Date of Submission


Degree Type


Degree Name

Doctor of Psychology (PsyD)



Department Chair

Robert A. DiTomasso, Ph.D., ABPP, Chair, Department of Psychology

First Advisor

Virginia Salzer, Ph.D. , Chairperson

Second Advisor

Elizabeth Gosch, Ph.D.

Third Advisor

Edna B. Foa, Ph.D.


The present study focused on the therapeutic alliance in Prolonged Exposure Therapy for adolescent girls with sexual abuse-related Posttraumatic Stress Disorder (PTSD). Two major questions were investigated. First, how does the therapeutic alliance act as a moderating variable between outcome and three predictor variables (homework compliance, habituation to the trauma memory, and number of sessions completed)? Second, how is the relationship between outcome and these same variables affected by the strength of the therapeutic alliance? The data used in this study were taken from an existing database of information on 40 sexually abused adolescent girls with PTSD (mean age = 14.6; range = 13 to 17; SD = 1.44) treated at a rape crisis center in Philadelphia for an average of 12 sessions (range = 6 to 18; SD = 3.62). Of the sample, 25 were African American (62.5%), 8 were Caucasian (20.0%), 5 were Hispanic (12.5%), and 2 were biracial (5.0%). Child Posttraumatic Stress Scale self-report scores, homework compliance ratings, peak Subjective Units of Distress ratings, number of sessions completed, and Working Alliance Inventory – Observer version ratings were collected from the database. Therapeutic alliance as a moderator variable was tested through a regression equation with the addition of totaling the product of the moderator and each predictor variable (homework compliance, habituation to the trauma memory, and number of sessions completed). The mediator qualities of these same three variables were also tested through a series of three regression models. Therapeutic alliance was not found to be meaningfully related to treatment outcome in this sample. As a result, alliance was not a moderator of the relationship between outcome and the predictor variables. The mediating variables were also not found to be affected by the alliance. While a number of possible confounds (small sample size, selection bias, exclusion of early dropouts, timing of assessment, and rater perspective) may have affected these results, the implications of the findings for current clinical practice call into question assumptions about therapeutic alliance in the treatment of adolescent girls with sexual assault-related PTSD.