Development and Validation of the Mental Health Provider Stigma Inventory

Background and Purpose:  Mental health provider stigma has potentially far reaching implications for client outcomes and the provision of services to marginalized and vulnerable populations.  Although stigmatization of mental health clients is well explored in the literature, the majority of stigma reduction strategies are not aimed at mental healthcare providers.  The Mental Health Provider Stigma Inventory (MHPSI) was developed to address this gap in the literature.  The MHPSI assesses stigma on three domains: attitudes, behaviors, and coworker influence.  Hypotheses of the validation study included finding adequate internal consistency, confirming the three factor model, and that construct validation would provide evidence that the MHPSI measured the constructs it was intended to measure. 

Methods:  Content validation was conducted by eight expert panelists, all with mental health experience.  After this process, a 41-item MHPSI was piloted among a purposive, non-probability sample of 212 mental health professionals and graduate student counseling interns.  The online survey included the MHPSI, a single-item validity indicator, two validated scales measuring therapeutic relationships and depersonalization, and a demographic questionnaire.  The three hypothesized subscales were then evaluated for reliability, factor structure, and construct validity.  Kurtotic items were removed from analysis, and as data were MCAR (χ² = 603.952, p = .336), missing values were replaced using expectation maximization.  Correlations between items were investigated to identify items with duplicate content.  Then, alpha-if-item-deleted indices were examined for each subscale.  Internal consistency was assessed using Cronbach’s alpha, stratified alpha, and standard error of measurement.  Factor structure was verified using a confirmatory factor analysis and further adjustments were made to optimize indices.  Finally, convergent and discriminant construct validity was assessed using Pearson’s correlations.

Results:  The MHPSI was refined during reliability and validity analyses, and a 24-item final version of the scale emerged.  Reliabilities were strong: Cronbach’s α for each of the proposed subscales ranged from .83 to .93; the global stratified α was .95.  Factor analysis confirmed the hypothesized structure, with fit indices ranging from adequate to excellent.  Evidence of construct validation endorsed hypotheses about how scores on the MHPSI subscales would relate to theoretically related constructs.  All correlations were significant at alpha = .05. Specifically, the attitudes subscale correlated negatively with therapeutic relationships (r² = .18), the behaviors subscale correlated positively with frequency (r² = .29) and strength (r² = .22) of depersonalization, and the coworker influence subscale correlated positively with a single-item validity indicator (r² = .17).

Conclusions and Implications:  Preliminary psychometrics of the MHPSI suggest that the scale is a reliable and valid measure of provider stigma among the sample.  The MHPSI is anticipated to take ten minutes for respondents to complete, making it an accessible and innovative tool for the diverse range of persons who provide mental health services both formally and informally in a variety of settings.  Initial validation suggests that the MHPSI may be a valuable tool for professional training, continuing education, and workplace assessment.  Results on the scale may help researchers or program administrators develop targeted workshops to optimize service delivery in a specific workplace setting.