An Assessment of Practical Solutions for Structural Equation Modeling with Complex Sample Data

Abstract This article discusses 5 approaches that secondary researchers might use to obtain robust estimates in structural equation modeling analyses when using data that come from large survey programs. These survey programs usually collect data using complex sampling designs and estimates obtained from conventional analyses that ignore the dependencies in complex sample data may not be robust. The results from a simulation study that examined 5 methods of estimation under 6 types of sampling designs and different population conditions are shared and applied analysts are encouraged to consider using the pseudomaximum likelihood for linearization estimation of asymptotic covariance matrices currently available in some software programs for structural equation modeling analyses.