As researchers soon discover, the inclusion of noisy (irrelevant) variables in cluster analyses can obscure or distort Atrue@ subgroup structures. This problem, identified and discussed by Milligan (1980), has prompted the search for methods that identify noisy variables and either down-weight or remove them. Several researchers have investigated this problem and have met with limited success (DeSarbo, Carroll, Clark, and Green 1984, De Soete 1986, 1988). Recently, Donoghue (1995) and Carmone, Kara, and Maxwell (forthcoming, 1999) have proposed screening methods to identify and eliminate noisy variables.

Researchers of cluster analysis are becoming increasingly interested in methods that finds a set of variable weights that optimize some index of “clusteriness” such as the modified Rand index. This paper describes a sample replication method for finding a set of SYNCLUS-based variable weights for the practical case in which the weights are to be applied to large data bases constituting hundreds or even thousands of observations. A Monte Carlo simulation of synthetic data sets is used to test the replicated variable weights estimation procedure.

A recent critique of Sawtooth Software’s Adaptive Conjoint Analysis model proposes an iterative procedure for adjusting self-explicated response scales to agree more closely with the respondent’s partial profile evaluations. The present paper implements this suggestion in the context of two variants of a full-profile, nonadaptive hybrid model. The model is described formally, and then it is applied to an illustrative data set.

This paper describes PRIDEL, a model for product line pricing that uses conjoint input data. The model entails three types of optimization: (a) optimal product line pricing, conditional on specified levels of non-price attributes; (b) non-price level optimization, conditional on specified price levels; and (c) overall optimization of product line prices and non-price attribute levels. The model is described and applied to an actual (disguised) case, involving product line pricing for home video games.

This article is a response to Richard Johnson's (2005) memoir. It notes some of the important contributions that Johnson has made to marketing research and reflects on his unique personal and professional characteristics.

Collecting and analyzing respondents’ conjoint data is an essential part of the analytical process. As noted by John Hauser and Vithala Rao in their excellent discussion of conjoint analysis over the past 30 years (see Chapter 6), marketing researchers have expended intense efforts on data collection and partworth estimation. Partworths have been estimated by full profile, Adaptive Conjoint Analysis, hybrid conjoint, and categorical conjoint, with or without empirical or hierarchical Bayes enhancement. Partworth measurement and estimation processes are central to the accuracy and usefulness of all conjoint studies. Accordingly, much has been written about the pros and cons of various conjoint data collection and parameter estimation methods.

Models for optimal product positioning have received considerable attention by marketing researchers and marketing scientists over the past decade. Typically, optimizing models take the viewpoint that the manager wishes to find a specific vector of product attribute levels that, in the face of competitors' product profiles, maximizes the firm's market share (or, perhaps, return) over some designated planning horizon. This class of models emphasizes long run strategic modeling.In contrast, the authors introduce a tactical, short-term model, called SALIENCE, whose purpose is to allocate sales efforts in such a way as to increase the relative importance of attributes for which the sponsoring firm's current product has a (possibly temporary) differential advantage. In this case emphasis is on short-run, tactical decision making.We describe the SALIENCE model, both informally and mathematically. The model is applied, illustratively, to a real (disguised) study of overnight air shipment delivery. (c) 2004 Elsevier B.V. All rights reserved.

On behalf of the Wharton School, Dean Patrick Harker, and myself, I would like to welcome those of you who have traveled to be with us to recognize Paul Green's many contributions.

Paul Green and a Brief History of Marketing Research.- 1. When BDT in Marketing Meant Bayesian Decision Theory: The Influence of Paul Green's Research.- 2. Applications of Multivariate Latent Variable Models in Marketing.- 3. Multidimensional Scaling and Clustering in Marketing: Paul Green's Role.- 4. Market Research and the Rise of the Web: The Challenge.- 5. Thirty Years of Conjoint Analysis: Reflections and Prospects.- 6. Conjoint Analysis, Related Modeling, and Applications.- 7. Buyer Choice Simulators, Optimizers, and Dynamic Models.- 8. A 20+ Years' Retrospective on Choice Experiments.- 9. Evolving Conjoint Analysis:From Rational Features/Benefits to an Off-the-Shelf Marketing Database.- 10. The Vagaries of Becoming (and Remaining) a Marketing Research Methodologist.- 11. The Journal of Marketing Research: Its Initiation, Growth, and Knowledge Dissemination.- 12. Personal Reflections and Tributes from the May 2002 Conference Celebrating Paul Green's Career.- 13. Continuing the Aldersonian Tradition.- 14. Reflections and Conclusions: The Link Between Advances in Marketing Research and Practice.- Appendix: Paul E. Green's Curriculum Vita.

Methodological developments in marketing research would go for naught if the methodology and research techniques were not applied by practitioners (i.e., industry consultants, marketing research firms, and intra-company professionals) and adopted by management. As noted throughout this book, one of the hallmarks of work by pioneers such as Wroe Alderson has been an intense interest in both research and practice. While individuals have spanned these boundaries, there is still work to do in building bridges between research and practice. This will ensure that new ideas are disseminated quickly into the hands of those who can benefit most from them and also will help maintain the relevance of research advances in the field.

Conjoint analysis is a class of techniques for analyzing consumers’ preferences and trade‐offs regarding their selection of products and services. Typically, conjoint analysis has been applied to established markets such as frequently purchased packaged goods, consumer durables, communication services, and business‐to‐business products. Recently, marketing researchers have extended conjoint methodology to cope with the measurement of buyer trade‐offs associated with “really new” products and services, for which there is little or no prior buyer knowledge or experience. The researcher’s task is twofold: to educate the potential buyer regarding the pros and cons of the new product/service while, at the same time, obtaining the respondent’s evaluation of the new product/service itself. This article describes the application of conjoint techniques to a new service, TrafficPulse, that enables subscribers to obtain continuous 24/7 updates on traffic conditions, travel times, and alternative routes, should congestion occur. In particular, describes how traditional conjoint analysis can be embellished to obtain relevant information about consumer evaluations of new goods and services prior to their actual use by prospective consumers. In short, the prospective consumer can be “educated” about the new product/service before obtaining evaluation of its potential worth. The paper also shows how conjoint analysis can be modified to accommodate restrictions on various attribute levels, how the use of BASES‐like norms can be incorporated, and how optimization algorithms can be used at either the single product or multiple product (i.e. product line) level.

Sometimes simulator designers are so enamored of the technology to create detailed road scenes that they lose sight of the goals that guide simulator design, with usability and usefulness suffering as a result. The third generation UMTRI Driving Simulator was designed with 4 goals in mind: (1) facilitate running experiments, (2) support demonstrations, (3) facilitate configuration verification and troubleshooting, and (4) provide high quality audio and video signals. The time spent on demonstrations (to sponsors, the public, and the media) can equal the time testing subjects and the impact of those demonstrations can be significant. Further, the time spent troubleshooting equipment, especially the cameras, cables, amplifiers, etc. associated with the audiovisual system can far exceed the time spent on demonstrations and testing subjects. This paper lists 30 recommendations for designing a simulator audiovisual system. For many recommendations, specific model numbers or web sites are provided. Noteworthy recommendations in support of the testing and demonstration goals (1 and 2) concern (a) the camera locations to consider (face, screen, interior, feet), (b) hiding cameras (yes, using lipstick cameras), (c) monitoring of the experiment (the operator should see everything), (d) audio and video switching (use a one, not multiple mixers for each input and output modality), (e) assuring visitors can see and hear everything subjects see and hear, (f) accommodating TV crews, (g) light control (use black carpet), and (h) power (provide UPS for all critical items). To facilitate troubleshooting (goal 3), label and diagram everything, use geared tripod heads to aim the LCD projectors, and provide access to the front and back of all equipment. Finally, to assure signal quality (goal 4), use broadcast quality components, especially shielded cable, and double check all hand made cables. Although many of the recommendations may seem obvious in hindsight, many were not thought of during planning of a new simulator by a team with considerable experience with using and developing driving simulators. Users of existing simulators and those building new simulators should benefit from the recommendations in this paper.

The Journal of Marketing Research has now reached almost four decades of growth since its initiation in 1964. It continues to be a rich source of innovative ideas in research methodology and knowledge diffusion for both practitioners and academics. This article pays tribute to the many scholars who have contributed to the Journal of Marketing Research's success and continued service to marketing researchers worldwide. The authors describe the early years of the Journal of Marketing Research and illustrate how it has become the premier journal in marketing research methodology. In so doing, they discuss its important role in the dissemination of knowledge through the Advanced Research Techniques Forum and other special interest marketing groups. The authors also stress the significant role of software development in the implementation of the increasingly complex computations associated with advanced marketing research methods.

The art (and science) of successful product/service positioning generally hinges on the firm's ability to select a set of attractively priced consumer benefits that are:valued by the buyer,distinctive in one or more respects,believable,deliverable, andsustainable (under actual or potential competitive abilities to imitate, neutralize, or overcome)in the target markets that the firm selects. (While easily said, meeting all of these conditions is often the exception rather than the rule.)For many years, the ubiquitous quadrant chart has been used to provide a simple graph of product/service benefits (usually called product/service attributes) described in terms of consumers' perceptions of the importance of attributes (to brand/supplier choice) and the performance of competing firms on these attributes. This paper describes a model that extends the quadrant chart concept to a decision support system that optimizes a firm's market share for a specified product/service. In particular, we describe a decision support model that utilizes relatively simple marketing research data on consumers' judged benefit importances, and supplier performances on these benefits to develop message components for specified target buyers. A case study is used to illustrate the model. The study deals with developing advertising message components for a relatively new entrant in the US air shipping market. We also discuss, more briefly, management reactions to application of the model to date, and areas for further research and model extension. (C) 2002 Elsevier Science B.V. All rights reserved.

We norm (defined as the limit of an L p norm as p approaches zero). In Monte Carlo simulations, both K-modes and the latent class procedures (e.g., Goodman 1974) performed with equal efficiency in recovering a known underlying cluster structure. However, K-modes is an order of magnitude faster than the latent class procedure in speed and suffers from fewer problems of local optima than do the latent class procedures. For data sets involving a large number of categorical variables, latent class procedures become computationally extremly slow and hence infeasible. We conjecture that, although in some cases latent class procedures might perform better than K-modes, it could out-perform latent class procedures in other cases. Hence, we recommend that these two approaches be used as "complementary" procedures in performing cluster analysis. We also present an empirical comparison of K-modes and latent class, where the former method prevails.