Journal of Marketing Research

The authors report on a procedure for exploiting the information content of rank ordered choice sets to estimate efficiently the parameters of the multinomial logit model formulation of the stochastic utility model of choice behavior. The availability of rank ordered choice set data leads to an “explosion” or decomposition procedure for exploiting such extra information. This “explosion” process involves the decomposition of a ranked choice set into a series of unranked and statistically independent choice sets. In relation to explosion strategies, several heuristics and an analytical procedure for determining the “optimal” explosion depth are discussed in detail. The results of a Monté Carlo study of the small sample properties of the conditional logit estimation procedure (the maximum likelihood estimation procedure used to develop parameter estimates of the multinomial logit model formulation of the stochastic utility model) are reported and interpreted. A college choice empirical application illustrates the procedures developed.

Recently interest has centered on extending decompositional methods for modeling consumer preferences to include group-level effects. The motivation for this effort is the general belief that respondent background characteristics and situation variables frequently interact with product attributes in jointly influencing preferences over a set of multiattribute alternatives. In much the same way, respondent and situation factors can interact with product attribute perceptions in perceptual mapping applications to influence the respondent's ratings of the brands. The authors describe how respondent background characteristics and situation variables can be accounted for as sources of variation which can potentially influence the reduced space configuration of brands. Essentially the approach can be viewed as chaining multivariate analysis of variance and standard perceptual mapping techniques. Two illustrative examples are provided. In one example the results show that the uncovered spatial configuration of brands was influenced not only by the perceptual differences between the brands, but also by a respondent-related factor.

A modified model for examining power and its sources in marketing channels is developed and tested. The results of testing this model indicate that its predictive ability is no better than that of previous approaches. Some unexpected results obtained in this and other studies are discussed in the context of an overall framework of power sources and their consequences in marketing channels.

Though most observers acknowledge that consumers make inferences among product attributes, the influence of such inferences on product evaluation is much less clear. Study respondents evaluated products for which information on one of two attributes was systematically omitted. A general model is built to estimate the directional effect of inferences on product evaluation. The effect of inferences to a missing attribute is statistically significant and in the expected direction. In one case, the marginal value of the remaining attribute (price) reverses in sign because of an inference. Thus, inferences are theoretically important and a potentially troublesome issue in the modeling and measurement of consumer choice processes.

The authors describe an approach to measuring self- and cross-price/demand relationships via conjoint analysis. The modeling “trick” is to characterize each brand as a separate price-related factor in which respondents are asked to indicate their subjective likelihood of buying a set of J brands, each priced at some experimental level. A first-order Markovian model is postulated to represent price-induced brand switching. The price-demand model is then fitted (by generalized least squares) to share of choices, estimated after T time periods. The model is illustrated empirically in the context of a consumer nondurable good.

The concept of customer orientation in salespeople is defined, a scale is developed to measure the degree to which salespeople engage in customer-oriented selling, and the properties of the scale are reported. A test of the nomological validity indicates the use of customer-oriented selling is related to the ability of the salespeople to help their customers and the quality of the customer-salesperson relationship.

Although empirical evidence shows that advertising wears out over time and that advertising pulsation may be superior to a strategy of constant spending, current advertising models neither represent the wearout phenomenon nor consider pulsation as an optimal strategy. In this article, the wearout phenomenon is represented in a model which distinguishes between level stimuli and differential stimuli of advertising. The model is applied to several brands, and both advertising stimuli are found to be significant. A pulsation strategy is shown to be optimal under both an unconstrained and a constrained advertising budget.

The author introduces relative demand, defined as the proportion of a country's personal consumption expenditure spent on a good, and describes relative demand as a function of the level of saturation of the good. He compares parameters of these functions among countries and discusses their use in predicting demand.

Many, if not most, of the important functional relationships in marketing are inherently nonlinear. Fitting a nonlinear function to data involves first the specification of a model, and then the estimation of the model's parameters. Where theory gives insufficient guidance for the specification of a model, specification based on observation of the scatterplot may be required. In many instances it is difficult to specify a model accurately solely on the basis of scatterplot observation. In those instances the researcher's power of observation can be enhanced by use of a distribution-free method of approximating the underlying function. The authors discuss alternative distribution-free methods, and demonstrate how the implicit model specification process of exhaustive trial and error curve-fitting can be replaced by an explicit model specification process based on observing an approximation of the underlying function.

The relative efficacy of two types of incentives for producing responses to a mail survey was investigated. Personal cash payments of varying amounts enclosed with the questionnaire were compared with promises of a contribution to a charity of the respondent's choice and with a no-incentive control. Contrary to a previous research finding reported in the literature, the charity incentive did not produce a significantly greater return rate than was obtained with the no-incentive control. Personal cash incentives produced a significantly greater response rate than either the no-incentive control condition or the charity-incentive condition. Results of the study are interpreted as consistent with a cognitive dissonance theory of mail questionnaire response.

Random selection of eligible persons within households typically is done by procedures developed by Kish or by Troldahl-Carter-Bryant. These procedures require different tasks from respondent and interviewer. The differences have been thought to affect cooperation rates and/or respondent/household characteristics. The authors examine the Kish procedure and two versions of the Troldahl-Carter-Bryant procedure and conclude that Kish and one version of Troldahl-Carter-Bryant are the preferred methods.

The authors describe alternative approaches for determining rank-order agreement on a common set of brands between two groups of subjects as well as within each group. Specific attention is given to the application of U-statistics to the two-group rank-order problem in consumer and marketing research.

This note adds further evidence of the importance of questionnaire layout, particularly for self-administered questionnaires.