I am the stats advisor to a specialist clincial journal, and yesterday
reviewed a meta analysis written by three clinicians from a well-known
USA medical research centre. This was the second time it had been round,
and I was suspicious of the results, and looked again at the tables of
the data for the 9 studies included in the analysis. The data looked a
bit odd, and when I went to the library and checked the original papers
I found that they had copied the standard errors from some papers and
the standard deviations from others, and assumed they were all s.d.'s. I
expect most of you can guess what the reviewer's comments were like!
However, it did lead me to wonder just how often this happens, as many
journals don't have statistical advisors. The main journals do, but not
many of the specialist area ones.
Irene Stratton
|