Knowledge that Transforms

Several issues relating to goodness of fit in structural equations are examined. The convergence and differentiation criteria, as applied by Bagozzi, are shown not to stand up under mathematical or statistical analysis. The authors argue that the choice of interpretative statistic must be based on the research objective. They demonstrate that when this is done the Fornell-Larcker testing system is internally consistent and that it conforms to the rules of correspondence for relating data to abstract variables.

The statistical tests used in the analysis of structural equation models with unobservable variables and measurement error are examined. A drawback of the commonly applied chi square test, in addition to the known problems related to sample size and power, is that it may indicate an increasing correspondence between the hypothesized model and the observed data as both the measurement properties and the relationship between constructs decline. Further, and contrary to common assertion, the risk of making a Type II error can be substantial even when the sample size is large. Moreover, the present testing methods are unable to assess a model's explanatory power. To overcome these problems, the authors develop and apply a testing system based on measures of shared variance within the structural model, measurement model, and overall model.

Fishbein's attitude theory posits that beliefs are the only mediators of attitude formation and change. The validity of this proposition for consumers’ beliefs about product attributes and brand attitudes was exmained in the context of an advertising effects study. To manipulate product attribute beliefs and to create settings in which other mediation processes might occur, the authors exposed subjects to simple advertisements that contained either a verbal claim or visual information. Level of repetition also was varied. As expected, product attribute beliefs mediated attitude formation. However, another variable, termed attitude toward the advertisement, also mediated brand attitudes and purchase intentions. The authors discuss alternative explanations for the results and offer suggestions for future research.

The author examines the reliability and validity of measures of organizational characteristics used in previous marketing studies in the areas of strategic planning and distribution channels. Key informants in 506 wholesale-distribution companies provided reports on (1) characteristics of the firm's product portfolio and (2) characteristics of the firm's power-dependence relations with its major suppliers and customers. In contrast to previous investigations, which sampled only a single informant per unit of analysis, data were collected from multiple informants in each firm. Results showed that informant reports often achieved convergent and discriminant validity when variance due to methods factors was explicitly modeled. However, partitioning of variance according to trait, method, and random error components showed that informant reports often exhibited less than 50% variance attributable to the trait factor under investigation. Implications of the findings are discussed for those marketing studies which focus on organizations or organizational subunits as the unit of analysis.

The author comments on Fornell and Larcker's article in the February 1981 issue of JMR. He indicates some limitations of the analyses, pinpoints where they can be misleading, and introduces some new notions.

Although the concept of moderator variables has been used extensively in marketing-related studies, much confusion persists as to how they are defined and identified. To alleviate this confusion, the authors present a typology of moderator variables with a framework for identifying their presence and type. Simulated data are used to illustrate and validate the proposed framework.