Robert M. Thorndike

The role of disgust and fear in blood and injection—related fainting symptoms: A structural equation model

Structural equation models were used to examine the relationship of blood, injection, and injury (BII) fears, disgust sensitivity, and trait anxiety to having experienced fainting-related symptoms in 722 university students. The latent variable representing BII fears was highly positively related to faint symptoms whereas that relating to disgust experiences was inversely related to fainting. A second disgust sensitivity variable was not directly related to fainting. Trait anxiety added nothing directly to the model, although it was related to BII latent variable. Additional analyses included a gender-related cross-validation which showed that models developed on one gender group did not fit well when applied to the other gender group. Blood drawings were the most common faint eliciting situations and were found to be reportedly more likely to elicit faint symptoms than were injections.

A Study of the Stability of Canonical Correlations and Canonical Components

OVER the last 10 years considerable progress has been made in the theory and availability of canonical correlation analysis. Cooley and Lohnes (1962) drew attention to the method and provided one of the early programs for performing canonical analyses. Horst (1961) generalized Hotelling’s (1936) solution for the two set case to encompass m sets of variables. This development opened the way for Cooley’s solution for &dquo;multiple partial canonical correlation&dquo; (Cooley, 1967) in which the effect of one (or more) set of variables is partialed out of the relationship between two other sets. In addition, Meredith (1964) has shown how canonical analysis may be performed on the true score components of data which are less than perfectly reliable. The number of empirical examples of canonical correlation in the literature has not kept pace with the theoretical advances. Cooley (1965) suggested three possible reasons for this: (1) the difficulty in computation; (2) availability of other, more familiar, methods for studying the relationships between two sets of vari-

Would the Real Factors of the Stanford-Binet Fourth Edition Please Come Forward?:

Publication of the Stanford-Binet Intelligence Scale, Fourth Edition has resulted in publication of at least six factor analyses of the instrument. After the principles for factor analysis are discussed, four of these analyses are reviewed and criticized in detail, and two others are noted. After the shortcomings of previous analyses are described, factor analyses of the standardization data are reported that are based on principal axis factorings of the age-specific and age-grouped correlation matrices with Oblimin rotation. At ages 2 through 7 the instrument produced two correlated factors of verbal and nonverbal ability. Above age 7 there are three correlated factors: verbal ability, abstract/spatial ability, and memory. The correlations among the primary factors are taken as evidence of a second-order factor of general intelligence. The factors remain stable across age levels. Analysis of the median correlations across all ages yielded evidence of the four factors hypothesized by the test constructors.

Stability of the WAIS‐R and WISC‐III factor structure using cross‐validation of covariance structures

The stability of the Verbal Comprehension, Perceptual Organization, and Freedom from Distractibility factors of the WAIS-R and WISC-III was tested using cross-validation of covariance structure models, a methodology that employs unrestricted and restricted factor analyses. Stability was indicated if the goodness of fit of restricted models generated from unrestricted factor patterns did not degrade when applied to other age groups. If goodness of fit degraded inconsistently, the factor structure of one or both instruments was unreliable. If goodness of fit changed in some systematic fashion, an argument for true intellectual changes could be made. Results indicated consistency across age groups. Therefore, it was concluded that the three-factor structure is robust and reliable across age groups and instruments.

Stability of the WPPSI-R and WISC-III Factor Structure Using Cross-Validation of Covariance Structure Models:

The reliability of the Verbal Comprehension and Perceptual Organization factors of the WPPSI-R and WISC-III was examined by use of cross-validation of covariance structure models, a new statistical methodology that employs both unrestricted and restricted factor analyses. Reliability of the factor structure would be indicated if the goodness of fit of restricted factor analytic models generated from unrestricted factor patterns did not degrade when tested on other age groups. If the goodness of fit changed in an inconsistent fashion, that would indicate that the factor structure of one or both of the instruments is unreliable. Finally, if the goodness of fit changed in a systematic fashion, getting worse with greater age differences, then an argument for true intellectual changes could be made. Results revealed general consistency in the goodness of fit across age groups. Therefore, it was concluded that a two-factor model that consists of the Verbal Comprehension and Perceptual Organization factors is reliable across age groups and instruments.

Canonical Correlation Analysis

Publisher Summary Canonical correlation analysis is viewed as the most general of the traditional least-squares methods for the analysis of data structures. This chapter describes set and canonical correlation analysis as a widely applicable taxonomy for data analysis, shows how they are related, and explains how all other least-squares procedures can be derived as conceptual special cases. One factor that contributes to poor methodology in canonical analysis is that there are no fully adequate computer programs generally available to researchers. Each major statistics package can perform a canonical analysis, but none of them include all of the desirable features. Furthermore, the chapter discusses the traditional canonical analysis model, shows its relations to principal components analysis, and reviews significance tests and other interpretive aids such as redundancy and the rotation of canonical components. Finally, the chapter concludes with a discussion of methodological issues in the use of canonical analysis, including stepwise procedures and the need for cross-validation.

Canonical Analysis And Predictor Selection.

This paper addresses theoretical and practical issues in the use of canonical analysis for prediction. A major objection to canonical prediction has been that the variates that are predicted are of low utility and do not reflect practical concerns. This problem may be overcome by a stepwise approach to the selection of predictor variables such that those predictors which predict the desired criterion are retained. In addition, canonical analysis may provide a solution to the problem of multidimensional criteria.

Stability of stated reinforcers as measured by the reinforcement survey schedule

The Reinforcement Survey Schedule (RSS) is designed to assess therelative reinforcing values of a variety of stimuli and stimulus situations. While the scale apparently is a valid assessment device, its stability over time had not been established. The present study is a test-retest reliability study of the RSS over three time intervals of 1,3, and 5 weeks, respectively. Subjects were 313 undergraduate psychology students at Western Washington State College. Median correlations for the 140 items, across subjects were .734, .665, and .709 for 1, 3, and 5 week intervals, respectively. Median correlations for subjects, across items for the same intervals were .829, .775, and .801. These reliability coefficients are of sufficient magnitude to render the scale useful as a clinical and research instrument. Implications and further refinements are discussed.

The Influence of Television on Measured Cognitive Abilities A Study with Native Alaskan Children

In 1979 and again in 1982 data were collected in rural Alaska to assess certain effects of commercial entertainment television, which began in 1977 in selected rural sites. The findings reported here include two major culture groups: the northern Inupiat Eskimo (four villages) and the southwestern Tlingit and Haida (two villages). In 1979, half of the villages had received television for two years; by 1982 all villages had reception capabilities, allowing for a longitudinal as well as cross-sectional study of the differential effects of television on children. Age, sex, culture area, and amount of television watched were major predictor variables while scores on four cognitive tests (CEFT, Kohs Blocks, Raven Matrices, and PPVT) were the criterion variables. Data analysis was a three-stage procedure in which the independent variables were treated as sets. There was no evidence that television by itself had a major effect on cognitive abilities. Numerous interactions show that television in Alaska did not, during the period of the study, have a uniform effect on the children in the study. The social learning concept of reciprocal determinism can be used to explain the complex ways in which television interacts with person variables and other environmental variables to influence test scores.

STUDYING CANONICAL ANALYSIS: COMMENTS ON BARCIKOWSKI AND STEVENS

In their Monte Carlo study of canonical analysis, Barcikowski and Stevens evaluated the relative stability of canionical weights and loadings. This paper identifies some weaknesses in their study, suggests directions for future research in this area, and discusses interpretation of canonical analysis both in development and in cross-validation.