Multilevel confirmatory factor analysis of the Systems of Care Implementation Survey (SOCIS)

Abstract

A major impediment to obtaining national information on systems of care implementation has been the lack of a psychometrically sound large-scale survey instrument. The present study provided information on the factorial and concurrent validity of the Systems of Care Implementation Survey scales. Multilevel confirmatory factor analysis and multilevel regression analysis were used to test these indicators of internal and external validity. Two hundred twenty-five counties were randomly selected and stratified by population size and poverty level. Nine hundred ten informants responded to the survey questionnaire, M = 4.04 informants per county (SD = 3.17). Results indicated that all models had at least adequate fit to the data, with nine of the 14 factor models having excellent fit. Overall, 11 of the 14 factors had some indication that receiving federal funding to create systems of care was associated with higher scores on the factors. Implications for future research were discussed.