3.1 Socio-demographic characteristics of the participants
As to gender, male accounted for 14.37% and female accounted for 85.63%. The mean age was 21.97 years (SD = 1.54) ranging from 19 to 29 years. As to the educational background, most of them (51.37%) were university graduates, followed by those having completed junior college (39.76%), and graduates with master’s degrees or above (8.87%), respectively. There were vast majority of participants (79.20%) could choose nursing at first. The average score of each dimension ranged from 4.87 to 9.97, and the average score on the TSSNGN was 44.75 (SD = 8.23) (See Table 1).
Table 1
Socio-demographic characteristics of participants (n=327)
Characteristics
|
|
n(%)
|
Mean (SD)
|
Gender
|
Male
|
47(14.37)
|
|
|
Female
|
280(85.63)
|
|
Age (in year)
|
|
|
21.97(1.54)
|
Education level
|
Junior college
|
130(39.76)
|
|
|
University
|
168(51.37)
|
|
|
Master’s degree or above
|
29(8.87)
|
|
If you have any brothers or sisters
|
Yes
|
116(35.47)
|
|
|
No
|
211(64.53)
|
|
Marital status
|
Single
|
322(98.47)
|
|
|
Married
|
5(1.53)
|
|
Was nursing your first choice?
|
Yes
|
259(79.20)
|
|
|
No
|
68(20.80)
|
|
service relations
|
staffing of public institution
|
32(9.79)
|
|
|
Contract
|
295(90.21)
|
|
Score of Y1
|
|
|
7.37(1.65)
|
Score of Y2
|
|
|
9.91(1.85)
|
Score of Y3
|
|
|
5.01(1.6)
|
Score of Y4
|
|
|
7.61(1.82)
|
Score of Y5
|
|
|
9.97(2.47)
|
Score of Y6
|
|
|
4.87(1.31)
|
Score of TSSNGN
|
|
|
44.75(8.23)
|
3.2 Validity and reliability of the instrument
3.2.1 Content validity
The content validity testing expert panel consisted of six people. In summary, the score of I-CVI (Item Content Validity Index) was 0.83-1.0 and the S-CVI (Scale Content Validity Index) was 0.92. It met with Lynn’s [19] criteria (a minimum I-CVI of .78 for 6 to 10 experts) and it had an S-CVI/Ave of 0.90 or higher and had good content validity. Thus, all 18 items were selected in the final questionnaire.
3.2.2 Construct validity
For all 18 items in the preliminary Korean version, six factors were used. Confirmatory factor analysis (CFA) was performed on the TSSNGN items. The Chinese TSSNGN followed the division established by the original version. The parameter estimates of the CFA of the Chinese version of the TSSNGN are shown in Figure 1. Examination of normalized factor loadings for 18 items was determined to be 0.51-0.85. The entire standardized factor loading was statistically significant and all the item was more than the minimum standard of 0.50. All the items loaded significantly onto their respective factors. The results showed that the non-standardized value of C. R was ranged 8.10 to 18.79, which satisfied the analytical condition above 1.96 (see Table 2).
CFA was done on these items. The model fit indexes were as follows: The c2 value of the 18 items in Model 1 was 463.57 (df = 120, p < 0.001), indicating that the p value was less than 0.05. Since the c2 value increases with the sample size, the correct c2 value can only be obtained if the sample size is appropriate. Because in this study, the confirmation of the model fit by c2/df was 3.86, which failed to meet the acceptability criterion of 3 or less. Meanwhile, the model fit indexes were CFI = 0.88, GFI = 0.85, IFI = 0.89, TLI = 0.85, RMR = 0.05, and RMSEA = 0.10, which did not meet the standard and should be adjusted. Therefore, Model 4 was constructed after three adjustments by applying the Modified Index (MI). The fitness analysis of model 4 showed that c2/df =3.09, RMSEA =0.07, RMR =0.03, GFI =0.90, IFI =0.92, TLI =0.90, and CFI =0.92, which was deemed relatively good by improving all fit of the model (see Table 3).
Table 2
Analysis of convergent validity of items
|
Items No.
|
B
|
SE
|
β
|
C.R.
|
P
|
CR
|
AVE
|
Cronbach’s
|
Y1
|
1
|
1.00
|
-
|
0.63
|
-
|
<0.001*
|
0.71
|
0.45
|
0.79
|
2
|
1.25
|
0.15
|
0.80
|
8.26
|
3
|
1.29
|
0.16
|
0.57
|
8.25
|
Y2
|
4
|
1.00
|
-
|
0.64
|
-
|
<0.001*
|
0.69
|
0.37
|
0.64
|
5
|
0.73
|
0.09
|
0.51
|
8.10
|
6
|
0.77
|
0.09
|
0.54
|
8.29
|
7
|
0.76
|
0.09
|
0.57
|
8.40
|
Y3
|
8
|
1.00
|
-
|
0.73
|
-
|
<0.001*
|
0.70
|
0.50
|
0.67
|
9
|
0.99
|
0.08
|
0.69
|
12.02
|
Y4
|
10
|
1.00
|
-
|
0.85
|
-
|
<0.001*
|
0.79
|
0.57
|
0.79
|
11
|
0.92
|
0.07
|
0.74
|
13.95
|
12
|
0.82
|
0.06
|
0.65
|
12.97
|
Y5
|
13
|
1.00
|
-
|
0.83
|
-
|
<0.001*
|
0.89
|
0.66
|
0.89
|
14
|
0.88
|
0.06
|
0.77
|
15.40
|
15
|
1.05
|
0.06
|
0.83
|
18.79
|
16
|
1.05
|
0.06
|
0.81
|
18.07
|
Y6
|
17
|
1.00
|
-
|
0.78
|
-
|
<0.001*
|
0.77
|
0.62
|
0.77
|
18
|
1.04
|
0.07
|
0.80
|
14.83
|
*p<0.05.
Table 3
Analysis of construct validity
|
χ2(p)
|
df
|
CIMIN/df
|
GFI
|
IFI
|
TLI
|
CFI
|
RMSEA
|
RMR
|
Model 1
|
463.57*
|
120
|
3.86
|
0.85
|
0.89
|
0.85
|
0.88
|
0.10
|
0.05
|
Model 2
|
416.32*
|
119
|
3.50
|
0.86
|
0.91
|
0.88
|
0.90
|
0.09
|
0.04
|
Model 3
|
394.01*
|
118
|
3.34
|
0.87
|
0.91
|
0.89
|
0.91
|
0.08
|
0.04
|
Model 4
|
361.50*
|
117
|
3.09
|
0.90
|
0.92
|
0.90
|
0.92
|
0.07
|
0.03
|
*p<0.05.
3.2.3 Convergent validity
According to Fornell and Larcker [22], there were three methods to test the convergent validity: the standardized factor loading, AVE and CR [23] considered that the standardized factor loading of each item should not be less than 0.50, or the factor load of each item was statistically significant. And then the scale could be considered to have convergent validity if either of the two conditions was met. The standardized factor loadings for the 18 items ranged from 0.51 to 0.85, with each item greater than 0.50. Therefore, it was confirmed that the standardized factor loadings in this study met the criteria for all items.
The AVE was 0.37~0.66 in Table 2. Generally speaking, AVE was higher than 0.5, but 0.4 is acceptable. Because it was said that if AVE was less than 0.5, but the composite reliability is greater than 0.6, the convergent validity of the construct was still sufficient according to Lam [24]. CR should be 0.70 or higher [25]. The CR was 0.69~0.89 in this study, meeting the minimum criteria. Therefore, since the TSSNGN met the minimum criteria for standardized factor loading, AVE, and CR, the convergent validity of the items measuring the transition shock experience of newly graduated Chinese nurses was confirmed (see Table 2).
3.2.4 Discriminant validity
To verify the discriminant validity of the TSSNGN, firstly, the six latent variables constituting the concept should be significant; secondly, the correlation coefficient should be less than 0.50; finally, the correlation coefficient should be less than the square root of AVE. In this study, the six factors were significantly correlated (p<0.05). The AVE of Y1 to Y6 was 0.45, 0.37, 0.50, 0.57, 0.66 and 0.62, respectively (see Table 2). The AVE square root of Y1 to Y6 was 0.67, 0.56, 0.71, 0.75, 0.81 and 0.79, Correspondingly (see Table 4), which all correlation coefficients were less than the AVE square root, indicating that each latent variable had a certain correlation with each other and a certain degree of differentiation, thereby confirming discriminant validity in this scale (see Table 2 and 4).
Table 4
Analysis of criterion-related validity
|
Y1
|
Y2
|
Y3
|
Y4
|
Y5
|
Y6
|
Y1
|
1
|
|
|
|
|
|
Y2
|
0.10*
|
1
|
|
|
|
|
Y3
|
0.08*
|
0.16*
|
1
|
|
|
|
Y4
|
0.13*
|
0.18*
|
0.28*
|
1
|
|
|
Y5
|
0.10*
|
0.17*
|
0.26*
|
0.30*
|
1
|
|
Y6
|
0.10*
|
0.16*
|
0.25*
|
0.27*
|
0.32*
|
1
|
Square root of AVE
|
0.67
|
0.56
|
0.71
|
0.75
|
0.81
|
0.79
|
*p<0.05.
3.2.5 criterion-related validity
Stress perception has a certain positive effect on the pressure caused by people entering a new environment. To test the criterion-related validity, we analyzed the correlation between the Chinese version and perceived stress. Generally, a correlation of 0.40-0.80 is recommended when testing the criterion-related validity. Analysis showed that the scale of TSSNGN and PSS had statistically significant positive (+) correlation (r=0.97, p<0.01). In addition, the scale of TSSNGN showed significant positive (+) correlations, specifically PSS-1 (r=0.88, p<0.01), PSS-2 (r=0.93, p<0.01). Moreover, the correlation coefficients of all factors in the scale ranged from 0.43 to 0.93. Therefore, it could confirm the criterion-related validity of the Chinese version of the PSS (See Table 5).
Table 5
Analysis of criterion-related validity of the Chinese TSSNGN and PSS
Scales
|
Y1
|
Y2
|
Y3
|
Y4
|
Y5
|
Y6
|
Total of TSSNGN
|
PSS-1
|
PSS-2
|
Total of PSS
|
Y1
|
1
|
|
|
|
|
|
|
|
|
|
Y2
|
0.44*
|
1
|
|
|
|
|
|
|
|
|
Y3
|
0.28*
|
0.59*
|
1
|
|
|
|
|
|
|
|
Y4
|
0.39*
|
0.59*
|
0.70*
|
1
|
|
|
|
|
|
|
Y5
|
0.31*
|
0.61*
|
0.67*
|
0.69*
|
1
|
|
|
|
|
|
Y6
|
0.31*
|
0.57*
|
0.67*
|
0.63*
|
0.79*
|
1
|
|
|
|
|
Total of TSSNGN
|
0.57*
|
0.61*
|
0.63*
|
0.85*
|
0.88*
|
0.82*
|
1
|
|
|
|
PSS-1(The negativity subscale)
|
0.80*
|
0.66*
|
0.65*
|
0.75*
|
0.69*
|
0.63*
|
0.88*
|
1
|
|
|
PSS-2(The positivity subscale)
|
0.43*
|
0.87*
|
0.79*
|
0.81*
|
0.79*
|
0.72*
|
0.93*
|
0.78*
|
1
|
|
Total of PSS
|
0.62*
|
0.83*
|
0.77*
|
0.83*
|
0.79*
|
0.72*
|
0.97*
|
0.96*
|
0.92*
|
1
|
*p<0.01.
3.2.6 Reliability
The Cronbach's alpha of the six factors in the 18-item Chinese version were 0.79, 0.64, 0.67, 0.79, 0.89 and 0.77, respectively. The total scale was 0.92 (see Table 2)