Misplaced confidence

Jane Hutton of Warwick University criticises the recent funding council report on patterns of participation ("Hefce access study 'flawed'", February 11).

I have no wish to defend the report in general, but her letter in the same issue reveals that one of her objections is the absence of confidence intervals. These are a guide to the variability of results due to random sampling.

However, the funding council's report was based on "population" data for each unit and level of analysis, with no sampling. The results will contain an error component, stemming from missing data and perhaps double-counting of individuals, but confidence intervals will not resolve this.

Stephen Gorard
York University