A Comparison of Variable- and Person-Oriented Approaches in Evaluating a Universal Preventive Intervention

Area of Science:

• Child Psychology
• Developmental Psychology
• Educational Psychology

Background:

• Prevention programs like the PAX Good Behavior Game (PAX) often involve complex, multilevel, and longitudinal data with multiple outcomes.
• Evaluating these programs requires sophisticated statistical methods to capture nuanced effects.

Purpose of the Study:

• To present and compare variable-oriented and person-oriented statistical approaches for evaluating interventions with multilevel, longitudinal, multivariate outcomes.
• To demonstrate how these complementary methods offer distinct yet valuable insights into program effectiveness.

Main Methods:

• Utilized data from the Manitoba PAX Study, employing multilevel linear regression (variable-oriented) and latent transition analysis (person-oriented).
• Applied both approaches to analyze multiple outcome variables (emotional, behavioral, relationship problems) across multiple time points.
• Examined the influence of gender and socio-economic status on program effects.

Main Results:

• The variable-oriented approach identified overall PAX program effects on each outcome variable.
• The person-oriented approach revealed how the intervention influences the transition of multiple outcomes over time, varying by student risk profiles.
• Both methods provided insights into how gender and socio-economic status moderate program impact.

Conclusions:

• Combining variable-oriented and person-oriented approaches provides a more comprehensive evaluation of complex prevention programs.
• These complementary statistical strategies enhance our understanding of intervention effects on individual trajectories and overall outcomes.
• The findings have implications for optimizing the evaluation of interventions targeting child behavior and well-being.