Quantifying the dyadic interaction: A methodological framework for conducting observational moment-by-moment research

Systematic observation is a valuable method for studying interactive phenomena in dyads, as it captures sequences of behavior, potentially enhancing validity. However, observational studies remain underutilized compared with other tools, such as self-report questionnaires, because they require time and resources. This work aims to facilitate observational research by providing a step-by-step guide for preprocessing and exploring data from systematic observations, enabling flexible data handling beyond traditional sequential analyses. To illustrate the guide's application, we used a dataset of 154 psychotherapy sessions collected longitudinally across different treatment phases from 32 patient-therapist dyads. We coded sessions using the Therapeutic Relationship Coding System (TRCS), which captures moment-by-moment verbal interactions relevant to the therapeutic alliance. Post-session questionnaire data were collected using the Working Alliance Inventory (WAI). An R script and a set of functions were developed to assist researchers with data preprocessing and to generate descriptive visualizations addressing the following questions: (1) What behaviors (and sequences of interacting behaviors) are more frequent?; (2) How do behaviors evolve over time; (3) Do behavioral changes differ across groups? After applying our proposed analytical tools, we interpret our results in light of the dominant theoretical framework on the therapeutic alliance.