The conclusion that the proportion of good degrees has risen slightly between 1994-95 and 1998-99 (Analysis, 色盒直播S , June 22) is consistent with analyses I have undertaken. But analyses of whole-institutional data have limited value.
The trend in the overall proportion of good degrees for an individual institution masks considerable variation between subject areas. A more detailed analysis shows that, for example, there is over the same period a statistically robust upward trend in education and engineering, whereas biological sciences and computer science are essentially flat. Although some plausible causes for the institution-level differences are pointed to in the article, the complexity underlying the trend data requires a deeper scrutiny if it is properly to be understood.
Mantz Yorke
Professor of higher education
Liverpool John Moores University