Knowledge that Transforms

When comparing alternate product formulations, or evaluating competitors' formulations, researchers often conduct blind, forced choice product tests. In this study, the authors present an approach to designing these tests and evaluating their results. Two types of tests are considered: repeat paired comparisons formats, and formats consisting of several triangle tests and a single paired comparison. The approach consists of psychophysical assumptions, discrimination and preference constructs, and a set of analytical techniques. Using this approach, a researcher can compare product test formats to see which one most efficiently estimates a given construct of interest. The approach can also be used to gauge sample size requirements for a particular product test format and to estimate confidence intervals for an observed product test result. To illustrate the approach the authors compare two popular product test formats—the Four Paired Comparisons, and the Double Triangle Single Preference.

This paper presents research initiated in 1978 to examine the marketing effectiveness of the U.S. Navy recruiting program and to quantify the relationship between marketing efforts and enlistment achievements. The major component of the research was a one-year, controlled experiment in which levels of Navy recruiters and advertising were systematically varied and a detailed and comprehensive data base assembled. This paper describes the experiment and presents analyses of the experimental data. These analyses led to a number of conclusions regarding the impact of various factors on enlistment numbers and quality: The number of recruiters had a significant impact. Effectiveness of recruiters depended on recruiter tenure with clearly discernable learning and delearning effects. Advertising expenditures of certain types were effective, while others were not. There was a wide variation in the impact of the effective media. Socioeconomic factors such as unemployment and urbanization had major impacts. Navy marketing efforts led to an expansion of the total market for enlistments in addition to their impact on Navy enlistments. The study was carried out in a highly collaborative mode, with frequent meetings and briefings with all levels of Command. This not only facilitated control and direction of the research, but also led to a number of adaptive initiatives undertaken by policy makers. These included reallocation of the advertising budget, development of a tenure-based recruiter incentive system and a change in emphasis from recruit accessions (deliveries) to contracts (sales).

This descriptive study explores the reasons for integration of the personal selling function, i.e., the use of employee “direct” salespeople rather than manufacturers' representatives “reps”. A hyp...

Problems arise in generalizing research results based on artificial objects to the case of real products. The few previous investigations of this issue may have obtained inflated assessments of predictive validity by focusing unrealistically on stimuli with isomorphic features and perceptions. Such isomorphism occurs when each feature is matched with a uniquely appropriate perceptual response. To handle the case of nonisomorphism, where features and perceptions do not show a one-to-one correspondence, we propose a method to assess the convergent validity between preference structures for artificial objects and real products via the use of shadow features and imputed preferences. In this approach, shadow features (analogous to the economist's “shadow prices”) are weighted by utilities estimated via conjoint analysis on artificial objects to derive imputed preferences toward real products. The correlation of imputed with actual preferences gauges the convergent validity between artificial and real preference structures. This estimate of convergent validity suggests the maximum predictive validity that one can reasonably expect in generalizing results based on artificial objects to the case of real products (for similar samples of subjects and sets of stimuli). We illustrate the proposed method in a study that collected data on perceptual and affective responses to real and artificial recordings of performances by male pop singers. This illustrative application suggests that, even with satisfactory results in separate analyses for artificial objects (R̄ AF = 0.81) and real products (R̄ AS = 0.82), one may find only moderate convergent validity (r̄ AA = 0.52) and disappointing predictive validity (r 15 = 0.40) in applying utilities estimated on artificial objects to the prediction of preferences toward real products.

Based on a general production function, this study estimates productivity gains and losses for the various outputs of food retailing over the period 1959 to 1979. Because it allows for the incorporation of all factors of production rather than just labor, the methodology employed in this study is superior to conventional methods of studying productivity change. Our estimates indicate that, between 1959–1979, total factor productivity in food retailing rose at a slower rate than total factor productivity in manufacturing. While total factor productivity in food retailing rose from 1959 until about 1972, it remained constant or declined between 1972–1979. We discuss the policy and managerial implications of our results, and provide suggestions for further research.

In this paper a methodology for hierarchical market structure analysis is derived and illustrated. A probabilistic model is developed which provides a general, flexible framework which can be used to test hypothesized hierarchical market structures. Because the general probabilistic model can be translated in terms of a restricted latent class model it can be easily implemented. Among its other benefits, the proposed methodology, utilizing panel data, tracks households over a number of switches, furnishes explicit statistical tests for alternative hypothesized tree structures, as well as individual parameters, and can potentially identify clusters of households who structure the market in similar ways.

There are many situations in marketing in which several alternative quantitative models may be built to model a particular marketing phenomenon or system. Few methods exist for comparing the fit of such models if the models are not nested, especially if their performance on each of several criteria is important. This paper proposes a Bayesian cross-validated likelihood (BCVL) method for comparing quantitative models. It can be used when the models are either nested or nonnested, and is especially useful for nonnested models. A simulation based upon a typical marketing modeling situation shows the incremental benefit of using the BCVL method rather than existing techniques, and explores the circumstances under which BCVL works best. The applicability of the BCVL method is demonstrated using several typical marketing modeling situations.

Applications of market share models which (implicitly) rely on Luce's choice axiom have been widely criticized because they cannot account for the effects of differential product substitutability and product dominance. Three types of choice models—Tversky's Elimination-By-Aspects model, Tree Models, and Generalized PROBIT—have been offered as solutions to the problems identified with the Luce model, but they each suffer from limitations which have prevented their widespread application in marketing contexts. Tversky's elegant EBA model has not been widely used because it requires a large number of parameters and no special-purpose parameter estimation software has yet emerged. Tree models have been offered as more parsimonious special cases of EBA, but they are more restrictive in that they presume: (1) that products, and the process of choosing from among them, can be characterized in terms of hierarchical, attribute-based trees; and (2) that the aspects governing choice are well-known. Generalized PROBIT can paramorphically handle the problems with the Luce model, but parameter estimation software has proved problematic because it cannot guarantee a globally optimum solution. This paper proposes a new class of market share models. Rather than model the choice process explicitly, the new models simply scale the effects competing products have on each other's market share. These competitive effects are scaled in the context of a class of market-share models which: (1) do not assume a tree-like structure for the competing products; (2) do not presume any a priori knowledge about the attributes governing choice; (3) are characterized in terms of parameters that can be estimated using ordinary least-squares; and (4) provide clear managerial insight into the sources of competition. The paper begins with a brief review of previous work. Following the review, the paper offers a theorem and proof which guarantees the existence of the new class of models. The empirical validity and strategic utility of the models are demonstrated using two separate sets of data.

A probabilistic Rasch model is advocated for brand-attribute measurements which replace ubiquitous mean ratings. The usefulness of this model is then extended by showing that distinct latent processes, one extreme value and the other logistic, imply common probability structures for both its classical form and the generalization developed here. If given data reject the classical structure, an extended analysis is carried out in which logistic coefficients are estimated for the general model. These values are then used in a generalized-least-squares (GLS) procedure for estimating and testing the brand-attribute locations. An illustrative multiattribute analysis is given in which logistic coefficients and locations are found for 16 soft drinks on the continua of preference, sweetness, and fizziness.

The processes of consumer preference formation during product learning are analyzed. A hypothesis about how individuals form attribute expectations is used to derive a dynamic multinomial logit model of individual choice which endogenously recognizes product learning. The model and its underlying assumptions are then tested in the context of an interactive grocery store learning game. The results support most elements of the proposed model structure. An assumption of temporal stationarity in attribute salience, however, could not be supported. Approaches to implementation of the model are discussed, and implications for marketing management and research in individual choice modeling are addressed.