This chapter presents a model describing the evolution of preferences as a stochastic process. These preferences are represented by weak orders, i.e. rankings with possible ties, on a set of alternatives, and can be modified under the influence of ‘tokens’ of information delivered by the environment according to a stochastic mechanism. The parameters of this mechanism can be estimated from the data and are descriptive of the environment. The potential effect of a token is to move an alternative up or down in an agent’s ranking. Attitude change is modeled by the stepwise transitions between the weak orders, which takes the form of a Markov process. The model permits exact predictions (up to a small number of parameters) of panel data in which the judges have been required to repeatedly evaluate the alternatives at times t 1,…,tn . An illustrative application of this model is described in a companion paper (Regenwetter, Falmagne, & Grofman, 1995). That illustration uses NES Thermometer (Rating) data on the 1992 presidential candidates.
Falmagne, J-C., Regenwetter, M., & Grofman, B. (2019). A Stochastic Model for the Evolution of Preferences. In A. A. J. Marley (Ed.), Choice, Decision, and Measurement: Essays in Honor of R. Duncan Luce (pp. 111-129). Routledge. https://doi.org/10.4324/9781315789408-8