143
Hội thảo Khoa học Quốc tế
...
4. RESULTS AND DISCUSSION
4.1. Results
The concepts scale was preliminarily assessed by the Cronbach Alpha reliability and the EFA
analysis. It is then verified through aggregation reliability, convergence value, and discriminant
value using Confirmation Factor Analysis (CFA). SEM (Structural Equation Modeling) is used to
test theoretical models and hypotheses. The estimation method is ML (Maximum likelihood).
The proposed initial scale has 04 single concepts: Entrepreneurial perceived desirability,
Entrepreneurial perceived feasibility, Entrepreneurial Implemental mindsets, Entrepreneurial
implementation intentions. The results of the pre-test and confirm test showed that 05
observation variables were eliminated: 02 Entrepreneurial implementation intentions variable,
02 Entrepreneurial Implemental mindsets variable and 01 Entrepreneurial perceived feasibility
variable. These observation variables were excluded in the preliminary analysis because of the total
variance coefficient <0.4, factor loading<0.5 and not convergent. The results of the scale test are
presented in Table 3.
Table 3 Summary table of measurement results
Concept
Number of
observation
variables
Reliability
Average
Variance
Extracted
Cronbach
Alpha
Composite
Reliability
Entrepreneurial
implementation intentions
(INT)
03
0,777
0.778
0,540
Entrepreneurial Implemental
mindsets (IMP)
03
0,785
0,787
0,552
Entrepreneurial
perceived
desirability (DES)
06
0,871
0,872
0.534
Entrepreneurial
Perceived
feasibility (FEA)
04
0,852
0,854
0,600
Source: data processing from the authors
The results of the CFA analysis with the critical model show that the model has chi-squared
statistical data of 221.280 with 98 degrees of freedom, p = 0.000. If the degree of freedom is
CMIN / df = 2,258 <3, satisfies the compatibility requirement. Different indicators such as GFI =
0.941, TLI = 0.950, CFI = 0.959, RMSEA = 0.054 <0.80, were satisfactory. Therefore, it can be
concluded: Critical models achieve compatibility with market data.
The SEM results (Fig. 2) showed that the model had a chi-squared statistic of 260,118 with 99
degrees of freedom, p = 0.000. If the degree of freedom is CMIN / df = 2.627 <3.0, satisfactory
compatibility. Other indicators such as GFI = 0.932, TLI = 0.935, CFI = 0.946, RMSEA = 0.061
<0.80, were satisfactory. The Heywood phenomena (negative variance of negative error) did not
appear in the estimation of CFA, SEM and most standardized residuals were less than |2.28|.
Therefore, it can be concluded, this model is appropriate for market data.
Estimates of the major parameters presented in Table 4 of all relationships were statistically
