Dennis Hocevar

A New, More Powerful Approach to Multitrait-Multimethod Analyses: An Application of Second-Order Confirmatory Factor Analysis.

The advantages of applying confirmatory factor analysis (CFA) to multitrait-multimethod (MTMM) data are widely recognized. However, because CFA as traditionally applied to MTMM data incorporates single indicators of each scale (i.e., each trait/method combination), important weaknesses are the failure to: (1) correct appropriately for measurement error in scale scores; (2) separate error due to low internal consistency from uniqueness due to weak trait and/or method effects; (3) test whether items or subscales accurately reflect the intended factor structure; and (4) test for correlated uniquenesses. However, when the analy;is begins with multiple indicators of each scale (i.e., items or subscales), second-order factor analysis can be used to address each of these problems. In this approach first-order factors defined by multiple items or subscales are posited for each scale, and the method and trait factors are posited as second-order factors. This paper illustrates models that incorporate multiple indicators of each scale. The advantaot ) of their application are discussed. A three-page list o.. references and the description of the models, mentioned in the text, supplement the paper. (Author/JAZ) *********************************************************************** Reproductions supplied by EDRS are the best that can be made from the original document. *******;c*************************************************************** A New, More Powerful Approach to Multitrait-Multimethod Analyses: An Application of Second-order Confirmatory Factor Analysis

A Taxonomy and Critique of Measurements Used in the Study of Creativity

Creativity measurement itself has been a creative endeavor for both researchers and practitioners. When viewed as a group, the most salient characteristic of creativity measurements is their diversity. The initial purpose of this review is to integrate creativity measurements into a meaningful taxonomy and to illustrate the diversity of the available measurements by citing key examples of the many and varied ways in which creativity has been operationalized. It also is hoped that the numerous examples will give researchers a concise but thorough picture of the many options available when a measure of creativity is needed. The second goal of this review is to use the taxonomy as a framework for discussing the creativity construct in terms of several psychometric characteristics—namely, reliability, discriminant validity, and nomological validity. The third goal is to describe an analytic framework in which measurement issues can be better addressed.

Intelligence, Divergent Thinking, and Creativity.

Abstract Guilfords Alternate Uses, Plot Titles, and Consequences tests were given to 94 university students along with the Concept Mastery Test, a traditional measure of verbal intelligence. These measures were correlated with an inventory of creative activities and accomplishments. A composite index of ideational fluency correlated with four creativity indices: Crafts, Performing Arts, Math-Science, and Total Creativity, while the Concept Mastery Test correlated with three indices: Art, Literature, and Total Creativity. With the exception that verbal intelligence was a better predictor of creativity in literature, no statistical difference between the predictive accuracies of ideational fluency and verbal intelligence were found. The need to re-examine the widely accepted association of divergent thinking with creativity was discussed.

Self‐regulation, goal orientation, self‐efficacy, worry, and high‐stakes math achievement for mathematically gifted high school students 1 , 2

Our study used a structural equation modeling framework to investigate the effects of gender, self‐efficacy, learning goal orientation, self‐regulation, and worry on high‐stakes mathematics achievement in a sample of mathematically gifted, primarily Asian American, high school students. It is one of the few studies that investigated the joint effect of such variables in a high‐stakes environment (i.e., an Advanced Placement calculus exam). Our analyses showed that self‐efficacy is positively related to math achievement, is moderately and positively related to self‐regulation, and is highly and negatively related to worry, and that learning goal orientation (or intrinsic value) is positively related to self‐regulation and worry but is not related to self‐efficacy or high‐stakes mathematics achievement. With respect to gender, young men were less worried and had higher self‐efficacy for math than young women. Finally, self‐regulation was negatively related to worry, but surprisingly, was not related to high...

The Effects of Scoring Formulas on the Discriminant Validity of Tests of Divergent Thinking

Two multitrait-multimethod studies were conducted to investigate the effects of two scoring formulas on the discriminant validity of the dimensions of divergent thinking. In the first study, 39 fifth-grade students were given the Torrance Tests of Creative Thinking. In the second study, 60 college students were given Alternate Uses, Plot Titles, and Consequences. Findings indicate that the discriminant validity of subjectively determined originality scores is enhanced when the scores are divided by the total number of responses (yielding a percentage score). However, when the percentage formula was used to derive flexibility scores from the Torrance Tests or an originality score based on statistical rarity from the Alternate Uses test, the coefficient alpha reliabilities dropped to near zero. This finding suggests that the utility of the percentage formula is limited to subjectively determined originality scores.

Levels of aggregation in higher level confirmatory factor analysis: Application for academic self‐concept

The effects of levels of aggregation on measures of goodness of fit and higher order parameter estimates obtained from confirmatory factor analysis (CFA) were investigated. For a higher order model of academic self‐concept, 3 levels of aggregation were considered—disaggregated, partially disaggregated, and partially aggregated. In the disaggregated model, measured variables represented individual items. In the partially disaggregated model, testlets (groups of 4 items) represented measured variables. In the partially aggregated model, subscale scores represented measured variables. Three indexes of fit were employed: the Tucker‐Lewis Index (TLI), the Comparative Fit Index (CFI), and chi‐square. Solutions for the disaggregated models consistently evidenced poor fit. TLI and CFI values for partially disaggregated and partially aggregated solutions were satisfactory. Standardized parameter estimates were similar across all solutions. Implications of these findings are discussed with consideration of other re...

The Factorial Invariance of Student Evaluations of College Teaching

University instructors who taught the same course at least four times over a 4-year period were evaluated by different groups of students in each of these four classes (N = 316 instructors, 1,264 classes, 31,322 students). Factor analyses of the class average ratings clearly replicated the nine evaluation dimensions found in earlier studies. A multitrait-multimethod (MTMM) matrix was formed in which the nine dimensions were the multiple traits, and the four different offerings of the same course were the multiple methods. Across all nine dimensions the average convergent validity was r = .68. MTMM analyses with both the Campbell-Fiske MTMM criteria and confirmatory factor analysis (using LISREL) demonstrated substantial convergent and discriminant validity, but little method/halo effect. The confirmatory factor analysis also demonstrated the generalizability of the multivariate structure of the ratings across the four sets of courses. Taken together, these findings provide stronger evidence for the multidimensionality of student ratings than does any previously published research and describe a powerful analytic tool for use in student evaluation research.

Familiar Correlates of Gang Membership: An Exploratory Study of Mexican-American Youth

The differences between the families of youths who are gang members versus youths not in gangs were investigated. Data were collected from thirty mothers using a semi-structured interview. Results indicated that family related variables distinguish the families of gang members from the control families. Specifically, youths in gangs are more likely to come from families which put less emphasis on intrafamilial socialization, youth supervision, and outward expression of affection. Furthermore, mothers of youths who were gang members appeared more dissatisfied and fatalistic. Theoretical and practical implications are discussed.

Personality and Demographic Characteristics of Mothers and their Ratings of Child Difficultness

The relationships between an indicator of child difficultness and a set of maternal and familial factors were examined in three samples: ages 3-12 months; 12-24 months; and 24-36 months. After initial attempts at using a traditional psychometric approach to measuring difficultness had resulted in a methodologically questionable indicator, factor analytical techniques were employed to develop an alternative indicator of difficultness for each sample. Maternal anxiety was significantly correlated with child difficultness in all three samples. Mothers effectiveness, nervousness, extroversion, and contentment showed the strongest correlations in the age group 12-24 months, and particularly in the male subsample. Familial and demographic factors showed no associations with difficultness in any of the samples. The similarity between these findings based on a Danish sample and those reported in previous studies involving other nationalities confirms the notion of a reliable, relatively culture-independent influence of personality characteristics of the rater on measures of infant and child difficultness.

Cultural and Sexual Differences in Test Anxiety, Trait Anxiety and Arousability Egypt, Brazil, and the United States

Cultural and sexual differences in test anxiety were investigated in samples of high school students in Egypt (N= 277), Brazil (N = 234), and the United States (N = 141). Measures of trait anxiety and trait arousability were also included. Compared to the United States greater test anxiety was found in Egypt on both the worry dimension and the emotionality dimension. Greater test anxiety was also found in Brazil, but only on the emotionality dimension. Compared to both the United States and Brazil, greater trait anxiety and arousability were reported by high school students in Egypt. Finally, in all three cultures females reported greater worry, emotionality, trait anxiety, and arousability than males. A possible explanation for the cultural differences in test anxiety is the important consequences of high school testing in Egypt and Brazil.