In this paper we investigate explorative approaches to group comparison in PLS Path Modeling. Starting from the model estimated on the whole population, we intend to verify if different sub-populations lead to different path models and path coefficient estimates. We propose to face the problem by means of two different approaches. In the first one, a model building strategy is applied to sub-populations in order to verify if they belong to the same model; a procedure for path coefficients comparison can be then applied. In the second one, partial analysis precedes the PLS algorithm in order to estimate the model net of the effect of a categorical variable defining the groups; a comparison is then made between the partial models and the global one. In both approaches, the comparison will focus on communality/redundancy indexes.