What counts and what doesn't in rankings? 2

You acknowledged last year that the Times Higher Education-QS World University Rankings were not well suited to recognise the particular characteristics of institutions such as the London School of Economics and Political Science and the School of Oriental and African Studies, which research and teach in the social sciences.

Publication and citation practices vary dramatically across disciplines. Published outputs in biomedicine, for example, are short, numerous, multi-authored and cited more quickly and more often - by a huge margin - compared with outputs in the vast majority of social sciences.

This means that the "citations per staff" measure in the main rankings table systematically disadvantages institutions with more researchers in the less highly cited disciplines.

In addition, this citation measure is based on a count of papers appearing in the Scopus database that ignores outputs in books. In some of the social science disciplines, publications in books are at least as important as journal articles, and we estimate that your "citations per staff" measure has excluded a third or more of the outputs of almost half of our academics here at the LSE.

Most importantly, our work suggests that the published data are incorrect. We sought to replicate the citation data in the 2007 rankings. According to our calculations, the "citations per paper" was out by a factor of two.

QS, the firm that compiles the tables, has not been able to explain the discrepancy or to produce a justification of its own figures. This year's result appears to be another serious underestimate.

You will be the first to recognise the impact that your tables can have on university reputations. Accuracy and transparency should, therefore, be the least we deserve.

Howard Davies, Director, London School of Economics and Political Science.