51¹ú²úÊÓƵ

In claiming that citation counting is unlikely to deliver, Ole Petersen (Opinion, December 15) overlooks a point missed by many.

Yes, citation counting can be an unreliable way to judge an individual, for a variety of reasons, some of which he rehearses. But as a measure of an aggregation of people of different ages and research styles, such as the members of a university department, it works well because its failings tend to affect all aggregations similarly. This is not just an opinion; there is good empirical evidence for it.

The mean citation counts of whole departments correlate extremely highly with research assessment exercise ratings.

Andy Smith
Royal Holloway, 51¹ú²úÊÓƵ of London