Bridging Disciplinary Boundaries (January 11 - 14, 2007)

Purpose: If the variable of social support is to be used in both social work research and practice applications, the measurement models for this concept should be confirmed in other samples to which social work research often applies. To meet this aim, the factor structure of the Medical Outcomes Study's Social Support Survey (Sherbourne & Stewart, 1991) was tested in a sample of mothers whose children were treatment for emotional problems at the time of data collection. Additionally, the items with low factor loadings in the original study were removed to generate both a 12-item and 4-item social support scale in hopes that these shortened scales may encourage greater utilization of this instrument in certain settings. Methods: Data were obtained from participants in studies 2 and 3 (N = 320). Confirmatory factor analysis (CFA) was utilized to confirm the factor structure of the MOS-SSS. For the 18-item and 12-item scales, models tested included both a one-factor model (in which all subscales are correlated at 1.0) and a second-order model with subscales. Chi-squared difference tests were also used to evaluate the “better fit” of models nested in each other. Results: While all models tested were adequate at describing the data, the 12-item higher-order model had the “best fit.” While the subscales remain highly correlated, this model described the data better than a model in which all subscales would be considered correlated to 1.0. Implications: The 12-item MOS-SSS should be considered for utilization by researchers and clinicians who may be interested in the subscales, yet the 4-item could also be utilized as it also maintains structural integrity with adequate internal consistency.