Measurement Invariance of HIV-Related Stigma Scales among Men Who Have Sex with Men (MSM) and Non-MSM Populations: Implications for Comparative Studies in China

Tianyue Mi ^1* Xueying YANG ¹ Lan Guanghua ² Shan Qiao ¹ Zhiyong Shen ² Yuejiao Zhou ² Xiaoming Li ¹

• ¹ University of South Carolina, Columbia, United States
• ² Guangxi Center for Disease Control and Prevention, Nanning, China

Background Measurement invariance ensures that scales used in research measure the same constructs across different groups. As HIV-related stigma scales are increasingly used in studies involving men who have sex with men (MSM) and non-MSM populations, it is crucial to evaluate the equivalence of these measures. This study examines the measurement invariance of internalized, anticipated, and enacted HIV-related stigma scales between MSM and non-MSM populations in China.Methods Data were derived from two studies: a prospective cohort study with 193 MSM and 579 non-MSM, and a cross-sectional survey of 402 MSM. Participants completed the 8-item internalized, 9-item anticipated, and 16-item enacted HIV-related stigma scales. Confirmatory factor analysis was used to test measurement invariance by progressively adding equality constraints to the models for each stigma dimension.Results Partial scalar measurement invariance was achieved for the internalized stigma scale, allowing the intercepts of items 2, 3, and 6 to vary (χ 2 =89.32, df=43; CFI=.986; TLI=.981; RMSEA=.043, 95%CI [.030, .056]; SRMR=.033), indicating that the zero points of item 2 ("I feel ashamed of having HIV"), item 3 ("Having HIV makes me feel unclean"), and item 6 ("I feel guilty because I have HIV") were different between MSM and non-MSM. Partial residual measurement invariance was established for the anticipated stigma scale by allowing the residuals of item 2 to vary (χ 2 =93.57, df=66; CFI=.994; TLI=.993; RMSEA=.027, 95%CI [.012, .038]; SRMR=.022), indicating that the item variance that could not be explained by the factor was different between MSM and non-MSM. For the enacted stigma scale, partial scalar invariance was achieved by allowing the threshold of item 7 to vary (χ 2 =314.74, df=219; CFI=.987; TLI=.986; RMSEA=.027, 95%CI [.020, .034]; SRMR=.088), indicating that the threshold of item 7 was different between MSM and non-MSM.The study supports the use of these HIV-related stigma scales for comparing MSM and non-MSM populations, though caution is needed as some items demonstrated partial measurement invariance. These findings provide a foundation for future research and interventions aimed at reducing HIV-related stigma across diverse groups.