Essays on attribute inattention choice behavior

Abstract

The main objective of this dissertation is to explore attribute non-attendance choice in food consumption research under the discrete choice framework. The standard choice analysis based on random utility maximization assumes that an agent evaluates every attribute of alternatives and selects his or her most preferred option that maximizes utility in a given choice situation. However, recent empirical evidence reveals that decision makers may ignore a certain attribute presented in a choice set. My dissertation research investigates inattention choice behaviors using stated and revealed preferences data. The first essay, “Out-of-sample Validity of Random Response Share Approach”, applied the Random Response Share (RRS) approach that was proposed by Malone and Lusk (2018) for investigating inattention choice in choice experiments. The aim of the RRS approach is to identify and purge inattention observations in analysis. We applied the RRS and assessed the out-of-sample predictive performance of the RRS using 60 months of choice experiment data from 61,592 U.S households. Our results show that the RRS is not a dominant strategy to the conventional multinomial logit model in terms of out-of-sample forecasting accuracy. However, the RRS could be a way to deal with attribute nonattendance when also considering the socio-economic characteristics of respondents because it is not harmful compared to the predictive accuracy of the traditional multinomial logit model. In the second essay, “Incorporating Choice Heuristics in Analysis of Decision Making”, we investigated consumers’ heuristic choices when purchasing hotdog sausage products. This study applied the IRI marketing data set into the latent class structure of the discrete choice models to explore choice heuristics based on different attribute processing at the level of the household. The main contribution of this study is to incorporate attribute inattention into discrete choice model using actual market data, instead of stated choice data. The estimation results based on multiple models reveal that marginal utilities and willingness to pay estimates for attributes of hotdog products are sensitive to model selection. Our empirical analysis suggests that accounting for heterogeneous decision rules could provide better model fit. Thus, researchers need to consider the heterogeneous decision rules as an alternative to the classic assumption that all attributes are considered in choice situations by decision makers to better understand consumers' choices and provide more accurate policy implications. To sum up, the traditional assumption of full attribute consideration may be strong and restrictive to reflect consumer decision making rules. Recent studies are attempting to relax this assumption and reflect real choice environments. Considering ANA-based choice behaviors may help improve understanding of consumer preference through better analysis of decision making. I hope that this dissertation on attribute inattention choices will be a steppingstone to additional research in the field of discrete choice analysis.