Purpose To compare the relationship of the eight SF-36 v1 subscale scores to the summary scores of the PCS and MCS derived from two different scoring algorithms: one based on the original scoring method (Ware, Kosinski and Keller, SF-36 physical and mental health summary scales: a users manual. The Health Institute, New England Medical Centre, Boston, MA, 1994); and the other based on scoring algorithms that use parameters derived from structural equation modelling. Further, to provide SF-12 scoring algorithms similarly based on structural equation modelling. Methods The Australian Bureau of Statistics 1995 Australian National Health Survey dataset was used as the basis for the production of coefficients. There were 18,141 observations with no missing data for all eight SF-36 subscales following imputation of data items, and 17,479 observations with no missing data for the SF-12 data items. Data were analysed in LISREL V8.71. Structural equation models were fit to the data in confirmatory factor analyses producing weighted least squares estimates, which overcame anomalies found in the traditional orthogonal scoring methods. Results Models with acceptable fits to the hypothesised factor structure were produced, generating factor score weighting coefficients for use with the SF-36 and SF-12 data items, to produce PCS and MCS summary scores consistent with their underlying subscale scores. Conclusions The coefficients generated will score the SF-36 summary PCS and MCS in a manner consistent with their subscales. Previous Australian studies using version 1 of SF-36 or SF-12 can re-score their summary scores using these coefficients.
- SF-36 summary scores
- Structural Equation Model