CHAPTER 4: DATA ANALYSIS AND RESULTS
4.2. Measure assessment 1. Reliability analysis
To ascertain the reliability of the measurement scales and to check the degree to which the items that make up the scale “hang together”, Cronbach alpha
coefficient is calculated. Cronbach s alpha checks the internal consistency ‟
reliability of scales. It checks if whether the items that make up the scale actually measure the same underlying construct (Pallant, 2001). For scale to be reliable, its Cronbach alpha value should be above 0.6 (George & Mallery, 2003).
The above guideline indicates that the higher the Cronbach s alpha value is, the ‟ more reliable are the items measuring a give construct. Cronbach s alpha closer ‟ to 1.0 is preferred. A Cronbach s alpha value of 0.9 and above was regarded as ‟ the most reliable of scales, while a scale that has a Cronbach s alpha value that is ‟ below 0.5 is regarded as unreliable and cannot be used to measure a given
construct.
[ 37 ]
Table 9: Reliability analysis for each factor
Corrected
2.941
PEU06 10.95 4.069 .569 .886
PEU07 11.52 3.471 .779 .804
PEU08 11.24
3.769
Perceived ease of use Cronbach's Alpha
ATT11
SN13
SN16
.530 .841
Subjective norm Cronbach's Alpha .846
PBC19 11.09 3.188 .692 .808
PBC20 10.37 3.261 .582 .849
PBC21 11.50 2.644 .769 .770
PBC22 10.95 2.770 .720 .793
Perceive behavioral Cronbach's Alpha .848
control
INT23 6.90 2.124 .771 .887
INT24 7.19 1.995 .856 .811
INT25 7.39 2.248 .787 .872
Intention to use MB Cronbach's Alpha .901
[ 38 ]
Table 8 depicts a summary of the beta scores of all the response ranking of the factors that affect the intention to use mobile banking in HCM. According to the data analysis, the factor "Perceived Usefulness" Cronbach's alpha coefficient is 0.625. After delete PU03 item of this factor, then the factor "Perceived
Usefulness" Cronbach's alpha coefficient increased to 0.652. Other factors exhibit a Cronbach s alpha coefficient from 0.846 to 0.901. Among the factors, the factor‟
“intention to use mobile banking” has the highest ranking of Cronbach alpha of 0.901, followed by the factor “perceived ease of use” and “attitude” with 0.867.
The factor “subjective norm” has the lowest ranking with 0.846. Hence, six variables were retained.
4.2.2. Exploratory Factor Analysis (EFA)
The test of the measurement model includes the estimation of using item-to-total correlations (>0.5) and the convergent and discriminant validity of the instrument items. The convergent validity is demonstrated when items load highly (loading>
0.5) on their associated factors.
Dependent variable and the independent variables were analyzed independently of each factor. All the variables were extracted through the Principal axis
factoring, using the Promax rotation method. After three times of running factor analysis, the result shows that all remained variables are greater than 0.5 and a significant loading on an acceptable factor. Advanced data processing steps as follows:
The first time running factor analysis for independent variables, there are six factors with the eigenvalues that are higher than 1 (see Appendix 5). The variables PU05, PEU06, PBC20 have a cross-loading (see Appendix 6).
Therefore, they were deleted from the analysis and the loadings recalculated.
The second time running, six factors were extracted (see Appendix 8).
The last time, running factor analysis for dependent variables.
[ 39 ]
Rotation was converged in six iterations. The generated factors, the attributes of each factor loadings are present in Table 9. All items have large and significant loadings on their corresponding factors. The composite reliabilities of the
different measures included in the model ranged from 0.645 to 0.955. Further, the shared variance is less than the amount of variance extracted by the indicators measuring the constructs. Thus, the convergent and discriminant validity are meet. Taken together, the evidence indicates the scales' had adequate
psychometric quality for usage in the next stage of analysis.
After processing factor analysis for the independent variables and the dependent variable by the Promax method has five factor are formed, there are three items (observed variables) are removed because values less than 0.5. The 22 items were selected and have values above 0.5, with significant dependent variable. The factors form after the implementation of EFA: perceived usefulness, perceived ease of use, attitude, subjective norm, perceived behavioral control, and intention to use mobile banking.
[ 40 ]
Table 10: Key dimensions, items Dimension
Factor Loading
PU01 - Flexibility to conduct banking business 24 hours per .742
X1 -day
Perceived
PU02 - Make banking transactions quickly .648
usefulness
PU04 - MB transactions relevant to my work .497
PEU07 - Instructions in MB system are clear and .729
understandable
X2
-PEU08 - MB has many flexible ways to search your required .801
Perceived
information
ease of use
PEU09 - I feel that user-friendliness of MB services is .979
important
ATT10 - Using MB would be a wise idea .779
X3
-ATT11 - Using MB is a good idea .894
Attitude
ATT12 - I like to use MB .921
SN13 - People important to me would think that using MB
.688
would be a wise idea
SN14 - People important to me would think that using MB is a .653
good idea
X4 –
SN15 - Most people important to me would think that I should .957
use MB
Subjective
SN16 - My family important to me would think that using MB .860
norm
would be a wise idea
SN17 - My family important to me would think that using MB .902
is a good idea
SN18 - My family important to me would think that I should .783
use MB
X5
-PBC19 - I would be able to operate MB .958
Perceived
PBC21 - I have the knowledge to use MB .752
behavioral
PBC22 - I have the ability to use MB
.621
control
Y –
INT23 - Plan to use MB .815
Intention
INT24 - Intention to use it within the next tree months .948
to use
INT25 - Add MB to my favorite apps
mobile
.841
banking
[ 41 ]