ETAM에 의한 슈퍼컴퓨터 활용태도에 관한 실증적 연구

Abstract

We have come to the era that knowledge and information are essential to the international competitiveness of our country. The society of knowledge and information demands profound science and technology for international competitiveness. Furthermore, R&D institutes will need a lot of information for their research activities. The information for R&D requires higher level of measures in speed, performance and simulation. The supercomputer as a high performance computing system, supports to reach the required level of those measures. The supercomputer also contributes to improve the research environment of R&D institutes. The supercomputer can be used in the R&D organization if end-users are quite cooperative. Unfortunately, resistance from managers and researchers to accept the supercomputer has been quite strong. To get over this problem and to make end-users more familiar with the supercomputer, we should understand the factors behind their acceptance and rejection. It is believed that the stability of supercomputer infrastructure performane and the system level are influential factors for the end-users to realize the usefulness of supercomputer. It is also important to study the end-users' perception on supercomputer use, because this may give a clue how to effectively manage the use. In this study, the end-users' perception on supercomputer use is considered by applying Extended Technology Acceptance Model(ETAM). ETAM is widely used by researchers and practitioners to predict and explain the end user's acceptance of new information technology. It postulates that the two variables, perceived usefulness and perceived easy of use, are compelling factors in deciding the end-users' willingness to use. While some researchers follow the TAM, other researchers extend the model and apply it to explain adopting new information technology. But, it is unclear that what external variables would affect the attitude of usage and the intention to use. External variables such as objective system design characteristics, training, computer self-efficacy, and the nature of implementation process are proved to influence the intention to use, and ultimately to the usage, via their influence on the perceived usefulness and the perceived easy of use. In this study, external variables such as CEO support, the stability of supercomputer infrastructure performance and system level are used. Accordingly, this study focus on the perception of end-users for supercomputer actual use. This study has been conducted to explain peoples' attitude toward supercomputer actual use by measuring their intentions, perceived usefulness, perceived easy of use and other related variables. The validity of all the measures was assessed by the factor analysis. The factor-loading values for the measures were above the conventional acceptance level of 0.6. The reliability of all the factors was assessed by the Cronbach's alpha reliability coefficient. The data were analyzed using path analysis. A survey of 272 end-users(male 255, female 17) for a supercomputer was performed to test the suggested model. All the hypothesis proposed on the model were accepted and proved to be statistically significant. The followings are the summary of the results of our research. First, CEO support is posited to exert direct impact on perceived usefulness of supercomputer actual use. The path coefficient on this path is 0.392 and t-value is 2.258. Thus, Hypothesis 1(H1) is supported at the significance level. Second, perceived easy of use for supercomputer is statistically significant, derived from two factors such as the quality of supercomputer infrastructure and that of system level. The effect of supercomputer infrastructure quality on the perceived easy of use is stronger than that of system level. The path coefficient from a supercomputer infrastructure quality to the perceived easy of use is 0.583 and t-value is 4.360. The path coefficient from the system level to the perceived easy of use is 0.362 and t-value is 3.418. Third, the perceived usefulness is influenced by one path from the perceived easy of use, and it is statistically significant. The path coefficient from the factor is 0.263 and t-value is 2.293. Forth, the perceived usefulness, with the path coeffcient 0.132, affects the preference for the supercomputer actual use. The t-value is 2.088 Fifth, the perceived easy of use , with the path coefficient 0.507, influences on the preference for supercomputer actual use. The t-value is 3.988. Sixth, the support of CEO, with the path coefficient 0.230, affects the supercomputer actual use. Its t-value is 3.939. Seventh, the preference for supercomputer, with the path coefficient 0.384, influences on the supercomputer actual use and its t-value is 6.275. Therefore, the perceived easy of use directly impacts on the preference for supercomputer more stronger than does the perceived usefulness. Also, the perceived usefulness directly influences on the intention to revisit supercomputer actual use. As a result, all the hypotheses proposed on the model were proved to be statistically significant. The future directions of research are suggested as follows. First, although the results show that supercomputer infrastructure quality and system level affect the perceived easy of use, it is important to realize that other external factors may also play an important role in affecting the perceived easy of use. Second, in the study, moderating variables such as skill, challenge and championship as personal characteristics are used. In the future research, hopefully other variables can be included beyond those personal characteristics. In conclusion, we hope that R&D organizations can utilize this result in improving and developing the use of information technology to get international competitiveness.