Resumé

More choice for parents and students is a major recent change in school policies in most Western countries. In Sweden, these changes coincided with the decentralization of power and responsibilities within primary and secondary education from central to local government. This gave municipalities an important role in the actual implementation of choice policies. The question addressed here is how and why local authorities vary with regard to their promotion of choice in schooling. A theory of local government school choice policies is developed, focusing on the social and political composition of the locality. Four factors are seen as likely to be particularly important: the strength of liberal conservatism, the size of the middle class, ethnic diversity and urban location. The empirical investigation is based on data from the 288 municipalities in Sweden. All four factors turn out to be important, but contrary to previous assumptions, choice policies are not primarily linked to the success of the Moderate (Conservative) Instead, the strength of the middle class is the major explanation. In Sweden, but also in other Western societies, the middle class has emerged as the dominant social stratum. Middle class parents, who themselves are well educated, take a keen interest in their children's education and are more likely to demand greater opportunities for choice. In the final analysis, this is linked to the changing position of the Social Democratic Party and its aim to broaden its electoral basis by appealing to middle class voters.

Introduction

The notion of school choice has entered the Swedish school policy agenda comparatively late. The comprehensive schools, uniformly modeled according national requirements, were streamlined in order to provide equal education for everyone, but also to facilitate changes of schools as a result of population mobility. However, other reasons for choosing a different school than the one nearby were not recognized. Pupils were expected to attend at the school closest to their homes, and alternative, private schools have always been rare. Hence, only in exceptional cases were changes of school a result of parental priorities, rather than family mobility. The opportunities to choose between programs within schools were greater, but in the main, these occurred at secondary level (Amman & Jönsson 1993).

---

* Anders Lidström, Department of Political Science. Umeå University, S-901 87 Umeå, Sweden. E-mail: anders.lidstrom(a)pol.umu.se

From the end of the 19705, the question of school choice has gradually emerged on the policy agenda. During the last 15 years, the Moderate (Conservative) Party has consistently argued for greater scope for parental choice of schools, in particular with regard to private or independent schools. The 1980s witnessed gradual changes in the regulation of school choice, aimed at opening up greater opportunities for choice for parents (Skolverket 1993b). These changes mainly occurred under bourgeois governments, some were initiated under the Social Democrats, however with less enthusiasm. Indeed, the Social Democratic position changed considerably the 1980s (Schiillerqvist 1995). The differences in views between the two major parliamentary blocs no longer concern whether to allow parents to choose which school their child should attend or if there should be independent schools, but rather how the independent schools should be funded. The bourgeois parties fa vor a centrally decided guarantee of public funding combined with a right for schools to charge fees. However, the Social Democratic government has implemented a system which leaves decisions about the amount of support to the local authorities. Nevertheless, is supposed to be set at a level which provides equal conditions for independent and municipal schools (Skolverket 1996). Although only 2.7 percent of students attended independent schools in 1997, the share has been increasing during the 1990s (Skolverket 1998). A special feature of Swedish school choice is parents' right to choose at what age their children start school. Normally, children begin school the year they are seven. However, since 1991, local authorities may provide piaces for six-year olds if that is the wish of their parents and resources are available. In 1997, parents obtained the right to demand such a place (Skolverket 1993a).

The more prominent position of school choice on the Swedish school policy agenda corresponds to an international movement. A number of other western countries have, from the mid-19705, introduced similar policies, giving parents and pupils greater opportunities for choice (Boyd & Kerchner 1988; Boyd 1992; Daun 1993; OECD 1994; Boyd 1996; Walford 1996; Cohn 1997). The timing cannot be a coincidence; choice policies are likely to be responses to more general changes taking place in the western world. These include a general economic restructuring, a better educated population, the spread of Thatcherite and Reaganite ideas and a declining confidence in the ability of the welfare state to provide quality services. Indeed, behind these policies, Brown detects more fundamental transformations the value orientation of education. An ideology of meritocracy has been replaced by an ideology of parentocracy: There is a 'move towards a system whereby the education a child receives must conform to the wealth and wishes of parents rather than the abilities and efforts of pupils' (Brown 1994,51).

However, the way these policies have taken shape in each country and each locality varies. In the Swedish case, there is a general national choice policy. For example, from 1989, the School Act gives parents the right to choose which school their child attends, but only if this does not cause unreasonable economic and organizational problems for the local authority. The school may be either public or independent. However, those living closest to a school are always given precedence if the number of available piaces is limited. Each local authority is free to develop these policies further through its own measures. Indeed, this may be one option open to the local political decision makers to fill the policy vacuum that has emerged in the wake of the decentralization process (Hudson & Lidstrom, forthcoming; 1991). Since 1975, powers and responsibilities in school matters have been gradually transferred from national government to the municipalities, giving local authorities greater opportunities to shape more distinctive local school systems (Lidstrom & Hudson 1995).

However, central government still retains important functions. It establishes national curriculum for all public and private education, it sets the legal framework and it provides considerable financial support. Thus, parliament and government specify the main structure and content of the education system, and they also set educational goals for students and schools. The Swedish National Agency for Education evaluates goalachievements supports local changes. There is a strong belief in equality and a common national standard in education. In terms of educational achievements, differences between schools are smaller in Sweden than in any other country (Svenska Kommunförbundet 1998).

Nevertheless, this central regulation should be understood as a frame within which local authorities and schools have considerable freedom to shape the organization and content of schooling. In recent years, local variation has increased. Local authorities now have greater freedom to decide what courses to offer and their content, and it is more common for schools to emphasize a particular profile in terms of subjects or pedagogy. In addition, a general trend is that local authorities increasingly adjust their courses to the needs of local economic interests (Lidström 1998).

The purpose of this article is to analyze how the decisions to introduce local school choice policies at local government level in Sweden can be understood. Which local authorities provide parents with greater opportunities choice and why? My argument is that the presence of local choice policies is linked to the particular social and political composition of the locality. Among the factors that may make decision makers more prone to develop such policies, the strength of the middle class is expected to be particularly important.

Local government school choice policies will refer to municipal attempts
to facilitate or stimulate parental choice in any of the following fields:

selecting an independent instead of a public school, changing from the closest to another public school, and deciding to let one's child start school the year it turns six, Le., one year earlier than normally. Local government provision of a more choice-oriented upper secondary school {kursutformad gymnasieskola) will also be regarded as a way of increasing choice opportunities.

As a next step, an empirically oriented theory of local government school choice policies is developed. This will be used as a tool in the subsequent analysis. Quantitative data on school policies at local government level and on a number of other variables will be used as a means of empirically testing the theoretical assumptions.

A Theory of Local Government School Choice Policies

How can local variation in school choice policies be explained? In order to provide a framework for the analysis, we will develop a number of hypotheses the basis of previous studies in the field. These hypotheses will then be linked into a coherent model. Thus, the theory is not generated according to a deductive format on the basis of overarching axiomatic assumptions. Rather, it has been developed more inductively, based on generalizations from empirical observations and research. An underlying belief is that the local social and political context is crucial for an understanding of the shaping of local choice policies. However, it must be emphasized that while the context provides incentives, the actual decisions are taken by actors. Indeed, this would correspond to the assumption expressed by the Swedish Parliament, that decentralizing powers and responsibilities local authorities will make them more responsive to local conditions.

Thus, the basic question may be formulated as what conditions are likely to be favorable for the development of local school choice policies? We assume that four conditions are particularly important and these are expressed four hypotheses. They concern liberal conservatism, the middle class, ethnic diversity and urbanism. To a considerable extent, these are also expected to reflect parents' demands for opportunities for choice.

The Liberal Conservatism Explanation

Extension of individual choice is a classical liberal idea. In their original
forms, conservatism and socialism emphasized a view in which the individualwas
larger collective entities, e.g., a nation, a class or a

common social cause. Liberalism, however, has from the very beginning stressed the role of the individual. John Stuart Mill emphasized the principle of individual liberty. Society should be constructed to provide the largest possible scope for individual initiatives. The role of the state should be limited to protecting citizens from each other.

The present versions of these ideologies have modified the original assumptions and also borrowed ideas from other thought systems. Not least, modern conservatism has been influenced by the liberal tradition. Indeed, a struggle between traditional conservatism and neo-liberalism has characterized conservative parties in the western world during the last few decades (Levitas 1986; Girvin 1988).

The influence of liberal ideas on conservative parties seems to have been particularly strong at the end of the 1970s and during the 1980s. This may be regarded as a reaction to the continuous growth of welfare states after the Second World War, to an increasing tax burden and to the perception that the freedom of individuals to make decisions about their own lives was narrowing. A large public sector was seen as hampering individual initiatives thereby reducing the efficiency and effectiveness of the economic system. This was a favorable ground for a re-emergence of the classical liberal ideas of pure market economy, reduced welfare state, and greater individual freedom, expressed by, for example, Hayek (1944) and Nozick (1974).

This tendency towards a liberalization of the conservative parties reached a peak with the heyday of Thatcherism and Reaganism. They represented more classical liberal ideas in the sense described above and managed first to get support from their parties and then to stay in power practically throughout the 1980s. Their policies set an example for conservative parties in other countries, and made a lasting imprint on the ideology of modern conservatism (Levitas 1986; Girvin 1988).

In line with these ideological changes, modern conservative parties have become keen advocates of policies that enhance individual choice. Apart from promoting a general reduction of the public welfare commitment, the parties have also favored the introduction of market-like conditions within the public sector and new means for individual choice. Voucher systems, privatization and outsourcing of public services are just a few examples of new policies.

In Sweden, the Moderate Party changed its ideological emphasis during the 19705. Freedom of choice and individual independence were new slogans in the 1976 election campaign (Ljunggren 1992). In 1978, liberalism was for the first time written into the party manifesto, when it was stated that '(M)oderate policies are anchored in conservative ideology and combinethis liberal ideas' (quoted in Ljunggren 1992, 286). Gradually, liberalism was established as an ideological source of even greater importancethan

tancethanconservatism (Hylen 1991). During the 1980s, more radical versions of liberalism influenced sections of the party, in particular the youth and student organizations. This coincided with an increase in electoralsupport the party. In the 19705, the party received 10-15 percent of the votes, whereas election results below 20 percent were unusual in the 1980s and early 19905. The political climate in Sweden in the late 1980s has been labeled hogervågen (a shift to the right), during which right-wing and neo-liberal ideas not only became more central in the public debate, but also influenced political decision makers from other parties (Boréus 1994).

Thus, these ideological shifts within the Moderate Party and its relative increase in strength during this period may be one explanation of the development of school policies that emphasize greater individual choice. The Moderate Party clearly altered its school policies over the last two decades. As late as in the mid-19605, promoting the establishment of independent was not a prominent feature of the Party's education policies (Lundahl 1989). Now it is the major proponent for support to individual and the introduction of voucher systems, permitting parents to choose between independent or public schools (Schüllerqvist 1995). Major changes were introduced by bourgeois governments, in particular by the four-party coalition between 1991 and 1994, when there were two Moderate Ministers of Education.

We expect these explanations with regard to the ideology of liberal conservatism be valid also at the local level. When local authorities are given the scope to decide whether and how to establish local choice policies, they are likely to follow the same party political pattern as at national level. Our first hypothesis is that local government school choice policies are more developed by local authorities with strong Moderate Party representation.

The Middle Class Explanation

One megatrend in the western world during the last three or four decades is the transformation of the workforce from manual to non-manual. This corresponds to the decline of manufacturing industry and agriculture as employers and the growth of the service sector, including the public welfare sector (OECD 1992). In essence, the development may be characterized as a growth of a new middle class, which, in particular, has taken place at the expense of the working class.

Class analysis is always controversial. There is no obvious way of distinguishingdifferent and stratification theory repeatedly addresses the question of what to base class identification on (Crompton 1993; Butler & Savage 1995). The problems are particularly difficult if the aim is to analyze how classes change over time. Nevertheless, if class is to be a fruitfulanalytical

fulanalyticalconcept, it is worth making the effort to try to clarify how it
can be defined. There are a number of alternatives.

While the classical Marxist definition was based on a dichotomy reflecting antagonism between labor and capital, modern theories have suggested a more diversified approach, attempting to take into account the more diffuse and fragmented class lines of today. These theories recognize that many individuals occupy middle positions where they may not own capital, but still exert decision making powers in the industrial process and over labor (cf. Wright 1985). However, these theories share with Marxism the belief that the relationship to the means of production is the major criterion for distinguishing classes.

In recent years, alternative theories have emerged which take into account other sources of power. One example, introduced in particular by Goldthorpe (1982), is the concept of the service class, which consists of white-collar workers with non-routine tasks, such as professionals, managers administrators. Professionals provide specialized knowledge and managers/administrators possess delegated authority on behalf of the employing class. The underlying formative concept is trust. The employing class - in public or private business - has invested trust in the service class in a way that makes it different from, for example, other white-collar employees the working class.

Another key notion in the service class concept is the emphasis on cultural capital or education. Other scholars have more explicitly based their class schemes on the possession of educational assets. Giddens (1981) identifies three classes in a capitalist society: one that owns capital, one with educational resources and one that provides labor. The class with educational is typically labeled the new middle class. This contrasts to the old middle class, which was much smaller and to a significant extent consisted of middle managers in manufactural production. The new middle class is not a coherent or stringent actor, and it is not organized in the same way as the working class. As Crompton observes, 'the term encompasses a wide variety of occupational groupings, distinguished only by the fact that they are not manual workers' (Crompton 1993, 175). However, a common denominator of the middle class is its educational resources. To a large extent, the middle class has reached its position because of its education.

The middle class has been growing continuously during the last few decades, and this increase is likely to continue in most western countries. Each year, the new generation that enters the work force is better educated than the generation that retires. This ongoing process is the result of postwar reforms, which aimed at giving more people access to higher education. A consequence is the gradual growth of the middle class, which is likely to continue well into the 21st century.

What makes these changes particularly interesting in this context is their value implications. A well-educated middle class, together with the expansion the service sector and the entry of women into the labor market, are regarded as major features of the post-industrial society (Bell 1973). In repeated studies in Europe and North America, Inglehart (1977; 1990) has detected a shift in the value pattern. New, post-material values tend to emphasize individualism, freedom and quality of life, participation and influence, sustainability and the preservation of nature. The younger generations, also the middle class, in terms of the well-educated and the professionals, seem to carry the new values (cf. also Scarborough 1995).

Corresponding changes have occurred in Sweden (Lidström & Hudson 1995). According to data from Swedish election studies, the manual workers' share of the adult population decreased from 51 percent to 38 percent between 1960 and 1991, whereas the non-manual workers' {tjänstemäri) share during the same period increased from 27 percent to 51 percent (Oskarson 1994). Changes in the same direction, although less dramatic, are reported by Ahrne et al (1995). However, compared with other Western nations, the Swedish middle class consists of a larger proportion of economically women and of public sector employees (Ahrne et al 1995). Also in Sweden, the middle class is likely to grow further along with a better educated population. In 1970, eight percent of the adult population had a post-secondary degree. In 2015, this number is expected to be 29 percent (Statistiska Centralbyrån 1994). It should be emphasized that a society dominated by the middle class is not necessarily more equal than the traditional industrial society. On the contrary, during the recent growth of the middle class, there are signs of increasing differences in income and wealth in the Swedish population.

The observation about middle class individualism mentioned earlier has also been detected in Sweden (Pettersson & Geyer 1992). In addition, this class exhibits a different form of democratic behavior than the working class. Instead of utilizing traditional ways of influence through political parties and collective organizations, the middle class favors more individually orientated means, for example individual contacts with decision and more ad hoc organizing (Petersson et al. 1989).

A point already emphasized is the specific relationship between the middle class and education. Not only have large segments achieved their positions because of their education, they also tend to take a keen interest in the education of their children (Beare 1993; cf. also Gewirtz et al. 1995). Therefore, we will expect the more individualistic values of the middle class to make an imprint on the shaping of the local school systems. This adjustment middle class preferences may have the effect of reducing the influence of other social strata on education.

Several investigations have shown that well-educated parents are more

in favor of choice policies in schools. They know more about choice
opportunities, and they are more likely to prefer independent schools for
their children (Skolverket 1993b).

Thus, it seems likely that the existence of a large middle class in a local community will be favorable for the emergence of local choice policies. Therefore, our second hypothesis is that local government school choice policies are more developed in local authorities with a larger middle class.

The Ethnic Diversity Explanation

Not long ago, a common feature of the Swedish society was its ethnic homogeneity. Apart from the Sami minority in Lapland, the Swedes largely consisted of one major ethnic group. As late as the early 19705, Sweden scored very low on measures of ethnic-linguistic and religious fragmentation & Ersson 1987). However, since the 19305, immigration to Sweden has been greater than emigration. In particular during the 19605, recruiting labor from other countries was a means of solving domestic problems with workforce shortage. In addition, there has been scope for refugees from war-ridden countries to seek a haven in Sweden.

In a few decades, this has transformed Swedish society (Ålund & Schierup 1991). Today, one eighth of the population in Sweden was born abroad or have at least one parent who was. Thus, the Swedish population has become more heterogeneous. It is more common that languages other than Swedish are spoken or that religions other than Protestantism are practised.

Several more heterogeneous Western countries have school systems that provide considerable scope for ethnic and religious minorities to run their own schools more independently. Schools are a major means in the attempt to preserve sub-cultures (Daun 1993). In particular, Catholic minorities have been eager to provide their own schools in countries such as the USA and Australia.

There are no data available about how different religious minorities choose schools in Sweden. However, the ethnic factor is obviously important. survey by the National Agency for Education (Skolverket) in 1994/95 shows that parents who are born outside Sweden are more likely to choose a different public or independent school than the one closest to their home. In particular if the parents come from another Nordic country, but also those who are born outside the Nordic countries exert their right to choose more than native Swedes (Skolverket 1996).

Accordingly, we expect that local authorities with large ethnic or
religious minorities will be more in favor of choice policies. This is a way of
adjusting the school systems to the specific needs of the local population.

Therefore, our third hypothesis is that local government school choice
policies are more developed in local authorities with large ethnic or religious
minorities.

The Urbanism Explanation

Even if there is strong demand for different choice alternatives, an important is still a certain proximity to alternative schools. Generally, it should be expected that parental choice is exerted to a larger extent if there are reasonable alternatives close by or if there is a transport system which facilitates daily traveling between school and home. Thus, urban settings are likely to promote choice whereas this is probably rare in the countryside. Perhaps not surprisingly, people in the larger Swedish cities are more in favor of choosing a different public or independent school than the one closest to their home (Skolverket 1993b).

In terms of local government policies, it seems more reasonable for municipalities with large urban populations to develop policies promoting choice. Our fourth hypothesis is that local government school choice policies are more developed by local authorities in urban settings.

A Combined Model

Each of the four hypotheses that have been identified are expected to contribute explaining why some local authorities develop choice policies and others do not. However, in practice, these likely explanations interact. One possibility is that they overlap and partly cover the same phenomenon. Another is that one variable may exert its influence through another and may, inadvertently, be counted twice. The task in this section is to try to disentangle the specific role of each variable by constructing a combined model which specifies how they are linked theoretically to each other.

We expect urbanism to play a role as background variable in relation to each of the other likely explanations. The middle class, the strength of liberal conservatism and ethnic diversity are likely to be larger in larger cities and their suburbs, but we also expect urbanism to be important by itself without mediation through other variables.

The middle class explanation is separate from, but nevertheless closely linked to, the size of the Moderate Party. This party has its stronghold among middle-class voters and leaders of small businesses (Gilljam & Holmberg 1995). Hence, we expect the size of the middle class to influence the local strength of the Moderate Party. Finally, ethnic diversity is not likely to be directly linked to either middle class or Moderate Party

strength. However, an indirect link is likely to exist, since all these variables
are expected to be features of urban areas. Figure 1 summarizes the model.

An alternative theory has been presented by The Swedish Agency for Education in a study of choice policies during 1992/93 (Skolverket 1993b). After an investigation of school choices in nine local authorities, more general conclusions were drawn about the relationship between structural factors and the propensity for local authorities to promote parental choice. In essence, three structural factors, which are positively related to choice policies, were identified: The number of inhabitants per km², geographical proximity to schools and the presence of independent schools in the local area.

For several reasons, this Skolverket model seems less appropriate than the one suggested in Figure 1. First, the relationship between the different explanatory variables is not discussed. In particular, the first two variables are likely to be highly overlapping. The greater the population density, the closer the schools are. In terms of our model, these two variables correspond the urbanism explanation. Second, the Skolverket model is clearly more limited, as both the role of the middle class and the political context are omitted, and the importance of ethnic diversity is not taken into account. To some extent, but probably not fully, these factors may coincide with the third variable in the Skolverket model, i.e., presence of independent schools. However, in the subsequent empirical analysis, we will also test the explanatory power of this model.

Measuring School Choice Policies

The empirical analysis of the theory outlined above is undertaken with the help of quantitative data. A quantitative approach has been chosen, as it permits generalizations from the conditions studied for all Swedish local authorities. The data set consists of information about all 288 Swedish local authorities, although there are some omissions with regard to a few variables. The data base not only contains information about local school policies and practices, it also permits these to be analyzed in relation to different social, economic and political characteristics of the local area. The data has been collected from different sources. Two questionnaire surveys of local government education officers were undertaken.¹ In addition, the National Agency for Education generously provided us with data collected by its regional officers in October 1995. Finally, publicized statistics from the Swedish Statistical Central Bureau and the National Agency for Education about conditions in local authorities have been included.

Local government school choice policies are the dependent variable in the analysis. As previously mentioned, these refer to attempts to facilitate or stimulate parental choice in the following fields: selecting an independent instead of a public school; changing from the nearest to another public school; and deciding to let one's child start school a year earlier. Local government provision of a more choice-oriented upper secondary school is also included. Introduction of voucher systems and stimulating schools to develop distinct profiles are regarded as additional and supporting ways of promoting choice.

The dependent variable is measured in four different ways. By focusing on assessments of policy as well as what local authorities actually do, the stability of the model can be controlled. Each measure is represented by an index based on a set of variables. These variables are summarized in Table 1.

All indexes represent different ways of operationalizing local government school choice policies and were constructed by adding the values of the relevant variables. They are all ordinal scales where zero indicates no expression school choice policy whatsoever, and the maximum value represents the presence of all types of choice policies. Indexes 1 and 2 are based on our own questionnaire surveys. The chief education officers were asked to assess how important different school choice policies were for the political majority of the council during the two election periods 1991-1994 and 1994-1998.² This provides an opportunity to analyze changes over time between two very different periods. The 1991 local and national elections resulted in a bourgeois landslide at the same time as the Social Democratic party had its worst election result since 1928. In 1994, however, the pendulum back. The bourgeois parties suffered considerable losses, and the Social Democrats regained their majority in many councils.

The figures in the table indicate that the importance of choice policies has decreased as the bourgeois parties have lost ground. The only exception concerns the policy to stimulate schools to develop their own profiles, which seems to be regarded as important, regardless of political majority. However, analyses are required for a more thorough understanding of the relationships.

Indexes 3 and 4 focus on what local authorities actually do in order to promote parental and student choice. They are both based on data collected by the Swedish Agency for Education. Index 3 concerns a number of possible mainly in relation to upper secondary education. Index 4 is about whether local authorities provide information to parents about choice opportunities at primary and lower secondary levels. As the data about the variables included in index 4 illustrate, local authorities are clearly less willing to inform about the right to choose independent schools than about choice opportunities within the public system.

Testing the Model

The empirical test of the model has been conducted in several steps. Initially,
the bivariate relationships between different indicators of the hypothesized

explanations and the indexes were explored. Positive, significant and often strong relationships emerged. However, the final test required a multivariate approach, since the combined effect of the possible explanations and the relative weight between them are major concerns in the study.

The process of selecting the indicators of the independent variables required consideration. Choosing the size of the Moderate Party in local councils as a measure of Liberal Conservatism was fairly obvious. The middle class, on the other hand, is not easily operationalized because of the theoretically unclear status of the concept. Different alternatives were investigated, including level of education of the local population and employment in the service sector, but in the end the share of the population representing intermediate and higher non-manual employment³ was selected. measure is more closely linked to Goldthorpés concept of the service class. However, all the examined alternatives are highly intercorrelated, coefficients (rxy) varying between .90 and .98. Despite the theoretical vagueness of the concept of the middle class, its empirical representation is distinct.

Choosing a measure of ethnic diversity was problematic. There are no data available about the representation of different religious denominations at local level in Sweden. However, different measures of the number of non-Swedish residents were tested in the bivariate analyses. The strongest relationships emerged when the share of non-Nordic residents was correlated measures of the dependent variable. To some extent, this may also indicate the presence of religious denominations other than Protestantism. final variable, urban location, is represented by a dummy variable.

Table 2 summarizes the multivariate analyses. The relative impacts of
the four independent variables have been tested in relation to all four
measures of the dependent variable.4

A number of observations can be made on the basis of the regression analyses. The variation in the first index, representing perceived local government choice policy 1991-94, is to a considerable extent explained by the four independent variables (R² = .42). Also, all four seem to contribute the explanation, even if middle class appears as the most important of the factors.

This pattern changed during the next election period, represented by index 2. First, the total explained variance decreases radically, suggesting that variables other than the four hypothesized factors have become more relevant, or that school choice policies no longer follow a clear and identifiablepattern. particular, ethnic diversity and urban location no longer have the same importance. It should be kept in mind that local government school choice policies have actually decreased in importance after the 1994 elections (cp. Table 1). Therefore, we may conclude that the remaining

policies are more spread out in different kinds of local authorities in the
1994-98 period than in the beginning of the decade.

However, the second, and our major, conclusion concerns the explanatory of the middle class which emerges as the major factor behind local government school choice policies. Indeed, its position as an explanatory factor has become stronger than during the previous period. Even if local government school choice policies are less common, those remaining are more closely connected with a strong middle class. The party political variable has also become more prominent, but not to the same extent as the middle class.

The middle class is again the major explanation with regard to the third index, which concerns local government decisions resulting in greater choice opportunities, particularly with regard to upper secondary schools. Basically, the pattern emerging resembles the one behind index 1, even if the coefficients are lower, with the exception of the middle class. This may be regarded as support for the overall model, which is obviously relevant for what local authorities actually do, as well as for perceptions of policies.

The fourth index, which represents information measures undertaken by the local authority about choices available with regard to primary and lower secondary education, exhibits a slightly different pattern. The model seems less appropriate, with an explained variance of .21. It emphasizes

urban location and non-Nordic citizenship, whereas the other two hypothesizedexplanations
not significant.

We can now reconnect with the model in Figure 1, which also outlined possible relationships between the independent variables in the analysis. Urban location was assumed to have a background function in relation to the other three independent variables, and Moderate Party strength was expected to be associated with the size of the middle class. The path analysis in Figure 2 summarizes the disentangling of these relationships with regard to the first index.

As expected in the combined model, urban location is closely associated with the other three independent variables. A strong relationship also exists between the size of the middle class and the strength of the Moderate Party. Urban location appears as more important when its indirect effects through the other variables are taken into account. The path analyses for the other indexes provide fairly similar results, with the exception that the associations the independent variables and the dependent variable are different, which has already been shown in Table 2.

Generally, our theory seems to be able to identify fairly accurately the pattern behind why local government school choice policies emerge in some settings, but not in others. However, there is an alternative. Previously in the article, an explanation, suggested by The Swedish Agency for Education,was (the Skolverket model). This emphasized three factors, i.e., number of inhabitants per km², geographical proximity to schools and

presence of independent schools nearby. Data is not available for the second factor, but proximity is likely to overlap with population density. The remaining two variables are tested against our measures of school choice policies. As Table 3 indicates, the Skolverket model is less successful in explaining the variance than our model.⁵

Conclusions

On the whole, our theory has been fruitful in explaining why some local authorities develop school choice policies and others do not. Clearly, the size of the middle class is a previously underestimated factor of considerable importance. In the Swedish context, choice policies have mainly been regarded a result of the 'turn to the right' and the new conservative ambitions to provide greater choice opportunities. However, in the light of this study, this political factor is less important than the extent of middle class domination in the local community. Further, this tends to occur in urban rather than rural locations.

The growth of the middle class is a major contemporary megatrend. The middle class has specific preferences, not least with regard to education policy. Individualism and greater choice are important components in its value orientation. With time and as more of those entering the work force have a higher education, the middle class will continue to grow. The middle class will come to dominate in more communities and, therefore, it is not unreasonable to expect that this process of adjusting school policies to the needs and preferences of the middle class will continue. Perhaps we arc only witnessing the beginning of middle class influence on local school systems. Such a reinforcement of the position of the middle class may eventually exclude the preferences of other social strata, in particular the working class and the disadvantaged.

The Social Democratic position on these issues has changed in recent decades. Once hostile towards choice policies, the party now accepts and even promotes them. Admittedly, the Social Democratic Party does not exhibit the same enthusiasm as the Moderates. Nevertheless, parents' and students' rights to choose a school and an acceptance of independent schools are now a part of Social Democratic policies. This may be regarded as a move to the right, but a more reasonable interpretation is to see it as an adjustment to the preferences of the growing middle class. Since the traditional working class is slowly but steadily decreasing in number, the electoral basis of the Social Democratic Party is under threat. To compensate this, the Party has attempted to increase its support in the middle class (Svensson 1994; Esping-Andersen 1994). Hence, the adoption of middle class preferences can be seen as an effort to attrået these voters, nationally and locally. School choice policies are parts of a wider strategy by the Social Democratic Party to maintain its position in Swedish politics in a time of major social transformations. These attempts to capture the voters in the middle of the left-right scale may also be seen as an adjustment predicted by the median voter theory (Downs 1957).

---

NOTES

1. The first, distributed in the spring of 1995, received replies from 75.0 percent of the local authorities, and the second, undertaken in the beginning of 1996, reached a response rate of 89.2 percent.

2. Please note that the election period in Swedish local and national elections was extended from three to four years in 1994. Local and national elections are held on the same day.

3. In the official Swedish census statistics, the working population is divided into three major groups! manual workers, non-manual workers and employers. However, in particular first two have been further divided into sub-categories, mainly on the basis of required education level. Non-manual workers, for example, are divided into four categories: two lower, intermediate and higher (Statistiska Centralbyrån 1992).

4. In addition to the estimations presented in the table, possible interaction effects have been examined. In particular, it could have been suspected that the combined presence of a strong Moderate Party and a large middle class would enhance the emergence of local choice policies even further. However, only very marginal effects of this kind were detected. The statistical significance of the coefficients in this article is indicated in the following way: Significant at the .001-level (***), .01-level (**) and .05-level (*). Finally, it should be observed that index 3, representing decisions, is only calculated for the 212 local authorities with upper secondary schools.

---

5. The presence of independent schools was not included among the hypothesized explanations of local government school choice policies. However, adding this to the test of the four variables in Table 6 makes little difference. On the whole, its impact is absorbed by other variables. However, it appears important in relation to index 3, which represents decisions by local government in relation to upper secondary education contributes to an increase in the explained variance from .35 to .39. The presence of independent schools receives a Beta weight of .19**. The Beta weights of the four hypothesized variables decreases slightly.

ACKNOWLEDGMENT

This research has been funded through a research grant from the Swedish National Agency of Education (project number 15001). I am indebted to the members of the International Working Group on Policy Implementation and Management (Fredrikshavn 23-25 August, 1997) and the participants at the working group on Swedish politics at the meeting of the Swedish Political Science Association (Uppsala 5-7 October, 1997) for valuable comments on previous versions of this article.

REFERENCES

Ahrne, G., Ekerwald, H. & Leiulfsrud, H. 1995. Klassamhållets forandring. Lund: Arkiv
forlag, 4:e upplagan.

Arnman, G. & Jonsson, I. 1993. Konkurrens for stimulans. Stockholm: Skolverket.

Beare, H. 1993. 'Different Ways of Viewing School-site Councils: Whose Paradigm Is in Use Here?,' in Beare, H. & Boyd, W. L., eds., Restructuring Schools: An International Perspective the Movements to Transform the Control and Performance of Schools. London: The Falmer Press.

Bell, D. 1973. The Corning of the Post-Industrial Society. A Venture in Social Forecasting.
New York: Basic Books.

Boréus, K. 1994. Hogervåg. Nyliberalism och kampen om språket i svensk offentlig debatt
1969-1989. Stockholm: Tidens forlag.

Boyd, W. L. 1992. 'Parental Choice of Schools: An International Movement,' in Miron, G.,
ed., Towards Freee Choice and Market-Oriented Schools. Problems and Promises. Stockholm:
of International Education/Skolverket.

Boyd, W. L. 1996. 'The Politics of Choice and Market-oriented School Reform in Britain
and the United States: Explaining the Differences,' in Chapman. J. D. et al., eds., The
Reconstruction of Education. Quality, Equality and Control. London/New York: Cassell.

Boyd, W. L. & Kerchner, C. T., eds. 1988. The Politics of Excellence and Choice in Education.
New York: The Falmer Press.

Brown, P. 1994. 'Education and the Ideology of Parentocracy,' in Halstead, J. M., ed..
Parental Choice and Education. Principles, Policy and Practice. London: Cogan Page.

Butler, T. & Savage, M., eds. 1995. Social Changeand the Middle Classes. London: UCL Press.

Crompton, R. 1993. Class and Slratification. An Introduction to Current Debates. Cambridge:
Policy Press.

Daun, H. 1993. Omstrukturering av skolsystemen. Decentralisering, valfrihet och privatisering.
En internationell oversikt. Stockholm: Skolverket.

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

Gewirtz, S., Ball, S. J. & Bowe, R. 1995. Markels, Choice and Equity in Education.
Buckingham & Philadelphia: Open University Press.

Giddens, A. 1981. The Class Slructure of the Advanced Societies. London: Hutchinson. 2nded.

Girvin, 8., ed., 1988. The Transformation of Contemporary Conservatism. London: Sage
Publications.

Goldthorpe, J. 1982. 'On the Service Class, its Formation and Future,' in Giddens, A. &
MacKxnzie, G., eds., Classes and the Division of Labor: Essays in Honour of Ilya Neustadt.
Cambridge: Cambridge University Press.

Hayek, F. A. 1944. The Roadto Serfdom. London: Routledge & Kegan Paul.

Hylen, J. 1991. rosterlcindel framst' K.onservul ism och libcriiliMti mc'?* '-< i.'i rptirtiet 1904—
1985. Stockholm: Norstedts Juridikforlag.

Inglehart. R. 1977. The Silent Revolution Princeton: Princeton University Press.

Lane, J. E. & Ersson, S. O. 1987. Politics and Society in Western Europe. London: Sage
Publications.

Levitas, R., ed., 1986. The Ideology ofthe New Right. Cambridge: Polity Press.

Lidstrom, A. 1991. Discretion. An Art ofthe Possible. Department of Political Science, Umeå
University.

Lidstrom, A. 1998. 'Utbildning och kommunal utvecklingspolitik.' Kommunal ekonomi och
politik 2, 21-42.

Lidstrom, A. & Hudson, C. 1995. Skola i forandring. Decentralisering och lokal variation.
Stockholm. Nerenius & Santérus forlag.

Ljunggren, S. B. 1992. Folkhemskapitalismen. Hogerns programutveckling under efterkrigstiden.
Stockholm: Tidens forlag.

Lundahl, L. 1989. / moralens, produktionens och det sunda fornuftets namn. Det svenska
hogerpartiets skolpolitik 1904-1962. Lunds Universitet: Pedagogiska institutionen.

Nozick, R. 1974. Anarchy, State and Utopia. Oxford: Basil Blackwell.

Oskarson, M. 1994. Klassrostning i Sverige. Rationalitet, Lojalitet eller bara slentrian.
Stockholm: Nerenius & Santérus forlag.

Peterson, 0., Westholm, A. & Blomberg, G. 1989. Medhorgarnasmakt. Stockholm: Carlssons.

Scarborough, E. 1995. 'Materialist-Postmaterialist Value Orientations,' in Deth, J. W. van
& Scarborough, E., eds., The Impact of Values. Beliefs in Government volume four. Oxford:
Oxford University Press.

Schiillerqvist, U. 1995. 'Forskjutningen av svensk skolpolitisk debatt under det senaste
decenniet,' in Englund, T., ed., Utbildningspolitiskt systemskifte. Stockholm: HLS forlag.

Skolverket. 1993a. Tidigare skolstart. Om kommuners mote med en reform. Skolverkets
rapport nr 39.

Skolverket. 1993b. Val av skola. Rapport om valfrihet inom skolpliktens ram låsåret 1992/93.
Skolverkets rapport nr 40.

Skolverket. 1996. Att vdlja skola - effekter av valmojligheter i grundskolan. Skolverkets rapport
nr 109.

Skolverket. 1998. Skolan. Jdmforelsetalfor skolhuvudmån. Organisation - Resurser - Resultat.
Delrapport, mårs 1998. Skolverket rapport nr 146.

Statistiska Centralbyrån. 1992. Folk- och bostadsrdkningen 1990. Del 5: Forvårvsarbetande
och yrke.

Statistiska Centralbyrån. 1994. Trender och prognoser '94. Befolkningen, uthildningen, arbetsmarknaden.

Svenska Kommunforbundet. 1998. Aktuelt om skolan. Augusti 1998.

Svensson, T. 1994. Socialdemokratins dominans. En studie av den svenska socialdemokratins
partistrategi. Stockholm: Almqvist & Wiksell forlag.

Walford, G. 1996. School Choice and the Quasi-market. Oxford Studies in Comparative
Education 6. Wallingford: Triangle Books.

Wright, E. O. 1985. Classes. London: New Left Books.

Ålund, A. & Schierup, C. U. 1991. Paradoxes of Multiculturalism. Essays on Swedish Society.
Aldershot: Avebury.