Using Big Data in Psychotherapy Research: Possibilites and Perils

Traditionally, psychotherapy research has used efficacy and effectiveness studies. Efficacy studies have been considered the gold standard for studying clinical interventions and effects due to their stringent controls. While not as scientifically rigorous as efficacy studies, effectiveness studies examine clinical interventions with larger and more diverse populations and more real-tolife treatment protocols. Unfortunately, effectiveness studies tend to be based on retrospective report, sometimes many months or even years after the conclusion of psychotherapy. The growth of technology, in particular smartphone applications (apps) has opened the door to a form of effectiveness study that allows for real-time data collection. The Therapy Outcome Management System (TOMS) is an iOS app that offers the potential to collect “big data.” The data set for this study included 323 Norm Development Associates, who tracked the outcome and alliance of a total 4,110 clients. Five research questions were considered. The first research question yielded a statistical difference in therapeutic alliance variables between short-term and long-term psychotherapy, indicating that long-term psychotherapy patients report statistically improved therapeutic alliance across 3 of the 5 variables used. The second research question yielded no significant differences in therapeutic alliance between gender-matched and mixedgender client-therapist dyads. The third research question investigated theoretical orientation/modality in relation to treatment duration and outcome rating, with three of the four variables used to measure outcome revealing differences. The fourth research question observed therapy outcome and treatment duration in various countries where psychotherapy is being conducted, with significant differences observed. The fifth research question observed outcome and therapeutic alliance between various diagnostic groups. Individuals with a child/adolescent disorder were found to have had statistically higher average outcome ratings in the first 8 sessions when compared to mood disorders, and those in the substance abuse disorder group reported statistically lower therapeutic alliance scores when compared to adjustment disorder, anxiety disorder, and child/adolescent disorder. The 5 research questions reported here illustrate and explore the potential of a novel research method involving big data. Implications and limitations are considered