The case for using country-specific scoring coefficients for scoring the SF-12, with scoring implications for the SF-36

Purpose: To examine the validity of using the same scoring coefficients across countries for the SF-12. Methods: We test the equality of scoring coefficients derived for a contraction of the SF-36, the Short Form 12 (SF-12), using a large international database drawn from nine countries, to test equality between Australia and twelve other country/language groups. First, we checked that the theoretical structure of the SF-12 as set out by Ware and colleagues, but including a correlation between physical and mental health, provided an adequate fit to the data for each country/language group in a confirmatory factor analysis. We then compared Australia to all of these country/language groups in multiple-group models to assess whether a model producing common factor score coefficients provided an adequate fit to the data. We also derived Chi-squared tests for the differences between the restricted and unrestricted models, to test the equality of the factor score coefficients across countries. Results: We found that the theoretical structure of the SF-12, with a correlation between physical and mental health, provides an adequate fit to the data for all country/language groups except Hungary. Further, all the unrestricted multiple-group models provide an adequate fit to the data. In contrast, none of the multiple-group models restricted to common parameters provide an adequate fit to the data. The significance tests confirm that the constraints on parameter values produce significantly different models to the unrestricted models. Conclusions: We conclude that researchers should derive their own country-specific scoring coefficients for physical and mental health summary scores.