A proposal for handling missing data

A method for dealing with the problem of missing observations in multivariate data is developed and evaluated. The method uses a transformation of the principal components of the data to estimate missing entries. The properties of this method and four alternative methods are investigated by means of a Monte Carlo study of 42 computer-generated data matrices. The methods are compared with respect to their ability to predict correlation matrices as well as missing entries. The results indicate that whenever there exists modest intercorrelations among the variables (i.e., average off diagonal correlation above .2) the proposed method is at least as good as the best alternative (a regression method) while being considerably faster and simpler computationally. Models for determining the best alternative based upon easily calculated characteristics of the matrix are given. The generality of these models is demonstrated using the previously published results of Timm.     

Predecisional information acquisition: Effects of task variables on suboptimal search strategies

Effective overall performance in judgment tasks generally involves both acquisition of information and integration of the information acquired. When information is costly the decision maker must balance acquisition costs against improved decisional accuracy, a complex balancing problem in which, laboratory evidence suggests, humans often do poorly. The three experiments reported here extend this earlier evidence to different task structures, subject pools, incentive systems, information volumes, decision aids, and kind of data sources. Though each of these factors was found to affect performance, the general finding was of persistent underpurchase (buying less information overall than is optimal) and mispurchase (buying poor sources when better sources are available at the same cost), with significant inflation of overall costs. It is proposed that unaided human judgment is unequal to the complexity of the cost/benefit trade-offs involved in acquiring costly information, and that formal decision analysis should be preferred whenever the stakes justify.     

Factor analysis of the Wechsler adult intelligence scale-revised (WAIS-R): An examination of the standardization sample

Orthogonal and oblique factor analyses were performed using the Wechsler Adult Intelligence Scale-Revised (WAIS-R) standardization sample (Wechsler, 1981). Using objective criteria, it was decided to retain two factors. Evidence for a strong general intelligence (g) factor was found from both the orthogonal and oblique rotations. Whereas the verbal IQ provided an acceptable estimate of the Verbal Comprehension factor, a purer measure was found in the form of a Verbal Comprehension Deviation Quotient containing only the Information, Vocabulary, Comprehension, and Similarities subtests. The performance IQ was found to be an inadequate measure of the Perceptual Organization factor. A Perceptual Organization Deviation Quotient, containing only the Block Design and Object Assembly subtests, was recommended as an effective alternative. Analysis of the variance components for each subtest at every age level revealed that a substantial proportion of subtests at a wide range of age levels evidence either high or intermediate levels of specific variance. These results are discussed in terms of interpreting individual subtest scores and patterns. Each of the study's major findings are compared with prior work done with the Wechsler Adult Intelligence Scale (WAIS) (Wechsler, 1955).     

Me,us and them: Political identification and the third-person effect in the 1993 Australian federal election

Three days prior to the 1993 Australian federal election 54 Australian university students who identified with one of the two major political parties were surveyed regarding their perceptions of media campaign impact on self and others. Results provided evidence of a third-person effect (Davison, 1983) wherein respondents judged others us more influenced by the election campaign than themselves. Consistent with predictions derived from social identity theory and self-categorization theory (e.g. Turner, Hogg, Oakes, Reicher and Wetherell, 1987), political ingroup members were also judged as less injuenced by campaign content than political outgroup members. Respondents who identified strongly with their preferred party judged self and ingroup members as less influenced by campaign content than did other respondents, and showed more evidence of positive intergroup differentiation. At the same time, however, these respondents exaggerated self–ingroup differences, challenging the theoretical assumption that intergroup diferentiation is associated with ingroup assimilation. Judgements of media impact on self and other also depended on the direction of the campaign message. Respondents believed ‘voters in general’ were persuaded in line with the intent of campaign content, while outgroup members were seen to be persuaded by material favouring their own side but to be uninfluenced by counter-attitudinal content. Election propaganda, irrespective of direction, was seen to amplify existing party preferences in self and ingroup members. Hence the relative invulnerability of self to media impact was pronounced when respondents judged the impact of pro-outgroup messages. Results suggest that perceptions of self–other differences in media vulnerability are influenced by the subjectively salient social relationship between self and other, and are governed by motivational needs, such as self-esteem, social-identity, and differentiation from others (cf. Brewer, 1991; Hogs and Abrams, 1993).     

Effects of food priming on instrumental acquisition and performance

Four experiments investigated the effects of a priming food reinforcement, given 0 or 75 sec pretrial, on runway performance of rats. The studies differed in the use of between-versus-within-subject designs, and by using food or water as the goal reinforcer. In Experiment 1, using food as the goal reward, subjects primed with food 0 sec pretrial conditioned slower than subjects primed 75 sec pretrial. In Experiment 2, using water as the goal reward, subjects primed with food 0 sec pretrial conditioned faster. These differences were evident on both prefed trials and on nonprefed test trials. Experiments 3 and 4 showed an opposite pattern of results when within-subject comparisons of 0- and 75-sec pretrial intervals were used: Food priming immediately pretrial facilitated food-rewarded running but inhibited water-rewarded running. The results suggest prefeeding has differential effects on acquisition and performance of instrumental behavior, and also depending upon the similarity of the priming and goal reinforcers.     

Demonstrating the effect of context on order effects for an army air defense task using the patriot simulator

The results reported herein support the hypotheses that (1) situation-specific, contextual features of a task can cause people to use explanation-based reasoning (Pennington and Hastie, 1993); (2) such reasoning can cause experienced personnel, both individually and in two-person teams, to reinterpret the meaning of the same information when it is presented in two different ordered sequences; and (3) the result will be primacy or recency (or no) effects depending on whether the most recent conflicting information can be explained away or not, respectively. These results extend the belief-adjustment model proposed by Hogarth and Einhorn (1992), which does not address information reinterpretations, and always predicts recency effects for an evaluation task with a short series of conflicting information. More generally, the results demonstrate the importance of situation-specific, contextual features in understanding judgment processes. © 1997 John Wiley & Sons, Ltd.     

The role of esteem and coping in response to a threat communication

Two experiments examined the relations among self-esteem, perceived competency to cope, and actual coping behaviors following a threat communication. Leventhal's “parallel response model” (in Advances in experimental social psychology, L. Berkowitz (Ed.), New York: Academic Press, 1970, Vol. 5) predicts that low-esteem subjects will show deficits in both competency and coping behaviors. Experiment 1 manipulated threat level of a tetanus communication. Low-esteem subjects showed coping deficits on measures of free associations, free recall, fatalism, and coping. Threat groups differed only on fear and danger measures. Experiment 2 manipulated the fear level of an antismoking film and used false feedback to alter perceived competency. Positive feedback increased perceived competency to quit smoking among low-esteem subjects only. Without feedback, low-esteem subjects reduced smoking less than high-esteem subjects; positive feedback reversed the pattern. The discussion argued that, consistent with Leventhal's model, the low-esteem coping deficit has two independent causes: (1) excessive concern with fear, and (2) inadequate perceived competency.     

Parent and Teacher Ratings in the Assessment and Prediction of Antisocial and Delinquent Behaviors

ABSTRACT Subjects consisted of a sample of two cohorts of approximately 100 boys each whose behaviors were rated by their parents and teachers. Criterion variables included antisocial behavior, based on parent, child, teacher, and interviewer reports, and delinquency, based on parent and child reports in addition to cumulative arrest data taken from juvenile court records. The data suggest that mothers are focused on the daily, irritating behaviors of their sons. Teachers, on the other hand, appear to focus on a relatively small number of items (e.g., child physically attacks others, associates with deviant peers), and thereby provide ratings that are better predictors of delinquency and arrest. When the mothers' ratings were constrained to include only items that were salient for teachers, their predictive validity coefficients approached the magnitude of the teacher coefficients.