by Oliver (1999), the scale has been inherited and developed by many studies in recent years (see Appendix 2).
4.3 Qualitative Research
4.3.1 Qualitative research objectives
Qualitative research was conducted with the aim of discovering and correcting the scale in the influence of CSR on the loyalty of depositing customers at Vietnamese commercial banks in Ho Chi Minh City. The scale formed in the qualitative research will be the basis for building the official survey questionnaire to collect data for the next quantitative research.
4.3.2 Design of qualitative research
Three common tools for scientific research in business are group discussion, one-on-one discussion and observation (Nguyen Dinh Tho, 2011). In which qualitative research only uses two tools, group discussion and one-on-one discussion. According to Ly Thuc Hien (2011), one-on-one discussion is suitable for research topics that are highly individual, not suitable in a collective environment or because of competition, but research subjects do not participate in group discussions. Group discussion is more suitable for exploratory research on consumer habits and attitudes; develop data to build quantitative research questionnaires. In this study, the author chooses a group discussion tool for qualitative research. The questions in the discussion outline are built by the author based on the scales of previous studies. Appendix 3 .
According to Ly Thuc Hien (2011), a real group usually has 8-10 respondents, a small group of about 4 respondents. In this study, the author selects 8 respondents who have been depositing money at Vietnamese commercial banks in Ho Chi Minh City to participate in the group discussion. The respondents are selected according to the principle: the higher the homogeneity in the group, the easier it is to discuss; the member has not participated in a similar discussion before or at least for a period of 6 months to 1 year; members do not know each other (Ly Thuc Hien, 2011). The list of respondents is listed in Appendix 4. In this list, there are 4 respondents who are employees
transactions directly with customers depositing money at Vietnamese commercial banks in Ho Chi Minh City. The author chooses these 4 respondents because they are the ones who often interact with depositors, they will probably understand the factors related to customer-oriented CSR implementation, consider consent or complementary to the factors that are the consequences of CSR and the antecedents of customer loyalty. The remaining four respondents are depositors at Vietnamese commercial banks in Ho Chi Minh City. Ho Chi Minh, the author also discusses to see the level of understanding of the concepts and statements used in the research scales. The author is the one who arranges the group discussion and conducts the discussion. essay. The discussion proceeds as follows:
Step 1: Before conducting group discussion, the author investigates and selects individuals who are working at Vietnamese commercial banks in Ho Chi Minh City and customers who have used deposit products of these banks.
Step 2: Contact the respondents to choose an appropriate time and place and conduct a group discussion.
Step 3: During the discussion, the author notes the respondents' opinions on the contents related to the group discussion ( Appendix 3 ).
Step 4: Synthesize all the data collected in the group discussion, from this data adjust the proposed questionnaire into the official one.
4.3.2 Qualitative research results
The eight respondents participating in the group discussion generally have a high degree of homogeneity, the common age group is from 24 to 30 years old, the occupation is mainly office workers, etc., therefore, the group discussion is easier. The respondents were all asked questions in the order of the group discussion outline ( Appendix 3). In which, records from 4 respondents who are bank employees showed. They completely agree with the factors proposed in the research model in chapter 2, that is, there are 4 factors that are the consequences of CSR and these 4 factors affect the loyalty of depositors:
(1) Customer identification; (2) Customer satisfaction; (3) Customer's feelings (4) Customer's trust. In which the element "Customer identification" is also
have an impact on “Customer Emotions” and “Customer Satisfaction”; The factor “Customer trust” not only affects loyalty but also affects “Customer satisfaction” and “Customer emotion” in addition to “Loyalty” also affects “ “Customer Satisfaction” deposit. They find that in the process of dealing with customers, when they can prove these products with many benefits and incentives, customers tend to be more satisfied, happy and comfortable in contact with the bank. than. Some respondents also believe that these customers will deposit more or renew when due, and even recommend their relatives and friends to deposit money at the bank.
For questions to check the clarity of the statements in the scales. Respondents who are depositors as well as bank employees also answered that most of the statements were clear. Remittance respondents who were able to understand these statements demonstrate that the statements can be used to construct a formal survey for quantitative research. However, there are also some statements that are unclear or inappropriate. The author adjusts these statements accordingly according to the opinions of the respondents participating in the group discussion. Details of scale calibration are presented in tables 4.1 to 4.6:
Corporate Social Responsibility Scale Table 4.1 Corporate Social Responsibility Scale after adjustment
The source | The scale | Correction | Code chemistry |
Nguyen Hong Ha (2016) | This bank provides accurate information private deposit products to customers. | No correction | TN1 |
This bank cares about benefits of customers (consultation, customer care, ...). | No correction | TN2 | |
Banks treat customers fairly row. | No correction | TN3 | |
Product price is affordable, suitable for product quality. | Reasonable service fees suitable for the product. | TN4 | |
This bank has support programs support cultural activities, community and | No correction | TN5 |
Maybe you are interested!
- The influence of corporate social responsibility on the loyalty of customers depositing money at Vietnamese commercial banks - 2
- The Need To Study The Loyalty Of Depositors At Vietnamese Commercial Banks
- Actual Situation Of Deposit Mobilization Activities Of Vietnamese Commercial Banks In The City. Ho Chi Minh
- Analysis Of Multi-Group Model According To Customer Characteristics
- Suggestions On Implementing Social Responsibility Customer-Oriented Approach
- The influence of corporate social responsibility on the loyalty of customers depositing money at Vietnamese commercial banks - 8
social (studying for students, helping tough people,…) |
Customer loyalty scale
Table 4.2 Customer loyalty scale after adjustment
The source | The scale | Correction | Encode |
Oliver (1999) | If someone I need product advice deposit, I will always recommend this bank. | No correction | TT1 |
I often say well about this bank when who that asked about it. | No correction | TT2 | |
I will always go to the bank every time want to send money. | No correction | TT3 | |
This bank is the first choice when I need to use a new deposit product. | No correction | TT4 |
Customer identification scale
Table 4.3 Customer identification scale after correction
The source | The scale | Correction | Code chemistry |
Pérez and Bosque (2015a). | I see a strong resemblance between me and bank | No correction | ND1 |
This bank matches my personality | No correction | ND2 | |
I feel great being a guest This bank's products | No correction | ND3 | |
I enjoy saying that I am customer of this bank | No correction | ND4 | |
I feel strongly attached to the bank this | No correction | ND5 | |
I clearly feel about the membership of this bank. | No correction | ND6 |
Customer emotion scale
Table 4.4 Customer's emotional scale after adjustment
The source | The scale | Correction | Code chemistry |
Vlachos (2012) | I love this bank | No correction | CX1 |
I love this bank | Remove from scale | ||
This bank makes me feel so happy | This bank does I feel very interesting | CX2 | |
This bank is the most popular | No correction | CX3 |
mine | |||
Qualitative NC | I feel comfortable every time I trade deposit at this bank. | CX4 |
Customer confidence scale
Table 4.5 Customer confidence scale after adjustment
The source | The scale | Correction | Code chemistry |
Martínez and Bosque (2013) | This bank's deposit service makes me feel safe. | I feel safe when deposit money in this bank. | NT1 |
I believe in service quality deposit from this bank. | No correction | NT2 | |
Guaranteed bank deposit service quality assurance. | Reputable bank for with deposit customers | NT3 | |
This bank cares about customers deposit goods. | No correction | NT4 | |
This bank is honest with customers deposit goods. | No correction | NT5 |
Customer satisfaction scale
Table 4.6 Customer satisfaction scale after adjustment
The source | The scale | Correction | Code chemistry |
Pérez & Bosque (2015a) | Deciding to choose to deposit money at the bank This item is the right decision. | No correction | HL1 |
I feel happy when I decided to choose this bank. | I feel satisfied when decided to deposit money at this bank | HL2 | |
Banks can provide services deposit service that I need. | No correction | HL3 | |
I am satisfied with the bank I am use. | No correction | HL4 |
To summarize the results of qualitative research:
- Scale of CSR (TN): 5 observed variables from TN1 to TN5.
- Customer identification scale (ND): 6 observed variables from ND1 to ND6.
- Customer emotion scale (CX): 4 observed variables from CX1 to CX4.
- Customer trust scale (NT): 5 observed variables from NT1 to NT5.
- Scale of customer satisfaction (HL): 4 observed variables from HL1 to HL4.
- Customer loyalty scale (TT): 4 observed variables from TT1 to TT4.
Based on the results of qualitative research, the author added demographic variables (age, gender, education level and occupation) to form a formal questionnaire (see Appendix 5 ) for the study. quantitative research.
4.4 Quantitative Research
Quantitative research is carried out to analyze the data and the appropriateness of the research model as well as to test the proposed hypotheses.
Quantitative research taking data from interview results by questionnaire for customers depositing money at Vietnamese commercial banks in Ho Chi Minh City. The questionnaire includes variables that have been adjusted from qualitative research. The study used a 5-level Likert scale with opposite form (opposite scale): level 1 corresponds to the degree of completely disagree and level 5 corresponds to the degree of completely agree. The customer survey questionnaire is presented in the appendix ( Appendix 5 ).
Sampling method : This study selects a sample by convenience, non-probability method. This method makes the respondents easily accessible, they are ready to answer the questionnaire and can save time and cost for the researcher.
Sample size : According to Hair et al. (2010) to be able to conduct EFA exploratory factor analysis, the ratio of observations/measured variables is 5:1. According to Tabachnick and Fidell (2007, citing Nguyen Hong Ha, 2016), the sample size must ensure the formula: n ≥ 50 + 8p. Where n: is the required minimum sample size and p: is the number of concepts in the model. Therefore, to ensure representativeness and provision for incomplete surveys. In this study, the author uses 6 concepts with 28 observed variables, the expected number of customers to be surveyed is about 300 people.
Data collection method: Surveys are sent directly to customers at Vietnamese commercial banks in Ho Chi Minh City. Ho Chi Minh City or sent online via email and Google's support tools to customers. Data collection period is July 2017.
Data analysis method: from the collected data, the author uses SPSS 23 software to make descriptive statistics, evaluate the reliability of the scale, analyze the EFA exploratory factor, and analyze the confirmatory factor. CFA determination, research model analysis and hypothesis testing, multigroup analysis using AMOS 20 software.
Descriptive Statistical Analysis
Simple statistics such as frequency, percentage, mean, standard deviation in SPSS 23 software are used to describe the characteristics of the research sample including: age, sex, education level, income input and scale research concepts.
Check the reliability of the scale using Cronbach's Alpha coefficient
Cronbach's Alpha coefficient is used to check the rigor of the scale to eliminate inappropriate variables.
The criteria used when making the scale reliability assessment:
+ type of observed variables with small variable-total correlation coefficient (less than 0.3) (Nguyen Dinh Tho, 2011).
+ The higher Cronbach's Alpha coefficient, the better. However, when Cronbach's Alpha is too large (>0.95), it is easy to overlap in the scale, meaning that many variables in the scale have no difference or the research data cannot be trusted. According to Nguyen Dinh Tho (2011), a scale has good reliability when it varies from 0.75 to nearly 0.95, but Cronbach's Alpha coefficient from 0.6 or more is acceptable.
EFA . exploratory factor analysis
EFA analysis is used to evaluate the convergence value and discriminant value of the factors in the scale, so it is best to analyze the EFA factor simultaneously for all variables instead of for each research concept. The EFA analysis method belongs to the group of interdependent multivariate analysis, that is, there is no independent or dependent variable, but it relies on the correlation between the variables. Two commonly used extractions are “Principle Component Analysis-PCA” with Varimax rotation and “Principle Axis Factoring-PAF” with Promax rotation. According to Gerbing
& Anderson (1988), Principal Axis Factoring - PAF extraction method with Promax rotation will reflect the data structure more accurately than Principal Components extraction method with Varimax rotation, Promax rotation will be interested in the structure of the scale. , the concepts after drawing can be correlated with each other, and as well as a clear distinction between the factors. This is suitable for performing CFA and linear structural analysis (SEM) in the next step. Therefore, in this study, the author uses the PAF extraction method with Promax rotation. Criteria used in EFA analysis:
- KMO>=0.5, Bartlett test has statistical significance (sig<0.05) 3 .
- Factor Loading factor >=0.5 (Hair et al., 2010) 4 .
- Total variance extracted (TVE) ≥ 50% (Gerbing & Anderson, 1988) 5 .
- Eigenvalue coefficient must have the value ≥ 1 (Gerbing & Anderson, 1988) 6 .
- Difference | Factor Loading| largest and | Factor Loading| any must >=0.3 (Gerbing & Anderson, 1988).
If the results of EFA analysis of these criteria are satisfied, the model is suitable.
Confirmatory Factor Analysis
Confirmatory factor analysis (CFA) is one of the statistical techniques of linear structural modeling (SEM). CFA allows us to test whether the measurement model meets the requirements and how well the observed variables represent the factors, test discriminant values, convergent values. Unlike the exploratory factor analysis EFA, which must evaluate the whole, the CFA method evaluates each concept locally, or it is best to evaluate each pair of concepts or evaluate all the concepts at the same time.
3 KMO is an indicator used toconsider the appropriateness of EFA, 0.5≤KMO≤1 then factor analysis is appropriate. Bartlett's test considers the hypothesis that the correlation between the observed variables is zero in the population. If this test is statistically significant (Sig < 0.05), the observed variables are correlated with each other in the overall population (Nguyen Khanh Duy, 2009).
4 Factor loading is the criterion to ensure the practical significance of EFA. Factor loading 0.3 is considered minimal, Factor loading 0.4 is considered important, Factor loading 0.5 is considered significant.
farewell. In addition, there are some conclusions: If the Factor loading criterion is selected ≥ 0.3, the sample size should be at least 350, if the sample size is about 100-350, then the Factor loading criterion should be chosen ≥ 0.55, if the sample size is about < 100, then the factor loading criterion should be chosen. Factor loading must be 0.75 (Hair et al., 2010)
5 Total variance extracted (TVE) shows how much percentage of the variables are extracted.
6 Is the sum of the squares of the weights of the variables on a factor column. It represents the degree of volatility explained by a factor.
concept in the critical model. A critical model is a model in which all research concepts are freely related to each other. Using the critical model not only tests the set of scales for each research concept, but also tests the discriminant validity between the concepts (Hoang Hai Yen, 2015). The criteria used when analyzing CFA:
- The degree of fit of the model with the actual information : The model is considered appropriate when the Chi-squared has the P value > 5%. However, since Chi-squared has the disadvantage of being highly dependent on the sample size. The larger the sample size, the larger the chi-squared, which reduces the fit of the model. Other indicators that replace Chi-squared to assess the relevance of data to reality are:
+ Chi-squared adjusted for degrees of freedom CMIN/DF <3 (Carmines and McIver, 1981; quoted Hoang Hai Yen, 2015). According to Hair et al. (2010): n < 200, CMIN/df < 3, n > 200, CMIN/df < 5, where n is the sample size.
+ RMSEA (Root Mean Square Error Approximation) index ≤ 0.08, RMSEA ≤
0.05 is considered very good (Hair et al., 2010).
+ TLI (Tucker and Lewis Index); Comparative Fit Index (CFI), GFI fitness index. TLI, CFI, GFI > 0.9 (Hair et al., 2010). However, values for GFI, TLI and CFI at 0.8 to 0.9 are also acceptable (Gerbring and Anderson, 1988).
- Evaluate the combined reliability ( ρc ) and extracted total variance ( ρvc) . The scale is reliable when both the combined reliability c and the total variance extracted vc are valid .
>0.5 (Hair et al., 2010).
- Unidirectionality of the scale: the scale is unidirectional when there is no correlation between the errors of the observed variables (Nguyen Khanh Duy, 2009).
- Evaluation of the convergent value of the scale: the scale reaches the convergent value when the observed variables of a research concept have high correlation or high standardized (≥0.5) and achieve level of statistical significance (P < 0.05) (Gerbring and Anderson, 1988).
- Evaluation of discriminant value. The scale set achieves discriminant value when the internal correlation coefficient between the component concepts of a research concept and the external correlation coefficient between other research concepts is 1 (Nguyen Khanh Duy, 2009).
SEM . linear structural model analysis
The SEM method is used to evaluate the causal relationship between the components, more specifically, to estimate the impact of the independent variables on the dependent variable. In hypothesis testing and research models, the linear structural model also has advantages over the multivariable regression method because it can calculate the measurement error, test the reliability of the associated model. The correlation coefficients in Model Fit are also more reliable than conventional tests.
Bootstrap test
This test method is used to evaluate the reliability of the estimates, if the standard deviation appears not statistically significant (>0.05), it can be concluded that the model is reliable.
Multi-group analysis
Multi-group analysis aims to examine the difference in research results for each qualitative group with different customer characteristics (gender, service time, income) and bank characteristics (group of commercial banks). with high state ownership, group of joint stock commercial banks), banking products (savings deposits, payment deposits).
4.5 Research results
4.5.1 Sample descriptive statistics
Data is collected by distributing survey questionnaires directly or by email to customers of Vietnamese commercial banks in Ho Chi Minh City. Ho Chi Minh. The author approached the survey subjects through tellers and consultants at Vietnamese commercial banks in Ho Chi Minh City and asked them to send the survey to customers. Besides, the author has worked at a number of banks in the past, the author contacted old customers via mail, phone and asked them to assist in doing online surveys. Before entering raw data into the machine, the author checks and removes the votes
Invalid survey. Of the 309 survey results received. The author removed the votes of respondents who did not belong to the survey subjects, did not fill in all information, did not answer all the questions. As a result, there were only 267 valid answer sheets and were entered into the computer.
During the survey, some questions were intentionally repeated many times such as ND1 and ND7 in the customer identification scale, NT2 and NT6 in the trust scale, TT2 and TT5 in the customer loyalty scale. (see Appendix 5 for details). If the results of the sentences are different, indicating that the respondents are not paying attention to the question, those results should be discarded. After entering the data, the author conducts the second data filtering by using Excel functions to detect these unreliable survey tables. As a result, there are only 236 valid survey votes left. Thus, the data has been cleaned and continued to be included in the next analysis steps. The sample includes 236 customers, still ensuring the required sample size requirements in the NC.
Table 4.7 Statistics of the number of surveys
Form survey | Initial survey | Filter 1 | Filter 2 | |||
Generate | Received | Invalid | Illegal | Trust | Unreliable | |
Direct | 200 | 153 | 142 | 11 | 137 | 5 |
Online | - | 156 | 125 | thirty first | 99 | 26 |
Total | - | 309 | 267 | 42 | 236 | thirty first |
( Source: author's calculation)
In general, the results show that the direct survey has a higher degree of validity, out of a total of 153 survey panels, only 11 are invalid, accounting for 7.2%. Similar to the data cleaning time, the exclusion rate is only 3.6%, corresponding to 5 survey panels. Because, surveying actual customers, the concentration of customers is high, they are less influenced by many surrounding factors. This, in contrast to the online survey, the respondents may choose the grand result without reading the question, leading to the lack of confidence in the survey results.
The results of the sample synthesis show that the proportion of men and women is fairly evenly distributed, among 236 customers participating in the survey, there are 135 male customers, accounting for 57.20% and 101 female customers, accounting for 42.80%. The majority of customers participating in the survey are aged from
25 to 40, followed by customers under 25 years old, at least the group of customers over 60 years old with the rate of 8.05%, equivalent to 19 survey participants.
Table 4.8 Sample statistics by sex and age
Variable | AGE | % | ||||
<25 | 25-40 | 41-60 | >60 | |||
SEX | male | 36 | 52 | 36 | 11 | 57.20 |
Female | 32 | 28 | 33 | 8 | 42.80 | |
Total | 68 | 80 | 69 | 19 | 236 | |
% | 28.81 | 33.90 | 29.24 | 8.05 | 100 |
( Source: author's calculation from survey results)
Regarding education: subjects in the survey sample have a university degree, accounting for a high proportion of 80 people, accounting for 33.9%. Next is the subjects with high school education with 68 people, accounting for 28.8%. Survey participants with graduate degrees also accounted for a fairly high proportion with 38 people accounting for 16.1%.
Regarding income: the income level of most of the respondents is from 5,000,000 to less than 10,000,000 VND with 117 people, the rate is 49.6%. The group of customers with an income of over VND 20,000,000 includes 47 guests, a rate of 19.9%. And the remaining income from 10,000,000 to less than 20,000,000 and 5,000,000 dong are behind with the corresponding ratio
17.4% and 13.1%.
Table 4.9 Summary of sample descriptive statistics
Sample information | Frequency | Ratio (%) | |
Academic level | High School | 68 | 28.8 |
Intermediate college | 50 | 21.2 | |
University | 80 | 33.9 | |
After university | 38 | 16.1 | |
Income | < 5,000,000 VND | thirty first | 13.1 |
5,000,000 – less than 10,000,000 VND | 117 | 49.6 | |
10,000,000 VND – less than 20,000,000 VND | 41 | 17.4 | |
20,000,000 or more | 47 | 19.9 | |
Products Products | Deposit payment | 108 | 45.8 |
Saved money | 128 | 54.2 | |
Delivery time Translate | Less than 1 year | 82 | 34.7 |
1 to 3 years | 110 | 46.6 | |
More than 3 years | 44 | 18.6 |
( Source: author's calculation from survey results)
Regarding deposit products, survey respondents used savings more, with 128 people accounting for 54.2%, the rest were customers using deposits.
payment.
Regarding the transaction time, there are 110/236 surveyed customers who have transacted with the bank for 1-3 years, 82 customers have transacted for less than 1 year and 44 customers have transacted for more than 3 years, accounting for 18.6 percent. %.
41
37
32 29
24
20
14 11
9
7 4
3
2
1 1
first
As a result from Figure 4.2 , BIDV's customers participated in the survey the most with 41 customers, followed by Vietinbank with 37 customers. In the group of 4 banks with the largest number of customers participating in the survey, there are 3 commercial banks with state-owned capital accounting for a high proportion of charter capital, namely BIDV, Vietinbank and Vietcombank. This can also be easily explained because these are three large banks, which have been operating for a long time and have received a good reputation from consumers. In addition, these banks have a wide variety of deposit products, thus receiving many choices of customers. Some other banks have a high percentage of customers participating in the survey, such as ACB, Sacombank, and SCB with 32, 24 and 20 customers respectively.
Figure 4.2 Customer size of the banks in the sample
4.5.2 Analysis of the reliability of the scale
The reliability of the scale is assessed by Cronbach's Alpha reliability coefficient. According to Nguyen Dinh Tho (2011) to calculate Cronbach's Alpha, the scale must have at least 3 measurement variables. The concepts in this study all have at least 4 measurement variables, so there are enough conditions to calculate Cronbach's Alpha coefficient.
Table 4.10 Analysis results of Cronbach's Alpha
Variables close | Average of the scale if variable type | Scale variance if variable type | Variable correlation total | Cronbach's Alpha if type variable | Cronbach's Alpha |
Corporate Social Responsibility: n=236.5, 5 variables. | |||||
TN1 | 14.326 | 17,634 | .806 | .923 | .935 |
TN2 | 14.314 | 17,050 | .823 | .920 | |
TN3 | 14,360 | 16,589 | .857 | .914 | |
TN4 | 14.326 | 17,336 | .837 | .917 | |
TN5 | 14.403 | 17,790 | .807 | .923 | |
Customer identification : n=236, 6 variables. | |||||
ND1 | 16,640 | 20,197 | .630 | .915 | .913 |
ND2 | 16,627 | 19,843 | .699 | .906 | |
ND3 | 16.576 | 17,973 | .828 | .887 | |
ND4 | 16.572 | 18.033 | .804 | .891 | |
ND5 | 16,589 | 18,167 | .805 | .891 | |
ND6 | 16.572 | 18,671 | .775 | .895 | |
Emotions of customers : n=236, 4 variables. | |||||
CX1 | 10.538 | 7.509 | .718 | .870 | .888 |
CX2 | 10.513 | 6,864 | .749 | .858 | |
CX3 | 10,407 | 6,642 | .780 | .845 | |
CX4 | 10,441 | 6.716 | .774 | .848 | |
Customer trust : n=236.5 5 variables. | |||||
NT1 | 14,720 | 12,117 | .690 | .883 | .895 |
NT2 | 14,797 | 11,678 | .702 | .881 | |
NT3 | 14,712 | 10,963 | .775 | .864 | |
NT4 | 14.648 | 11,199 | .771 | .865 | |
NT5 | 14,581 | 11,096 | .772 | .865 | |
Customer satisfaction : n=236, 4 measurement variables. | |||||
HL1 | 11,085 | 5.057 | .625 | .780 | .819 |
HL2 | 11,000 won | 4,630 | .661 | .762 | |
HL3 | 11.072 | 4,739 | .660 | .763 | |
HL4 | 10,966 | 4.671 | .619 | .783 | |
Customer loyalty : n=236, 4 measurement variables. | |||||
TT1 | 10,669 | 7,201 | .694 | .859 | .877 |
TT2 | 10,725 | 6.771 | .676 | .866 | |
TT3 | 10,695 | 6.153 | .800 | .816 | |
TT4 | 10,712 | 6,189 | .780 | .824 |
(Source: Author analysis from SPSS)
CSR scale: Cronbach's Alpha coefficient reached .935 and the lowest variable-total correlation coefficient reached .806 . The variables of this scale that continue to be used for EFA analysis are TN1 to TN5 and no observed variables were excluded.
Customer identification scale: Cronbach's Alpha coefficient reached .913 and the lowest variable-total correlation coefficient reached .630 . In which, if ND1 is removed, Cronbach's Alpha coefficient increases to .915. Because this difference is not significant, keeping this variable reduces the reliability of the scale not much, so the author decided to keep the ND1 variable to continue the EFA analysis. The variables of this scale that continue to be used for exploratory factor analysis are ND1 to ND6.
Customer emotion scale: Cronbach's Alpha coefficient reaches .888 and there is no variable correlation coefficient – the sum of the components is less than 0.3. Therefore, keeping 4 variables CX1, CX2, CX3, CX4 to continue EFA analysis.
Customer confidence scale: Cronbach's Alpha coefficient reached .895 and the lowest variable-total correlation coefficient reached .690 . The variables of the scale (NT1 to NT5) continued to be used for exploratory factor analysis and no observed variables were excluded. Customer satisfaction scale: Cronbach's Alpha coefficient reaches .819 and there is no variable correlation coefficient – the sum of the components is less than 0.3 (the lowest coefficient is .619) so this scale is reliable. All variables were retained for EFA analysis. Customer loyalty scale: Cronbach's Alpha coefficient reached .877 and the lowest variable-total correlation coefficient reached .676 Therefore, this scale is reliable. All variables were retained for EFA analysis.
4.5.3 Exploratory factor analysis EFA
According to Nguyen Dinh Tho (2011), the dependent variable cannot be combined with the independent variable to process EFA at the same time when using the Varimax perpendicular rotation and using the factor value generated by EFA for further analysis. according to. The author uses Promax rotation for exploratory factor analysis for all variables in the model. The shortened results are shown in Table 4.11 (see Appendix 7 for details):