Resumé

The left-right positions of the political parties in Sweden, Norway, Denmark, Finland, and Iceland are compared from the late 1970s to the beginning of the 19905. To locate the parties, survey data on the voters' self-placements along the left-right continuum are used. In order to study changes in the left-right polarity and the degree of consensus along the left-right continuum in each of the countries, we use the mean party positions to calculate three different measures of party distances. The wing party distance is that between the party farthest to the left and the party farthest to the right. The rival party distance is that between the Social Democratic Party and the traditional Conservative Party. Finally, the mean party distance is the average distance between all pairs of parties. One of the main conclusions is that in Sweden and Iceland the left-right continuum seems to contract, whereas in Norway and Denmark the left-right polarity and the distances between the parties are increasing. In today's Nordic party space, the distance between left and right is longest in Denmark and shortest in Norway. Eventually, 39 Nordic parties are brought together on the same left-right scale. The analysis reveals that there are some clearly distinguishable clusters of parties or party families in the Nordic countries, such as, for example, the socialist parties, the social democratic parties and the conservative parties. Other party groups differ greatly in left-right position, like the progressive parties, the liberal parties and the centre parties.

Introduction

The golden age of mapping the Nordic voters' perceptions of the party space goes back a good 20 years (e.g. Converse 1966; Särlvik 1968, 1976; Converse & Valen 1971; Pesonen 1973; Rusk & Borre 1976; Worre 1976; Petersson & Valen 1979; Holmberg 1981). After a period of few advances, there are now signs of a renewed interest (e.g. Hardarson & Kristinsson 1987; Nannestad 1989; Tonsgaard 1989; Bennulf & Holmberg 1990; Gilljam & Holmberg 1990; Listhaug, Macdonald & Rabinowitz 1990, 1994; Rabinowitz, Macdonald & Listhaug 1991; Gilljam & Holmberg 1993; Oscarsson 1995; Pesonen, Sänkiaho & Borg 1993; Bennulf 1994). The most central question today is to what extent the left-right dominance and the unidimensionality of the 1960s and 1970s have been replaced by a party space with two or even more dimensions.

By using different types of dimensional analysis, scholars have confirmed
that the left-right dimension is today more subject to significant competition
than in the previous two or three decades. The left-right dimension still

dominates when voters are asked to evaluate the parties on some sort of feeling thermometer. But a uni-dimensional party space is no longer completely sufficient to grasp how the voters perceive the distances and the competition between the parties. An awkward problem in this regard is to interpret, in a meaningful way, the party positions along possibly new dimensions (e.g. Nannestad 1989; Listhaug, Macdonald & Rabinowitz 1990; Oscarsson 1993; Bennulf 1994).

In this article, we do not deal with possible new political dimensions and what these might represent. We prefer to stay on more solid ground, and have therefore limited the analysis to the still-dominant left-right dimension. With respect to geography, we are, however, more inclusive: the following analysis of the voters' perceptions of party positions along the left-right dimension includes all five Nordic countries. The aim is to bring together all the parties in the five countries along the same left-right scale.

Since our data cover the period from the 1970s to the 19905, we also deal with the question whether the left-right dimension in the different countries is contracting - which in such a case could indicate a larger degree of consensus and cooperation in politics. Three different measures of the party space will be used: (1) the wing party distance (WPD), which is the distance between the party farthest to the left and the party farthest to the right; (2) the rival party distance (RPD), which is the distance between the Social Democratic Party and the traditional Conservative Party; and (3) the mean party distance (MPD), which is the sum of all distances between pairs of parties divided by the total number of distances.¹

There are basically three ways to determine how the voters perceive the party locations in the party space. A dimensional analysis, built on how the voters evaluate the parties on a feeling thermometer, always gives a solution open to interpretations. In such an analysis, the voters' evaluations are summarized in one, two or possibly more dimensions, but it is generally unclear what type of dimensions these are. The ordering of the parties along the left-right dimension is recognized by most people. However, if the analysis unveils other dimensions, the number of questions rapidly increases.

Another, more direct method is to take the step from evaluations to cognitions, and focus on how the voters place the parties along preconstructed scales. In the Nordic countries, the left-right scale is of course the most frequently used (e.g. Särlvik 1976; Holmberg 1981; Borre 1983; Hardarson & Kristinsson 1987; Aardal & Valen 1989). However, following the American tradition, Norwegian scholars have asked the voters to place the parties along several other types of issue dimensions as well (e.g. Aardal & Valen 1989, 1993; Macdonald, Listhaug & Rabinowitz 1991).

A third possibility to obtain a party map of how the voters perceive the

party positions is to follow the roundabout way of asking how the voters place themselves along different scales. A party position along a scale is, in this case, the average self-placement for the people voting for the party. Since we wish to compare all five Nordic countries, and since the most adequate data do not exist, we have chosen to operate in this roundabout manner. The following analysis is built, in other words, on the Nordic voters' self-placements along the left-right dimension.

Data

The data for this analysis are taken from the election studies in Sweden (1979, 1982, 1985, 1988 and 1991), Norway (1977, 1981, 1985 and 1989), Denmark (1977,1979,1984 and 1990), Iceland (1987 and 1991), and Finland (1991). Unfortunately, there are no relevant data from the Danish elections of 1987 and 1988, and in the Finnish case the time-series is limited to one data point (1991).

The Swedish, Norwegian, Danish and Finnish election studies are carried out as face-to-face interviews, while the Icelandic scholars rely upon telephone interviews. The Norwegian, Danish and Icelandic studies are postelection studies, which means that the interview question on party choice has been used to group the respondents according to political affiliation. In Sweden, half of the interviews are conducted before the election, and the other half after the election; therefore the respondents have been grouped according to party preference and not according to party choice. The Finnish study is a panel with both pre- and post-interviews, and since the question on left-right self-placement is only included in the pre-election study, we again rely upon party preference.

The interview question on left-right placement differs somewhat between the countries. A delicate problem in this regard is that the number of points along the left-right scale varies between different studies. Sweden and Iceland use an 11-point scale (0-10), Denmark and Finland use a 10-point scale (1-10), while Norway first used a 9-point scale, and from 1985 onwards a 10-point scale (1-9 and 1-10 respectively). In the following analyses, the Danish, Finnish and Norwegian scales have been transformed to 11-point scales (0-10).²

Sweden

The voters' map of the Swedish party positions along the left-right scale
1979-91 - operationalized as the average self-placement for groups of

people with the same party preference - can be found in Fig. 1. On the 1979 scale, we find the five traditional parliamentary parties - the Left Party (left), the Social Democratic Party (sd), the Centre Party (c), the Liberal Party (lib) and the Conservative Party (con) - and the Christian Democratic Party (chr), which was formed in 1964 but did not enter Parliament until 1991. On the 1982 scale, we also find the newly created Green Party/The Environmentalists (env); and on the 1991 scale New Democracy (nd) appears, which is the latest addition to the Swedish party bouquet.

According to the voters, all the parties, with the exception of the Greens, have moved toward the right during the last decade. The shift toward the right between 1979 and 1991 is largest for (in order) the Christian Democratic Party (0.9 units), the Left Party (0.6 units) and the Social Democratic Party and the Liberal Party (both 0.5 units). The Green Party's leftward shift was 0.5 units between 1985 and 1991. The largest change between two consecutive elections is the Christian Democratic Party's shift to the right between 1988 and 1991 (0.8 units).³

The main result for Sweden is a somewhat compressed left-right dimension
with shorter distances between the parties (cf. Gill jam & Holmberg

1993; Oscarsson 1993). The wing party distance (i.e. the distance between the Left Party and the Conservative Party) has decreased from 5.30 units in 1979 to 4.87 in 1991 (-0.43 units). The rival party distance (the distance between the Social Democrats and the Conservatives) has decreased from 3.93 in 1979 to 3.60 in 1991 (-0.33); the highest figure, however, was 4.12 in conjunction with the wage-earners' funds debate of the 1982 election. The mean party distance (the sum of the distances between all pairs of parties divided by the number of distances) has declined from 2.29 in 1979 to 1.88 in 1991 (-0.41).⁴ The shrinking distances between the Swedish parties indicate a larger degree of consensus along the left-right dimension and therefore, possibly, a larger potential for new political dimensions.⁵

Norway

The Norwegian voters' map of the left-right party positions contains, with one exception, the same parties as in Sweden, i.e. the Socialist Left Party (soc), the Social Democratic Party (sd), the Liberal Party (lib), the Centre Party (c), the Christian Party (chr), the Progressive Party (pro) and the Conservative Party (con). The newly created Green Party is missing from the scale because of its small size; in the 1989 election it only received 0.4 percent of the votes. The greatest differences in comparison with the positions of the Swedish parties are that the Liberal Party is situated relatively farther to the left and the Progressive Party relatively farther to the right (see Fig. 2).

The shifts in the Norwegian voters' party map between 1977 and 1989 have been toward the left in four cases and toward the right in three cases. Shifts to the left are observed for, in order, the Social Democratic Party (0.9 units), the Centre Party and Christian Party (both 0.5 units), and the Liberal Party (0.4 units). However, the movement of the Liberal Party is u-shaped, and hides the largest change of all - from 5.1 in 1977 to 4.0 in 1981, down to 3.5 in 1985, and back to 4.7 in 1989. The Progressive Party shows the largest shift to the right (0.8 units).⁶

Norway is certainly not a case of increased consensus and decreased party distances. On the contrary, the three measures of party space show an increased polarization. The wing party distance has grown from 4.49 in 1977 to 4.53 in 1989 (+0.04 units), the rival party distance has increased from 2.48 in 1977 to 3.55 in 1989 (+1.07), and the mean party distance has risen from 1.87 in 1977 to 2.84 in 1989 (+0.97). However, the largest distance in the Norwegian par⁺y space was not found in 1989 but in 1985, when the first right-wing led conservative government since 1928 had completed its first four-year period in office (WPD 4.91, RPD 3.64 and MPD 3.22).⁷

When changes along the left-right dimension in Norway and Sweden are compared, the three measures of distance show a decreased polarity in Sweden and an increased polarity in Norway. By the end of the timeperiod, the rival party distance is on the same level in both countries (3.60 and 3.55 respectively). The wing party distance, however, continues to be larger in Sweden than in Norway (4.87 versus 4.53). When it comes to the mean party distance, the countries have switched places. The average distance between the political parties is now significantly larger in Norway than in Sweden (2.84 vs. 1.88).⁸

Denmark

From the Danish election studies it is possible to map the positions of the nine largest parties along the left-right dimension. To the left of the Social Democratic Party (sd) we find the Socialist People's Party (spp) and even further to the left the Socialist Left Party/Left Alliance (soc/la), which was not represented in the Parliament following the 1990 election.⁹ Farthest to the right we have the Progressive Party (pro), and in the middle and to the

right of the middle the traditional non-socialist parties: the Radical Liberals (rl), the Christian People's Party (chr), the Agrarian Liberal Party (a), and the Conservative Party (con). The Centre Democrats (cd), a break-away from the Social Democratic Party in 1973, have no direct counterpart in the other Nordic countries. Note that Denmark, like Norway, does not have a significant green party (0.9 percent of the votes in the 1990 election).

In contrast to Sweden and Norway, the Danish Progressive Party is situated more to the right. Also in contrast to Sweden, but like Norway, the Danish Radical Liberals are relatively close to the Social Democratic Party. Even more notably for today's Danish party system is that the old farmers' party, the Agrarian Liberals, is found to the right of the Conservative Party (see Fig. 3).

The two greatest changes on the voters' map of the Danish party space are the Agrarian Liberal Party's and the Progressive Party's shifts toward the right by a total of 1.1 units between 1979 and 1990. The Socialist Left Party also showed a relatively large shift toward the right (0.7 units); this change, however, should be viewed in light of the party's election cooperation with the Communist Party in the 1990 election. Three Danish parties have moved somewhat toward the left during the period - the Centre Democrats (0.5 units), the Socialist People's Party (0.3 units) and the Social Democratic Party (0.2 units).

In Denmark, the distances between the political parties have increased,
precisely as seen in Norway. The wing party distance has grown, from what

was previously a Nordic record of 5.87 in 1979, to a new record level of 6.29 in the 1990 election (+0.42). The rival party distance has increased by the same amount - from a relatively modest value of 2.64 in 1979 to 3.06 in 1990, which is still approximately half a unit less than in Norway and Sweden. The average distance between the parties (MPD) has also grown by a slim half unit - frcm 2.98 in 1979, surpassed only by the Norwegian record of 3.22 in 1985, to the new Nordic record of 3.38 in 1990 (+0.40).

Iceland

The voters' map of the Icelandic party space in the elections of 1987 and 1991 is displayed in Fig. 4. The four traditional parties in Iceland are the large non-socialist Independence Party (ip), the agrarian Progressive Party (pp), the relatively small Social Democratic Party (sd), and the relatively large left-socialist People's Alliance (pa). Iceland differs from the other Nordic countries due to its lack of a liberal party and, perhaps most importantly, because of its unique, combined women's and environmental party, called the Women's Alliance (wa), which has been represented in the Parliament since 1987. In the election of 1987, also the Citizens' Party (cip) and the Union of Regional Equality (ure), (break-aways from the Independence Party and the Progressive Party, respectively) were represented in the Parliament. But in contrast to the Women's Alliance, these newcomers were not successful in repeating their success in the election of 1991.¹⁰

According to the Icelandic voters' party map, the parties to the left of the middle (the People's Alliance and the Women's Alliance) have shifted toward the right, whereas the parties to the right and in the middle (the Independence Party, the Progressive Party, and the Social Democratic Party) have moved toward the left. The two largest changes between 1987

and 1991 are the leftward movement of the Progressive Party (0.5 units) and the rightward movement of the People's Alliance (0.3 units). The result of these changes is that the Progressive Party is found just to the left of the Social Democrats in 1991.

Just as in Sweden - but unlike Norway and Denmark - the distances between the Icelandic parties along the left-right dimension have telescoped. The wing party distance (i.e. the distance between the People's Alliance and the Independence Party) has decreased from 5.12 in 1987 to 4.66 in 1991 (-0.46), while the mean party distance has declined from 2.07 in 1987 to 1.79 in 1991 (-0.28).ⁿ The rival party distance remains unchanged (2.28 in both 1987 and 1991), but owing to the fact that the position of the Social Democrats is relatively far to the right, the RPDvalue is already significantly lower than that in the other Nordic countries. The mean party distance in the last Icelandic election (1.79 in 1991) is clearly less than that in Denmark (3.38 in 1990) and Norway (2.84 in 1989) and a trace less than in Sweden (1.88 in 1991).

Finland

The Finnish party system contains, in principle, the same parties as the other three large Nordic countries. For Finland it is notable that there are two liberal parties, one being the Swedish People's Party (spp, 5.5 percent in the 1991 election) and the other the Liberal People's Party (lib, 0.8 percent in 1991). The other Finnish parliamentary parties, from left to right, are: the Left Union (lu), the Social Democratic Party (sd), the Green League (g), the Finnish Rural Party (frp), the Finnish Christian League (cl), the Finnish Centre Party (c), and the Conservative Party (con). The voters' map of the party positions along the left-right dimension in the 1991 election is presented in Fig. 5.

Questions on how the voters place themselves along the left-right dimension have unfortunately not been included in earlier Finnish election studies. Therefore, it is not possible to comment on whether the Finnish party space is changing. The wing party distance in Finland in 1991 (5.71) was less than that in Denmark, but larger than in the rest of the Nordic countries at approximately the same time. The rival party distance (the distance between the Social Democrats and the Conservative Party) was larger in Finland (3.84) than in the other countries, which is somewhat surprising in light of the fact that these two parties had participated in the same government in the years preceding the 1991 election. The mean party distance in Finland in 1991 (1.93) was in approximately the same range as that in Sweden and Iceland; that is to say, considerably less than that in Norway and certainly less than that in Denmark.

The Five Countries Compared

As with most political parties, those reviewed in this article have siblings, or at least relatives, in the other Nordic countries. These relations do not, however, need to indicate identical or almost identical positions along the left-right dimension. Therefore, it is now interesting to compare the party maps in the five countries: Which Nordic party is, according to the voters, farthest to the left and which is farthest to the right? Where on the map do we find the different social democratic parties? And how great are the differences between the traditional right-wing parties, the different protest parties, as well as the traditional farmer parties, respectively? The answers to these and similar questions can be found in Fig. 6, which shows the 39 party positions along the left-right dimension for Norway in 1989, Denmark in 1990, and Sweden, Iceland and Finland in 1991.

It is primarily Denmark which is responsible for the relatively high polarity along the left-right dimension in the Nordic party space. The Danish Progressive Party (position 8.4) takes the flank position farthest to the right even on the Nordic left-right dimension; the Danish Progressives are found a whole 0.9 units farther to the right than the three closest parties on the scale, which are the Swedish Conservative Party, the Finnish Conservative Party, and the Danish Agrarian Liberal Party (all at position 7.5). Farthest to the left we find the Finnish Left Union (1.8) and the Danish Left Alliance (2.1); just to the right of these two parties, the

remaining socialist left parties are situated - the Danish Socialist People's
Party (2.5), the Swedish Left Party and the Icelandic People's Alliance
(both at 2.6,) as well as the Norwegian Socialist Left Party (2.7).12

The largest empty space along the Nordic left-right scale is the distance of 1.0 units between the left-socialists and the social democratic parties. Farthest to the right amongst the left-socialists is the Norwegian Socialist Left Party (2.7) and farthest to the left amongst the social democratic parties we find the Norwegian and the Finnish social democratic parties (both at 3.7). The social democratic parties in the four largest countries are relatively well grouped around the 4.0 position, with the Norwegian and the Finnish parties farthest to the left (3.7), the Danish party farthest to the right (4.3), and the Swedish People's Party in-between (3.9). The Icelandic Social Democratic Party is clearly to the right of its Nordic brother parties. Iceland's representative in the social democratic span is, instead, the Women's Alliance (4.0) which has approximately the same left-right position as the Swedish Social Democrats.

The Nordic left-right party space further contains two, fairly well-connected, trans-national, party families. All five traditional conservative parties are found within the small interval of 7.3 to 7.5, with the Norwegian and Danish Conservative Parties and the Icelandic Independence Party a little to the left of the Finnish and Swedish Conservatives. The Christian parties are found in the interval of 5.7 to 6.4, with the Swedish Christian Democratic Party and the Danish Christian People's Party to the right (both at 6.4), the Norwegian Christian Party closer to the centre (5.7) and the Finnish Christian League in-between (6.0).

Within the remaining Nordic party families, there is a larger internal spread along the left-right dimension. The liberal parties are found in two discernible groups. The Finnish Swedish People's Party (6.6), the Swedish Liberal Party (6.4) and the Finnish Liberal Party (6.1) are situated clearly to the right of the Icelandic Citizen's' Party (5.2), the Danish Radical Liberals (5.1) and the Norwegian Liberals (4.7).

As for the protest parties and the traditional farmers' parties, it is uncertain whether one can speak of some type of party families. The protest parties have an interval range of three units, with the Danish Progressive Party farthest to the right (8.4), followed in turn by the Norwegian Progressive Party (7.1), the Swedish New Democracy (6.3) and the Finnish Rural Party (5.4). The farmers' parties range over almost as large an interval, with the Danish Agrarian Liberal Party farthest to the right (7.5), the Icelandic Progressive Party farthest to the left (4.8) and the Swedish, Finnish and Norwegian Centre Parties in-between (6.1, 6.0 and 5.2 respectively).

Finally, the results for the green parties are somewhat ambiguous. In
Iceland, there is no explicit environmental party and the Danish and

Norwegian Green Parties are too small to be included in this study. If we take the liberty to count the Norwegian Liberals and the Icelandic Women's Alliance as their respective country's environmental party, the green parties are found within a one-unit interval slightly to the left of the middle point - the Finnish Green League at 4.8, the Norwegian Liberals at 4.7, the Swedish Green Party at 4.5 and the Icelandic Women's Alliance at 4.0.

Mapping the Nordic Party Space

To a large degree, Denmark sets the framework for the voters' map of the Nordic party positions along the left-right dimension. The wing party distance in Denmark was as high as 6.29 in 1990, which is a good half unit more than in Finland in 1991 (5.71), and about one and a half units more than in Sweden in 1991 (4.87), Iceland in 1991 (4.66) and Norway in 1989 (4.53). The high degree of polarity in Denmark depends primarily on the position of the Progressive Party far to the right, while the relatively short wing party distance in Norway primarily depends on the Socialist Left Party's position somewhat to the right of the left-socialist parties in the other countries. The distance between left and right is, in other words, longest in Denmark and shortest in Norway.

Considering the rival party distance, variations between the five countries have a very different pattern. In the early 19905, the longest distance between the Social Democrats and the traditional Conservative Party is found in Finland (3.84 units), while the shortest distance is in Iceland (2.28), with Sweden (3.60), Norway (3.55) and Denmark (3.06) in-between. The results perhaps indicate that a government shift, signifying a power displacement from socialist to non-socialist, or the reverse, could have the relatively greatest political consequences in Finland and the least consequences in Iceland.

The mean party distance, lastly, can give a hint of eventual empty spaces for new parties between the existing flank parties, and may also provide a hint on how far the average potential party switcher must stretch if he chooses to leave his present party. The mean party distance is longest in Denmark (3.38) and in Norway (2.84), and least (in order) in Finland (1.93), Sweden (1.88) and Iceland (1.79). Denmark, with its already rich party bouquet, consequently could provide a good opening for further new additions between the Progressive Party and the Left Alliance. On the other hand, Iceland, Sweden and Finland, ceterisparibus, could be expected to show a high degree of voter volatility.

The longitudinal analysis has not resulted in any uniform answer to the
question of whether the party map in the Nordic countries has changed
toward conflict or consensus along the left-right dimension. In Sweden and

Iceland, the distances between the parties have shrunk, while the changes in the Norwegian and Danish party systems show an increased polarization. Developments in Finland remain to be outlined, since the 1991 election study so far provides the first point in the Finnish time-series.

The voters' map of the party positions along the left-right dimension is of course significant for understanding party competition and possibilities for party cooperation in politics. In this perspective, it appears that Denmark and Norway are on the road toward more instability, while Sweden and Iceland show signs of more tranquillity.

To say that the left-right dimension is dominant is not the same as saying it provides the whole picture. The tendency for new dimensions to be formed around environmental questions, moral issues, family politics, as well as refugee and immigrant questions has grown stronger. A more complete answer to the question of conflict or consensus in politics should therefore be explored in a multidimensional party space. The voters' map of the Nordic party space needs to be complemented by at least a green, a white-ethical and a xenophobic-cosmopolitan dimension.

---

NOTES

1. The three measures of the party space for all countries and all time-points are listed in Appendix A. The mean party distance (MPD) is calculated from all parties in the respective election year. This allows for testing the hypothesis that the addition of new parties means that the party system is perceived to be less polarized. The calculation of MPDs based upon the same number of parties at different time-points is listed in the footnotes.

2. The 1-9 and 1-10 scales have been transformed in the following way: First, the original values were subtracted by 1 which gave the scales a zero point. The 9-point scale now ranged from 0 to 8 and the 10-point scale from 0 to 9. Thereafter, all scale values were multiplied by a number which made the highest scale value equal to 10. The scale values on the 9-point scale were multiplied by 10/8s and the scale values on the 10point scale were multiplied by 10/9s.

3. The Swedish data also include information about where on the left-right scale the voters have placed each of the eight parties. For the voters' placement of their own party, the changes between 1979 and 1991 were in the same direction as for the selfplacements. However, taking into consideration all the voters' placement of the respective parties, there are two differences in comparison to the self-placements and placements of the own party. In the eyes of all voters, the Centre Party and the Conservative Party have shifted somewhat to the left between 1979 and 1991 - the Centre Party from 6.2 to 5.7 and the Conservative Party from 8.9 to 8.7.

4. The mean party distance, when basedon the distances between the five traditional parties - left, sd, c, lib and con - (MPDS), has decreased from 2.62 in 1979 and 2.63 in 1982 to 2.45 in 1991. If chr is included, the mean party distance (MPD6) has decreased from 2.29 in 1979 and 2.30 in 1982 to 2.15 in 1991.

---

4. The mean party distance, when basedon the distances between the five traditional parties - left, sd, c, lib and con - (MPDS), has decreased from 2.62 in 1979 and 2.63 in 1982 to 2.45 in 1991. If chr is included, the mean party distance (MPD6) has decreased from 2.29 in 1979 and 2.30 in 1982 to 2.15 in 1991.

5. The decreased polarization along the left-right dimension is even more pronounced if the three measures of the party space are instead based on where along the leftright scale the voters placed the different parties. The changes between 1979 and 1991 are as follows: WPD for self-placement -0.43, WPD for all the voters' placement of the parties —0.55, and WPD for the voters' placement of their own party —0.57. The corresponding changes of the RPD are -0.33, -0.80, and -0.33, and the corresponding changes of the MPD are -0.41, -0.92 and -0.91.

6. The Norwegian data also include information about where on the left-right scale the voters have placed the different parties. Taking into consideration all the voters' placements of the different parties, there are two differences in comparison with the self placements. In the election of 1981, the Liberals (5.0) did not stand to the left of the Social Democrats (3.9). The other difference is that all the voters perceive that the Progressive Party (8.8) in 1989 stood to the right of the Conservative Party (8.0). According to the voters' placements of the parties, it was only the Progressive Party which shifted toward the right between 1977 and 1989 (1.5 units). The remaining parties have moved toward the left: lib by 1.4 units, c by 1.0 units, chr by 0.8 units, sd by 0.6 units, and con and soc by 0.2 units. When it comes to the voters' placement of their own party, the changes between 1977 and 1989 follow in the same trend as for all the voters' placement of the respective parties.

7. The increased left-right polarity, with the exception of the rival party distance, becomes more obvious if the distances are calculated based on the voters' party placements. The changes are as follows: WPD for self placements +0.04, WPD for all voters' placements of the parties +0.84, and WPD for the voters' placement of their own party +0.52. The corresponding changes of the RPD are +1.07, +0.43, and +0.49, and the changes of the MPD are +0.97, +1.93 and +1.70.

8. The conclusion holds true even when consideration is given to the increased number of parties in Sweden. MPD for Norway in 1989 is 2.84 while MPDS and MPD6 for Sweden in 1991 are 2.45 and 2.15 respectively.

9. In 1990, the Socialist Left and the small Communist Party joined together for the election under the name of the Left Alliance. Therefore, the scale notations for 1990 refer to the Left Alliance and not the Socialist Left.

10. The absence of a liberal party in Iceland is a "modified truth", since the Citizens' Party is now called the Liberal Party (1.2 percent of the votes in the 1991 election). Note also that the Icelandic Progressive Party should not be confused with the corresponding party labels in Norway and Denmark. The Progressive Party was created back in 1916 and is Iceland's equivalent to the traditional farmer/centre party in the other Nordic countries.

11. The decrease in the MPD-value from 2.07 in 1987 to 1.79 in 1991 (-0.28 units) is based on seven parties in 1987 and on six parties in 1991, which reflects the picture in Fig. 4. If the measure is based on the same five parties both in 1987 and in 1991 (the four traditional parties and the Women's Alliance), the decrease is still -0.28 units - from 2.33 in 1987 to 2.05 in 1991. If the MPD-value is based on seven parties in 1987 and five parties in 1991 (i.e. only the parliamentary parties), the decrease would be only marginal - from 2.07 in 1987 to 2.05 in 1991 (-0.02 units). Lastly, if the MPDvalue is based on six parties in both 1987 and 1991 (i.e. without including the Union of Regional Equality in both 1987 and 1991) the decrease is -0.45 units, from 2.24 in 1987 to 1.79 in 1991.

12. The distance between the Danish Progressive Party and the smallest of the three neighbours immediately to the left on the scale, i.e. the Danish Agrarian Liberal Party, is significant at the 0.025 level. The distance between the Finnish Leftist Union and its nearest neighbour on the scale, i.e. the Danish Left Alliance, is, however, not significant at any reasonable level of significance. The distance between the Left Alliance and the third party from the left, i.e. the Danish Socialist People's Party, is significant at approximately the 0.075 level. A complete account of the number of observations, means and standard deviations for each party can be found in Appendix B.

ACKNOWLEDGEMENTS

The authors express their thanks to Sami Borg, Peter Esaiasson, Ingemar Glans, Donald Granberg, Olafur Hardarson, Sören Holmberg, Ola Listhaug, Susan Marton, the Danish Data Archive (DDA), the Norwegian Social Science Data Service (NSD), and the Swedish Social Science Service (SSD) for invaluable data and for valuable comments on an earlier draft of this article. We are also grateful for financial support from the Swedish Council for Research in the Humanities and Social Sciences.

REFERENCES

Aardal, B. & Valen, H. 1989. Questionnaires from the 1989 Norwegian Election Study. Oslo:
Institutt for samfunnsforskning.

Aardal, B. & Valen, H. 1993. Questionnaires from the 1993 Norwegian Election Study. Oslo:
Institutt for samfunnsforskning.

Bennulf, M. 1994. Miljöopinionen i Sverige. Lund: Dialogos.

Bennulf, M. & Holmberg, S. 1990. "The Green Breakthrough in Sweden", Scandinavian
Political Studies 13, 165-184.

Borre, O. 1983. "Ideologi og partivalg", in Borre, O. et al., Efter vælgerskredet. Analyser av
folketingsvalget 1979. Aarhus: Politica.

Converse, P. E. 1966. "The Problem of Party Distances in Models of Voting Change", in
Jennings, K. & Zeigler, L. H., eds. The Electoral Process. Englewood Cliffs: Prentice-Hall.

Converse, P. E. & Valen, H. 1971. "Dimensions of Cleavage and Perceived Party Distances
in Norwegian Voting", Scandinavian Political Studies 6, 107-152.

Gilljam, M. & Holmberg, S. 1990. Rött Blått Grönt. En bok om 1988 års riksdagsval.
Stockholm: Bonniers.

Gilljam, M. & Holmberg, S. 1993. Väljarna inför 90-talet. Stockholm: Norstedts.

Hardarson, O. & Kristinsson, G. 1987. "The Icelandic Parliamentary Election of 1987",
Electoral Studies 6. 3, 219-234.

Holmberg, S. 1981. Svenska väljare. Stockholm: Liber.

Listhaug, 0., Macdonald, S. E. & Rabinowitz, G. 1990. "A Comparative Spatial Analysis of
European Party Systems", Scandinavian Political Studies 13, 227-254.

Macdonald, S. E., Listhaug, O. & Rabinowitz, G. 1991. "Issues and Party Support in
Multiparty Systems", American Political Science Review 85, 1107-1131.

Nannestad, P. 1989. Reactive Voting in Danish General Elections 1971-1979. Aarhus: Aarhus
University Press.

Oscarsson, H. 1993. Politiska avstånd. Väljarnas uppfattningar av de politiska avstånden
mellan partierna i det svenska partisystemet. Göteborgs universitet: Statsvetenskapliga
institutionen.

Oscarsson, H. 1995. Partierna i rymden. Svenska väljares uppfattningar av partiavstånd och
konfliktstrukturer 1968-1994. Göteborgs universitet: Statsvetenskapliga institutionen.

Pesonen, P. 1973. "Dimensions of Political Cleavages in Multiparty Systems", European
Journal of Political Research 1, 109-132.

Pesonen, P., Sänkiaho, R. & Borg, S. 1993. Vaalikansan ännivalta. Poorvo: Werner Söderström
Osakeyhtiö.

Petersson, O. & Valen, H. (1979). "Political Cleavages in Sweden and Norway", in Scandinavian
Political Studies 2, 313-331.

Rabinowitz, G., Macdonald, S. E. & Listhaug, O. 1991. "New Players in an Old Game: Party
Strategy in Multiparty Systems", Comparative Political Studies 24, 147-185.

Rusk, J. G. & Borre, O. 1976. "The Changing Party Space in Danish Voter Perceptions,
1971-73", in Budge, 1., Crewe, I. & Farlie, D., eds., Party Identification and Beyond.
Representations of Voting and Party Competition. London: John Wiley.

Särlvik, B. 1968. "Partibyten som mått på politiska avstånd och dimensioner i partisystemet",
Sociologisk forskning 5, 35-80.

Särlvik, B. 1976. Mapping the Party Space: Distances, Evaluations and Ideological Perspectives.
An Analysis Framework Applied to Data on Electors and Elected in Sweden 1964-1973.
Paper presented at the Edinburgh IPSA Congress.

Tonsgaard, O. 1989. "Høyre og venstre i dansk politik", in Elklit, J. & Toonsgaard, 0., eds.,
To folketingsvalg. Aarhus: Politica.

Worre, T. 1976. "Partisystemets ideologiske dimensioner", in Borre, 0., Nielsen, H. J.,
Sauerberg, S. & Worre, T., Vælgere i 70'erne. Resultater fra interviewundersøgelser ved
folketingsvalgene i 1971, 1973 og 1975. Köpenhamn: Akademisk Forlag.