Knowledge that Transforms

The authors develop a flexible model to analyze relative preference scores: the binomial/Dirichlet model. This model assumes that (1) individual respondents make independent draws from binomial distributions when stating their preferences and (2) the latent (unobserved) preference parameters vary across respondents according to a Dirichlet distribution. Through the analysis of 44 tests that include from two to five products each, the authors show that the model fits the data relatively well. A multinomial/Dirichlet extension of the model that applies to repeat preference tests of two items provides a better fit than an alternative mixture model despite fewer parameters. To test two items and obtain an accuracy of +.05 with a 95% confidence interval for the mean preference intensities, a multinomial/Dirichlet model requires two paired comparisons (made at two points in time) per respondent and a sample size of 400; these requirements represent half the required number of preference measurements per respondent and half the required sample size of alternative methods. Although the illustrative examples refer to the comparison of known brands and unidentified products, the proposed methodology can be applied to many contexts, including the evaluation of product profiles in conjoint analysis.

Response latencies provide information about consumers' choice behavior in a conjoint choice experiment. The authors use filtered response latencies to scale the covariance matrix of a multinomial probit model and show that this leads to better model fit and holdout predictions, even if the response latencies in the holdout task are not used. The authors provide an empirical application along with a tentative explanation for the findings of the effect of response latencies.

The authors develop a model to decompose the effects of television advertising for a toll-free referral service, at the hourly level. The model estimates which ad works, when, in which station, and for how long. Results of the analysis show that ads do stimulate direct response, but their effects dissipate very rapidly. Effectiveness and profitability vary substantially by creative, television station, and station x time of the day. The results underscore the need for managers to undertake such analyses and for researchers to use such a disaggregate approach.

In this article, the authors develop a class of models to reconstruct brand-transition probabilities when individual brand purchase sequence information is not available. The authors introduce two general model forms by assuming different underlying mechanisms for individual heterogeneity in brand switching. The first model form captures individual heterogeneity by a latent class structure. The second model form captures individual heterogeneity by postulating that the brand-choice probabilities follow a Dirichlet distribution, which yields the popular Dirichlet multinomial formulation. Monte Carlo simulations are performed with a view toward assessing whether individual transition probabilities can be captured from knowledge of only aggregated brand choices. Results indicate that the proposed method can indeed capture individual brand-transition probabilities under several different conditions. An empirical application illustrates how these models can be used to provide important information on individual brand transitions and the role of marketing-related covariates.

Commercial market research firms provide information on advertising variables of interest, such as brand awareness or gross rating points, that are likely to contain measurement errors. This unreliability of measured variables induces bias in the estimated parameters of dynamic models of advertising. Consequently, advertisers either under- or overspend on advertising to maintain a desired level of brand awareness. Monte Carlo studies show that the magnitude of bias can be serious when conventional estimation methods, such as ordinary least squares and errors in variables, are employed to obtain parameter estimates. Therefore, the authors have developed two new approaches that either reduce or eliminate parameter bias. Using these methods, advertisers can determine an unbiased optimal advertising budget, even if advertising variables are measured with error. The application of these methods to estimate the extent of measurement noise in empirical advertising data is illustrated.

Different streams of research offer seemingly conflicting predictions as to the effects of analyzing reasons for preferences on the attitude-behavior link. The authors apply these different theoretical accounts to a new product scenario and identify conditions under which analyzing reasons for brand preferences can increase or decrease the predictive value of reported preferences. Consistent with dual-process theories of persuasion, in Study 1 the authors find that reasons analysis increases the link between attitude and behavior when the measure of behavior closely follows attitude measurement. In contrast, and consistent with research by Wilson and colleagues (e.g., Wilson et al. 1989 ) on the disruptive effects of reasons analysis, the authors find that thinking about reasons significantly decreases the attitude–behavior correlation when the observed behavior occurs after a substantial delay. Study 2 not only replicates this finding but also suggests that the timing of the reasons task can be an important moderator of the disruption effect. Specifically, the authors draw on the literature on accountability effects to show that even when there is a delay between attitude and behavior measurement, reasons analysis leads to an increase in the attitude–behavior link, as long as reasons are analyzed after attitude measurement. Finally, in Study 3, the authors validate the account of the effects of reasons analysis by obtaining parallel findings for attitude persistence. Together, the studies offer preliminary advice to both practitioners and academics regarding the potential effects of asking consumers to think about why they like or dislike certain products.

In marketing, qualitative data are used in theory development to investigate marketing phenomena in more depth. After qualitative data are collected, the judgment-based classification of items into categories is routinely used to summarize and communicate the information contained in the data. In this article, the authors provide marketing researchers with a method that (1) provides useful substantive information about the proportion and degree to which items belong to several categories and (2) measures the classification accuracy of the judges. The model is called the fuzzy latent class model (FLCM), because it extends Dillon and Mulani's (1984) latent class model by freeing it from the restrictive assumption that all items are crisp for a given categorization. Instead, FLCM allows for items to be either crisp or fuzzy. Crisp items belong exclusively to one category, whereas fuzzy items belong—in varying degree—to multiple categories. This relaxation in the assumption about the nature of qualitative data makes FLCM more widely applicable: Qualitative data in marketing research are often fuzzy, because they involve open-ended descriptions of complex phenomena. The authors also propose a moment-based measure of overall data fuzziness that is bounded by 0 (completely crisp) and 1 (completely fuzzy).