ÖйúAƬ

You report the finding from the Office for Fair Access that neither the size nor the availability of a bursary had a discernible effect on whether a?student from a?poor background would finish a course or not (¡°Bursaries fail to help poor students stay the course¡±, News, 6?March).

However, Offa¡¯s analysis is based on completion rates at institutional level. Epidemiologists have long known that the absence of an effect at group level (university here) need not imply absence of that effect at the individual level (the student) ¨C the so-called ecological fallacy. The data available to Offa did not permit an individual-level analysis.

Using we have shown that, after accounting for other factors that affect a student¡¯s likelihood to complete their studies (A?level tariff and socio-economic background), not only does the award of a bursary increase the likelihood of completion, but it does so to an extent that almost eliminates the effect of socio-economic background.

Bursaries are awarded to students, not to universities. A full understanding of their effects requires the analysis of data on individual students.

Amanda Chetwynd
Peter J. Diggle
Lancaster University