Journal of Marketing Research

The Fishbein intention model often produces weak and inconsistent predictions in marketing applications. Of equal concern, high multicollinearity between independent variables is typical. A new intention model was developed and tested. Results show high stability, reliability, and predictiveness across diverse test products, with very low intercorrelations between explanatory variables.

Repetition of a pattern of television commercials caused wearout in viewers’ evaluation of the commercials and the products being advertised. As predicted by an information processing view, wearout was not forestalled by strategies designed to enhance attention.

A three (message appeal: supportive, brand X, and comparative) by two (product class utility: social and functional) by three (information load: low, medium, and high) factorial experimental design (with covariates) was used to explore communications effectiveness of the three modes of advertising appeals. ANOVA results show that the comparative mode appears to be superior only for the affective responses to the ad itself. The brand X mode of message appeal generally is superior to other modes for cognitive, affective, and conative responses toward the promoted brand. However, the differences are only partially significant at p < .05. In no case is the supportive mode superior. Some weak evidence of information overload is also uncovered. Several personality variables did not have significant effects on the relative effectiveness of the different appeals. Lack of substantial interaction effects among the three factors indicates that advertisers can determine message appeal, amount of information provided, and type of products to be promoted independently of each other.

Easily implemented dynamic models for pricing and production or ordering decisions are developed for products or services whose value to the consumer may be changing over time. The models are easily solved by brute force dynamic programming. Implementation examples of models for broadcast pricing are presented.

The lognormal distribution is shown to account well for purchase frequency heterogeneity in the U.S. dentifrice market, including subgroups of households. Renewal process assumptions are used to define the market population. As the definition would exclude some very infrequent buyers, left-truncated lognormal distributions are fitted to the data.

The authors report the sample sizes required to obtain maximum likelihood estimates of parameters in a zero-order model. In the typical formulation of the model, parameter estimation requires about 2000 individuals with five purchases per consumer. The original parameters can be easily transformed, however, into a market share estimate and a polarization index that convey the key information sought from the model. Estimation of the transformed parameters requires smaller samples of about 400 individuals with five purchases per consumer.

Three factor analytic procedures, principal components analysis, image analysis, and alpha factor analysis, are compared by means of a Monté Carlo simulation. The simulation allows systematic tests of the ability of each method to recover prespecified factor patterns. Image and alpha analysis are found to be more accurate than principal components analysis, especially for the type of problem to which factor analysis is applied in marketing. Recommendations for applied research and suggestions for future methodological work are made on the basis of the study findings.

The role of response latency in the measurement of preferences is investigated by means of a causal modeling method. The convergent validity and predictive validity of response latency measures are examined and compared with those of constant sum and paired comparison procedures.

Often the response rate for a given market segment is a function of the relevance of the topic to that segment. Consequently, in situations where the topic interest differs among segments, it is not correct to use the proportion of returned responses that indicate interest in the topic as the estimate of the population proportion. The authors describe three different models useful in assessing differential response rates and the methods for estimating each model. An application of real survey data is used to compare and contrast the different estimates.

Two experiments were conducted to test the generality of the door-in-the-face compliance technique from nonbusiness to business contexts. Results indicate that the compliance-gaining procedure generalizes if a concession is emphasized by making the second request a smaller version of the first request rather than a new request, and if the second request is made large enough to avoid ceiling effects.

An often-cited study of the price-perceived quality relationship is reexamined in detail. Relationships for which extremely high levels of statistical significance are claimed are shown to be neither statistically nor substantively significant. In addition, some deficiencies in design seriously threaten the validity of the investigation.

A mailed questionnaire study examined the effect of a 25-cent monetary inducement on three components of response quality: item omission, response error, and completeness of answer. The monetary inducement is found to decrease item omission and response error significantly. Some research support is shown for improving completeness of answer.