Measuring Likeability and Impressions of Depressive Affect: a Psychometric Instrumentation Study

Purpose: The causes and consequences of mental illness stigma remain poorly understood. Moreover, few validated instruments exist to measure mental illness stigma or its consequences. This study was based on the premise that when someone is diagnosed with Major Depressive Disorder, stigma is likely to manifest in two ways: 1) decreased likeability of and 2) stronger impressions of depressive affect in the diagnosed person (because depressive affect is the most salient feature of the diagnosis). In order to investigate these issues, instruments are needed to measure 1) likeability and 2) impression of depressive affect. The only previously existing measure of likeability found in the literature exhibits serious flaws in content validity. No measures of impression of depressive affect were found in the literature, and clinical measures of depression were deemed inappropriate for the purpose at hand. Therefore the present study develops and validates a Scale of General Likeability (SGL) and an Impression of Depressive Affect Scale (IDAS).

Method: A convenience sample of 293 undergraduate students was drawn to examine the reliability, construct validity, and factorial validity of the IDAS and SGL scales. A 2x2 experimental design was used to assess construct validity (convergent and discriminant) by manipulating the factors of depressive affect and likeability in the content of vignettes to determine whether participants' scale scores (on the final scale items) vary as hypothesized in rating the vignette protagonist. Alpha reliability analyses were used to select the best-performing items from the original pool of items. These analyses were conducted for the sample as a whole and for each of four sub-samples that were randomly exposed to different levels of depressive and likable content. Multiple group confirmatory factor analysis (CFA) was used to provide additional evidence of item performance. Finally, factorial validity was examined with structural equation modeling (SEM).

Results: Good to excellent reliability was observed for both scales. Both final scales showed excellent reliability (>.90) for the sample as a whole and good reliability (>.80) for each of these sub-samples. The multiple group method applied to the final scales showed high intended factor loadings on both final scales and comparatively low unintended factor loadings, for all retained items. An SEM model showed adequate fit for the two-factor model (chi-square/df ratio = 1.91, RMSEA = .056, NNFI = .99, CFI = .99, GFI = .91). Moreover, construct validity was supported as both final scales responded to experimental manipulation of vignette content as hypothesized.

Implications: Findings support the reliability, factorial validity, and construct validity of the SGL and IDAS. It is hoped these scales will be used in studies on the effect of mental illness diagnostic labels on impressions of the diagnosed person's likeability and level of depressive affect. Such studies may contribute to the understanding of mental illness stigma and thus aid efforts to combat it through education and intervention with social work students and practitioners. Both scales also have potential value outside the field of mental illness stigma.