Exploratory Structural Equation Modeling as a Tool for Validation of Scales Used To Measure Research Self-Efficacy and Resilience in Research Training Programs for Underrepresented Students

Abstract

Using multiple forms of structural equation modeling, this study demonstrates a progression of analyses from the most basic confirmatory factor analysis through more complex exploratory structural equation modeling to interrogate both the construct and discriminant validity of the RSES and CD-RISC-10 when used as an outcome assessment for research training programs that target underrepresented students. Not only does this study inform the understanding of the factor structures for both measures, but it also demonstrates the utility of using more complex modeling for localized instrument validation. The psychometric assessment of the CD-RISC-10 found adequate model fit using CFA of the original factor structure, but improved model fit utilizing a two-factor structure defined through ESEM. However, the modifications made to the original RSES before being applied to this population resulted in more model fit and factor loadings inconsistent with its original form. Potential adjustments to the RSES and suggestions for interpretation are discussed.