Grades across Universities over Time

Abstract

This paper examines the determinants of degree outcomes in a sample of UK universities from 2004 to 2012. We use stochastic frontier methods to account for differences in efficiency across universities and over time. The quality of the student intake and the university's research performance are the main determinants of degree outcomes. There is no evidence of grade inflation at the boundary between upper and lower second-class honours. Decomposing the determinants of degree outcomes, we find that good entry grades yield a higher return in traditional universities than in new universities. Although high-quality universities award more good degrees, we find little evidence that universities of different quality differ in their propensity to inflate grades over time.

Grade inflation in UK higher education

“…In contrast to Johnes and Soo (2013), we find no statistical evidence that student assessment of teaching via the NSS impacts on degree outcome. Unlike the performance specifications employed by Johnes and Soo (2013), the NSS score is entered using a one-period lag to reflect the fact that the results from the survey are generally available after the final degree classifications are awarded.…”

“…This also implies that a university with an average intake of students with three As at A-level will award 4% more 'good' degrees than a university with an intake profile which on average achieves three Bs at A-level, and this result is intuitively plausible. This effect is marginally smaller than the 5% differential reported by Johnes and Soo (2013).…”

“…Unlike the performance specifications employed by Johnes and Soo (2013), the NSS score is entered using a one-period lag to reflect the fact that the results from the survey are generally available after the final degree classifications are awarded. 8 We argue that lecturers alter their behaviour in response to the NSS score the following year to improve 'league table' ranking.…”

“…We employ the 'true' (university) random effects estimator (TRE) advanced by Greene (2005), which develops previous stochastic random effects panel models in that it allows for time-varying university inefficiency to be separated from cross-section university heterogeneity (see also, Belotti 2012; Belotti et al 2012). 7 Following Johnes and Soo (2013), and for the purpose of comparison we transform the continuous variables into natural logarithms.…”

“…In a more recent study, Johnes and Soo (2013) exploit data from the Sunday Times University Guide for the period 2005-2012 and employ an institution-level true random effect stochastic frontier estimator. They find evidence of 'grade inflation' in the academic years 2011 and 2012 in their most austere model which only controls for student pre-entry A-level point scores but incorporates year dummies.…”