How to Capture Model(Mis-)Specification in Structural Equation Modeling

摘要

How well can fit and construct validity criteria capture mis-specified models and which combination of fit and construct validity criteria is the best to detect misspecified models while retaining many correctly specified models? Data are generated by means of a comprehensive Monte Carlo study.We use 7 experimental factors(misspecification severity,location of misspecification,number of exogenous constructs,number of endogenous constructs,average number of items per construct,sample size,and average indicator loading per construct)and consider 3,060 different constellations.In total,we arrive at 153,000 models.