Validity testing
Both the Kaiser-Meyer-Olkin (KMO) and Bartlett Sphericity tests were performed to determine if the data collected by the questionnaire were suitable for the factor analysis. The test results showed that the KMO values of the Activity subscale and Participation subscale were .914 and .893, respectively; KMO values > .5 indicated that there were common factors in the questionnaire items, i.e., the factors were independent. The results of Bartlett’s chi-square test were all statistically significant (p < .001), indicating that the factors were independent and exclusive. Therefore, the questionnaire data collected in this study were suitable for factor analysis.
EFA was used to extract the construct of the scale. The Activity subscale had 3 factors with an eigenvalue > 1, explaining 75.176% of the total variance in the scale. However, according to the scree plot, the curve flattened after the fourth factor. Therefore, based on the standard and scree plots for the factors with eigenvalues > 1.0 and combined with clinical experience and the factorial structure of the raw scores,30-33,36 we postulated that the Activity subscale of the Chinese version of the PaArticular Scales had 3 main factors: lower-limb activity, upper-limb activity, and self-care activity. Table 3 shows the factor structure after rotation. A factor with an eigenvalue > 1 was obtained from the Participation subscale, explaining 62.83% of the total variance in the scale. However, according to the scree plot, the curve flattens after the second factor. The Participation subscale of the Chinese version of the PaArticular Scales had only 1 major factor: participation. Tables 3 and 4 show the factor structure after rotation of the Activity and Participation subscales.
For content validity, correlation coefficients for factors 1, 2, and 3 and the item-to-subscale were obtained; the ranges for these coefficients were .725-.888, 706-.886, and .622-.853, respectively, and the Cronbach’s α was .958, .951, and .910, respectively.
For the criterion-related validity and according to the classification proposed by Cohen,37 Pearson’s product-moment correlation revealed that the correlation coefficient (r ) between the Chinese version of the PaArticular Scales and the WHODAS 2.0-36 items was .770, which was interpreted as a large coefficient, with ap -value less than .001, indicating a highly significant result. The correlation coefficient (r ) between the Chinese version of the PaArticular Scales and the WHOQOL-BREF was -.553, which was interpreted as a large coefficient, with a p -value less than .001, indicating a highly significant result. The correlation coefficients between the Activity subscale and the WHODAS 2.0 ̵̵36 items and WHOQOL-BREF were .722 and -.502, respectively, the correlation coefficients between the Participation subscale and the WHODAS 2.0 ̵̵36 items and WHOQOL-BREF were .742 and -.580, respectively; all the correlation coefficients were highly significant (Table 5).