In this study, seven two-stage approaches in detecting continuous manifest moderating effects on construct relationships are compared in terms of bias and power by using Monte Carlo simulation. The simulation is conducted for each of the conditions formed by different skewness, kurtosis and intercorrelations for exogenous constructs and the moderator, different factor loadings, error variances, and sample sizes under specified models. The results indicate that different approaches tend to produce little bias difference but lead to power difference, varying with different situations. Recommendations for the proper use of the two-stage
approaches are provided based on the simulation results obtained.