The current prevalence of elevated internalized stigma was defined as an item mean score of 1.5 or higher on mean aggregated scale score not including the stigma resistance subscale (this criterion represented the “midpoint” on the 0–3 item scale (Ritsher & Phelan, 2004
; Brohan et al., 2010
). Overall, 36.1% of the sample (52 out of 144 participants) had elevated scores following this criterion. Of the participants in the New York sites, 31.1% (23 out of 74) had elevated scores, while 41.4% (29 out of 70) of the participants in Indiana had elevated scores. shows that a larger percentage of women than men had elevated internalized stigma scores (41.4% vs. 34.8%), but this difference was not statistically significant. There were no differences in likelihood of having an elevated score by race/ ethnicity. The mean total score was 1.29 (standard deviation = 0.57).
Association of Variables with Elevated Internalized Stigma
Analyses then investigated whether mean scores varied based on demographic variables. shows that most of the demographic variables studied (sex, race/ethnicity, age at first hospitalization, and total number of hospitalizations) were not significantly related to mean scale scores. However, there was evidence of a curvilinear relationship between age and internalized stigma scores, with individuals in the middle of the age distribution scoring highest. To explore this relationship, age was recoded into four categories: youngest–34, 35–44 (early midlife), 45–54 (late midlife), and 55–oldest and the relationship between age groups and mean scores was examined using analysis of variance. With this re-categorized age variable, there was a significant association between age and internalized stigma (F(3,140) = 4.75, p = 0.003). Specifically the youngest and oldest age categories had lower mean scale scores [1.00 (standard deviation = 0.51) and 1.10 (standard deviation = 0.56), respectively], while the two midlife categories had higher mean scores [1.50 (standard deviation = 0.55) and 1.39 (standard deviation = 0.56) for younger and older midlife participants, respectively]. A multiple comparison test [the Tukey Honestly Significant Difference] revealed that the young midlife participants differed significantly from both the youngest and oldest age groups, while the older midlife participants differed significantly from the young adult participants.
Association of Variables with Mean Internalized Stigma of Mental Illness Scale
When the subscales were examined, none were found to be significantly related to demographic variables, with the exception of education, which was negatively correlated with stereotype endorsement (r = −0.256, p = 0.002), indicating that more education was associated with less stereotype endorsement; and number of hospitalizations, which was negatively correlated with stigma resistance (r = −0.193, p = 0.021), indicating that participants with more past hospitalizations were less likely to endorse positively phrased items.
Despite the minimal associations with demographic variables, mean scale scores varied significantly (F = 5.24, p = 0.02) by recruitment site (New York versus Indiana), with Indiana participants showing higher mean total scores (1.41 vs. 1.19). The two sites differed significantly in a number of variables that may account for this difference, including age (participants in Indiana tended to be older), education (participants in Indiana had significantly more education), race (participants in Indiana were more likely to be European-American), gender (participants in Indiana were more likely to be male), and age at first hospitalization (participants in Indiana had later ages at first hospitalization). However, as previously noted, only age was significantly related to the mean internalized stigma score. To examine if age differences accounted for differences between the sites, we conducted an analysis of variance controlling for site and found that age and site had independent significant relationships with the internalized stigma scale score.