Factor Structure and Rasch Performance of the Shona Multidimensional Scale of Perceived Social Support among Zimbabwean Caregivers: A cross-sectional psychometric validation study

Abstract Background The MSPSS is widely used to assess perceived social support, but cross-cultural adaptations often yield factor structures that differ from the original three-factor model. This reflects the sociocultural nature of social support as a subjective latent construct. We translated, culturally adapted and evaluated the Shona MSPSS among caregivers in Zimbabwe. Methods The MSPSS was translated using forward–backward translation, reconciliation, expert review and cognitive debriefing. It was administered to caregivers of typically developing children (TDC; n = 771) and caregivers of children with cerebral palsy (CP; n = 461). Psychometric evaluation combined classical test theory methods (factor analyses, internal consistency, test–retest reliability, known-groups validity and construct validity) with Rasch analysis using the Partial Credit Model. Results EFA and CFA supported a three-factor structure representing family, friends and significant other support, although CFA fit indices were mixed. Internal consistency was good for subscales (α = 0.833–0.892) and the total scale (α = 0.868). Four-week test–retest reliability was excellent (ICC = 0.980; 95% CI: 0.959–0.990; n = 33). Caregivers of TDC reported higher support than caregivers of children with CP, and higher support was associated with lower caregiver burden, lower psychiatric morbidity and better health-related quality of life. Rasch analysis showed acceptable reliability for the original 12-item scale but violations of model fit, threshold ordering, local independence, unidimensionality and item invariance. After dichotomising responses and removing item 5, the revised 11-item version achieved acceptable fit (χ²(66) = 81.76, p = 0.091; RMSEA = 0.001), acceptable reliability (PSI = 0.731; α = 0.722) and unidimensionality, although residual local dependency and age-related DIF for item 8 remained. Conclusions The MSPSS-Shona showed strong classical test theory evidence of reliability and validity and replicated the original three-factor structure. The 12-item version is supported as a reliable ordinal measure, whereas interval-level scoring may require the revised 11-item dichotomised version. Further independent validation is needed before routine metric use.