Search

This paper proposes a new approach to modeling heterogeneity in choice data that can accommodate fixed-point ratings data and text. This paper proposes a new approach to modeling heterogeneity in choice data that can accommodate fixed-point ratings data and text. Respondent choices, survey responses, and narratives are combined to form latent archetypes that provide an integrated description of respondents in terms of the objects and drivers of their wants. We propose a measure of coherence to assess the value of integrating these data elements and demonstrate the value of integrating text data into an analysis of choice and scaled response data. A conjoint data set is used to illustrate the model where we find that the text data helps clarify the origin of demand. [ABSTRACT FROM AUTHOR]

A firm contemplating improvements to its product attributes would be interested in the dollar value the market attaches to any potential product modification. In this paper, we derive a measure of market value such that the comparison of the measure against the incremental unit cost of the attribute improvement is key in deciding whether or not the attribute improvement is profitable. Competition from other brands, the potential for market expansion, and heterogeneity in customer preference structures are explicitly modeled using the multinomial logit framework. The analysis yields a closed form expression for the market's value for an attribute improvement (MVAI). A key result we obtain is that customers should be differentially weighted based on their probability of purchasing the firm's product. In particular, customers who exhibit a very high or very low probability of choosing the firm's product should receive less weight in determining MVAI. Because the probability of choice varies across products, the answer to the question of how much the market values an improvement depends on which firm is asking the question. It is shown that customers whose utilities have a greater random component should be weighted less. Furthermore, the measure developed is robust to the influence of outliers in the sample. An empirical illustration of the MVAI measure in the context of a new product development study is provided. The study illustrates the advantages of the proposed measure over currently used approaches and explores the possibility of competitive price reactions. [ABSTRACT FROM AUTHOR]

Consumers use different attribute partworths online and offline; these differences are consistent across consumers, and we propose a method for adjusting them. Despite the growth of online retail, the majority of products are still sold offline, and the "touch-and-feel" aspect of physically examining a product before purchase remains important to many consumers. In this paper, we demonstrate that large discrepancies can exist between how consumers evaluate products when examining them "live" versus based on online descriptions, even for a relatively familiar product (messenger bags) and for utilitarian features. Therefore, the use of online evaluations in market research may result in inaccurate predictions and potentially suboptimal decisions by the firm. Because eliciting preferences by conducting large-scale offline market research is costly, we propose fusing data from a large online study with data from a smaller set of participants who complete both an online and an offline study. We demonstrate our approach using conjoint studies on two sets of participants. The group who completed both online and offline studies allows us to calibrate the relationship between online and offline partworths. To obtain reliable parameter estimates, we propose two statistical methods: a hierarchical Bayesian approach and a k-nearest-neighbors approach. We demonstrate that the proposed approach achieves better out-of-sample predictive performance on individual choices (up to 25% improvement), as well as aggregate market shares (up to 33% improvement). Data are available at https://doi.org/10.1287/mksc.2018.1124. [ABSTRACT FROM AUTHOR]

Existing research on choice designs focuses exclusively on compensatory models that assume that all available alternatives are considered in the choice process. In this paper, we develop a method to construct efficient designs for a two-stage, consider-then-choose model that involves a noncompensatory screening process at the first stage and a compensatory choice process at the second stage. The method applies to both conjunctive and disjunctive screening rules. Under certain conditions, the method also applies to the subset conjunctive and disjunctions of conjunctions screening rules. Based on the local design criterion, we conduct a comparative study of compensatory and conjunctive designs--the former are optimized for a compensatory model and the latter for a two-stage model that uses conjunctive screening in its first stage. We find that conjunctive designs have higher level overlap than compensatory designs. This occurs because level overlap helps pinpoint screening behavior. Higher overlap of conjunctive designs is also accompanied by lower orthogonality, less level balance, and more utility balance. We find that compensatory designs have a significant loss of design efficiency when the true model involves conjunctive screening at the consideration stage. These designs also have much less power than conjunctive designs in identifying a true consider-then-choose process with conjunctive screening. In contrast, when the true model is compensatory, the efficiency loss from using a conjunctive design is lower. Also, conjunctive designs have about the same power as compensatory designs in identifying a true compensatory choice process. Our findings make a strong case for the use of conjunctive designs when there is prior evidence to support respondent screening. [ABSTRACT FROM AUTHOR]

Random effects or mixed logit models are often used to model differences in consumer preferences. Data from choice experiments are needed to estimate the mean vector and the variances of the multivariate heterogeneity distribution involved. In this paper, an efficient algorithm is proposed to construct semi-Bayesian D-optimal mixed logit designs that take into account the uncertainty about the mean vector of the distribution. These designs are compared to locally D-optimal mixed logit designs, Bayesian and locally D-optimal designs for the multinomial logit model and to nearly orthogonal designs (Sawtooth (CBC)) for a wide range of parameter values. It is found that the semi-Bayesian mixed logit designs outperform the competing designs not only in terms of estimation efficiency but also in terms of prediction accuracy. In particular, it is shown that assuming large prior values for the variance parameters for constructing semi-Bayesian mixed logit designs is most robust against the misspecification of the prior mean vector. In addition, the semi-Bayesian mixed logit designs are compared to the fully Bayesian mixed logit designs, which take also into account the uncertainty about the variances in the heterogeneity distribution and which can be constructed only using prohibitively large computing power. The differences in estimation and prediction accuracy turn out to be rather small in most cases, which indicates that the semi-Bayesian approach is currently the most appropriate one if one needs to estimate mixed logit models. [ABSTRACT FROM AUTHOR]

The use of adaptive designs in conjoint analysis has been shown to lead to an endogeneity bias in part- worth estimates using sampling experiments. In this paper, we re-examine the endogeneity issue in light of the likelihood principle. The likelihood principle asserts that all relevant information in the data about model parameters is contained in the likelihood function. We show that, once the data are collected, adhering to the likelihood principle leads to analysis where endogeneity becomes ignorable for estimation. The likelihood principle is implicit to Bayesian analysis, and discussion is offered for detecting and dealing with endogeneity bias in marketing. [ABSTRACT FROM AUTHOR]

Firms develop products by manipulating the attributes of offerings, and consumers derive utility from the benefits that the attributes afford. While the field of marketing has long been aware of the distinction between attributes and benefits, it has not developed methods for understanding how attributes and benefits are related. This paper develops a benefit-based model for conjoint analysis that assumes consumers satiate on attributes that are perceived to provide the same benefit. A latent-variable model is proposed that estimates the map between attributes and benefits, and is applied to data from two conjoint studies involving a durable product and a household consumable. The model is shown to fit the data better, provide improved predictions, and lead to different product design implications than the standard conjoint model. Data, as supplemental material, are available at . [ABSTRACT FROM AUTHOR]

Features involving the taste, smell, touch, and sight of products, as well as attributes such as safety and confidence, are not easily measured in product research without respondents actually experiencing them. Moreover, product researchers often evaluate a large number of these attributes (e.g., >50) in applied studies, making standard valuation techniques such as conjoint analysis difficult to implement. Product researchers instead rely on ratings data to assess features for which the respondent has had actual experience. In this paper we develop a method of monetizing rating data to standardize product evaluations among respondents. The adjusted data are shown to increase the accuracy of purchase predictions by about 20% relative to existing methods of scale adjustment, leading to better inference in models using ratings data. We demonstrate our method using data from a large scale product use study by a packaged goods manufacturer. Data, as supplemental material, are available at . [ABSTRACT FROM AUTHOR]

As technology advances, new products (e.g., digital cameras, computer tablets, etc.) have become increasingly more complex. Researchers often face considerable challenges in understanding consumers' preferences for such products. This paper proposes an adaptive decompositional framework to elicit consumers' preferences for complex products. The proposed method starts with collaborative-filtered initial part-worths, followed by an adaptive question selection process that uses a fuzzy support vector machine active learning algorithm to adaptively refine the individual-specific preference estimate after each question. Our empirical and synthetic studies suggest that the proposed method performs well for product categories equipped with as many as 70 to 100 attribute levels, which is typically considered prohibitive for decompositional preference elicitation methods. In addition, we demonstrate that the proposed method provides a natural remedy for a long-standing challenge in adaptive question design by gauging the possibility of response errors on the fly and incorporating the results into the survey design. This research also explores in a live setting how responses from previous respondents may be used to facilitate active learning of the focal respondent's product preferences. Overall, the proposed approach offers new capabilities that complement existing preference elicitation methods, particularly in the context of complex products. Data, as supplemental material, are available at . [ABSTRACT FROM AUTHOR]

Extant research on choice designs in marketing focuses on the construction of efficient homogeneous designs where all respondents get the same design. Recently marketing scholars proposed the construction of efficient heterogeneous designs where different respondents or groups of respondents get different subdesigns, and demonstrated substantial efficiency gain when such heterogeneous designs are employed. A significant hurdle in the widespread adoption of heterogeneous designs is the high computation cost, even when the number of subdesigns contained in the heterogeneous design is restricted to be small. In this paper we propose a new approach for the construction of efficient heterogeneous choice designs. In contrast to extant approaches that are based on an exact design framework where it is computationally prohibitive to do an exhaustive search to find a globally optimal design, our proposed approach is based on the continuous design framework where well-established mathematical theories can be leveraged for quick identification of a globally optimal design. The proposed approach makes it feasible to generate a highly efficient choice design that is completely heterogeneous--a unique subdesign for each individual respondent in the choice experiment. The proposed approach is the first in the marketing literature to find a completely heterogeneous choice design with assured high global design efficiency using the continuous design framework. Results from simulation and empirical studies demonstrate superior performance of the proposed approach over extant approaches in constructing efficient heterogeneous choice designs. [ABSTRACT FROM AUTHOR]

Quantity discount pricing is a common practice used by business-to-business and business-to-consumer companies. A key characteristic of quantity discount pricing is that the marginal price declines with higher purchase quantities. In this paper, we propose a choice-based conjoint model for estimating consumer-level willingness to pay (WTP) for varying quantities of a product and for designing optimal quantity discount pricing schemes. Our model can handle large quantity values and produces WTP estimates that are positive and increasing in quantity at a diminishing rate. In particular, we propose a tractable WTP function that depends on both product attributes and product quantity and that captures diminishing marginal WTP. We show how such a function embeds standard WTP functions in the quantity discount literature as special cases. We also demonstrate how to use the model to estimate the consumer value potential, which is the product of the premium a consumer is willing to pay and her volume potential. Finally, we propose a parsimonious experimental design approach for implementation. We illustrate the model using data from a conjoint study of online movie rental services. The empirical results show that the proposed model has good fit and predictive validity. In addition, we find that marginal WTP in this category decays rapidly with quantity. We also find that the standard choice-based conjoint model results in anomalous WTP distributions with negative WTP values and nondiminishing marginal willingness-to-pay curves. Finally, we identify four segments of consumers that differ in terms of magnitude of WTP and volume potential, and we derive optimal quantity discount schemes for a monopolist and a new entrant in a competitive market. [ABSTRACT FROM AUTHOR]

Choice-based conjoint analysis is a popular marketing research technique to learn about consumers' preferences and to make market share forecasts under various scenarios for product offerings. Managers expect these forecasts to be "realistic" in terms of being able to replicate market shares at some prespecified or "basecase" scenario. Frequently, there is a discrepancy between the recovered and base-case market share. This paper presents a Bayesian decision theoretic approach to incorporating base-case market shares into conjoint analysis via the loss function. Because defining the base-case scenario typically involves a variety of management decisions, we treat the market shares as constraints on what are acceptable answers, as opposed to informative prior information. Our approach seeks to minimize the adjustment of parameters by using additive factors from a normal distribution centered at 0, with a variance as small as possible, but such that the market share constraints are satisfied. We specify an appropriate loss function, and all estimates are formally derived via minimizing the posterior expected loss. We detail algorithms that provide posterior distributions of constrained and unconstrained parameters and quantities of interest. The methods are demonstrated using discrete choice models with simulated data and data from a commercial market research study. These studies indicate that the method recovers base-case market shares without systematically distorting the preference structure from the conjoint experiment. [ABSTRACT FROM AUTHOR]

In their purchase decisions, online customers seek to improve decision quality while limiting search efforts. In practice, many merchants have understood the importance of helping customers in the decision-making process and provide online decision aids to their visitors. In this paper, we show how preference models which are common in conjoint analysis can be leveraged to design a questionnaire-based decision aid that elicits customers' preferences based on simple demographics, product usage, and self-reported preference questions. Such a system can offer relevant recommendations quickly and with minimal customer input. We compare three algorithms--cluster classification, Bayesian treed regression, and stepwise componential regression--to develop an optimal sequence of questions and predict online visitors' preferences. In an empirical study, stepwise componential regression, relying on many fewer and easier-to-answer questions, achieved predictive accuracy equivalent to a traditional conjoint approach. [ABSTRACT FROM AUTHOR]

Polyhedral methods for choice-based conjoint analysis provide a means to adapt choice-based questions at the individual-respondent level and provide an alternative means to estimate partworths when there are relatively few questions per respondent, as in a Web-based questionnaire. However, these methods are deterministic and are susceptible to the propagation of response errors. They also assume, implicitly, a uniform prior on the partworths. In this paper we provide a probabilistic interpretation of polyhedral methods and propose improvements that incorporate response error and/or informative priors into individual-level question selection and estimation. Monte Carlo simulations suggest that response-error modeling and informative priors improve polyhedral question-selection methods in the domains where they were previously weak. A field experiment with over 2,200 leading-edge wine consumers in the United States, Australia, and New Zealand suggests that the new question-selection methods show promise relative to existing methods. [ABSTRACT FROM AUTHOR]

The authors propose two variants of lexicographic preference rules. They obtain the necessary and sufficient conditions under which a linear utility function represents a standard lexicographic rule, and each of the proposed variants, over a set of discrete attributes. They then: (i) characterize tile measurement properties of the parameters in the representations; (ii) propose a nonmetric procedure for inferring each lexicographic rule from pairwise comparisons of multiattribute alternatives; (iii) describe a method for distinguishing among different lexicographic rules, and between lexicographic and linear preference models; and (iv) suggest how individual lexicographic rules can be combined to describe hierarchical market structures. The authors illustrate each of these aspects using data on personal-computer preferences. They find that two-thirds of the subjects in the sample use some kind of lexicographic rule. In contrast, only one in five subjects use a standard lexicographic rule. This suggests that lexicographic rules are more widely used by consumers than one might have thought in the absence of the lexicographic variants described in the paper. The authors report a simulation assessing the ability of the proposed inference procedure to distinguish among alternative lexicographic models, and between linear-compensatory and lexicographic models. [ABSTRACT FROM AUTHOR]

Consumer choice in surveys and in the marketplace reflects a complex process of screening and evaluating choice alternatives. Behavioral and economic models of choice processes are difficult to estimate when using stated and revealed preferences because the underlying process is latent. This paper introduces Bayesian methods for estimating two behavioral models that eliminate alternatives using specific attribute levels. The elimination by aspects theory postulates a sequential elimination of alternatives by attribute levels until a single one, the chosen alternative, remains. In the economic screening rule model, respondents screen out alternatives with certain attribute levels and then choose from the remaining alternatives, using a compensatory function of all the attributes. The economic screening rule model gives an economic justification as to why certain attributes are used to screen alternatives. A commercial conjoint study is used to illustrate the methods and assess their performance. In this data set, the economic screening rule model outperforms the EBA and other standard choice models and provides comparable results to an equivalent conjunctive screening rule model. [ABSTRACT FROM AUTHOR]

Almost universally, research and practice suggest that a brand that increases its product assortment, or variety, should benefit through increased market share. In this paper, we show this is not always the case. We introduce the construct "assortment type" and demonstrate that the effect of assortment size on brand share is systematically moderated by assortment type. We define an "alignable" assortment as a set of brand variants that differ along a single, compensatory dimension such that choosing from that assortment only requires within-attribute trade-offs. In contrast, we define a "nonalignable" assortment as a set of brand variants that simultaneously vary along multiple, noncompensatory dimensions, demanding between-attribute tradeoffs. In turn, we argue that an alignable assortment can efficiently meet the diverse tastes of consumers, thereby increasing brand share, but that a nonalignable assortment increases both the cognitive effort and the potential for regret faced by a consumer, thereby decreasing brand share. We term this effect "overchoice." Across three studies, we provide evidence of overchoice and tie the effect to the effort and regret brought about by nonalignability. In the process, we demonstrate that simplification of information presentation, reversibility of choice, and a reduction in underlying nonalignability serve to reduce or eliminate this effect. [ABSTRACT FROM AUTHOR]

Many theories of consumer behavior involve thresholds and discontinuities. In this paper, we investigate consumers' use of screening rules as part of a discrete-choice model. Alternatives that pass the screen are evaluated in a manner consistent with random utility theory; alternatives that do not pass the screen have a zero probability of being chosen. The proposed model accommodates conjunctive, disjunctive, and compensatory screening rules. We estimate a model that effects a discontinuous decision process by employing the Bayesian technique of data augmentation and using Markov-chain Monte Carlo methods to integrate over the parameter space. The approach has minimal information requirements and can handle a large number of choice alternatives. The method is illustrated using a conjoint study of cameras. The results indicate that 92% of respondents screen alternatives on one or more attributes. [ABSTRACT FROM AUTHOR]

This paper develops a model for capturing continuous heterogeneity in the joint distribution of reservation prices for products and bundles. Our model is derived from utility theory and captures both within- and among-subject variability. Furthermore, it provides dollarmetric reservation prices and individual-level estimates that allow the firm to target customers and develop customized and nonlinear pricing policies. Our experiments show that, regardless of whether the products are durables or nondurables, the model captures heterogeneity and predicts well. Models that assume homogeneity perform poorly, especially in predicting choice of the bundle. Furthermore, the methodology is robust even when respondents evaluate few profiles. Self-stated reservation prices do not have any informational content beyond that contained in the basic model. The direct elicitation method appears to understate (overstate) the variation in reservation prices across consumers for low-priced (high-priced) products and bundles. Hence this method yields biased demand estimates and leads to suboptimal product-line pricing policy. The optimization results show that the product-line pricing policy depends on the degree of heterogeneity in the reservation prices of the individual products and the bundle. A uniformly high-price strategy for all products and bundles is optimal when heterogeneity is high. Otherwise, a hybrid strategy is optimal. [ABSTRACT FROM AUTHOR]

A computationally attractive model for the analysis of conjoint choice experiments is the mixed multinomial logit model, a multinomial logit model in which it is assumed that the coefficients follow a (normal) distribution across subjects. This model offers the advantage over the standard multinomial logit model of accommodating heterogeneity in the coefficients of the choice model across subjects, a topic that has received considerable interest recently in the marketing literature. With the advent of such powerful models, the conjoint choice design deserves increased attention as well. Unfortunately, if one wants to apply the mixed logit model to the analysis of conjoint choice experiments, the problem arises that nothing is known about the efficiency of designs based on the standard logit for parameters of the mixed logit. The development of designs that are optimal for mixed logit models or other random effects models has not been previously addressed and is the topic of this paper. The development of efficient designs requires the evaluation of the information matrix of the mixed multinomial logit model. We derive an expression for the information matrix for that purpose. The information matrix of the mixed logit model does not have closed form, since it involves integration over the distribution of the random coefficients. In evaluating it we approximate the integrals through repeated samples from the multivariate normal distribution of the coefficients. Since the information matrix is not a scalar we use the determinant scaled by its dimension as a measure of design efficiency. This enables us to apply heuristic search algorithms to explore the design space for highly efficient designs. We build on previously published heuristics based on relabeling, swapping, and cycling of the attribute levels in the design. Designs with a base alternative are commonly used and considered to be important in conjoint choice analysis, since they provide a way to compare the utilities of profiles in different choice sets. A base alternative is a product profile that is included in all choice sets of a design. There are several types of base alternatives, examples being a so-called outside alternative or an alternative constructed from the attribute levels in the design itself. We extend our design construction procedures for mixed logit models to include designs with a base alternative and investigate and compare four design classes: designs with two alternatives, with two alternatives plus a base alternative, and designs with three and with four alternatives. Our study provides compelling evidence that each of these mixed logit designs provide more efficient parameter estimates for the mixed logit model than their standard logit counterparts and yield higher predictive validity. As compared to designs with two alternatives, designs that include a base alternative are more robust to deviations from the parameter values assumed in the designs, while that robustness is even higher for designs with three and four alternatives, even if those have 33% and 50% less choice sets, respectively. Those designs yield higher efficiency and better predictive validity at lower burden to the respondent. It is noteworthy that our "best" choice designs, the 3- and 4-alternative designs, resulted not only in a substantial improvement in efficiency over the standard logit design but also in an expected predictive validity that is over 50% higher in most cases, a number that pales the increases in predictive validity achieved by refined model specifications. [ABSTRACT FROM AUTHOR]

The drive to satisfy customers in narrowly defined market segments has led firms to offer wider arrays of products and services. Delivering products and services with the appropriate mix of features for these highly fragmented market segments requires understanding the value that customers place on these features. Conjoint analysis endeavors to unravel the value, or partworths, that customers place on the product or service's attributes from experimental subjects' evaluation of profiles based on hypothetical products or services. When the goal is to estimate the heterogeneity in the customers' partworths, traditional estimation methods, such as Least squares, require each subject to respond to more profiles than product attributes, resulting in lengthy questionnaires for complex, multiatiributed products or services. Long questionnaires pose practical and theoretical problems. Response rates tend to decrease with increasing questionnaire length, and more importantly, academic evidence indicates that long questionnaires may induce response biases. The problems associated with long questionnaires call for experimental designs and estimation methods that recover the heterogeneity in the partworths with shorter questionnaires. Unlike more popular estimation methods, Hierarchical Bayes (HB) random effects models do not require that individual-level design matrices be of full rank, which leads to the possibility of using fewer profiles per subject than currently used. Can this theoretical possibility be practically implemented? This paper tests this conjecture with empirical studies and mathematical analysis. The random effects model in the paper describes the heterogeneity in subject-level partworths or regression coefficients with a linear model that can include subject-level covariates. In addition, the error variances are specific to the subjects, thus allowing for the differential use of the measurement scale by different subjects. In the empirical study, subjects' responses to a full profile design are randomly deleted to test the performance of HB methods with declining sample sizes. These simple experiments indicate that NB methods can recover heterogeneity and estimate individual-level partworths, even when individual-level least squares estimators do not exist due to insufficient degrees of freedom. Motivated by these empirical studies, the paper analytically investigates the trade-off between the number of profiles per subject and the number of subjects on the statistical accuracy of the estimators that describe the partworth heterogeneity. The paper considers two experimental designs: each subject receives the same set of profiles, and subjects receive different blocks of a fractional factorial design. In the first case, the optimal design, subject to a budget constraint, uses more subjects and fewer profiles per subject when the ratio of unexplained, partworth heterogeneity to unexplained response variance is large. In the second case, one can maintain a given level of estimation accuracy as the number of profiles per subject decreases by increasing the number of subjects assigned to each block. These results provide marketing researchers the option of using shorter questionnaires for complex products or services. The analysis assumes that response quality is independent of questionnaire length and does not address the impact of design factors on response quality, If response quality and questionnaire length were, in fact, unrelated, then marketing researchers would still find the paper's results useful in improving the efficiency of their conjoint designs. However, if response quality were to decline with questionnaire length, as the preponderance of academic research indicates, then the option to use shorter questionnaires would become even more valuable. [ABSTRACT FROM AUTHOR]

Digitization of content is changing how consumers and firms use purchase and rental markets. Low transaction costs make accessing content easier for consumers. Digital technology enables firms to create nondurable "rental" versions of their content and to restrict content to the purchasing consumer, effectively shutting down resale markets. To empirically analyze the interaction of purchase and rental markets, I design a preference measurement tool to recover consumers' intertemporal preferences through current-period choices alone. I then use these preferences to solve for a dynamic equilibrium between consumers and the firm. In the context of the online home video market, I find that when the firm is able to commit to holding prices fixed forever, providing content through the purchase market alone is sufficient. However, when the firm is unable to commit, it should serve both purchase and rental markets. Canonical theory models would predict exclusive rentals, but the purchase option enables indirect price discrimination in practice. I also find that when consumers place a premium on accessing new content, they are less likely to intertemporally substitute, thereby increasing the firm's pricing power. Consistent with theory, commitment to future prices increases profits considerably. This finding supports the rigid pricing structure of such retailers as Apple, despite studios' push toward more pricing flexibility. [ABSTRACT FROM AUTHOR]

In most marketing experiments, managerial decisions are not based directly on the estimates of the parameters but rather on functions of these estimates. For example many managerial decisions are driven by whether or not a feature is valued more than the price the consumer will be asked to pay. In other cases, some managerial decisions are weighed more heavily than others. The standard measures used to evaluate experimental designs (e.g., A-efficiency or D-efficiency) do not accommodate these phenomena. We propose alternative "managerial efficiency" criteria (M-errors) that are relatively easy to implement. We explore their properties, suggest practical algorithms to decrease errors, and provide illustrative examples. Realistic examples suggest improvements of as much as 30% in managerial efficiency. We dose by considering approximations for nonlinear criteria and extensions to choice-based experiments. [ABSTRACT FROM AUTHOR]

Greedoid languages provide a basis to infer best-fitting noncompensatory decision rules from full-rank conjoint data or partial-rank data such as consider-then-rank, consider-only, or choice data. Potential decision rules include elimination by aspects, acceptance by aspects, lexicographic by features, and a mixed-rule lexicographic by aspects (LBA) that nests the other rules. We provide a dynamic program that makes estimation practical for a moderately large numbers of aspects. We test greedoid methods with applications to SmartPhones (339 respondents, both full-rank and considerthen-rank data) and computers (201 respondents from Lenk et al. 1996). We compare LBA to two compensatory benchmarks: hierarchical Bayes ranked legit (HBRL) and LINMAP. For each benchmark, we consider an unconstrained model and a model constrained so that aspects are truly compensatory. For both data sets, LBA predicts (new task) holdouts at least as well as compensatory methods for the majority of the respondents. LBA's relative predictive ability increases (ranks and choices) if the task is full rank rather than consider then rank. LBA's relative predictive ability does not change if (1) we allow respondents to presort profiles, or (2) we increase the number of profiles in a consider-then-rank task from 16 to 32. We examine trade-offs between effort and accuracy for the type of task and the number of profiles. [ABSTRACT FROM AUTHOR]

Adaptive metric utility balance is at the heart of one of the most widely used and studied methods for conjoint analysis. We use formal models, simulations, and empirical data to suggest that adaptive metric utility balance leads to partworth estimates that are relatively biased--smaller partworths are upwardly biased relative to larger partworths. Such relative biases could lead to erroneous managerial decisions. Metric utility-balanced questions are also more likely to be inefficient and, in one empirical example, contrary to popular wisdom, lead to response errors that are at least as large as nonadaptive orthogonal questions. We demonstrate that this bias is because of endogeneity caused by a "winner's curse." Shrinkage estimates do not mitigate these biases. Combined with adaptive metric utility balance, shrinkage estimates of heterogeneous partworths are biased downward relative to homogeneous partworths. Although biases can affect managerial decisions, our data suggest that, empirically, biases and inefficiencies are of the order of response errors. We examine viable alternatives to metric utility balance that researchers can use without biases or inefficiencies to retain the desired properties of (1) individual-level adaptation and (2) challenging questions. [ABSTRACT FROM AUTHOR]

Increasingly, researchers in marketing are recognizing the "lability" of attribute importance weights derived from measurement techniques, such as conjoint analysis. As has been suggested by Simonson and Tversky, attribute importance weights can be sensitive to competitive product context and to purchase situation. This paper describes and applies a procedure for adjusting conjoint importance weights to predict consumers' actual or potential product choices. We discuss the approach from both a descriptive and prescriptive viewpoint. In particular, the latter perspective provides strategic insights into how attribute importance modifications can increase brand share. An industry case, based on real data, is used to illustrate the approach. (Competitive Strategy, Scaling Methods) [ABSTRACT FROM AUTHOR]

This paper describes a case application of SIMOPT, a product positioning model and decision support system. [ABSTRACT FROM AUTHOR]

This paper looks at the comparative reliability of different methodological variants of the conjoint analysis procedure. It differs from previous studies in that it looks at three methods of data collection (Full Profile, Trade-off Matrices, and Paired Comparison) and two levels of a key attribute (price) across five different product categories. In addition it tests these manipulations using two different reliability assessment procedures. The results show that most manipulations have a significant effect on the reliability scores and many interaction terms are significant. (Conjoint Analysis; Data Collection; Product Categories; Reliability Assessment) [ABSTRACT FROM AUTHOR]

This paper presents a simultaneous approach to estimating partworths and aggregating consumers in conjoint analysis. First, based on a logit choice model, a "ridge"-like procedure for estimating individual partworths is proposed, and using simulation, the reliability of the estimator is examined. A method of aggregating consumers homogeneous in preference is then presented. The effectiveness of the aggregation procedure is tested on the simulated data. Finally, to illustrate how the total procedure can be applied to the real world, a set of preference data for Japanese automobiles is used. (Logit Model; Market Segmentation; Clustering; Information Measure) [ABSTRACT FROM AUTHOR]

Formulas are derived which estimate the accuracy of conjoint analysis in predicting preferences in a validation sample. This accuracy turns out to depend on (among other things) which model of preference is used (e.g., whether interactions are added, whether partworth or linear functions are used). I first show a paradoxical result that simpler models often yield higher predictive accuracy, even when a more complex model ts the true one. The reason for this is that the additional parameters of the complex model are estimated with larger variance, which tends to overwhelm the benefits of using the true model. I then shift my criterion from predicting an individual's preferences, to predicting market share, which is of most interest to managers. Under this criterion my conclusions reverse, and I show that a true model (even when complex) is much more likely to yield higher predictive accuracy than a simpler incorrect model. This reverses some previous conclusions in marketing, and confirms that finding the correct model of consumer preference is important in improving prediction. Results from four previous empirical papers are correctly predicted by these formulas, as well as results from additional Monte Carlo studies. [ABSTRACT FROM AUTHOR]

This paper surveys economic choice theory, stressing developments that permit use of data from psychometric and conjoint experiments to produce market demand forecasts. Alternatives to the widely used multinomial logit model are summarized, and a new method for estimating multinomial probits is described. An integration of choice models with attitudinal scaling and perceptual mapping, within a latent variable system, is described. Estimation of such systems under either "random effects" or "fixed effects" descriptions of heterogeneity across individuals is discussed. Issues in the use of choice models to describe responses from conjoint experiments are presented. New regression diagnostic tests for the consistency of multinomial logit representations are discussed. [ABSTRACT FROM AUTHOR]

In conjoint analysis, a consumer's utility function for a continuous attribute is usually estimated using a part worth function. However, one may also use continuous functions. The purpose of the paper is to investigate through simulation the combined influence of the number of degrees of freedom, the nature of the "true" utility function and the amount of error in the data on the selection of a utility function. The focus is on functions that are known or expected to be monotone within the range of attribute levels of interest. One can then choose among linear, quadratic and part worth functions. The results show (a) that estimation procedures with monotonicity constraints should be used, and (b) that it is best to use quadratic or part worth functions rather than linear functions to minimize potential losses in predictive validity. (Preference Model; Conjoint Analysis; Constrained Utility Functions) [ABSTRACT FROM AUTHOR]

Consumers' preferences can often be represented using a multimodal continuous heterogeneity distribution. One explanation for such a preference distribution is that consumers belong to a few distinct segments, with preferences of consumers in each segment being heterogeneous and unimodal. We propose an innovative approach for modeling such multimodal distributions that builds on recent advances in sparse learning and optimization. We apply the model to conjoint analysis where consumer heterogeneity plays a critical role in determining optimal marketing decisions. Our approach uses a two-stage divide-and-conquer framework, where we first divide the consumer population into segments by recovering a set of candidate segmentations using sparsity modeling, and then use each candidate segmentation to develop a set of individual-level heterogeneity representations. We select the optimal individual-level heterogeneity representation using cross-validation. Using extensive simulation experiments and three field data sets, we show the superior performance of our sparse learning model compared to benchmark models including the finite mixture model and the Bayesian normal component mixture model. Data, as supplemental material, are available at . [ABSTRACT FROM AUTHOR]

In conclusion, the authors' results on simplifying preference models would seem to be relevant for a wider range of choice simulators and for up to 32 trials per subject. Further, hybrid designs can be used to improve accuracy in the preference model while not fatiguing subjects. Commentators have pointed the way to a valuable research program to find out: how fatigue increases with the number of trials, how best of model the choice simulator, and how to adjust for advertising and distribution. As one continues on this research program, professionals and academics will converge on more accurate and effective ways to predict choice. Apparently, practitioners are already using these extensions, though very little is published yet on these applications. For example, empirical studies on the form of actual preference distributions across consumers, and on the magnitude of correlations among products, would be very useful in designing better choice simulators. The author's paper did not consider the potential effects of subject fatigue on predictive accuracy of conjoint analysis.

Although the assumption of utility-maximizing consumers has been challenged for decades, empirical applications of alternative choice rules are still very new. We add to this growing body of literature by proposing a model based on the idea of a "satisficing" decision maker. In contrast to previous models (including recent models implementing alternative choice rules), satisficing depends on the order in which alternatives are evaluated. We therefore conduct a visual conjoint experiment to collect search and choice data. We model search and product evaluation jointly and allow for interdependence between them. The choice rule incorporates a conjunctive rule for the evaluations and, contrary to most previous models, does not rely on compensatory trade-offs at all. The results strongly support the proposed model. For instance, we find that search is indeed influenced by product evaluations. More importantly, the model results strongly support the satisficing stopping rule. Finally, we perform a holdout prediction task and find that the proposed model outperforms a standard multinomial logit model. [ABSTRACT FROM AUTHOR]

We develop and test an incentive-compatible Conjoint Poker (CP) game. The preference data collected in the context of this game are comparable to incentive-compatible choice-based conjoint (CBC) analysis data. We develop a statistical efficiency measure and an algorithm to construct efficient CP designs. We compare incentive-compatible CP to incentive-compatible CBC in a series of three experiments (one online study and two eye-tracking studies). Our results suggest that CP induces respondents to consider more of the profile-related information presented to them compared with CBC. [ABSTRACT FROM AUTHOR]

We study how multiattribute product choices are affected by peer influence. We propose a two-stage conjoint-based approach to examine three behavioral mechanisms of peer influence. We find that when faced with information on peer choices, consumers update their attribute preferences in a Bayesian manner. This suggests that greater uncertainty in the attribute preferences of a focal consumer and lesser uncertainty in preferences of peers both lead to greater preference revision. Greater number of peers is associated with greater preference revision, although the extent of preference revision diminishes with increasing number of peers. Furthermore, to address the significant time and costs associated with collecting sociometric data, we estimate the accuracy of predicted consumer choices when peer influence data are unavailable. Online social network membership and frequency of peer interactions provide better proxies than more common demographic similarity measures. These findings have key implications, especially for word-of-mouth marketing. [ABSTRACT FROM AUTHOR]

Brand preferences and marketplace demand are a reflection of the importance of underlying needs of consumers and the efficacy of product attributes for delivering value. Dog owners, for example, may look to dog foods to provide specific benefits for their pets (e.g., shiny coats) that may not be available from current offerings. An analysis of consumer wants for these consumers would reveal weak demand for product attributes resulting from low efficacy, despite the presence of strong latent interest. The challenge in identifying such unmet demand is in distinguishing it from other reasons for weak preference, such as general noninterest in the category and heterogeneous tastes. We propose a model for separating out these effects within the context of conjoint analysis, and we demonstrate its value with data from a national survey of toothpaste preferences. Implications for product development and reformulation are explored. [ABSTRACT FROM AUTHOR]

We introduce methods from statistical learning theory to the field of conjoint analysis for preference modeling. We present a method for estimating preference models that can be highly nonlinear and robust to noise. Like recently developed polyhedral methods for conjoint analysis, our method is based on computationally efficient optimization techniques. We compare our method with standard logistic regression, hierarchical Bayes, and the polyhedral methods using standard, widely used simulation data. The experiments show that the proposed method handles noise significantly better than both logistic regression and the recent polyhedral methods and is never worse than the best method among the three mentioned above. It can also be used for estimating nonlinearities in preference models faster and better than all other methods. Finally, a simple extension for handling heterogeneity shows promising results relative to hierarchical Bayes. The proposed method can therefore be useful, for example, for analyzing large amounts of data that are noisy or for estimating interactions among product features. [ABSTRACT FROM AUTHOR]

It has been shown in the behavioral decision making, marketing research, and psychometric literature that the structure underlying preferences can change during the administration of repeated measurements (e.g., conjoint analysis) and data collection because of effects from learning, fatigue, boredom, and so on. In this research note, we propose a new class of hierarchical dynamic Bayesian models for capturing such dynamic effects in conjoint applications, which extend the standard hierarchical Bayesian random effects and existing dynamic Bayesian models by allowing for individual-level heterogeneity around an aggregate dynamic trend. Using simulated conjoint data, we explore the performance of these new dynamic models, incorporating individual-level heterogeneity across a number of possible types of dynamic effects, and demonstrate the derived benefits versus static models. In addition, we introduce the idea of an unbiased dynamic estimate, and demonstrate that using a counterbalanced design is important from an estimation perspective when parameter dynamics are present. [ABSTRACT FROM AUTHOR]

We propose and test new adaptive question design and estimation algorithms for partial-profile conjoint analysis. Polyhedral question design focuses questions to reduce a feasible set of parameters as rapidly as possible. Analytic center estimation uses a centrality criterion based on consistency with respondents' answers. Both algorithms run with no noticeable delay between questions. We evaluate the proposed methods relative to established benchmarks for question design (random selection, D-efficient designs, adaptive conjoint analysis) and estimation (hierarchical Bayes). Monte Carlo simulations vary respondent heterogeneity and response errors. For low numbers of questions, polyhedral question design does best (or is tied for best) for all tested domains. For high numbers of questions, efficient fixed designs do better in some domains. Analytic center estimation shows promise for high heterogeneity and for low response errors; hierarchical Bayes for low heterogeneity and high response errors. Other simulations evaluate hybrid methods, which include self-explicated data. A field test (330 respondents) compared methods on both internal validity (holdout tasks) and external validity (actual choice of a laptop bag worth approximately $100). The field test is consistent with the simulation results and offers strong support for polyhedral question design. In addition, marketplace sales were consistent with conjoint-analysis predictions. [ABSTRACT FROM AUTHOR]

The usefulness of a technology product for an end-user often depends on the availability of complementary software products and services. Computers require software, cameras require film, and DVD players require movie programming in order for customers to value the whole product. This phenomenon, where the demand for hardware products is mediated by the supply of complementary software products, is called an indirect network externality. Indirect network externalities create a two-way contingency between the demand for the hardware product and the supply of software products, and result in a strategic interdependence between the actions of hardware manufacturers and the actions of software providers. Indirect network externalities are gaining economic significance in technology markets, because hardware and software are typically provided by independent firms, and both sets of firms have an incentive to free- ride on each others' demand creation efforts. Despite the ubiquity of this phenomenon, it has largely been ignored in the marketing science literature. We present a conceptual and operational model for the evolution of markets with indirect network externalities. The key feature of our framework is to model the market-mediated dependence between the actions of hardware manufacturers and software complementors, created by the direct dependence of consumer demand for the whole product on the actions of manufacturers as well as complementors. In addition, we incorporate marketing-mix effects on consumer response, as well as heterogeneity in consumer preferences for hardware and software attributes. We model consumer response using a latent-class choice model. To estimate the complement or response functions, we use a modified Delphi technique that allows us to convert qualitative response data into quantitative response functions. We integrate the consumer and complement or response models to create a simulation model that generates forecasts of market shares and sales volumes for competing technologies, as a function of marketing-mix effects and exogenously specified regulatory scenarios. The modeling framework is of interest to new product modelers interested in creating empirical models and decision-support systems for forecasting demand in technology markets characterized by indirect network externalities. The decision-support aspects of the modeling framework should appeal to managers interested in understanding and quantifying the complex interplay between hardware manufacturers and software complementors in the evolution of markets with indirect network externalities. We present an application of the modeling framework to the U.S. digital television industry, and use the framework to characterize the competition among analog and digital TV technologies. Our results suggest that complementor actions play an important role in the acceptance of digital TV technologies in general, and high definition television (HDTV) in particular. We find that forecasts that ignore the influence of indirect network externalities would be seriously biased in favor of HDTV. We illustrate how the modeling framework can be used to identify and profile customer segments in the digital TV market based on their utility for hardware-related features as well as programming-related features. We also illustrate the decision-support capabilities of the modeling framework by evaluating the sensitivity of the forecasts to varying marketing, regulatory, and complementor response scenarios. We derive implications for marketing and public affairs policies of the hardware manufacturers. The developments in the digital TV industry generally support our finding that HDTV will be a niche product, and will diffuse slower than originally expected due in part to the lack of programming. The delays in the introduction of digital TV to the marketplace also suggest that most forecasts for infrastructure-intensive technologies like digital TV may be too optimistic simply because they underestimate the delays in agreeing upon technology standards and resolving regulatory debates. [ABSTRACT FROM AUTHOR]

Experimental conjoint choice analysis is among the most frequently used methods for measuring and analyzing consumer preferences. The data from such experiments have been typically analyzed with the Multinomial Logit (MNL) model. However, there are several problems associated with the standard MNL model because it is based on the assumption that the error terms of the underlying random utilities are independent across alternatives, choice sets, and subjects. The Multinomial Probit model (MNP) is well known to alleviate this assumption of independence of the error terms. Accounting for covariances in utilities in modeling choice experiments with the MNP is important because variation of the coefficients in the choice model may occur due to context effects. Previous research has shown that subjects' utilities for alternatives depend on the choice context, that is, the particular set of alternatives evaluated. Simonson and Tversky's tradeoff contrast principle describes the effect of the choice context on attribute importance and patterns of choice. They distinguish local contrast effects, which are caused by the alternatives in the offered set only, and background contrast effects, which are due to the influence of alternatives previously considered in choice experiments. These effects are hypothesized to cause correlations in the utilities of alternatives within and across choice sets, respectively.The purpose of this study is to develop an MNP model for conjoint choice experiments. This model is important for a more detailed study of choice patterns in those experiments. In developing the MNP model for conjoint choice experiments, several hurdles need to be taken related to the identification of the model and to the prediction of holdout profiles. To overcome those problems, we propose a random coefficients (RC) model that assumes a multivariate normal distribution of the regression coefficients with a rank one factor structure on the covariance matrix ... [ABSTRACT FROM AUTHOR]