Predicting purchase decisions with different conjoint analysis methods - A Monte Carlo simulation

To forecast purchase decisions, different conjoint-based approaches have been discussed. Nevertheless, there is no clear evidence on which variant performs best. This study uses a Monte Carlo simulation to systematically compare different choice-based models and different models of a modified traditional conjoint variant, namely limit conjoint analysis (LCA), which allows for integrating choice decisions. All models compared, except the aggregate logit model, are rather robust. However, the hierarchical Bayes approaches perform best with both choice-based and limit data. The limit models are more efficient than those based on choice data. Thus, to predict purchase decision in practice, the limit hierarchical Bayes model should be considered first.