A Regression Framework for Effect Size Assessments in Longitudinal Modeling of Group Differences

Area of Science:

• Prevention Science
• Clinical Psychology
• Psychiatry
• Quantitative Psychology

Background:

• Growth modeling analysis (GMA), including multilevel and latent growth modeling, has seen a significant rise in application for intervention effect testing.
• Calculating standardized effect sizes, like Cohen's d, for GMA findings presents conceptual challenges.
• Existing methods for effect size calculation in GMA are not always integrated with classical analyses.

Purpose of the Study:

• To review conceptual issues in calculating Cohen's d for growth modeling analysis.
• To introduce a novel, integrative framework for effect size assessment that encompasses GMA.
• To provide a unified approach for calculating effect sizes across diverse analytical methods.

Main Methods:

• Conceptual review of effect size calculation in GMA.
• Introduction of an integrative framework based on linear regression models.
• Demonstration of calculating effect sizes using familiar statistics (e.g., regression coefficient, standard deviation, study duration).

Main Results:

• The proposed framework allows for the calculation of model-based effect sizes (Cohen's d) within GMA.
• The integrative approach subsumes GMA within a broader linear regression structure.
• Effect sizes can be consistently calculated for both cross-sectional and longitudinal data.

Conclusions:

• A unified framework simplifies effect size estimation in intervention research utilizing GMA.
• This approach enhances comparability between GMA findings and traditional meta-analytic results.
• The method facilitates a more standardized approach to quantifying intervention impact across various study designs.