Hierarchical Models of Attitude

White South Africans' Reactions to Black Advancement: A Two-Sample Confirmatory Investigation of the Structure of Attitude Using an Analogy to the Multitrait-Multimethod Design

The tripartite (affective, conative, cognitive) theory of attitude has been investigated in a number of empirical studies, with findings mostly favoring the theory. Little attention has been paid, however, to other important characteristics of attitude. One of these is multiplexity which refers to the number of separate domains into which an attitude object can be partitioned. In this study, a data design was used which made it possible to investigate trait and domain structure simultaneously. A questionnaire measuring affective, conative, and cognitive responses to three aspects of black advancement was administered to two groups of white South Africans: English speakers employed by a large private-sector company and Afrikaans speakers employed by the government. Confirmatory techniques were employed to investigate the structure of the data. Single-group analysis procedures adapted from Widaman (1985) were initially performed to establish a model satisfactory for both samples. Multi-group procedures were then performed on the two samples to investigate group differences in data structure. The structure was very similar in the two samples with the exception that domain variances were smaller in the English-speaking sample.

An Information Theoretic Approach for Identifying Shared Information and Asymmetric Relationships Among Variables

Behavioral researchers are often faced with the need to identify complex multivariate relationships. Statistical information theory provides a framework for quantifying in a single value the proportion of total information in one set of measures (Y) explained by another set of measures (X). It also quantifies the amount of redundant information and allows for asymmetry of explained information between variables. The general information theoretic approach is presented and illustrated using measures of affect, cognition and behavior. A statistically significant and asymmetric information theoretic relationship is found among the variables: affect (like/dislike) provides a higher percentage of information about behavior (shopping frequency) than behavior does about affect. In addition, affect provides a higher percentage of information about behavior than does perceived location convenience.

A Factorial Evaluation of Effects of Model Specification and Error on Parameter Estimation in a Structural Equation Model

In an experiment manipulating artificial data in a factorial design, model misspecification and varying levels of error in measurement and in model structure are shown to have significant effects on LISREL parameter estimates in a modified peer influence model (Duncan, Haller & Portes, 1971). The extent of error present in both the measurement and the model have significant direct effects on estimated parameter values, and the two sorts of error interact as well. In specification, incorrect omission of parameters seems to have more impact on parameter error than does incorrect inclusion of parameters.