Modeling Preference Data with the Basic Local Independence Model

The link between two probabilistic models named "true and error" (TE) model and basic local independence model (BLIM) is provided. The TE has been proposed in 2007 in the research ?eld on preference studies, whereas the BLIM was proposed in 1988 in the area of knowledge assessment. It is proven in this article that the TE is a special case of an adaptation of the BLIM to forced-choice pair comparison tasks. Moreover, a new version of the model, named sBLIM, is presented in which speci?c constraints on the error parameters can be introduced. The BLIM and the sBLIM were compared in both simulations and empirical applications.Results suggest that these two models are suitable also for studying human preferences. In fact, three well-known studies have successfully replicated, that is Tversky's experiment (Tversky, 1969) on the intransitivity of preference and the two experiments on the positive andnegative prospects by Kahnemann (1979). Finally, some new results about the identifiability of the BLIM arise. In fact, it seems that some identifiability problems that are common in the application of the model to knowledge assessment studies do not occur in the study of preference.