In sci.med.nutrition Eric S <erics_news_addyl at yahoo.dawt.cawm> wrote:
> Well my thought was that (assuming they got their data by asking
> subjects about their daily habits) caffeine was not actually the
> variable being measured, but merely the one they chose to report. It
> seems likely coffee (and possibly also tea, cola and other caffeine
> containing consumables) intake was estimated by the subjects, and
> caffeine intake further estimated from that (adding additional
> uncertainty due to quite significant variation in caffeine content in
> coffee by preparation method, roast, beans used, etc). So unless I'm
> wrong about how they gathered their data, their data about
> caffeine-intake is one estimate further away from actual measurement
> than their data about about coffee / tea / etc intake.
And they seemed to have been able to measure it to a resolution of 100
micrograms/day.
And they seem to be relying on recall for a comparison between two groups,
one of which can be expected to have much poorer recall than the other.
Hmmm.
