Purpose: This study conducted a Monte Carlo simulation to deal with the issues of alternative methods to replace the composite scores, which were the integrated scores of indicators, with a measurement model for testing in order to solve the problems with model estimation in a research that applied SEM. Methods: The investigator directed the simulation to compare the estimated structural coefficients from the original model containing with those in the reduce model that was replaced with the composite scores or integrated scores of those indicators. The simulation was carried out in three cases according to the organizational methods of composite scores that considered the correlation coefficient and factor loading characteristics of indicators. Results: first, the invariance of the estimated structural coefficient wasn’t maintained when the factor coefficients of indicators were heterogeneous in case of a unidimensional structure among indicators. Secondly, a structural coefficient was underestimated regardless of heterogeneity among indicator factor coefficients in case of a multidimensional structure among indicators. The study also showed the effects of composite scores on the estimation of structural coefficients by comparing their estimation results according to the sample size of analysis data in each experiment.