구조방정식 모델과 경로분석에서 측정오차가 경로계수에 미치는 영향에 관한 연구|RISS 상세보기

다국어 초록 (Multilingual Abstract)

Structural equation modeling and path analysis have a common character that can simultaneously estimate several causal relationships among variables, but these two technique are categorized by the existence of measurement errors. While structural equation modeling estimates path coefficient without measurement errors of observed variables, path analysis estimates path coefficient with them. To solve this limitation, several statistical techniques which calculating measurement errors in the path analysis model have been introduced, especially, ‘(1-α)×variance’ technique is generally used among them. However, very few empirical studies that examine the impact of ‘(1-α)×variance’ technique on path coefficients have been researched.
Based on this issue, in this study, data were classified as 4 samples (2×2) in terms of measurements reliability and variance that influence measurement errors, and then the statistical results of samples were compared. Also, Cronbach"s alpha and composite reliability were used as reliability coefficient individually in ‘(1-α)×variance’ technique.
The study shows that the results of ‘(1-α)×variance’ technique in path analysis are much similar to the those of structural equation modeling than those of common path analysis. Especially, as the measurement errors has increased, the path coefficients in ‘(1-α)×variance’ technique has shown accurate results. However, there is no big differences between when cronbach"s alpha used as reliability coefficient and composite reliability used as reliability coefficient in ‘(1-α)×variance’ technique.

번역하기