FACTORS AFFECTING USER BEHAvIOR OF TECHNOLOGY APPLICATION OF MANAGEMENT INFORMATION SYSTEM

This research aimed to determine the factors influencing the behavior of management information system users at the National Accreditation Committee. The theory used was based on the Technology Acceptance Model (TAM) theory extension, and the analytical method used was the Structural Equation Model (SEM). The factors were according to Davis's modified theory, where external factors were the quality of internet connection and leadership support. Internal factors included perceived ease of use and perceived usefulness, attitudes toward using, behavioral intention to use, and actual system use. This research was conducted on 128 respondents representing the Conformity Assessment Bodies (CABs). This research showed that leadership support had a significant effect on perceived ease of use and perceived usefulness. Meanwhile, the quality of the internet connection only affected the perceived usefulness. Leadership support, quality of internet connection, perceived ease of use, perceived usefulness, and attitudes toward using were factors that affected user behavior intention to use. This research's managerial implication is to strengthen the server and increase the leadership's motivation towards the CAB employees by socializing accreditation applications.


INTRODUCTION
Currently, the development of technology and industry proliferates. In terms of service to consumers, companies and governments continue to innovate to provide the best service for technology-based service users. The National Accreditation Committee (KAN) is a government institution responsible for providing accreditation services for Conformity Assessment Bodies (CABs). KAN always tries to improve services from year to year. This is shown by the number of CABs accredited in 2018, reaching 2019 CABs. Meanwhile, the number of KAN secretariat personnel was only 100 people. The limited number of human resources causes users in CABs to complain about accreditation services, such as the delay in the KAN secretariat response. Therefore, since 2015, KAN has made the KAN Management Information System application technology called KAN MIS to provide accreditation services for CABs. However, the use of this application has found difficulty. Until 2019, there were still 5% of CABs in inspection bodies and certification bodies that have not used the KAN MIS application technology. There were also complaints regarding the internet connection quality, hard-to-understand use of the application (perhaps due to being absent in the socialization), the lack of leadership support to try new technologies, and some not functioning features such as limited download capacity.
The issue of internet connection quality was considered to influence the perception of the ease and the perceived usefulness of the KAN MIS application because the CABs located all over Indonesia had a problem with internet network when uploading the accreditation documents. This problem can be seen from the CAB complaints submitted in the KAN service satisfaction survey. In addition, the CAB leadership support was also an issue that affects the perceived ease of use and perceived usefulness as technology will be successfully implemented if the leader provides support for the application of new technology. Venkatesh et al. (2003) studied the unified theory of acceptance and use of technology (UTAUT) formulated with four core determinants of behavioral intention and usage, namely performance expectancy, effort expectancy, social influence, and facilitating conditions. Neufeld and Dong (2007) studied the impacts of charismatic leadership on the four UTAUT beliefs that, in turn, influenced behavioral intention and use. Dumpit and Fernandez (2017) studied the Technology Acceptance Model (TAM) about the impact of the quality of internet connection on perceived usefulness and perceived usefulness, and perceived ease of use; the impact of subjective norm on behavioral intention to use and actual system use. Al-Somali et al. (2009), which studied the constructs developed based on TAM, suggested that the internet connection quality had a direct relationship with perceived ease of use. Schepers et al. (2005), which also studied the constructs extended based on TAM, suggested that leadership style affected the perceived ease and perceived usefulness, especially in charismatic/transformational leadership, about how leaders inspire and motivate. Venkatesh et al. (2012) added the impacts of hedonic motivation, price value, and habit on behavioral intention and technology use as moderated by age, gender, and experience.
The application of new technology was not easy, according to the Technology Acceptance Model (TAM) developed by Davis in 1989, due to the external factors/variables, perceived usefulness (PU), perceived ease of use (PEOU), attitude toward using (ATU), behavioral intention to use (BI), and actual system use (AU). The external variable includes all variables not included in the system model (Davis 1985). There were several external variables in the researches of the TAM extension. For example, age, level of education, organizational employment category (Hubona and Kennick 1996), charismatic leadership (Schepers et al. 2005;Neufeld and Dong, 2007), quality of internet connection (Al-Somali et al. 2009;Fonchamnyo, 2013;Dumpit and Fernandez, 2017).
This research employed two external variables, namely the quality of internet connection (QIC) and leadership support (LS). QIC is an important component of internetbased applications (Fonchamnyo 2013). LS refers to a transformational leadership style with charismatic characteristics, which has individual consideration, encourages intellectual, inspires, and motivates (Schepers et al. 2005). The internal variables in this research consisted of PEOU, PU, and ATU. PEOU is a person's level of confidence that using a certain technology/system will make the business easier (Saade and Bahli, 2005). PU is the degree to which someone believes the system will contribute to achieving certain goals (Lu et al. 2003). ATU is an attitude towards the use of a system in the form of acceptance or rejection as a result when someone uses a technology/system in their work (Fishbein and Ajzen, 1975). This research focused on the factors influencing the KAN MIS' users' behavior, namely the CABs, inspection bodies, and certification bodies.

METHODS
This research was conducted at KAN, located at Menara Thamrin, 11th floor, Jl. M. H. Thamrin Kav. 3 Kebun Sirih, Central Jakarta 10340. This research was conducted for three months, from March 2020 to May 2020.
The data used in this study were primary and secondary data. Primary data were obtained from questionnaires given to CABs, inspection bodies, and certification bodies as accreditation services users. Secondary data were obtained by taking data published by KAN and BSN and conducting literature studies on theses, international and national scientific journals, and the internet. This research utilized online questionnaires because the CABs officer locations in this study were all over Indonesia. The answers to questionnaire questions were in the form of assessments from respondents measured using a Likert scale of 1 to 5. The assessment started from strongly-disagree with the lowest weight to strongly-agree with the highest weight. This research was a quantitative research with variables referring to the TAM external variables in the form of quality of internet connection (QIC) and leadership support (LS), perceived usefulness (PU), perceived ease of use (PEOU), attitude toward using (ATU), the tendency of user behavior to continue to use technology or behavioral intention to use (BI), and actual system use (AU). This study's sample was determined using purposive sampling with the following criteria: having CABs certification body scheme and the CABs inspection body scheme registered in the KAN directory; owning active numbers enabled to access the questionnaire through Google Form; having experience in using the KAN MIS application. The number of respondents in this study was calculated using the Slovin formula, with the total number of population members as many as 188 CABs, inspection bodies, and certification bodies. It resulted in 128 respondents representing CABs nation-wide office locations. The questions sent to the respondents had previously been tested for validity with product-moment correlation and reliability by comparing the Cronbach's alpha value using IBM SPSS Statistics 21 software.
Structural Equation Modeling (SEM) was applied in this research due to its higher flexibility for social science researchers to connect theory and data (Ghozali, 2014). This study used the Partial Least Square (PLS) approach as it is not based on many assumptions, and the number of samples does not have to be large (around 100). Statistical calculations were assisted by SmartPLS version 2 software. The analysis stages using the PLS approach started from the measurement model (outer model), which specified the relationship between the latent variables and the manifest variables/ indicators, to the structural model (inner model), which specified the relationship between latent variables based on substantive theory and hypothesis testing.
The outer model assessment was conducted by determining the convergent validity of each construct. Convergent validity is used to measure the validity of each indicator by looking at the loading factor value. A value of outer loading or loading factor is said to be valid if it correlates more than 0.5 on the measured variable. The next discriminant validity assessment was the degree of conformity between variables or attributes measured by measuring instruments and theoretical concepts. Reflective discriminant validity can be used in several measures, such as cross-loading and comparing the square root of average variance extracted (AVE) value of each construct with the correlation between constructs and other constructs in the model. If the AVE square root value of each construct is higher than the correlation value between the construct and other constructs in the model, it is said to have good discriminant validity. Discriminant validity is assessed based on the cross-loading measurement with the construct. Cross-loading is used to determine the relationship between variables and indicators.
Composite reliability measures internal consistency. Its coefficient uses confirmatory factor analysis in the structural equation modeling approach. The use of composite reliability is approved to have a better advantage than usual reliability. The accepted limit value of composite reliability is 0.7.
Inner model evaluation describes the relationship between latent variables. It aims to see the relationship between constructs and the significance value and R-square (R2) value. The relationship between constructs can be seen from the path coefficient estimation, which is the estimated value for the structural model's path relationship. The significance value can  (2013) and Sholiha and Salamah (2015).
This research referred to the TAM theory presented by Davis in 1989 ( Figure 1). The use of TAM, apart from explaining a system, can be accepted or rejected. This model provides input to improve the technology system to make it more acceptable to its users. TAM is the best approach to explain and predict the acceptance of information system users (Wang et al. 2005).
This study used the TAM model flow, as shown in Figure 2. This research is a TAM theory extension with external variables consisting of quality of internet connection and leadership support. Combining these two external variables has never been done in previous research and was considered appropriate to be applied to the problems in this study. In this study, there were seven studied variables: quality of internet connection, leadership support, perceived ease of use, perceived usefulness, attitude toward using, behavioral intention to use, and actual system use. The significance level used in this study was 5%. The hypothesis was accepted if the t-value was more than 1.96 and rejected if the t-value was lower than 1.96. Figure 2 shows the seven studied variables: (1) quality of internet connection (QIC) consisting of 4 indicators: ease of internet access (QIC 1), the ability of the internet network to handle the online accreditation process accurately (QIC 2), internet network capability that supports 24 hours (QIC 3), and internet network capability guarantees all processes are complete (QIC 4) , according to Al-Somali et al. (2009) and Dumpit and Fernandez (2017); (2) leadership support (LS) consisting of 4 indicators: leadership instructions to use the system (LS 1), the leader thinks the system will save costs (LS 2), motivation from the leader (LS 3), and the leader provides the opportunity to training (LS 4), according to Schepers et al. (2005), Neufeld and Dong (2007), Venkatesh et al. (2016); (3) perceived ease of use (PEOU), consisting of 6 indicators: easy to use (PEOU 1), easy to learn (PEOU 2), interaction with the system is clear and understandable (PEOU 3), easy to control (PEOU 4), easy to understand (PEOU 5), and requires skills (PEOU 6), according to Davis et al. (1989), Wang et al. (2005), and Nasri and Charfeddine This hypothesis is supported by Schepers et al. (2005). 5. H5: Perceived ease of use (PEOU) affects perceived usefulness (PU) in using the KAN MIS application. This hypothesis is supported by Davis et al. (1989), Wang et al. (2005), Shyu and Huang (2011), and Manalu et al. (2020). 6. H6: Perceived ease of use (PEOU) affects attitude toward using (ATU) in using the KAN MIS application. This hypothesis is supported by Davis et al. (1989), Shyu and Huang (2011), and Nasri and Charfeddine (2012). 7. H7: Perceived usefulness (PU) affects behavioral intention to use (BI) in using the KAN MIS application. This hypothesis is supported by Davis et al. (1989), Wang et al. (2005), and Shyu and Huang (2011) 8. H8: Perceived usefulness (PU) affects attitude toward using (ATU) in using the KAN MIS application. This hypothesis is supported by Davis et al. (1989) and Nasri and Charfeddine (2012). 9. H9: Attitude toward using (ATU) affects behavioral intention to use (BI) in using the KAN MIS application. This hypothesis is supported by Davis et al. (1989)

RESULT
The tendency of user behavior to continue to use (behavioral intention to use) the KAN MIS application was reflected from the conformity assessment bodies intending to use, choosing to use, and being committed to using the KAN MIS application. The actual system use was indicated by the categories of satisfaction of the conformity assessment bodies in use, increasing the intensity of use, each month will use, frequently use, and continue to use the KAN MIS application. The results of the evaluation of the outer model were as follows: If there were indicators that had a loading factor value of < 0.5, a recalculation of the model must be carried out so that the loading factor of all reflective indicators was > 0.5 as a criterion for the validity test of convergent latent constructs (Ghozali, 2014). Also, the model requirement had good validity if each latent variable with a reflective indicator had an average variance extracted of > 0.5. Figure 3 shows that all indicators had a loading factor value of > 0.5, which means that all indicators had met the convergent validity requirements. Table 1 shows that the average variance extracted value of each latent variable had a value of > 0.5, which means that the PLS model fulfilled good convergent validity requirements.
The next measurement, namely the reliability test, was used to prove the consistency and accuracy in measuring the construct. The reliability test was conducted by measuring composite reliability and Cronbach's alpha for latent variables. The ones with the value of > 0.7 were reliable. Based on Table 1, all latent constructs had a good, accurate, and consistent reliability because they met the requirements with the composite reliability and Cronbach's alpha values of > 0.7 for each latent construct.
The discriminant validity test was carried out on the principle that the gauges (manifest variables) of different constructs should not be highly correlated (Ghozali, 2014). Table 2 shows that the correlation value between constructs as a whole was less than the root value of the average variance extracted. It could be said that the model has met the conditions for discriminant validity. It means that the indicators in the latent construct were correct in measuring the latent variables.
The goodness of fit obtained in this research model was 0.799. The value was more than zero, which means that this model was relevant as a prediction model. The bootstrapping results in Table 3 can answer the hypothesis in this study with the following explanation:

Evaluation of the Structural Model (Inner Model)
The structural model is used to describe the relationship from each variable to another and indicators to variables. The structural model is evaluated by looking at the significance of the relationship between the constructs. This can be seen from the path coefficient, which describes the strength of the relationship between the constructs. The sign in the path coefficient must be in accordance with the hypothesized theory, and the significance of the path coefficient can be seen from the t-statistic value obtained from the bootstrapping process (resampling method). The PLS structural model in this study is portrayed in Figure 3.

Effect of Quality of Internet Connection on Perceived Ease of Use and Perceived Usefulness
Quality of internet connection had a significant influence on perceived ease of use with t-statistical value > t-table (1.96) at a significance level of 5%. Quality of internet connection had an effect on perceived ease of use by 0.432, which means the better the quality of internet connection (reflected by its four indicators), the better the perceived ease or perceived ease of use will be. In the study, the internet connection quality did not affect the perceived usefulness because of the t-statistic value (0.69) ≤ t-table (1.96) at the significance level of 5%.

The Effect of Leadership Support on Perceived Ease of Use and Perceived Usefulness
Leadership support had a significant influence on perceived ease of use with a t-statistic value > t-table (1.96) at a significant level of 5%. Leadership support had an effect on perceived ease of use by 0.468, which means that the better Leadership support is reflected by the indicator of leadership instructions to use the system, the leader thinks that the system will save costs, motivation from the leadership, the leadership provides opportunities for training, the better the perceived ease of use will be. The structural model of perceived ease of use generated an R-square value of 70.4%. It means that the diversity of perceived ease of use that the model could explain was 70.4%, while other factors outside the model explained the remaining 29.6%. Leadership support had a significant effect on perceived usefulness with statistical value > t-table (1.96) at a significant level of 5%. Leadership support had an effect on the perceived usefulness by 0.329, which means that the leading indicator reflects the better leadership support for using the system, the leader considers that the system will provide costs, motivation from the leader, and the leadership provides opportunities for training, thus increasing the perceived benefits.

Effect of Perceived Ease of Use on Perceived Usefulness and Attitude Toward using
The perceived ease of use had a significant effect on the perceived usefulness with the t-statistic value > t-table (1.96) at a significant level of 5%. The perceived ease of use affected the perceived usefulness by 0.608, which means that the better the perception of ease of use (reflected by its six indicators), the better the perception of usability will be. The perceived structural usefulness resulted in an R-square value of 82.8%, meaning that the perceived diversity of usefulness that the model was able to explain was 82.8%. In comparison, the remaining 17.2% was the fact by other factors outside the model. The perceived ease of use had a significant effect on attitudes towards using because of the t-statistic value > t-table (1.96) at the significance value of 5%.

Effect of Perceived Usefulness on Behavioral Intention to Use
Perceived usefulness had a significant influence on behavioral intention to use with a t-statistic value > t-table (1.96) at the significance value of 5%. Perceived usefulness affected behavioral intention to use by 0.423, which means that the better the perceived usefulness (reflected by its eight indicators), the increased the behavioral intention to use will be. The structural model of behavioral intention to use generated an R-square value of 83.8%, meaning that the diversity of perceived usefulness that could be explained by the model is 83.8%, while other factors outside the model explained the remaining 16.2%.

Effect of Perceived Usefulness on Attitude Toward Using
Perceived usefulness had a significant effect on attitude toward using the t-statistic value > t-table (1.96) at the significance level of 5%. Perceived usefulness had an effect on attitude toward using by 0.830, which means that the better the perceived usefulness (reflected by its eight indicators), the better the attitude toward using will be. The structural model of attitude toward using generated an R-square value of 83.9%, which means that the diversity of attitude toward using that could be explained by the model was 83.9%, while the remaining 16.1% was explained by other factors outside the model.

Effect of Attitude Toward Using on Behavioral Intention to Use
Attitude toward using had a significant effect on behavioral intention to use with a t-statistic value > t-table (1.96) at the significance level of 5%. Attitude toward using had an effect on behavioral intention to use by 0.513, which means that the better the attitude toward using (reflected by its three indicators), the better the behavioral intention to use. The structural model of behavioral intention to use generated an R-square value of 83.8%. The diversity of actual system use that the model could explain was 83.8%, while other factors outside the model explain the remaining 16.2%.

Effect of Behavioral Intention to Use on Actual System Use
Behavioral intention to use had a significant effect on actual system use with the t-statistic value (1.149) > t-table (1.96) at the significance level of 5%. Behavioral intention-to-use affected the actual system use by 0.843, meaning that the better the behavioral intention (reflected by its three indicators), the better the actual system use will be. The structural model of actual system use generated an R-square value of 71.1%, which means that the actual system use diversity that the model could explain was 71.1%, while the remaining 28.9% was explained by other factors outside the model.

Managerial Implication
The results of this study provide several inputs that can be applied by management. The practical implication is to improve internet network capabilities by strengthening the server capacity of the KAN MIS application with internet network investment, increasing the motivation of the leadership towards the employees of the CABs with the socialization of the use of the KAN MIS application on a regular basis involving the leadership of the CABs in the framework of updating the application, increase the convenience in using the application. Top management is expected to receive input from this research so that the KAN MIS application can be more useful in the future.

Conclusions
This study showed that the factors that significantly influenced the behavior of using the KAN MIS application were the quality of internet connection, leadership support, perceived ease of use, perceived usefulness, and attitudes toward using. By extending TAM, this study revealed results that support arguments made by other researchers claiming that quality of internet connection affected perceived ease of use, and leadership support (charismatic or transformational leadership style) affected perceived ease of use and perceived usefulness. This study also confirmed perceived ease of use and perceived usefulness as the determinants of attitude toward using that affected behavioral intention to use and actual system use.