Events
Colloquium: Chuan Hong |
| Time: Jan 20, 2016 (04:00 PM) |
| Location: Parker Hall 249 (tentative) |
|
Details: Speaker: Chuan Hong, The University of Texas Health Science Center (UTHealth) Title: Detecting publication bias in multivariate meta-analysis Abstract: Publication bias occurs when the publication of research results depends not only on the quality of the research but also on the direction, magnitude, or statistical significance of the results. The consequence is that published studies may not represent all valid studies undertaken, and this bias may threaten the validity of systematic reviews and meta-analyses. However, both detecting and accounting for publication bias are challenging in a multivariate meta-analysis setting because some studies may be completely unpublished while others may selectively report only part of multiple outcomes. We propose a pseudolikelihood-based score test for detecting publication bias in multivariate random-effects meta-analysis. To the best of our knowledge, this is the first test for detecting publication bias in a multivariate meta-analysis setting. Through simulation studies, the proposed test is found to be more powerful than the existing univariate tests. In addition, by empirically evaluating 169 systematic reviews with multiple outcomes from the Cochrane Database, the proposed multivariate test is shown to identify more studies with publication bias than existing univariate tests |