Factor Analysis of Continuance Intention to Use QR Code Mobile Payment Services: An Extended Expectation-Confirmation Model (ECM)

ABSTRACT


Introduction
The increasingly developed technology has made progress with the presence of a transaction method using mobile payment or m-payment as a digital payment.M-payment has now dominated the market in both developed and developing countries (Franque et al., 2021).Several innovations have been implemented in the m-payment application service, one of which is the Quick Response (QR) code technology is a code that has a black and white grid arrangement and is usually used to store information that can be read by a cell phone camera (Gao et al., 2018).QR code m-payment is used by opening the m-payment application to display or scan QR code.According to a Kadence (2021) survey, it shows that OVO, GoPay, and ShopeePay are the three most widely used m-payment applications in Indonesia with a percentage of OVO 31%, GoPay 25%, ShopeePay 20%, DANA 19%, and LinkAja 4%.The three applications are OVO, GoPay, and ShopeePay.They also have a QR code service for transaction processing.M-payment applications are very convenient payment method and can be integrated with QR codes (Silalahi et al., 2022).
The main advantage of using QR code m-payment is that users can make payments easily, anywhere and anytime.However, there are still obstacles related to the intention to continue using 224 H3: Perceived Usefulness (PU) has a positive effect on Satisfaction (S) H4: Perceived Usefulness (PU) has a positive effect on Continuance Intention (CI)

Trust
Trust is the user's level of belief in the QR code m-payment service.Lack of security will also increase the risk and uncertainty of QR code m-payment after adoption of usage, so that it can affect the intention of continued use.Several studies have shown the effect of trust on continuance intention in various contexts, such as m-payment (Cao et al., 2018), m-wallet (Kumar et al., 2018), and m-banking (Susanto et al., 2016).When users feel a service is reliable, it will lead to a sense of satisfaction.If the user does not trust it, then there is a possibility of dissatisfaction with the service, resulting in a negative evaluation (Cao et al., 2018).Trust has been shown to have a significant positive effect on satisfaction (Cao et al., 2018;Poromatikul et al., 2019;Susanto et al., 2016).
H5: Trust (T) has a positive effect on Satisfaction (S) H6: Trust (T) has a positive effect on Continuance Intention (CI)

Effort Expectancy
Effort expectancy in this study is the user's perception of the ease of using the QR code m-payment.Effort expectancy in UTAUT can be used to investigate continuance intention (Venkatesh et al., 2011).Marinković et al. (2020) used the UTAUT and stated that effort expectancy determines satisfaction.Furthermore, effort expectancy affects continuance intention in using m-payment (Singh, 2020).The research of Gao et al. (2018) also shows that there is a positive effect of effort expectancy with continuance intention in the use of QR codes in m-payments.H7: Effort Expectancy (EE) has a positive effect on Satisfaction (S) H8: Effort Expectancy (EE) has a positive effect on Continuance Intention (CI)

Perceived Risk
Perceived risk is the user's response to uncertainty and adverse consequences when using a system or service (Yuan et al., 2016).QR code m-payment services are often associated with risks related to privacy, loss of personal data, and transactions, which are of greatest concern to consumers (Gao et al., 2018).Some users may be concerned if the service provider discloses their personal data to other companies or people.Perceived risk has a significant negative effect on satisfaction in the use of m-payment (Chen & Li, 2017;Yuan et al., 2016).There is a negative effect between perceived risk and continuance intention of using m-payment services (Chen & Li, 2017;Rouibah et al., 2016;Shao et al., 2019).H9: Perceived Risk (PR) has a negative effect on Satisfaction (S) H10:Perceived Risk (PR) has a negative effect on Continuance Intention (CI)

Social Influence
Social influence, in the context of this study, is specified as the influence of the family or friends environment.Since people observe the actions of others when using m-payment services, they are influenced by those around them (Gao et al., 2018).Previous research discovered that social influence explains positively on continuance intention to use m-payment (Gao et al., 2018;Lu et al., 2017).H11: Social Influence (SI) has a positive effect on Continuance Intention (CI)

Satisfaction
Satisfaction is defined as a user's feeling of satisfaction in using the QR code m-payment service.However, when users are dissatisfied with the product, they can stop using it (Kumar et al., 2018).In accordance with several previous studies, continuance intention is affected by satisfaction (Joo et al., 2017;Kumar et al., 2018;Yuan et al., 2016).H12: Satisfaction (S) has a positive effect on Continuance Intention (CI) The research model used is shown in Figure 1.

Method
This study uses quantitative methods in testing the extended ECM by combining ECM and UTAUT and adding trust and perceived risk to explain continuance intention to use QR code m-payment services.Quantitative methods are used through surveys using questionnaires.The questionnaire consists of 34 items.Questionnaires were distributed to users of QR code m-payments such as OVO, GoPay, or ShopeePay via WhatsApp, Instagram, and Twitter.

Participants
The sampling technique used is purposive sampling.The specified criteria are users who have used QR code m-payment OVO, GoPay, or ShopeePay and are at least 17 years old.Using a minimum sample of 250, because in PLS-SEM the sample size of more than 250 can increase accuracy and consistency (Sholihin & Ratmono, 2021).This study obtained 313 valid participants.Based on gender, 183 (58.5%) of the 313 participants were female and 130 (41.5%) were male.Most of the participants were 17-25 years old 240 (76.7%), and most of them had a high school education or equivalent 197 (62.9%).Students dominated 207 (66.1%) of the total 313 participants.The most participants were ShopeePay users, with 284 (90.7%).Furthermore, most of the participants used the QR code m-payment more than once a week 159 (50.8%).Table 1 shows the profile of participants.

Measures
The data was obtained from a survey using a questionnaire.The questionnaire consists of eight constructs and 34 items.They are perceived usefulness (PU) (four items), confirmation (C) (four items), satisfaction (S) (five items), effort expectancy (EE) (four items), social influence (SI) (five items), perceived risk (PR) (five items), trust (T) (four items), and continuance intention (CI) (three items).The research questionnaire consists of demographic information and questions of construct.

Data Analysis
Data analysis is a process after collecting participant data or other data sources (Sugiyono, 2013).
Then the data analysis was carried out by dividing it into two steps are demographic analysis and statistical analysis.Demographic analysis by processing and analyzing demographic data from the results of distributing questionnaires.The next stage is statistical analysis to test the model and test the proposed hypothesis using the Partial Least Square-Structural Equation Modeling (PLS-SEM) approach with SmartPLS version 3.

Individual Item Reliability
Individual item reliability is done by checking the outer loading value of each variable.The indicator or item is declared valid if the outer loading value is above 0.7 (Hair et al., 2017).Based on Table 2, the outer loading is more than 0.7, so it can be said to be valid and can be continued on the test afterwards.

Internal Consistency Reliability
The internal consistency reliability test is based on the value of composite reliability and Cronbach's alpha values with the condition that the value is above 0.7 (Hair et al., 2017).Based on Table 3, composite reliability and Cronbach's alpha have values above 0.7, so that the variable is in accordance with the requirements.

Convergent Validity
The convergent validity test is carried out based on the value of the average variance extracted (AVE) and accepted if it is more than 0.5 (Hair et al., 2017).Table 4 shows that The AVE for every variable is more than 0.5.(Hair et al., 2017;Yamin & Kurniawan, 2011).Fornell-larcker based on the AVE root value on the correlation between variables, provided that each variable's value must be larger than the correlation with other variables and the requirement for the measurement of cross loading is that the value of the indicator on the variable itself must be greater than the value of the indicator on other variables.(Hair et al., 2017).Based on Table 5 and 6, the cross loading and fornell-larcker values are appropriate so that it can be continued in the next test.Path coefficient which shows the magnitude of the influence of each independent variable (Widarjono, 2020).If the path coefficient value is more than 0.1, then it will affect the model.Table 7 indicates that perceived usefulness, effort expectancy, and perceived risk on continuance intention have path coefficient value below 0.1, indicating that they have no effect.While the other eight variables have a significant positive effect and perceived risk on continuance intention has a significant negative effect. ) If the R 2 value is around 0.75, it is said to be strong, around 0.50 is said to be moderate, and around 0.25 is said to be weak (Hair et al., 2017).According to the results in Table 8, the dependent variables are continuance intention and perceived usefulness, which can be said to be moderate.Meanwhile, satisfaction can be said to be strong.It can be interpreted that perceived usefulness, perceived risk, trust, effort expectancy, social influence, and satisfaction explain continuance intention moderately.In addition, confirmation moderates the perceived usefulness.While the confirmation, perceived usefulness, trust, effort expectancy, and perceived risk explain satisfaction strongly.

T-Test
Table 9 presents the results of hypothesis test on the structural model.The hypotheses were tested by a two-tailed test using the bootstrapping method with a significance level of 5%.If the t-value is greater than 1.96 then the hypothesis can be accepted.The results show that there are nine accepted hypotheses because the t-value is greater than 1.96.The other three were rejected because they had a t-value of less than 1.96.10 show that confirmation of perceived usefulness has a large effect.Then, the three pathways that have a moderate effect are confirmation on satisfaction, effort expectancy on satisfaction, and social influence on continuance intention.The three pathways that have no effect are perceived usefulness, effort expectancy, continuance intention, and perceived risk on continuance intention.While the other five variable relationships have a small effect.

Predictive Relevance (𝑄
2 ) The predictive relevance test must have a value larger than zero so that the variable has a predictive relationship with other variables (Hair et al., 2017).In Table 11, the overall value of the dependent variable, which is above zero, has a predictive relationship with other variables.

Relative Impact (𝑞
If the relative impact value is about 0.02 then it has a small effect, has a medium effect if the relative impact value is around 0.15 and has a large effect if the relative impact value is 0.35 (Hair et al., 2017).Based on Table 12, the relationship between confirmation and perceived usefulness which has a large effect, and social influence on continuance intention has a medium effect, perceived usefulness, effort expectancy, continuance intention, and perceived risk on continuance intention have no effect.While the other ten hypotheses have a small effect.

Model Fit
Model fit aims to assess how well the hypothesized model structure fits the empirical data to help identify model specification errors (Hair et al., 2017).The fit models used in this study were standardized root mean square residual (SRMR), chi-square statistic (χ²), and normal fit index (NFI).The results show that the SMRM is 0.051, which means that the suitability of this research model is said to be good (good fit).The chi-square value is 1609.276,which means that this research model has a good fit.While the NFI value, which is 0.845, which means the suitability of this research model is still acceptable (marginal fit).Values ranging from 0.80 < NFI < 0.90 are still acceptable (Khairi et al., 2021).The results obtained for the fit model are shown in Table 13.

Discussion
Based on the findings, there are three hypotheses supported from 12 hypotheses.Perceived usefulness, effort expectancy, and perceived risk have no effect on continuance intention.Trust, social influence, and satisfaction have an effect on continuance intention.The three factors that have a large effect on continuance intention are social influence, followed by satisfaction, and trust.
The results indicate that perceived usefulness is positively influenced by confirmation.Previous literature has also confirmed this relationship (Bhattacherjee, 2001;Franque et al., 2021;Mouakket, 2015;Yuan et al., 2016).Based on the results that have been obtained, it shows that users feel that their initial expectations are appropriate, either the suitability of the service level or service function, and that the user feels that the QR code service really has benefits, so that it affects the perceived usefulness factor.Furthermore, users feel that if their initial expectations are appropriate and fulfilled, it tends to increase satisfaction.Developers and service providers such as OVO, GoPay, and ShopeePay must provide quality services to exceed user expectations and make users feel satisfied.
Perceived usefulness determines satisfaction positively, but does not determine on continuance intention.Previous literature has also confirmed this relationship (Bhattacherjee, 2001;Lai et al., 2016;Susanto et al., 2016).This finding contradicts previous research on perceived risk and continuance intention (Bhattacherjee, 2001;Franque et al., 2021;Liébana-Cabanillas et al., 2021).Based on the results that have been obtained, it shows that users will feel satisfied if they feel useful or beneficial when using the QR code m-payment service, such as making transactions faster because scanning or showing the code is easier, and can be done anywhere and anytime.However, the perceived usefulness of users such as ease, speed, and effectiveness in transactions, does not affect continuance intention to use QR code m-payment service.When users feel the benefits obtained from the service, it is not necessarily the case that they will use it again.Developers and service providers further improve performance to be able to increase the effectiveness of user performance, including in terms of data security and transaction reader speed.
Trust has a significant positive effect on satisfaction and continuance intention.Previous literature has also confirmed this relationship (Cao et al., 2018;Kumar et al., 2018;Poromatikul et al., 2019;Susanto et al., 2016).Based on the results obtained, it shows that when users feel that the QR code m-payment service is trustworthy, secure, and reliable, it will foster a sense of satisfaction and increase continuance intention.Developers and service providers must continue to improve QR code services with more attention to convenience and can further improve the security system strictly to maintain user trust when transacting.
Effort expectancy has a significant positive effect on satisfaction, but has no effect on continuance intention.Previous literature has also confirmed this relationship (Marinković et al., 2020).However, these results are in line with previous research between effort expectations and continuance intention (Chopdar & Sivakumar, 2019;Zhao & Bacao, 2020).Therefore, the smaller the effort, the more user satisfaction there will be.However, effort expectancy does not affect continuance intention to use QR code m-payment.All of the items show that QR code m-payment service is easy to learn and use, but this convenience is not what affects on continuance intention.This can be due to direct transactions or cash, which can also be done easily, so that convenience is not their consideration.The recommendations that can be given are that service providers must update the ease of service, for example, by simplifying the payment process or transactions and being able to provide digital literacy to the public.
Perceived risk has a significant negative effect on satisfaction, but has no impact on continuance intention.Previous literature has also confirmed this relationship (Chen & Li, 2017;Yuan et al., 2016) and Liébana-Cabanillas et al. (2015) found that perceived risk has no impact on continuance intention.Based on the results that have been obtained, it shows that the perceived risk of users of the QR code m-payment service will affect user satisfaction.However, users are not worried about the risks and it will not affect the user's intention to continue using the service.This can happen if the majority of respondents come from gene Z.Gen Z is not too concerned with risk because most of them already know how it works and want to do things quickly, easily, and instantly.So, the recommendations that can be given are that service providers must pay more attention to and improve security and convenience to avoid scams, or fraud that can occur in QR code services.In addition, the company can also provide warnings and directions to users regarding the potential for crime.
Social influence has a significant positive effect on continuance intention.Previous literature has also confirmed this relationship (Gao et al., 2018;Lu et al., 2017).It shows that continuance intention is influenced by the social environment, such as family, friends, relatives, or co-workers who have used the QR code m-payment service and recommend it.Social influence has the biggest effect on continuance intention so that social influence is the main factor affecting continuance intention to use QR code m-payment services.Recommendations that can be given are more socialization or dissemination of information to the public to influence the public to use the QR code m-payment service and also collaborate with various merchants to better reach users.
Continuance intention is positively influenced by satisfaction.Previous literature has also confirmed this relationship (Joo et al., 2017;Kumar et al., 2018;Yuan et al., 2016).The author assumes that satisfaction in using the QR code m-payment service is one of them due to promos so that users feel happy and interested to continue using it.So, service providers must continue to ensure that the QR code service can function properly.Provide promos in the form of discounts or cashback for users when making transactions to provide user satisfaction.
This study has several limitations, based on the demographic results of participants, users of the QR code m-payment service are dominated by ShopeePay users, so this study explains more about ShopeePay users than OVO and GoPay.The results are also more dominated by students.The results of data collection also found 17 invalid data that had to be deleted.
Future research can use different research subjects on m-payment applications that have a QR code feature for transactions, such as the type of mobile banking application.The research model can be further developed, such as combining the ECM model with other models.Perceived usefulness, effort expectancy, and perceived risk have no effect on continuance intention, so that further researchers can review each of the indicators on these variables.

Conclusion
The concept of continuance intentions are still not described in detail in this context based on the IS study.To fill this gap, it is proposed to combine ECM and UTAUT to investigate the continuance intention to use QR code m-payment.The results represent the factors that affect continuance intention to use QR code m-payment services are trust, social influence, and satisfaction.Social influence is the factor that has the greatest effect on continuance intention followed by satisfaction, and trust.Trust, perceived risk, social influence, perceived usefulness, effort expectancy and confirmation determine satisfaction.But, perceived usefulness, perceived risk and effort expectancy and have no effect on continuance intention.Based on this, social influences such as family, friends, relatives, or co-workers have the most important role in affecting the continuance intention to use QR code m-payment.The QR code service providers carry out more socialization or disseminate information to the public to influence the public's intention to use the QR code m-payment service and collaborate with various merchants to better reach users and affect the continuance intention to use QR code m-payment.

Table 1 .
Demographic information of the participants

Table 2 .
Outer loading value

Table 3 .
Composite reliability and cronbach's alpha value

Table 4 .
AVE value

Table 5 .
Cross loading

Table 8 .
Coefficient of determination

Table 10 .
Effect size

Table 12 .
Relative impact

Table 13 .
Results of model fit