Number's up (or down)

While berating scholars for their abuse of statistics, your article "Not so good with numbers" (6 May) was illustrated with a photograph of a researcher jotting down his results and thinking: "I'll just round that up ... or maybe down."

However, the topic of rounding numbers is one that Times Higher Education failed to handle properly in the article titled "Sector offers UK real value for money, v-cs claim" (22 April). It stated that "among students starting full-time first degrees in the UK in 2007-08, the proportion who had dropped out a year later was 8.6 per cent, down from 9 per cent in 2006-07".

But the 9 per cent statistic is clearly a value that has been rounded to a whole number, otherwise it would properly be written as "9.0 per cent". As such, it represents an unspecified value somewhere between 8.5 per cent and 9.5 per cent. Thus, it is possible, on the evidence presented, that the dropout rate had increased.

The error was compounded in a further statement that "undergraduate entrants from state schools rose to 89 per cent in 2008-09, up from 88.5 per cent the previous year". In this case there is no evidence for an increase since the "raw" statistic of 88.5 per cent is being compared with a rounded value of 89 per cent, and the number 88.5 conventionally rounds up to 89.

Therefore, the article's standfirst, "Figures show state school entrants on the rise and dropout rate falling", was potentially invalid on both counts.

Ben Atkinson, Hertfordshire.