Data pruning in consumer choice models

Common, if not ubiquitous, Marketing practice when estimating models for scanner panel data is to: (a) observe the data, (b) prune the data to a “manageable” number of brands or SKUs, and (c) fit models to the remaining data. We demonstrate that such pruning practice can lead to significantly different (and potentially biased) elasticities, and hence different managerial/practical outcomes, especially in the context of model misspecification. We first justify our claims theoretically by writing the general problem in a classic missing-data framework and demonstrate that commonly used pruning mechanisms (gleaned from current academic Marketing literature) can lead to a nonignorable missing data mechanism. Secondly, we summarize an extensive set of simulations that were run to understand the driving factors of that bias. The results indicate much greater pruning bias in those cases where model fit is poor (small ), random utility errors are correlated with the covariates, or the model is misspecified (e.g., a homogeneous logit is specified when a mixed-logit is true). Empirically, we also demonstrate our findings on the well-cited and highly utilized fabric softener data of Fader and Hardie (1996). Our empirical findings suggest a number of estimates that vary according to the way in which the data is pruned including the magnitude of market mix and attribute elasticities, and purchase probabilities, but that the pruning effect is smaller for better fitting models.

Heterogeneity distributions of willingness-to-pay in choice models

We investigate direct and indirect specification of the distribution of consumer willingness-to-pay (WTP) for changes in product attributes in a choice setting. Typically, choice models identify WTP for an attribute as a ratio of the estimated attribute and price coefficients. Previous research in marketing and economics has discussed the problems with allowing for random coefficients on both attribute and price, especially when the distribution of the price coefficient has mass near zero. These problems can be avoided by combining a parameterization of the likelihood function that directly identifies WTP with a normal prior for WTP. We show that the typical likelihood parameterization in combination with what are regarded as standard heterogeneity distributions for attribute and price coefficients results in poorly behaved posterior WTP distributions, especially in small sample settings. The implied prior for WTP readily allows for substantial mass in the tails of the distribution and extreme individual-level estimates of WTP. We also demonstrate the sensitivity of profit maximizing prices to parameterization and priors for WTP.

Estimation bias in choice models with last choice feedback

The study compares two estimation methods for choice models with last choice feedback, using simulated and real data. The first method ignores the impact of unobserved heterogeneity on observed choices via presample choices, while the second method approximates this impact by a stochastic relationship. In panels with less than 10 choices per panelist, the first method overstates the impact of last choice on current choice and understates the impact of intrinsic preferences (i.e., brand intercepts). The second method performs significantly better than the first method. Under both methods, an increase in the number of heterogeneous coefficients in the model tends to increase the bias in the estimates. The largest bias occurs when lagged choice coefficients are heterogeneous.     

When worlds collide: The implications of panel data-based choice models for consumer behavior

Panel data, both diary and scanner, have been analyzed by marketing scientists for over thirty years. One of the important uses of panel data is to better understand consumer behavior by developing and testing hypotheses using the revealed preference data rather than experimental data that uses only self-reported behavior or behavior in a simulated choice environment. The purpose of this paper is to suggest areas of research where panel data can be used to better understand the underlying behavior of the panel members.     

Advances in computation,statistical methods,and testing of discrete choice models

This paper gives a brief overview of recent developments in computation, estimation, and statistical testing of choice models, with marketing applications. Topics include statistical models for discrete panel data with heterogeneous decision-makers, simulation methods for estimation of high-dimension multinomial probit models, specification tests for model structure and for brand and purchase clustering, and innovations in numerical analysis for estimation and forecasting. In collaboration with Denis Bolduc, David Bunch, Michael Keane, Don Kridel, and Steve Stern.     

Efficient choice set designs for estimating availability cross-effects models

Batsell and Polking proposed a discrete choice model which incorporates the availability (presence or absence) of competing brands into the utility of each brand under study. The information on relative impacts of adding or deleting brands is of strategic interest, and models that do not incorporate such effects may be misleading. The designs suggested by Batsell and Polking have 2^m–m–1 choice sets. Even with as few as 10=m brands, this requires over 1000 choice sets. In this paper we provide a catalog of designs for estimating cross effects models in as few as 2m–1 choice sets. This will make cross effects modelling practical in a wide range of academic and commercial settings.     

Modeling preference evolution in discrete choice models: A Bayesian state-space approach

We develop discrete choice models that account for parameter driven preference dynamics. Choice model parameters may change over time because of shifting market conditions or due to changes in attribute levels over time or because of consumer learning. In this paper we show how such preference evolution can be modeled using hierarchial Bayesian state space models of discrete choice. The main feature of our approach is that it allows for the simultaneous incorporation of multiple sources of preference and choice dynamics. We show how the state space approach can include state dependence, unobserved heterogeneity, and more importantly, temporal variability in preferences using a correlated sequence of population distributions. The proposed model is very general and nests commonly used choice models in the literature as special cases. We use Markov chain monte carlo methods for estimating model parameters and apply our methodology to a scanner data set containing household brand choices over an eight-year period. Our analysis indicates that preferences exhibit significant variation over the time-span of the data and that incorporating time-variation in parameters is crucial for appropriate inferences regarding the magnitude and evolution of choice elasticities. We also find that models that ignore time variation in parameters can yield misleading inferences about the impact of causal variables. This paper is based on the first author's doctoral dissertation.     

Homogeneity and heterogeneity in stochastic models of brand choice behavior

Heterogeneity of consumers is one of the cornerstones of empirical findings and theories in marketing. It serves, for example, as the foundation for such areas as market segmentation and product differentiation. This paper attempts to trace and clarify the evolution over the last twenty years of the homogeneity assumptions in the area of stochastic models of brand choice behavior. In analyzing individual choice behavior by means of stochastic models, all individuals were often assumed to possess the same set of transition probabilities or follow the same stochastic process. However, empirical studies at the individual level indicate that individuals are actually non-homogeneous in those probabilities and processes. In this article we provide an analytical proof that if the behavior of individuals is specified to be homogeneous when it is not, wrong inferences about the type of stochastic process individuals follow and about the expected behavior of the total population will be drawn. Ways to remedy these problems by allowing for heterogeneity are reviewed. The implications of heterogeneity and our findings in the various application areas which utilize stochastic choice models are examined.     

Category management and consumer choice

This paper shows the results of an exploratory analysis of the structure of a complex product category: Wine. In complex categories, a usual strategy of consumers is the partitioning of alternatives into homogeneous subsets and the sequential eliminations of subsets until a product/brand is chosen from among few homogeneous alternatives in the last subset. To identify if there is such a kind of strategy and the product attributes involved is of great interest for the retailer. After a discussion of the different modelling alternatives of the choice process, the authors provide an application of the additive trees (ADDTREE) model to explore the hierarchical structure. The ADDTREE results provide a first overview of the competitive market structure of the wine category: competition becomes more intense as the wine category is partitioned by, first, the type of wine criteria and, second, the designation of origin (DO) criteria.     

Demographics in consumer behavior

Four criticisms against using socioeconomic and demographic (SED) factors in consumer behavior are reviewed: dissatisfaction with models of consumption behavior developed by economists and sociologists, obsolescence of SED factors in mass consumption societies, poor predictions produced by SED factors, and a grass-is-greener attitude held by consumer researchers. The insights offered substantially hurt the validity of these criticisms. Strategies for better theory and research in consumer behavior using SED factors are described.     

Relationship Quality and the Theory of Planned Behavior models of behavioral intentions and purchase behavior

Using real-life purchase behavior data of apparel and survey information, this study compares the Relationship Quality and the Theory of Planned Behavior models. The attitude towards the buying behavior, the subjective norm and perceived behavioral control (antecedents of the buying intention in the Theory of Planned Behavior) are better predictors of behavioral intentions than Relationship Quality. In both models intentions fully mediate the impact of attitudinal antecedents on behavior, both in terms of purchase incidence and purchase behavior (amount spent, number of visits, and types of products bought). Frequency and recency of prior buying behavior and, to a lesser extent, its monetary value, predict subsequent purchase incidence, above and beyond the impact of attitude and intention. Attitudinal antecedents of behavior significantly predict buying behavior, but they become insignificant when buying behavior is included in the model.     

Profiling the reference price consumer

Researchers in marketing have devoted considerable attention to understanding how price impacts the purchase decision. Some individuals, termed memory-based reference price (MBR) consumers, take into account price expectations developed from past purchase behavior when making a current choice. Other individuals, termed stimulus-based reference price (SBR) consumers, make choices by constructing a reference point from the currently observed distribution of prices. Using a latent class model of structural heterogeneity applied to purchase histories from the toilet tissue category, we classify households in terms of the pricing mechanism used in buying decisions. We find strong evidence that memory-based (internal) reference price consumers are more price sensitive than other consumers. Moreover, we find that variables associated with the accessibility of price information are predictive of consumer use of memory-based reference prices. Managerial implications of these results are discussed.     

Information processing and innovative choice

In this paper we examine how the information processing of subjects who make an innovative choice (innovators) differs from that of subjects who make a noninnovative choice (noninnovators). The task involves selection of an alternative within a range of prerated product category innovativeness. We propose that subjects who seek 1) impersonal/uncontrollable sources, 2) higher levels of information, 3) more detailed (versus summary) data, and 4) noncomparative (versus comparative) data are more likely to make innovative choices. The research method is a computerized process tracing experiment utilizing Search Monitor (Brucks 1988).The authors wish to express their gratitude to Merrie Brucks for the use of and help with Search Monitor and to Eric Johnson, Dave Schmittlein, and Mita Sujan for helpful comments on an earlier version of this paper.     

Positive affect moderates the impact of assortment size on choice satisfaction

The present research proposes that positive affect mitigates choice overload. Results from four studies show that whereas people in neutral affect feel less satisfied when choosing from a large relative to a small assortment, people in positive affect do not experience a decline in satisfaction and may even feel more satisfied when choosing from a large, relative to a small assortment. It is proposed that positive affect has these effects by shifting people's attention from the difficulty of the choice task to the quality of the assortment, as a basis for judgment of choice satisfaction.     

Cooling off and backing out: Understanding consumer decisions to rescind a product purchase

Consumers sometimes have the right to exit a sales contract during what is known as a cooling-off period. Sales process research generally does not address cases when consumers withdraw from sales contracts during this period. Since securing product sales involves substantial marketing and sales costs, a need exists to better understand not only consumer rescission decisions and their legal context but also the managerial implications of the cooling-off period. This exploratory qualitative study examines purchase rescinding and develops a conceptual model using timeshare as the context. Results suggest that rescission relates to a mismatch between product features and personal circumstances, post-purchase concerns about product value, reassessment of financial capability, reflections on sales presentations, and cautionary influences of reference groups.     

The dynamics of networked power in a concentrated business network

The purpose of this paper is to investigate the dynamics of networked power in a concentrated business network. Power is a long standing theme in inter-organizational research, yet there is a paucity of studies about how power emerges and is constructed over time at the network level. The paper adopts process, systems and network theory to interpret a rich single case study from the food industry. Three power mechanisms are identified, gatekeeping, decoupling and resource allocation, which form the basis of a model of networked power dynamics. Empirically tracing the dynamics of networked power highlights the economic contents of interactions. The paper extends current understandings of power as ‘conflict and coercion’ to include influencing, leveraging and strategic maneuvering in the actual performance of networked power.     

Regret and performance uncertainty in consumer repeat choice

The role of regret on consumer choices has been well recognized in risky purchasing contents. Rather than outcome comparison, regret should also be contingent on the extent of consumer expectation uncertainty on performance. To explore this issue, the paper first draws some propositions based on a Bayesian updating model assuming consumer expectation as a kind of distribution. By two experiments, we show that consumers change their choice intention for alternatives with different performance uncertainty even though zero disconfirmation and no regret are experienced. Consumer experience and risk attitude moderate the change pattern.     

Quick and easy choice sets: Constructing optimal and nearly optimal stated choice experiments

In this paper we compare a number of common strategies for constructing discrete choice experiments. Two of the strategies, including one based on theoretical constructions for optimal discrete choice experiments, produce designs that are better than those that come about from random grouping and from using the L^MA construction. A simple account of this theoretical construction is given.     

Determinants of retail bank choice in Nigeria: a focus on gender-based choice decisions

As the Nigerian retail banks customers become more sophisticated, it is now very important that retail banks determine the factors that are important and relevant to the customers’ retail bank choice decisions. By using a survey of retail bank customers, this study evaluates the relative importance attached to retail banks’ choice criteria used by male and female customers in Nigeria. The results show that there are some differences in choice factors used by male and female customers in selecting a retail bank for patronage. The recommendation is that bank managers should take both female and male market segments into consideration when making retail bank strategic planning in order to become competitive. The findings can assist retail bank managers in refining their marketing strategies as a means of overcoming the intense competition that exists in the Nigerian banking system.