Psychometric properties of the Chinese version of the ASCOT-SCT4 in Hong Kong: a validation study using classical test theory and Rasch analysis
Abstract
Hong Kong’s rapid population ageing increases demand for long-term care (LTC) and places pressure on public expenditure. The Adult Social Care Outcomes Toolkit Four-level Self-Completion Tool (ASCOT-SCT4) is a preference-based instrument specifically developed to measure social care-related quality of life (SCRQoL); however, evidence regarding the construct validity and reliability of its Chinese version in Hong Kong is lacking. This study evaluates these psychometric properties among older community care users using both classical test theory (CTT) and Rasch analysis. A cross-sectional survey was conducted with 312 older adults aged ≥ 60 years receiving community care and support services; 119 participants completed a follow-up retest approximately 1–4 weeks later. CTT analyses were conducted to assess internal consistency, factor structure, construct validity via hypothesis testing, and test-retest reliability. Rasch analysis was employed to evaluate item fit, response category ordering, person-item targeting, reliability (person and item), and differential item functioning (DIF) across age and gender groups. The CTT analyses showed acceptable internal consistency (Cronbach’s alpha = 0.79), good test-retest reliability (intraclass correlation coefficient = 0.809), and sufficient construct validity (> 75% of the hypotheses were confirmed). A one-factor structure was supported; however, the Dignity item had a low factor loading (0.269). The Rasch results indicated satisfactory item fit (Infit/Outfit statistics within 0.50–1.50), ordered response thresholds, and acceptable reliability (person: 0.721, item: 0.964), with separation indices of 1.606 (person) and 5.182 (item). However, a measurement gap was observed for respondents with higher (better) SCRQoL levels. No evidence of DIF was detected. The Chinese version of the ASCOT-SCT4 is a valid and reliable instrument for use among older community care users in Hong Kong. Nevertheless, the suboptimal performance of the Dignity item and limited measurement coverage at higher SCRQoL levels warrant further validation and potential refinement.