The Quality of Higher Education and Employability of Graduates

Abstract

In this paper, employability is regarded as an aspect of quality of higher education, or more precisely, the benefit and usefulness of the study programme for career and work tasks. The analyses are based on a comparative survey among graduates in 13 countries, five to six years after graduation. The information about employability, or the usefulness of the study programme related to work is based on the graduates’ own perceptions. The extent to which employability is related to characteristics of the study programme is investigated, in addition to country differences and to what extent the differences are affected by the graduates’ labour market experiences. The analyses show that study programme characteristics have great impact on the value of the programme in the world of work. These characteristics also have an impact on the initial country differences with regard to employability. The quality indicators have minor influence on the chance of obtaining a job but significant effect on doing the job.

Keywords:

• country differences
• employability
• quality
• study programme

Notes

[1] Some readers might be interested in the formula used. The estimates of the binary logistic regressions are made according to the formula P= e^z/ (1+ e^z) where P is the probability of whether to a very high extent the study programme was a good basis for starting work and Z = the intercept plus the effects of the control variables (z = B₀ + B₁X₁ + B₂X₂ …), including the effects of study programme characteristics.

[2] The descriptive results in Tables 1, 2 and 3 are weighted to 2000 cases for each country so as to prevent certain countries from dominating the mean results across all countries. In the REFLEX project the ‘weighting coefficient used also corrects for over‐ or under‐representation of certain levels or fields of higher education compared to population data. The multivariate analyses use unweighted data whereby a random sample of no more than 2000 cases per country has been drawn’ (Allen & van der Velden, 2007, p. 3).

[3] Another possibility could be to include controls for unemployment level in the different countries. Then the dummy variables for country would have to be left out, because it is not possible to keep both in the regression model. Then interesting information will be lost. In addition, the effects of the general unemployment level(s) vary unsystematically and do not give significant results. On the other hand, the individual’s labour market experience makes a significant contribution, as seen in Table 4.