Statistical significance tests for binormal ROC curves

Statistical significance tests are derived and evaluated for measuring apparent differences between an obtained and an expected binormal ROC curve, between two independent binormal ROC curves, and among groups of independent binormal ROC curves. A binormal ROC curve is described by two parameters which represent the spread of the means and the ratio of the standard deviations of the two underlying Gaussian decision variable distributions. To test the significance of apparent differences between or among ROC curves, approximate χ² statistics for each of the three tests were constructed from maximum likelihood estimates of the two parameters defining the binormal ROC curve. The performance of each test statistic was evaluated by simulating five-category rating scale data with equal numbers of noise and signal-plus-noise trials (set at 50, 250, and 500) for each of three typical ROC curves. For the significance test involving only one ROC curve, rating scale data were generated from the chance diagonal of the ROC space also. Although test performance was found to be somewhat dependent on the number of trials and on the location of the ROC curve in the ROC space, comparisons of the obtained and expected fractions of (falsely) significant results at various α levels showed the proposed statistical significance tests to be reliable under practical experimental conditions.     

A set-theoretic random utility model of choice behavior

A random utility model of choice was developed by combining the basic ideas of the well-known theories of Thurstone and Restle. The new model has exactly the same number of free parameters as Tversky's Elimination-by-Aspects model. Furthermore, both models were found to fit, with equal accuracy, the data reported by Rumelhart and Greeno, and Tversky. It was concluded that although the two theories are not identical, they may be difficult to discriminate empirically.     

Religion,psychology, and mental health: The problems of partnership

While a current and increasingly popular trend is toward partnerships between religion and psychology in the area of mental health, and especially toward the use of psychological ideas and techniques by pastoral counselors, modern psychology and traditional religion are divided in significant ways over several normative issues. Too often these issues are minimized or overlooked. But unless these issues—which carry significant sociocultural implications as well as implications for our definitions of individual well-being—are acknowledged and subjected to critical considerations, partnerships between religion and psychology must be far more cautiously constructed and must be subjected to far more systematic thinking about the differences which separate the two.This article is based on a paper delivered at the Seventh Annual Clinical-Community Conference on Psychology and Religion at the University of Maryland in November, 1978.     

Fluid and crystallized intelligence: effects of diffuse brain damage on the WAIS

In the 1940's two forms of intelligence, fluid and crystallized, were postulated. Active mental processing (WAIS performance subtests) were fluid while well learned abilities (WAIS verbal subtests) were crystallized. Brain damage was considered to affect fluid abilities more than crystallized. WAIS scores comparing normal and diffusely organically damaged subjects indicated that, as hypothesized, the verbal subtests were more affected than the performance subtests.     

THE VALIDITY and FAIRNESS OF EMPLOYMENT and EDUCATIONAL TESTS FOR HISPANIC AMERICANS: A REVIEW and ANALYSIS

This study examined the questions of whether employment and educational tests demonstrate either differential validity or test unfairness to Hispanic Americans relative to the majority group. Relevant data from 19 published and unpublished studies formed the basis of the review and analysis. Results for employment tests, based on a very large amount of data, indicated that differential validity occurs no more frequently than would be expected on the basis of chance plus the operation of various statistical artifacts. The frequencies of slope and intercept differences between Hispanic and majority regressions were also within the chance range. A similar pattern of results obtained for the more limited amount of data available on educational tests. Taken together, the results provide strong evidence that tests are neither differentially valid for, nor unfair to, Hispanics. These findings are consistent with the results of previous studies of test validity and fairness for blacks.     

Memory aids people use: Two interview studies

In two interview studies subjects were asked what memory aids they used and how often. In particular, the reported frequencies of use of external aids such as shopping lists and memos were compared with those of internal aids that have been investigated by psychologists, such as the method of loci. Subjects in the first study were 15 male and 15 female university students, and in the second, a group of 30 adult women, most of whom were housewives. In both studies the aids reported to be used most were external ones. However, it is pointed out that the most commonly used methods of remembering must be internal, although these are not usually considered to be aids. Some difficulties that arise in the assessment of memory in interview and questionnaire studies are discussed.     

Gender typing of game choices and qualities of boys' play behavior

Relationships between gender choices and both movement patterns and social behavior were studied in first- and second-grade boys. Three-child, structured play groups were each composed of a boy whose mother saw him as high masculine in play preference on the Games Inventory, one seen as average, and one low masculine. Behaviors rated from videotapes included gender presentation variables, (e.g., leg separation), indicators of dominance and personal comfort (e.g.,range of movements), and indicators of social skill and peer response (e.g., interaction initiations). The low masculine boys were found to be the most feminine in their gender presentation, least dominant and aggressive, and the least socially successful of the boys. The greatest difference was between the low and the high masculine boys. The average masculine boys' scores were generally intermediate,but more similar to the low masculine boys on some variables and more similar to the high masculine boys on others. These conclusions apply to a group interaction play task, but not to an initial noninteractive play task. A secondary study in which girls played with low and average masculine boys is also reported. Here it was found that low masculine boys were generally intermediate between average boys and girls on gender presentation and dominance variables, but lowest of the groups on social interaction variables. The authors appreciate the comments of Peter Bentler and Spencer Thompson on a draft of this paper, the assistance of Laura McCain in running subjects and scoring tapes, and the expertise of Shu-Yeng Wong for computer programming.     

Scaling clinical judgments of symptom pathology by means of the psychophysiological method of magnitude estimation

Summing scores across heterogeneous symptom items without consideration of their differing psychopathological significance has been criticized as producing an inadequate picture of an individual's clinical status. The purpose of this study was to derive symptom item weights representing clinically judged seriousness of each symptom through the application of Stevens's psychophysical method of magnitude estimation. A nationwide sample of 129 clinicians rated the pathological significance of 221 symptom items in a design such that every rater rated 121 items, 21 of which were rated by all raters and 100 of which were rated only by the A or B subgroup to which each rater was randomly assigned. Each item was rated as to the seriousness of the pathology it would represent if manifested by either a boy child, girl child, boy adolescent, or girl adolescent, with one-fourth of the raters assigned to each condition. The results of 211 two-way analyses of variance revealed that age and age and sex in interaction, but not sex alone, significantly influenced the clinical ratings. The resulting magnitude estimation ratings of symptom pathology ranged from 1.0 to 9.9. They were demonstrated to have satisfactory reliability and convergent validity and to have the psychophysical characteristics of a prothetic continuum.A preliminary version of this paper was presented at the 85th Annual Convention of the American Psychological Association, San Francisco, August 1977. The research was supported by HEW grant MH24886. The authors wish to express appreciation to Walter W. Wengel for his work in developing the median quotient ratings used in the study, to Terrance Auger for making available unpublished raw data, and to Merton S. Krause for helpful suggestions regarding the interpretation of results.     

Conceptual development in learning-disabled and normal boys

In two related studies, a pictorial sorting task served as the means of measuring conceptual development in LD and normal boys. The studies found no differences in the bases on which groups were formed (perceptible attributes, function, or nominal class membership) or in the manner in which groups were structured (thematically, in complexive groups or superordinate groups). Furthermore, parallel developmental trends were found in both LD and normal subjects groups. These findings are not consistent with early studies of concept formation in brain-damaged subjects, most of whom were retarded. Subjects in the present studies were selected on the basis of clinical diagnosis of LD and average IQ. The findings of the present studies suggest that generalizations about conceptual abilities of LD children of normal intelligence should not be based on early studies of the effects of brain damage.