This study has been held to measure the subjective health status of employees ar governmental research instituter through the world-widely used SF-36 health survey, and the need of rest of the validity of SF-36 using statistical tests.
In order to test hypotheses regarding the differentials of health status of employees at government research institutes, and the hypothesized structure of SF-36 by exploratory factor analysis and structural equation model, 2,511 workers from 46 governmental research institutes were randomly selected to answer a self-rated questionnaire regarding their self-perceived health status.
The results are as the following :
First, as the rest of the SF-36´s scaling assumption, the distribution of responses to items within the same scale and item variances are roughly equal, thus proving the equal variance hypothesis. Excluding all the items in SF, the item internal consistency showed results larger than 0.4. Also, due to numerous items in 3 scales, we could see that the item discriminant validity had somewhat dropped. but concluded that if did not largely violate the scaling assumption on the whole.
Second, in order co test the hypothesized structure of SF-36(given model), a factor analysis was held with 35 items, which we excluded one item from the usual 36. However, based on the given model, the 8 factors were nor extractable. Therefore in this experiment, the 8 factors were fixed in order to test the given model through the SF-36. Using the maximum likelihood method and analyzing the data, the extracted 8 factors did not fully march the items of the 8 scaler of the given model. Also, from the results of analysis of the 8 scaler, 2 factors had eigenvalues larger than 1, and therefore indicated that the SF-36 was structured co represent two major dimension of health. The PF, RP and RE showed high results in the first factor, Physical Health, compared to the Mental Health factor. In comparison, the GH, VT and MH were relatively higher in the Mental Health factor. However the SF was loaded similarly in bath Physical Health and Mental Health factors.
Third, the SF-36 did nor sufficiently qualify as a ´good rnodel´ in the structural equation model. The structural equation model, made of 35 items, is divided into 8 first order factors, and once again divided into two second order factors, Physical Health and Mental Health. Thereupon we selected the most adequate alternative model with multiple possibilities. In the selected alternative model, 35 items of the SF-36 were divided in 8 scaler. The physical functioning 2 scale, the physical functioning 2 scale, the bodily pain scale, the vitality and mental health I scale, the vitality and mental health 2 scale, the role-physical limitation scale, the role-emotional limitation scale, and the general health scale. The first of the two second order factors, the Physical Health factor included physical functioning 1 and 2 scale, the role-physical limitation scale and the role-emotional limitation scale. The Mental Health factor included the vitality and mental health 1 and 2 scale, and the general health scale. The bodily pain scale was set to influence both second order factors. Just as the verification results of the alternative model did nor fully satisfy the fitness index as a ´good model´, however the alternative model did show a higher fitness index.
|