The Continuous Assessment of Interpersonal Dynamics (CAID) is an observational coding method that enables continuous tracking of warmth and dominance in both members of a dyad as an interaction unfolds. Research using this tool has revealed dynamic patterns relevant to psychopathology and psychotherapy, suggesting considerable potential for clinical assessment and research. However, CAID data are sensitive to a variety of person and situational factors, and the way that these factors combine to influence CAID data is poorly understood. We examined data from ten raters using CAID to assess moment-to-moment warmth and dominance in opposite-sex married dyads (N = 137 couples) interacting in four distinct situations. Using Generalizability Theory methods, we decomposed sources of variance in CAID data and estimated multiple forms of between- and within-person reliability. Results revealed how spouses’ moment-to-moment behaviors varied as a function of person, sex, dyad, rater, situation, and relevant interactions between these factors. Based on these results, we provided guidelines for interpreting CAID data at different levels of aggregation in clinical research and practice.
