An Introduction to Using Multivariate Multilevel Survival Analysis to Study Coercive Family Process
Multivariate multilevel survival analysis is introduced for studying hazard rates of observed emotional behavior relevant for coercion theory. Finite time sampling reliability (FTSR) and short-term retest reliability (STRR) across two occasions (sessions) of observation during structured problem-solving tasks several weeks apart were determined for hazard rates of emotional behaviors for parent–child dyads. While FTSR was high (.80–.96), STRR was low (.16–.65), suggesting that emotional behaviors in the context of parent–child social interaction are not very stable over a period of several weeks. Using latent variable structural equation models that corrected for the low STRR, two hazard rates were predictive of change in child antisocial behavior over a 3-year period (kindergarten to third grade) net of initial child antisocial behavior. Low levels of parent positive emotion and increases from session 1 to 2 of child neutral behavior both accounted for unique variance in third grade antisocial behavior.