More Ambitious Educational Choices in Urban Areas

Abstract

Urban–rural disparities in educational outcomes have so far primarily received attention in U.S.-based research. These studies show that pupils in rural areas are at a disadvantage compared with pupils in (sub)urban areas. This article aims to examine urban–nonurban differences in educational choice in a European context, namely Flanders (the northern part of Belgium). To do so, we make use of data gathered from 1,339 parents of pupils in a sample of 53 primary schools (24 urban, 29 nonurban). We find that pupils in urban areas make more ambitious choices and that this is partly explained by local labor market conditions.

Keywords

high school programs urban social school choice urban education

Introduction

Research into the existence of rural–(sub)urban disparities in educational outcomes has so far almost entirely been restricted to U.S.-based studies (e.g., Reeves, 2012; Reeves & Bylund, 2005; Roscigno & Crowley, 2001; Roscigno, Tomaskovic-Devey, & Crowley, 2006; Strayhorn, 2009). These studies show that students who attend schools in rural areas perform worse than students at (sub)urban schools. By contrast, in Europe research has tended to focus attention mainly on the potential impact of more immediate locales, such as neighborhoods, on the educational outcomes of its inhabitants (e.g., Andersson & Subramanian, 2006; Kauppinen, 2008; Sykes & Kuyper, 2009; Sykes & Musterd, 2011). While some of these studies demonstrate the existence of neighborhood effects on educational achievement (Andersson & Subramanian, 2006; Sykes & Kuyper, 2009), others show that these are mediated through school-level effects (Kauppinen, 2008; Sykes & Musterd, 2011). Nevertheless, some European scholars have occasionally brought the question of regional differences in educational outcomes to the fore (de Boer, van der Werf, Bosker, & Jansen, 2006; Dronkers, van Erp, Robijns, & Roeleveld, 1998; Gambetta, 1996). In his study on higher education participation in Italy, Gambetta (1996) showed that participation rates differ between North, Central, and South Italy. He explained these differentials by referring to differences in labor market opportunities between these broad geographical regions. Dronkers et al. (1998) and de Boer et al. (2006) focused on regional differences in the transition between primary and secondary education in the Netherlands. Dronkers et al. (1998) found that pupils in highly urbanized regions are more likely to get the advice to enroll in academic tracks in secondary education than pupils in less urbanized regions. de Boer et al. (2006) were especially interested in the situation in one specific rural Dutch province, namely Friesland (in the North West of the Netherlands), as opposed to the rest of the Netherlands with regard to educational recommendations. They found that pupils in Friesland do less often receive the advice to start secondary education in the academically oriented electives than pupils in the rest of the Netherlands. An explanation in terms of labor market opportunities similar to that of Gambetta (1996) did not prove fruitful in this case. However, the authors stressed the fact that provinces might be too crude units of analysis to study the impact of regional characteristics.

Most of the studies examining regional differences in educational outcomes invoke differences in regional opportunity structures to explain these differentials. Alongside these occasional endeavors, there is also a research tradition directly focusing on how local labor markets influence educational decision making (e.g., Raffe & Willms, 1989). Raffe and Willms (1989) found that unemployment rates in regions in Scotland were positively related to the decision to stay on at school after compulsory education. They also found that local employment structures with a high share of jobs in service industries—requiring a more highly qualified workforce—were significantly related with higher staying-on rates, among otherwise comparable pupils. The reasoning is that local opportunity structures influence pupils’ subjective evaluations of the benefits of staying on at school (see Rosenbaum, 2001).

However, unlike in the United States, there seems to be no European research tradition that systematically inquires into rural–urban disparities in educational outcomes, despite research that demonstrates the impact of local opportunity structures on, for example, young people’s migration propensity (Thissen, Droogleever Fortuijn, Strijker, & Haartsen, 2010). Taking into consideration the wider context in which education takes place proffers a more encompassing view than the overly individualistic models which are prevalent in research in sociology of education (Roscigno & Crowley, 2001). As Roscigno et al. (2006) rightly state, “Such views are limited and should be supplemented with an understanding that the dynamics of stratification are ultimately manifested at a more local level . . .” (p. 2139). So, in essence, taking into account potential urban–nonurban differences in educational outcomes means providing a more complete image of educational processes. Moreover, except the fact that a focus on urban–nonurban disparities might contribute to the scholarly understanding of educational processes, an interest in potential disparities between urban and nonurban areas in Europe is warranted in view of the challenges such areas are currently confronted with. While urban areas in Europe face the challenge of new immigration from Eastern Europe and rapid population growth, rural areas face dejuvenation as a result of brain drain caused by a decline in employment opportunities (Thissen et al., 2010). From a policy perspective, it is therefore interesting to see how education in urban and nonurban areas fares.

A crucial educational outcome in most continental European countries is the educational choice made by pupils at the transition from primary to secondary education, for this choice is very consequential for a pupil’s future educational and occupational prospects (Vanderheyden & Van Trier, 2008). While scholars throughout Europe have displayed a particular interest in social inequalities in educational choice at the transition from primary to secondary education (Boone & Van Houtte, 2013, for Flanders—Belgium; Ditton & Krüsken, 2006, for Germany; Jaeger, 2009, for Denmark; Kloosterman, Ruiter, de Graaf, & Kraaykamp, 2009, for the Netherlands), the potential impact of contextual, school-level variables has received much less attention (for notable exceptions, see Jonsson & Mood, 2008; Kauppinen, 2008; Schulze, Wolter, & Unger, 2009). Possible spatial disparities in educational choice seem to be lacking altogether, except for one study in Turkey by Aypay (2003) who found that students in rural areas prefer academic education over vocational education. However, the question is whether patterns found in Turkey can be indicative of potential spatial disparities in northern Western Europe. In this study, we aim to explore the impact of attending a primary school in an urban versus a nonurban environment on choice of type of secondary education in Flanders (the northern, Dutch-speaking part of Belgium). If differences between urban and nonurban areas can be established, a second objective is then to explain these differentials in terms of the labor market opportunities characteristic of these areas.

Inequalities of Place

Inequalities of place with regard to educational outcomes have been a recurring topic of interest in American sociology at least since Coleman et al.’s (1966) Equality of Educational Opportunity. Differences in the resources available to schools in different geographical areas and their consequences for educational achievement have been at the core of this concern. Studies have repeatedly shown that students in rural areas perform worse than students in (sub)urban areas (Reeves, 2012; Reeves & Bylund, 2005; Roscigno & Crowley, 2001; Roscigno et al., 2006; Strayhorn, 2009). However, few attempts have been made at providing a systematic explanation for these geographical disparities (Roscigno & Crowley, 2001). Roscigno and Crowley (2001) have proposed an argument, which meant both a theoretical and a methodological advancement in this particular field of study. By simultaneously considering differences in family- and school-level processes between rural and nonrural areas, their argument offers a more encompassing view than prior studies that had taken only one of these institutional spheres into account. Moreover, previous research had probably overestimated the direct effects of processes in families by not considering school-level processes (Roscigno, 1998).

According to Roscigno and Crowley (2001), local labor market opportunities lay at the basis of urban–rural differences in educational outcomes through their influence on family- and school-level processes. In comparison with urban labor markets, rural labor markets would more often be characterized by low-wage, labor-intensive jobs (e.g., in agriculture and industry) and low-wage service sector jobs. As a consequence, there should be a higher concentration of disadvantaged families in rural areas, who have limited resources at their disposal for their children’s education. The prevalence of low pay jobs should also have negative repercussions for the tax revenues in rural areas, which in turn depresses the amount of resources available in rural areas to spend on education. Furthermore, the authors suppose that there might be an unwillingness to allocate resources to education for fear that it might in the end benefit other, nonrural locales through processes of brain drain.

Besides disparities in the resources available to families and schools in urban versus rural areas, the authors also stress differences in investments made at family and school level between urban and rural areas (Roscigno & Crowley, 2001). As a result of the lack of resources families in rural areas possess, they are supposed to invest less cultural capital in their children and invest less in household educational items (e.g., books, newspapers), both of which have been shown to have a positive impact on children’s scholastic achievement. In addition, the authors suspect that there could be a direct effect of local labor market opportunity on parents’ expectations with regard to their children’s educational career. Similarly, they expect teachers in rural schools to have lower expectations of their pupils (Roscigno & Crowley, 2001). The theoretical argument proposed by Roscigno and Crowley (2001) could indeed be confirmed as they found that rural versus nonrural differences in achievement and dropout could, to a large extent, be explained by differentials in resources at family and school level and concomitant differentials in investments in these two institutional spheres (Roscigno & Crowley, 2001). More recently, Roscigno et al. (2006) have shown that this reasoning could also explain achievement and dropout differentials between, on the one hand, students who attend schools in suburban areas and, on the other, students who attend schools in rural or inner city areas. However, while differences in resources and investments could explain all of the inner city versus suburban gap in achievement and dropout status, part of the rural versus suburban gap remained unexplained, indicating that other processes should be at work as well.

In Europe, research has tended to direct its attention to the potential impact of more immediate locales such as neighborhoods on young people’s educational outcomes (Andersson & Subramanian, 2006; Brannstrom, 2008; Garner & Raudenbush, 1991; Kauppinen, 2008; Sykes & Kuyper, 2009; Sykes & Musterd, 2011). While some of these studies include controls for urbanization, the primary focus clearly is on how neighborhood (dis)advantage affects pupils’ scholastic achievement, thereby displaying a similar interest in local opportunity structures, albeit at a more local geographical level. However, rather than exerting an influence sui generis on educational achievements, neighborhood (dis)advantage seems to operate mainly through processes at the school level (Kauppinen, 2008; Sykes & Musterd, 2011).

Some European scholars have occasionally focused their attention on more global geographical units, such as regions (Gambetta, 1996), conurbations (Dronkers et al., 1998), and provinces (de Boer et al., 2006), and their relation to particular educational outcomes. In his study on higher education participation in Italy, Gambetta (1996) found differences in participation rates between South, Central, and North Italy. Remarkably, the highest participation rates were found in the disadvantaged South. According to Gambetta, this finding can be explained through local labor market characteristics, especially unemployment rates. The reasoning goes that in regions where unemployment rates are high, students stay longer in education as opportunity costs are lower than in regions where jobs are readily accessible upon completion of compulsory schooling. This hypothesis, denoted as parking theory, runs counter to the propositions that derive from Roscigno and Crowley’s (2001) theoretical statement; while in the latter, areas characterized by depressed economic conditions would lead to lower educational expectations, the former theory predicts that these conditions will increase the propensity of staying on in education. However, unlike Roscigno and Crowley, Gambetta did not have data at his disposal to put this theory through a test.

In the Netherlands, some scholars have been particularly interested in spatial disparities in the recommendations given to pupils by primary school teachers at the transition from primary to secondary education (de Boer et al., 2006; Dronkers et al., 1998). Dronkers et al. (1998) found that pupils in the so-called Randstad cities—Rotterdam, Amsterdam, The Hague, and Utrecht—do more often receive the advice to enroll in academically oriented electives in secondary education than pupils with a similar achievement background in the rest of the Netherlands. Part of this association could be explained by the fact that these cities count fewer low educated parents and more parents with an immigrant background. de Boer et al. (2006) found that pupils in the rural province of Friesland are less often advised to enroll in academically oriented secondary education than comparably achieving pupils in the rest of the Netherlands. While some of this association could be explained by taking into account the background characteristics of pupils, a sizable amount of under-advising remained visible. An explanation in terms of local labor market opportunities by comparing the advices given to pupils in Friesland with the advices given in provinces that resemble Friesland with regard to their labor markets was refuted in this particular case. However, the authors stressed the fact that provinces might be too broad geographical units to examine such effects, as there is substantial within-province variation in labor market opportunity.

While past research has tended to explain spatial disparities in educational outcomes by referring to local opportunity structures and more specifically local labor market characteristics, most of these studies have taken for granted this assertion rather than effectively testing it. In this study, we aim to explore urban versus nonurban differences in educational choice in a European context and their possible relation to local labor market conditions.

Labor Markets and Course-Taking Patterns

By and large, there seems to be a consensus among scholars that differences in opportunity structures underlie differences in educational outcomes according to place (de Boer et al., 2006; Gambetta, 1996; Roscigno & Crowley, 2001; Roscigno et al., 2006). Places differ with regard to their labor market situation and this would then influence investments in education.

Alongside research examining differences in educational outcomes between places, there is also a research tradition directly focusing on how local labor markets influence educational decision making (e.g., Raffe & Willms, 1989; Rice, 1999). These studies show that local unemployment levels are positively related to the decision of staying on in education beyond compulsory education. Moreover, Raffe and Willms (1989) show that local labor markets characterized by a high proportion of jobs in service industries are associated with higher staying-on rates, among otherwise comparable pupils. What these studies demonstrate is that local labor markets influence people’s subjective evaluations of the benefits that they will derive from an investment in continued education. The more students have to win from an extended stay in education, the more they will be inclined to stay on.

Research that deals with the potential influence of local labor market characteristics on course-taking patterns during compulsory education seems to be scarce (de Boer et al., 2006). Nevertheless, there are reasons to suppose that local opportunity structures also influence this particular educational outcome. In fact, areas differ with regard to their occupational structure. While some places are characterized by a labor market that is still predominantly dependent on low-wage, labor-intensive jobs (in industry and agriculture), in other places it is clearly geared at a more highly educated workforce. Moreover, these occupational structures tend to be quite stable over time. As a result, there might be less incentives for pupils attending schools in areas in which the local economy mainly depends on low-skilled labor to choose for those academic courses logically leading to higher education. Conversely, for those pupils who attend schools in places where the local economy chiefly requires highly educated workers, the incentive to enroll in academically oriented secondary education might be much greater.

Study Setting

Flanders, the Northern, Dutch-speaking part of Belgium is a highly urbanized region, characterized by urban sprawl, that is, outward expansion of urban areas into previously rural areas (Poelmans & Van Rompaey, 2009). This phenomenon, caused by economic growth in the 1960s and 1970s and a permissive spatial policy, has led Flanders to be one of the most urbanized regions in Western Europe. Therefore, one can hardly speak about a clear urban–rural divide in Flanders (Antrop, 2004). For this reason, in this study, we will make a rather rough distinction between urban areas, that is cities, and nonurban areas, that is municipalities which are generally less densely populated than these cities and depend on them for certain services (Loopmans et al., 2011; Schreurs, 1986).

The outcome variable in this study is educational choice as choice during compulsory education is a prominent feature of most continental education systems. Moreover, educational decisions made throughout compulsory schooling are very consequential for future educational and occupational prospects (Vanderheyden & Van Trier, 2008). In Flanders, compulsory education starts at the age of 6 (Department of Education, 2008). At that age, pupils begin with primary education, which is undifferentiated and lasts 6 years. After these 6 years, at the age of 12, pupils make the transition to secondary education, which is tracked. Roughly speaking, one can distinguish four hierarchically ordered tracks, that is, academic education, technical education, arts education, and vocational education (see Figure 1). While students who graduate from any of these four tracks—those in vocational education have to attend an additional 7th year of schooling to graduate—are entitled to enroll in higher education, their chances of success largely depend on which of these tracks they completed (Vanderheyden & Van Trier, 2008). Students with a certificate of academic education are much more likely to be successful in higher education than students with a certificate of technical or vocational education.

Figure 1.

Schematic presentation of the Flemish educational system.

However, upon completion of primary education pupils along with their parents are confronted with a basic choice between A-stream and B-stream. B-stream is said to offer education for those pupils who are less fit for theoretical tuition. Actually, B-stream groups those pupils who do not reach the learning objectives that should be reached by the end of primary education. Yet, the great majority of pupils enter A-stream in the first year of secondary education. A-stream is said to offer a common curriculum to all pupils, which prepares them to make a choice between one of the four previously mentioned tracks after 2 years of secondary school. Nevertheless, there is some differentiation within A-stream in the form of optional courses such as Latin, modern sciences, technology, and arts. Whereas Latin and modern sciences prepare pupils for academic education, technology and arts prepare for technical and arts education, respectively. Pupils who choose to enroll in academically oriented electives within the A-stream can always choose to quit these courses and to enroll in nonacademically oriented electives. The reverse movement—from nonacademically oriented electives within the A-stream to academically oriented electives within the A-stream—however, is much more difficult to achieve (Van Damme et al., 1997). So while pupils who choose to enroll in academically oriented tracks have all options open to them, their peers who choose to enroll in nonacademically oriented electives already preclude certain choices. Knowing that chances of success in higher education differ according to the track in which one finally graduates, educational decisions made at the onset of secondary education can be consequential for one’s future opportunities.

Education in Flanders is costless and parents are entirely free to choose the school of their preference. Moreover, there are no standardized tests or entrance requirements for enrollment in A-stream or B-stream, or the options within A-stream. As a result, pupils along with their parents have a great deal of discretion in deciding which educational alternative to enroll in. This openness to individual decision making has led to processes of self-selection (Boone & Van Houtte, 2013), comparable with those found in similar educational systems in neighboring countries (Ditton & Krüsken, 2006: Germany; Jackson, Erikson, Goldthorpe, & Yaish, 2007: United Kingdom; Kloosterman et al., 2009: the Netherlands). In fact, pupils from higher socioeconomic backgrounds are more likely to choose the academically oriented electives within A-stream—that is, Latin and modern sciences—than pupils from lower socioeconomic backgrounds, even if these latter achieved comparably well (Boone & Van Houtte, 2013).

To our knowledge, there are no studies on Flanders that focus on potential differences in educational decision making depending on attending a primary school in an urban rather than a nonurban area, despite the fact that research in the Netherlands has suggested that such differences might exist (Dronkers et al., 1998).

Data

The data used in this study were collected as part of a research project on processes of educational choice at the transition from primary to secondary education funded by the Flemish Ministry of Education. Survey data were gathered during the months of May and June of 2008 from 1,339 parents of pupils in their last year of primary education in a sample of 53 primary schools. Participating schools were selected using a disproportionally stratified sample on the basis of three criteria, namely geographical spread, school sector and location—in an urban or nonurban environment. Samples of schools were drawn from official records of primary schools from the Flemish Educational Department. Selected schools were contacted and asked to participate. Of all the schools we contacted, 41.36% accepted to participate in this study. This low rate of positive responses is due to the fact that schools in Flanders are swamped with requests to participate in research. As schools often respond to this kind of requests using a logic of “first come, first served,” this data collection probably suffered more from negative responses, as it started at the end of the school year. We have no indication that this could be of influence on the results of our study. Once schools had accepted to participate, questionnaires were brought to the school sites by a researcher and collected about a month later. Somewhat less than 87% of all parents of pupils in last year of primary education in those schools participated, yielding information on 1,339 pupils. As we focus attention on particular decisions and due to missing data on particular variables, the number of cases used will vary according to the analysis (see “Design and Variables” section).

In addition, to retrieve data on local labor market characteristics, we turned to official statistics of the Flemish office for employment mediation (VDAB, 2008).

Variables

We successively considered three different dichotomous dependent variables: first, choice between A-stream (1) and B-stream (0); next, within the A-stream, choice between the academically oriented electives—Latin, modern sciences—(1) and the nonacademically oriented electives—technology, arts—(0); and, finally, within the academically oriented electives, choice between Latin (1) and modern sciences (0). We consider these particular choices as research has shown that students with a certificate of academic education are much more likely to study at university level rather than at colleges of higher education and are more likely to be successful in either type of higher education than pupils with a certificate of technical or vocational education (Vanderheyden & Van Trier, 2008). Moreover, recent research has shown that students having studied Latin in secondary education are more likely to succeed in their first year at university than pupils who did not study Latin (Pinxten et al., 2015). As can be seen from Table 1, only 5.4% of the parents indicated to have chosen to enroll their child in B-stream. This finding is not surprising, as B-stream is mainly recruiting pupils from primary schools for special education.

Table 1.

Descriptive Statistics for the Dependent and Independent Variables: Frequencies, Means, Minimum and Maximum Values, Standard Deviations.

Variables	M	SD	Minimum	Maximum
Independent variables
School level (N = 53)
Local labor market	70.94	14.43	39.28	94.13
Pupil level
Gender (N = 1,339)
Female	52.1%
Male	47.9%
SES (N = 1,274)	5.25	1.98	1	8
Ethnic origin (N = 1,337)
Belgian or Western European origin	88.5%
Non-Belgian or non-Western-European origin	11.5%
Mother’s educational attainment (N = 1,296)
Lower secondary education	13.4%
Higher secondary education	41.9%
Tertiary secondary education	44.7%
GPA (N = 1,118)	80.67	8.18	47	96
Dependent variables
A- or B-stream (N = 1,277)
A-stream	94.6%
B-stream	5.4%
A academic or A nonacademic within A stream (N = 1,172)
A academic	79.3%
A nonacademic	20.7%
Latin or modern sciences (N = 889)
Latin	43.6%
Modern sciences	56.4%

Note. GPA = grade point average; SES = socioeconomic status.

Urban Versus Nonurban

The main independent variable in this study was geographical location of the primary school, in an urban or nonurban area. To determine whether a school was located in an urban rather than in a nonurban environment, we used the official classification used by the Flemish Department for Environmental Planning (Loopmans et al., 2011). This classification, inspired on central place theory (Christaller, 1933), is mainly based on the amount of services available to the population in a certain municipality. The kinds of services taken into consideration to determine whether an area can be characterized as an urban rather than a nonurban area comprise health care facilities (hospitals, medical centers), recreation facilities (sports facilities, hotels, restaurants, bars), education and retail services, and also administrative facilities (principal seats of provinces or districts, courts of justice, principal seats of federal police services), so-called “counter services” (banks, post offices, temporary employment agencies, principal seats of tax and employment services), and transport facilities (train stations, public transportation bus services). On the basis of this classification, we made a rather rough distinction between places that can be characterized as cities (urban = 1), and places that cannot be characterized as cities (nonurban = 0). Population density per square kilometer was significantly higher in areas that we characterized as urban (mean = 946 inhabitants/square kilometer) than in areas that we characterized as nonurban (mean = 450 inhabitants/square kilometer). In all, 24 schools were located in areas that can be qualified as urban, the remaining 29 schools are located in nonurban areas.

Local Labor Market

Local labor market opportunity structure was measured as the proportion of jobs in the service sector in the primary school’s municipality. The service sector comprises all jobs in the tertiary and quaternary sectors. The basic characteristic of these sectors is that they provide services rather than end-products. While the tertiary sector refers to services such as transport, distribution, and entertainment, the quaternary sector refers to all knowledge-based services such as consulting, education, research and development, financial planning, and information technology. Jobs in the service sector and especially those in the quaternary sector require a more highly educated workforce than those in the industrial sector (secondary sector) and the agricultural sector (primary sector; see Bell, 1976; Powell & Snellman, 2004). The average percentage of jobs in tertiary and quaternary sectors in the municipalities in this study is 70.94 (SD = 14.43, Table 1). The correlation between urban–nonurban location and percentage of jobs in tertiary and quaternary sectors is .467 (p < .01).

Socioeconomic Status (SES)

Parental SES was measured by asking both parents what their occupation was at the time of the survey, or in case they were unemployed, what their previous occupation had been. Answers were then recoded by the researcher according to the classification by Erikson, Goldthorpe, and Portocarero (1979). Scores range from 1 to 8, in which 1 stands for unskilled manual labor, and 8 for managers, professionals, and company holders. To determine family SES, the highest of both scores was then used. The mean SES in the sample is 5.25 (SD = 1.98; Table 1).

Grade Point Average (GPA)

Pupils’ prior achievement was measured by asking parents what their children’s GPA was at the end of 5th year of primary school. As a lot of Flemish primary schools do not work with GPAs, 16.5% of the pupils in the sample have missing values on this variable. The mean for this variable is 80.67% (SD = 8.18; Table 1).

Mother’s Educational Attainment

The educational attainment of the mother of the pupil was measured by asking parents to indicate what the highest credential obtained by the mother of the child was. Parents could choose among the following: no education, primary education, lower secondary education, higher secondary education, tertiary not at university level, and tertiary at university level. Answers were recoded into three categories: lower secondary education or less (13.4%), higher secondary education (41.9%), and tertiary education (44.7%; Table 1). In multivariate analysis, we make use of two dummy variables, with higher secondary education as a reference category.

Gender

As for gender (male = 0, female = 1), 52.1% of the pupils in the sample were girls (Table 1).

Ethnic Origin

As common in Flanders (Belgium) and the Netherlands (Timmerman, Hermans, & Hoornaert, 2002), pupils’ ethnic origin was determined by asking parents the birthplace of the pupil’s maternal grandmother. If the pupil’s maternal grandmother was born in Belgium or another Western European country, the pupil got value 0, else the pupil got value 1. The sample consisted of 88.5% pupils with Belgian or Western European origin (Table 1).

Design

The aim of this study was to establish whether attending a primary school in an urban area rather than in a somewhat more isolated nonurban area influences educational decision making in Flanders (the northern, Dutch-speaking part of Belgium), and if so, whether such differentials can be explained by local labor market characteristics.

We assume that parents in Flanders make a series of binary choices when deciding between the educational alternatives available at the onset of secondary education. First, they have to choose whether to send their children to A- or B-stream; subsequently, whether within A-stream they will take up an option leading to academic education or rather an option that leads to technical or arts education; and finally, if they choose an academic option, whether it will be Latin or modern sciences.

As the data set is made up of a clustered sample of pupils nested within schools and involves data at different levels (pupil- and school level), the use of multilevel modeling is most appropriate (Snijders & Bosker, 1999). Furthermore, as the dependent variables are dichotomous, we make use of hierarchical logistic regression models. It is common in multilevel analyses to begin with examining unconditional models—that is without specifying any determinant—to determine the amount of variance that occurs among schools, but in hierarchical logistic models it is not appropriate to partition the variance in outcome into its between- and within-components. However, the between-school variance component estimated in an unconditional model does give an idea of whether or not the between school variance is significant and can be modeled (Frost, 2007; Lee & Burkam, 2003).

In the first model, we examine the possible effects of attending a primary school in an urban versus a nonurban environment, without further controls. However, as noted by Manski (2000), it is vital to take into consideration correlated or selection effects when examining contextual effects. Therefore, in the second model, we control for pupil’s SES, GPA, gender, ethnic origin, and mother’s educational attainment to rule out possible selection effects. If geographical setting of the primary school, that is, in an urban versus a nonurban area, is still related to pupils’ educational choices, in the next model, we then enter a measure of local labor market opportunity to examine whether this can account for the impact of attending an urban versus a nonurban school. All independent variables are grand mean centered, except for the dichotomous and dummy variables (gender, ethnic origin, and mother’s educational level), which are uncentered for reasons of interpretation, and slopes are allowed to vary across schools.

Results

A- Versus B-Stream

For the choice between A- and B-stream, the between-school variance in the unconditional model (τ₀ = .001) appeared to be nonsignificant (p > .50), meaning that there was no variance to be explained at the school level. As a result, it was of little use to take school-level variables into consideration. Moreover, preliminary analyses revealed that the choice between A- and B-stream is primarily determined by pupils’ achievement (Boone & Van Houtte, 2010). We therefore focused attention on the other choices, namely the choice within A-stream between academically oriented electives and nonacademically oriented electives, and within the academically oriented electives, between Latin and modern sciences.

Academically Oriented Electives Versus Nonacademically Oriented Electives

For the choice within A-stream, between academically oriented and nonacademically oriented electives, the between-school variance estimated in an unconditional model indicated that it was useful to estimate a model taking into consideration school-level variables (τ₀ = .508, p < .001). In the first model without additional controls (Table 2), we found that attending a school in an urban environment has a borderline significant (p = .06, only 6% chance that this finding is coincidental) positive effect on choice between academically and nonacademically oriented electives. Pupils attending schools in urban areas were more likely to choose to enroll in the academically oriented electives within A-stream (probability of 82.8%) than pupils attending schools in nonurban areas (probability of 75.5%). To ascertain whether this effect was not due to differences between the pupil populations attending these schools (selection effects), in the second model, we controlled for pupils’ SES, GPA, gender, ethnic origin, and mother’s educational attainment. The effect of attending school in an urban area became clearer on inclusion of these controls. In fact, among otherwise comparable pupils, those attending primary schools in urban areas appeared to be more likely to choose the academically oriented electives than those attending primary schools in nonurban areas. The second model also confirms the determining influence of pupils’ SES and GPA on pupils’ choice between academically and nonacademically oriented electives (see also Boone & Van Houtte, 2013). Pupils who stem from high SES families and pupils who achieved well in primary education were more likely to choose to enroll in the academically oriented electives. Furthermore, girls appeared to be more likely to opt for the academically oriented electives than comparable boys. In the third model, we included the measure capturing the local labor market opportunity structure. Upon inclusion of this variable, the urban versus nonurban effect was clearly attenuated and lost its significance altogether (p = .208). However, the proportion of jobs in the tertiary and quaternary sectors in a municipality was not in itself significantly related to the choice between academically and nonacademically oriented electives in Model 3 (p = .232). Nevertheless, additional analyses have shown that both the urban versus nonurban variables as well as local labor market opportunity are significantly positively related to the choice between academically and nonacademically oriented electives.

Table 2.

Hierarchical Logistic Modeling Estimates of Choice for Academically Oriented Electives Versus Choice for Nonacademically Oriented Electives.

	Model 1 (N = 1,172)	Model 2 (N = 938)	Model 3 (N = 938)
Intercept	1.137***	1.604***	1.723***
	3.119	4.976	2.276
	(0.143)	(0.198)	(0.230)
School level
Urban versus nonurban	.440^†	.648*	.436
	1.553	1.913	1.546
	(0.230)	(0.300)	(0.342)
Local labor market			.013
			1.014
			(0.011)
Individual level
Gender		.438*	.428*
		1.550	1.536
		(0.199)	(0.200)
SES		.307***	.309***
		1.360	1.362
		(0.064)	(0.063)
Ethnic origin		.612	.543
		1.843	1.721
		(0.603)	(0.600)
Mother’s educational attainment
≤Lower secondary		−.215	−.239
		0.806	0.787
		(0.410)	(0.410)
Higher education		.078	.052
		1.081	1.053
		(0.244)	(0.244)
GPA		.243***	.244***
		1.275	1.276
		(0.021)	(0.021)

Note. The three rows for each parameter represent gamma coefficients, odds ratios (in italics), and standard errors (in parentheses). GPA = grade point average; SES = socioeconomic status.

†

p = .06. *p < .05. ***p ≤ .001.

Latin Versus Modern Sciences

With regard to the choice between Latin and modern sciences (Table 3), the between-school variance estimated in an unconditional model indicated that it was useful to estimate a model taking into account school-level variables (τ₀ = .290, p < .001). The first model containing only the urban versus nonurban variable showed that attending a school in an urban area increased the propensity of choosing Latin (p = .001). Pupils attending schools in an urban area were more likely to choose Latin than pupils attending schools in nonurban areas. In the second model, we controlled for pupils’ SES, GPA, gender, ethnic origin, and mother’s educational attainment. Upon inclusion of these variables, the effect of the urban versus nonurban variable became more pronounced. This means that among otherwise comparable pupils, those who attended a primary school in an urban area were more likely to choose Latin than those who attended a primary school in a nonurban area. Furthermore, the determining impact of pupils’ SES and GPA on choice between Latin and modern sciences was once again clearly visible. The higher the SES of a pupil and the better a pupil achieved in primary school, the more inclined he or she was to opt for Latin in the first year of secondary education. In the third model, we included the measure capturing local labor market opportunity structure and found that the effect of attending a primary school in an urban area dropped and lost in significance. The variable for local labor market opportunity appeared to have a positive influence on the propensity of choosing Latin (p = .066, or only 6.6% chance that this finding is coincidental). The greater the share of jobs in the tertiary and quaternary sectors in a municipality, the greater the likelihood of pupils attending schools in these municipalities to choose Latin.

Table 3.

Hierarchical Logistic Modeling Estimates of Choice for Latin or Modern Sciences.

	Model 1 (N = 889)	Model 2 (N = 719)	Model 3 (N = 719)
Intercept	−.605***	−1.014***	−.954***
	0.546	0.363	0.385
	(0.138)	(0.258)	(0.262)
School level
Urban versus nonurban	.648***	.903***	.660*
	1.911	2.467	1.934
	(0.180)	(0.249)	(0.302)
Local labor market			.018^†
			1.018
			(0.010)
Individual level
Gender		−.002	.034
		0.998	1.035
		(0.199)	(0.203)
SES		.234**	.234**
		1.263	1.263
		(0.072)	(0.073)
Ethnic origin		.695	.692
		2.004	1.998
		(0.482)	(0.464)
Mother’s educational attainment
≤Lower secondary		−.044	−.130
		0.957	0.878
		(0.394)	(0.389)
Higher education		−.002	.019
		0.998	1.019
		(0.257)	(0.263)
GPA		.251***	.250***
		1.285	1.263
		(0.029)	(0.030)

Note. The three rows for each parameter represent gamma coefficients, odds ratios (in italics), and standard errors (in parentheses). GPA = grade point average; SES = socioeconomic status.

†

p = .066. *p < .05. **p ≤ .01. ***p ≤ .001.

Discussion and Conclusion

In most European education systems, the transition from primary to secondary education is a crucial branching point in a pupil’s school career. The choice between academically and nonacademically oriented electives can be very consequential for further life chances. So far, scholars have mainly been interested in the impact of pupils’ individual characteristics on their choice between academically and nonacademically oriented courses, demonstrating the influence of parental SES time and time again. In this study, we wanted to explore the impact of a particular contextual variable, namely primary school’s location in an urban or nonurban environment, on pupils’ choice at the transition from primary to secondary education in Flanders (the northern, Dutch-speaking part of Belgium).

We found that pupils who attend primary schools in urban areas tend to aim higher than pupils who attend primary schools in nonurban areas. In fact, pupils attending schools in urban areas are more likely to choose the academically oriented electives and more likely to choose Latin than comparable pupils attending schools in nonurban areas. These findings are all the more remarkable, given the fact that Flanders is known to be one of the most urbanized regions in Western Europe (Poelmans & Van Rompaey, 2009). So while from the outside Flanders may be considered as one large urban community, this study suggests that the specific context in which schooling takes place matters. Attending a school in an area that can be qualified as a city owing to the fact that it gives access to a lot of important services, such as health care, increases pupils’ chance of choosing the academically oriented electives at the start of secondary education.

The analyses further suggested that a part of this association can be explained by local labor market characteristics. The fact that pupils attending a primary school in an urban environment are more inclined to choose the academically oriented electives seems to be related to the fact that the labor market in those areas is more heavily dependent on a highly educated workforce. Conversely, the fact that pupils attending schools in nonurban areas tend to choose more for the nonacademically oriented electives appears to be connected to the fact that the labor market in those areas is less dependent on highly educated work. However, our study does not tell us how local labor market characteristics influence decision making in schools and families. It merely suggests that these urban–nonurban differences in educational choice are related to the opportunity structure characteristic of urban and nonurban areas.

Future research should examine how precisely these labor market characteristics influence processes of educational decision making at the school- and the family level. Could it be that teachers in nonurban schools expect less from children as they take the local labor market opportunities as a frame of reference? Could it be that parents who send their children to nonurban schools expect less from their children? These are questions that this study could not answer.

Based on the findings of this study, one could say that pupils attending primary schools in nonurban areas are at a disadvantage compared with pupils in urban areas. In fact, pupils attending primary schools in a nonurban environment are less likely than comparable pupils attending schools in urban areas to choose for these educational pathways that leave all future options open. Local educational boards could be made conscious of this particular course-taking pattern, so that they may encourage pupils in nonurban areas to enroll in academically oriented electives within first year of secondary education at an equal level.

While research on urban–nonurban differences in educational outcomes had so far been restricted to U.S.-based studies, this study shows that such research is also warranted in a European context. Even in the highly urbanized region of Flanders, we could establish urban–nonurban disparities in an educational outcome, which is crucial in this specific setting, namely the educational decision made by pupils at the transition from primary to secondary education. This finding demonstrates that taking into consideration the wider context in which education takes place is essential for getting a more complete understanding of educational processes. The present study is only a starting point in regard to this.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the Flemish Ministry of Education and Training, OBPWO 07.03, and by the Special Research Fund of Ghent University, project 01J15110.

Author Biographies

Simon Boone, PhD in sociology (Ghent University), is a sociologist of education. He is currently associated to the Department of Political Science of the Vrije Universiteit Brussel, Belgium. His research interests include inequality of educational opportunity, education systems and mixed method research designs.

Mieke Van Houtte, PhD in sociology, is full professor and head of the research team CuDOS at the Department of Sociology at Ghent University (Belgium). Her research interests cover diverse topics within the sociology of education, particularly the effects of structural and compositional school features on several outcomes for students and teachers.

References

Andersson

Subramanian

S. V.

(2006). Explorations of neighbourhood and educational outcomes for young Swedes. Urban Studies, 43, 2013-2025. doi:10.1080/00420980600897834

Antrop

(2004). Landscape change and the urbanization process in Europe. Landscape and Urban Planning, 67, 9-26. doi:10.1016/S0169-2046(03)00026-4

Aypay

(2003). The tough choice at high school door: An investigation of the factors that lead students to general or vocational schools. International Journal of Educational Development, 23, 517-527. doi:10.1016/S0738-0593(03)00040-3

Bell

(1976). The rise of post-industrial society: A venture in social forecasting. Harmondsworth, UK: Penguin.

Boone

Van Houtte

(2010). Sociale ongelijkheid bij de overgang van basis- naar secundair onderwijs. Eindrapport van het OBPWO-project 07.03 [Social inequality at the transition from primary to secondary education. Final report of the OBPWO-project 07.03]. Gent: Universiteit Gent.

Boone

Van Houtte

(2013). In search of the mechanisms conducive to class differentials in educational choice: A mixed method research. The Sociological Review, 61, 549-572. doi:10.1111/1467-954X.12031

Brannstrom

(2008). Making their mark: The effects of neighbourhood and upper secondary school on educational achievement. European Sociological Review, 24, 463-478. doi:10.1093/esr/jcn013

Christaller

(1933). Die zentralen Orte in Süddeutschland [The central places in Southern Germany]. Jena, Germany: Fischer.

Coleman

J. S.

Campbell

E. Q.

Hobson

C. J.

McPartland

Mood

A. M.

Weinfeld

F. D.

York

R. L.

(1966). Equality of educational opportunity. Washington, DC: U.S. Department of Health, Education and Welfare.

10.

de Boer

van der Werf

M. P. C.

Bosker

R. J.

Jansen

G. H.

(2006). Onderadvisering in de provincie Friesland [Underadvising in the province of Friesland]. Pedagogische Studiën, 83, 452-468.

11.

Department of Education. (2008). The Flemish educational landscape in a nutshell. Brussels, Belgium: Die Keure.

12.

Ditton

Krüsken

(2006). Der Übergang von der Grundschule in die Sekundarstufe I [The transition from primary to secondary education]. Zeitschrift für Erziehungswissenschaft, 9, 348-372. doi:10.1007/s11618-006-0055-7

13.

Dronkers

van Erp

Robijns

Roeleveld

(1998). Krijgen leerlingen in de grote steden en met name in Amsterdam te hoge adviezen? De relaties tussen taal- en rekenscores en advies binnen en buiten de Randstad onderzocht [Do pupils get higher advices in large cities and in particular in Amsterdam? An investigation into the relation between Dutch and math scores and advice within and outside Randstad]. Tijdschrift voor Onderwijsresearch, 23, 17-30.

14.

Erikson

Goldthorpe

Portocarero

(1979). Intergenerational class mobility in three West European countries: England, France and Sweden. British Journal of Sociology, 30, 415-441.

15.

Frost

M. B.

(2007). Texas student’s college expectations: Does high school racial composition matter? Sociology of Education, 80, 43-66.

16.

Gambetta

(1996). Were they pushed or did they jump? Individual decision mechanisms in education. Boulder, CO: Westview Press.

17.

Garner

C. L.

Raudenbush

S. W.

(1991). Neighborhood effects on educational attainment: A multilevel analysis. Sociology of Education, 64, 251-262. doi:10.2307/2112706

18.

Jackson

Erikson

Goldthorpe

Yaish

(2007). Primary and secondary effects in class differentials in educational attainment. Acta Sociologica, 50, 211-229. doi:10.1177/0001699307080926

19.

Jaeger

M. M.

(2009). Equal access but unequal outcomes: Cultural capital and educational choices in a meritocratic society. Social Forces, 87, 1943-1972.

20.

Jonsson

J. O.

Mood

(2008). Choice by contrast in Swedish schools: How peers’ achievement affects educational choice. Social Forces, 87, 741-765.

21.

Kauppinen

(2008). Schools as mediators of neighborhood effects on choice between vocational and academic tracks of secondary education in Helsinki. European Sociological Review, 24, 379-391. doi:10.1093/esr/jcn016

22.

Kloosterman

Ruiter

de Graaf

P. M.

Kraaykamp

(2009). Parental education, children’s performance and the transition to higher secondary education: Trends in primary and secondary effects over five Dutch school cohorts (1965-99). The British Journal of Sociology, 60, 377-398. doi:10.1111/j.14684446.2009.01235.x

23.

Lee

V. E.

Burkam

D. T.

(2003). Dropping out of high school: The role of school organization and structure. American Educational Research Journal, 40, 353-393. doi:10.3102/00028312040002353

24.

Loopmans

Van Hecke

De Craene

Martens

Schreurs

Oosterlynck

(2011). Selectie van kleinstedelijke gebieden in Vlaanderen. Een onderzoek in opdracht van Vlaamse Overheid Departement Ruimtelijke Ordening, Woonbeleid en Onroerend erfgoed, Afdeling Ruimtelijke Planning [Selection of (small) urban places in Flanders. A research commissioned by the Flemish Government Department of Spatial Planning, Housing Policy and Immovable Heritage, Section Spatial Planning]. http://www2.vlaanderen.be/ruimtelijk/docs/Studie_kleinstedelijkegebieden.pdf

25.

Manski

C. F.

(2000). Economic analysis of social interactions. Journal of Economic Perspectives, 14, 115-136. doi:10.1257/jep.14.3.115

26.

Pinxten

De Fraine

Van Den Noortgate

Van Damme

Boonen

Vanlaar

(2015). ‘I choose so I am’: a logistic analysis of major selection in university and successful completion of the first year. Studies in Higher Education, 40, 1919-1946.

27.

Poelmans

Van Rompaey

(2009). Detecting and modelling spatial patterns of urban sprawl in highly fragmented areas: A case study in the Flanders-Brussels region. Landscape and Urban Planning, 93, 10-19. doi:10.1016/j.landurbplan.2009.05.018

28.

Powell

Snellman

(2004). The knowledge economy. Annual Review of Sociology, 30, 199-220.

29.

Raffe

Willms

J. D.

(1989). Schooling the discouraged worker: Local-labour market effects on educational participation. Sociology, 23, 559-581.

30.

Reeves

E. B.

(2012). The effects of opportunity to learn, family socio-economic status, and friends on the rural math achievement gap in high school. American Behavioral Scientist, 59, 887-907. doi:10.1177/0002764212442357

31.

Reeves

E. B.

Bylund

R. A.

(2005). Are rural schools inferior to urban schools? A multilevel analysis of school accountability trends in Kentucky. Rural Sociology, 70, 360-386. doi:10.1526/0036011054831215

32.

Rice

(1999). The impact of local labour markets on investment in further education: Evidence from the England and Wales youth cohort studies. Journal of Population Economics, 12, 287-312.

33.

Roscigno

V. J.

(1998). Race and the reproduction of educational disadvantage. Social Forces, 76, 1033-1060.

34.

Roscigno

V. J.

Crowley

M. L.

(2001). Rurality, institutional disadvantage, and achievement/attainment. Rural Sociology, 66, 268-292.

35.

Roscigno

V. J.

Tomaskovic-Devey

Crowley

(2006). Education and the inequalities of place. Social Forces, 84, 2121-2145. doi:10.1353/sof.2006.0108

36.

Rosenbaum

(2001). Beyond college for all: Career paths for the forgotten half. New York, NY: Russell Sage Foundation.

37.

Schreurs

(1986). Een plattelandsatlas voor Vlaanderen [A rural atlas for Flanders]. Leuven, Belgium: Stichting Plattelandsbeleid VZW (Vereniging Zonder Winstoogmerk).

38.

Schulze

Wolter

Unger

(2009). Bildungschancen von Grundschülern: Die Bedeutung des Klassen- und Schulkontextes am Übergang auf die Sekundarstufe I [Educational opportunities of primary school pupils. The significance of class- and school context at the transition from primary to secondary education]. Kölnischer Zeitschrift für Soziologie und Sozialpsychologie, 61, 411-435. doi:10.1007/s11577-009-0072-7

39.

Snijders

T. A. B.

Bosker

R. J.

(1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London, England: Sage.

40.

Strayhorn

T. L.

(2009). Different folks, different hopes: The educational aspirations of black males in urban, suburban and rural high schools. Urban Education, 44, 710-731. doi:10.1177/0042085908322705

41.

Sykes

Kuyper

(2009). Neighbourhood effects on youth educational achievement in the Netherlands: Can effects be identified and do they vary by student background characteristics? Environment and Planning, 41, 2417-2436. doi:10.1068/a41255

42.

Sykes

Musterd

(2011). Examining neighbourhood and school effects simultaneously: What does the Dutch evidence show? Urban Studies, 48, 1307-1331. doi:10.1177/0042098010371393

43.

Thissen

Droogleever Fortuijn

Strijker

Haartsen

(2010). Migration intentions or rural youth in the Westhoek, Flanders, Belgium and the Veenkoloniën, The Netherlands. Journal of Rural Studies, 26, 428-436. doi:10.1016/j.jrurstud.2010.05.001

44.

Timmerman

Hermans

Hoornaert

(2002). Allochtone jongeren in het onderwijs. Een multidisciplinair perspectief [Allochtonous youngsters in education. A multidisciplinary perspective]. Leuven, Belgium: Garant.

45.

Van Damme

De Troy

Meyer

Minnaert

Lorent

Opdenakker

M.-C.

Verduyckt

(1997). Succesvol Doorstromen in de Aanvangsjaren van het Secundair Onderwijs [Successful transitions through the first years of secondary education]. Leuven, Belgium: Acco.

46.

Vanderheyden

Van Trier

(2008). Sociale Afkomst, Onderwijs en Maatschappelijke Positie van Vlaamse Jongeren bij Begin van de Loopbaan [Social background, education and social position of Flemish youngsters at the beginning of their professional career]. In Vranken

Jan

. (Eds.), Armoede en Sociale Uitsluiting: Jaarboek 2008 (pp. 155-176). Leuven, Belgium: Acco.

47.

VDAB (Vlaamse Dienst voor Arbeidsbemiddeling en Beroepsopleiding). (2008). http://arvastat.vdab.be/arvastat/index.html