Correlation limitations

April 18, 2013

The fact that at the institutional level, total quality-related (QR) cash from the Higher Education Funding Council for England and total research council funding are highly correlated is neither new nor surprising. It also misses the point. (“Alternative to REF delivers similar results on QR funding”, News, 11 April.)

Our analysis of research assessment exercise results, cited by the Centre for Business Research report covered by the article, shows that, notwithstanding the high correlation, the differences between QR-based and research council-based allocations can be very large.

More importantly, research quality varies substantially between departments within institutions. The research excellence framework, like the RAE before it, recognises this by judging research quality at the subject rather than the institutional level.

The aggregation of QR and research council income at the institutional level also exaggerates the correlation between the two. At the subject level it is typically weak.

Peter Diggle and Amanda Chetwynd
Lancaster University


You report a strong correlation between the amounts institutions earn from QR and research council funding.

This suggests that the costly and labour-intensive research councils could be scrapped and their funding allocated based on each institution’s QR income. The government would save a large amount of money by doing so. David Willetts, please take note.

Ben Atkinson

Have your say

Log in or register to post comments