PUBLICATION BIAS IN META ANALYSES FOR MENTAL HEALTH

DESCRIPTION (Applicant's abstract): The goal of this project is to develop software to address publication bias in meta analysis. The software will incorporate various computational approaches including procedures that focus on the presence of bias, others that yield an adjusted effect size, and others that address the robustness of the conclusions. These procedures will be integrated with graphical approaches, so that it is clear why bias appears to exist and how adjustments to effect size are being made. These procedures are meant to be applied as a key component of any meta analysis, to assess the likely impact of publication bias on the conclusions. When the impact is severe, this information will allow the researcher to avoid potentially serious consequences. When the impact is modest or trivial, this information is also critically important as it speaks to the validity of the analysis and should be reported as a key part of the results. The module will include functions to import data from any program that is being used to run the meta analysis. PROPOSED COMMERCIAL APPLICATION: The project will produce a computer program that allows researchers to assess the potential impact of publication bias on a meta analysis. This program will have wide application in the fields of mental health research, gerontology, AIDS research, and cancer research as well as the social sciences. It will be distributed by Biostat, SPSS and by Lawrence Erlbaum Associates.