Biography

Dr. Burchinal is a Research Professor of Psychology and Director of the Data Management and Statistics Core at the FPG Child Development Institute at the University of North Carolina at Chapel HillShe served as the primary statistician for many educational studies of early childhood, including the Abecedarian project, Cost, Quality and Outcomes Study, and the NICHD Study of Early Child Care. As an applied methodologist, she helped to demonstrate that sophisticated methods such as meta-analysis, fixed-effect modeling, hierarchical linear modeling, piecewise regression, and generalized estimating equations provide developmental researchers with advanced techniques to address important issues for research and policy. In addition, she has pursued her substantive interest in early education as a means to improve school readiness for at-risk children.

2-051 - Best Practices in Statistical Methodology for Developmental Science Research

Session Type: Invited State of the Art Symposium

Integrative Statement

Best practices in statistical methodologies for developmental science research describe statistical approaches that are believed to provide the strongest methodologies for this field. Within the past ten year, there have been major advances in the statistical methods used by developmental researchers. This symposium describes best practices in three areas: causal inference and bias reduction, mediation and moderation, and longitudinal data analysis. Each of these areas are crucial for addressing developmental research questions. Elizabeth Votruba-Drzal will describe issues involved in addressing causal inferences from developmental studies and will present approaches that can be used with observational data that can reduce selection bias. Todd Little will discuss issues involved with testing hypotheses of mediation and moderation that are typically central to developmental theories, and will provide guidance regarding best practices for identifying the most appropriate and powerful methods for testing mediation and moderation in developmental studies. Margaret Burchinal will talk about issues involved with testing hypotheses regarding development using longitudinal data, and will describe variable-centered and person-centered longitudinal analysis approaches. Each presenter will discuss the issues, the statistical methods, and provide an exemplary analysis.

Sub Unit

• SRCD Biennial Meeting

Chair

• Margaret Burchinal, University of North Carolina, Chapel Hill

Individual Presentations

• Causal Inference - Presenting Author: Elizabeth Votruba-Drzal, University of Pittsburgh

• Mediation and Moderation - Presenting Author: Todd D Little, Texas Tech University

• Longitudinal Methods - Presenting Author: Margaret Burchinal, University of North Carolina, Chapel Hill