Howard T. Everson

Confirmatory measurement model comparisons using latent means

Confirmatory factor analysis (CFA) is often used to verify measurement models derived from classical test theory: the parallel, tau-equivalent, and congeneric test models. In this application, CFA is traditionally applied to the observed covariance or correlation matrix, ignoring the observed mean structure. But CFA is easily extended to allow nonzero observed and latent means. The use of CFA with nonzero latent means in testing six measurement models derived from classical test theory is discussed. Three of these models have not been addressed previously in the context of CFA. The implications of the six models for observed mean and covariance structures are fully described. Three examples of the use of CFA in testing these models are presented. Some advantages and limitations in using CFA with nonzero latent means to verify classical measurement models are discussed.

Isolating Gender Differences in Test Anxiety: A Confirmatory Factor Analysis of the Test Anxiety Inventory

Using confirmatory factor analysis (CFA) with nonzero latent means, the factor structure of the Test Anxiety (Attitude) Inventory (TAI) across genders was investigated. Two underlying factors, worry and emotionality, have been identified in previous research efforts. A total of 501 college freshmen (219 men and 282 women) from a large urban university participated in the current study. Results support the invariance of the traditional two-factor structure for both males and females. Gender differences, however, were found in the unique factor variances, the factor covariance matrices, and in the latent means corresponding to the worry and emotionality factors. Implications for assessment and research using the TAI are discussed.

Beyond Individual Differences: Exploring School Effects on SAT Scores

This article explores the complex, hierarchical relation among school characteristics, individual differences in academic achievement, extracurricular activities, and socioeconomic background on performance on the verbal and mathematics Scholastic Aptitude Test (SAT). Using multilevel structural equation models (SEMs) with latent means, we analyzed data from a national sample of college-bound high school students. A nested series of SEMs were fit simultaneously to eight subgroups (disaggregated by both gender and ethnicity) of high school students. Our analyses suggest that (a) multilevel SEMs provide a reasonably good fit to the data, (b) family background influences SAT scores directly and indirectly, learning opportunities in and outside the school curriculum are related to SAT performance, and (c) the characteristics of the schools matter when it comes to performance on the SAT. We argue that context matters and that researchers ought to move beyond analyses of individual differences when attempting to understand performance on large-scale standardized tests.

Cognitive interference or skills deficit: An empirical test of two competing theories of test anxiety

Abstract Two competing explanations of the role of test anxiety, cognitive interference or poor academic skills, are examined empirically using structural equation models of reading comprehension and mathematics. We examined the performance of 211 participants (96 males and 115 females) on two standardized achievement tests. Prior achievement (a proxy for academic skill), test anxiety, and their interaction were represented in the structural equation models as causal influences on performance. Results suggest that the cognitive component of test anxiety (worry) and prior academic achievement contribute independently, not interactively, to performance on both measures. Findings are discussed in terms of their implications for test anxiety theory and achievement testing.

Leveraging the Power of Peer-Led Learning: Investigating Effects on STEM Performance in Urban High Schools.

The Peer Enabled Restructured Classroom (PERC) is an instructional innovation developed to address gaps in science, technology, engineering, and math (STEM) in urban high schools. The PERC model changes instruction from teacher led to peer led by bringing peer students into the classroom to lead small-group work. Our study sought to provide empirical evidence in support of the peer-led model as a means of improving STEM learning for tutored students in urban schools. We used propensity score matching to evaluate the innovations impact on students’ achievement on standardized end-of-course tests in two 9th-grade courses – Integrated Algebra and Biology. Results suggest that by the 2nd year of implementation, enrolment in PERC Biology increased the likelihood of passing. Similar effects were not observed for PERC Integrated Algebra, but when comparing cohorts, we found that the 2nd year was twice as likely to pass as the 1st year. We discuss implications for programme improvement.