From Arnold Kester, Maastricht University
The funnel plots in your article misrepresent an important point. The plots
as they are printed on page 31 suggest that it is mainly large studies with
negative results that are likely to go missing through publication bias. In
reality, as is aptly illustrated in the cited article by Givens et al, small
studies with negative or neutral results are more likely never to get
published.
This would create a gap in the lower right part of the funnel plot rather
than the upper right. Since the results of smaller studies will spread much
wider than those of large studies, this gap will be even more conspicuous than
the one illustrated.
The Netherlands
