Online Program Home
My Program

Abstract Details

Activity Number: 542 - New Research Synthesis Methods in Data Science
Type: Invited
Date/Time: Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
Sponsor: International Chinese Statistical Association
Abstract #300019
Title: Innovative Methods for Assessing Publication Bias in Meta-Analysis
Author(s): Lifeng Lin*
Companies: Florida State University
Keywords: Funnel plot; Meta-analysis; Publication bias

Publication bias (PB) is a major threat to the validity of meta-analyses. Funnel plots have been a popular tool for assessing PB. They usually present treatment effects against their standard errors (SEs), and the contours of different significance levels have been used to distinguish PB from other factors that may cause the plots’ asymmetry. However, treatment effects and their SEs may be associated even if no PB exists, so SE-based funnel plots may lead to false positive conclusions. Also, the missingness of studies may relate to their sample sizes. Therefore, funnel plots based on sample sizes (N-based) can be an alternative. We derive contours for N-based funnel plots, which help meta-analysts properly interpret such plots’ asymmetry. In addition, we introduce six measures to quantify the magnitude of PB. Currently, no guides are available for judging whether the PB is minimal or substantial. Based on the measures’ empirical distributions among a large collection of meta-analyses, we propose cutoffs that categorize the magnitude of PB as unimportant, moderate, substantial, and considerable. They help evidence users grade the certainty in evidence impacted by PB.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program