Gary J. Coles

A NEW METHODOLOGICAL REVISION OF EKMAN'S CONTENT MODEL OF MULTIDIMENSIONAL SIMILARITY ANALYSIS

A new methodological revision to Ekmans content model of multi- dimensional similarity analysis is proposed. Whereas Ekman factor analyzes matrices of mean similarity estimates, the Stone-Coles revision involves the factor analysis of correlational similarities obtained by intercorrelating all column vectors of an Ekman mean similarity matrix. The method proposed solves certain mathematical objections to the Ekman model. 14 investigations of multidimensional similarity conducted by Ekman and his associates were reanalyzed from the Stone-Coles correlational similarity approach. A compari- son of the results indicate that this new methodological construct yields interpretable factorial descriptions of the judged similarity of stimulus-objects which are more meaningful and more parsimonious than the results yielded by the Elanan methodology.

Multidimensional Scaling Used to Evaluate Students Residing in a Rehabilitation Unit

The Stone-Coles revision of Ekmans “content” model of multidimensional similarity analysis was applied to judgments of person similarity made by four mental health professionals. The person-stimuli evaluated were college student clients (disturbed Ss) and volunteers (normal Ss). Two students in the latter category also served as judges. All judges appeared to share a common frame of reference, as only one factor emerged from a factor analysis of interjudge correlations. The factor analysis of a matrix of correlational similarities calculated from a matrix of mean similarities yielded four factors. These factors were first interpreted using clinical judgment and subsequently these interpretations were checked by correlations with MMPI variables. The first three factors were labeled: I. Action-oriented vs Overcontrolled, II. Sex, and III. Severity of Disturbance. A small fourth factor was uninterpretable.

Psychology graduate students' multidimensional perceptions of their psychology faculty ☆

Abstract A recently proposed multidimensional similarity analysis methodology was used to analyze the dimensionality of a faculty group, the faculty which comprises an academic department of psychology, as perceived by advanced graduate students matriculating in that department. The two (or three) extracted judgmental-perceptual dimensions accounted for 70 % (or 79 %) of the judgmental variance; these dimensions appeared to be psychologically meaningful, i.e., they were readily interpretable. An inverse analysis of the judges was also accomplished and led to interpretable results. The success of this exploratory application of a multidimensional scaling procedure suggested further uses of such methodologies in investigations of other forms of social judgment-evaluation.

Methodological Note: An Application of Three-Way Analysis of Variance in the Estimation of Inter-Judge Reliability

An application of three-way analysis of variance to the estimation of inter-judge reliability in a psychological scaling investigation is described.

Instructors' Multidimensional Perceptions of Their Students.

Two instructors of a single psychology course made similarity estimations of students who were enrolled in this course. This judgmental data was multidimensionally scaled using the Stone-Coles paradigm. Three judgmental dimensions were extracted: academic achievement, sex membership, and liberal-conservative attitudes and behaviors. A suggestion was made that multidimensional scaling may represent an approach to grading if grading is to be based solely on observations-evaluations of teachers.

Multidimensional Scale Analysis of Elementary Teachers' Impressions of Selected Social Reinforcers

A recently proposed multidimensional similarity analysis methodology was used to analyze the dimensionality of a set of selected classroom social reinforcers. Judges were elementary school teachers who frequently administered such reinforcers. The three extracted evaluation dimensions, accounting for 81% of the judgmental variance, were all readily interpretable. The success of this exploratory application of a multidimensional scaling procedure in the realm of educational concerns suggested further uses for such investigative methodologies with regard to other forms of educational judgment-evaluation.

MULTIDIMENSIONAL SCALING OF ELEMENTARY TEACHERS' IMPRESSIONS REGARDING A SET OF STUDENT BEHAVIOR PROBLEMS

A recently proposed multidimensional similarity analysis methodology was used to analyze the dimensionality of a set of selected behavior problems sometimes exhibited by elementary students in classroom situations. Judges were elementary school teachers who frequently encountered such behavior problems. The three extracted evaluation dimensions, accounting for 82% of the judgmental variance, were all readily interpretable. The success of this exploratory application of a multidimensional scaling procedure to educational concerns suggested further uses for such investigative methodologies with regard to other forms of educational judgment-evaluation.

AN EXPLORATORY DIRECT-ESTIMATION APPROACH TO LOCATION OF OBSERVERS IN PERCEPT SPACE'

A least squares technique for estimating the loadings that additional/new variables might receive on factors derived in an analysis of a different set of variables was adopted for use with a recently proposed methodological revision of Ekmans “content” model of multidimensional similarity analysis. An experiment was then conducted in which psychologist-judges directly estimated the degree of perceived similarity between their particular viewpoints with respect to “personality” and 15 well-known personality theories. The procedure was used to locate the psychologists in the four-dimensional space previously constructed to account for the rated similarities among the same theories.

AN ESTIMATION OF r BASED ON SUBJECTIVE SIMILARITY EVALUATIONS

Using linear regression information presented by Ekman and his associates, a model which allows one to predict correlation coefficients (rij) from similarity estimations (sis) was devised and used to build a conversion table. An argument favoring possible use of such predicted rijs was made. The association of sij to rij was discussed.

Conditioned word meaning and the interstimulus interval

Word meaning responses were established for groups having one of four interstimulus intervals (ISIs): 2 sec backward, 1 sec backward, 1 sec forward, and 2 sec forward. There was no significant difference in the degree of conditioning shown by the four groups, and the acquisition of word meaning was due almost completely to the performance of Ss who were aware of the relations between CSs and USs. The appropriateness of labeling the acquisition of word meaning classical conditioning is questioned.