is one of the most frequently utilised theoretical perspectives in studying the development of the welfare state, has neglected this theme. According to this perspective, representatives of social classes, political parties and trade unions are assumed to promote the interests of classes in the process of decision making. However, power resources theory

---

* Pauli Forma, Local Government Pensions Institute of Finland, PL 425, FIN-00101 Helsinki, Finland.

does not have much to say about how the will of the social classes at the grassroots level is transformed into the aims and plans of the political actors. The same weakness can be found in the structural-political perspective (see Uusitalo 1984; Alestalo et al. 1985), which is based on premises quite similar to those of power resources theory. Hence, the interesting link between social classes and their political representatives is missing from the central perspectivesutilised studying the development of the welfare state.

In representative democracy, ordinary people participate in political decision making via their representatives, members of parliament (MPs). The parliament is, in addition to trade unions and other interest groups, the most important institution that mediates between the preferences of the citizens and actual political decision making. However, this link between the grassroots level and the level of political decision making is not unproblematic. concern about whether or not their MPs are optimally their interests is reflected in everyday discussions. More academically, this theme is related to the problems of dealing with interest mediation, the main focus being on how the preferences and interests of the people are mediated between the grassroots level and the level of political decision making.

This article approaches the mediation of interests in the parliamentary arena from two different theoretical angles. Firstly, the member of parliament his or her voter can be seen as an example of the principal-agent relationship, in which issues of information and monitoring are crucial. Secondly, the representative role adopted by the MP can have important effects on the mediation process. Our empirical analysis is based on two opinion-survey data sets. The first one highlights the opinions of the citizens. Hence, these data are similar to those of typical opinion studies. However, the second data set is different in that it consists of a questionnaire, including questions to those of the citizen survey, which was posted to candidates in the 1995 parliamentary elections. Thus, our two data sets make it possible to compare the opinions of the citizens with the opinions of the general-election candidates.¹ For this comparison, three opinion scales, which roughly measure the old tripolar model of political organisation presented Valen & Rokkan (1974), were constructed. These scales measure the 'class interests' of the farmers (land), workers (labour) and business (capital). Thus, in addition to examining the representation of interests, our research design allows us to test this old model of political organisation with recent data.

The outline of the article is as follows. In the theoretical section, the basic themes of principal-agent relationships are reviewed and their implications for interest mediation are elaborated. After that, we discuss representative roles. In the empirical section, we look first at how candidates conceive of representation. After that, the opinions of the candidates and citizens are

compared. In the last section the results are summed up and their implicationsfor
field of research are discussed.

Citizens and Their Agents

The principal-agent relationship consists of the situation where one actor, the agent, commits to do something on behalf of another actor, the principal Ross 1973, 134; more generally, Stiglitz 1989). The principal and the agent make an agreement that the agent does something for the principal. of this kind of relationship are easy to find: lawyer-client, doctor-patient, politician-citizen and employer-employee are all everyday examples of principal-agent relationships. Obviously the most central problem these relationships is that there is no guarantee that the agent works for the principal after the agreement has been made. As Moe (1984, 756) has stated:

there is no guarantee that the agent, once hired, will in fact choose to pursue the principal's best interests or to do so efficiently. The agent has his own interests at heart, and is induced to pursue the principal's objectives only to the extent that the incentive structure imposed in their contract renders such behaviour advantageous.

Thus, from the principal's point of view, an incentive system that guarantees promotion of his or her interests is needed. Hence, a crucial part of the principal-agent theory concerns the monitoring of the agent. The principal information about the circumstances and actions of the agent in order to monitor him or her. However, information is often distributed asymmetrically so as to benefit the agent (see Moe 1984, 756; Calvert 1986, 50-52). When the asymmetric nature of information is focused on, the concepts adverse selection and moral hazard are important. The former refers to a situation where the agent has information about his or her own knowledge, skills, preferences and diligence which is not available to the principal. From the principal's point of view these are highly important issues when deciding whether or not to employ a potential agent. Hence, the principal can be forced to make a decision about employing an agent with weak or perhaps even badly skewed information. The principal cannot be sure who would be the best agent for him or her. Adverse selection problems belong to the point in time when the agreement is considered. Moral hazard problems become acute later on when the agreement has already been made. This concept refers to a situation in which the principal does not have an opportunity to see in what kind of circumstances the actions of the agent are being taken. More precisely, the principal does not know if the agent is doing a good or a bad job. In the following, we will discuss in more detail the relevance of these concepts to the subject of this article.

From the citizen's point of view, problems begin as early as elections, when citizens have to choose between the agent candidates available. The citizen does not know which of the candidates will best promote his or her interests if elected. It is the candidates themselves who are most familiar with their own skilis and knowledge. This is an example of the adverse selection problem discussed above. Voters do not knowhow the candidates would promote their interests if elected and they have to make their choices under uncertainty (Wiberg & Salonen 1991, 154). However, the choices have to be made somehow. Here the concept of signalling, developed by Spence (1973), is helpful. Because the citizens do not know much about the skilis of the candidates, they have to use, for example, the education and work experience of the candidates as a measure of the aptitude of the candidates for the work of an MP. Candidates may use their educational and career achievements to show that they would be good MPs if elected. In a similar vein, ideology can be taken as a signal. The ideology of a political party is a label that identifies whose interests a party promotes (see Holcombe & Gwartney 1989, 671). The citizens can assume that, for example, political parties on the left work best for the workers' interests. Hence, ideology lessens the need for collecting information because citizens do not have to evaluate the candidates of each party separately (Downs 1957,98-100).

Post-election problems are a little different. The agents have now achieved their aim of getting elected. The question arises of whether or not the agent works on the issues on which she or he promised to work during the campaign (see e.g. Ferejohn 1986). It is extremely difficult for the citizen to monitor his or her agent: if the principal wants to monitor the agent perfectly he or she would need to expend a lot of time and energy sitting in the parliament building listening and watching his or her agent working. Only a few citizens have the opportunity to undertake this kind of monitoring. however, parties help the citizens, since they keep track of their MPs and their actions. This is rational behaviour on the part of the parties because if they failed to do so competing parties could publish negative information concerning their MPs (Holcombe & Gwartney 1989).

The above discussion highlights the problems of monitoring the agent. Another problem is that citizens have very few opportunities to intervene in MPs' work. Normally the period between elections is long, four years in the case of Finland. Between the elections citizens rarely have the means to punish their MP if they are not satisfied with his or her work. This kind of situation, where the principal lacks opportunities to punish the agent, is assumed to reduce the willingness of the principal to monitor the agent (Holcombe & Gwartney 1989, 670). More precisely, this may be a source of political alienation. In the following section we move on to our second theoretical perspective and discuss the importance of the representative roles.

The Impact of the Representative Role

In direct democracy citizens participate directly in political decision making. However, because this kind of democracy is possible only in small communities participation of citizens is usually manifest in representative democracy. Citizens choose their representatives, who then make political decisions on their behalf. However, there are some problems with representative which are the subject of this article. Should the MPs in parliament mirror the preferences of the citizens as closely as possible? Or do the citizens allow their representatives to work in parliament in the way in which they themselves think best (see e.g. Pitkin 1967; Riker 1982)?

The question of the extent to which a member of parliament reflects the will of his or her constituency is fundamental here. This theme can be analysed using different models of representation. In their classic study The Legislative System, Wahlke et al. (1962, 272-80) elaborated three models of representation. a trustee is a representative who works independently of his or her constituents. When considering issues on the political agenda, he or she tries to do what he or she thinks to be just or legal. A trustee thinks that he or she has more information about political issues than the people he or she represents and therefore the electors do not have much of a say in political issues. The second model, the delegate, is the opposite. The delegate thinks her or his work in the parliament should be based on the preferences of her or his constituents rather than her or his own choices. Delegates collect information the views of constituents before making decisions. Some delegates think that this information is the ultimate basis of decision making and overrides their own opinions. However, other delegates may not consider type of background information to be so conclusive. The third model of representation, politico, is a mixture of the first two. Depending on the issue at hand, representatives may behave as either a trustee or a delegate. It is also possible that representatives act according to the politico model when balancing their own interests with those of their constituency.

The model of representation adopted by an MP is crucial when it comes to mediation of interests. It can also be claimed that different models of representation are based on different understandings of the nature of interests. discussion concerning the concept of interest, a distinction between objective and subjective interests has been made (see e.g. Reeve & Ware 1983; Hindess 1989; Coleman 1990). The concept of objective interest assumes the interests of an actor derive directly from the actor's location in society and the social structure: workers have workers' interests, pensioners pensioners' interests and women have women's interests, no matter what they themselves think. On the other hand, when subjective understanding is employed, no one else can say better than the actor himself or herself what his or her interests are.

This discussion about objectivity or subjectivity of interests is also relevant the theme of this article. If the objective concept of interest is employed, the representative is not obliged to clarify what the opinions of the electors are. It is useless to ask the electors because their interests are already known, since they derive from their location in the social structure. But, when the subjective concept of interest is adopted, it is important for the representative to know the preferences and opinions of those he or she represents. No one other than the electors themselves can know the interests of the electors. Now, if this discussion is related to the models of represenation above, it can be argued that the trustee model employs an objective concept of interest. The representative does not ask his or her constituents about their opinions because he or she does what is considered to be in the objective interests of those he or she represents. The delegate utilises the subjective concept because he or she wants to know about the opinions and preferences of the electors. Lastly, the politico may employ the subjective concept of interest in some cases and the objective concept in others.

In sum, the representation of citizens' interests in parliament encounters several difficulties. The problems of moral hazard and adverse selection can cause difficulties in monitoring the agent. Even if the principal succeeds in monitoring the agent it can be difficult to punish him or her. The member of parliament may act according to a model of representation which has negative for the representation of interests. In the following section, we will present the research design of our study and move on to examine the representation of interests empirically.

Research Design

Data

The empirical analysis in our study is based on two different opinion surveys. The first one represents the adult population of Finland (n = 1737) and was collected at the end of 1994 and the beginning of 1995. The sample is the Finnish part of the International Survey of Economic Attitudes project (ISEA) (see Forma & Kangas 1999 for a more detailed description). The questionnaire included questions dealing with the public sector, working life, economics and social policy. The response rate for this survey was 56 percent. One obvious reason for the fairly low response rate was the length of the questionnaire (37 pages). However, the structure of the sample corresponds well with the structure of the Finnish population as regards age, gender and socioeconomic position.

The second survey was posted to a sample of candidates in the 1995

parliamentary elections in Finland (n = 529). The sample was constructed by randomly selecting names to be included in lists of candidates nominated by the political parties. The survey was posted just before the elections, which obviously had a negative impact on the response rate, which was 49 percent. It was thought that collecting the data after elections would not be a good idea. All the most important political parties in the Finnish political system are included in the sample. Numbers of candidates according to political party are: National Coalition Party (73 candidates), Centre Party of Finland (82), Green League (97), Finnish Social Democratic Party (91), Left Alliance (95) and 'Other Parties' (89). The structure of the candidate sample was compared with background information (Statistics Finland 1995) about all candidates in the 1995 elections. The structure of the sample corresponded almost exactly to the structure of the whole candidate populationaccording gender. When it came to education, candidates in our sample were slightly more educated than the other candidates. The questionnairesfor and candidates differed slightly from each other: the citizen questionnaire was longer than the candidate questionnaire. However, the two questionnaires included a set of questions that were the same, which makes it possible to compare the opinions of the citizens and the candidates.

Hypotheses

Traditionally, political organisation in the Scandinavian countries has been based strongly on the class structure of society. Each social class has had its own political party, which has represented class interests in political decisionmaking, trade unions, which have taken responsibility for representingeconomic (see Alestalo & Kuhnle 1987). Valen and Rokkan (1974, 333) have presented a tripolar model that describes political organisation in the Scandinavian countries by separating three factors of production: land, labour and capital. Wages and social benefits are in the workers' interests, the business sector is interested in prices and taxes, and farmers are interested in prices and support for agriculture (Nousiainen 1971; Alestalo et al. 1985; Salminen 1993, 44-47). In Finland, the Centre Party of Finland (the former Agrarian Party) has been the most important political representative of the farmers. The National Coalition Party has represented the interests of the business sector. The specific feature of the Finnish political system has been that the left side of the party spectrum has been divided between two parties, the Social Democratic Party and the Left Alliance (the former Communists). Studies focusing on the development of the welfare state have shown that, for example, the development of the welfare state in the Nordic countries has been the result of struggles between

different social classes (see Alestalo et al. 1985; Esping-Andersen & Korpi
1987; Salminen 1993).

Recently, social class has not been one of the most fashionable perspectives social sciences. Therefore, the starting point of this article may seem odd or old-fashioned. It is true that the trend in Western societies has been towards declining class voting (Clark & Lipset 1991, 403), which has been interpreted as an element of the diminishing significance of social class in general (Clark et al. 1993). This trend can also be found in Finland (see Pesonen et al. 1993). However, the choice to focus on classes can also be defended and claims about the 'death of classes' may be exaggerated, since there still seem to be issues where social class matters. The welfare state seems to be one such issue, since several studies have reported that the social class of the respondent is one of the most powerful explanations of opinions towards the welfare state (see e.g. Ervasti & Kangas 1995; Kangas 1995a; Svallfors 1995).

Class-based political organisation has also been a cornerstone in power resources theory, which has been frequently used to explain the development welfare states (see Stephens 1979; Esping-Andersen 1985; Esping- Andersen & Korpi 1987). However, the question of how the interests of the people are transmitted from the grassroots level to the level of political decision making has been beyond the scope of power resources theory (see Korpi 1989) and research focusing on this topic has been rare (see, however, Kangas 1995b; Forma 1996; Saari 1996). This article aims to shed some light on the theme of interest mediation. The empirical analysis is based on Valen and Rokkan's (1974) model discussed above. The model is operationalised the following three hypotheses.

Hypothesis 1 - Land

Farmers are interested in the maintenance of their employment and minimum
for agricultural products. The operationalisation of the land
interest consists of the following two statements:

(a) Government should subsidise agriculture to protect farm jobs.
(b) Government should guarantee minimum prices for agricultural pro-

ducts.

Hypothesis 2 - Labour

Workers are interested in centralised bargaining in the labour markets and
they also feel sympathy toward trade unions in general. The operationalisation
the labour interest consists of the following three statements:

(a) Wages and salaries should be set nationally in negotiations between
government, national union confederations and employers' groups.

(b) All important economic decisions should be made jointly by the government,
unions and employers' groups.

(c) There should be trade unions in all government enterprises and private
business.

Hypothesis 3 - Capital

Entrepreneurs support bargaining at company level and they also want to
restrict governmental intervention in labour-market bargaining. The capital
interest is operationalised by utilising the following three statements:

(a) Wages should be negotiated in each individual company, rather than
for the industry as a whole.

(b) Employers should have the right to negotiate earnings with each individual

(c) Government should not get involved in disputes between unions and
employers.

All the above-mentioned questions had the same scale of responses: 'Strongly agree', 'Agree', 'Don't know', 'Disagree' and 'Strongly disagree'. Additive scales measuring the 'class interests' of different classes were constructed. scales consist of issues that are in the 'objective interests' of the social classes. The land scale consists of questions measuring farmers' interests. The labour scale includes three variables measuring workers' interests. the capital scale consists of three variables measuring the interests of business people. The weakness of our analysis is that our questionnaires questions that could be used for operationalisation of the knowledge interest which has been assumed to be important especially to the middle classes (see e.g. Uusitalo 1985). Details of the scales are provided in Table 1. According to Cronbach's alpha test the reliabilities of the scales are satisfactory.

In examining the representation of interests, the starting point is that the closer the opinions of the candidates and the supporters of the parties are, the better their interests are represented. If the difference in opinion between candidates and citizens is negative, i.e. candidates' opinions towards own class interests are more negative than the opinions of the citizens, there are problems in the representation of interests. The opposite is also possible: if the difference is positive, i.e. candidates' opinions are more positive towards class interests than are the opinions of the citizens, this means that candidates represent class interests more eagerly than the supporters of the parties wish. Each social class should have its own political which represents the interests of the class. In order to examine this, the following hypothesis is submitted.

Hypothesis 4

Each social class has its own political representatives, which are predicted
to be as follows:

(a) The Centre Party of Finland represents farmers' interests.

(b) The National Coalition Party and the Centre Party of Finland represent
entrepreneurs' interests.

(c) The Leftist parties (Left Alliance and Social Democratic Party) represent
interests.

Although the Greens are not easily accommodated in the land-labourcapital we will also compare the opinions of the candidates and supporters of this party. At the beginning of our empirical analysis, we will examine who the MP candidates think they will be representing if they are elected. After that, we will see whether or not classes differ from each other regarding their opinions on land, labour and capital interests. Claims about the declining significance of class might suggest that there should not be large differences between social classes. Next, we will see how these class interests are transmitted from the grassroots level to the level of political decision making. This is done by comparing the opinions of party supporters parliamentary candidates. Lastly, we will see whether or not some candidates represent interests better than others, by examining what kind of impact gender, education or commitment to class representation has on the representation of interests.

Results

Opinions of the Candidates on Representation

The discussion above suggested that the model of representation adopted

by MPs can be crucial when it comes to the representation of interests. Therefore, we included a set of questions about representation in the candidates'questionnaire. exact question used was: 'Who do you think members of the parliament are representing?' The options given were: 'The whole nation', 'My own electoral district', 'My own voters', 'The supporters of the party' and 'My own social class'. Five responses were offered for each option: 'Strongly agree', 'Partially agree', 'Do not know', 'Partially disagree' and 'Strongly disagree'. Table 2 presents the percentage shares of affirmative answers to the questions among different parties.

In general, representing 'My own electoral district' receives the highest and representing 'My own social class' the lowest support. The strongest support for representing 'The whole nation' comes from candidates of the biggest parties in the Finnish political system, the National Coalition Party, the Centre Party of Finland and the Social Democratic Party. On the other hand, the Greens have much more negative feelings towards this kind of wide representation. Candidates of the Social Democratic Party, the National Coalition Party and the Centre Party of Finland also emphasise the importance of representing their own electoral district. Candidates of the Social Democratic Party, the Left Alliance and the Centre Party of Finland are quite positive towards representing their own voters, and the two Leftist parties emphasise representing the supporters of their own party. The P values in the bottom row of the table indicate that the most significant differences between political parties are found when it comes to representing 'The whole nation', 'My own electoral district' and 'My own voters'. The differences between parties are not statistically significant when it comes to opinions on the representation of supporters of the party or one's own social class."

Interestingly, representing one's 'own social class' does not receive strong support among the candidates. The explanation for this could be that candidates are laundering their interests (Goodin 1986; see also Elster 1985). According to the laundering hypothesis, it is not legitimate to publicly selfish interests. In our case, this could mean the interests of the actor's own social class. In order to maximise votes, it is more rational to reveal more general preferences and interests, for example the interests of the whole nation or the electoral district. However, it is important to notice that laundering preferences do not rule out acting selfishly: candidates may say that they represent wide segments of the nation while in faet they promote the interests of much narrower groups.

Opinions on Land, Labour and Capital

After examining opinions on representation among the candidates we are now ready to analyse attitudes towards class-related issues. According to the hypotheses presented above, farmers should have the most positive opinions towards the land interest, workers should have the most positive feelings towards the labour interest, and capital interest should be emphasisedby In addition, it is also interesting to see what the opinions of the middle classes are, although their own interest, knowledge, is not included in our research design. According to the results presented in Figure 1, our hypotheses are confirmed. The opinions of social classes towardstheir

wardstheirown, objective class interests are most positive. Interestingly, the opinions of the two middle-class groups are quite close to those of the workers. The explanation for this might be that scales measuring capital and labour interest consist of questions that are similarly important to all wage earners, not only for workers.

Whither, then, the representation of interests? How similar are the opinions of the candidates and the citizens? Our assumption here is that workers' interests are represented by leftist parties (the Left Alliance and the Social Democratic Party), the interests of the farmers by the Centre Party of Finland, and the interests of business people by the National Coalition Party and also the Centre Party of Finland. Figure 2 presents a comparison of the opinions of citizens and candidates with regard to the land interest. The longer the bars, the more positive the opinions are. When it comes to interest representation, bars representing the candidates' and citizens' opinions should be as close to each other as possible. If the darker bar (candidates) is longer than the lighter one (citizens), candidates have more positive opinions on the land interest and vice versa. The asterisks indicate the statistical significance (Mann-Whitney's [/-test is used) of the difference between the opinions of the candidates and the citizens.

Our hypothesis regarding the representation of the land interest is supported,since opinions are most positive among candidates and supportersof Centre Party of Finland. When it comes to the representation of farmers' interests, the opinions of the candidates and the supporters of the Centre Party of Finland are very close to each other. Thus, farmers can be quite satisfied with the representation of their class interest. Opinions

are quite similarly distributed among the Left Alliance, the National CoalitionParty the Green League. Both the candidates and the supporters of these parties feel fairly strong sympathy towards farmers and their interests. The land interest is supported least at the candidate level in the Social Democratic Party. Interestingly, there is a substantial and statisticallysignificant between Social Democratic candidates and party supporters: supporters of the party appear to understand farmers' interests nearly as much as the candidates and supporters of the other parties. However,the Democratic candidates reveal much more hostile opinions. This finding obviously indicates that the contradiction between farmers and workers and their interests is stronger at the candidate level than at the citizen level.

Figure 3 presents the results regarding the labour interest. Opinions reveal a clear left-right distinction with regard to the labour interest. Support for the workers' interest is stronger at the left side of the party spectrum than at the right side. Counting Greens out, there are statistically significant differences between candidates and citizens in all political parties. The differences are quite interesting. Among leftist parties the opinions of the candidates towards the labour interest are more positive than the opinions of party supporters. However, among rightist parties the situation is the opposite: candidates reveal more negative attitudes than supporters of these parties. Thus, it seems that candidates are defending the interests of their own social class even more strongly than are people at the grassroots level. In addition, the difference of opinion between candidates and party

supporters among rightist parties is very much in line with the finding presented above regarding the land interest and Social Democratic candidates: candidates seem to be more strongly antipathetic to the class interests of the competing classes than are party supporters at the grassrootslevel.

In Figure 4 the results regarding the capital interest are presented. Here, too, a clear left-right distinction is found. Support for the capital interest is stronger among the rightist parties than among the leftist parties. Interestingly, this distinction seems to be stronger among the candidates than among the citizens. Differences between candidates and party supporters substantial when it comes to the Social Democratic Party and the Left Alliance, while the candidates reveal much more hostile attitudes. As regards the Green League and the Centre Party of Finland, the opinions of the candidates and party supporters are quite similar. However, there is a statistically significant difference within the National Coalition Party: candidates support the interests of capital more enthusiastically than do party supporters.

To sum up, the above evidence indicates that there are significant differencesbetween opinions of the candidates and the parties' supporters. Therefore, the representation of class interests is not self-evident or clear. Two findings are worth emphasising. First, it seems that with regard to labour and capital interests, candidates are promoting these interests even more enthusiastically than their supporters expect. Secondly, also regarding labour and capital interests, candidates see the interests of the competing class in a much more negative light than do party supporters: candidates of

the rightist parties have negative opinions on issues that are in the interests of the workers and leftist candidates are antipathetic towards the interests of entrepreneurs. Also, Social Democratic candidates have negative feelings towards the interests of farmers. Thus, our findings indicate that class struggle is tougher at the candidate level than it is at the citizen level. Citizens have much more understanding for the interests of the other classes than the candidates have.

Comparing means, as was done above, may hide interesting differences between the citizens and different kinds of candidates. It might be that differences of opinion between some groups of candidates and citizens are smaller than the means may suggest and that, therefore, some candidates may represent interests better than others. Earlier research has indicated that gender and education may be relevant factors here. For example, Holmberg & Esaiasson (1988, 126) found that female members of parliament better than men their voters' opinions on different political issues. When it came to educational level, members of the parliament with higher education were more aware of the preferences of their voters than were less educated members of parliament. The relatively small size of the candidate data prevents us from undertaking a very detailed analysis here. However, it is possible to control for some of the background variables. In the following, we will examine the impact of MPs' gender and educational level on the fit between the opinions of candidates and citizens. Furthermore, will examine whether those candidates who emphasise the representation social class (see above Table 2) represent class interests more enthusiastically than others.

Table 3 presents the differences in opinions between citizens and different groups of candidates. The figures have been calculated by subtracting the scale value for the opinions of the group of candidates from the scale value for the opinions of the citizens. The negative sign indicates that candidates have more negative opinions than do supporters of the party, and the positive sign indicates the opposite. Asterisks indicate the statistical significance the difference of opinion between the candidate groups and the citizens.

Gender seems to be unimportant when it comes to the representation of the land interest. The only statistically significant difference is found with regard to the Social Democratic Party, in which female candidates' opinions differ significantly from the opinions of Social Democratic Party supporters.With to the labour interest, it seems that the male candidates of the Social Democratic Party represent the workers' interests better than do the female candidates. Also, the capital interest seems to be better represented by the male candidates of the National Coalition Party. Becauseof limited size of the candidate data, we have to use fairly rough categories when analysing the impact of education: the education variable

was recoded into two classes: 'lower' and 'higher' schooling (see Table 3 footnotes for details). The representation of the land interest through the Centre Party of Finland is not very dependent on the educational level of the candidate. However, as regards the labour interest, education has some impact: less educated candidates of the leftist parties represent labour interests better than do candidates with higher education. Similarly, less educated candidates of the National Coalition Party seem to represent the interests of capital better than do candidates with higher education. Results indicate that the more educated the candidates are the more they understand the interests of other classes.

On the far right side of the table we analyse the impact of commitment to class representation on the fit between the opinions of the candidates and the citizens. Results indicate that those candidates of the Social Democratic Party who emphasise the representation of social class have very negative feelings towards the land interest. The situation is similar with regard to candidates of the Centre Party of Finland and the labour interest. Hence, it seems that candidates who emphasise class representation do not have substantively more positive opinions towards their own class interests. However, they do differ from other candidates by having a more negative attitude towards the interests of the competing social classes.

Conclusions and Discussion

We began our analysis by examining candidates' opinions towards representation. was some weak evidence that the candidates preferred a wide model of representation to a narrow one. In the next step of the analysis we operationalised Valen's and Rokkan's (1974) tripolar model, land-labour-capital. The purpose was to see how this old model describing political organisation in the Scandinavian countries works in the 1990s when tested against opinion data. The aims of the analyses were, first, to see whether opinions of the social classes differ from each other. Do the classes support issues that are close to their predicted, objective class interests? The second aim was to see how the opinions of the candidates and party supporters correspond with each other. The assumption here was that the closer the opinions of the candidates and party supporters are the better the interests are represented.

The results of these comparisons indicated that the opinions of the social classes can be understood from the viewpoint of objective interests. Workers, entrepreneurs and farmers emphasise issues that derive from their location in the social structure. With regard to the representation of interests, the results indicated that this process is not self-evident and clear in all political parties. When it came to the Centre Party of Finland and farmers' interests, the opinions of the candidates and supporters of the party were very close to each other. However, regarding labour and capital interest we found evidence that candidates place even more emphasis on the representation of class interests than do the voters themselves. Furthermore,the seem to have more negative feelings toward competing class interests than the citizens have: candidates of the leftist parties, for example, have a much more hostile attitude towards the interests of capital than do the supporters of leftist parties. These findings are in line with some previous studies. For example, McClosky et al. (1960) state that the party elite has a greater need to defend their party's standpoint and they also have

more interest in developing a consistent set of attitudes. The party elite is
also more motivated not only to support the party appropriate to their
beliefs but also to learn how it differs from the opposition party.

We also examined what kind of impact candidates' gender, education and commitment to class representation had on the fit between the opinions of the candidates and the citizens. The results indicated that the gender and educational level of the candidate matter when it comes to consistency between the opinions of the candidates and the citizens. Less educated candidates were relatively focused on their own class interests while more educated candidates also understood the interests of other classes. The commitment a candidate to class representation also had some impact on the consistency between the opinions of the candidates and the citizens. Those candidates who emphasised the representation of social class had more hostile opinions towards the interests of competing classes. Hence. from the viewpoint of the citizens, it does matter to whom the vote is given - some agents seem to work for the people better than others.

Discussion about the objectivity and subjectivity of interests is relevant when interpreting the results of this study. Who is a good agent, a good representative of interests? Does a good agent follow as closely as possible the opinions of his or her principal? Or does a good agent promote as strongly as possible those things that are assumed to be in the objective interests of the principal? There is no correct answer to this question, since the answer depends on our understanding of the concept of interest (see above). Some differences in opinion measured here can be interpreted from the viewpoint of strategic choices. For example, collective bargaining in the labour markets influences the power of trade unions. It can be argued that the elite knows better than the average member or party supporter the importance of this kind of strategic choice. The candidates of the leftist parties are more aware than voters of the effects of bargaining at the level of the individual firm. Therefore, the candidates of the leftist parties may adopt a much more critical attitude to questions like this. Candidates of the leftist parties may think that collective bargaining is in the objective interests the worker.

Other studies that have contrasted opinions of the citizens to opinions of elite groups in Finland have also found differences between the elite groups and the general population. For example, a study focusing on opinionstowards services revealed opinion discrepancies between the municipal elite and the population (see Blomberg & Kroll 1999). According to this study, these differences even grew during the 19905. The authors suggestthat reason for the opinion gap could be the future political interests of the municipal elite. The elite might think that it is more advantageousto the municipal debt or to refrain from raising taxes and not to listen to the demands of citizens who do not want cuts in welfare

services. Future political interests of the elite come close to strategic choices as discussed above. Another interesting finding in a study by Blomberg and Kroll (1999) was that opinions of the political elite were closer to opinions of the population than were opinions of the administrative elite at the municipal level. Also other studies point out that opinions of the political elite are closer to opinions of the population than are views of the other elite groups (see e.g. Forma 1999). Thus, while the evidence found in this study shows differences between the MP candidates and party supporters, it should be kept in mind that contrasting the opinions of party supporters to those of other elite groups could reveal even more dramatic discrepancies.

The results of this article indicate that the old land-labour-capital distinction still has at least some validity with regard to political organisation Finland. When issues relevant to class are focused on, differences in opinion between classes can be found. Although many authors have claimed that social classes have lost their importance, it seems that classes still have relevance at least when it comes to public opinion (see e.g. Ervasti & Kangas 1995; Kangas 1995a; Svallfors 1995). Second, we found evidence that each social class still has its own political party which understands the interests of that class best. In Finland the Centre Party of Finland seems to represent the interests of the farmers, the Left Alliance and Social Democratic represent the interests of the workers, and the Centre Party of Finland with the National Coalition Party represents the interests of business

Comparisons between candidates and citizens indicated clear differences of opinion which suggest that the representation of interests is not straightforward. frequently utilised power resources perspective, which has been used to study the development of the welfare state, has not discussed or tried to empirically analyse problems dealing with interest representation. discussion about the driving forces behind different welfare programmes explanation for this or that outcome has been attributed to the different classes. For example, the working class has been credited with the universalism of the Scandinavian welfare state (Esping-Andersen 1985). According to Baldwin (1990) it was the middle class which fought for universalism. in that discussion social classes refer to political parties and trade unions, not the people belonging to the working class or to the middle class. The results of this article show that the preferences of a social class at the grassroots level can be very different from preferences at the elite level.

Another implication of the findings concerns the use of the results of public-opinion studies in general. Opinion studies often do not discuss the impact of public opinion on political decision making. It can be argued that the way in which opinion researchers interpret the impact of public opinion on political decision making is over-optimistic. For example, several researchershave

searchershavebeen very optimistic about the future of the welfare state after observing that support for the welfare state at the level of the citizens is strong (see e.g. Taylor-Gooby 1995). However, the differences in opinions between parliamentary candidates and citizens form the basis for a predictionthat opinion and political decision making do not necessarily go hånd in hånd.

The evidence found in this study suggests that more studies that operate both at the citizen level and at the elite level are needed in order to obtain a deeper understanding of how interests mediate from the grassroots level to the elite level. It is clear that types of research design other than the one utilised in this study are also needed. Although opinion studies do not reveal much about the actual process of interest mediation, they do provide some interesting information about this subject. If no differences between the representatives and the represented people were found there would not be much to study. However, the fact that differences are found indicates that there really is something at stake here.

---

NOTES

1. The empirical strategy used here has its weaknesses. First, opinions do not tell much about actual behaviour. Second, political decision making is not only preference aggregation; log rolling and compromises are also parts of the process. Third, there is no guarantee of the reliability of answers in this kind of survey. For instance, candidates may express strategic opinions rather than their actual opinions. However, the method has clear merits, too: the greatest advantage is that the variables compared are commensurate. Comparing opinions and political decisions would be more difficult.

2. This measure is a bit problematic, since it is difficult to know exactly what 'own class' really means for the respondent. One might wonder whether it means the class of the politicians or their class before their political career. However, since this is a survey for MP candidates, aiming to become members of parliament, we can quite firmly assume that the respondents relate it to their present class position.

ACKNOWLEDGEMENTS

The author is indebted to Katja Forssén, Olli Kangas, Heikki Paloheimo, Juho Saari, Kari
Salavuo, Matti Wiberg and an anonymous referee for their helpful comments on previous
versions of this article.

REFERENCES

Alestalo, M., Flora, P. & Uusitalo, H. 1985. 'Structure and Politics in the Making of the Welfare State: Finland in Comparative Perspective", in Alapuro, R., Alestalo. M.. Haavio- Mannila, E. & Våyrynen, R., eds. Small States in Comparative Perspective. Oslo: Norwegian University Press.

Alestalo, M. & Kuhnle, S. 1987. 'The Scandinavian Route: Economic, Social, and Political Developments in Denmark, Finland, Norway, and Sweden', in Erikson. R.. Erik. H.. Ringen, S. & Uusitalo, H., eds. The Scandinavian Model. Welfare States and Welfare Research. New York: M. E. Sharpe.

Baldwin. P. 1990. The Politics of Social Solidarity. Cambridge: Cambridge University
Press.

Blomberg, H. & Kroll, C. 1999. 'Who Wants to Preserve the Scandinavian Welfare State? Attitudes to Welfare Services among Citizens and Local Government Elites in Finland, 1992-6', in Svallfors, S. & Taylor-Gooby, P., eds, The End of the Welfare State? Responses to State Retrenchment. London: Routledge.

Calvert, R. 1986. The Models of Imperfect Information in Politics. New York: Harwood
Academic Publishers.

Clark, T. N. & Lipset, S. M. 1991. 'Are Social Classes Dying?', International Sociology 4,
397-410.

Clark, T. N., Lipset, S. M. & Rempel, M. 1993. 'The Declining Political Significance of Social
Class', International Sociology 8, 293-316.

Coleman, J. 1990. Foundations of Social Theory. Cambridge: Belknap Press of Harvard
University Press.

Downs, A. 1957. An Economic Theory of Democracy. New York: Harper & Row.

Esping-Andersen, G. 1985. Politics against Markets. Princeton: Princeton University
Press.

Esping-Andersen, G. & Korpi, W. 1987. 'From Poor Relief to Institutional Welfare States: The Development of Scandinavian Social Policy', in Erikson, R., Hansen, E., Ringen, S. & Uusitalo, H., eds, The Scandinavian Model. Welfare States and Welfare Research. New York: M. E. Sharpe.

Ferejohn, J. 1986. incumbent Performance and Electoral Control', Public Choice 50, 5-25.
Formå, P. 1996. The Politics of buerest Mediation: The Case of Universalistic Social Policy
in Finland. University of Turku, Department of Social Policy, Series B/7.

Forma, P. 1999. Interests, Institutions and the Welfare State. Studies on Public Opinion
towards the Welfare State. Stakes, Research Reports 102. Saarijårvi: Gummerus.

Forma, P. & Kangas, O. 1999. 'Need, Citizenship or Merit?', in Taylor-Gooby, P. &
Svallfors, S., eds, State Transformation and Public Opinion. London: Routledge.

Goodin, R. 1986. 'Laundering Preferences', in Elster, J. & Hylland, A., eds, Foundations of
Social Choice Theory. Cambridge: Cambridge University Press.

Hindess, B. 1989. Political Choice and Social Structure. An Analysis of Actors, Interests and
Rationality. Worcester: Billing.

Holcombe, R. & Gwartney, J. 1989. 'Political Parties and the Legislative Principal-Agent
Relationship', Journal of Institutional and Theoretical Economics 145, 669-75.

Holmberg, S. & Esaiasson, P. 1988. De folkvalda. En bok om riksdagsledamoterna och den
representativa demokratin i Sverige. Stockholm: Bonniers.

Kangas, O. 1995a. 'Attitudes on Means-Tested Social Benefits in Finland', Acta Sociologica
38,299-310.

Kangas, O. 1995b. interest Mediation and Popular Will', in Svallfors, S., cd., In l/ie Eye ofthe
Beholder. Opinions on Welftire and Juslice in Comparative Perspeclive. Umeå: Scandbook.

Korpi, W. 1984. Working Class in U'elfarc Capitalism. London: Macmillan.

Korpi, W. 1989. 'Power, Politics and State Autonomy in the Devclopment of Social Citizenship.
Rights During Sickness in Eighteen OECD Countries since 1930', American
Sociological Review 54, 309 -29.

McClosky, H.. Hoffman, P. J. & O'Hara, R. 1960. 'Issne Conilict and Consensus among
Party Leaders and Followers', American Polilical Science Review 54. 406 27.
Moe, f. 1984. The New Economics of Organization', American Journal of Polilical Science
28. 739-77.

Nousiainen, J. 1971. The Finnish Political System. Cambridge: Harvard University Press.

Reeve. A. & Ware, A. 1983. 'Intcrcsts in Political Theory', British Journal of Polilical Science
13. 379-400.

Riker, W. 1982. Liberalism against Populism. San Fransisco: W. H. Freeman.

Ross, S. 1973. 'The Economic Theory of Agency. The Principal Problem', American Economic
Review 63, 134-9.

Saari, J. 1996. Sosiaalipolitiikka markkinariippuvuuden våhentdjånå. Stakes, tutkimuksia 69.
Saarijårvi: Gummerus.

Salminen, K. 1993. Pension Schemes in the Making. The Central Pension Security Institution.
Studies 2/1993. Helsinki: Hakapaino.

Spence, M. 1973. 'Job Market Signaling', Quarterly Journal of Economics 87, 355-74.

Stiglitz, J. 1989. 'Principal and Agent', in Eatwell, J., Milgate, M. & Newman, P., eds,
Allocation, Information and Markets. London: Macmillan.

Svallfors, S. 1995. 'The End of Class Politics? Structural Cleavages and Attitudes to Swedish
Welfare Policies', Acta Sociologica 38, 53-74.

Taylor-Gooby, P. 1995. 'Who Wants the Welfare State? Support for State Welfare in
European Countries', in Svallfors, S., ed., In the Eye of Beholder. Opinions on Welfare and
Justice in Comparative Perspective. Umeå: Scandbook.

Uusitalo, H. 1984. 'The Development of the Welfare State: The State of the Art', European
Journal of Political Research 12, 403-22.

Uusitalo, H. 1985. 'Luokkarakenne ja intressien organisoituminen', Sosiologia 22, 165-76.

Wahlke, J., Ealau, H., Buchanan, W. & Ferguson, L. 1962. The Legislative System.
Explorations in Legislative Behaviour. London: John Wiley.

Wiberg, M. & Salonen, H. 1991. 'Poliittiset pååmies-agentti-vuorovaikutustilanteet',
Politiikka 33, 149-62.