Term-time employment and academic attainment: evidence from a large-scale survey of undergraduates at Northumbria University

Abstract

This study examines the growth in term-time employment and its impact upon academic attainment among full-time undergraduates at Northumbria University. The study utilises data from three large-scale surveys carried out in each Spring Term between 1999 and 2001. The growth in term-time employment coincides with change in the funding arrangements for students in higher education. This study shows that students in employment are drawn disproportionately from less well-off backgrounds. Many students see employment in term time as a means of keeping borrowing down; they also indicate employment as having a negative impact on their studies. The study investigates the impact on the attainment of students by broad subject group. For three of the seven subject groups investigated (including the largest) the adverse impact of employment on attainment was found to be substantial and statistically significant (the difference in performance between employed and non-employed students in the three groups most affected was more than three percentage points). In contemplating new funding arrangements for students in higher education, the Government should be concerned about the efficiency (loss of attainment) as well as the equity/fairness consequences of the arrangements

Notes

A copy of the questionnaire can be found in Barke et al., 2000.

A small number of students included in a row may have deferred entry to study and may face the funding arrangements of an earlier year, also a small number may have special support arrangements.

The rise in term-time employment among students may also be restricting the employment opportunities of the less skilled thus conflicting with Government attempts to improve their labour market prospects. Northumbria students are filling over 1600 full-time-equivalent jobs in the North East; students provide strong competition in the labour market, they impose few long-term costs as employees and are a very attractive workforce to employers at wages little above the national minimum wage.

Derived from student-reported occupations of their parents or guardians.

Some tables comparing what appear to be the same groups of students may have different totals due to a small number of respondents not answering all questions.

This is true whether one adopts a human capital (Becker, 1975) or a screening perspective (Weiss, 1995) on the role of education.

The possibility that hours of employment are met entirely from what would otherwise be leisure time is remote; this would only occur if a student had rather unusual (discontinuous) preferences.

Minimum admission requirements vary less within subject groups than across the university as a whole. Admission standards are high by comparison with most new universities (Sunday Times—Good University Guide). Competition from old universities means that many students fall within a narrow band of A level points.

The Mann–Whitney U test is used to determine whether there is a statistically significant difference in the attainment of two groups of students. Students in the groups to be compared are ranked according to their attainment and the performances of the median student in each group are then compared. This test is a nonparametric version of the t-test; the preferred option of comparing student attainment using t-tests was unavailable, as the student grades were not normally distributed. Since when using this test we do not make assumptions about the shape of the underlying distribution other than that they are independently and identically distributed, the test is not as powerful as the t-test. Each time the test is used a standard one-sided p value is reported. The ‘p = values’ within tables are used for deciding whether to reject the null hypothesis that there are no population differences between the groupings, meaning that any observed differences in attainment are due entirely to chance. The values indicate asymptotic probabilities of committing a type I error or rejecting the null hypothesis when in fact it is true. For most tables, if the p value is greater than 0.05 (i.e. 5% chance of committing a type I error) any observed difference between categories is not taken as large enough for a statistical difference to be assumed. However, for some tables the lack of statistical difference may be due to a small number of observations rather than the groupings having the same characteristics.

This pattern of a lower impact in first year is found also in the other subject groups.

Outliers are not shown, these are defined as observations which are more than 1.5 box lengths above the upper edge of the box or below the lower edge of the box. The average number of outliers is between 2 and 3 per grouping.