Practical Issues with Aggregate and Individual-Level Meta-Analytic Methods for Combining Results from Longitudinal Studies

• Activity Number: 626
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Ryan Williams*+ and Terri Pigott
• Companies: University of Memphis and Loyola University Chicago
• Keywords: meta-analysis ; ipd meta-analysis ; longitudinal data analysis ; integrative data analysis

Abstract

Meta-analysis is a common method for synthesizing existing evidence on a given topic. Traditional meta-analytic methods combine summary information provided in study reports. This form of aggregate data meta-analysis is commonly used to synthesize treatment effect estimates and bivariate correlations in the social and behavioral sciences. These methods are only useful when a collection of effect sizes estimate a common parameter. When, for example, treatment effect estimates are unconditionally estimated in some study and conditionally estimated in others, aggregate data meta-analysis is not easily able to combine that body of evidence. This paper will discuss some practical limitations to aggregate data meta-analysis in the context of combining conditional effect size estimates from ordinary least squares regressions. Then it will discuss some of the benefits of using individual participant data meta-analytic methods and how individual participant data may be synthesized with aggregate data. The paper will conclude with a discussion of how to build capacity for this area of research through replication and data sharing initiatives.