A0831
Title: Optimal sample size planning for longitudinal multisite experiments to investigate the main and moderator effects
Authors: Wei Li - University of Florida (United States) [presenting]
Spyros Konstantopoulos - Michigan State University (United States)
Zuchao Shen - University of Georgia (United States)
Abstract: Longitudinal multisite experimental designs are commonly employed in educational interventions, where, for example, students from the same schools are randomly assigned to either a treatment or control group and subsequently followed and measured over time. One objective of longitudinal studies is to examine how treatment effects evolve over time. Additionally, educational researchers are interested in assessing whether changes in treatment effects vary among subgroups of students or schools. These student and school characteristics, often referred to as moderators, can be investigated through interaction analyses between the treatment and specific student or school characteristics. A crucial consideration in designing longitudinal experiments is determining the sample size allocation across levels and treatment conditions to ensure sufficient power to detect the effect of interest. Researchers typically plan their longitudinal studies with budget constraints in mind, as different sampling plans under the same budget can yield varying levels of statistical power. The contribution to the literature is that it provides optimal sample size computation methods for three-level longitudinal multisite experiments to explore main and moderator effects and implements these methods into an R package and a Shiny App to assist applied researchers in planning longitudinal experiments.
