A new look at the attribution model: Considerations for the measurement of public mental illness stigma.

Abstract

Multiple versions of the Attribution Model and the corresponding Attribution Questionnaire have been used to assess public mental illness stigma. The objective of the current study was to examine (a) the factor structure of the Attribution Questionnaire and (b) relationships between constructs in the Attribution Model. Analyzing a sample of 334 U.S. adults recruited from Amazon Mechanical Turk, the authors employ confirmatory factor analyses to test three proposed factor structures of the Attribution Questionnaire and latent variable path analyses to reexamine relationships between variables in the stigmatization of people who experience mental illness. Confirmatory factor analyses of three previously examined versions of the Attribution Model revealed that the model proposed by the initial version of the Attribution Questionnaire had the best fit with the data comparative fit index (CFI = 0.92, root-mean-square error of approximation [RMSEA] = 0.07, standardized root-mean-square residual [SRMR] = 0.05). Subsequent path analyses among contructs in the model revealed acceptable model fit (CFI = 0.92, RMSEA = 0.07, SRMR = 0.06) and individual paths largely supported the hypotheses suggested by the Attribution Model. Analyses supported the original version of the Attribution Model and questionnaire with slight modifications, demonstrating that attributions of dangerousness and personal responsibility are associated with the endorsement of coercive treatment, and that attributions about dangerousness are associated with a lower desire to help. These findings suggest modifications in the current measurement of public mental illness stigma.