Age-level differences in work values as measured by the Values Scale using national norms were studied for 323 Hispanic or African American students in 9th grade, 12th grade, or college. These students differed according to grade level in their values as they related to work. (SLD)

The stability of a 2-factor model proposed for the Gibb Experimental Test of Testwiseness was studied with 173 undergraduates. Confirmatory factor analysis results indicate that data fit the two-factor model and a simpler one-factor model. The Gibb test could be characterized as tapping a general testwiseness proficiency. (SLD)

Effects that a creativity training program would have on divergent production subtests of the Structure-of-Intellect Learning Abilities Test were studied with 27 college students receiving training and 29 comparisons. Training appeared effective in increasing the production of units but not the production of relations or transformations. (SLD)

Confirmatory factor analysis results for the Attitudes toward Diversity Scale (ATDS) using 349 adults from different occupations indicated that the ATDS is 3-dimensional as hypothesized. Results support its high internal consistency and that it is not affected by a social desirability response style. (SLD)

Applying 2 different models of test construction to test results for a pool of more than 190,000 high school students found no systematic bias against groups with smaller or no representation in the test standardization sample. These results support the integrity of widely used sampling and item selection procedures. (SLD)

Univariate procedures proposed by M. Brown and A. Forsythe (1974) and the multivariate procedures from D. Nel and C. van der Merwe (1986) were generalized to form five new multivariate alternatives to one-way multivariate analysis of variance (MANOVA) for use when dispersion matrices are heteroscedastic. These alternatives are evaluated for Type I…

The separate calibration t-test approach of B. Wright and M. Stone (1979) and the common calibration between-fit approach of B. Wright, R. Mead, and R. Draba (1976) appeared to have similar Type I error rates and similar power to detect item bias within a Rasch framework. (SLD)

The use of Monte Carlo methods demonstrated that a formula presented by M. G. Kendall for estimating Pearson's rho from tau is somewhat more accurate than a formula presented by K. Pearson for estimating Pearson's rho from a Spearman's rho coefficient. Implications for meta-analysis of correlations are discussed. (SLD)

This article addresses the characterization and circumvention of E. H. Simpson's paradox in designs involving two ordered variables and illustrates the recommended procedure using a two-sample application involving estimation of correlation and a single-subject application involving the estimation of lag(1) autocorrelation. (Author/SLD)

A linear regression approach to determining the number of factors in factor analytic solutions is presented. It provides objectivity while producing the same results as the visual scree test. Experience with the method to date indicates that the results are consistent with visual solutions. (SLD)

A large-sample post hoc procedure for pairwise comparisons of proportions that was recently proposed by M. K. Cox and C. H. Key is discussed. Problems with the procedure are reviewed, and a corrected procedure is described. This method is extended to planned comparison of proportions. (Author/SLD)

The construct validity of scores on the Behavior Rating Scale of the revised Bayley Scales of Infant Development was studied with national standardization samples of children ranging from 1 to 42 months and a variety of factor analytic methods. The factor structure became more complex at older ages. (SLD)

A method is presented for partitioning the sums of squares of treatments and interactions to reflect specific treatment and interaction contrasts. The method is illustrated by partitioning the sums of squares in a split-plot factorial analysis of variance design. (Author)

The accuracy of the varimax and promax methods of rotation of axes in reproducing known factor matrices was examined. It was found that only when all tests are univocal, or nearly so, could one be reasonably confident that an obtained factor matrix faithfully reproduces a contrived matrix. (Author/JKS)

The analysis of variance model for estimating reliability in norm referenced tests is extended to criterion referenced tests. The essential modification is that the criterion or cut-off score is substituted for the population mean. An example and discussion are presented. (JKS)