원문정보
초록
영어
The use of structural equation modeling based on the formative measurement has been increasing in the IS community. However, there has been the lack of discussions on the merits/demerits of formative measurement despite its requisite limitations in theory testing and knowledge build-up. Based on real survey data, we conducted rigorous empirical tests to uncover implications of formative measurement associated with interpretational compounding and weak external consistency. The findings are resounding: (1) The formative approach resulted in a significant interpretational compounding in which standardized weights of the measures of a formative construct widely fluctuated according to the change of dependent constructs; (2) The estimation of formative models is not necessarily consistent with the degree of external consistency highlighting the limitation of positioning the formative construct as a mediating, point variable; and (3) Formative models consistently inflated path coefficients and model fit indices increasing the chance of Type 1 error.
목차
Introduction
Theoretical discussions
Interpretational Confounding
External Consistency
Empirical demonstration
Interpretational Confounding
External Consistency
Conclusion
References