Objective: To evaluate the reliability of self-reported sexual behaviors by a self-admin-istered questionnaire designed to evaluate the effectiveness of HIV prevention education.
Methods: Reliability was assessed by a test-retest study design in 176 female students of a medical technology college and a nurse's school. The questionnaire was anonymous and consisted of 32 main and 20 branching questions categorized into 8 sections including HIV/sexually transmitted diseases (STD)-related knowledge and sexual behaviors. Subjects were asked to respond twice to the same questionnaire with a one-week interval. Informed consent was obtained for the purpose and the necessity of the study and subjects were told that participation was voluntary and no disadvantage would be caused by declining or withdrawing from the study. Reliability was statistically evaluated by means of Pearson's correlation coefficients, concordance rates, kappa coefficients and weighted kappa coefficients.
Results: The overall participation rate was over 90% with almost complete linkages between test and retest responses. While the mean correlation coefficient was 0.83 for HIV/STD-related knowledge, kappa coefficients were 0.97 for sexual experience and 0.98 for the age of initial sexual intercourse, suggesting that self-reported sexual behaviors were reliable to an extent similar to that demonstrated in previous studies abroad. Subjects dropped out of the study were minimal and unlikely to have influenced the results because no bias was detected in sexual behaviors. Though the present results may not be directly applicable to other populations because the subjects were all of medical backgrounds and the interval between the tests was only one week, they provided the first evidence of the reliability for a sexual behavioral survey using self-administered questionnaires in Japan.
Conclusion: Present results suggested that sexual behaviors of Japanese youth could be investigated with sufficient reliability using self-administered questionnaires.
View full abstract