CMStatistics 2022: Start Registration
View Submission - CMStatistics
Title: A single framework for the analysis of effect sizes in cross-sectional and longitudinal studies Authors:  Kaidi Kang - Vanderbilt University (United States) [presenting]
Kristan Armstrong - Vanderbilt University (United States)
Suzanne Avery - Vanderbilt University (United States)
Maureen McHugo - University of Colorado Anschutz Medical Campus (United States)
Stephan Heckers - Vanderbilt University (United States)
Simon Vandekar - Vanderbilt University (United States)
Abstract: Effect size indices are useful tools for communicating study findings. Reporting effect size estimates with their confidence intervals (CIs) can be an excellent way to simultaneously communicate the strength of the observed evidence and the precision of the evidence. However, the existing effect size indices are all highly restricted by model type. They are limited to the cross-sectional study setting, and the indices for longitudinal analysis are poorly defined. These restrictions unavoidably lead to difficulties in effect size communication not only between cross-sectional studies using different models but also between cross-sectional and longitudinal studies and longitudinal studies with different numbers of measurements. We previously proposed a robust effect size index (RESI) which is advantageous over common indices, especially because it is widely applicable across different models such that cross-sectional studies using different models can easily report effect sizes with CIs in an analysis of variance (ANOVA) table format. In the current research, we extended the framework to longitudinal analysis and defined a new RESI that is not affected by the number of measurements. Thus, researchers studying the same scientific questions but using different data types or study designs (cross-sectional or longitudinal) can easily communicate their observed effect sizes and CIs without having to translate between them.