Combining group-based trajectory modeling and propensity score matching for causal inferences in nonexperimental...

Area of Science:

• Developmental psychology
• Psychopathology research
• Causal inference methods

Background:

• Understanding how significant life events influence developmental trajectories is crucial in human development and psychopathology research.
• Existing methods often struggle to establish clear causal links between turning-point events and behavioral changes.
• Observational longitudinal data offers rich insights but requires robust analytical approaches for causal inference.

Purpose of the Study:

• To present a novel statistical method for transparent causal inference on the impact of turning-point events on developmental trajectories.
• To integrate finite mixture modeling of trajectories with propensity score matching for enhanced analysis.
• To provide a framework for analyzing how specific events alter behavioral pathways over time.

Main Methods:

• Combines finite mixture modeling for trajectory analysis with propensity score matching.
• Uses propensity scores to balance observed covariates between groups.
• Employs trajectory groups to control for pre-treatment measures and characterize subject classes.

Main Results:

• The method allows for more transparent causal inferences regarding the effects of life events on developmental paths.
• Demonstrates application in analyzing the impact of gang membership on violent delinquency.
• Successfully characterizes subject classes, including those lacking suitable matches.

Conclusions:

• The proposed method offers a powerful tool for causal inference in longitudinal developmental research.
• It enhances the understanding of how discrete events shape individual behavioral trajectories.
• Applicable to various fields studying developmental changes and event impacts.