Identifying the Factors of Online Game Acceptance Using Technology Acceptance Model

In the last two decades, many companies create online games. Online game is a game that is connected to the internet, where players can play and communicate with other players in different places at the same time. There are many types of games, such as adventure, strategy, cooking, make up, and so on. That is why a lot of people love to play online game. This study tries to identify the factors that support online game acceptance by students of Atma Jaya Yogyakarta University, because many students at this university play online games. The method used for this analysis is Technology Acceptance Model, which has main construction namely Perceived easeof-use (PEOU), Perceived usefulness (PU), Attitude toward use (ATU), Intention to use (ITU) and Actual Use (AU), and additional variables of Social Influence (SI), Personal (P) and Excitement (E) added by the researchers. The data obtained to conduct this analysis used a questionnaire addressed to students. There are eleven hypotheses that serve as a reference in analysing the relationship between variables. The result shows that each variable has a significant relationship in online game acceptance.


Introduction
Online games are games that are connected to the internet network. Online games began to appear in Indonesia in 2001, with the launch of Nexia Online, an RPG game released by BolehGame with simple Other researchers look at online games from the other side. Online games can be a part of learning, helping school students and college students learn and gain knowledge through online games [17]. This study can determine the perception of students in accepting online games for education [18]. The results show that many students enjoy online games for their education or learning media [17], [19]. This can help the university in increasing the knowledge of their students.

Research Design
This study uses TAM to determine the variables needed in the analysis, both internal and external variables. Internal variables consist of Perceived ease-of-use (PEOU), Perceived usefulness (PU), Attitude toward use (ATU), Intention to use (ITU) and Actual Use (AU), and additional (external) variables consist of Social Influence (SI), Personal (P) and Enjoyment (E).
This research was conducted on certain representative populations and samples. In this study, the population was directed to Universitas Atma Jaya Yogyakarta (UAJY) that has 11,307 students. The sample in this study refers to UAJY students who actively play online games. This sampling used simple random sampling. The number of samples was calculated using the Slovin formula, with a critical value of 5%, and the results obtained were 386 people. This has met the sample size in the Structural Equation Model (SEM) with the estimation model using a minimum Maximum Likelihood (ML) of 386 samples.

Research Questionnaire
The statements for the questionnaire can be seen in Table 1. The questionnaire was compiled based on eight previously defined variables. Each variable consists of several statements, which later the respondent will give an opinion on the statement, whether they agree, neutral, or disagree. Online games are easy to play.
[20] Online games are easy to learn.
[12] It is easy for me to become proficient in playing online games. Perceived usefulness (PU) Playing online games helps me to improve my ability to play online games.
[10] Playing online games is very important in my life.
[20] I love to spend my free time playing online games. Online games are efficient to fulfill my needs.
[21] Attitude toward use (ATU) I responded positively about the existence of online games.
[10] I like playing online games.
[12] I think playing online games is a good idea.
[20] Intention to use (ITU) I will continue to play online games [13] I intend to play online games I will be playing online games for a long time I will recommend to others to play the online games I play [21] Actual Use (AU) Playing online games is a solution for me to get rid of boredom.
[20] I like the game in the online games.
[14] I like the services provided in online games. I am satisfied playing online games. Social Influence (SI) My friends think that I should play online games.
[12] My playmates think that I should play online games. My college friends think that I should play online games.
[13] Personal (P) I am confident in my ability to play online games.
[10] I have the necessary skills in playing online games. Enjoyment Playing online games is interesting to me. [13] Indonesian Journal of Information Systems (IJIS) Vol.

Research Model
Eight variables, consisting of Perceived ease-of-use (PEOU), Perceived usefulness (PU), Attitude toward use (ATU), Intention to use (ITU), Actual Use (AU), Social Influence (SI), Personal (P), and Enjoyment (E) are modeled to develop hypotheses. This model can be seen in Figure 1.

Data Analysis
The data analysis technique used in this analysis is a quantitative analysis using the Structural Equation Model (SEM). SEM is a multivariate statistical analysis technique that analyzes structured relationships. This technique is a combination of factor analysis and multiple regression analysis [31]. In this analysis, there are several stages, namely the descriptive analysis stage, the measurement analysis stage, and the structural analysis stage. Descriptive analysis was carried out using SPSS Ver.20, useful for providing an overview to the reader so that it could be understood easily, while measurement analysis was carried out using AMOS Ver.20, useful for testing whether the model used was in accordance with the data obtained, and lastly, structural analysis was also carried out using AMOS Ver.20, useful for testing the relationship between the analyzed factors and proving the hypothesis.

Data Collection Result
Data collection was carried out by distributing questionnaires to UAJY students using online (32.4%) and offline media (67.6%). The questionnaire was distributed to all study programs at UAJY and successfully collected as many as 500 responses.

Screening Process Results
This screening process aims to filter the data and to avoid invalidity in data analysis. The first process is to check the standard deviation with the result of deleting thirteen data because these data have a standard deviation of zero and are interpreted to have the same value. The next process is checking for missing data. The results obtained in Table 3 indicate that no data is empty, which means that all data has been provided.

Degree of Freedom Process Results
The relationship between the degree of freedom in this analysis before the model testing is carried out is the understanding of model identification. Identification of a model indicates whether there is sufficient information available to identify a solution from a structural understanding. The test result seen in Table 4 is showing positive result which indicates that the data can be used for the analysis.

Assessment and Estimation Process Results
The assessment is intended to determine the extent to which the data 'fit' with the model that has been made; whether the model is valid, and the sample data taken can show the strength of a model in explaining an event or phenomenon. Meanwhile, the estimation is used to see the strength of the relationships between variables in the model [31]. The assessment process is carried out using the Maximum Likelihood Estimation (MLE) technique, which is based on the covariance matrix of the sample with the population. The results obtained indicate that the relationship among the variables is quite close and the direction of some relationships is in the same direction, and some are different.

Outlier Test Results
Outlier test is a test on data that appears and has unique characteristics that are far different from the other data and appear in the form of extreme values. From the tests that have been carried out, there were 26 outlier data, so the researcher deleted the data. The results can be seen in Figure 2.

Validity and Reliability Test Results
This test is done by testing the validity and reliability of each instrument. The purpose of testing the validity and reliability is to ensure that the questionnaire that has been compiled will be able to measure symptoms and produce valid data. The results obtained from this validity test (Table 5) are valid because it has a significance value less than 0.05 and for reliability testing (Table 6) it has a Cronbach's alpha value greater than 0.6.

Measurement Model Analysis Results
A measurement model measures the goodness of fit of the model. This measurement was carried out using Cmin/df, RMR, RMSEA, GFI, AGFI, NFI, and CFI (Table 7). This measurement is carried out to ensure that the measurement model meets the criteria of goodness of fit. In previous studies [32], RMSEA and RMR values below 0.08 were said to be sufficient to proceed to hypothesis testing. The convergent validity test aims to ensure that the indicators that are theoretically related to a factor have a high correlation. In this study, the validity test was proven by looking at the Average Variance Extracted (AVE) value and the loading instrument value for each factor. The criterion for achieving convergent validity is that the AVE value of each factor exceeds 0.5. For the loading factor value, all indicators have a value of more than 0.5 and are significant [33]. The results obtained indicate that the loading and AVE factor values have met the criteria. Then proceed with the discriminant validity test by looking at the square root value of the AVE which is greater than the latent factor. The results obtained indicate that the values of several variables have the square root value of AVE greater than the latent factor, but several variables show numbers greater than 0.5. This is still acceptable and if it is squared it is still below the value of the square root of the AVE and proves that each variable is indeed different from one another [31].
The next process is reliability testing. This test uses composite reliability and Cronbach's alpha as testing methods. The value of most composite reliability has exceeded the recommended value, which is 0.7 [33]. However, the limit value cannot be applied explicitly, but must also consider several aspects [34].

Structural Analysis Results
Structural analysis is useful for seeing the relationship between the dependent variable and the independent variable. However, before conducting a structural test, there are initial assumptions that must be met. These assumptions are the assumptions of normality, multicollinearity, and homoscedasticity. The normality assumption aims to find out that the data is close to the normal assumption by looking at the range of values from -2.58 to +2.58 and the results obtained indicate that the data is normally distributed. Then multicollinearity aims to determine the relationship between the independent variables by looking at the VIF value of less than ten and tolerance of more than 0.1 and the results show that there is no multicollinearity in the data. Next is the homoscedasticity test using a graph and the results show that the points are spread out and do not form a certain pattern which means that each indicator group is in the same variance among the members of the group.
This structural analysis resulted in a better goodness of fit value, after improvements were made from the previous model. The refined model can be seen in Figure 3. The structural model that has been refined connects external factors, namely social influence (SI), personal (P), and enjoyment (E).

Figure 3. Structural Model
After the model is refined, there is an increase in the goodness of fit value from the initial to the final model. The RMR and RMSEA values are getting smaller and the GFI, AGFI, and NFI numbers are getting closer to 1 which indicates that the model is 'fit'. These values can be seen in Table 8.

Hypothesis Test
Hypothesis testing was carried out using the AMOS version 20 application. This test assessed the relationship between factors by looking at p as significance. The hypothesis test model can be seen in Figure 4.

Figure 4. Hypothesis testing model
Hypothesis testing is done by looking at the significant value less than 0.05 and the CR value more than 1.96. After knowing the relationship between variables, then proceed with finding out the strength of the relationship between these variables. Numbers above 0.7 or above 0.5 are generally used as a reference for the closeness between two variables [31]. In addition to the analysis of the relationship between constructs, the researcher also added testing of the relationship between exogenous variables. In the model, there are three exogenous variables, namely social influence (SI), personal (P), and enjoyment (E). This relationship has a significance value of less than 0.5, so there is a significant relationship between these exogenous variables. Then proceed with the test of the strength of the relationship between exogenous variables. Each variable has a close relationship. However, the relationship between the social influence (SI) and the enjoyment (E) variable, is still said to be less close because it has a correlation value of less than 0.5. After conducting several tests to see the relationship between variables, it can be concluded that the overall hypothesis test results can be seen in Table 9 and Table 10.  The level of ability in playing online games is different so that a person's ability to play online games does not significantly affect the usefulness of online games. A person's ability to use technology is influenced by that person's experience [35]. H2: PEOU affects ATU Yes Accepted The ease of playing online games can affect students' attitudes towards existing online games, such as responding positively and if playing online games is a good idea [21].

H3: PU affects ATU
Yes Accepted Students need something they might not be able to achieve in real life, such as building a city or building an empire. The game can be used to realize these needs. Thus, students responded positively to the existence of online games