China Sample Factor Analysis

CHAPTER 4 RESULTS AND DISCUSSION

4.3 Factor Analysis

4.3.1 China Sample Factor Analysis

Table 4.9 Validity Analysis Results of Each Variable Component

1 2 3 4 5 6 7 8

DE1 0.795 - - - -

DE2 0.688 - - - -

DE3 0.736 - - - -

DE4 0.784 - - - -

DE5 0.725 - - - -

DE6 0.670 - - - -

DE7 0.845 - - - -

RE1 - - - 0.713

RE2 - - - 0.747

RE3 - - - 0.708

RE4 - - - 0.765

TR1 - - - 0.80

TR2 - - - 0.80

TR3 - - - 0.73

TR4 - - - 0.79

EM1 - - - 0.784 - - - -

EM2 - - - 0.796 - - - -

EM3 - - - 0.799 - - - -

EM4 - - - 0.808 - - - -

CN1 - - - - 0.768 - - -

CN2 - - - - 0.775 - - -

CN3 - - - - 0.746 - - -

Component

1 2 3 4 5 6 7 8

CN4 - - - - 0.813 - - -

NA1 - - - 0.850 - -

NA2 - - - 0.871 - -

NA3 - - - 0.825 - -

NA4 - - - 0.901 - -

CR1 - - 0.591 - - - - -

CR2 - - 0.614 - - - - -

CR3 - - 0.621 - - - - -

CR4 - - 0.590 - - - - -

CR5 - - 0.738 - - - - -

CR6 - 0.677 - - - - -

PE1 - 0.647 - - - -

PE2 - 0.604 - - - -

PE3 - 0.682 - - - -

PE4 - 0.633 - - - -

PE5 - 0.605 - - - -

PE6 - 0.771 - - - -

Total 15.13

3.173 2.462 2.157 1.58 1.398 1.13 5

1.100

Cumulative

13.13 1

22.31 2

31.24 6

39.88 2

48.21 1

56.41 2

64.6 72.15 4

KMO 0.946

Bartlett's Test 11811.628 (P = 0.000)

Table 4.10 Common Method Deviation Test

Component Initial Eigenvalues Extraction Sums of Squared Loadings

Rotation Sums of Squared Loadings

Total % of Variance

Cumulative

Total % of Variance

Cumulativ e %

Total % of Variance

Cumulative

1 15.135 38.807 38.807 15.135 38.807 38.807 5.121 13.131 13.131

2 3.173 8.137 46.944 3.173 8.137 46.944 3.580 9.181 22.312

3 2.462 6.312 53.256 2.462 6.312 53.256 3.484 8.934 31.246

4 2.157 5.532 58.787 2.157 5.532 58.787 3.368 8.636 39.882

5 1.580 4.052 62.839 1.580 4.052 62.839 3.248 8.328 48.211

6 1.398 3.584 66.422 1.398 3.584 66.422 3.199 8.202 56.412

7 1.135 2.911 69.333 1.135 2.911 69.333 3.193 8.188 64.600

8 1.100 2.821 72.154 1.100 2.821 72.154 2.946 7.554 72.154

According to the result analysis in Table 4.9 and 4.10, the KMO value is 0.946, significantly higher than the standard 0.70, Bartlett's sphericity test value is 11811.628, and the significance p value is 0.00, so it is suitable for factor analysis.

Using principal component analysis, factors with eigenvalues greater than 1 were extracted, and as a result, a total of 8 common factors were extracted, and the rotational cumulative sum of squares was 72.154%, which was greater than 60%.

After rotating by the orthogonal rotation method, 39 question options could be classified as 8-category factors each item had a loading higher than 0.5, indicating that the extracted 8 factors contained more comprehensive information and did not appear to have high dual factor loading, and each observed variable was aggregated into each dimension according to theoretical presuppositions. Synthesis of the above analyses illustrates that the scales selected in this paper have good construct validity.

The most commonly used Harman single factor method was used to conduct the common method deviation test on the data in this paper, that is to do the exploratory factor analysis with the full scale Title items together, using principal component analysis, to draw the components with eigenvalues greater than 1, the test results are shown in table, a total of eight common factors with eigenvalues greater than 1 were extracted, and the cumulative variance explained rate was 72.154%, and the explanatory amounts of individual factors are all lower than 40%, the absence of a common factor explains most of the variation amount, can illustrate that this study does not have a serious problem of common method bias, can be conducted empirical results analysis.

2) Confirmatory factor analysis

Confirmatory factor analysis is a type of statistical analysis of survey data that tests whether the relationship between a certain factor and the corresponding observed variables conforms to the researcher's prespecified theoretical relationship.

This study uses AMOS 21.0 to analyze the validation factor analysis (CFA) of this scale, and establishes a validation factor model based on the results of validation factor analysis. The suitability of the confirmatory factor model built in this paper was assessed by judging the structural equation fit metrics, which, if the criteria were met, would indicate that the model built in this paper can effectively measure the relevant latent variables. Fitting indexes such as ² df, RMSEA, GFI, AGFI, NFI and CFI were selected to judge the degree of fit between the model and data. Specific criteria were: ² df <3, RMSEA<0.08, GFI, AGFI, NFI and CFI values were higher than 0.8. Standardized factor loading values ranged from 0.5-0.95, with combined reliability (C.R) greater than 0.7. Meanwhile, the convergent validity of each dimension was judged by calculating the average variance extracted (AVE), for which the AVE value healed larger, for which the percentage of variation explained by the latent variable healed larger, and for which the relative measurement error healed smaller (Wu, 2010). The diagram of CFA model in this study is shown in Figure 4.1.

Figure 4.1 Confirmatory Factors Analysis of Model Graph (diagram of CFA)

Table 4.11 Results of Confirmatory Factor Analysis

The Path Estimate S.E. C.R. P CR AVE

DE7 <--- DE 0.844 - - - 0.909 0.589

DE6 <--- DE 0.696 0.048 16.110 ***

DE5 <--- DE 0.758 0.049 18.180 ***

DE4 <--- DE 0.805 0.049 19.867 ***

DE3 <--- DE 0.711 0.049 16.603 ***

DE2 <--- DE 0.699 0.049 16.219 ***

DE1 <--- DE 0.841 0.051 21.280 ***

RE4 <--- RE 0.779 - - - 0.823 0.539

RE3 <--- RE 0.732 0.064 14.532 ***

RE2 <--- RE 0.665 0.062 13.162 ***

RE1 <--- RE 0.756 0.067 15.004 ***

TR4 <--- TR 0.878 - - - 0.889 0.667

TR3 <--- TR 0.782 0.041 19.431 ***

TR2 <--- TR 0.761 0.042 18.632 ***

TR1 <--- TR 0.841 0.043 21.725 ***

EM4 <--- AC 0.888 - - - 0.899 0.690

EM3 <--- AC 0.797 0.041 20.600 ***

EM2 <--- AC 0.768 0.042 19.395 ***

EM1 <--- AC 0.863 0.042 23.552 ***

CN4 <--- CC 0.889 - - - 0.917 0.735

CN3 <--- CC 0.828 0.042 22.666 ***

CN2 <--- CC 0.828 0.039 22.688 ***

CN1 <--- CC 0.882 0.040 25.561 ***

NA4 <--- NC 0.922 - - - 0.899 0.692

NA3 <--- NC 0.733 0.039 18.657 ***

NA2 <--- NC 0.854 0.036 24.369 ***

NA1 <--- NC 0.807 0.040 21.996 ***

CR6 <--- OCWB 0.825 - - - 0.905 0.613

CR5 <--- OCWB 0.789 0.052 18.843 ***

The Path Estimate S.E. C.R. P CR AVE CR4 <--- OCWB 0.752 0.051 17.599 ***

CR3 <--- OCWB 0.786 0.052 18.749 ***

CR2 <--- OCWB 0.750 0.052 17.546 ***

CR1 <--- OCWB 0.792 0.052 18.957 ***

PE6 <--- ICWB 0.858 - - - 0.922 0.663

PE5 <--- ICWB 0.803 0.043 20.637 ***

PE4 <--- ICWB 0.809 0.045 20.879 ***

PE3 <--- ICWB 0.781 0.042 19.738 ***

PE2 <--- ICWB 0.800 0.044 20.533 ***

PE1 <--- ICWB 0.831 0.046 21.862 ***

Note: *P<0.05; **P<0.01; ***P<0.001

According to the data in Table 4.11, the standardized factor load of each item is greater than 0.5, indicating that each item can well explain its dimension.

Combinatorial Reliability (CR) is one of the criteria for judging the internal quality of the model, which reflects whether all the items in each latent variable consistently explain the latent variable. It can be seen from the above table that the combined reliability CR is greater than 0.7, indicating that all test items in each latent variable can consistently explain the latent variable.

Table 4.12 Fitting Index of the Model The

Reference

𝜲^𝟐⁄𝒅𝒇 RMR GFI AGFI NFI TLI CFI RMSEA

Statistics 1.651 0.047 0.886 0.868 0.909 0.958 0.962 0.039 Reference < 3 < 0.05 > 0.8 > 0.8 > 0.9 > 0.9 > 0.9 < 0.08

According to the data in Table.4.12, the 𝛸²⁄𝑑𝑓 = 1.651, GFI = 0.886, AGFI = 0.868 NFI = 0.909, TLI = 0.958. CFI = 0.962, RMSEA = 0. 039. In summary, all indicators of the confirmatory Factor Analysis in this paper have reached the standard, and the overall fitting degree of the model is good.

Table 4.13 Discriminant Validity Analysis

DE RE TR AC CC NC OCWB ICWB

DE 0.767 - - - - - - -

RE .384** 0.734 - - - - - -

TR .446** .441** 0.816 - - - -

AC -.366** -.381** -.306** 0.830 - - - -

CC -.429** -.425** -.410** .557** 0.857 - - -

NC -.124** -.097* -.109* .216** .248** 0.831 - -

OCWB .569** .557** .553** -.572** -.638** -.294** 0.782 - ICWB .587** .590** .583** -.535** -.606** -.247** .742** 0.814

Note: *P<0.05; **P<0.01; ***P<0.001

It can be seen from Table 4.13 that by calculating the ave values of each dimension, they are 0.589, 0.539, 0.667, 0.690, 0.735, 0.692, 0.613 and 0.663 respectively, which are greater than 0.5. From table 4.13, it can be seen that by calculating the correlation matrix of each variable, the square root distribution of ave of each dimension is 0.767, 0.734, 0.816, 0.830, 0.857, 0.831, 0.782 and 0.814, which are greater than the correlation coefficient between each dimension. Therefore, it shows that the scale has good convergent validity and differential validity.

In summary, according to the results of data analysis, the constructive validity, convergence validity and discriminant validity of the scale are good, and the scale can be used as a research tool.

4.3.2 Factor Analysis of Thai Sample

Dalam dokumen PSYCHOLOGICAL CONTRACT VIOLATION, ORGANIZATIONAL COMMITMENT AND EMPLOYEE COUNTERPRODUCTIVE BEHAVIOR AMONG CHINESE AND THAI ENTERPRISES: IN CROSS-CULTURAL CONTEXT (Halaman 123-131)