open menu

An exploratory analysis of factors associated with decreased representation in higher education

Research overview

Understanding the factors that influence representation in higher education is of considerable public interest against a policy background in which widening participation within under-represented groups is a key aim. This study was designed to identify the most important among thirteen diverse factors that potentially influence representation, specifically considering the transition from secondary education.


Methods

The capabilities of the Administrative Data Research Centre – Northern Ireland were piloted with the creation and secure analysis of a large-scale linkage between two administrative datasets, the 2011 Northern Ireland Census and student enrolment data held by the Higher Education Statistics Agency (HESA). The Census dataset provided rich background information and the HESA enrolment dataset indicated which people made the transition to higher education at United Kingdom (UK) higher education institutions (HEIs). When linked at the individual level, a unique research dataset was created that was used to identify factors at the individual, household and area levels that were associated with variation in enrolment in Higher Education.

The study cohort consisted of all young people with at least five GCSEs (A*-C) who had turned 18 during the 2010/2011 academic year (i.e. were potentially in the final year of secondary education) and who were enumerated in the 2011 Northern Ireland Census (n = 14,895). The characteristics of those enrolling/not enrolling for undergraduate degrees at UK HEIs in the 2011/2012 academic year were compared and statistical models (logistic regression) used to estimate the strength of the association between each factor and the probability of enrolment.

The association with enrolment was first estimated for each factor separately using a series of unadjusted models to represent basic patterns observed in the data. However as some factors are correlated, this method may mislead if strong apparent associations between particular factors and enrolment are actually driven by other factors. Therefore, models were extended to determine the relative influence of each factor on enrolment whilst statistically adjusting for the influence of other factors (multiple regression). Finally, fitted regression models were used to test a particular hypothesis of interest, that Protestant males from deprived areas were disproportionately less likely to enrol in higher education than Catholic males from similar areas.


Key findings

Just over half of the cohort enrolled in higher education (8476 people). At the individual level there was variation in enrolment by religion and sex. Protestants were less likely to enrol than Catholics and males were less likely than females. Long hours of work outside school and poor health were both associated with decreased enrolment. At the household level there were strong gradients favouring those from households with high socio-economic status, and a small advantage for those from two-parent households. Area of residence was also associated with variation in enrolment, with those from areas deprived in terms of education far less likely to enrol than those from less deprived areas.

In terms of the relative strength of associations, household social class and housing tenure/value were most strongly associated with enrolment (those in rented accommodation were almost 70% less likely to enrol than those in the most expensive houses) followed by area level deprivation and long working hours. There were moderately strong associations with household car access, individual health status, religion and sex (males were 27% less likely to enrol than females). The associations with household structure and rurality (urban vs. rural) were much weaker (10% – 20%) and no association was found between enrolment and ethnicity.

Differences between adjusted and unadjusted estimates were large enough to influence conclusions regarding several factors. Associations with enrolment were substantially weaker after adjustment: at the area level for income deprivation and rurality; at the household level for car access and household structure and at the individual level for working hours. For these factors simple unadjusted analyses (e.g. cross-tabulations) are potentially misleading and adjusted models should be used instead.

Protestant males were less likely to enrol than Catholic males with similar individual and household characteristics. This differential was constant across the spectrum of area deprivation, rather than being confined to the most deprived areas.

An important caveat is that these findings concern the outcome of several decisions leading from school to higher education rather than the transition from A-level to higher education alone. Information on attainment at A-level was not included in the analysis because under this study design inclusion would have prevented valid investigation of the influence of household factors on enrolment. Future studies with cohorts based on A-level rather than GCSE results might therefore reveal different associations between enrolment and explanatory factors to those reported here.


Conclusions

The use of adjusted statistical models in this analysis has revealed for the first time the relative strength of associations between factors at the individual, household and area levels and enrolment in higher education for Northern Ireland domiciled young people. In general, household factors were more strongly associated with enrolment than area deprivation, which was in turn more strongly associated than a range of individual level factors. These findings indicate that household background has a profound influence on the decisions surrounding transition to higher education and that policies to increase enrolment might be best targeted towards households with fewer resources. Our approach also revealed that the pattern of lower enrolment among Protestant compared with Catholic males extended beyond the most deprived areas where it has previously been observed. However, given the absence of some key factors in this model on the transition from school to further and higher education it’s also quite possible that the findings of this exploratory model may differ substantially from that of a more complete model.  So care is needed at this stage in interpreting the findings of this initial exploratory study and further research is needed to investigate this in more detail.


Date Approved: June 2015


Project team

This research was conducted at the Administrative Data Research Centre – Northern Ireland

Michael MacNeill (Department of Employment and Learning)

David Wright (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Stephen Donnelly (Department of Employment and Learning)

David Patton (Department of Employment and Learning)

Dermot O'Reilly (Administrative Data Research Centre Northern Ireland, Queen's University Belfast)

Laura Smyth (Department of Employment and Learning)


Page last updated: 19/06/2017