Factor analysis and scale revision.

This article reviews methodological issues that arise in the application of exploratory factor analysis (EFA) to scale revision and refinement. The authors begin by discussing how the appropriate use of EFA in scale revision is influenced by both the hierarchical nature of psychological constructs and the motivations underlying the revision. Then they specifically address (a) important issues that arise prior to data collection (e.g., selecting an appropriate sample), (b) technical aspects of factor analysis (e.g., determining the number of factors to retain), and (c) procedures used to evaluate the outcome of the scale revision (e.g., determining whether the new measure functions equivalently for different populations).

Predicting medical students' academic performances by their cognitive abilities and personality characteristics.

PURPOSE: To utilize multiple regression analyses with both linear and quadratic models to explore and confirm the relationships among students' cognitive abilities, personality traits, and medical school performances at the University of Los Angeles, California, UCLA School of Medicine. METHOD: Ninety-seven "not-disadvantaged" students' pre-medical grade-point averages (GPAs), Medical College Admission Test (MCAT) scores, and personality traits as measured by the Comrey Personality Scales (administered at their application to medical school in 1985) were used to predict their medical school performances according to several criteria. "Disadvantaged" students were excluded from the study because their poor performances on all criteria would confound the relationships of personality, cognitive ability, and performance. RESULTS: The MCAT score was a strong predictor of medical school performances, particularly those criteria measured by medical school GPAs and the National Board of Medical Examiners examination scores, but its predictive power dropped sharply when clinical performance and personal suitability were part of the performance evaluation. Specific personality traits not only strengthened the predictive power of cognitive and personality variables jointly, they became the primary predictors of clinical performance and personal suitability. A single personality-profile index failed to show any power of prediction. As expected, quadratic relationships were found between some personality traits and some medical school performance measures. CONCLUSION: According to these data it is not realistic to use one or two personality traits to predict personal suitability on all medical performance measures. The multiple regression analysis results indicate that different personality characteristics are incorporated in different types of medical performances, which is perhaps why the overall personality-profile score failed to predict any of the studied criteria. Because of the study's relatively small sample size, however, cross-validation studies are necessary to confirm these findings.

A profile of personality for a Russian sample: as indicated by the Comrey Personality Scales.

The purpose of our study was to present a personality profile, based on data from the Comrey Personality Scales (CPS), for a diverse, Russian sample. The CPS was administered to 457 subjects drawn from convenience samples throughout St. Petersburg, Russia. In Part 1 of the study, the structure of the Russian sample's personality was identified. Data from the CPS were factor analyzed. Findings showed strong support for the generalizability of Comrey's (1970) eight-factor personality taxonomy to this sample. The CPS's factors derived from the Russians were compared and found to be similar in structure to samples derived from American, Australian, Brazilian, Israeli, Italian, and New Zealand samples. In Part 2, mean differences on the CPS's scales between the Russian sample and the American sample were assessed. Results showed that although the structure of the Russians' personality was similar to that of the Americans' personality, significant differences in the elevation of scale scores did exist. For example, as a whole, the Russians tended to be more defensive, lethargic, egocentric, and unstable. Contrary to popular expectations, however, the Russians were not more conforming.

Eyseneck Personality Inventory Item Factor Structure.

The Eysenck Personality Inventory (EPI) was administered to a sample of 583 Australians. Responses to the 57 items were intercorrelated and factor analyzed by three different methods: (a) extract two principal factors and rotate them by varimax; (b) extract as many factors as possible by the minimum residual method, determine the correct number of factors using the Tandem Criteria method and then rotate that number of factors by the Tandem Criterion I method; (c) using ones in the diagonals, extract all factors with positive eigen values and rotate these factors by varimax. All three methods obtained major Neuroticism and Extraversion factors but the third method gave the poorest confirmation of the expected factor structure for these items. Of the remaining two methods, the second is preferred for testing the expected underlying factor structure for these items since it permits items to remain on the same factor only if they are correlated with each other and it reveals what other constructs may be required in addition to those hypothesized to account for all the common factor variance.

Factor Analysis of the Comrey Personality Scales in an Australian Sample.

The Comrey Personality Scales (CPS) were administered to a sample of 669 Australians. The 40 subscales that define the eight CPS factors were intercorrelated along with two validation scale scores, age and sex. This matrix of correlations was factor analyzed to determine if the same structure of personality traits previously found in other groups would also hold for Australians. All eight CPS personality factors appeared as expected, showing excellent agreement with past findings in other groups. This agreement in factor structure occurred despite many significant differences in personality factor score means between this sample and the normative group sample. These findings lend additional support to the premise that the eight personality factors measured by the CPS have considerable stability across cultural boundaries.

Personality characteristics of Hare Krishnas.

Personality characteristics of 132 males and 94 females of the Hare Krishna movement were assessed using the Comrey Personality Scales (CPS). Subjects were sampled from eight U.S. sites. Their age averaged about 30 years, and their time in the movement averaged 8.6 years. The most prominent and surprising finding is the hallmark characteristic of the Hare Krishna personality, a strong compulsivity trait common to both genders, which is slightly stronger in males. Reduced trust in society was exhibited by both gender averages, with slightly less trust among females. Both these averages, however, were within the normal range for individuals. With the exception of compulsivity, average male CPS scores were within the normal psychological range but differed significantly from the normative male group indicating idiosyncratic traits. Average female CPS scores did not differ significantly from the normative female group except for compulsivity and trust. Study of CPS score distributions about group means supported these findings.

Millon MCMI Scales Factor Analyzed and Correlated with MMPI and CPS Scales.

The Millon MCMI was administered in translation to 527 applicants for drivers' licenses in Israel who were required to undergo psychological screening. Three hundred of these subjects also took the Comrey Personality Scales (CPS) and the other 227 also took the MMPI. A factor analysis of the 20 MCMI scales was carried out and these scales were also correlated with the CPS and MMPI scales. A meaningful five-factor solution was obtained and canonical correlations were computed between the MCMI scales and those from the other two tests. Evidence from this and previous studies suggests that the MCMI scales overlap each other more than they should for purposes of optimum differential diagnosis.

A Method for Removing Outliers to Improve Factor Analytic Results.

Bad data due to faked responses, errors, and other difficulties can distort correlations among variables leading to poor factor analytic results based on matrices of such correlations. A method of detecting potentially bad data cases, or outliers, is presented which is based on the average squared deviation of a given subject's cross product of standard scores from the average over all correlations in the matrix. Results of applying both this program and the BMD 10M outlier program to the same data examples are given. About 40 to 60 percent of the cases identified as outliers by the two programs were the same cases. Many cases identified as outliers proved not to be "bad data", however, so these programs should be used to identify cases that need scrutiny rather than as the sole basis for eliminating data.

Personality Factor Structure in Psychiatric Outpatients and Normals.

The Comrey Personality Scales (CPS) were administered to 394 psychiatric outpatients. The 40 subscales that define the eight CPS factors were intercorrelated and factor analyzed to determine if the same structure of personality traits fits beth normal and psychologically disturbed individuals. The factor structure in this study was very close to that found previously in six other widely different kinds of populations. These results support a previously stated conclusion that tests of normal personality traits are appropriate for use in assessment with psychologically disturbed individuals.

Personality structure of Mexican Americans using the Comrey Personality Scales.

The Comrey Personality Scales (CPS) were used to examine the personality structure of 176 Mexican American college students. A majority of the eight factors maintained substantial loadings and corresponded closely with Comrey's normative sample as well as with Brazilian and New Zealand samples. Two factors, Conformity and Orderliness, showed the weakest fit across all four groups. Overall, the CPS shows adequate measurement properties for "normal" personality assessment among Mexican Americans and is preferable to instruments such as the Minnesota Multiphasic Personality Inventory (MMPI). However, the external validity of the CPS with clinical populations must first be determined prior to its use to assess psychopathology among Mexican Americans.

Distortions In A Commonly Used Factor Analytic Procedure.

A currently popular procedure in empirical factor analytic studies is to use unities in the main diagonal as communality estimates, extract all factors with eigenvalues of 1.0 or higher, and rotate these factors by varimax. This procedure and others are applied to several previously published correlation matrices and some artificial matrices. This procedure results in the retention of too many factors, unrealistic elevation of the amount of common factor variance analyzed, and distortions in the conclusions drawn from the factor analytic investigation. Ways of avoiding these difficulties, are discussed.

An Empirical Comparison Of Two Minimum Residual Factor Extraction Methods.

Comrey and Harman & Jones have proposed different methods of factor analyzing a correlation matrix using on!y the off-diagonal elements. The purpose of such procedures is to avoid using the diagonal communality elements of the matrix theft are generally unknown and must be estimated. This study was undertaken to provide an empirical comparison of the two methods. The Comrey method was considerably faster but the Harman & Jones method produced higher derived communalities. Both methods gave very similar empirical rotated solutions. Implications of the results are discussed. This study was carried out in response to a question (personal communication) posed by Professor Jura Nunnally about the nature of the Comrey and Harman-Jones solutions. The authors wish to express their thanks to Harry Harman for providing a copy of his minres computer program.