The reproduction of geographically advantaged student groups - trends in metropolitan student enrolment at Swedish university colleges

ABSTRACT

This study examines how Swedish students originating from metropolitan areas have used university colleges to access higher education. In the 1970s, as part of a series of reforms to the Swedish higher education system, university colleges were established. One reason being to make higher education more accessible to students outside the metropolitan areas. However, students from metropolitan areas have increasingly enrolled in these institutions over time. Multinomial logistic regression was used to examine the pre-entry attributes of metropolitan students from four enrolment cohorts (1995, 1998, 2001, and 2004) and patterns in their enrolment at university colleges. The results indicate that over time metropolitan students from less advantaged backgrounds have increasingly attended university colleges, particularly those whose qualifications excluded them from entering the more prestigious universities.

KEYWORDS:

• Higher education
• widening participation
• access
• enrolment

Introduction

Widening participation in Swedish higher education has been high on the political agenda since the 1940s. One major step in widening participation was the implementation of the Higher Education Act of 1977 (SFS, 1977, p. 263), which set forth a major re-organization of the higher education system by increasing the number of higher education institutions and distributing them more equally around the country. Consequently the student population became more diverse. Students from the working classes and those living outside metropolitan areas¹ had increased access to higher education (Berggren & Cliffordson, 2012; HSV, 2011).

Although one could argue that the initiative to expand higher education access to non-metropolitan areas was successful, inequalities in access have persisted (Berggren, 2008) and have increased between some student groups (Berggren & Cliffordson, 2012; Salmi & Bassett, 2014). For example, several of Sweden’s oldest, and what are perceived as most prestigious higher education institutions, are located in and around metropolitan areas. This meant that many students originating from Sweden’s metropolitan areas can be considered geographically advantaged because they have access to these institutions without moving.

Where students attend higher education can be a signal of quality to future employers, so attending a prestigious university can have implications for social mobility and status attainment (Rivera, 2011). However, participation in Swedish higher education grew by approximately 129% between 1977 and 2016 (UKÄ, 2018b), leading to greater competition at Sweden’s most prestigious institutions. Hence, not all metropolitan students may be able to enter the old, prestigious universities proximate to where they grew up. They may need to adjust their educational aims. Such an adjustment may be to apply for admission at a less prestigious higher education institution (Kivinen et al., 2001) such as a university college. Hence the purpose of this study is to examine how Swedish students originating from metropolitan areas have used university colleges to access higher education. The following questions guide this study.

• Which groups of Swedish metropolitan students, with respect to parental education, upper secondary school grades and gender, enrol in university colleges?

• How have these enrolment patterns changed over time?

Answers to these questions are important to understand how geographically, socially and meritocratically advantaged groups of students may use university colleges to reproduce their social positions. While geographical advantage refers to having access to prestigious higher education institutions without moving, in this study, social advantage refers to having tertiary educated parents and meritocratic advantage as having high upper secondary school grades. These latter two sources of advantage are sometimes referred to as educational capital (i.e. educational resources) (Bourdieu, 1984; Krigh & Lidegran, 2010).

Student participation in Swedish higher education

From the late 1980s and through the early 2000s, student participation in Swedish higher education increased dramatically (see Figure 1). The number of study places in higher education institutions grew from about 75,000 to 400,000 (SOU, 2007, p. 35). This resulted in the number of degrees awarded during this period rising from about 30,000 in 1990 to more than 75,000 in 2017 (UKÄ, 2018a).

Figure 1. Registered students at the bachelor and master levels by academic year and gender 1977/78-2016/17. (Source: UKÄ, 2018b)

The reasons for this growth are in part due to a series of reforms to the higher education system that began in the 1960s and extended into the 1990s. As part of the 1977 re-organization of the higher education system (SFS, 1977, p. 263), 17 university colleges were established and short vocational schools were incorporated into the system. For example, teacher training institutes became housed within university colleges (HSV, 1998). Initially this led to a large increase in the number of women participating in higher education, particularly in the less prestigious programmes and institutions, but this trend has since changed. Women have become more equally represented in prestigious sectors of Swedish higher education, and in some programs they have come to dominate (Börjesson & Broady, 2016). However, access to Sweden’s old, prestigious higher education institutions became even more competitive as these institutions enrolled a smaller share of students as compared to the period before the re-organization of the higher education system (Börjesson & Broady, 2016). In other words, the main growth in student participation occurred at university colleges.

Another possible reason for rising student enrolments during this period may be due to an economic recession and higher education being used as a political strategy to reduce unemployment rates (Börjesson & Broady, 2016). In the early 1990s, Sweden experienced a financial crisis as a result of a housing bubble.² Then, in the latter part of the 1990s and early 2000s many employees in the technology industries lost their jobs. This was caused by the bursting of the dot-com bubble.³ Corresponding to these events, Sweden entered a recession and experienced low rates of job stability, particularly among youth (16–24 years old) (Statistics Sweden, 2005). An alternative to unemployment is to enrol in higher education and earn credentials to avoid potential future bouts of unemployment. Hence, students who otherwise may not have initially planned to enrol in higher education due to poor grades, for example, may have decided to enrol due to instability in the labour market.

The spatial arrangement of Swedish higher education

The spatial arrangement of higher education refers to the geographical distribution of higher education institutions (e.g. their locations in or near cities) and various aspects that characterize their education and research endeavours (Haley, 2017). An awareness of this arrangement is important for understanding the sorts of educational opportunities available in metropolitan areas and elsewhere and how this arrangement might relate to students’ choice of a higher education institution. Prior to the reforms of the 1960s and 1970s, higher education in Sweden was an elite privilege that was offered by eight universities. These old, well-established universities were located in or near the large metropolitan areas, that is, Stockholm-Uppsala, Gothenburg, and Lund-Malmö (Hallonsten & Holmberg, 2013). With the re-organization of several vocational schools (e.g. healthcare and teacher training institutes) into university colleges, access to higher education opened to students in small urban areas and offered an alternative route into higher education for individuals in metropolitan areas living within commuting distance or willing to move (HSV, 1998). Additionally, some university colleges started as branch campuses of the old universities and later became standalone institutions. Changing institutional affiliations and status in the higher education system raises the complexity in understanding the sorts of education available to students in different regions over time.

Take Malmö University, for instance. Malmö University began as a teacher training institute before being re-organized as a branch campus of Lund University. Then in 1998, Malmö University College was formed and teacher training was transferred to the university college. This meant that suddenly the teacher training that was available to students in Malmö went from being a programme offered at an old university to a program offered by just a university college (Egidius, 2001). Since that time, Malmö University College has been granted the status of a full research university, albeit a new university. In other words, while most university colleges are located outside the large metropolitan areas, that has not always been the case over time.

Through the various reforms, a hierarchy of higher education institutions emerged, with old universities, new universities, and university colleges being marked by different characteristics. University colleges in Sweden are generally small and medium-sized institutions that primarily offer bachelor degrees; however, a few of them have been granted the ability to award doctoral degrees and gained the status ‘new universities’ since their establishment. Disciplines generally taught at university colleges include the humanities, social sciences, health sciences, and engineering. While there is some overlap in the sorts of educational programs offered at each type of higher education institution, the most selective, longer professional programs (e.g. medicine, law, engineering) tend to be offered at old and prestigious universities (Börjesson et al., 2020). Further, these university colleges primarily have education-oriented missions while old universities, for instance, maintain combined missions of research and education. Such divisions are reinforced through the Swedish government’s research budget allocation. Only 12% of the total annual research budget from the government is allotted to university colleges and new universities combined; the remainder is awarded to old universities (Hallonsten & Holmberg, 2013). These divisions may influence and perpetuate perceived conceptions of prestige, or lack thereof, in relation to Swedish higher education institutions among youth and their parents.

Students’ choice of higher education institution

International research shows that the reputations of higher education institutions have a role in students’ choice of institution (McManus et al., 2017). In Sweden, prior research shows that there is a relationship between the different types of higher education institutions and the backgrounds of students who enrol in them. For instance, Cliffordson and Gustafsson (2007) found that Swedish students with greater social or meritocratic advantage generally attend the old universities. Börjesson and Broady (2016) offer further support to these findings. They found that students from higher social classes, including those with greater financial resources, or who have tertiary educated parents tend to enrol in old universities. They also found that students with high upper secondary school grades or scores on the Swedish Scholastic Assessment Test (SweSAT) more often enrol in programs and higher education institutions that are perceived as prestigious. While students’ educational resources, such as parental education and grades, have been linked with students’ choice of higher education institution and study program, gender is more closely related to their choice of program (Berggren, 2008; Börjesson & Broady, 2016; UKÄ, 2018c). These prior studies offer a good foundation for understanding students’ choice of a higher education institution in Sweden at specific points in time, but what they have not examined are changes in students’ choice of higher education institutions over time by their educational resources, gender and geographical origin. Therefore, this study contributes a longitudinal perspective on these relationships.

Earlier research in the UK (Gibbons & Vignoles, 2009; Mangan et al., 2010), Australia (Parker et al., 2016), Norway (Helland & Heggen, 2017), and the USA (Griffith & Rothstein, 2009) shows a trend that having high geographical advantage, such as living near elite or prestigious institutions, increases the likelihood that a young person will attend one of these institutions. The influence of distance on institutional choice in the USA is most likely due to a concentration of elite higher education institutions in the northeastern part of the country. A concentration of prestigious universities in metropolitan areas in other countries such as Norway and Australia also likely contribute to the relationship of distance on institutional choice. While the relationship of distance and proximity between a student’s home and their choice of higher education institution has been previously researched, what is less known is in which types of higher education institutions students from different geographical areas, particularly metropolitan areas, access higher education. Hence, this study contributes to filling this knowledge gap by focusing on the enrolment patterns of Swedish students originating from metropolitan areas.

Notions of place attachment and place identity can offer insight to understanding geographical differentiation in educational choice. For example, place attachment refers to the features of a place that serve as criterion for choosing one place over another (Pollini, 2005). The assumption is that students have ‘insider advantages’ (i.e. geographically embedded assets) attached to their place of origin that makes moving to other places for higher education less advantageous (Dahl & Sorenson, 2010; Wikhall, 2002). Place attachment also corresponds with theories of place identity, which refers to the relationship between people and places and the role that place has in shaping an individual’s identity and perceptions of belonging and life purpose (McAndrew, 1998). Numerous studies have used these theoretical perspectives to understand students’ choice of higher education study location and their transition from home to university (see e.g. Chow & Healey, 2008; Cicognani et al., 2011; Gabriel, 2006; Holton, 2015). In these studies, the focus is on students’ perceptions and feelings with regards to place and belonging. However, in the data available for this study, there is no information on students’ perceptions or sense of belonging, which means that the focus must be on the choice of higher education institution that students actually make.

Social reproduction through enrolment in higher education

Dominant theoretical explanations for enrolment patterns in higher education have primarily come from social or cultural reproduction theories with ties to education, such as those of Pierre Bourdieu. The focus has generally been Bourdieu’s conceptualization of ‘habitus’ and how students’ socially determined preferences by way of habitus have influenced their enrolment behaviours and preference of academic programs and higher education institutions. Habitus is in effect an individual’s common-sense understanding of the social world and how to behave within it (Bourdieu, 1989). Bourdieu (1989) describes an individual’s habitus as ‘schemes of perceptions, thought and action’ (14). In other words, habitus offers individuals a sense of what is forthcoming, what they can or should expect and anticipate.

Habitus is a reflection of an individual’s social position. According to Bourdieu, an individual’s social position is synonymous with the amount and composition of their accumulated capital (i.e. their resources) (Bourdieu, 1985, 1986, 1989). The primary forms of capital according to Bourdieu (1985) are economic and cultural capital because he considers these as fundamental differences between individuals that cannot be ignored. In the context of higher education, students’ educational capital (e.g. having tertiary educated parents or high qualifications to satisfy admissions requirements), a sub-form of cultural capital (Bourdieu, 1984), is likely to be important in shaping their choice of a higher education institution.

To illustrate the relationship of habitus and an individual’s social position in the context of higher education enrolment, students whose families have little knowledge or experience with higher education (i.e. low educational capital) may perceive higher education as something that is not for them and subsequently not enrol in higher education (Archer & Hutchings, 2000). In the frame of institutional choice, students who attend prestigious higher education institutions are likely to have powerful social positions through inherited financial resources and networks from their parents (Bourdieu & Passeron, 1990; Kivinen et al., 2001). Thus, their social position may provide them with a sense of belonging in the more prestigious institutions.

A central component to habitus is temporal consciousness, meaning that an individual’s perceptions, thoughts, and actions are not only based on their social position but they are also based on an individual’s past experiences within that social position (Atkinson, 2013, 2018). Atkinson (2018) states, ‘Recognizing the socially conditioned temporal horizon of consciousness as the heart of habitus does indeed undermine any notion that human beings are free-floating, without a past, totally unconstrained, or indeed mechanically reacting to external constraints or enablements’ (4). In the higher education context, this suggests that an expanding and more competitive educational system may provoke changes in an individual’s response. For example, if an expanding higher education system increases competition to the extent that metropolitan students with low upper secondary school grades can no longer gain admission to the old, prestigious universities proximate to where they grew up, then these students may have to apply to other, less prestigious higher education institutions in order to gain entry to higher education.

Bourdieu (2000) also describes changing contexts such as this, where an individual’s expectations based on their social position does not match its real value, thus influencing a response contrary to what would otherwise be anticipated. From this perspective, the assumption is that socially and meritocratically advantaged groups of young people originating from metropolitan areas would be better equipped to respond to a changing higher education system than less advantaged students from the same areas. For example, these advantaged groups of young people might have the means to bear the costs of a move from the family home. Socially advantaged groups are also more likely to anticipate such changes and respond in due course in order to maintain their preferential access to the educational system (Hardy, 2008). In other words, socially advantaged groups of students are better equipped to identify educational strategies that enable them to maintain their advantaged social position, whether this is making use of university colleges to enter the higher education system or some other approach.

Data

Register data from the Gothenburg Educational Longitudinal Database (GOLD) are used in the analysis. Register data are collected by Swedish government agencies for administrative purposes. In Sweden, individuals are given a unique social security number at birth or upon arrival to Sweden as immigrants, which allow social and demographic data to be collected and tracked across time. The benefit of using this type of data is that an overview of students’ actual enrolment behaviour in higher education is made possible, rather than retrospective views on their behaviour, such as would be collected through survey research. Hence, register data make it possible to research questions that require large amounts of specific information on groups of individuals over time (Mellander, 2017), such as is required by the research questions guiding this study.

The focal population in this paper is individuals born 1973 to 1982 who resided in Sweden at age 16 and who began studying at a higher education institution in Sweden by age 22 (N = 279,932), which is the average age at entry to higher education among students from these birth cohorts. The reason for focusing on young students is because they have not likely established themselves in a geographical location yet. Mature students entering higher education may have selected their higher education institution based on proximity to where they have already established themselves in a home or employment.

Only individuals who studied at least 75% full-time equivalent or 45 higher education credits during their first two years in higher education were considered in this study. The 75% full-time equivalent parameter is based on established requirements for students to obtain a government loan to pursue higher education studies (SFS, 1999, p. 1395). By using this parameter, only those individuals whose main activity is studying are included.

The following two figures, one for students originating from metropolitan areas (N = 79,079) and the other for students originating from non-metropolitan areas (N = 200,853), show the proportion of students born 1973–1982 who entered each of the three types of higher education institutions by age 22. Students’ geographical origin represents their place of residence one year prior to beginning higher education. This study uses the Swedish Association of Local Authorities and Regions classification of metropolitan municipalities to define metropolitan areas, that is, municipalities with over 200,000 inhabitants (SKL, 2011).

Information on students’ first enrolment in higher education was used to identify the type of institution they entered. To connect the birth cohorts to the time of first enrolment, the birth cohorts are discussed within the frame of enrolment cohorts from this point forward. For example, the 1973 birth cohort is now referred to as the 1995 enrolment cohort.

While Figure 3 shows that the proportion of students from non-metropolitan areas entering each of the three types of higher education institutions stayed fairly consistent across the birth cohorts, a noticeable shift in the proportion of students from metropolitan areas attending old universities and university colleges is evidenced in Figure 2. This figure illustrates a gradual increase in the proportion of metropolitan students entering university colleges. Moreover Table 1 indicates that the proportion of metropolitan students attending university colleges increased from 9% among students in the 1995 cohort to 25% among students in the 2001 enrolment cohort before decreasing slightly to 23% among students in the 2004 cohort.

Table 1. Pre-higher education entry attributes of students originating from metropolitan areas, by enrolment cohort

	1995		1998			2001		2004
	%	N	%		N	%	N	%		N
Gender
Women	53	3,616	54		4,132	56	4,543	56		5,041
Men	47	3,237	46		3,469	44	3,578	44		3,882
Parental Education
Non-Tertiary Educated	40	2,706	40		3,019	39	3,157	38		3,420
Tertiary Educated	60	4,147	60		4,582	61	4,964	62		5,503
Upper Secondary Grades
1 Lowest	14	935	11		843	15	1,251	17		1,526
2	20	1,390	19		1,405	22	1,767	23		2,048
3	26	1,791	28		2,104	26	2,108	26		2,353
4 Highest	36	2,441	39		3,004	31	2,550	29		2,588
Unknown/Missing	4	296	3		245	6	445	5		408
HEI Type
Old University	87	5,962	81	6,137		70	5,667		72	6,475
New University	4	288	5	347		5	384		5	434
University College	9	603	14	1,117		25	2,070		23	2,014
N (Total)	100	6,853	100	7,601		100	8,121		100	8,923
	31,498

Note: HEI Type = higher education institution type

Figure 2. Percentage of students from metropolitan areas who entered higher education between 1995 and 2004, by type of higher education institution

Figure 3. Percentage of students from non-metropolitan areas who entered higher education between 1995 and 2004, by type of higher education institution

In other words, these initial findings show that university colleges originally met the policy goals of increasing access to higher education to less urban communities but were diverted from this goal over time by increasing numbers of metropolitan students. This warrants further investigation into the pre-entry attributes of metropolitan students entering university colleges and how these pre-entry attributes trend across the enrolment cohorts. For the sake of parsimony, four enrolment cohorts were selected for an empirical analysis: 1995, 1998, 2001, and 2004. Hence, a selection of 31,498 individuals was included in the analysis (see Table 1).

Variables

The dependent variable in the empirical analysis – Higher Education Institution Type – refers to the first higher education institution students attended. This variable is divided into three categories – old university, new university, and university college. These categories are meant to differentiate the institutions based on degree and program offerings, institutional missions (e.g. research or teaching focused) and perceived prestige within society.

The following are independent variables that were included in the analysis.⁴ The variables represent students’ attributes prior to entering higher education. Cultural capital is operationalized by students’ educational capital (i.e. parental education and upper secondary grades). Table 1 offers a description of these variables for each cohort.

Gender refers to a student’s biological sex as registered at age 16.

Parental education is divided into two categories: tertiary educated and non-tertiary educated. These categories refer to whether or not students had at least one parent who completed two or more years of higher education by the time they reached age 16.

Upper Secondary Grades is organized into quartiles. An additional category for unknown/missing grades was included to acknowledge individuals who may have entered higher education through other qualifications such as through the Swedish Scholastic Assessment Test (SweSAT).

Method

Contingency tables for the pre-entry attributes on enrolment in university colleges are reported in Table 2.⁵ These frequencies are influenced by the distribution on and effect of other variables. To isolate the influence of the different variables in Table 2 to some degree, multinomial logistic regression was used. Multinomial logistic regression is used to examine metropolitan students’ pre-entry attributes and enrolment at different types of higher education institutions in Sweden.

Table 2. Percentage and total number of metropolitan students in each enrolment cohort studying in university colleges by gender, parental education, and upper secondary grades

	1995		1998		2001		2004
	%	(N)	%	(N)	%	(N)	%	(N)
Gender
Women	54	325	59	660	60	1,247	64	1,286
Men	46	278	41	457	40	823	36	728
Parental Education
Non-Tertiary Educated	48	291	51	567	48	991	50	1,010
Tertiary Educated	52	312	49	550	52	1,079	50	1,004
Upper Secondary Grades
1 Lowest	26	156	22	246	25	527	29	586
2	30	182	30	335	33	677	31	621
3	30	177	31	350	28	572	25	505
4 Highest	11	69	14	156	9	194	10	206
Unknown/Missing	3	19	3	30	5	100	5	96
N (Total)	100	603	100	1,117	100	2,070	100	2,014
	5,804

The results are presented as average marginal effects (AMEs). AMEs emphasize the practical significance of results instead of emphasizing the sign and statistical significance of effects (Williams, 2012). Emphasizing patterns in results and their practical significance over statistical significance is particularly important in studies that use register data since no statistical generalization is needed for population data (Gorard, 2012). Therefore, statistical significance is not reported. Furthermore, unlike odds ratios, which are commonly used to report results from logistic regression models in educational research, AMEs are insensitive to differences in unobserved heterogeneity (i.e. unobserved differences), which can vary across compared groups (Mood, 2010). Unobserved heterogeneity relates to a common problem in social science research where it is impossible to include all potential variables in a regression model. Therefore, AMEs can be used to draw comparisons across groups, such as the cohorts of students in this study, with a lower risk of the results being influenced by omitted variables. Predictive margins were also used to examine interaction effects between parental education and upper secondary school grades.

Results

The multinomial logistic regression model estimated the influence of metropolitan students’ pre-higher education entry attributes on enrolment at different types of higher education institutions (i.e. old universities, new universities, and university colleges). Based on the aim of this study, the interpretation of results focuses on enrolment at university colleges. Table 3 shows the average marginal effects on enrolment at university colleges. Results of the multinomial logistic regression model for enrolment at old universities and new universities are shown in Appendices A and B.

Table 3. Average marginal effects of metropolitan students’ pre-higher education entry attributes on enrolment at university colleges

	1995		1998		2001		2004
	Coef.	(S.E.)	Coef.	(S.E.)	Coef.	(S.E.)	Coef.	(S.E.)
Gender (ref = Men)
Women	.012	(.007)	.048	(.008)	.066	(.009)	.081	(.008)
Parental Education (ref = TertiaryEducated)
Non-TertiaryEducated	.019	(.007)	.042	(.008)	.057	(.010)	.070	(.010)
Upper Secondary Grades (ref = 4 Highest)
1 Lowest	.137	(.013)	.242	(.017)	.348	(.015)	.301	(.014)
2	.102	(.010)	.189	(.012)	.306	(.013)	.219	(.011)
3	.070	(.008)	.113	(.009)	.191	(.011)	.132	(.010)
Unknown/Missing	-	-	-	-	.150	(.021)	.151	(.021)

Note: “-“ indicates there are too few cases for the estimate to be reliable.

The results indicate that upper secondary grades have the largest influence on metropolitan students’ enrolment at university colleges. To a lesser degree, gender and parental education also influence enrolment at university colleges. Moreover the influence of each of these pre-entry attributes changed to a large extent between the 1995 and 2004 cohorts. To illustrate, the influence of gender on students’ probability of enrolment at university colleges appears to have grown larger between the 1995 and 2004 cohorts. While women in the 1995 cohort had an increased probability of enrolment at university colleges compared to men (.012), women in the 2004 cohort had an even greater probability of enrolment at university colleges (.081). Put another way, women in 1995 had just a 1.2% increased probability of enrolment at university colleges compared to men while women in 2004 had an 8.1% increased probability compared to their male counterparts.

The results also show that students in the 1995 cohort with non-tertiary educated parents were just 1.9% more likely to attend a university college than their peers with tertiary educated parents. However, the influence of having non-tertiary educated parents on enrolment at university colleges appears to have increased over time. For the 2004 cohort, students with non-tertiary educated parents had a 7% increased probability of attending university colleges compared to those with tertiary educated parents.

Regarding upper secondary school grades, the results show that the probability of a metropolitan student enrolling at a university college increases with low grades. The results also show that the degree of influence that low upper secondary school grades have on the probability of enrolling at a university college increased between the 1995 and 2004 cohorts. For example, while metropolitan students with low upper secondary grades in 1995 had a 13.7% increased probability of enrolling at a university college compared to students with the highest grades, their counterparts in the 2004 cohort had a 30.1% increased probability.

Interaction effects were estimated to examine how the influence of upper secondary school grades on enrolment in university colleges differs between groups of students with tertiary educated and non-tertiary educated parents. The purpose was to understand how metropolitan students from socially advantaged backgrounds, but with low meritocratic advantage, use university colleges to access higher education. The results shown in Table 4 indicate that although these students have increasingly used university colleges to access higher education, their counterparts from less socially advantaged backgrounds use university colleges to a greater extent. Students from the 2004 enrolment cohort with non-tertiary educated parents and low upper secondary school grades had a 44.3% probability of entering university colleges. Conversely their counterparts with tertiary educated parents had a 34.4% probability. The results for the earlier cohorts show a similar pattern, though the degree of difference between the two groups of students steadily increased over time with each cohort. To illustrate, in the 2004 cohort there was a difference of about 10 percentage points in the probability of entering university colleges between groups of students with tertiary educated and non-tertiary educated parents. Within the 1995 cohort, there was only a difference of about 3 percentage points in probability between the two student groups.

Table 4. Predictive margins illustrating interaction effects between parental education and upper secondary grades on enrolment in university colleges

	1995		1998		2001		2004
	Coef.	(S.E.)	Coef.	(S.E.)	Coef.	(S.E.)	Coef.	(S.E.)
Tertiary Educated Parents and Upper Secondary Grades
1 Lowest	.153	(.013)	.264	(.017)	.394	(.015)	.344	(.014)
2	.119	(.010)	.215	(.012)	.352	(.013)	.266	(.011)
3	.090	(.007)	.145	(.008)	.243	(.010)	.186	(.008)
4 Highest	.026	(.003)	.045	(.004)	.067	(.005)	.068	(.005)
Unknown/Missing	-	-	-	-	.203	(.019)	.202	(.019)
Non-Tertiary Educated Parents and Upper Secondary Grades
1 Lowest	.187	(.015)	.338	(.018)	.474	(.016)	.443	(.014)
2	.147	(.011)	.279	(.014)	.430	(.014)	.356	(.013)
3	.112	(.009)	.194	(.011)	.308	(.012)	.258	(.011)
4 Highest	.033	(.004)	.063	(.005)	.092	(.007)	.102	(.007)
Unknown/Missing	-	-	-	-	.262	(.023)	.280	(.024)

“-“ indicates there are too few cases for the estimate to be reliable.

Discussion

Using concepts from Bourdieu, this study attempts to make sense of how changing higher education contexts due to system expansion relates to changes in how students originating from metropolitan areas access higher education. The results of this study show that as higher education participation grew and entry to old universities became more competitive, the enrolment behaviours of some groups of metropolitan students shifted over time. Hence, an expanding higher education system may have created a mismatch between students’ expectations and their possibilities for accessing higher education, prompting some groups of metropolitan students to enrol in university colleges at a higher rate than would otherwise be anticipated.

One example of changing enrolment behaviours is that of an increasing probability of metropolitan women enrolling in university colleges. Women have traditionally dominated academic programs leading to careers in the public service sectors (e.g. teaching and nursing) (UKÄ, 2018c), and since programs leading to these careers have been trademarks of the university colleges since their inception (HSV, 1998), women have generally attended university colleges to a greater extent than men (Börjesson & Broady, 2016). However, the probability for Swedish women from metropolitan areas to enrol in university colleges compared to men grew at a disproportionate rate over time (see Table 2). Because the higher education context changed, women might have started viewing university colleges as more opportunistic educational institutions in which to reproduce their social position. For example, university colleges are generally perceived as less competitive in terms of gaining admission, yet they offer academic programs that more often attract women. Thus, women interested in pursuing careers in the public service sector may view these institutions as an increasingly advantageous place to obtain the credentials needed to gain entry to these careers. These results contrast with prior research in Norway that showed no gender differences in students’ choice of higher education institution (Helland & Heggen, 2017).

The increasing probability of metropolitan students with non-tertiary educated parents enrolling in university colleges offers another example of shifting enrolment behaviours related to changes in the higher education system. Students with non-tertiary educated parents (i.e. with low social advantage) have over time been increasingly more likely to attend university colleges compared to their peers with tertiary educated parents. This may be because parental education provides some indication of the cultural capital (Bourdieu, 1985) students were exposed to during their youth and the knowledge and guidance that may have been available to help them capitalize on higher education (e.g. knowledge of the most opportune higher education institutions to attend). Consequently, when the higher education system expanded, students with non-tertiary educated parents may not have received guidance from their parents on the most opportune institutions to attend in order to capitalize on higher education and saw university colleges as equally suitable.

The strongest example of shifting enrolment behaviours related to a changing higher education system is that of students entering higher education with low upper secondary school grades (i.e. low meritocratic advantage). Findings from this study suggest that as enrolment at the old universities grew more competitive and the share of students able to enrol in these institutions grew smaller (Börjesson & Broady, 2016), metropolitan students with low grades may have been increasingly forced to seek access to the higher education system through university colleges. This suggests that these students had to shift their expectations because the anticipated value of their upper secondary school grades no longer matched what was required for entry to the old universities. In other words, their temporal consciousness (Atkinson, 2013, 2018) may have influenced their decision to seek entry at the university colleges. Naturally, some students, regardless of meritocratic advantage, may never have aspired to attend an old university and instead chose a university college based on a number of other reasons, such as a desire to attend a particular program of study or to be near their social networks (Hemsley-Brown & Oplatka, 2015).

A further example is found when considering the relationship between students’ social and meritocratic advantage (Table 4). Regardless of social advantage, students with low meritocratic advantage have over time increasingly attended university colleges. However, students with both low meritocratic and low social advantage have an even higher probability of using university colleges to access higher education in relation to their peers with high social advantage. This finding suggests that some metropolitan students with low meritocratic advantage and high social advantage may have used their capital (Bourdieu, 1985) to respond to a changing higher education system in a way that they perceived would better maintain their social position. For example, by accessing higher education through one of the new universities.

The growing enrolment of metropolitan students in university colleges has important implications for widening participation initiatives. On one hand the results show that university colleges have been successful in providing access to higher education for student groups that have historically been excluded (e.g. women and students from non-academic backgrounds). On the other hand, the results show that university colleges have increasingly served the needs of students originating from metropolitan areas when one purpose for establishing university colleges was to increase access to individuals residing outside the major metropolitan areas (SFS, 1977, p. 263). Consequently, the results also support prior international research that widening participation initiatives can become less efficient over time (Harrison & Waller, 2017), specifically those initiatives geared towards reducing geographical hindrances in accessing higher education.

Further, the finding that metropolitan student groups with low social and meritocratic advantage have over time increasingly attended university colleges suggests that there are growing divisions between student groups that attend university colleges and those that attend the old universities. Thus, not only are there differences in degree and program offerings, research funding, and perceived prestige between universities and university colleges (Hallonsten & Holmberg, 2013), but there are also differences between the institutions in the type and amount of educational capital that students possess. Hence, despite efforts to reduce inequalities in access to higher education, there are still differences in how students with different forms of advantage access higher education.

Furthermore, earlier research shows that metropolitan students who attend higher education in small urban areas do not stay in these areas after ending their studies but instead return to the metropolitan areas from which they originated (Haley, 2018; HSV, 2011). Consequently, this hinders university colleges that are located in small urban areas in contributing to regional development and innovation, an expectation of the Swedish government (SOU, 2000, p. 87). If increasing numbers of students from metropolitan areas attend university colleges in small urban areas, the university colleges should further invest in programs to connect these students with regional employers with the aim of retaining them in the region following their graduation.

Conclusion

The results of this study contribute to discussions on how the selection to different social positions in society takes place and to what extent different groups of metropolitan students make use of university colleges. This study illustrates how the probability for certain groups of metropolitan students to enrol in university colleges has grown over time and in the context of expanding higher education enrolments and competition. While socially and meritocratically advantaged groups of metropolitan students continue to enter higher education through old, prestigious universities, socially and meritocratically disadvantaged groups have increasingly used university colleges to access higher education. Thus, it could be said that over time university colleges have increasingly attracted geographically advantaged, yet socially and meritocratically disadvantaged groups. This suggests that while university colleges may have initially served a broad range of student groups from surrounding regions that over time they started serving students from less advantaged backgrounds regardless of their region of origin, further widening divisions in the Swedish higher education system.

Disclosure statement

There is no potential conflict of interest.

Additional information

Funding

This work was supported by the Swedish Research Council 2012-5031.

Notes

1. Some courses are offered through distance education but less than one per cent of individuals included in this study completed all their higher education credits through distance education.

2. The causes of the housing bubble in Sweden were similar to the 2008 subprime mortgage crisis and subsequent economic recession in the USA (i.e. economic deregulation and increased lending from banks).

3. The dot-com bubble represented a period of large growth in Internet usage and subsequent high investment in Internet-based companies. Around the turn of the century spending on technology became unstable as companies and people feared what social or economic problems might occur when clocks and calendars changed from 1999 to 2000, but in actuality there was little impact.

4. Students’ choice of educational program also relates to their choice of higher education institution. Field of study could not be examined in this study because contingency tables show that the data became too disaggregated. Doing so may result in an ethical issue with some students becoming identifiable, in addition to too few cases needed for statistical reliability.

5. Contingency tables for the other categories of the outcome variable (i.e. old universities and new universities) are available upon request to the author.

Appendix A.

Average marginal effects of metropolitan students’ pre-higher education entry attributes on enrolment at old universities

Table

	1995		1998		2001		2004
	Coef.	(S.E.)	Coef.	(S.E.)	Coef.	(S.E.)	Coef.	(S.E.)
Gender (ref = Men)
Women	−.008	(.008)	−.048	(.009)	−.064	(.010)	−.088	(.009)
Parental Education (ref = Tertiary Education)
No Tertiary Education	−.021	(.008)	−.051	(.009)	−.060	(.010)	−.070	(.009)
Upper Secondary Grades (ref = 4 Highest)
1 Lowest	−.177	(.014)	−.257	(.017)	−.379	(.015)	−.350	(.014)
2	−.145	(.011)	−.229	(.013)	−.329	(.013)	−.271	(.012)
3	−.118	(.010)	−.158	(.010)	−.240	(.012)	−.176	(.011)
Unknown/Missing	−.040	(.017)	−.061	(.023)	−.176	(.022)	−.167	(.011)

Appendix B.

Average marginal effects of metropolitan students’ pre-higher education entry attributes on enrolment at new universities

Table

	1995		1998		2001		2004
	Coef.	(S.E.)	Coef.	(S.E.)	Coef.	(S.E.)	Coef.	(S.E.)
Gender (ref = Men)
Women	−.004	(.005)	.000	(.005)	−.003	(.005)	.007	(.005)
Parental Education (ref = Tertiary Education)
No Tertiary Education	.002	(.005)	.009	(.005)	.003	(.005)	.000	(.005)
Upper Secondary Grades (ref = 4 Highest)
1 Lowest	-	-	-	-	-	-	-	-
2	-	-	.040	(.007)	-	-	-	-
3	-	-	.045	(.006)	-	-	-	-
Unknown/Missing	-	-	-	-	-	-	-	-

“-“ indicates there are too few cases for the estimate to be reliable. Most estimates for Upper Secondary Grades are not shown because there are too few cases of high upper secondary school grades, which is the reference category.