Statistical Verification of Computational Rapport Model
This work addresses the problem of verifying computational models of rapport for researchers in human-computer interaction or communication studies, but it is incremental as it builds on existing theoretical models.
The paper tackled the lack of solid verification for computational rapport models by applying structural equation modeling to theoretical models on friend and stranger dyads, identifying unfavorable paths and modifying the model to include nonverbal behaviors like gaze and smile, with fit indices showing improved goodness of fit.
Rapport plays an important role during communication because it can help people understand each other's feelings or ideas and leads to a smooth communication. Computational rapport model has been proposed based on theory in previous work. But there lacks solid verification. In this paper, we apply structural equation model (SEM) to the theoretical model on both dyads of friend and stranger. The results indicate some unfavorable paths. Based on the results and more literature, we modify the original model to integrate more nonverbal behaviors, including gaze and smile. Fit indices and other examination show the goodness of our new models, which can give us more insight into rapport management during conversation.