Modeling Psychophysiological Processes in dyadic interactions

Open Access
Liu, Siwei
Graduate Program:
Human Development and Family Studies
Doctor of Philosophy
Document Type:
Date of Defense:
May 11, 2012
Committee Members:
  • Michael J Rovine, Dissertation Advisor
  • Michael J Rovine, Committee Chair
  • Peter Cm Molenaar, Committee Chair
  • Susan Mc Hale, Committee Member
  • Runze Li, Committee Member
  • Matthew Goodwin, Special Member
  • dyadic analysis
  • psychophysiology
  • cortisol
  • time series
  • electrodermal activity
Psychophysiological interactions constitute an important aspect of interpersonal relationships. Although behavioral scientists often conceptualize interpersonal units (e.g., dyad, family, or group) as dynamic systems, in which social partners are interwoven in terms of their physiology, psychology, and behavior, relatively little research has examined these dynamic processes, especially at the physiological level. This is partly due to the underdevelopment of statistical methods to analyze repeated measures data from interdependent individuals. The general purpose of this dissertation is to apply state-of-the-art methodology in longitudinal dyadic analysis to address empirical questions related to interpersonal physiological processes, and to develop new statistical approaches that researchers can use to model dyadic interactions. This dissertation consists of three studies. Study 1 investigated coregulation of diurnal cortisol pattern in married couples using data from the Penn State Hotel Work and Well-Being project. Saliva samples containing cortisol were obtained from 28 heterosexual couples four times a day for four consecutive days. Multilevel modeling was conducted to examine whether husband and wife coregulate in their cortisol awaking responses (CAR) and diurnal cortisol slopes (DCS). Results showed synchrony (concurrent covariation) in DCS between spouses. For CAR, the strength of synchrony was stronger in couples characterized by higher levels of spousal strain and disagreement. In addition, cross-lagged models revealed that an individual’s diurnal cortisol pattern on a particular day depended on his/her own, but not his/her spouse’s, cortisol pattern on a previous day. Stability in CAR was stronger in couples reporting higher levels of spousal support. The discussion highlights the family systems perspective and the methodological advantages of the analytic technique. Study 2 and 3 were based on a project examining dynamic interactions in physiological arousal between children with Sensory Processing Disorder (SPD) and their therapists, as indicated by electrodermal activity (EDA). Study 2 compared three approaches for handling missing data in multivariate time series, which is common in studies of physiological interactions. A recursive prediction routine developed by the author was found to outperform listwise deletion and data imputation using sample means and variances in recovering the parameters of the time series models. Study 3 introduced an advanced signal processing technique - time-frequency analysis (TFA) - to study interpersonal dynamic interactions. A program called SAM-MOW (Spectral Analysis for Multivariate data using a MOving Window) was developed by the author to carry out TFA automatically. The method was used to analyze EDA data from one child-therapist dyad during therapy. The analysis revealed a unidirectional influence from the therapist to the child (i.e., the child’s EDA was predicted by the therapist’s EDA). In addition, the magnitude of influence appeared to be contingent on guided activities during the therapy. The discussion focuses on the strengths and weaknesses of this novel methodological approach. In sum, this dissertation focuses on the application and development of statistical methods to study interpersonal physiological interactions from a dynamic systems perspective. It contributes both to our knowledge of physiological processes underlying dyadic interactions, and to methodological development in this area.