Comparing generalised maximum entropy and partial least squares methods for structural equation models

The generalised maximum entropy (GME) method is presented for estimating structural equation models, where a real data set of the Service&MotorVehicle Industry in Sweden is used to show the implementation of the method. Monte Carlo simulation comparisons are also made between GME and partial least squares (PLS) methods in the presence of messy data. Three cases are considered: outliers, missing data and multicollinearity. It is shown that this method can be considered a valid alternative to the conventional method of PLS, where the results of GME, in terms of mean squared error, outperform the PLS results in some respects.