The therapeutic community: multivariate prediction of retention

Multiple regression procedures were employed to study the profile of resident time in program, demographic, sociobackground, and psychological variables in relationship to retention. The sample consisted of 809 Phoenix House residents who were primarily male (81.4%) and Black (64.9%). A psychological battery was administered on a cross-sectional basis. The multiple R of .44 and R2 of .19 were large for this type of research. Time in program accounted for over half the variance explained. Ten other measures contributed significantly to the explained variance. Residents who dropped out were less likely to be Blacks and more often came from lower SES families but with less family deviance; they made fewer prior attempts at giving up drugs and had lower education levels. Psychologically, they were generally more defensive and disturbed.