Data collected from sample 2 were used to assess thegeneralizability of the scale to a sample of community-recruitedadults living with chronic illness. Internal consistency was re-tested withinsample 2.
An analysis of factorial invariance was also conducted to test formeasurement equivalence among samples 1 and 2.In the start, the CIASSincluded 53 items: 17 items representing anticipated stigma from friends andfamily members, 19 items representing anticipated stigma from work colleagues,and 17 items representing anticipated stigma from healthcare workers. Their itemswere developed based on reviewing different literature stigma anticipated andexperienced by people living with a variety of chronic illnesses.
Qualitative representationsof stigmatizing treatment & the context in which the treatment occurredwere also noted. Participants were asked to rate the possibility that theywould encounter these stigmatizing experiences in the future on a scale rangingfrom 1 (very unlikely) to maximum 5 (very likely). Each item was evaluated for presencein the ?nal version of the scale. The evaluation involved examining theitem-scale correlations, item variances, and item means. Items were also chosenthat correlated well with the overall scale, had relatively high variance, andhad means that were far from the extremes of the scale.
The process resulted inthe selection of 12 items, 4 per each scale, that were evaluated forreliability and validity.Indicators of reliability were calculated for both the entire12-item scale and each 4-item subscale using the data from sample 1. The entireCIASS was highly internally consistent with an ordinal alpha of 0.95.Additionally, the friends and family, work colleagues, and healthcare workerssubscales were internally consistent, with ordinal alphas of 0.92, 0.95, and0.95 respectively.
Test-retest reliability was also calculated by comparing the time 1 & 2scores of the 38 participants from sample 1 who completed the CIASS twice.Participant scores on the entire scale were highly correlated at 0.82(P<0.001).
Additionally, participant scores on the friends and family, healthcareworkers, and work colleagues subscales were correlated at 0.67, 0.61, and 0.83 (allP <0.001) respectively.The structural validity and construct validityof the CIASS were evaluated using a con?rmatory factor analysis (CFA) with thedata from sample 1.
The Chi-square was small but statistically signi?cant, X2(51) = 88.59, p = 0.0008. However, the root mean square residual (RMR) was 0.05and the goodness-of-?t index (GFI) was 0.
99, it shows that the model is strongto for the data. The entire scale was again highly consistentwith an ordinal alpha of 0.93. The friends and family, work colleagues, andhealthcare workers subscales were also highly consistent with ordinal alphas of0.93, 0.93, and 0.95 respectively