Latent variable transformation using monotonic B-splines in PLS Path Modeling

PLS Path Modeling is a widely used approach in marketing. In that field, relationships between latent variables are frequently nonlinear. This nonlinearity is usually defined by a piecewise linear function. In this talk, we present an approach to include non linear transformations of the latent variables in the PLS path modeling algorithm with monotonic B-splines. We use optimal scaling methods to obtain an optimal transformation of the latent variables explaining a target latent variable. Applications on customer satisfaction data and on simulated data are presented.