Friday, July 02, 2010

B - Sources of dissemination bias

Bowden J, Jackson D, Thompson SG. Modelling multiple sources of dissemination bias in meta-analysis. Statistics in Medicine 2010;29:945-955. DOI: 10.1002/sim.3813

Asymmetry in the funnel plot for a meta-analysis suggests dissemination bias, perhaps caused by publication bias or selective reporting by authors. Studies with statistically significant results or larger effect sizes are more likely to be published, giving a biased picture of the evidence. Considering a realistic scenario in which multiple processes, involving both authors and journals, statistical methods can be used look for potential effects of multiple biases in meta-analysis.

No comments: