DIGITAL GAP ANALYSIS IN THE USE OF MOBILE BANKING APPLICATIONS IN THE CITY OF BANDUNG

. This research discusses the digital divide in the use of mobile banking applications in the city of Bandung, Indonesia, where the use of mobile banking is still relatively minimal compared to other bank services. The study aims to analyze the digital divide based on four stages of using digital technology, which are motivation, physical access and material access, mobile banking skills, and usage. The author uses quantitative methods with the PLS-SEM multivariate analysis technique to determine the relationship between predetermined variables. For data processing, the author uses WarpPLS 8.0 where the data processed is based on samples taken through non-probability sampling using the purposive sampling method with a total of 300 respondents taken from part of the population of Bandung City who use mobile banking applications. Based on research that has been carried out by the author, it was obtained that the variables motivation, physical access and material access, mobile banking skills, and usage are stated to be significant. This shows that there is no digital divide in the use of mobile banking applications in the city of Bandung.


INTRODUCTION
The current population growth in Indonesia is closely related to technological growth. This is because with population growth, the needs needed by the community will increase, so that this can affect the development of new technologies so that they can meet the needs of the population (Nick Wilkinson, 1973). The technology that is quite widely used by the people of Indonesia is Information and Communication Technology (ICT) such as smartphones and the internet.
A research that is relevant to the findings above is "Factors affecting the adoption of mobile banking in Jordan: Extending the UTAUT model" (2018) by Al-Jabali, Alnsour, and Al-Jabali. This study examines the factors influencing mobile banking adoption in Jordan using the extended UTAUT (Unified Theory of Acceptance and Use of Technology) model. One of the main findings of this study is that physical accessibility and availability of adequate materials is very important in influencing the use of mobile banking in the country.
The novelty that can be carried out in future research is to look at other moderator variables that can affect the use of mobile banking applications, such as psychological, cultural, and environmental factors. In addition, research could be carried out in different areas to see how these factors vary in different geographic and social contexts.
The number of smartphone users in Indonesia in 2021 will reach 180.4 million users, while internet users will reach 210.03 million users in 2022. In addition, there are also innovations made to ICT in the banking sector such as mobile banking (Khairani & Giri, 2020) . Even though mobile banking is an innovation that can be useful, in fact, mobile banking users are still relatively few when compared to the number of customers (Arditya & Giri, 2021) However, based on data obtained by the Indonesian people, especially residents of the city of Bandung, the number of people using the internet with the aim of accessing finance is still low, with only 163 thousand out of 2.1 million internet users in 2021. Apart from that, based on the data, mobile application users Currently, banking in Indonesia is still relatively minimal, where only 36.5 million customers use the mobile banking application in 2021. Based on this, we can see that there is a digital divide that occurs in Indonesia and in the city of Bandung. The phenomenon of this digital divide is a gap in the use of a technology, one of which is basically caused by digital skills (Hukama & Giri, 2023). Therefore the authors are interested in conducting research with the title "Digital gap analysis in the use of mobile banking applications in the city of Bandung".

RESEARCH METHODS
In this study the authors used quantitative research methods with the PLS-SEM multivariate analysis technique. The strategy that the writer uses in obtaining data is by means of a survey which amounts to 300 respondents. As for sampling, the authors used a non-probability technique using a purposive sampling method. WarpPLS 8.0 software will be used to process the data that has been obtained.

RESULTS AND DISCUSSION Measurement Model Test Results (Outer Model) Convergent Validity
The value needed so that an indicator can be explained by the construct that is measured can be seen from the outer loading value where it requires a value of > 0.5 and an average variance extracted (AVE) of > 0.5 (Sholihin & Ratmono, 2021:45 (2022) Based on the results of the tests that have been carried out, it can be concluded that each indicator can be explained by the construct being measured. This is because the values obtained from each indicator have met the required value so that an indicator can be said to be valid.

Discriminant Validity
This discriminant test is intended to determine how much a variable differs from other variables. This test can be seen with the cross-loading approach, where the value required to be valid is that the loading value of the variable being tested is greater than the loading value for other constructs.  (2022) Based on the table it can be concluded that the discriminant validity test with the crossloading approach is said to be valid because it meets the necessary conditions. In addition to using the cross-loading approach, discriminant validity tests can be carried out using the Fornell-Larcker criterion approach. To be valid, this approach requires that the AVE root value in the diagonal element is greater than the value in the non-diagonal column (correlation).  (2022) Based on the table it can be concluded that the discriminant validity test with the crossloading approach is said to be valid because it meets the necessary conditions. In addition to using the cross-loading approach, discriminant validity tests can be carried out using the Fornell-Larcker criterion approach. To be valid, this approach requires that the AVE root value in the diagonal element is greater than the value in the non-diagonal column (correlation).

Reliability Test
To test reliability, researchers can use a composite reliability value with a value of ≥ 0.7, and Cronbach's alpha with a value of ≥ 0.6 (Sholihin & Ratmono, 2021: 43-44 (2022) Based on the tests that have been carried out, the results show that the confidence level of the gauge is reliable or reliable. This is because the value obtained from the test results meets the requirements needed for a measure to be said to be reliable.  (2022) To find out the measure of the predictive power of the model, the value of R2 is needed. The R2 value of 0.75 can be said to be at a substantial level, 0.50 is moderate, and 0.25 is weak. Based on the results contained in the table, the R2 value is obtained where the results show that the R2 value on the three exogenous variables has a weak and moderate effect. The following is an interpretation of the R2 value that has been obtained: The motivation variable has an influence of 16.8% on the physical access and material access variables and around 83.2% is identified as being influenced by other variables during the study; Physical access and material access variables have an influence of 44.4% on the mobile banking skills variable and around 55.6% are identified as influenced by other variables during the study; The mobile banking skills variable has an influence of 38.4% on the usage variable and around 61.6% is identified as being influenced by other variables during the study.

Q Square (Stone-Geisser/Q 2 )
Q Square is one of the stages in the evaluation of structural models where this measure is predictive relevance. Models with predictive validity must have a Q2 value greater than zero to be said to be good (Sholihin & Ratmono, 2021: 55). In the table it can be seen that the estimation results for the Q2 value obtained are above zero, this indicates that the results have good predictive validity.

Effect Size (f 2 )
The effect size is a measure of the change in the R2 value when an independent variable is removed from the framework so that it is known how the independent variable affects the dependent construct (Sholihin & Ratmono, 2021: 84). This effect size is divided into three groups, namely 0.02 which can be said to have a weak impact, 0.15 is said to be an influence with a medium impact, and 0.35 can be said to have a large effect (Kock, 2013;Hair et al., 2013).  (2022) The following is an interpretation of the effect size value that has been obtained: the estimation results show that the effect of the motivation variable (MOT) on the physical access and material access (PMA) variables is 0.125. This result is classified as a medium size effect group, indicating that motivation has a moderate effect on physical access and material access; the estimation results show that the effect of physical access and material access (PMA) variables on mobile banking skills (MBS) is 0.418. This result is classified as a large effect size group, indicating that physical access and material access have an important influence on mobile banking skills; and the estimation results show that the influence of the mobile banking skills (MBS) variable on usage (USG) is 0.388. This result is classified as a large effect size group, indicating that mobile banking skills have an important influence on usage.  (2022) In testing hypothesis 1, it was found that the motivation variable (MOT) had a positive and significant influence on physical access and material access (PMA) variables based on a p-value of <0.001 and a path coefficient value of 0.311. In testing hypothesis 2, it was found that the physical access and material access (PMA) variables had a positive and significant influence on the mobile banking skills (MBS) variable based on a p-value of <0.001 and a path coefficient value of 0.652. In testing hypothesis 3, it was found that the mobile banking skills (MBS) variable had a positive and significant influence on the usage variable (USG) based on a p-value of <0.001 and a path coefficient value of 0.624. In testing hypothesis 4 it was found that the gender moderator variable showed a positive but not significant influence in moderating the relationship between the motivation variable (MOT) and the physical access and material access (PMA) variables based on a p-value of 0.109 and a path coefficient value of -0.071. In testing hypothesis 5, it was found that the moderator variable age showed a positive but not significant influence in moderating the relationship between the motivation variable (MOT) and the physical access and material access (PMA) variables based on a p-value of 0.339 and a path coefficient value of -0.024. In testing hypothesis 6, it is found that the moderator education variable shows a positive but not significant influence in moderating the relationship between the motivation variable (MOT) and the physical access and material access (PMA) variables based on a p-value of 0.205 and a path coefficient value of -0.047. In testing hypothesis 7 it was found that the gender moderator variable showed a positive but not significant influence in moderating the relationship between physical access and material access (PMA) variables with the mobile banking skills (MBS) variable based on a p-value of 0.251 and a path coefficient value of -0.039. In testing hypothesis 8, it was found that the age moderator variable showed a positive but not significant influence in moderating the relationship between physical access and material access (PMA) variables with the mobile banking skills (MBS) variable based on a p-value of 0.121 and a path coefficient value of -0.067. In testing hypothesis 9, it was found that the moderator education variable showed a positive but not significant influence in moderating the relationship between physical access and material access (PMA) variables with the mobile banking skills (MBS) variable based on a p-value of 0.318 and a path coefficient value of 0.027. In testing hypothesis 10, it was found that the gender moderator variable showed a positive but not significant influence in moderating the relationship between the mobile banking skills (MBS) variable and the usage variable (USG) based on a p-value of 0.119 and a path coefficient value of 0.068. In testing hypothesis 11, it was found that the moderator variable age showed a positive but not significant influence in moderating the relationship between the mobile banking skills (MBS) variable and the usage variable (USG) based on a p-value of 0.460 and a path coefficient value of 0.006. At last, in testing hypothesis 12, it was found that the moderator education variable showed a positive but not significant influence in moderating the relationship between the mobile banking skills (MBS) variable and the usage variable (USG) based on a p-value of 0.093 and a path coefficient value of -0.076.

CONCLUSION AND SUGGESTIONS Conclusion
This study found that physical and material access variables were influenced by motivation, mobile banking skills were influenced by physical and material access, and usage was influenced by mobile banking skills. The study also found that there was no significant effect of moderator variables such as gender, age, and education on the relationship between latent variables. Therefore, it can be concluded that there is no digital divide in the use of mobile banking applications in Bandung. For future research, the authors suggest exploring