Integrated writing assessments are tasks that require students to construct an essay using evidence from source texts. Topic models are statistical models used to analyze text. This study analyzes two integrated writing tasks from students in grades 6 (N=1,250) and 8 (N=1,250) using topic models. Independent evaluations of the essays were also…

Current descriptions of results from hierarchical linear models (HLM) and hierarchical generalized linear models (HGLM), usually based only on interpretations of individual model parameters, are incomplete in the presence of statistically significant and practically important "slopes as outcomes" terms in the models. For complete description of…

Uses examples of hierarchical linear modeling (HLM) at local and national levels to illustrate proper applications of HLM and dummy variable regression. Raises cautions about the circumstances under which hierarchical data do not need HLM. (SLD)

Using a degenerate multivariate normal model for the distribution of organismic variables, the form of least-squares regression analysis required to estimate a linear functional relationship between variables is derived. It is suggested that the two conventional regression lines may be considered to describe functional, not merely statistical,…

The problem of chance model modifications under varying levels of sample size, model size, and severity of misspecification in confirmatory factor analysis models was examined through Monte Carlo simulations. Findings suggest that practitioners should exercise caution when interpreting modified models unless sample size is quite large. (SLD)

A new method of measuring item bias based on the latent class model proposed by the author is suggested. A test for item bias is also suggested that is based on standard asymptotic results. (Author/DWH)

Point-estimate and confidence-interval procedures are proposed for the (1) combined Mann-Whitney U test; (2) combined two-sample normal-scores test; (3) combined matched-pair Wilcoxon signed-ranks test; and (4) combined matched-pair normal-scores test. The four techniques are illustrated with hypothetical examples and are considered either…

The partial correlation coefficient is derived analytically under exemplary factor patterns. In these patterns, variables are described as an additive composition of a set of orthogonal factors, including general, common, and specific factors. Viewed in this framework, it is evident that the partial correlation may yield spurious results.…

A heuristic example is presented in which parsimony as typically operationalized in indices of fit may not be desirable. In a simplex model of longitudinal data, the failure to include correlated uniquenesses relating to the same indicators on different occasions will typically lead to systematically inflated estimates of stability. (SLD)

Problems of estimation and interpretation are discussed that arise when a statistical package, which does not incorporate a dedicated weighted least-squares (WLS) routine, performs WLS regression by misapplication of a case-weighting strategy. A strategy is offered for adjusting WLS regression estimates after a case weighting strategy has been…

An error in an essential equation within the article by Williams and Zimmerman is corrected, and the algebraic inequalities are translated into questions a researcher can ask about simple or residualized difference scores. Williams and Zimmerman acknowledge the error and note that main conclusions are not affected. (SLD)

A mathematical link between test reliability and test validity is derived, taking into account the correlation between error scores on a test and error scores on a criterion measure. When this correlation is positive, the "paradoxical" nonmonotonic relation between test reliability and test validity occurs universally. (Author/BW)

Adequacy coefficients can be derived by calculating the cosines of the angles between factors' actual and theoretically expected locations within factor space. The use of adequacy coefficients during instrument development is discussed. The development of an instrument to measure thinking interests provides a heuristic framework for the…

Six estimators of treatment effect when assignment to treatment groups is based on the covariate are compared in terms of empirical standard errors and percent relative bias. Results show that simple analysis of covariance estimator is not always appropriate. (Author/GK)

The nonparametric hypothesis-testing model of M. L. Puri and P. K. Sen (1969, 1985) circumvents concerns regarding the inapplicability of nonparametric statistics outside simple cases and the lack of pertinent computer programs. The breadth and flexibility of the model are illustrated with several examples. (TJH)