Methods: Participants were 290 individuals with SMI recruited from community-based psychosocial rehabilitation programs. Two theoretical models – a traditional Sociological Model (Link, et al., 1989: without affect) and an alternative Social Psychological Model (with affect) – were tested through SEM (Structural Equation Modeling). Analyses examined whether affect, which has not been adequately addressed in the traditional Sociological Model, has a significant impact on self-esteem above and beyond beliefs in devaluation/discrimination. Analyses controlled for symptoms and rejection experiences as well as demographic and mental health background variables. Each measurement model was developed with exploratory and confirmatory factor analyses based on the literature.

Results: The Social Psychological Model explained more variance in self-esteem than the Sociological Model.  Specifically, the results of measurement models indicated that self-esteem consists of two independent dimensions – i.e., self-worth and self-deprecation (Owens, 1993). SEM results indicated that affective characteristics are more influential than beliefs in predicting self-esteem and that the factors predictive of the two dimensions differ from each other.

Implications:  Findings imply that: (1) interventions should consider affect to be an important intervention target for improving self-esteem, in addition to beliefs in devaluation-discrimination; (2) social work interventions should target specific dimensions of self-esteem, either self-worth or self-deprecation, depending on the clinical characteristics of clients; and (3) different interventions are required to influence self-worth versus self-deprecation.