著者

肥田野 直

出版者

東京大学教育学部

雑誌

東京大学教育学部紀要 (ISSN:04957849)

巻号頁・発行日

no.18, pp.p49-82, 1978

---

Department of Educational Psychology Relationships between social desirability (SD) scale values based on different assessment procedures were studied as follows : 1) The social desirability of 526 items from the MMPI Tokyo University Revised Edition were rated on the Heineman 5-point scale by subjects consisting of 100 male and 125 female university students. They were first asked to rate the degree to which the yes (true) response to each item was favorable on the 5-point scale. The most favorable rating was scored 1 and least favorable scored 5. Then subjects were asked to rate the no (false) response to the same item on the same scale. The mean scores of each subjects group was used as the S-D scale value for each item. It was found that the S-D scale value of "yes" response and that of "no" response were almost complementary. Descrepancies between the scale values of the "yes" response and those of the "no" response were found with some items. Most of them had high S-D scale values on the "yes" response, but did not have very low S-D scale values on the "no" response; that is, they were judged as desirable on both "yes" and "no" response conditions. On the other hand, a few items which had low S-D scale values on the "yes" response were found not to have very high scale value on "no" response. Less descrepancies were found in female subjects than in male subjects. 2) The subjects consisted of 195 male and 67 female university students who were asked to rate the social desirability of 500 items from TPI on the Edwards 9-point scale. According to this scale the most desirable rating was given 9 points. Correlations between two sets of the S-D scale value on the "yes" response to 263 items which were common to both the MMPI Tokyo University Revised Edition and the TPI were sought : Correlation coefficients were found to be -.947 for male and -.938 for female subjects respectively. The negative efficients were due to the rating scales which had reversed directions. Those high correlation coefficients suggested the stability of the S-D scale values of the same items put in different personality inventories. 3) The S-D scale values of items from the TPI obtained from 262 subjects, males and females combined, were analysed by means of principal component factor analysis. The highest factor loading found in the correlation matrix between subjects was.936, and the next highest was.017. The fact that there was only one dominant factor suggested that the subjects were homogeneous in terms of social desirability rating. However, among 10 factors extracted from correlation matrix between items the highest loading was only.262, the rest of the loadings were much lower. This indicates that the subjects' criteria for desirability rating varied from item to item. 4) Percentages of "yes" response (probability of endorsement) and percentages to "no" response were obtained in normal adult sujbects (300 male and 300 female) with items from MMPT Tokyo University Revised Edition administered under regular instruction. Correlations between those percentages and the S-D scale values were obtained. Coefficients were higher with "no" response than with "yes" response in both male and female subjects; that is, -.668 with "yes" response and -.702 with "no" response in males and -.594 with "yes" response and -.625 with "no" response in females. When subjects were asked to fake themselves desirable, the correlation became much higher ; coefficient was -.517 with regular instruction and -.946 with faking instruction in "yes" responses, and -.536 with regular instruction and -.940 with faking instruction in "no" responses. In this experiment 125 female students were administered the MMPI Tokyo University Revised Edition two times with 2 weeks interval. In the first time the inventory was administered with regular instruction and in the next time it was given with the instruction to fake.