2018/01/26 by Sophia Kyriakou, Kyriakou, Sophia, Ioannis Kosmidis +3
Decision Sciences · #62F03 #62F12 #62P10 #Applications (stat.AP) #FOS: Computer and information sciences #Meta-analysis and systematic reviews #Methodology (stat.ME)
paper · pdf · doi:10.48550/arxiv.1801.09002
openalex publication_date 2018/01/26 · openalex created_date 2022/09/29 · openalex updated_date 2026/07/28
Random-effects models are frequently used to synthesise information from\ndifferent studies in meta-analysis. While likelihood-based inference is\nattractive both in terms of limiting properties and of implementation, its\napplication in random-effects meta-analysis may result in misleading\nconclusions, especially when the number of studies is small to moderate. The\ncurrent paper shows how methodology that reduces the asymptotic bias of the\nmaximum likelihood estimator of the variance component can also substantially\nimprove inference about the mean effect size. The results are derived for the\nmore general framework of random-effects meta-regression, which allows the mean\neffect size to vary with study-specific covariates.\n