In structural equation modeling, the concern of researchers arises when they want to measure the effect of one or more variables on a categorical variable. The problem becomes more serious when the categorical variable has more than two dimensions.

 For example, you can define the Result of the match latent variable that measures via the Result measurement variable(See Introduction to SEM-PLS). You give the Result variable in a football match the numbers 0, 1, 2. This means that the Result variable is 2 if the football team wins,  1 if the game is tied, and  0 if the team loses. Now, suppose we want to examine the effect of the amount of training and performance on the match on the Result of the match. You may mistakenly convert a category variable to a numeric variable. In this case, with this definition of the Result variable, the results obtained will be completely wrong. It would be best to use Multinomial SEM-PLS to examine the effect of training and performance on the Result of the match variable. The SEM-PLS software allows you to perform the multinomial model using structural equation methods based on the PLS method(See Multinomial Regression and Path Analysis).