The art of analyzing developmental trajectories: recommendations for developmental trajectory group analyses

The application, analysis, and interpretation of developmental trajectory group analyses, a statistical method applied to public health and other fields. Utilizing simulation studies and TRacking Adolescents’ Individual Lives Survey (TRAILS) cohort data, the thesis scrutinizes the performance of Latent Class Analysis (LCA), Latent Class Growth Analysis (LCGA), or Growth Mixture Modeling (GMM) under various conditions. One study highlights the impact of class separation and sample size, revealing GMM as generally more robust, particularly if data are less ideal. Another study delves into variance constraints in GMM, emphasizing the delicate balance required to avoid bias. The notorious cat's cradle pattern is explored, revealing that misinterpretations are more likely to arise from model specifications rather than from intrinsic data patterns.
The thesis concludes with guidance for researchers, stressing the necessity of a nuanced understanding of methods, theory, and interpretation. The thesis underscores the importance of recognizing trajectory classes as a statistical data summary rather than substantive subgroups. As a metaphor, trajectory group analyses are like baking the perfect chocolate chip cookie, emphasizing the need for a comprehensive approach akin to knowing ingredients, understanding the impact of process decisions, and contemplating the end result.