Distance and academic performance in higher education

ABSTRACT

This paper examines whether one of the determinants of academic performance in higher education is the geographical distance separating the place of study and the place of family residence. Twelve years of data on students from a public university are utilized to estimate a model of relative academic performance, with an instrumental variables two-stage least squares estimator to account for possible endogeneity bias. The results indicate that distance is a negative determinant of academic performance and provide a new factor to take into account in reflexions over the impact of the spatial organization of higher education institutions.

KEYWORDS:

• higher education
• academic performance
• geographical distance

JEL:

• D91
• I24
• I25
• R12
• R23

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the valuable comments made by two anonymous referees.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

ORCID

Carlos Vieira http://orcid.org/0000-0001-8112-2736

Isabel Vieira http://orcid.org/0000-0002-2983-7555

Luis Raposo http://orcid.org/0000-0002-0638-029X

SUPPLEMENTAL DATA

Supplemental data for this article can be accessed at https://doi.org/10.1080/17421772.2017.1369146.

Notes

1 Any bias from omitting students who fail to graduate would not operate in the opposite direction from the study findings, since Luo et al. (2006) show that distance negatively impacts persistence.

2 The Portuguese public higher education system is divided into two subsystems (14 universities and 20 polytechnics). Selection of applicants is centrally organized and most students (90% of entrants) apply through a ‘National Contest’. There are also private universities and polytechnics, mainly attended by those unable to obtain the grades needed for public institutions (with some exceptions in Lisbon and Porto).

3 On average, in the last decade, the University of Évora received per year 4617 applications from the national contest candidates, of whom 888 were admitted and enrolled.

4 These include programmes where access requires local auditions (music and drama); students whose parents are public servants abroad; foreign diplomats in Portugal; army officers; students from former Portuguese colonies; international students; athletes; mature students; and holders of higher education or post-secondary technical diplomas.

5 The Bologna process aimed at creating a European Higher Education Area by 2010. Taking part was voluntary and Portuguese universities began the reforms in 2006. The main changes were shorter first degrees (from four or five to a three-year period, in most cases), more student-focused learning approaches, more elective courses, possibility of taking some courses abroad, and a stronger emphasis on employability.

6 These and further robustness tests can be found in the supplemental data online.

7 Future complementary research, using, for example, survey data, may help control for other determinants, albeit at the cost of other distortionary elements, since it relies on the willingness, sincerity and memory of respondents.

8 We have also estimated the empirical model including the students’ order of preferences as an explanatory factor. The results are not shown (but are available in the supplemental data online) since this variable was not statistically significant and the number of observations is smaller as the information was missing from many student records. This alternative model did not change the sign, magnitude or significance of the estimates of the other determinants. It is, however, important to be aware of this variable in future research since a strong relationship with distance may signal potential selection bias.

9 An anonymous referee aptly suggested this gender difference could have emerged simply because men display a greater deviation in academic performance (Table 1). In this case, however, normalizing the dependent variable in order to avoid this problem does not qualitatively alter the overall results nor the relative differences between male and female students (see the supplemental data online).

10 The supplemental data online presents the results of estimating the model with other interaction terms (none statistically significant) between the main variable of interest, distance, and all other variables in the base model.

11 This result is confirmed when the model is re-estimated only for students under the Bologna reform (see Table 9A in the supplemental data online).

12 We thank an anonymous referee for the suggestions on these possible causes for the non-linearity.

Additional information

Funding

The authors gratefully acknowledge the financial support from the Fundação para a Ciência e a Tecnologia and FEDER/COMPETE [grant number UID/ECO/04007/2013].