Essays on Optimal Sequential Logit: A Discrete Choice Model of Preplanned vs Impulse Purchases



Journal Title

Journal ISSN

Volume Title



This dissertation is an effort to develop the Optimal Sequential Logit (OSL) model of sequential decisions that assumes the decision-maker anticipates future choice situations even when the details are not known with certainty. In the first essay, I elaborate on the model development by explaining the setting, the assumptions, the utility and probability functions. Then, I apply OSL to a consumer panel data of grocery shopping decisions. In the second essay, I compare OSL with sequential logit model, which I refer to as “myopic” sequential logit (MSL), to test the assumptions of the two models. OSL assumes that the decision maker is forward-thinking which means that she anticipates the unknown parts of her utility for the future choice situation and this affects her choice in the first stage while MSL assumes the decision-maker is myopic meaning that she does not have any anticipation of the second stage and/or she makes her decision independent of what she expects to happen in the future. I use simulated datasets with different percentages of myopic and forward-thinking decision makers to assess the performance of OSL and MSL in modeling choice behavior. I also apply MSL to the panel data used in essay 1 to compare estimation results in both models on real consumer choice data. In the third essay, I compare OSL to nested logit model (NL) on simulated and real data. On simulated data, I test the ability of the models in capturing the sequential nature of the decision. I use simulated data generated by sequential framework to compare OSL and NL in explaining the choice in a two-stage sequential decision. I also apply NL to consumer panel data from essay 1 and compare the results with those of OSL to evaluate the two models in modeling consumer decisions.



Choice Models, Logit Models, Consumer Choice, Impulse Purchase, Shopping List, Planned Purchase, Discrete Choice Models