Made available in DSpace on 2011-05-04T12:36:27Z (GMT). No. of bitstreams: 0 Previous issue date: 2009 / Estudos sobre a importância de rede e apoio social nos desfechos em saúde têm sido desenvolvidos nos últimos 30 anos. Rede social é definida como o grupo de pessoas com as quais o indivíduo mantém contato ou alguma forma de vínculo social. Apoio social refere-se à funcionalidade da rede social (apoio emocional, apoio material entre outros). O presente estudo dá continuidade à avaliação das propriedades psicométricas da escala de apoio social utilizada em um estudo de coorte de trabalhadores de uma universidade pública no Rio de Janeiro - Estudo Pró-Saúde por meio da avaliação de sua validade de constructo através da análise fatorial confirmatória. Essa investigação baseia-se nos dados de 4030 participantes obtidos na primeira fase desse estudo (1999), onde um questionário multidimensional foi utilizado. Nesse foi incluída a escala de apoio social elaborada para o Medical Outcomes Study (MOS), que foi adaptada para o português, no âmbito do estudo. Foi realizada a análise fatorial confirmatória com o intuito de avaliar os modelos com três, quatro e cinco fatores quanto ao ajuste, validade convergente e validade discriminante. (...) / Studies about the importance of social support on health outcomes have been developed since 30 years. Social network is defined as a group of people which a person has contact with or any kind of social connections. Social support is related to the functionality of the social network. The present study continues the psychometric testing of the social support scale, used in a cohort of public university servents of Rio de Janeiro - Pró-Saúde Study – through evaluation of the construct validity using the confirmatory factor analysis. This investigation was based on information about 4030 participants in the first stage of the study (1999), which a multidimensional instrument was used. In this study, it was included the social support scale of the Medical Outcomes Study (MOS), which was adapted to Portuguese, in the study‟s environment. The confirmatory factor analysis was used to test the three, four and five factors models adjustment, the convergent and discriminant validity. The estimation method used was the WLSMV (robust weighted least squares) recommended to models with multivariate non-normal distributed indicators which is the case of the categorical indicators of the MOS social support scale. The fit indices used were the CFI (comparative fit index), the TLI (Tucker Lewis index), the RMSEA (root mean square error of approximation), the χ² statistic, and the WRMR ((weighted root mean square residual). The statistic program used was the Mplus. The four factor model was the one which presents the best goodness-of-fit measures (CFI=0,897; TLI=0,987). However, the residuals were high (RMSEA=0,131; WRMR=3,727), indicating a poor adjust. The convergent validity was good, presenting values above 0,50 of extracted variance and values above 0,70 of composed reliability in all dimensions. Only the material dimension presented good discriminant validity. In conclusion, nevertheless the four factor model presented some vantages compared with the other models, it is necessary implement changes to improve the adjust of the model. This modifications are related to itens exclusion, and not only about the dimentions addiction, which is suggested through insufficient results of the discriminant validity of that model - highly correlations in the “information” and “emotional‟ dimensions. A scale with a smaller numbers of itens should improve the measurements of social support and bring news conclusions in the investigations about the importance of social support in the social determinants and wellness.