Journal of Marketing Research

Marketing researchers operate in an environment which is ripe for legislation. Researcher-respondent privacy issues, unfair and deceptive practices among researchers, and the continual problem of “sugging,” or the impersonation of legitimate interviewers by sales agents, demand attention. The authors attempt to make marketing researchers more cognizant of current legislation and the potential implications of the Magnuson-Moss/Federal Trade Commission Improvement Act. Professionals in marketing research are called upon to respond in a positive and productive fashion to the public policy issues in their operating environment.

Durable goods buyers’ prepurchase information search activities were studied to determine whether distinctive patterns of information source usage could be identified. A method based on canonical analysis of retail, neutral, and personal source usage measures in conjunction with selected explanatory variables proved useful in distinguishing four different patterns. Tradeoffs were evident in the usage of and reliance on different information sources. The results emphasize the need for further study of consumer information need and information-gathering abilities.

Since 1971, interest in the use of decompositional multiattribute preference models in marketing has been increasing. The applications have varied in terms of the type of data used, behavior predicted, and methods used for estimating parameters. The authors examine the effect of different data collection and estimation procedures on parameter estimates and their stability and validity. An actual data base is used. A detailed comparison is made of the alternative approaches of parameter estimation and suggestions are given for the potential users of decompositional multiattribute preference models.

An experimental analysis of host selling, premium, and announcer television commercials directed toward children was conducted. Stage of cognitive development, sex, and race of the children were assessed as well. Premium format commercials were more effective than host and announcer commercials in producing desired behavioral changes. Girls manifested more favorable attitudes toward the advertised product but there was no difference in the percentage of boys and girls who selected the advertised product. No differences were found between white and black children in attitude or recall but a higher percentage of white children selected the advertised product than did black children.

Patterns of television audience duplication (overlap of audiences between media vehicles) in the United States are found to be more complex than those in the United Kingdom. A new model of duplication is developed and shown to be better than previous estimates of the Viewing Law in describing television audience duplication in the U.S.

The authors report the results of a 2 × 3 factorial experiment which measures high and low prejudice white consumers’ evaluations of ads for which models’ race is manipulated. The study disconfirms the results of a previous study which differ substantially from those of the general body of literature on the effects of black models. The response of consumers to black models is found to be consistent with previous research findings.

A flexible class of sigmoid or “S-shaped” functions useful in estimating the relationship between advertising and sales or market share is developed. The class, called the “odds” models, consists of functions able to depict relationships ranging from diminishing returns to nonsymmetric S-curves, including different saturation levels and nonzero intercepts. Which one of the member models applies to a given set of data can be determined on the basis of the estimated advertising coefficient, and thus it is easy to identify whether or not the relationship under analysis is S-shaped.

Path analysis is used to examine the relationships among the salesperson's perceptions of performance feedback, participation, role clarity, and job satisfaction. The results of the study indicate that performance feedback and participation in decision making are both positively related to role clarity. In addition, performance feedback is related primarily to lower order need fulfillment whereas participation and role clarity are related primarily to higher order need fulfillment.

This article is a review of the results, as are available, on the performance of the linear discriminant function in situations where the assumptions of multivariate normality and equal group dispersion structures are violated. Some new results are discussed for the case of classification using discrete variables, and in the case of both binary and continuous variables. In addition, alternative methods which have been proposed, and evaluated, for estimating misclassification error rates are thoroughly reviewed. In all cases, the material is reviewed in terms of practical significance, with particular emphasis on the conditions unfavorable to the performance of each procedure.

In the chi square analysis of cross-tabulated variables, the problem of low cell counts is usually solved by combining categories, at the cost of a loss of information, or by using larger samples, at added financial cost. Use of a chi square statistic based on an extension of the Fisher Exact Test can provide a third, less costly, alternative.

The author compares two discrimination procedures used for analyzing nominally scaled data sets, dummy variable multiple discriminant function analysis and multivariate nominal scale analysis. Both procedures are illustrated on an in-home consumption data base.

A simple structural model relating attitudes toward business to product satisfaction, experience in shopping, and search effort is proposed and tested by use of confirmatory factor analysis (COFAMM) and the LISREL model. The fact that the basic hypotheses of the model are supported suggests a positive relation between business attitude and product satisfaction and a negative one between business attitude and information search. The LISREL model is seen as a useful tool in future work in structural modeling.

The research described was an initial effort by the Bureau of the Census to employ laboratory methods to study alternative designs of a self-administered questionnaire for use in decennial censuses of population and housing. Though no one of the three kinds of investigation described is original, the combination of methods in a single program such that results from one kind of study explain and reinforce results from others is unusual.

The “jackknife” technique developed by Tukey (1958), although rarely used in marketing, promises to be widely applicable. The very name of the technique was chosen to express its versatility. The jackknife offers the researcher the opportunity to reduce bias, perform significance tests, and assess the validity and stability of analyses without requiring a large sample. The author describes the jackknife method, applies the method to an analysis of international marketing effectiveness, and discusses some of its unresolved problems.

A spaced advertising schedule is much more dollar-effective than a massed (pulsed) schedule, according to this reanalysis of the experimental data developed by Pomerance and Zielske. This finding is at variance with the conclusions drawn by other writers. The raw Pomerance-Zielske data are published here for the first time; until now only idealized representations of the data have been available, which has been the source of confusion and misunderstanding.

The advantages of telephone surveys would be negated if telephone subscribers were not representative of total populations, including nonsubscribers. Surprisingly, the literature on this subject is somewhat limited. A study was done to ascertain whether there are any important differences between telephone subscribers and nonsubscribers. From a combination of NORC surveys, analysis of more than 7500 respondents shows that, in most cases, less than 2% of the responses on a single item of a telephone sample will differ from equivalent responses of a sample of the total population.

Researchers have used either average correlation or the percentage of objects correctly predicted in assessing the accuracy of attitude models. Because the two measures are not directly comparable, the ability to contrast results from different studies is obviated. A method is presented for computing the average Spearman correlation coefficient directly from confusion matrices.

A 2 × 2 factorial experiment was designed to test the impact on mail survey response rate resulting from variations in paper trim size and number of printed pages in the questionnaire. ANOVA findings suggest 8½ × 11″ paper trim size produces a better response rate than an 8½ × 14″ paper trim size. Use of a one-sheet (front and back) versus a two-sheet (front only) questionnaire did not cause a significant difference in response rate; a hypothesized interaction effect was not found to be statistically significant.