However, the study, which, since its publication in the Lancet, has been the focus of bitter dispute, has been defended by its authors and other scientists. The new analysis used the same data but a different definition for what constituted “recovery”. Philip Stark, professor of statistics at Berkeley, argued that this simple change converted a “finding into an unfinding” — showing that there was no benefit from exercise or cognitive behavioural therapy.
He conducted the re-analysis because he said the original authors had weakened the criteria used to judge the severity of symptoms. “It was lax to the point where an individual who would have been deemed ill on entry could have been considered healthy at the end, even if they had deteriorated,” he said.