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.     

Incidental learning,attention, and curiosity in squirrel monkeys calorie deprived as infants

The long-term effects of infantile malnutrition on information processing were studied in squirrel monkeys. From 2 through 8 weeks of age, four low-calorie infants were fed reduced amounts of a high protein diet, limiting their mean body weight gain to 23% of that of four well-fed controls, with all of this gain occurring during the first 2 weeks on the low-calorie diet. Thereafter, both groups received the standard diet ad libitum. At 68 weeks of age, incidental learning was tested by introducing shape cues that were redundant to the solution of a previously learned, two-choice, color discrimination. Removal of the color cues showed that the low-calorie monkeys had learned significantly less about the shape cues than had the controls. The basis for this incidental learning failure was then examined with an embedded-figures dicrimination, to measure attentional ability, and a novel stimulus substitution task, to measure curiosity. The groups did not differ in detecting embedded figures, but the low-calorie monkeys were significantly less likely than controls to select or handle novel stimuli that had been substituted for previously learned discriminative cues. It was concluded that undernutrition in infancy may induce a long-term strategy of only learning information which leads to an immediate reward, while suppressing the acquisition of information due to intrinsic curiosity.     

Auditory processing of brain-damaged adults under competitive listening conditions

Right CVAs, mild left CVAs, and non-brain-damaged adults received the Token Test under three conditions: quiet, white noise, and speech babble. Left CVAs performed most poorly of all groups in all conditions experiencing difficulty with subtests III, IV, and V. Right CVAs were inferior to controls on each of these three subtests in the speech babble condition only. Thus, right CVA difficulty with both memory and syntactic tasks was demonstrated in speech babble. These results are explained by the hypothesis that both hemispheres are necessary to process auditory information under complex listening conditions. Total brain capacity of right CVA patients may be overloaded if either memory load or linguistic complexity is increased in the presence of certain types of listening competition.     

The availability explanation of excessive plausibility assessments

The assessment of hypotheses in hypothesis generation involves a comparison between those hypotheses that have been generated (specified) and those that are not generated (unspecified). Experiment 1 was an investigation of an “availability explanation” for subjects' overconfidence in the probability of specified hypotheses. The conjecture is that subjects have difficulty retrieving unspecified hypotheses from memory. Therefore, the underpopulated set of unspecified hypotheses is assessed as less probable then it actually is and the specified set is assessed as more probable. Two manipulations to increase the availability of unspecified hypotheses were investigated. One involved explicitly requesting subjects to populate the unspecified set; the other was a computer presentation of candidate unspecified hypotheses. Results of experiment 1 were that assessment overconfidence for both experimental groups was reduced. The results support the conjecture that the availability heuristic is at least partially responsible for subjects' overconfidence. The main result of experiment 2 was that the overconfidence bias persisted with different assessment methods for both subject-generated and experimenter-supplied hypotheses.     

Self-consciousness and reactance

Two studies examined the effects of dispositional self-consciousness on reactance. In Experiment 1, men who were high in private self-consciousness displayed greater reactance responses to a coercive communication attempt (as reflected by attitude reversal) than did men lower in private self-consciousness. In contrast, the effect of public self-consciousness in this context was to inhibit the expression of reactance. In Experiment 2, women high in private self-consciousness exhibited greater reactance responses to a self-imposed threat to their freedom of choice (as reflected by equivocation over two choice alternatives) than did women lower in private self-consciousness. The effect of public self-consciousness in this context was negligible. These findings replicate and extend a previous self-awareness and reactance finding; they provide additional evidence that manipulated self-awareness and dispositional self-consciousness converge on the same psychological entity; and they provide additional evidence that there are important differences between the dimensions of private and public self-consciousness.     

Stuttering and fluency: Exclusive events or points on a continuum?

In recent years, several studies have been conducted in which the fluency of stutterers was compared with that of normal speakers. These investigations have been of two basic types, some involving objective measures of physiologic, aerodynamic, or acoustic features of the subjects' fluent speech signal, and others involving experiments performed in which listeners attempted to distinguish the fluency produced by stutterers from that generated by normal speakers. The body of literature pertaining to each of these two types of investigations is reviewed and interpreted. The interpretations offered are then developed in an effort to shed further light on both fluency and stuttering. Implications for further research, as well as for the evaluation, treatment, and post-therapy assessment of stutterers are drawn.     

A perceptual comparison: All-voiced versus typical reading passage read by children

The present investigation tested the modified vocalization hypothesis (Hutchinson and Brown, Journal of Fluency Disorders, 1978, 3, 149) in young stuttering subjects. In this study, stuttering children ages 7–13 yr were instructed to read two experimentally constructed passages. One passage contained both voiced and voiceless phonemes; the second passage was constructed using only voiced phonemes. Results indicated there was neither a quantitative nor a qualitative difference in the stuttering blocks during the reading of either experimental passage. These results would appear not to be supportive of the modified vocalization hypothesis in children; however, as has been indicative of similar studies, 8 of 22 subjects did demonstrate more dysfluencies during the reading of the passage that contained both voiced and voiceless phonemes.