Checking the Reliability of the Factor Scale Before Conducting Exploratory Factor Analysis Efa

2.3.2. Check the reliability of the scale of factors before conducting exploratory factor analysis EFA

Cronbach's Alpha coefficient analysis is used to evaluate the reliability of scales in research. This coefficient is often used to measure the degree of correlation between items in a scale. To calculate Cronbach's Alpha for a scale, the scale must have at least three measurement variables (Nguyen Dinh Tho, 2011). The scales used in the research all have at least three measurement variables, so Cronbach's Alpha can be calculated for the scales.

According to Hair (2010), scale reliability is understood as the level at which the measurement of survey variables does not encounter errors and the results of customer interviews are accurate and consistent with reality [15].

Cronbach's Alpha coefficient is used to check the consistency and correlation between observed variables. This tool helps to eliminate inappropriate observations or scales first. Variables with an Item-Total Correlation coefficient greater than 0.3 and a Cronbach's Alpha coefficient greater than 0.6 are considered acceptable and suitable for inclusion in the next steps of analysis (Nunnally and BernStein, 1994) [16].

According to convention, a set of questions used to measure is considered good if it has a Cronbach's Alpha coefficient ≥ 0.8. According to Hoang Trong and Chu Nguyen Mong Ngoc (2008), many researchers agree that when Cronbach's Alpha is from 0.8 to nearly 1, the scale is good, from 0.7 to 0.8 is usable, from 0.6 or higher is usable in cases where the research concept is new or new in the research context.

The scale used in the study consists of five main components, including: (1) Personal loan products, (2) Loan procedures and processes, (3) Credit officers, (4) Tangible means and (5) Loan interest rates. The results of the reliability test of the scale through the Cronbach's Alpha coefficient are shown in the following table.

Table 2.14: Results of scale reliability test

Soy sauce

Coefficient

VARIABLE CODE

VARIABLE DEFINITION

total variable

Cronbach's Alpha if variable is excluded

1. Personal loan products

Cronbach's Alpha = 0.774

SP1 Diverse loan products and forms 0.641 0.699

	Flexible and reasonable loan term
SP3	Information on personal loan products Human resources are always fully provided by customers.	0.459	0.761
	Exactly
SP4	Banks regularly offer programs. Promotion on personal loan products	0.517	0.743
	very attractive
SP5 Personal loan products always 0.389			0.784
2. Loan procedures and processes Cronbach's Alpha = 0.823
TTQT1 Bank loan procedures are simple and easy 0.528			0.653
TTQT2 The bank's loan procedures are very reasonable and 0.581			0.615
TTQT3 Fast loan processing time and 0.476			0.682
TTQT4 Regulations and working procedures of the credit department 0.472			0.682
3. Credit Officer Cronbach's Alpha = 0.843
NV1 Bank credit officer has knowledge and 0.484			0.822
NV2 Credit officers have polite and friendly attitude 0.656			0.776
NV3 Credit staff handle business quickly, 0.689			0.766
NV4 Credit officers create trust for customers 0.631			0.784
NV5 Customers always receive guidance and support 0.632			0.783
4. Tangibles Cronbach's Alpha = 0.837
PTHH1 The bank facilities look very attractive 0.513			0.839

Maybe you are interested!

SP2 Personal loan products have a term of

0.751 0.661

Fully and promptly meet your needs

understand

easy to apply

Exactly

use is public, transparent

good professional competence

Exactly

row

timely from credit officer when needed

guide.

PTHH2 The bank's transaction place has very good equipment.

0.657 0.799

modern.

PTHH3 Reasonable and easy-to-receive transaction counter layout 0.762			0.770
PTHH4 Good service facilities (waiting space, newspapers, water 0.714			0.784
PTHH5 Bank employees dress politely and neatly. 0.563			0.825
5. Interest and loan fees Cronbach's Alpha = 0.890
LS_P1 Bank interest rate and loan fees are acceptable 0.537			0.695
LS_P2 The bank's lending interest rate and fee regulations are 0.622			0.638
Interest rates and bank loan fees are not too high. LS_P3 has a large difference compared to other banks over 0.498 location			0.713
LS_P4 Banks rarely change interest rates and lending fees in 0.527			0.697
6. Overall assessment Cronbach's Alpha = 0.841
DGC1	You appreciate our ability to meet your needs. bank loan demand	0.693	0.790
DGC2	You are satisfied with the values that the service brings. personal loans of the bank	0.641	0.813
	again
DGC3	You will continue to use the loan service. individual bank customers during the period	0.723	0.776
	next
DGC4	You will introduce to friends, relatives, Colleagues using customer loan services	0.643	0.812
	bank personal

know.

drink, convenient parking location…)

Okay

clear and reasonable

duration

Source: SPSS data processing results (2018)

The results of Cronbach's Alpha coefficient analysis show that all Cronbach's Alpha coefficients of the observed variable groups are greater than 0.7. In general, the research components ensure reliability to perform the necessary analysis of the research. All total variable correlation coefficients are greater than 0.3 and removing any observed variable will reduce the reliability of the scale. Thus, the observed variables ensure sufficient reliability to conduct the following analysis.

2.3.3. Exploratory factor analysis (EFA)

Factor analysis is a quantitative analysis method used to reduce a set of many interdependent observed variables into a smaller set of variables (called factors) so that they are more meaningful but still contain most of the content.

information of the original set of variables (Hair et al., 2010) [15]. When analyzing EFA, researchers often pay attention to a number of criteria:

- The KMO coefficient (Kaiser- Meyer-Olkin) is an indicator used to examine the appropriateness of EFA. The larger the KMO, the better because the common part between the variables is larger. The KMO value is between 0.5 - 1, then factor analysis is appropriate. Bartlett's test examines the hypothesis that the correlation between observed variables is zero in the population. If this test is statistically significant (Sig ≤ 0.05), then the observed variables are correlated with each other in the population enough to conduct EFA analysis [13].

- Factor loading coefficient (Factor Loading), according to Hair & the authors (2010), factor loading coefficient is an indicator to ensure the practical significance of EFA. Factor loading coefficient of 0.3 is considered the minimum level, from 0.4 or higher, factor loading coefficient is considered important, and from 0.5 is considered to have practical significance. In this study, if any observed variable has factor loading coefficient < 0.50, it will be eliminated [15].

- The scale is accepted when the total extracted variance is ≥ 50% (Nguyen Dinh Tho, 2011). Also according to Nguyen Dinh Tho (2011) [13], the total extracted variance from 60% or more is good.

- The stopping point when extracting factors with Eigenvalue coefficient must have a value ≥ 1 (Hair, 2010) [15]. Less important factors are eliminated, only important factors are retained by considering the Eigenvalue value. The results of exploratory factor analysis (EFA) are presented as follows:

Table 2.15: KMO and Bartlett'st test

KMO and Bartlett's Test

Kaiser-Meyer-Olkin Measure of

Sampling Adequacy.	0.700
Bartlett's Test of Sphericity	Approx. Chi-Square	1449,578
	Df	253
	Sig.	0.000

Source: SPSS data processing results (2018)

According to the table above, the p-value = 0.000 of Bartlett'st test allows us to safely reject the hypothesis H 0 (H 0 : Factor analysis is not suitable for the data). The KMO index = 0.700 (ranging from 0.5 to 1) shows that the model's suitability is at an acceptable level.

Components

Table 2.16: Total extracted variance

Initial Eigenvalues Rotation Sums of Squared

Loadings

Total	% of Variance e	Total %	Total	% of Variance e	Total %
1	4,114	17,888	17,888	3,105	13,498	13,498
2	3,347	14,551	32,439	3,007	13,075	26,573
3	2,557	11,115	43,554	2,820	12,260	38,833
4	2,127	9,246	52,801	2,492	10,834	49,667
5	1,541	6,702	59,503	2,262	9,836	59,503
6	1,026	4,462	63,965
7	0.859	3,734	67,699
8	0.797	3,463	71,162
9	0.776	3,375	74,538
10	0.724	3,148	77,686
11	0.697	3,029	80,715

Source: SPSS data processing results (2018)

Based on the table above, the total variance extracted is 59.503% > 50%, therefore, the analysis

factorization is appropriate.

Table 2.17: Factor rotation matrix

Group of factors

Name VARIABLE CODE 1 2 3 4 5

Vehicle

tangible

PTHH3 0.865

PTHH4 0.815

	PTHH2	0.797
Eigenvalue =	PTHH5	0.714
13,498 > 1	PTHH1	0.667
Credit Officer	NV3		0.832
use	NV2		0.750
	NV4		0.742
Eigenvalue =	NV5		0.701
13,075 > 1	NV1		0.659
Products for	SP2			0.875
science and technology loan	SP1			0.810
	SP4			0.670
Eigenvalue =	SP3			0.664
12,260 > 1	SP5			0.552
Interest and	LS_P2				0.770
loan fee	LS_P1				0.743
Science and Technology	LS_P3				0.716
Eigenvalue = 10,834> 1	LS_P4				0.696
Procedures, rules	TTQT2					0.782
Science and Technology	TTQT1					0.765
Eigenvalue =	TTQT4					0.695
9,836 > 1	TTQT3					0.680

loan program

Source: SPSS data processing results (2018)

The results of the EFA exploratory factor analysis showed that 5 factors were formed after rotating the factors. The factors kept the observed variables in the group and did not eliminate any observed variables. The factors all had Eigenvalue coefficients > 1, which is consistent with the requirements of the EFA exploratory factor rotation statistical technique.

2.3.4. Regression analysis

After evaluating the scale using Cronbach Alpha and EFA, the author used the average method of measurement variables (observed variables) for the factors to conduct regression analysis. The results of multiple regression analysis in this study were performed on SPSS statistical software with the ordinary least squares estimation method with the ENTER method.

This is an analysis step that helps determine the factors affecting the quality of credit services for personal loans of the bank through customer satisfaction. This is one of the important issues for the bank to be able to improve the quality of credit services for personal loans, at the same time improve the efficiency of the bank's business operations.

The regression model is presented as follows:

HL = β 0 + β 1 PTHH + β 2 NV + β 3 SP + β 4 LS_P + β 5 TTQT + e i

The symbols in the model are defined as follows:

- HL: Satisfied with personal loan service

- PTHH: Tangible means

- NV: Credit officer

- QT: Settlement work

- LS_P: Interest rates and fees for personal loans

- TTQT: Procedures and processes for lending to individual customers

- Coefficients: β 1 , β 2 , β 3 , β 4 , β 5 : Regression coefficients corresponding to independent variables.

- ei: Model residuals

Before testing the research model using multiple linear regression analysis, we need to consider the correlation between the variables of the model. Correlation matrix analysis uses Pearson Correlation coefficient to quantify the degree of closeness of the relationship between each different factor with Customer Satisfaction with Personal Loan Services, and the independent variables with each other.

Pearson Satisfaction 1 .253 .220 about serviceCorrelation				.616 .271 .148
Sig. loan (2-	.001 .005 )			.000 .001 .062
N	159	159	159	159	159	159
Pearson .253 ** DirectionCorrelation		1,033		.011	-.029	.019
Sig . (2- .001) imagetailed)		.675		.891	.713	.816
	N 159	159	159	159	159	159
	Pearson .220 ** Correlation	.033	1	-.008	-.008	-.004
credit	Sig. (2- .005	.675		.918	.924	.958
	N 159	159	159	159	159	159
Pearson .616 ** ProductCorrelation		.011 -.008		1	-.005	.009
loan Sig. (2- .000 Science and Technologytailed)		.891 .918			.952	.906
N	159	159	159	159	159	159

Table 2.18: Correlation matrix between variables

Satisfied with KHCN loan service

Tangible means

Credit Officer

Science and Technology Loan Products

Interest rates and fees for science and technology loans

Procedures and processes for lending to science and technology

Science and Technology

tailed

Staff

tailed)

Interest rate

Pearson

.271 ** -.029 -.008 -.005 1 .008

and fees forCorrelation

get a loan

Sig. (2-

.001 .713 .924 .952 .924

159

Procedure, Pearson

procedureCorrelation

Sig. loan (2-

.148

.062

)

.019 -.004

.816 .958

.009

.906

.008

.924

159

Science and Technology

tailed)

Science and Technology

tailed

**. Correlation is significant at the 0.01 level (2-tailed).

Source: SPSS data processing results (2018)

The test results show that the highest “correlation coefficient” between the dependent variable and the factors is 0.377 (the lowest is 0.08), these relationships are significant when sig < 0.05. It can be preliminarily concluded that these independent variables can be included in the model to explain the dependent variable. In addition, the correlation coefficients between the independent variables are mostly not statistically significant, it can be preliminarily concluded that there is no multicollinearity between the independent variables in the multiple regression model.

Comment