Multilevel (ML-ICLV) & Single Level Integrated Discrete Choice and Latent Variable (ICLV) Models Using Alternative Latent Structures' Conceptualizations

The aim of the present endeavor is to experiment on integrating discrete choice with latent variable (ICVL) models using alternative factorial structures’ conceptualizations and do so at both Single Level (Level 0) and Multilevel (ML-ICVL).

In doing, specific independent variables amenable to alternative latent variables’ conceptualization were selected. These included:
a) 1st-order latent variables (1st-order factors) (FM; FW),
b) 1st-order latent variables (1st-order factors) (FM; FW) forming a 2nd-order factor (F),
c) Multi-level (two-level) factorial structures (FML0; FML1 and FWL0; FWL1), and
d) Bi-Factor factorial structures (FM; FW; FG).

The results may be of use to researchers interested in using valid, reliable, and accurate structures of latent variables in ICLV models. We confirm that alternative latent structures of divergent factorial nature exist for the same observed variables, and may have different impact upon the dependent observed choice variable in the ICLV models. Second, DCE utility is conceptualized and estimated at both Level 0 and Level 1 and the differences are evident.