The size effect in the RAE results ("How the 'small world' effect skews results," THES , January 18) relates to unit of assessment size, not department size.
In the five largest units, each panel rated 80 or more departments and, on average, awarded 5 or 5* ratings to 30 per cent. In the five smallest, with fewer than ten departments, the panels awarded 5 or 5* to 82 per cent, on average. This effect was evident across all units of assessment. The correlation between size and percentage of 5 or 5* ratings is - 0.48. This is moderately large and statistically significant, so it should not be attributed to chance.
A superficially plausible explanation might attribute the effect to concentration of talent in small branches of research and dilution in large ones, but that does not fit the facts regarding competitiveness of recruitment and shortlist quality. It may be due, in part, to the fact that mutually supportive behaviour tends to thrive in small close-knit communities in which individuals know one another personally.
Andrew M. Colman
Professor of psychology
University of Leicester