3.1 Demographic characteristic of participants
As shown in Table 1, 749 students made up the sample. There are 337 males (44.99%) and 412 females (55.01%). As for specialties, 552 (73.70%) students majored in clinical medicine, 49 (6.54%) majored in medical inspection and 148 (19.76%) majored in nursing. Of the total sample, 57.81% came from junior grade (the fifth year to the second year), 36.45% from senior grade (the third year to the fifth year) and 5.74% from the school of postgraduate. 477 (63.68%) students had practice experiences. The mean age was 20.67 years old (standard deviation (SD)=1.88 years old) with a range from 17 to 28 years old. Mean score of the JSE-S in this sample was 81.66 (SD= 7.18).
3.2 Item statistics
The possible scores ranged from 0 to 100, and the actual scores ranged from 49 to 100. The quartile of the total score is 76, 81, 89, and 100, respectively. Item mean scores ranged from a low of 3.58 (SD=0.99) for item 8: “For more effective treatment, physicians must be attentive to their patients’ personal experiences” to a high of 4.43 (SD=0.62) for item 4: “Understanding body language is as important as verbal communication in the physicianpatient relationship”.
3.3 Item-total score correlations
As shown in Table 2, the item-total score correlations ranged from a low of 0.38 (item 1:” Physicians’ understanding of their patients’ feelings and the feelings of their patients’ families is a positive treatment factor”) to a high of 0.76 (item 15: “Empathy is a therapeutic skill without which the physician’s success is limited”). The median item-total score correlation was 0.65. All correlations were positive and significant in statistics (p<0.01), indicat that all items contributed positively to the total score and direction of scoring was true.
3.4 Internal consistency reliability
We calculated Cronbach’s coefficient alpha and the results indicated a satisfactory internal consistency of the scale (Cronbach’s coefficient alpha = 0.89, 95% CI [0.88-0.90]). The Cronbach’s coefficient alpha reliability for items under factor 1 (perspective taking), factor 2 (compassionate care), and factor 3 (“standing on patient’s shoes” replaced by “emotional detachment”) was respectively 0.91 (95% CI [0.89-0.92]), 0.82 (95% CI [0.79-0.84]), and 0.71 (95% CI [0.65-0.76]).
3.5 Exploratory factor analysis
We reaffirmed the underlying components of the JSE-S (Chinese version) by using the exploratory factor analysis. Oblique rotation (promax) was performed to examine if previously reported factor patterns would remain unchanged. We extracted remained 3 factors of the eigenvalue which greater than 1. The Scree test also showed that the plot of the eigenvalues leveled off after extraction of the third factor. The KMO measure was 0.92 and Bartlett’s test was significant at (0.05,190) =3536.02 (p<0.01), supporting the factorability of the data.
The three-component solution was also extracted as shown in Table 2, and it is slightly different from the previous results [2, 4]. The eigenvalues for the first, second and third retained factors were 7.97, 2.05, and 1.20, respectively. The first component, “Perspective Taking”, including eleven items, accounting for 39.82% of the variance. Ten items (out of 10 items under this factor in Hojat’s research [2]) had a coefficient greater than 0.40 on this factor. The item 20 (“I believe that empathy is an important therapeutic factor in medical treatment”) scored the highest factor coefficient (0.95). The second component, “Compassionate Care”, including seven items, accounted for 10.26% of the variance, and seven items (out of eight items under this factor in Hojat’s research [2]) had a coefficient greater than 0.40 on this factor. Item 18 (“It is acceptable for a physician to be touched by intense emotional relationships between patients and their families”) belonged to factor 2 originally now it was proposed to move to factor 3. The item 14 (“Emotion has no place in the treatment of medical illness”) had the largest factor coefficient (0.80), while the item 11 (“Patients’ illnesses can be cured only by medical or surgical treatment, therefore, physicians’ emotional ties with their patients do not have a significant influence in medical or surgical treatment”) had the lowest factor coefficient (0.44) on factor 2. The third component which was entitled as “Standing on Patient’s Shoes” explaining for 6.01% of the variance, two items of whom (out of two items under this factor in Hojat’s research [2]) had a coefficient greater than 0.40 on this factor. In addition, item 18 was proposed to move to factor 3. The factor loadings were 0.85 (item 3 “It is difficult for a physician to view things from patient’ perspectives”), 0.84 (item 6 “Because people are different, it is difficult to see things from patients’ perspectives “) and 0.66 (item 18 “It is acceptable for a physician to be touched by intense emotional relationships between patients and their families”), respectively.
3.6 Confirmatory factor analysis
In the light of the modification indices generated by AMOS program [43], correlations between error variances were permitted. The items contained in each factor in different models are shown in Table 3. Based on our results, we compared the original model with a proposed model by Hojat. The model fit indices for original and proposed modes are listed in Table 4. After SEM was performed, the proposed model showed acceptable model fit indices and reached a satisfactory model fit (c2=411.93, c2/df=2.48, CFI=0.93, IFI=0.93, GFI=0.90, TLI=0.92, RMSEA=0.063). However, the results for original model was c2=492.41, c2/df=2.95, CFI=0.91, IFI=0.91, GFI=0.88, TLI=0.89, RMSEA=0.072, model fit is less good than the proposed model. Standardized path coefficients with the original model and the proposed model were shown in Figure 1. The path coefficient ranged from 0.47 to 0.89 in the proposed model, and all the path coefficients were statistically significant. Both the c2 difference test and the all model fit indices suggest that the proposed model is a better fit than the original model.