구조방정식모형에서 비모수 부트스트랩 기법과 모수 부트스트랩 기법을 이용한 간접효과 유의성에 관한 연구 = A Study on The Significance of Indirect Effects Using Non-Parametric Bootstrap and Parametric Bootstrap in Structural Equation Modeling

다국어 초록 (Multilingual Abstract)

This study investigated the utilization of bootstrap techniques in assessing the significance of indirect effects within structural equation modeling. Non-parametric bootstrap was applied using raw data, while parametric bootstrap utilized correlation matrix data. The analysis revealed distinct differences in the significance levels of indirect effects between the first non-parametric bootstrap and parametric bootstrap methods, with the parametric bootstrap consistently yielding lower significance levels across all cases. Additionally, the parametric bootstrap exhibited a more stable and normal distribution-like pattern in all re-sampling distributions compared to the non-parametric bootstrap. Furthermore, in both non-parametric and parametric bootstrap methods, it was observed that as the number of re-sampling iterations increased, the significance levels decreased; however, beyond 500 iterations, both methods showed minimal changes in significance levels. Lastly, as the number of bootstrap iterations increased, both methods displayed stable re-sampling distributions. In conclusion, it was determined that the selection of bootstrap technique and the specification of the number of bootstrap iterations can influence the significance levels and statistical significance of indirect effects.