Content validity was confirmed by specialists (one medical doctor and one psychologist).
As for the surface validity, five non-target high school students were asked to respond to the questions to obtain their opinions on the difficulty of expression of questions and the length of time for responding, and they did not indicate any problems.
The construct validity was studied using principal factor analysis with Varimax rotation.
As a result of the first sample in June, factors with an eigenvalue of 1.0 or higher were found in seven out of 30 items. However, regarding the eigenvalue, the eigenvalue of the first factor was extremely high by 10.24 and the difference of it from the eigenvalue of the second factor, 1.88, was large. For confirmation, factor analysis was performed, assigning factor numbers from 1 to 7. As a result, the best arrangement of factors was the one-factor structure. This was consistent with the one-factor structure estimated from eigenvalues. To examine the sex difference of the factor structure, the factor analyses were performed on the male and female answers separately. The result was that the eigenvalue of the first factor was extremely high and the difference of it from the eigenvalue of the second factor was large in both male and female answers. Therefore, it was judged that this scale was appropriate to be understood by one factor, ‘psychosomatic complaint’. Further, at application of the factor analysis, a factor with low factor loading is judged such that the loading on the factor is low, and removal of item and so on are performed. In this scale, factor loadings were 0.3 or more in all items (Table 1). Additionally, the coefficient of determination of the first factor was 34.1%.
Table 1. Principal factor analysis of the psychosomatic complaints data
| 28. I feel gloomy.||0.78||0.72|
| 26. I lack tenacity.||0.75||0.71|
| 23. I get irritated without any reason.||0.74||0.71|
| 27. I somehow feel sick.||0.74||0.75|
| 11. I get tired easily.||0.72||0.68|
| 25. I get distracted.||0.71||0.72|
| 2. I feel heavy in the head.||0.67||0.70|
| 20. I get in no mood to do anything.||0.66||0.67|
| 5. I feel heavy in the stomach or I do have nausea.||0.63||0.50|
| 19. I feel like lying.||0.63||0.64|
| 9. I have a dizziness when I stand up.||0.62||0.57|
| 21. I feel difficulty in putting my thoughts together.||0.62||0.66|
| 22. I get tired of talking with other people.||0.62||0.56|
| 3. I have a stomachache.||0.60||0.41|
| 30. The relationship between myself and my friends gets bad.||0.58||0.45|
| 1. I have a headache.||0.54||0.56|
| 16. I have palpitation without any reasons.||0.54||0.58|
| 17. I suddenly feel difficulty in breathing.||0.54||0.54|
| 8. I have lumbago.||0.53||0.45|
| 7. I have stiffness in the shoulder.||0.51||0.45|
| 10. I catch a cold easily.||0.51||0.36|
| 12. I have ear ringing.||0.51||0.49|
| 24. I feel like shouting.||0.51||0.56|
| 4. I have diarrhea.||0.50||0.31|
| 14. I have eyestrain.||0.46||0.54|
| 6. I have a poor appetite in the morning.||0.43||0.36|
| 29. I feel I sleep badly.||0.41||0.43|
| 18. I have numbness or trembling in the arms or legs.||0.39||0.49|
| 13. I get eczema.||0.35||0.30|
| 15. I have a tic.||0.34||0.39|
Next, in order to verify if this scale can be applied to understand the health of high school students of the first grade to third grade, a survey using this scale was carried out on three grades of students simultaneously in June 1999. As a result of principal factor analysis, the eigenvalue of the first factor was extremely high by 9.32 and the difference of it from the eigenvalue of the second factor, 2.04, was large. It was therefore found that this scale was appropriate to be understood by one factor. Similarly to the first sample, factor analysis was performed, assigning factor numbers from 1 to 6. As a result, the best arrangement of factors was the one-factor structure. In this scale too, factor loadings were 0.3 or more in all items (Table 1). Additionally, the coefficient of determination of the first factor was 31.1%.
With respect to the concurrent validity, the relationship between the 30 items on psychosomatic complaints and GHQ-28 conducted the second sample, was studied. Using Pearson's product moment correlation, the correlation coefficient between their psychosomatic complaints and the GHQ-28 was γ = 0.65 in June 1998 and γ = 0.64 in March 1999. We examined whether this scale can comprehend a body symptom of a mental cause. The psychosomatic complaints score of the students who frequently utilized the school infirmary was compared with the mean value of other students. The psychosomatic complaints score of two non-target students with eating disorder and the one student who was diagnosed with neurotic disorder who frequently utilized the school infirmary, was compared with the mean value of other students. The total mean of the psychosomatic complaints score was 35.38 ± 16.02; the scores of the two students with the eating disorder was 56 and 62, and the score of the one student with a neurotic disorder was 68.
The Cronbach's α coefficient indicating internal consistency was high: 0.93 in 1997, 0.91 in 1998 and 0.92 in 1999.
The reproducibility was assessed using the first and the second samples. Using the results of the survey in June, the next survey after 4 months and the following survey after the next 8 months, correlations were examined among totals of 30 items of the scale. As a result, correlation between scales was γ = 0.5 or more.