An Empirical Two-Stage Choice Model with Varying Decision Rules Applied to Internet Clickstream Data

Abstract

In this article, the author proposes and applies an empirical two-stage choice model to Internet clickstream data that captures observed choices for two choice stages: products viewed and products purchased. The model allows for interdependences between choices within a stage as well as the use of varying decision rules in each stage. The author accommodates heterogeneity in preferences and in decision rules. The proposed model uses observed choices to infer both attribute preference ratings and criterion attributes. The results show empirically, in a field setting, that the product attributes evaluated in Stage 1 differ from those evaluated in Stage 2. Consistent with existing theory, consumers tend to use simpler decision rules based on a subset of attribute information in earlier stages. The author also finds some preliminary evidence that attributes such as price and size tend to be considered in only one of the two stages, whereas ingredient attributes tend to be considered in both stages. From a managerial perspective, the identification of criterion attributes, coupled with the preference parameters estimated by the model, can assist with the targeting and positioning of promotional strategies.

Choice modelling with search and sort data from an interactive choice experiment

“…Whilst models have begun to be developed in the marketing literature for online choice environments that are well suited to marketing problems (e.g. Moe 2006), such endeavours have received scant attention in the transportation field. Within the broader choice modelling literature, much attention has been given in recent years to modelling unobserved preference heterogeneity, principally through the now ubiquitous mixed logit and latent class models.…”

“…Smith et al (2007) used revealed preference data from an online travel agent (OTA) to estimate choice models, but did not present empirical results, or suggest that process data was utilised in the models. In contrast, Moe (2006) exploited process data collected from the internet, estimating a two-stage choice model that utilised product view and purchase data to model both of these choices. Collins et al (2012) estimated a choice model using both conventional stated preference (SP) data, and data collected from an interactive choice environment that mimicked an OTA, and found consistent results across the two datasets, however with less error variance associated with the OTA data.…”

“…The latter is consistent with the framework proposed by Hauser and Wernerfelt (1990), where a consideration set is gradually expanded as alternatives are evaluated. Moe (2006) also examines data that is consistent with exploratory search, whereas the current paper considers data from a reductive search process. The above outline of a reductive search loosely matches many online choice environments, including those for hotels, hire cars, and airline tickets, with the latter forming the empirical setting for the current paper.…”

Flexible Query Answering Systems 2015

“…Kim et al (2004) [11] used the clickstream data as implicit feedback to design a hybrid recommendation model, which resulted in better performance. Moe (2006) [17] proposed an empirical two-stage choice model based on clickstream data to capture observed choices for two stages: products viewed and products purchased. They found that the product attributes evaluated in Stage 1 differred from those evaluated in Stage 2.…”

Information Processing Pattern and Propensity to Buy: An Investigation of Online Point-of-Purchase Behavior

“…on the Internet Although our main contribution is to basic research in information processing, this paper relates to econometric studies that investigate conversion behavior on the Internet, i.e., converting visits into purchases (Moe 2006, Moe and Fader 2004, Montgomery et al 2004, Sismeiro and Bucklin 2004. These efforts typically focus on determining how customer and search characteristics relate to purchase behavior but do not study the information processing pattern and its relationship to propensity to buy at the point of purchase.…”