Terry L. Dickinson

A comparison of the behaviorally anchored rating and mixed standard scale formats.

Compared the behaviorally anchored rating (BAR) and mixed standard scale (MSS; F. Blanz and E. E. Ghiselli, see record 1974-04034-001) formats in terms of the multitrait-multimethod validity they engendered in rating scales. Rating scales were developed with the retranslation of expectations (RE) pr

Preferences for Recreational Environments: Theoretical Considerations and a Comparison of Models

Abstract Environments were conceptualized as providing a context within which valued psychological attributes could be experienced. Environmental preferences under this general theory were defined to be a function of evaluative and cognitive assessments of an environments psychological attributes. Since previous research was unclear as to the exact nature and validity of the evaluative assessments, four models of environmental preferences were investigated. The models differed in their inclusion or exclusion of evaluative assessments. Data were collected to implement and compare the four models using survey methodology. The model that distinguished between rewarding and costly psychological attributes was found to be most useful in explaining environmental preferences. The remaining three models were less comprehensive in their explanation of preferences. Finally, the reward‐cost distinction was discussed as being important in the understanding of environmental quality.

Mixed Standard Scale Response Inconsistencies as Reliability Indices

The mixed standard scale is a rating format which allows researchers to count illogical (i.e., internally inconsistent) response patterns. These counts have been interpreted as an index of rater reliability, based on the notion that illogical response patterns reflect carelessness in responding. This study investigated the meaning of these counts, using 943 accountants as raters. It was found that the counts of illogical response patterns were not related to reliability as measured by Cronbachs alpha, and alternative interpretations for such indices are suggested.

Trait Patterns, Background Characteristics, Managerial Styles, and Job Practices of Industrial Middle Managers

Abstract Cluster analysis was used to classify the trait patterns of middle managers in the United States into six personality types. These personality types ranged from low to high leadership, and they differed in terms of background characteristics, time spent on and importance of job functions, and managerial styles. The six types appear to be useful for understanding the personalities of middle managers and for making organizational decisions about them.

Book Reviews : David Freedman, Robert Pisani, and Roger Purves. Statistics. New York: W. W. Norton, 1978. Pp. xv + 506 + A-83.

book. Overall, the book is a scholarly effort, and it is most appropriate for the student who wishes to learn the utility of statistics for research and rigorous thinking. The book is divided into eight parts containing 29 chapters. Part I concerns the design of experiments, and in two chapters it distinguishes controlled experiments from observational studies. Several examples are used including the Salk vaccine field trial. Part II contains five chapters mostly on descriptive statistics such as histograms, the arithmetic mean, and standard deviation as well as their relation to the normal curve. The authors do not discuss the concept of variance, deferring it to later courses. In one chapter, the standard deviation is used to define measurement error for a series of repeated measurements. The calibration work done at the National Bureau of Stan-

13.95

Linear models are available for describing dominance and composition judgments. These models include parameters for the scale values of the stimuli, response order bias, and interaction between stimuli, all of which may be evaluated with ANOVA procedures.