PUBLICATION BIAS: THE ACHILLES’ HEEL OF SYSTEMATIC REVIEWS?
The term ‘publication bias’ usually refers to the tendency for a greater proportion of statistically significant positive results of experiments to be published and, conversely, a greater proportion of statistically significant negative or null results not to be published. It is widely accepted in the fields of healthcare and psychological research to be a major threat to the validity of systematic reviews and meta-analyses. Some methodological work has previously been undertaken, by the author and others, in the field of educational research to investigate the extent of the problem. This paper describes the problem of publication bias with reference to its history in a number of fields, with special reference to the area of educational research. Informal methods for detecting publication bias in systematic reviews and meta-analyses of controlled trials are outlined and retrospective and prospective methods for dealing with the problem are suggested.