Validity of the C-SCIB scale
In terms of expert validity, the I-CVI was 0.99 while the S-CVI as evaluated by four experts with regard to content suitability was 0.96, 1, 1, and 1, respectively; the S-CVI with regard to clarity of text was 1, 0.96, 1, and 1, respectively, with a mean of 0.99. The S-CVI/UA (universal agreement) was obtained by dividing the number of items in which a score of 4 of above was given by each expert with the number of total items (33). The results showed that the S-CVI/UA with regard to content suitability and clarity was 0.97 and 0.97, respectively. The I-CVI (0.99) and S-CVI (0.99) were both above 0.78 and 0.80. The expert validity of the scale is excellent as the experts gave a score of 4 or 5 for most items.
In this study, the QMAALD was used as a criterion and compared with the C-SCIB scale. The results revealed a significant but low correlation between the two scales. The dimensions of the C-SCIB scale, namely internal control, external control, emotional support, and tangible support, had a significant but low correlation with the QMAALD. This shows that the dimensions of the C-SCIB scale are correlated to the QMAALD, and even though the level of correlation is not high, the C-SCIB scale still exhibits some degree of predictive validity for QMAALD scores.
The BANSILQ and LAS were used as criteria and compared with the C-SCIB scale. The results revealed a significant and moderately positive correlation between the BANSILQ and the dimensions of emotional support and tangible support (p<0.01). As for the LAS, it was found to be significantly correlated to the dimensions of internal control and external control (p<0.01), with the correlation coefficients indicating a moderate correlation with internal control, a low correlation with external control, and a moderate correlation with the combined scores for internal and external control. This shows that C-SCIB scale possesses good concurrent validity.
Since the English language version of the scale was not subjected to a criterion-related validity test in previous studies, no comparisons were conducted. However, its subscales were compared, and the results indicated a high degree of correlation among its dimensions, with internal control and external control being moderately correlated (r＝0.55), internal control and support being moderately correlated (r＝0.51), and support and external control being highly correlated (r＝0.69). In our study, the correlations between the subscales were as follows: internal control and external control were found to be weakly correlated (r＝0.36), internal control and support were found to be moderately correlated (r＝0.46), and support and external control were found to be moderately correlated (r＝0.55). Despite the comparatively high degree of correlation among the subscales of the English-language scale, a significant correlation was still achieved in our study, indicating a link between control and support.
In this study, a confirmatory factor analysis was performed to examine the construct validity of the SCIB scale when applied to postpartum women in Taiwan. An item analysis was first carried out before conducting the abovementioned analysis and five items were found to have not met the verification standards, namely 1. The labor pains were too much for me to gain control; 6. I felt my body was experiencing an event that was out of my control; 17. I was unable to control the staff entering and leaving the delivery room; 29. I felt the medical staff were only doing things for their own convenience; and 33. The medical staff ignored the things I said. The confirmatory factor analysis was eventually conducted with three factors, 33 items, and a sample size of 228 participants, with the aim of verifying the overall model's goodness of fit. Although the results indicated a poor overall goodness of fit, they met the standards for parsimony goodness of fit. The researchers removed five items, reducing the total number of items from 33 to 28, and conducted another round of confirmatory analysis. Although the results were better compared to those for the initial 33 items, they still failed the standards for overall goodness of fit while meeting those for parsimony goodness of fit. Given that the confirmatory analysis results did not improve after several items were removed, the researchers decided to retain the original 33-item scale. In the original study, an exploratory factor analysis was performed, revealing factor loadings that were higher than 0.47 and a total variance explained of 55.8%. The original study primarily utilized statistical methods to identify common attributes shared by numerous variables, with the aim of establishing new hypotheses or developing new theoretical frameworks. In contrast, an exploratory factor analysis was not performed in our study, which only utilized confirmatory factor analysis to test and validate the scale model.
The above results indicate that the C-SCIB scale had good criterion-related validity and was able to effectively and quickly assess the support- and control-related conditions of women during childbirth, such that their birth experiences could be predicted. Despite the poor overall goodness of fit for construct validity, the results still met the standards required for parsimony goodness of fit and, thus, possess value as a reference.
The original English-language SCIB scale had an overall Cronbach’α coefficient of 0.95, and a coefficient of 0.86, 0.93, and 0.93 for internal control, external control, and support, respectively. In our study, the 33-item C-SCIB scale had an overall Cronbach’α coefficient of 0.81; a coefficient of 0.75, 0.71, and 0.88 for internal control, external control, and support, respectively; and a reliability in excess of 0.07. Even though these figures are lower compared to those in the original study, they nevertheless indicate a high degree of reliability. Furthermore, the scale's test-retest reliability (tested using the Pearson product-moment correlation coefficient) was found to be 0.96 for the overall scale, and 0.76, 0.78, and 0.99 for internal control, external control, and support, respectively, indicating the scale's stability. Since the English-language SCIB scale's test-retest reliability was not tested, no comparisons between the two scales were performed. However, the C-SCIB scale's reliability was higher than 0.70, which indicates that it has good test-retest reliability and stability.
In summary, the C-SCIB scale was found to have good reliability and stable test-retest reliability, and be correlated to the validity scales with respect to predictive and concurrent validity. Although the confirmatory factor analysis showed that overall goodness of fit was only for construct validity, its results still met the standards for parsimony goodness of fit. Like the original English-language scale, the C-SCIB scale was shown to have met the statistical standards for reliability and validity.