Julius A. Najab
jnajab at gmu dot edu
Research Interests
As a social scientist my interests center on program evaluation. My main focus of program evaluation creates many options to conduct research in a broad range of content areas. In order to study a variety of content areas I also invest much of my time in psychometrics, research methodology and statistics. I am also interested in an interdisciplinary approach toward psychology research. I am also interested in integrating evolutionary psychology and genetics into the my interdisciplinary studies.
Current Projects
Dissertation Topic Selection
Correlated Measurement Errors Meta-Analysis Structural Equation Models fit indices benefit from additional model pathways. A common practice is to correlate measurements errors as a method to improve model fit. Along with Liz Conjar, Richard Hermida, Jose Cortina, Seth Kaplan, Ron Landis, and Bryan Edwards we are investigating the practices and justifications of social scientists regarding the inclusion of correlated measurement error pathways.
Hierarchical Linear Modeling Education and psychological researchers frequently use hierarchical linear models (HLM). However, the analysis complex and not very well understood. Confusion in terminology between HLM and multi-level models (MLM) results in differential practices and possible errant inferences. Carrie Wiley and Mei-kuang Chen from the University of Arizona and I are collaborating on this project. We hope to clarify concepts and practices to the general research community.
Regression Diagnostics Linear Regression Diagnostics are incredible important to obtaining precise parameters estimates and hypothesis testing results. However, psychological researchers rarely if ever conduct regression diagnostics. Patrick McKnight, Jeff Steuwig and I are collaborating of a project to highlight the practice of researchers. Beyond highlighting faulty practice we hope to provide useful guidelines for identifying regression assumption violations.
Interrater Reliability Caroline Wiley, Simone Erchov, Patrick McKnight are developing a manuscript for a general psychological audience on the various interrater reliability and agreement estimations that are available.
Bayesian Analysis Bayesian analysis is an alternative method for interpreting probabilities and testing hypotheses than the Frequentist status quo. Together with David Cades, David Kidd, and Patrick McKnight we are composing a manuscript to describe the benefits and limitations under Bayesian analyses for psychological research.
Advanced Statistical Website Soon an instructional website for novices will be up. This website will demonstrate how to accomplish some basic Hierarchical Linear Model, Structural Equation Model, and Item Response Theory analyses. This project is funded in part the American Psychology Association’s Commission on Ethnic Minority Recruitment, Retention and Training Task Force. Stephanie Wong is the fellow graduate student collaborating. Jon Mohr and Patrick McKnight are serving as faculty supervisors.