Categorical Predictor Variables using SMART-PLS

Sometimes scholar(s) may need to include categorical predictor variables in their model. It is important to note that the categorical variables cannot be directly added to the model. Since the categories are represented by numbers and increase or decrease in the numbers do not depict an increase or decrease in the endogenous variable, the categories in the variable will have to be transformed. Here are the steps that one should follow when using categorical variables in SMART-PLS

• Create Dummy Variables for the categories in the variable. For example, let’s assume we have a variable Job Rank with three categories (Junior, Middle, Senior). Now we need to create two dummy variables (n-1). Although SPSS will give us three dummy variables, we will only add two in our model, the third one will serve as a reference category. The coefficients (effect) of the dummy variable are compared to the reference category.  (Procedure to create dummy variables is explained in detail in the video at the end of the notes.)
• Once, the dummy variables are created, for example there were three categories, and we created three dummy variables. Out of those three we only need to add two onto our model. In this case we add Middle and Senior. We are interested in assessing the impact of job rank on Commitment. The model will look like

• Next, Run Bootstrapping procedure, make sure you select path from weighting scheme. Here are the results from bootstrapping