Coast and City, It Matters Where You Live: How Geography Shapes Progression to Higher Education in England

Abstract

Progression to higher education in England varies markedly by region, with lower rates of participation outside of London. While some previous studies have explored challenges in accessing higher education in rural and coastal areas, there is a lack of research which considers both individual-level and geographic effects in relation to regional variations in HE progression. In this study, using multivariate regression analysis, we examine whether regional differences in transition to higher education can be explained by the rural/coastal nature of the geographic area in which young people grow up, by area-level deprivation, or by the characteristics of young people living within these regions. The analysis uses the Longitudinal Study of Young People in England, a representative cohort study. These data have been linked to information on the proximity to the coast. Consistent with other work, we find that individual differences and area-level deprivation predict HE aspirations and progression. The newly introduced coastal/rural indicator also predicts HE aspirations and progression, but this is mitigated by the inclusion of individual differences and area-level deprivation. However, we find that unexplained regional differences persist. In particular, the South West of England emerges as a regional cold spot for HE. Consequently, policy makers should consider the role that regional dynamics may have in influencing the choices and constraints faced by young people. The approach may also be applicable to understanding inequalities in progression to HE in other countries.

1. Introduction

Previous research has highlighted unequal progression into HE by region of England (County All Party Parliamentary Group 2018; Department for Education 2017). Particularly noteworthy here is the fact that London boasts a participation rate of 43.1%, with a South East participation rate of 36.6%. Other regions lag behind; the North East has a participation rate of 29.4%, and the South West figure is 32.1% (Department for Education 2017, p. 17). In addition to the South West being characterised by comparatively low progression into HE, social mobility rates are low. Devon ranks in the bottom ten least mobile counties in the UK, and the South West ranks among the least mobile regions (County All Party Parliamentary Group 2018).

The example of the South West of England is pertinent, with young people living in the region having markedly different challenges in accessing HE, when compared to their counterparts in London. The South West is both rural and coastal; it has the longest coastline in England. Of its population of over five million people, about a third live in settlements of fewer than 10,000 people, which is the largest proportion in England (LG Inform 2012). As researchers living and working in the South West of England, we were also motivated to understand better the backgrounds and educational outcomes of young people in the region.

What explains these differences in progression rates? The natural ‘null hypothesis’ would be to expect that there is nothing unusual about the South West. In other words, lower progression rates into HE in the South West would be fully explained once we account for the differences in: (1) the characteristics of individuals living in the South West; and (2) the rural and coastal composition of the region. It is these factors we investigate in this article. There is a wealth of literature on aggregate regional differentials relating to educational progression and outcomes. For example, there is literature on attainment gaps (Education Policy Institute 2017; Teach First 2020) and on the numbers of disadvantaged children entering HE (Department for Education 2020). However, we could not identify previous research that explored both individual-level and geographic compositional effects in relation to regional variations in HE progression. The intended contribution of this article is to address this gap.

We first review the literature relating to the compositional explanation. We then use data from the Longitudinal Study of Young People in England to explore the individual-level explanations; area-level deprivation; and thanks to a measure of distance to nearest coast, provided by Wheeler et al. (2012), we explore the geographic compositional effects. We conclude that known individual and geographic markers only partially explain the low transition rates into HE in the South West of England.

2. Literature Review

The focus of this study is to gain a better understanding of the relationship between regional differences and progression to HE. The literature review is split into three sections. Firstly, what is known about how the rural, urban, and coastal aspects of the areas in which young people live influence educational progression? Secondly, what are the main individual-level factors associated with differential progression to HE? And thirdly, how do locality and geography interact with such individual factors?

Many of England and Wales’ lowest rates of participation in Higher Education (HE) are found in rural, dispersed, and coastal localities. This is evident in the South West of England (for a map of regions of England, see Figure 1). Based on a composite measure of educational and employment-based indicators, the county of Devon has been ranked in the bottom ten counties for social mobility in England, with the rest of the South West ranked in the bottom twenty (County All Party Parliamentary Group 2018). In contrast, London is particularly advantaged in terms of educational and employment opportunities (Donnelly and Gamsu 2018a). The UK Social Mobility Commission (2017) identified that young people living in coastal locations had limited educational opportunities, and that the local authorities in which they lived frequently have insufficient numbers of disadvantaged students to benefit from university outreach programmes (see Wilkinson and Lane 2010). The local employment available in coastal and rural areas also means that there are limited career opportunities for graduates, which in turn decreases the incentive to study for HE qualifications among those young people who wish to remain in their local community (Wilson 2016). Consequently, these localities pose major challenges to Widening Participation (WP) practice, the bulk of which is focused on towns and cities (IntoUniversity 2015).

The most robust empirical work we could find, on HE access by region and on the South West, was by Manley and Johnston (2014) using school-level data from England. Accounting for the school-level characteristics of students, they found that young people in schools in the South West were notably less likely to attend HE: ‘schools located in all of the regions along England’s south coast are less likely to have their students proceed to a higher education qualification than those in London and the country’s midlands and north’ (Manley and Johnston 2014, pp. 271–72). Recent evidence suggests that young people aged 16–18 living in the South West are the most likely (among regions of England) to enter employment at the earliest opportunity (Sim and Elliot Major 2022). Whether this reflects an active choice or a constraint on the educational options available to young people is less clear. The South West is a diverse region with more advantaged occupations being concentrated around Bristol and Bath. Across the region as a whole, however, the region is marked with a greater prevalence of low-paid and insecure work, particularly in more rural and coastal areas (Sim and Elliot Major 2022).

Other than the study by Manley and Johnston, there was surprisingly little research on the dynamics of: (a) the composition of peripheral regions such as the South West; and (b) educational progression. The bulk of regional work is more descriptive. Here, we find a strong narrative of disadvantages driven by a range of factors. For example, Cornwall Council (2018) observed a proliferation of small schools, experiencing particular complexities in relation to finance, curriculum, and facilities; they noted that one Cornish secondary community college had ‘a prescribed admission number of only 600 students yet serve[d] a designated area of over 150 square miles’ (Cornwall Council 2018, p. 9). This means that travel and accessibility are key barriers to educational aspirations.

Spatial barriers and school dynamics also feature strongly in various national and international reviews which emphasise a wide variety of potential reasons for differences in accessing HE. A lack of infrastructure and transport often means that rural students encounter more financial, logistical, and emotional barriers than those from urban locations (Wilson 2016). Other factors identified included: a lack of quality teachers in schools; less access to guidance counsellors; fewer post-secondary engagement activities; high rates of staff sickness; poor student assessment structures; poor data management; poorly trained staff in key positions; and low staff morale (Department for Environment Food and Rural Affairs 2018). Owing to their remoteness and smaller size, rural schools also experience higher costs for the recruitment and retention of staff, and have fewer opportunities for economies of scale, or for efficiency savings from sharing resources and governance arrangements (Department for Environment Food and Rural Affairs 2017). Outside of school, in rural and coastal areas, a number of social and neighbourhood influences not conducive to pursuing HE have been cited. These include: adversity to debt (potentially as a consequence of low local wages); a limited number of role-models; and having high levels of attachment to their local area (Commission for Rural Communities 2012; County All Party Parliamentary Group 2018; Department for Education 2017; Donnelly and Gamsu 2018b; Echazarra and Radinger 2019; Education Policy Institute 2020; Future Leaders Trust 2015; Gorard 2012; House of Lords 2019; IntoUniversity 2015; Social Mobility Commission 2017; Spielhofer et al. 2011; Tieken and San Antonio 2016; Wilkinson and Lane 2010). IntoUniversity (2015, p. 30) also found a negative view of cities, together with a lack of confidence to live in urban areas, which is where we find the greatest density of Higher Education Institutions (HEIs). Despite these perceived barriers, not all coastal areas are disadvantaged or lack educational opportunities; cities and towns such as Brighton and Bournemouth have experienced urban renewal, with good access to local universities and educational facilities. Strategic change in education has also been credited with increasing academic progression in coastal schools in Lincolnshire (County All Party Parliamentary Group 2018). Similarly, Carmarthenshire Council has created a wide-ranging strategy to regenerate its rural communities, as recently as 2019. In summary, a plethora of potential barriers to accessing HE have been suggested, which reflect the different challenges faced by young people growing up in rural and coastal areas.

A second broad field of research has examined the role of individual background characteristics with regard to educational attainment and progression into HE. A significant predictor of student progression to HE is prior attainment. Chowdry et al. (2013) used linked administrative records to show how the progression gap into HE by social class reduced markedly after removing prior attainment. The attainment of good grades in General Certificate of Secondary Education (GCSE) qualifications (at age 15/16) has been positioned as the single most important factor that schools can influence towards student HE progression (Crawford 2014). In other words, social background manifests itself early in a young person’s life, through the attainments, or primary effects (see also Boudon 1974; Domingue et al. 2015). A significant body of social stratification literature has demonstrated how young people from more advantaged social classes, and living in the most advantaged geographic areas, are more likely to progress to HE (Department for Education 2020; Mountford-Zimdars et al. 2015). There are also clear associations between parental education and the educational outcomes of children (Dickson et al. 2013; Ermisch and Pronzato 2010; Gayle et al. 2003; Korupp et al. 2002; Strand 2014; Thaning 2018), and between ethnicity and educational progression (Bhopal 2020).

Research on educational aspirations has highlighted that there is not a socioeconomic gap (Archer et al. 2014; Baker 2017; Green et al. 2017). Higher levels of attainment encourage aspiration, rather than vice versa (Cummings et al. 2012; Gorard 2012). Focus on expectations rather than aspirations might be more meaningful, as they conceptually include an element of the individual’s estimation of whether a particular life outcome is likely (Harrison and Waller 2018). Importantly, expectations among disadvantaged young people tend to be considerably lower, compared to their own aspirations and the expectations of their relatively advantaged peers (Boxer et al. 2011; Khattab 2015), meaning there is a class gap in realising aspirations (Croll and Attwood 2013; St. Clair et al. 2013). By contrast, Spielhofer et al. (2011) found that work ethic and ambition did not differ between those living in urban and rural locations.

The third strand of literature we explored included studies that investigated regional and geographic factors alongside individual-level characteristics. These studies have generally identified the importance of accounting for the differing social characteristics of young people living in different areas. A report published by the DEFRA (2009) included a series of statistical models estimating participation in education at age 16/17, using the Longitudinal Study of Young people in England. Rural-urban differences in participation were explained by differences in the characteristics of young people living in these areas. Similarly, a report by the Regional Schools Commissioner for the South East and South London (2016) attributed differential progression between coastal and non-coastal schools to compositional effects, noting the higher number of white British free-school-meals pupils in coastal areas. In their research into educational participation in Hastings, a disadvantaged coastal town, Church et al. (2010) found that a third of the young people they interviewed saw going to university as impossible to afford; again, this indicates that compositional effects of the local economy potentially explain differential progression to HE. A 2006 study in this journal found that white rural boys were particularly likely to drop-out of school but to engage in informal life-long learning instead (Quinn et al. 2006). More recently, Davies et al. (2021) analysed elite HE participation using 5 years of Higher Education Statistics Authority (HESA) data. Progression rates to ‘elite’ universities were highest in rural areas. However, once the characteristics and backgrounds of young people were controlled for, progression rates were higher for those living in major urban areas. Davies et al. (2021) note that, relative to individual-level characteristics, place was a minor explanatory factor.

Perhaps the most thorough and robust study exploring the geographical variation in participation in HE, in conjunction with the individual-level characteristics of young people, was conducted by Gibbons and Vignoles in 2012. This examined the role of distance from universities, in terms of HE participation and choice of institution. The analysis used linked administrative data sets based on school-leavers in England in 2002. Raw distances to the nearest universities were notably higher for young people living in coastal areas of the South West of England. Despite this, once the backgrounds of young people were controlled for, little or no association was identified between distance to the nearest universities and participation in HE. For those young people with the requisite grades to attend an HE institution, distance was a relevant factor in predicting choice of institution, with (ceteris paribus) a tendency to choose institutions closer to home (Gibbons and Vignoles 2012). Willingness to travel further was associated with higher research quality of the institution.

In summary, several gaps in the existing research have been identified; this article seeks to contribute to the filling of these gaps. There has been surprisingly little recent UK research examining coastal and rural variation in educational outcomes. Some research has used administrative data to model educational outcomes for young people who have attended university, and the type of university they attended. However, there is a notable lack of research examining the educational outcomes of all young people in the years following the end of compulsory schooling. This is perhaps because these young people are not routinely captured by administrative data, which tends to record the experiences of those who remain in education. This is the first gap we want to fill.

Secondly, we have found only one study that combines compositional data of English regions with attainment and progression data. However, even this work by Manley and Johnston (2014) does not include individual-level measures, with the compositional measures recorded at the school level. Hence, there is no recent study we could find that combines individual-level and regional characteristics to further knowledge of educational progression in different regions in England. This is the second gap we want to fill. Our research questions, therefore, are as follows.

• What is the pattern of expectation of, and progression into, HE among young people in different regions in England, given the characteristics of these regions?
• Does regional context offer explanations for differential progression over and above individual-level explanations?

3. Materials and Methods

To answer our research questions, we required a data set that contained three key dimensions—geographic identifiers, individual-level characteristics, and HE progression. We decided to analyse the first cohort of the Longitudinal Study of Young People in England (LSYPE), which is known as Next Steps. These data sets include geographical identifiers and linked National Pupil Database (NPD) information on school qualifications (University College London et al. 2017, 2020, 2021). Next Steps is a nationally representative sample of year 9 pupils (aged 13/14) in schools in England. The young people from the Next Steps cohort have been interviewed eight times between age 13 (sweep 1 was collected in 2004) and age 25 (sweep 8 in 2016). The initial sample size in sweep 1 was of 15,770 young people in maintained schools, independent schools, and pupil referral units in England.

We considered two alternative data sets for analysis. One would have been a more recent cohort of the Longitudinal Study of Young People in England—called Our Future (Department for Education and Kantar Public 2018). However, as of 2022, this study only has 3 sweeps available, up to age 16; thus, it would not have enabled us to undertake the longitudinal tracking of whether a young person ever went into HE, as is possible with the Next Steps study. Similarly, the Millennium Cohort Study only has accessible data available up to age 17 and thus also misses this crucial overall outcome measure.

The Next Steps data set includes rich information on the attitudes of young people and their parents to school and continuing in education. It also includes measures of prior educational attainment, which is a prerequisite for continuing in further education and HE. Due to the sensitive nature of these variables, access to the data was via the UK Data Service Secure Lab. The Stata .do file containing the research code used to prepare and analyse the data is available as a supplementary file.

We used the following measures to operationalise our research questions. Firstly, for our outcome measures, we were interested in university aspiration, applications, and attendance. We used two survey questions to measure these concepts:

• How likely do you think it is that you will ever apply to go to university to do a degree? (sweep 4, age 17, W4HePoss9YP).
• Have you applied for a place on a university course which will start either this year (that is in September/October 2008) or next year (that is in September/October 2009)? (sweep 5, age 18, W5HeapplyYP).

We also used two derived variables indicating whether the young person is in HE, is intending to apply, or has applied—(sweep 6, aged 19, W6HEFlag2) and (sweep 7, aged 20, W7HEFlag2)—and a final summative variable indicating whether the young person has ever been to university (sweep 8, aged 25, W8EVERUNI). In terms of concepts that would explain any differences in our outcomes, we were interested in geographic, attainment, and further individual characteristics as our explanatory measures.

Geography: We used the UK government office region, which was based upon the location of the school the young person attended in sweep 1. We constructed a derived variable indicating whether the young person was living in an urban inland, urban coastal, rural inland, or rural coastal area, based on the address of the young person in sweep 1 (age 14). This was based upon two variables. The first was the 8-category rural urban classification based on output area (ONS 2016). This was collapsed to two categories, rural and urban. The second was an imported variable indicating proximity of the Lower Super Output Area to the nearest coast. This information was used by Wheeler et al. (2012), and it was made available to us. Output areas which were 5km from the coast or less were coded as coastal areas, with areas more than 5km from the coast as inland. The third indicator was the Index of Multiple Deprivation quintile based on where the young person was living in wave 1.

The literature review established the importance of prior attainment and attitudes on progression. These were operationalised as follows.

Prior Attainment: We included whether a young person gained 5 or more GCSEs at grades A*-C (or equivalent GNVQs), as a measure of minimum preparedness for eligibility for HE progression. This information is available in Next Steps through linked NPD data.

School Enjoyment: We included a measure of school enjoyment from sweep 4 (age 17, W4YelevenYP). The question was ‘How enjoyable would you say that your time at school in Year 11 (the 5th year) was?’.

Parental Aspiration: We used a variable from sweep 3 of Next Steps, based on what the main parent wanted the young person to do when they reached school leaving age (sweep 3, age 16, W3parasp2MP). The question was ‘Looking at this card, what would you yourself like (name of young person) to do when/now he/she has finished’. The responses were recoded into one of the two categories FT education or Other.

In terms of the individual-level characteristics of the young person, we used standard measures of sex, ethnicity, and socioeconomic background as follows.

Sex: This was coded by the interviewer as male or female (based on the question ‘Respondent is …’, which was recorded in sweep 4, age 17, W4SexYP). There was no option for non-binary in the data set. We used ‘male’ as the reference category for the multivariate analysis.

Ethnicity: The interviewer asked the young person ‘To which of the groups on this card would you say you belong?’ (sweep 4, age 17, W4ethgrpYP). There was a choice of 16 categories, from which a grouped variable was deposited, with 8 categories (White, Mixed, Indian, Pakistani, Bangladeshi, Black Caribbean, Black African, Other). This classification follows the approach recommended for social surveys by the UK Office for National Statistics (see Connelly et al. 2016).

Socioeconomic background: We used two standard measures for empirical work on the linkage between socioeconomic position and educational outcomes. The first is National Statistics Socioeconomic Class (NS-SEC, see Rose and Pevalin 2003). This is available for the household reference person, which is typically the main parent (wave 4, age 17, w4cnssecfam). We were able to include the highest educational qualification held by either the main or second parent in the analysis (wave 4, age 17, w4hiqualgfam).

While the data set had unique strength in combining relevant individual-level and geographic markers for our study, there were two limitations. Regarding geography, we would have liked to include a measure of distance from home to school. Prior work by DEFRA (2009) was able to include such a measure; however, our data set did not contain the necessary postcode data underpinning these calculations. Also, like many cohort studies, Next Steps has issues of sample attrition and differential non-response. Sample weights, which seek to adjust estimates to be representative of the general population, have been included in these analyses.

Our analytical strategy was as follows. Firstly, to show descriptive statistics of the relationship between our outcome variables (university aspiration and attendance) and regional and individual-level predictors. Secondly, to estimate a series of logistic regression models predicting whether a young person would apply for a university course this year or when they were age 18. The predictors expanded from using only region as an explanation, to including region and area type, and then to including region, area type, area-level deprivation, and individual-level explanations. Thirdly, to examine whether the patterns of association found in our second step held for other outcomes This meant establishing: how likely the young person expected they were to go to university at age 17; how likely they expected they were to make application to a university course (at age 18, model 2); whether they were attending or had applied to a university (at age 19, model 5, and at age 20, model 6); and whether they had ever attended a university (age 25, model 7).

4. Results

Our first analysis replicated the regional variations in university attendance reported by other studies; we found a story of London’s exceptionalism and a South West lag.

Table 1. Higher Educational outcomes by Region of England.

Government Office Region	North East	North West	Yorkshire & The Humber	East Midlands	West Midlands	East of England	London	South East	South West	All
	%	%	%	%	%	%	%	%	%	%
Likelihood of ever applying to university to do a degree (age 17)
Not at all likely	22	22	23	22	16	21	9	20	24	19
Not very likely	21	15	16	16	17	20	10	17	18	16
Fairly likely	18	19	20	20	20	21	19	19	22	20
Very likely	39	44	41	42	47	38	63	44	36	45
%	100	100	100	100	100	100	100	100	100	100
n	458	1439	1068	859	1205	1060	1590	1431	821	9931
Applied for a university course for this year or next (age 18)
No	62	58	62	60	60	62	45	57	67	58
Yes	38	42	38	40	40	38	55	43	33	42
%	100	100	100	100	100	100	100	100	100	100
n	407	1326	986	769	1114	998	1465	1313	752	9130
In or applying for Higher Education (age 19)
Not in HE	47	44	51	47	45	48	29	46	53	44
HE	53	56	49	53	55	52	71	54	47	56
%	100	100	100	100	100	100	100	100	100	100
n	373	1224	910	731	1059	929	1382	1247	709	8564
In or applying for Higher Education (age 20)
Not in HE	53	48	54	51	48	51	32	50	55	48
HE	47	52	46	49	52	49	68	50	45	52
%	100	100	100	100	100	100	100	100	100	100
n	329	1079	812	649	968	845	1227	1131	621	7661
Ever been to University (age 25)
No	58	52	54	51	54	52	39	50	58	51
Yes	42	48	46	49	46	48	61	50	42	49
%	100	100	100	100	100	100	100	100	100	100
n	277	844	608	516	688	675	921	866	488	5883

Source: Next Steps (LSYPE). Table is unweighted.

shows that young people in the South West were the least likely to expect to go to university, to progress to university, or to have attended university. At age 17 (sweep 4), 36% of young people in the South West of England stated that they were very likely to go to university. In contrast, 63% of young people in London said they were very likely to go to university. At age 18 (sweep 5), 33% of young people in the South West had applied for a university course. This was lower than for any other region of England. The percentage of young people who had applied for a university course was highest in London (55%). A similar pattern is noted for measures of participation in HE at ages 19 and 20. By age 25, 42% of young people in the South West had attended university, compared to 61% in London. Across the range of outcome measures, it was notable that participation among young people in London was markedly higher than for other regions of England.

The next question we considered was whether differences in HE outcomes might be associated with area-level characteristics; particularly, whether the area in which the young person lived was urban or rural, and whether the area was coastal or inland. Much of the South West of England is located on a peninsula, which includes many rural and coastal settlements. These area-level characteristics might explain the regional differences observed in HE outcomes.

In

Table 2. Higher Educational outcomes by area type.

Urban Rural Coastal Indicator	Urban Inland	Urban Coastal	Rural Inland	Rural Coastal	All
	%	%	%	%	%
Likelihood of ever applying to university to do a degree (age 17)
Not at all likely	17	27	21	14	19
Not very likely	15	21	17	20	16
Fairly likely	20	19	20	18	20
Very likely	48	33	43	47	45
%	100	100	100	100	100
n	6997	1242	1486	206	9931
Applied for a university course for this year or next (age 18)
No	57	69	57	56	58
Yes	43	31	43	44	42
%	100	100	100	100	100
n	6387	1155	1393	195	9130
In or applying for Higher Education (age 19)
Not in HE	43	56	43	42	44
HE	57	44	57	58	56
%	100	100	100	100	100
n	5990	1083	1304	187	8564
In or applying for Higher Education (age 20)
Not in HE	46	60	47	46	48
HE	54	40	53	54	52
%	100	100	100	100	100
n	5363	948	1183	167	7661
Ever been to University (age 25)
No	50	60	48	47	51
Yes	50	40	52	53	49
%	100	100	100	100	100
n	4089	710	957	127	5883

Source: Next Steps (LSYPE). Table is unweighted.

, variation by area-level characteristic is reported for the same set of HE outcome measures. The figures, for attending university (at age 17), application to a university course (at age 18), attending or having applied to a university (at ages 19 and 20), and having ever attended a university (age 25), are broadly similar for all area types, with the exception of urban coastal areas. At each stage, young people living in urban coastal areas were less likely to progress to university than those living in other area types. By age 25, young people growing up in urban coastal areas were notably less likely to have ever attended university (40%), compared to Urban Inland (50%), Rural Inland (52%), and Rural Coastal (53%).

We then investigated whether the characteristics of individuals living in different geographic areas—the South West versus London, and urban coastal areas versus other areas—explained differences in HE applications and attendance. The individual-level characteristics of different geographical areas are summarised in

Table 3. Individual-level characteristics by area type.

Urban Rural Coastal Indicator	Urban Inland	Urban Coastal	Rural Inland	Rural Coastal	All
	%	%	%	%	%
Achieved 5 or more GCSE/GNVQs at grades A*-C.
0–4 GCSEs A*-C	34	38	26	21	33
5+ GCSEs A*-C	66	62	74	79	67
IMD Quintile
Quintile 1–Most Deprived	34	25	1	*	27
Quintile 2	21	22	9	21	20
Quintile 3	17	23	27	33	19
Quintile 4	14	19	39	35	19
Quintile 5–Least Deprived	14	10	24	*	15
Family’s current NS-SEC class (from household reference person)
Higher managerial and professional	9	7	11	13	9
Lower managerial and professional	24	25	31	31	25
Intermediate	7	11	9	7	8
Small employers and own account	9	9	13	15	10
Lower supervisory and technical	7	9	7	5	7
Semi-routine	12	16	10	9	12
Routine	7	8	7	6	7
Not currently working	25	15	11	13	21
Highest qualification held in family (main or second parent)
Degree or equivalent	19	15	24	24	19
Higher education below degree level	15	16	21	24	16
GCE A Level or equivalent	16	21	20	19	18
GCSE grades A-C or equivalent	23	31	24	25	24
Qualifications at level 1 and below	6	7	5	*	6
Other qualifications	3	2	2	*	2
No qualification	20	8	4	*	15
Main parent aspiration end of school
Not FT Education	12	17	16	13	14
FT Education	88	83	84	87	86
Whether young person found year 11 enjoyable
Not at all enjoyable	4	4	4	*	4
Not very enjoyable	10	13	9	*	10
Quite enjoyable	55	58	60	59	56
Very enjoyable	31	25	27	30	30
Young person’s ethnic group (grouped)
White	60	92	97	95	70
Mixed	6	3	2	*	5
Indian	10	2	*	*	7
Pakistani	8	*	*	*	6
Bangladeshi	6	1	*	*	4
Black Caribbean	4	*	*	*	3
Black African	3	*	*	*	3
Other	3	1	*	*	2
Sex of young person (coded by interviewer)
Male	49	56	49	48	50
Female	51	44	51	52	50
%	100	100	100	100	100
n	6387	1155	1393	195	9130

Source: Next Steps (LSYPE), sweep 5 (age 18), unweighted, n = 9130. Where * is reported, n has been suppressed due to Statistical Disclosure Control.

. Young people growing up in urban coastal areas had lower GCSE attainment, with 62% gaining 5 or more GCSE passes at grades A*-C, compared to 66% in urban inland areas, 74% in rural inland areas, and 79% in rural coastal areas. There were modest differences in the social class distribution between urban coastal and other areas. A lower proportion of parents in urban coastal areas had attained a university degree (15%) than in urban inland areas (19%), rural inland areas (24%), and rural coastal areas (24%). There was slight variation in the levels of enjoyment of school expressed by young people in different areas. In urban coastal areas, a slightly higher proportion of parents (17%) did not want their children to continue in FT education at the end of compulsory schooling, compared to 12% in urban inland areas, 16% in rural inland areas, and 13% in rural coastal areas (13%).

A higher proportion of young people were white, in urban coastal (92%), rural inland (97%), and rural coastal areas (95%), than in urban inland areas (60%). A higher proportion of respondents to the survey in urban coastal areas were male (56%) than in other areas (which was between 48% and 49% in other area types).

To summarise our analysis so far, we found the following patterns of association: (1) young people were less likely to aspire to, progress to, or have attended HE if they grew up in South West England; (2) young people were less likely to aspire to, progress to, or have attended HE if they grew up in urban coastal areas; and (3) young people living in urban coastal areas had some individual-level characteristics that are associated with lower rates of participation in HE.

To understand the relative impact of individual-level versus geographic factors, we estimated, in

Table 4. Logistic regression models. Applied for a university course for this year or next (age 18).

Standard Errors: * p < 0.05, ** p < 0.01, *** p < 0.001	Model 1		Model 2		Model 3		Model 4
	B	SE	B	SE	B	SE	B	SE
Geography
North East	−0.62 ***	0.18	−0.55 **	0.18	−0.45 *	0.18	0.14	0.19
North West	−0.44 **	0.16	−0.41 *	0.17	−0.36 *	0.16	0.15	0.13
Yorkshire and The Humber	−0.58 ***	0.17	−0.61 ***	0.17	−0.60 ***	0.17	−0.06	0.15
East Midlands	−0.57 ***	0.15	−0.65 ***	0.16	−0.79 ***	0.15	−0.11	0.14
West Midlands	−0.54 ***	0.14	−0.59 ***	0.14	−0.59 ***	0.14	−0.31 *	0.13
East of England	−0.62 ***	0.15	−0.61 ***	0.15	−1.00 ***	0.15	−0.28 *	0.14
London (reference)	0.00		0.00		0.00		0.00
South East	−0.38 *	0.15	−0.34 *	0.15	−0.86 ***	0.15	−0.22	0.14
South West	−0.81 ***	0.15	−0.80 ***	0.16	−1.06 ***	0.15	−0.54 ***	0.14
Urban inland (reference)			0.00		0.00		0.00
Urban coastal			−0.40 **	0.13	−0.24 *	0.11	−0.25 *	0.11
Rural inland			0.24 **	0.08	−0.06	0.08	−0.06	0.08
Rural coastal			0.22	0.19	0.16	0.18	−0.11	0.20
Quintile 1–Most Deprived					−1.60 ***	0.10	−0.66 ***	0.12
Quintile 2					−1.12 ***	0.10	−0.46 ***	0.11
Quintile 3					−0.73 ***	0.09	−0.37 ***	0.09
Quintile 4					−0.24 **	0.08	−0.08	0.09
Quintile 5–Least Deprived					0.00		0.00
Attainment and background Factors
0–4 GCSEs A*-C							−2.66 ***	0.10
5+ GCSEs A*-C (reference)							0.00
Higher managerial and professional (reference)							0.00
Lower managerial and professional							−0.37 ***	0.10
Intermediate							−0.21	0.14
Small employers and own account							−0.55 ***	0.13
Lower supervisory and technical							−0.58 ***	0.15
Semi-routine							−0.64 ***	0.13
Routine							−0.78 ***	0.15
Not currently working							−0.67 ***	0.12
Degree or equivalent (reference)							0.00
Higher education below degree level							−0.29 **	0.10
GCE A Level or equivalent							−0.57 ***	0.10
GCSE grades A-C or equivalent							−0.83 ***	0.10
Qualifications at level 1 and below							−1.13 ***	0.18
Other qualifications							−0.88 ***	0.26
No qualification							−0.66 ***	0.14
Main parent aspiration: Not FT Education							−2.01 ***	0.17
Main parent aspiration: FT Education (reference)							0.00
Not at all enjoyable							−1.24 ***	0.25
Not very enjoyable							−0.62 ***	0.12
Quite enjoyable							−0.28 ***	0.06
Very enjoyable (reference)							0.00
White (reference)							0.00
Mixed							0.19	0.15
Indian							1.36 ***	0.16
Pakistani							0.82 ***	0.15
Bangladeshi							0.85 ***	0.17
Black Caribbean							0.49 *	0.23
Black African							1.01 ***	0.19
Other							1.01 ***	0.20
Male							−0.20 ***	0.06
Female (reference)							0.00
Observations	9130				9130		9130
Pseudo R2	0.011		0.015		0.042		0.296
AIC	12,299.4		12,257.8		11,923.4		8827.1
BIC	12,363.5		12,343.2		12,037.4		9126.2
Log likelihood	−6140.71		−6116.89		−5945.72		−4371.57

Source: Next Steps (LSYPE). Estimates are survey weighted and include survey design (PSU and Strata). Model fit statistics are based on models which are unweighted. Constant removed due to Statistical Disclosure Control.

, a series of logistic regression models where the outcome was whether a young person applied for a university course (for the current year or the next) when they were age 18 (i.e., at the end of compulsory schooling in England).

Model 1 includes only region as an explanatory variable. As expected, young people in the South West of England were significantly less likely to apply to university (p < 0.001). However, we also confirm the story of London exceptionalism, there being a higher rate of applications from London than anywhere else. A further explanatory variable is included in model 2, whether the area type in which a young person grew up was urban inland, urban coastal, rural inland, or a rural coastal area. Young people who grew up in an urban coastal area were less likely to apply to university than those from urban inland areas (p < 0.01) irrespective of the region they grew up in. The association with region, noted in model 1, persisted.

Model 3 includes region, area type, and area-level deprivation, in order to control for the observed association between HE outcomes and the urban/rural and coastal/inland area types. Young people in the South West of England were less likely to apply to university (p < 0.001) than those who lived in London, net of the effect of area type or area-level deprivation. Young people who lived in urban coastal areas were also less likely to apply to university (p < 0.05) than those who lived in urban inland areas (the reference category). A clear association with area-level deprivation was evident.

Model 4 includes region, area type, area-level deprivation, and a series of individual-level characteristics including GCSE/GNVQ attainment, parental highest NS-SEC, parental highest qualification, main parent aspiration, the young person’s enjoyment of school, and ethnicity. Compared to models 1 and 2, the overall goodness of fit of model 4 was greatly improved by the inclusion of these explanatory variables (Pseudo R² = 0.30).

For most regions of England, the inclusion of the individual-level characteristics did mediate the association between region and application to HE, the exceptions being the South West and West Midlands. Young people living in the South West of England were less likely than those living in London to apply to university (p < 0.01), net of area type, area-level deprivation, and individual-level characteristics. Those living in urban coastal areas were also less likely to apply to university (p < 0.01).

The regional difference in progression to HE is reported in Figure 2. This is based on the estimates from model 4. This chart shows comparison intervals based on quasi-variance standard errors, which enable comparison between all regions, and not exclusively between particular regions and the reference category (see Gayle and Lambert 2007). It can be observed that application to HE at age 18 is markedly lower in the South West of England than for other regions of England, with the exception of the West Midlands.

To examine whether this pattern held true for the wider range of HE outcomes, we estimated a series of multivariate models. The results of these models are reported in

Table 5. Ordinal logistic regression model (model 5) and logistic regression models.

Standard Errors: * p < 0.05, ** p < 0.01, *** p < 0.001	Model 5		Model 4		Model 6		Model 7		Model 8
	B	SE	B	SE	B	SE	B	SE	B	SE
Geography
North East	−0.01	0.14	0.14	0.19	0.16	0.20	−0.16	0.22	−0.39	0.23
North West	−0.17	0.12	0.15	0.13	0.19	0.15	0.02	0.16	−0.17	0.15
Yorkshire and The Humber	−0.24	0.12	−0.06	0.15	−0.32 *	0.15	−0.41 **	0.15	−0.28	0.16
East Midlands	−0.23	0.12	−0.11	0.14	−0.15	0.17	−0.35 *	0.16	−0.09	0.19
West Midlands	−0.20	0.12	−0.31 *	0.13	−0.34 *	0.15	−0.40 *	0.16	−0.64 ***	0.15
East of England	−0.42 ***	0.12	−0.28 *	0.14	−0.30	0.16	−0.39 *	0.16	−0.28	0.16
London	0.00		0.00		0.00		0.00		0.00
South East	−0.26 *	0.11	−0.22	0.14	−0.32 *	0.15	−0.52 ***	0.15	−0.24	0.16
South West	−0.37 **	0.12	−0.54 ***	0.14	−0.43 **	0.16	−0.52 **	0.17	−0.58 ***	0.17
Urban inland	0.00		0.00		0.00		0.00		0.00
Urban coastal	−0.14	0.08	−0.25 *	0.11	−0.16	0.10	−0.13	0.10	−0.02	0.13
Rural inland	−0.10	0.07	−0.06	0.08	0.06	0.10	−0.03	0.09	0.05	0.10
Rural coastal	0.09	0.15	−0.11	0.20	0.03	0.18	0.10	0.20	0.12	0.18
Quintile 1	−0.37 ***	0.10	−0.66 ***	0.12	−0.63 ***	0.12	−0.77 ***	0.13	−0.64 ***	0.14
Quintile 2	−0.27 **	0.09	−0.46 ***	0.11	−0.41 ***	0.11	−0.49 ***	0.12	−0.34 **	0.12
Quintile 3	−0.21 *	0.08	−0.37 ***	0.09	−0.33 **	0.10	−0.31 **	0.10	−0.19	0.11
Quintile 4	−0.09	0.08	−0.08	0.09	−0.10	0.10	−0.13	0.10	−0.02	0.11
Quintile 5	0.00		0.00		0.00		0.00		0.00
Attainment and background Factors
0–4 GCSEs A*-C	−1.77 ***	0.06	−2.66 ***	0.10	−2.06 ***	0.08	−2.18 ***	0.09	−2.11 ***	0.10
5+ GCSEs A*-C	0.00		0.00		0.00		0.00		0.00
Higher managerial and professional	0.00		0.00		0.00		0.00		0.00
Lower managerial and professional	−0.32 ***	0.09	−0.37 ***	0.10	−0.45 ***	0.12	−0.24 *	0.11	−0.56 ***	0.14
Intermediate	−0.38 ***	0.11	−0.21	0.14	−0.41 **	0.15	−0.27	0.15	−0.60 ***	0.17
Small employers and own account	−0.63 ***	0.11	−0.55 ***	0.13	−0.56 ***	0.14	−0.48 ***	0.14	−0.87 ***	0.16
Lower supervisory and technical	−0.67 ***	0.12	−0.58 ***	0.15	−0.56 **	0.17	−0.49 **	0.16	−0.79 ***	0.17
Semi-routine	−0.46 ***	0.11	−0.64 ***	0.13	−0.60 ***	0.14	−0.48 ***	0.14	−0.65 ***	0.17
Routine	−0.62 ***	0.12	−0.78 ***	0.15	−0.81 ***	0.17	−0.75 ***	0.17	−0.83 ***	0.18
Not currently working	−0.51 ***	0.11	−0.67 ***	0.12	−0.43 **	0.14	−0.34 *	0.14	−0.57 ***	0.16
Degree or equivalent	0.00		0.00		0.00		0.00		0.00
Higher education below degree level	−0.59 ***	0.08	−0.29 **	0.10	−0.65 ***	0.11	−0.50 ***	0.11	−0.45 ***	0.12
GCE A Level or equivalent	−0.84 ***	0.08	−0.57 ***	0.10	−1.05 ***	0.10	−0.91 ***	0.10	−0.68 ***	0.13
GCSE grades A-C or equivalent	−1.06 ***	0.08	−0.83 ***	0.10	−1.13 ***	0.11	−1.04 ***	0.10	−0.91 ***	0.12
Qualifications at level 1 and below	−1.19 ***	0.11	−1.13 ***	0.18	−1.40 ***	0.17	−1.40 ***	0.18	−1.19 ***	0.19
Other qualifications	−0.91 ***	0.19	−0.88 ***	0.26	−0.88 ***	0.26	−1.00 ***	0.23	−0.89 **	0.28
Main parent aspiration: Not FT Education	−1.00 ***	0.11	−0.66 ***	0.14	−1.09 ***	0.15	−1.13 ***	0.15	−0.94 ***	0.17
Main parent aspiration: FT Education	−1.38 ***	0.07	−2.01 ***	0.17	−1.50 ***	0.12	−1.77 ***	0.13	−1.33 ***	0.17
FT Education	0.00		0.00		0.00		0.00		0.00
Not at all enjoyable	−1.09 ***	0.14	−1.24 ***	0.25	−0.80 ***	0.21	−1.18 ***	0.22	−0.73 **	0.28
Not very enjoyable	−0.76 ***	0.08	−0.62 ***	0.12	−0.43 ***	0.11	−0.49 ***	0.12	−0.34 *	0.14
Quite enjoyable	−0.29 ***	0.05	−0.28 ***	0.06	−0.17 **	0.06	−0.13	0.07	−0.18 *	0.07
Very enjoyable	0.00		0.00		0.00		0.00		0.00
White	0.00		0.00		0.00		0.00		0.00
Mixed	0.56 ***	0.12	0.19	0.15	0.53 **	0.18	0.05	0.17	0.45	0.24
Indian	1.58 ***	0.13	1.36 ***	0.16	1.94 ***	0.15	2.10 ***	0.17	1.45 ***	0.22
Pakistani	1.43 ***	0.12	0.82 ***	0.15	1.23 ***	0.14	1.46 ***	0.17	1.03 ***	0.20
Bangladeshi	1.08 ***	0.14	0.85 ***	0.17	0.96 ***	0.20	1.30 ***	0.20	0.55 *	0.22
Black Caribbean	1.10 ***	0.18	0.49 *	0.23	0.90 ***	0.24	0.46	0.27	0.68 *	0.31
Black African	2.14 ***	0.19	1.01 ***	0.19	1.86 ***	0.27	1.85 ***	0.27	1.76 ***	0.33
Other	1.29 ***	0.17	1.01 ***	0.20	1.13 ***	0.21	1.20 ***	0.23	0.91 ***	0.26
Male	−0.32 ***	0.05	−0.20 ***	0.06	−0.28 ***	0.06	−0.18 **	0.06	−0.17 *	0.07
Female	0.00		0.00		0.00		0.00		0.00
Observations	9931		9130		8564		7661		5883
Pseudo R2	0.210		0.296		0.289		0.285		0.228
AIC	20274.3		8827.1		8441.8		7661.4		6377.8
BIC	20591.3		9126.2		8738.1		7953.0		6658.4
Log likelihood	−10093.15		−4371.57		−4178.90		−3788.68		−3146.90

Source: Next Steps (LSYPE). Estimates are survey weighted and include survey design (PSU and Strata). Model fit statistics are based on models which are unweighted. Constant removed due to Statistical Disclosure Control.

. Model 5 is an ordinal logistic regression where the outcome was how likely the young person expected, at age 17, they were to go to university. The remaining models were all logistic regression models, with the outcomes being: application to a university course (at age 18, model 4); attending or having applied to a university (at age 19, model 6, and at age 20, model 7); and having ever attended a university (age 25, model 8). The same set of explanatory variables was included in all five models reported, including the region, area type, and individual-level characteristics.

Young people from the South West of England were consistently less likely to progress to HE. Those from the South West were less likely than those from London to expect, at age 17, to: go to university (model 5, p < 0.01); be at university or to have applied at age 18 (model 4, p < 0.01); be attending university or to have applied to university at ages 19 and 20 (model 6, p < 0.01; model 7, p < 0.01); or to have attended university by age 25 (model 8, p < 0.01). The estimated effect of living in an urban coastal area was not statistically significant across the range of outcomes, perhaps suggesting that area-level deprivation or the composition of these areas might explain the raw differences observed from the descriptive statistics reported in Table 1, Table 2 and Table 3.

We undertook several further analyses to check the robustness and sensitivity of these findings. Owing to space constraints, these are included in the appendices. We estimated the same set of models using a different operationalisation of region (using whether the young person grew up in an urban or rural area, and the distance from the coast in km, measured as a series of categories). We also used sweep 5 outcomes based on 9130 cases, which allowed us to fit a multi-level model which accounted for school-level variation. These changes, in operationalising region or school, did not change the association between living in the South West of England and HE outcomes.

5. Discussion

The rates of HE transitions, and the levels of social mobility, in the South West of England, are among the lowest in England. We wanted to understand whether this is simply an effect of the composition of people living there, together with the prevalence of coastal and rural areas. In short, the answer is that neither the rural and coastal composition, nor area-level deprivation, nor who lives in the South West explain why this region has lower HE aspirations and progression than other regions in England.

The present work is the first study to go beyond national headlines of regional variations of low progression to university. Previous research has used administrative data, particularly school-level data from the Department for Education, and some studies have used individual-level data from the Pupil Level Annual Census, NPD, and HESA. While this has enabled researchers to understand the characteristics of those attending HE, far less is known about individuals who do not continue. We used detailed individual-level factors, including personal background, attainment, and aspirations, in modelling the chances of going to university. No previous research, for example, had the distance-to-coast measure involved in the investigation of progression.

In our descriptive analysis, we found that locations in the South West, and in urban coastal areas, were associated with lower likelihoods of aspiring to, progressing to, and having attended HE. We also found that young people living in urban coastal areas have some individual-level characteristics that are associated with lower rates of participation in HE. Other findings from the analyses were broadly consistent with previous research into individual differences in HE progression by social class, parental education, and ethnicity; HE progression among those living in urban areas (Davies et al. 2021); and lower rates of progression in the South West of England (Manley and Johnston 2014). There were no obvious areas of inconsistency with previous studies; rather, our research provided a fuller explanation of the individual-level and geographical variation in HE progression observed.

It was established that among the study respondents, aspirations (how likely they said they were to go to university) did not differ greatly from what the young people actually did (whether they actually applied for or went to university). In this study, aspirations were an accurate predictor of behaviour and action, which is a contribution to the body of literature on aspirations and expectations (Harrison 2018; Harrison and Waller 2018).

In our multivariate regression analyses, we then found that—even when we controlled for the background characteristics of young people, area-level deprivation, and the coastal/rural dynamics—young people from the South West of England are less likely to go to university. What is perhaps surprising is that the South West and West Midlands are the only regions in England where neither the compositional effects at the individual level, area-level deprivation, nor the rural-coastal indicators account for the lower levels of progression. In other words, for most areas in England, knowing the local deprivation, rural-urban and coastal-inland markers, as well as the individual-level composition is sufficient to explain HE aspirations and progressions; however, this standard model leaves the South West effect unexplained.

6. Limitations and Further Research

This study has investigated the progression to Higher Education accounting for the backgrounds, area-level deprivation, and the rural/coastal composition of the areas in which young people grow up. It should be noted that the patterns of association we observe should not be interpreted causally. Further research is required to understand the reasons for these associations. We were not able in this study to investigate accessibility to particular HE providers at the individual-level as the data do not permit this. Future studies might also consider alternative routes to Higher Education or the supply of local job opportunities available to young people in their local area. This is a topic that we are currently investigating in a separate study. It would also be of interest to understand, across regions of England, whether young people are able to commute or whether they would have to geographically relocate to access HE provision. Future research could also attempt to undertake more nuanced analyses of exploring any potential linkage between geography and subject choice in higher education, which in turn might be related to regional opportunity structures.

7. Conclusions

The policy implications of this study relate to the observation that the South West is a special case for those wishing to increase educational opportunities and progression. It might be that outreach work trialled in (for example) ethnically diverse urban areas simply cannot be translated into the South West context. It might be that the peripheral location of the South West, and the particular labour market opportunities, as well as the pattern of HE availability require a bespoke approach to supporting continuation. Perhaps dislocation is a more significant factor for students from smaller rural and coastal settings, when looking into residential non-local university experiences. Future studies might, therefore, seek to investigate the educational challenges faced by young people in rural and coastal areas.

It is clear from reviewing the literature that there are different barriers to accessing higher education for young people growing up in rural or coastal areas compared to the urban disadvantaged. Whilst our study controls for area-type and area-level deprivation, measures of area-level deprivation have been criticised as poorly representing the nature of disadvantage in rural areas (Burke and Jones 2019; Commins 2004). Transport and lacking internet connectivity are key barriers, with remote working having clear potential for education and work opportunities.

The wider research question that this type of research does not answer: Are the young people making the best decisions about their future? There has, rightly, been considerable push-back against ‘deficit’ discourses regarding the causes of traditionally underrepresented groups in HE. In highlighting the comparatively low HE progressions in the South West, we must take care not to fall into a similar trap. Our study has not included a measure of life satisfaction; and life satisfaction is achievable in other ways than through progression to HE (MacBrayne 1987). Rural and coastal lifestyles rooted within communities can create strong attachments and feelings of belonging (Nelson 2019). For some young people, it is possible that not leaving their communities in pursuit of education and social mobility might be the best life choice. This sentiment ties in with the shifting policy discourses towards the UK government’s ‘levelling up’ agenda (HM Government 2022). This agenda includes a range of objectives, among which a theme is to seek to ‘spread opportunities’ for people living in different regions of the UK without requiring that people become more geographically mobile. The present study is silent on the normative question of whether it is necessarily a good thing to increase progression into HE in the South West; but this should form part of the considerations of further, potentially more qualitative work.

To conclude, as researchers located in the South West of England, our findings develop our theoretical understanding of the drivers of differential progression. The longer-term ambition for this work is to understand the context of choice in the South West, and to support continuation in education where this is what young people desire; but also to understand and support alternative pathways, to enable choices leading to the greatest possible life satisfaction for these young people.