[Psychometric properties of the Functional Social Support Questionnaire and the Loneliness Scale in non-institutionalized older adults in Spain].

OBJECTIVES To examine the psychometric properties of the Social Support Questionnaire Duke-UNC (DUFSS) and the De Jong-Gierveld Loneliness Scale in a sample of non-institutionalized older adults. METHODS The sample consisted of 1,106 non-institutionalized older adults included in a national survey on quality of life. Both scales were analyzed according to classical test theory (acceptability, internal consistency, internal validity, convergent validity, discriminant validity and accuracy) and Rasch analysis. RESULTS The mean ± standard deviation scores were 44.95 ± 8.9 for the DUFSS and 1.92 ± 1.83 for the Loneliness Scale. Cronbach's alpha was 0.94 for the DUFSS and 0.77 for the Loneliness Scale. Factor analysis identified two factors in each scale (explained variance: 73.8% for the DUFSS and 67.7% for the Loneliness Scale). The instruments showed a correlation of -0.59 with each other. Rasch analysis of the DUFSS identified two dimensions with a good model fit, whereas the Loneliness Scale did not fit the Rasch model. CONCLUSIONS The DUFSS, with some modifications, meets the Rasch assumptions and provides linear measures. However, more Rasch analysis studies are needed for the Loneliness Scale. According to classical test theory, the DUFSS has good internal consistency for comparisons among people and the Loneliness Scale for comparisons among groups. Both scales have satisfactory construct validity.