Measurement invariance of health-related quality of life: a simulation study and numeric example

Abstract

Measurement invariance (MI) is a prerequisite to conduct valid comparisons of Health-related quality of life (HRQOL) measures across distinct populations. This research investigated the performance of estimation methods for testing MI hypotheses in complex survey data using a simulation study, and demonstrates the application of these methods for a HRQOL measure. Four forms of MI were tested using confirmatory factory analysis. The simulation study showed that the maximum likelihood method for small sample size and low intraclass correlation (ICC) performed best, whereas the pseudomaximum likelihood with weights and clustering effects performed better for large sample sizes with high ICC to test configural invariance. Both methods performed similarly to test other forms of MI. In the numeric example, MI of one HRQOL measure in the Canadian Community Health Survey was investigated and established for Aboriginal and non-Aboriginal populations with chronic conditions, indicating that they had similar conceptualizations of quality of life.