Social Skills Rating System (Gresham & Elliott, 1990) is one of the most widely used instruments for assessment and measurement of children’s and youth’s social skills both within research and clinical work. The validity and reliability of the English original of the Social Skills Rating System has been assessed in several different studies. The Norwegian language version is, however, less studied, and to my knowledge there has not been conducted any confirmatory factor analysis or examinations of measurement invariance of this version of the rating scales. The present study investigates factor structure and measurement invariance of teacher ratings using the Social Skills Rating System (SSRS-T) in a sample of Norwegian 4th through 7th graders. The sample used consists of 8016 students from 64 primary schools that participated in an effectiveness trial of the Norwegian School-Wide Positive Behavior Support model.
Four specific questions regarding the SSRS-T were addressed in separate analyses. First, the original three-factor structure proposed by Gresham and Elliot was examined in a confirmatory factor analysis for categorical indicators (CCFA). This factor structure was not replicated in the current sample. Second, a hypothesis about method effects caused by the different indicators’ assessing social skills either in situations involving peers or involving adults was tested in a CCFA. This hypothesis was also rejected. As neither of these models was found to acceptably fit the data, a model re-specification based on the original factor structure was conducted within the CCFA framework. An alternative factor structure that acceptably fit the sample data was found, and the results were replicated in a second sample. Third, measurement invariance of the alternative factor structure between boys and girls was investigated in multiple-groups CCFA. The results indicate that the measurement was invariant across child gender. Finally, the viability of using the re-specified factor structure in combination with the isolated uniqueness approach for parceling indicators was assessed in a confirmatory factor analysis for continuous variables using ML estimation. The results indicated tentative support for this approach.
The findings in this study indicate that there may be problems associated with the use of Gresham and Elliot’s original factor structure in the Norwegian version of the SSRS-T for primary school level. Future studies should closely examine the indicators used in the Norwegian SSRS-T for primary school level and further assess the original factor structure. The alternative factor structure proposed in this thesis shows promising psychometric qualities, including indices of convergent and discriminant validity and measurement invariance across child gender. However, the alternative factor structure should be replicated in other samples and also studied in relation to other operationalizations of social skills and related constructs, before more firm conclusions should be drawn.