Data Analysis and Results

Một phần của tài liệu Kỷ yếu hội thảo quốc tế dành cho các nhà khoa học trẻ khối trường kinh tế và kinh doanh năm 2021 (Volume 4) (Trang 436 - 440)

FINANCIAL TECHNOLOGY AND OTHER RELATING ISSUES

4. Data Analysis and Results

All of the constructs in this study were examined in terms of reliability, convergent validity, discriminant validity, and Latent variable interaction via SPSS 22 and AMOS 21.

Reliability was evaluated with the composite reliability values. Convergent validity shows the extent to that the measure of a construct. Convergent validity can be evaluated using three criteria suggested by Fornell et al. (1981): all indicator factor loadings should be significant

and exceed 0.70, construct reliabilities should exceed 0.80, and average variance extracted (AVE) by each construct should exceed the variance due to measurement errors for that construct. AVE should exceed 0.5 (Fornell et al., 1981). Principal Components Analysis with Varimax rotation was first conducted to extract five factors. The latent variable interaction was used to test the effect of the Moderator in the research. The results show that all items fit their respective factors quite well.

4.1 Demographic Analysis

Table 2 illustrates the characteristics of respondents. Most of the responses were generated by males, which occupied 63.05% of our sample size, whereas females accounted for around 36.95%. Regarding age, respondents under 35 years old dominated (62.07%). The respondents are mainly distributed under 10 million per month (76.60%) in terms of income.

In addition, most respondents are a student (31.287%) and office worker (30.79%).

Table 2. Sample Demographics

Demographic Category Count % Demographic Category Count %

Gender Male 256 63.05

Age

Under 25 92 22.66

Female 150 36.95 26-35 160 39.41

Career

Student 127 31.28 36-45 79 19.46

Office worker 125 30.79 Over 45 75 18.47

Self-employed 73 17.98

Monthly Income (Million VND)

Under 5 166 40.89

Housewife 61 15.02 5-10 145 35.71

Others 20 4.93 10-15 47 11.58

Over 15 48 11.82 4.2. Measurement Model Assessment

4.2.1. Reliability and Validity

Table 3. Results of Cronbach’s Alpha Check

Construct name Variable Cronbach’s α Construct name Variable Cronbach’s α Low Satisfaction at

Cash Payment LS 0.894 Inconvenience of Cash

Payment IC 0.855

Attractive Alternatives AL 0.822 Mobility MB 0.899

Mobile Payment

Knowledge KN 0.796 Personal Innovativeness IN 0.913

Perceived Risk RI 0.837 Switching Cost CO 0.953

Habit HA 0.913 Perceived Usefulness PU 0.859

Perceived Ease of Use PE 0.866 Intention to Switch IS 0.873 Cronbach's alpha was applied to check the internal consistency and reliability of the items. Cronbach's Alpha test results (Table 3) show the value in all cases (0.796 ~ 0.913) over 0.7, implying that the data are reliable. The Exploratory Factor Analysis (EFA) was

performed, which divided factors into 12 components. All items were well loaded with factor loading of more than 0.5.

4.2.2 Measurement Model Assessment

Confirmatory Factor Analysis (CFA) was assessed to check the reliability and construct validity. As the CFA results, the seven model-fit measures were satisfactory which is good evidence for the validity of the model (χ2 = 871.752; df = 584; χ2/df = 1.493; CFI = 0.967; NFI = 0.908; IFI = 0.968; RMSEA = 0.035). According to Table 3, the average variance extracted (AVE) for all cases (0.501~ 0.785) were higher than 0.5 and all CR value also exceeded 0.7 (0.801~0.916). These values showed strong evidence of convergent validity (Fornell et al., 1981). To test the discriminant validity of the constructs, we compared the square root AVE of each construct with the correlation coefficients. The correlation matrix in Table 4 illustrates that the highest value of correlation coefficient (0.644) is smaller than the lowest values of square root AVE (0.708) indicating the evidence of the discriminant validity (Fornell et al., 1981).

Table 4. Convergent Validity and Correlation Matrix of Latent Constructs

CR AVE AL MB LS PE IN IC KN PU IS RI

AL 0.824 0.610 0.781

MB 0.900 0.693 0.426 0.832

LS 0.896 0.682 0.452 0.506 0.826

PE 0.873 0.633 0.524 0.502 0.541 0.796

IN 0.916 0.785 0.161 0.228 0.250 0.333 0.886 IC 0.857 0.599 0.573 0.593 0.554 0.562 0.185 0.774 KN 0.801 0.501 0.254 0.357 0.463 0.355 0.178 0.387 0.708 PU 0.861 0.609 0.601 0.567 0.587 0.644 0.310 0.600 0.403 0.780 IS 0.875 0.637 0.577 0.643 0.558 0.546 0.195 0.665 0.490 0.632 0.798 RI 0.838 0.633 -0.186 -0.217 -0.194 -0.261 0.040 -0.184 -0.190 -0.299 -0.287 0.795 4.3. Structural Model Analysis

Structural equations modeling (SEM) was used to examine the hypotheses of the proposed model. The model fitting indices of the constructs model (χ2 = 973.189; df = 591;

χ2/df = 1.647; CFI = 0.957; NFI = 0.897; IFI = 0.957; RMSEA = 0.040) met the appropriate levels. Inspection of the path coefficients was assessed to check the research hypotheses.

Table 5 and Figure 2 show the results of the tests of the hypotheses, with fifteen of the seventeen hypotheses were adopted and two hypotheses were rejected. Regarding explanatory power, the model explained 65.0% of the variation in perceived usefulness, 50.4% of the variation in perceived ease of use. Moreover, the switching intention from cash payment to mobile payment was explained by 49.4% of the variance in the model. To summarize, Figure 2 presents the estimation results.

Table 5. Results of Estimated Structural Coefficients Hypotheses Path Std.

Weights S.E. C.R. P Results R2

H1a LS → PU .155 .043 2.794 .005 Supported

0.650

H2a IC→PU .128 .059 2.013 .044 Supported

H3a AL→PUS .256 .051 4.531 *** Supported

H4a MB→PU .171 .044 3.206 .001 Supported

H5a KN→PU .079 .056 1.620 .105 Not Supported

H6a IN→PU .099 .034 2.403 .016 Supported

H7a RI→PU -.116 .043 -2.714 .007 Supported

H1b LS→PE .187 .060 3.112 .002 Supported

0.504

H2b IC→PE .196 .083 2.817 .005 Supported

H3b AL→PE .218 .070 3.641 *** Supported

H4b MB→PE .122 .061 2.088 .037 Supported

H5b KN→PE .046 .079 .855 .392 Not Supported

H6b IN→PE .183 .047 4.138 *** Supported

H7b RI→PE -.124 .059 -2.669 .008 Supported

H8 PE→PU .173 .046 2.932 .003 Supported

H9 PE→IS .213 .063 3.372 *** Supported

0.494

H10 PU→IS .546 .086 8.157 *** Supported

Figure 2. The Result of Hypothesis Test

4.4. The Test of The Moderating Effect

A pairwise parameter comparison was analyzed to test the hypothesis regarding the moderation of switching cost and habit. To do so, respondents were divided into low switching cost group (Group A) with an average point less than or equal to 4 with 101 persons, the remaining 305 persons categorized in Group B with high switching cost.

Similarly, the high habit group (Group C) contained 122 respondents with an average point over 4, and 294 respondents belonged to the low habit group (Group D). To verify the difference between groups, pairwise parameter comparisons were conducted by computing the critical ratios for differences between parameters (Z-statistics) that were confirmed with statistical significance level DBP ±1.96. As the data analysis result, there is no difference between the low and high switching cost groups, thus, rejecting hypothesis 11.

As shown in Table 6, the relationship between the inconvenience of cash payment and perceived usefulness is affected by the difference test statistic -2.216** between the low and high habit groups. By contrast, it was found that the high habit group is more sensitive to the different test statistic 3.551*** in the effect of perceived ease of use on perceived usefulness.

In addition, the effect of mobile payment knowledge on perceived ease of use was more significant in group C than group D, and the different test statistics are -2.484**. However, it was found that group D was more sensitive to the different test statistic 2.114** in the effect of personal innovativeness on perceived ease of use. Finally, the relationship between perceived ease of use and switching intention is affected by the difference test statistic - 2.194** between group C and group D.

Table 6. Testing for Moderator Effect of Habit Hypotheses Path

Standardized Estimate Critical Ratio for Differences Between

Parameters

Result Group C (112) Group D (294)

H2a IC→PU 0.430** 0.072 -2.216** Difference H5b KN→PE 0.487** -0.001 -2.484** Difference H6b IN→PE 0.080 0.275*** 2.114** Difference H8 PE→PU -0.438* 0.223*** 3.551*** Difference H10 PE→IS 0.446*** 0.147 -2.194** Difference

*** p < 0.001, ** p < 0.01, * p < 0.05 level of significance

Một phần của tài liệu Kỷ yếu hội thảo quốc tế dành cho các nhà khoa học trẻ khối trường kinh tế và kinh doanh năm 2021 (Volume 4) (Trang 436 - 440)

Tải bản đầy đủ (PDF)

(646 trang)