[Development of a perceived social support scale: for a Japanese population]

Social support is increasingly attracting interest in as a factor in epidemiological research of disease risks. However, at present, few methods for measuring social support that can be credited as having reliability and validity exist in Japan. Therefore a perceived social support scale was developed in relation to a population study of a town in Fukuoka prefecture. A method for measuring functional support from several external sources was designed consisting of a self-administered test provided with only two available response options--yes & no. Pretests were conducted twice in order to evaluate content validity and adjust the number of determinants and resulted in spouse, family members, and friends being selected to be primary sources of support. A factor analysis was applied to select 10 conditions with high internal consistency for each of the sources. Responses were scored 0 or 1, for 'no' or 'yes,' respectively. Subsequently each set of 10 variables was summed into an index ranking social support conditions for each of the three sources. The three highest ranking indicators from each of the respective sources were then integrated into a single comprehensive total social support score of the individuals. The inquiry method was applied on 277 adults, aged from 40 to 69 years, in the town. Intercorrelation between the indicators of support derived from the three sources was low and coefficients of reliability ranged from 0.892 to 0.978. The score of each set of the 10 variables were incorporated into a principal components analysis in order to determine factorial validity. All of the first factor loadings of the items measuring support from both spouse and family were greater than 0.7 and high eigenvalues were obtained. Therefore the methods applied to spouse and family were considered adequate to measure a given construct. The items measuring support from friends were divided into two factors, emotional support and tangible support.