Resumé

Political parties are often said to be in decline. In Sweden, the major parties have lost significant amounts of members and activists in recent years. The purpose of this study is to investigate whether the decline in participatory linkage via parties has been accompanied by a representative Using data from the Swedish Election Studies and Parliamentary Studies, average self-placement on the left-right schema is compared between Members of Parliament, party activists, party members and voters. There is no evidence of a decline in the political representativeness the parliamentarians. There is, however, some evidence that members and activists have become radicalized compared to the voters. This is particularly true of the Social Democrats, and also to some extent in the Moderate Party. Thus, while there is no general decline in the representative linkage provided by the Swedish parties, there is some justification to speak of a representational decline for the Swedish parties as membership organizations.

Introduction

In the 1996 report from the Democratic Audit of Sweden, Petersson et al.
noted a negative trend in political participation in Sweden. Not everything
was gloomy, but:

Fewer and fewer members participate actively in the internal life of the organizations. A
considerable number of associations and parties are having great difficulty in recruiting
activists (Petersson et al. 1997, 21).

There is always a risk of exaggerating the problem. Indeed, Gilljam & Holmberg (1995) found no evidence that Sweden had changed from a participatory to a spectator democracy. What seems clear, however, is that participation, to an increasing extent, takes place outside the political parties. In 1968, over one in ten adults were individual members of a party. If collective members of the Social Democrats are included, the proportion was much higher; close to 25 percent according to the parties' own statistics.

---

* Anders Widfeldt. Department of Politics and International Relations. University of Aberdeen. Edward Wright Building, Aberdeen A824 3QY. UK. E-mail: polOS4«?abdn.ac.uk

By the mid-1980s, the proportion individual party members of the electorate had, in fact, increased somewhat, but since then there has been a negative development which has accelerated after 1990. Activity in political parties has also declined. In 1979, over 6.5 percent of the voters reported that they had attended at least one party meeting during the past year. In 1994, the proportion had gone down to four percent (Widfeldt 1999).

Given the documented loss of members and activists, is there a risk that the parties are losing touch with the public? Are the Swedish parties becoming insular sects? The purpose of this article is to investigate whether the documented decline in participatory linkage provided by the parties has been accompanied by a decline in representative linkage. The study will deal with the development between 1985 and 1994. In that period, the once stable party system has become fragmented. At the same time, support for the dominant Social Democrats has declined (Arter 1999a, 351ff; 1999b). The two traditional ideological blocs have shown signs of withering, as exemplified by the parliamentary cooperation between the Social Democrats Center Party in economic policy between 1995 and 1997. These changes have been accompanied by important developments in the political climate and culture: The welfare state has undergone a phase of profound reassessment (Arter 1999a, 188-95); the corporatist 'Swedish Model' of government is no longer valid (Rothstein 1998); unemployment has reached the highest levels since the 1930s depression; in late 1994, Sweden decided to join the European Union, something which seemed highly improbable in the late 1980s. In other words, the nine years between 1985 and 1994 cover many important and historic events. Before going on to the empirical evidence, some conceptual problems will be discussed.

In her famous book first published in 1967, Pitkin defined representation as 'acting in the interests of the represented, in a manner responsive to them' (Pitkin 1972, 76). Such a definition sets the focus on activity. It is also a definition which concerns the interests rather than the manifest opinions of the represented. Yet, much of the research on political representation has focused on representativeness, rather than representation the way Pitkin defines it. In other words, focus has mainly been on the degree of likeness between the representatives and the represented, normally because it has been assumed that representativeness is highly correlated to representation. If the representatives and the represented share enough important characteristics, the former can be assumed to act in the interests of the latter.

Political representativeness either means similarities in social characteristicsor in terms of opinion. Research on the social representativenessof tends to confirm the popular image of politicians as a social elite (e.g., Esaiasson & Holmberg 1996; Aberbach et al. 1981; and Norris & Lovenduski 1995). There is, however, no unanimity that

social characteristics are democratically important. Many would agree with Westerståhl's argument that representativeness in terms of opinion is what really matters. This perspective is manifested by the proportional electoral system. A proportionally elected assembly should in its composition reflect the opinion distribution in the electorate. This, in turn, means that it is assumed that the elected representatives have opinions which are representativeof who elected them (Westerståhl & Johansson 1981, 20). There are also several examples of scholars who argue that large discrepanciesin between politicians and the public is a democratic problem (e.g., Huber & Bingham Powell 1992; Dalton 1996, 240f).

The Study

This study will focus on political differences between Swedish parties and the voters. The main question is whether the differences between the parties and the public have increased over time or not and, thus, whether the political representativeness of the Swedish parties has decreased or not. Four party levels will be compared in two phases. First, a comparison involving only elected parliamentarians and the parties' respective voters. Second, a wider comparison involving party members and activists. Due to its continuing centrality in Swedish politics, the left-right dimension will be used as the analytical tool. There are examples of other dimensions: Most commonly referred to is the emergence of a green dimension, sometimes referred to as growth versus ecology, or the industrial-agrarian dimension. The entry of green parties in many European countries, but also the reorientation some older parties, in Sweden primarily the Center Party, are often cited as indicators that the green dimension has started to challenge the traditional left-right dimension. In Sweden, however, research evidence suggests that the dominance of the left-right dimension remains intact. Empirically, the left-right dimension in Sweden consists of issues such as privatization versus nationalization, taxes, and laissez-faire economics versus state interventionism (Bennulf 1994, 221; Gilljam & Holmberg 1995, 62).

A study of political representativeness is a comparison of political attitudes. Political attitudes can be measured objectively or subjectively. The objective method is based on the respondents' opinions on concrete issues, which can be summarized into an attitude index. Such an index does not take into account whether the respondents consider themselves left or right (for examples of studies based on an objective attitude index, see, e.g.. Esaiasson & Holmberg 1996, 92-101; Brown et al. 1999, eh. 4). In the subjectivemethod, the respondents' own perception of left and right is central. Here, respondents are asked where they would place themselves

on a numerical scale from left to right, where low numbers indicate a leftist position and high numbers a position to the right. This method, commonly referred to as left-right self-placement, does not take opinions on concrete issues into account (see, e.g., Esaiasson & Holmberg 1996, 101-4; Widfeldt 1995). The development of the representativeness of Swedish parliamentariansaccording the attitude index has been extensively reported elsewhere(Holmberg Studies using left-right self-placement also exist, but none in which the development throughout the 1985 to 1994 period has been studied. There is, furthermore, no study from Sweden using the subjectivemeasure includes parliamentarians and party members in the same analysis (Esaiasson & Holmberg 1996, 96ff include MPs and party members in their study, using the objective measure, with data from 1985 only).¹

According to the attitude index, MPs of all Swedish parties were to the left of their respective voters in the late 1960s (Holmberg 1974). This pattern has been given the epithet 'the floating corn-field.' Since then, however, the pattern has changed. Subsequent studies from 1985 and 1988 show that the 'corn-field' has been succeeded by a 'fan,' where MPs of rightwing are to the right of their voters, and vice versa on the left. In other words, there was a left-leaning elite in the 19605, but in the 1980s the pattern was one of ideological polarization among the elite, something which Sören Holmberg refers to as the 'elite conflict model' (Esaiasson & Holmberg 1996, 92). The metaphors 'corn-field' and 'fan' were coined by Westerståhl in 1981 (Westerståhl & Johansson 1981, 87). Westerståhl & Johansson's study showed that the change from a left-leaning elite to elite polarization could also be observed at the local government level (ibid., 86f)-² With the benefit of hindsight, and longitudinal research, it seems clear that the situation in the late 1960s was exceptional. The elite conflict model has been the only observed pattern in a number of different studies on Swedish politics in the 1980s and 19905. Studies using left-right selfplacement consistently resulted in the elite conflict pattern. This may partly be explained by the fact that no comparisons are possible from the late 19605, since Swedish MPs were not asked about their left-right selfplacement 1969. Hence, comparisons between MPs and voters are only possible from 1985, 1988 and 1994 (there was no parliamentary study in 1991), when the objective attitude index also resulted in the elite conflict pattern.

In this article, only the left-right self-placement method will be used. The left-right scale, or schema, is a useful instrument when measuring ideological orientation and its development over time. A person's selfplacementon left-right continuum can be seen as an aggregation ofthat person's political outlook. The scale has been shown to work well in comparisons between different countries, as well as over time. Most citizens

in Western democracies are familiar with the left-right continuum and are able to relate to it in a meaningful way. Self-placement on the scale tends to be correlated to opinions on concrete issues, and most voters are able to place the political parties along the continuum in a meaningful way (Fuchs & Klingemann 1990). In other words, this article will be based on the respondents' self-image of their general ideological outlook.³

One problem with the left-right schema is that the meaning of left and right changes over time. For example, in the 19605, a leftist outlook meant focus on issues such as workers' rights and nationalization of industries, while in the 1990s leftist views mean defense of the welfare state and keeping the service sector in public ownership. Thus, it may mean something different be in a given position on the left-right schema one year, compared to being in the same position several years later. Against this it could be said that stability on the left-right schema shows a similar ideological outlook, bearing in mind which particular issues are considered ideologically important divisive at a given time. Therefore, comparisons over time of leftright is a meaningful exercise, since the left-right schema can be argued to control for changes in the content of left and right, and shifts in the saliency of different left-right issues. It should also be noted that most of the analysis in this article will focus on a comparison of the differences between different party levels, i.e., not only on the actual positions of parliamentarians, members and voters, but also the relative differences between these party levels, and the extent to which the sizes of these differences have changed over time.

Data on voters and party members are taken from the Swedish Election Studies, based on in-person interviews (Holmberg 1994). Data on MPs are taken from mail questionnaires sent in connection with the Swedish parliamentary of 1985, 1988 and 1994.⁴ Four levels will be studied: elected parliamentarians, party activists, passive party members and party voters. First, however, the focus will be on voters and parliamentarians.

Political Representativeness of Swedish MPs 1985-1994

The development of the political representativeness of Swedish MPs between 1985 and 1994 is presented in Table 1. The entries are the average left-right self-placement on a scale from 0 to 10 in the respective groups. The representativeness of the MPs is shown in the 'difference' row at the bottom, which indicates the differences in average left-right self-placement between the MPs and the voters of their respective parties.

There is no general decline in the representativeness of Swedish MPs. In
fact, the gap between MPs and voters has, for the most part, shrunk. This is

particularly evident in the Center Party, but also to some extent in the Left Party and the Social Democrats. The People's Party Liberals and the Green Party display no particular trend. The only clear case of a decline in representativeness is the Moderate Party. In the 1980s, there was no statistically significant difference between the party's MPs and voters, but in 1994 the difference had increased to almost one full scale step. The Moderates are, however, the exception to the rule. Table 1 offers no support for the allegation that Swedish parliamentarians have moved away from the general public. The least representative MPs can be found in the Left Party; even if their representativeness has increased, the difference is consistently more than one scale step. Other research has shown that

differences of more than one full scale step in average left-right selfplacementarerelatively between different levels in the same party (see, e.g., Widfeldt 1995). Also the Social Democrats and People's Party Liberals display statistically significant differences on every time point. The most representative MPs can be found in the Christian Democrats, for whom data are only available for 1994.

A negative difference indicates that the MPs are to the right of the voters, and a positive difference indicates that the MPs are to the left of the voters. As can be easily seen, Swedish parliamentarians tend to be to the left of their respective party voters. This applies to all parties except the Moderates, whose MPs tend lean to the right, even if the difference is significant only in 1994. This pattern is closest to the 'fan,' or elite conflict, model, where MPs of leftist parties are to the left of their voters, and vice versa on the right. What this means is that there is a higher degree of polarization at the parliamentary than the electoral level. The difference between left and right is more pronounced in the political elite than among the voters. This is illustrated more clearly in Figure 1 below. As can be seen, the 'fan' model is quite weak in 1985 and 1988, because in the only case of MPs being to the right of the voters, the Moderate Party, the differences are not significant. In 1994, however, the elite conflict model is unambiguous.

Table 1 indicates that the Swedish parliamentarians have mostly moved towards the right during the nine year research period. The differences are for the most part not massive, but there was some movement towards the right between 1985 and 1994 in all parties except the People's Party Liberals. The trend is continuous in the Left Party and the Social Democrats; other words, the two leftmost parties have actively contributed to a rightward shift of the Swedish Riksdag. The voters have not, however, followed this trend. In most parties, there is no trend at all on the electoral level. What has happened in the Left Party and the Social Democrats is thus that the MPs have moved from left to right, while the voters have almost stood still. Since the voters of both parties have always been to the right of their respective MPs, the representativeness of the MPs has increased. As can be seen in Figure 1, the lines connecting voters and MPs in the Left and Social Democratic parties gradually approach vertical positions over time.

The 1985-1994 period was characterized by change and ideological reassessmentboth the Left Party and the Social Democrats. The latter party made several policy changes in the late 1980s, which were interpreted as moves away from traditional socialist values. The tax reform, carried through parliament together with the People's Party Liberals, which significantlyreduced progressiveness of the income tax, is an example. In the Left Party, the collapse of the communist regimes in Eastern Europe led to a reassessment of the Marxist-Leninist ideological baggage. A major consequencewas change of the party name, where the word 'Communists'

was dropped, at the 1990 party congress. Thus, it could be argued that both the Left Party and the Social Democrats moved ideologically towards the right between 1985 and 1994. The evidence in Table 1 and Figure 1 suggests that the parliamentarians of both parties have moved accordingly along the left-right continuum. The voters, however, have remained in a position closer to the middle. Therefore, it could be argued that the ideological reorientation of the Social Democrats and the Left Party, and the apparent adjustment among the parties' respective MPs, has moved both parties closer to their respective voters. This is quite interesting, not least in the light of the debate in the Social Democratic Party, where an internal opposition has claimed that the party has betrayed its ideals and its voters. The evidence presented here does not give the final answer, because much of the 'betrayal debate' is focused on concrete policy changes. It does lend some support to the allegation that the party elite has moved towards the right, but it is equally important to note that the ideological gap between Social Democratic MPs and voters appears to have decreased.

In the Center Party, the voters have moved to the left. The change is not huge, but there is a continuous trend. The MPs have moved back and forth, but were in 1994 further to the right than at the two earlier time points. As with the Social Democrats and the Left Party, the consequence has been that the Center Party MPs were more representative of the party's voters in 1994 than before. In 1985 and 1988, the MPs were to the left of the party's voters, but in 1994 the difference was no longer significant. This convergencejust the right of the mid-point of the left-right continuum has taken place during a period where the party tried to manifest its independence of the other non-socialist parties with which they had formed three different coalition governments between 1976 and 1982. This policy of relative independence,and of especially the Moderate Party, continued also after the Center Party had joined a Moderate led four-party government in 1991, and was highlighted when party leader Olof Johansson left the government shortly before the 1994 election. Much of the distancing from the right was from a green rather than leftist perspective - indeed, Johansson's resignation was over the construction of a bridge between Sweden and Denmark - but the party also dissented from the Moderates on welfare issues. Subsequently, the Center Party formed a parliamentary cooperation pact on economic issues with the Social Democrats between 1995 and 1997. It is interesting that the Center Party MPs moved slightly towards the right between 1988 and 1994, when the party's policies appeared to shift in the opposite direction. There is no ready explanation for this, but the evidence in Table 1 and Figure 1 suggests that the Center Party MPs became more representative of their voters during the process of policy adjustment. It should be noted, however, that the party lost many voters during the 1985-1994 period, so the move towards a higher degree of

left-right coherence has not prevented a substantial loss of electoral
support.

In certain respects, the People's Party Liberals went through a development to that of the Center Party during the nine year research period. The 1985 election was extremely successful, largely due to the popular party leader Bengt Westerberg. At that time, the party's policies were solidly right-of-center, with the critical edge directed towards the Social Democrats. While the criticism against the Social Democrats never ceased, there was a gradual movement towards the left after the 1988 election. It was never a question of a far-reaching ideological change, but there was a shift in emphasis. Westerberg became one of the hardest critics of the populist right New Democracy party when it appeared in 1991, and he also maintained a critical distance to the Moderates in the 1991-1994 coalition government. However, nothing suggests that the adjustment of the party's policies has had any major impact on the left-right positioning of the party's voters and MPs or on the MPs' representativeness. Throughout research period, there is a statistically significant distance between the MPs and the voters, with the former to the left of the latter, but there is no trend in the size of this difference. The voters were further to the left in 1994 than in 1988, but the difference is marginal if the comparison is between 1994 and 1985. There is virtual stability in the position of the MPs and it is also worth noting that the party's electoral support was halved between 1985 and 1994. However, this does not appear to have had any effect on the average left-right self-placement of the party's voters or MPs.

The Moderate Party is a deviant case as the only party where the MPs have become less representative over time. In 1985 and 1988, the parliamentarians well as the voters stayed in exactly the same positions, and there was no significant difference between the two levels. In 1994, the MPs had moved to the right and the voters towards the middle, so that the former were now significantly to the right of the latter. It is difficult to interpret this change in the light of any political development in the Moderate Party. Throughout the period, the Moderates maintained a 'modern conservative' profile with lower taxes, privatization and monetarist economic policies as key ingredients. The party had to make concessions to the coalition partners during the 1991-1994 government, but still implemented key policies, such as the privatization of public utilities. The party's MPs moved to slightly the right between 1988 and 1994, but whether this was a reflection of the experiences in the coalition government is difficult to say. The fact that the voters had moved to the left in 1994 could, to some extent, be the result of the increased electoral support compared to 1988. Many of the new voters came from parties to the left of the Moderates, and it is possible that they were reluctant to see themselves as being too far to the right.

The question to be answered in this section is not whether Swedish parliamentarians representative. This would require some sort of standard against which the differences between MPs and voters could be judged. It could be argued that Swedish MPs are not representative of the voters because there is usually a statistically significant difference between them. However, the main question was whether the Swedish MPs have become less representative over time or not, and the answer is no. There is no support for the allegation that the Swedish Riksdag has distanced itself from the electorate. There are traces of a rightward movement among the parliamentarians but, with the exception of the Moderate Party, that has only made them more representative.

Political Representativeness of Members and Activists, 1988-1994

It is sometimes assumed that party members are unrepresentative in their views. If the distance in opinion between party members and voters is too wide, there is a potential conflict between internal party democracy and societal democracy. McKenzie's famous exclamation that party democracy is incompatible with societal democracy is the best known exponent of this perspective (McKenzie 1982, 195). The view of party members as politically radical is, by and large, supported by other research. A study of 37 European at the end of the 1980s showed that the 'fan,' or elite conflict, model is also applicable to party members. In other words, that members of a leftist party are to the left ofthat party's voters, and vice versa on the right (Widfeldt 1995, 166-70). This observation also applies to Sweden throughout period between 1968 and 1994 (Widfeldt 1999, 249-57). Membership can be a democratic problem: Should party leaderships and parliamentarians listen to the radical members or the more moderate voters? The problem is particularly delicate for the MPs who are primarily accountable to the voters, but it is the party members who decide whether they are going to be renominated for the next election.

Earlier studies using left-right self-placement have compared only two levels, either MPs and voters (e.g., Esaiasson & Holmberg 1996; Holmberg 1996) or party members/activists and voters (Widfeldt 1995; 1999). This is relevant and legitimate, but cannot approach the dilemma discussed here. In order to shed some light on the conflict between societal and party democracy,at three levels will have to be studied: the elite level (MPs), the intermediate level (members) and the mass level (voters). To further lift the analytical value, the members can be separated into active and passive members. This is in line with May's well-known distinction between differentparty where the voters are the lowest and the leadership the

highest level. May's argument is that it is intermediate party strata, members and activists, that can be expected to be most ideologically radical. The reason is, among other things, that members have no material incentives, such as election or re-election to important public positions. Instead, the reason for their political involvement is ideological conviction which, unlike for parliamentarians, is not checked by any strategic considerationsaiming re-election. The result is a curvilinear ideological pattern in the party, referred to as 'May's Law of Curvilinear Disparity' (May 1973; Kitschelt 1989; Norris 1995). May does not subject his law to a far-reaching empirical test. Such a test is, however, provided by Norris' (1995) study on the British Labour and Conservative parties which came to a negative conclusion: There is no empirical support for May's Law in the leading British parties.

The ideological profile of the Social Democrats and the Center and Moderate Parties between 1985 and 1994 can be found in Figure 2. The remaining parties have been excluded, because the data include too few party members to allow statistical analysis. The figure includes the average left-right self-placement of four different party strata: MPs, active members, members and voters. Activists are operationalized as members who attend meetings a few times a year. This is, admittedly, a generous operationalization. However, it is a limitation that comes with secondary analysis of already collected data, where one is dependent on the operationalizations exist. It should also be noted that the number of cases already is low with this generous operationalization of activity, which means that it would be methodologically very difficult to use more restrictive when analyzing data from samples representing whole electorate. Passive members are, consequently, those who have not answered yes to the question whether they attend party meetings a few times a year.⁵ The inclusion of four levels in the analysis is sufficient for a test of May's Law (Norris argues that at least three levels are necessary (Norris 1995, 34)). If May's predictions are to be confirmed, the active members ought to be the most radical stratum. They are what May would call 'sub-elite activists'; that is members with little chance of a political career. Since such members have no rational incentives to move closer to the voters, but are politically active purely out of conviction, they can be expected to be the most radical party stratum (May 1973, 148ff).

The Social Democrats is the party that is most consistent with May's Law. On every time point, the activists are the most radical level, in other words place themselves furthest to the left. On the whole, the Social Democratsfit Law very well, even if the differences between the party strata are not always statistically significant. As mentioned in the previous section, the Social Democratic MPs moved towards the right between 1985 and 1994. During the same period, the party's active members have remained

almost totally static. The consequence is that the activists have, in relative terms, become more radical compared to the party's MPs as well as voters. Thus, the Social Democrats have, to an increasing extent, become compatiblewith Law. Any conclusions about the other two parties will have to be made with extreme care due to the low number of cases. There is only limited support for May's Law in the Moderate Party. In 1988, the patternwas with May's predictions. In 1985, however, the passive members were furthest to the right, and in 1994 the party's MPs, activists and passive members were all in exactly the same position. At no time point is there a statistically significant difference between the party's passive and active members. With a slightly more generous interpretation, however, the applicability of May's Law on the Moderate Party cannot be entirely discarded. Both the passive and active members were to the right of MPs as well as voters in 1985 and 1988, even if the difference between activists and MPs was significant only in 1988. In 1994, however, the difference between members and MPs disappeared. This was partly due to a radicalization of the MPs, but also due to the fact that the party members and activists went in the opposite direction, that is from right to left.

The Center Party is more difficult to interpret. To begin with, it is not exactly a straightforward question what 'center radicalism' is. It may be argued that it is to be as close to the center as possible. That, however, would be a simplification. On several issues, for example labor market relations and taxation, the Center Party is arguably still 'right of center,' despite the changes discussed in the previous section. On the other hand, the Center Party also stands for policies such as decentralization and environmental which are difficult to apply to the left-right schema. It could therefore be that May's Law would be more applicable to the Center Party if self-placement along a 'green-gray' scale was used instead. It should be noted, however, that 'green' politics tend to consist of valence issues, which provoke little open disagreement, while left-right issues are conflict, or position, issues (Butler & Stokes 1974, 292). Thus, it is doubtful whether positions along a green dimension are applicable to May's Law; indeed, May's discussion is explicitly in terms of left and right.

These conceptual problems do not prevent the Center Party from giving us some interesting observations. As could be expected, the Center Party is in its entirety safely tucked in the middle between the two flank parties, the Social Democrats and the Moderates. At all levels and all time points, however, the Center Party is slightly to the right of the exact mid-point, 5.0. If we therefore accept that the Center Party, despite having several diverse ideological ingredients, after all is slightly right-of-center, May's Law would mean that the members should be to the right of the MPs and the voters. There is some support for this in 1985, even though the differences between voters and passive and active members are not significant. In 1988,

however, the passive members are marginally to the left of the voters, which is not consistent with May's Law for a right-of-center party. In fact, the most striking impression of the Center Party is the lack of anything resembling a systematic relationship between the different party strata. In 1994, there was a zigzag pattern, which is not consistent with the Law. Another observation not consistent with May's Law is that Center Party MPs tend to be closest to the middle of all party strata. This was especially true in 1985 and 1988, while the differences had decreased in 1994 when the active members were in an almost identical position to the MPs. A somewhattentative would be that the Center Party MPs had, for some time, been ideologically prepared for the parliamentary pact on economicissues the Center Party and the Social Democratic government,which in the spring of 1995. The party's members and activists had been further to the right, but by 1994 they too had moved towards the middle, which meant that the party as a whole was now more prepared to cooperate with the Social Democrats.

Thus, there is some support in Figure 2 for May's prediction that party members and sub-elite activists are the most radical party strata. This applies in particular to the Social Democrats, but also to some extent to the Moderate Party. This is not consistent with Norris' findings on the British parties, but remember that her study was based on objective indicators, and this article is solely based on subjective ideological self-placement. Damgaard & Kristensen (1982, 43f) used subjective left-right self-placement in a study on ten Danish parties and found patterns clearly consistent with May's Law in the Socialist People's, Liberal and Conservative parties. In the Social Democrats, the MPs were furthest to the left. The Liberal party Venstre, often regarded as the closest equivalent to the Swedish Center Party, displayed a pattern in line with May's Law if the party is regarded as right of center. The time difference (the Danish study used data from 1979) means that any comparisons must be tentative; however, if the same label is applied to the Swedish Center Party, there is an interesting contrast between the two sister parties because there is no support for May's Law in the Swedish party.

As mentioned in the previous section, Table 1 and Figure 1 do not lend much support to the notion of a widening gap between voters and parliamentariansin Figure 2 shows that there are increasing ideological differences between members and voters in two of the three studied parties. This is more apparent in Table 2, where the differences between every party level are specified. In the Social Democrats and the Moderates, both active and passive members were further away from their respective voters in 1994 than before. This means that the electorally two biggest Swedish parties are subjected to potential cross-pressure from their radical members and activists, on the one hand, and their more middle-of-the-road voters on the

other. The situation would appear to be most complicated in the Social Democratic Party where there are clear signs of membership radicalization over time. This brings us back to the discussion about the dilemma between party democracy and societal democracy. According to the latter, the voters' opinion should be the most important, and according to the former the party members should take precedence. Thus, the observations in Table 2 could mean a dilemma for parliamentarians. The Social Democrats face more resistance internally than from the voters when it comes to, for example, reforming the public sector. Conversely, the Moderate members and activists could be expected to want to go further in, for example, tax cuts than the party's voters. If the parties choose to follow the members, they risk losing votes. If they decide to follow the voters, they risk loss of members and increased internal unrest. If we add the faet that party membership and activism are declining in Sweden, the results in Figure 2 and Table 2 could be interpreted very negatively. There is some justification to speak of a shrinking hard core of ideologically radical party activists, who are influential in the formally democratic party organizations. If this is a democratic problem, is ultimately a normative question. It could be argued that it is good for democracy that ideologically committed people become involved in democratic organizations in order to promote new ideas. However, membership radicalism combined with the decline in membershipand could be seen as a democratic liability.

There is a methodological problem with these findings: That voters are less radical could be explained by a 'regression to the mean' phenomenon, where less politically interested and committed citizens can be expected to be less likely to express political radicalism. Looking again at Figure 1, it is indeed true that the voters are closer to the middle than the MPs every year in the Left Party, Social Democrats and the Moderate Party. In the Center and Liberal parties, however, the MPs are closest to the mid-point every year, and in the Green Party this was the case in 1994. Figure 2 shows that the voters were closest to the mid-point of all four party strata for the Social Democrats and the Moderates on every time point. In the Center Party, however, it was the MPs in 1985 and 1988, and in 1994 the activists, who were the closest to the middle (although no intra-party differences in the Center Party were significant in 1994). Even if the 'regression to the mean' tendency appears to apply at least to parties commonly referred to as left or right, it is still relevant to study whether the size of the difference between voters and other party levels increases or decreases over time. And, indeed, the gap between voters and party members has increased in the Social Democratic Party and in the Moderate Party. In the Social Democratic Party, primarily the passive members have moved away from the voters, while there is a weak tendency in the same direction for the activists. In the Moderate Party, the distance between voters and passive as well as active

members has increased, even if the rate of change is marginal. Thus, there
is some justification to speak in terms of an increasing gap between the
party organizations and the general public.

That the Social Democratic members and activists have been radicalized relative to the voters is an important and interesting finding in its own right. First, because even if there is no general increase in membership radicalism in all the three studied parties, it is of course not unimportant that such a

development can be observed in the party with the highest number of voters as well as members. Second, because it suggests that there could be a relationship between a decline in membership and activity, on the one hånd, and membership radicalism on the other, in parties on the left. While one should beware of far-reaching conclusions based on the relatively limited evidence presented, it does seem as if the Social Democrats have retained a hard core of radical activists. In faet, also the passive members had moved to the left in 1994. A tentative explanation could be taken from Damgaard & Kristensen (1982, 40). Referring to Hirschman (1970), they argue that the existence of Voice in a party organization gives the more radical members the opportunity to stay in the party, rather than Exiting. This leads to a centrifugal, rather than centripetal, ideological pattern in the party. Followingthis it is possible that the most ideologically committed members of the Swedish Social Democrats have responded with Voice rather than Exit to the many difficult challenges faced by the party in the 19905. Another interesting observation is that the clearest case of membershipradicalization taken place in the party which best fits the traditional 'mass,' 'party democracy' or 'mass bureaucratic' models (Duverger 1964; Wright 1971; Panebianco 1988). The Social Democrats are, also after the abolition of collective membership in the late 1980s, by far the biggest membership party in Sweden, and has close traditional links to tråde unions and the labor movement. In recent years, however, the union links have become increasingly strained, and the evidence in Figure 2 and Table 2 suggests that the policies that have led to a strained relationship between the party and the unions have also been met with membership resentment.

Conclusion

Much of the debate in political science over, at least, the last decade and a half, has dealt with the crisis of political parties (see, e.g., Katz & Mair 1995; Webb 1995; Schmitt & Holmberg 1995; Håkansson 1995; Pierre & Widfeldt 1995; Back 1995). Crisis is, of course, a problematic concept. To merit the label 'crisis,' the observed situation has to relate to some defined criterion. One way of doing this is to set up criteria for an 'acceptable' situation, and test the observed data against that. In this vein, it could be said there is a representative crisis in Sweden, since MPs and members tend to be ideologicallyseparated the voters to a statistically significant extent. Another way of approaching the problem is to study the development over time. If a decline can be observed in something which is considered of central democratic importance, it could be possible to speak in terms of crisis. Seen that way, the findings in this article do not support the notion of a crisis at the

parliamentary level. In most parties, MPs are separated from the voters in terms of left-right self-placement, but there was no general increase in the discrepancy between 1985 and 1994. On the other hand, there are signs of membership radicalization in the Social Democratic Party and the Moderate Party. Thus, while there was no general decline in the representative linkage provided by the Swedish parties between the mid-1980s and mid-19905, there is some justification to speak of a representational decline for the Swedish parties as membership organizations during the same period.

---

NOTES

1. Comparisons between the left-right self-placement of Swedish MPs and Swedish voters are previously published in Holmberg & Esaiasson 1988 (data from 1985), Esaiasson & Holmberg 1996 (data from 1985 and 1988) and Holmberg 1996 (data from 1994).

2. The metaphors corn-field (in Swedish, sädesfälf) and fan (in Swedish, solfjäder) relate to the patterns that emerge if the left-right positions of the voters and the parliamentarians placed along identical horizontal scales, with the parliamentarian scale above the voter scale. If the positions of the voters and parliamentarians of the same party are connected with lines, the resulting pattern can be likened to a corn-field, with all connecting leaning in the same direction, or a fan, with the connecting lines on the right leaning to the right, and vice versa on the left. The 'fan' pattern is illustrated in Figure 1.

3. The scale used has eleven points, between 0 and 10, where the former is extreme left and the latter extreme right. In the Election Studies, which are basedon in-person interviews, the interviewers describe the scale to the respondents, and also show a card which displays the scale. In the parliamentary studies, which are basedon mail questionnaires, the scale is described in writing as well as graphically. In both studies, the mid-point 5 is explicitly described as 'neither left nor right.'

4. The survey questions used in this article are as follows: Election Studies: 1985 q. 30, 1988 q. 28 and 1994 q. 26. Parliamentary studies: 1985 q. 20,1988 q. 23 and 1994 q. 15.

5. For a discussion on the operationalization of party activity in the Swedish Election Studies, as well as other survey data, see Widfeldt 1999, 137-39.

REFERENCES

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

Arter, D. 1999a. Scandinavian Politics Today. Manchester: Manchester UP.

Arter, D. 1999b. 'Sweden: A Mild Case of "Electoral Instability Syndrome?",' in Broughton,
D. & Donovan, M., eds., Changing Party Systems in Western Europe. London: Pinter.
Bennulf, M. 1994. Miljöopinionen i Sverige. Stockholm: Dialogos.

Brown, A., McCrone, D., Paterson, L. & Surridge, P. 1999. The Scottish Electorate. The 1997
General Election and Beyond. Basingstoke: Macmillan.

Butler, D. & Stokes, D. 1974. Political Change in Britain. The Evolution of Electoral Choice
(2nd ed.). London and Basingstoke: Macmillan.

Back, H. 1995. 'Partikrisen, den nya politiken och den nya höeern,' Statsvetenskaplig Tidskrift
98,51-54.

Dalton, R. J. 1996. Citizen Politics in Western Democracies. Public Opinion and Political Parties
in Advanced Western Democracies (2nd ed.). Chatham: Chatham House.

Damgaard, E. & Kristensen, O. 1982. 'Party Government under Pressure.' in Anckar. D..
Damgaard, E. & Valen, H., eds., Partier, ideologier, våljare. En antologi. Turku: Publications
of the Research Institute of the Åbo Akademi Foundation.

Duverger, M. 1951/1964. Political Parties. London: Methuen.

Esaiasson, P. & Holmberg, S. 1996. Representation from Above. Members of Parliament and
Representative Democracy in Sweden. Aldershot: Dartmouth.

Fuchs, D. & Klingemann, H.-D. 1990. 'Partisanship and Political Behavior,' in Jennings, M.
K. & van Deth, J. W., eds., Continuities in Political Action. Berlin: de Gruyter.

Gilljam, M. & Holmberg, S. 1995. Väljarnas val. Stockholm: Norsteds Juridik.

Holmberg, S. 1974. Riksdagen representerar svenska folket. Empiriska studier i representativ
demokrati. Lund: Studentlitteratur.

Holmberg, S. 1994. 'Election Studies the Swedish Way,' European Journal of Political Research
25, 309-22.

Holmberg, S. 1996. 'Svensk asiktsöverensstämmelse,' in Rothstein, B. & Särlvik, 8., eds.,
Vetenskapen om politik. Festskrift till professor emeritus Jörgen Westerståhl. Göteborg:
Statsvetenskapliga institutionen, Göteborgs universitet.

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

Huber, J. D. & Bingham Powell, G. 1992. Congruence between Citizens and Policymakers in
Two Visions of Liberal Democracy. Ann Arbor and Rochester: University of Michigan and
University of Rochester (working paper).

Håkansson, A. 1995. 'Kris för de svenska partierna?,' Statsvetenskaplig Tidskrift 98, 45-51.

Kitschelt, H. 1989. 'The Internal Politics of Parties: The Law of Curvilinear Disparity
Revisited,' Political Studies 37,400-21.

May, J. D. 1973. 'Opinion Structure of Political Parties: The Special Law of Curvilinear
Disparity,' Political Studies 21, 135-51.

McKenzie, R. T. 1982. 'Power in the Labour Party: The Issue of Intra-Party Democracy,' in
Kavanagh, D., ed., The Politics of the Labour Party. London: George Allen and Unwin.

Norris, P. 1995. 'May's Law of Curvilinear Disparity Revisited: Leaders, Officers, Members
and Voters in British Political Parties,' Party Politics 1, 29-47.

Norris, P. & Lovenduski, J. 1995. Political Recruitment. Gender, Race and Class in the British
Parliament. Cambridge: Cambridge University Press.

Panebianco, A. 1988. Political Parties: Organization and Power. Cambridge: Cambridge UP.

Pitkin, H. F. 1967/1972. The Concept of Representation. Berkeley and Los Angeles: California
UP.

Rothstein, B. 1998. 'Breakdown of Trust and the Fall of the Swedish Model.' Paper presented at the 1998 Annual Meeting of the American Political Science Association, Boston, 3-6 September 1998, Panel on Social Capital and Social Policy in Advanced Industrial Democracies.

Schmitt, H. & Holmberg, S. 1995. 'Political Parties in Decline?,' in Klingemann, H.-D. &
Fuchs, D., eds., Citizens and the State. Oxford: Oxford University Press.

Webb, P. D. 1995. 'Are Political Parties in Decline?,' Party Politics 1, 299-322.

Westerståhl, J. & Johansson, F. 1981. 'Medborgarna och kommunen. Studier av medborgerlig
aktivitet och representativ folkstyrelse,' Ds Kn 1981, 12.

Widfeldt, A. 1995. 'Party Membership and Party Representativeness,' in Klingemann, H.-D.
& Fuchs, D., eds., Citizens and the State. Oxford: Oxford University Press.

Widfeldt, A. 1999. Linking Parties with People? Party Membership in Sweden 1960-1997.
Aldershot: Ashgate.

Wright, W. E. 1971. 'Comparative Party Models: Rational-Efficient and Party Democracy,'
in Wright, W. E., ed., A Comparative Study of Party Organization. Columbus, Ohio: Charles
E. Merrill Publishing Company.