Gritty Measures for Gritty Times: Evaluation of the Grit Scale

Background: Grit has been defined as passion and perseverance for long term goals. It is presented as a trait which measures success in certain situations such as retention at West Point, placement in the final round at the National Spelling Bee, and Ivy League GPA. The theory of grit is rooted in personality theory and is closely related to the construct of conscientiousness (Duckworth et al., 2007). Grit theory argues that grit is independent from IQ and essential to high achievement. However, the reliability of the current grit scale has been brought into question in several articles (e.g., Crede, Tynan, & Harms, 2017). These studies have questioned the structure of the grit scale as a hierarchal two- factor model and the degree to which it overlaps with the personality trait of conscientiousness. Many studies reporting on grit’s overall relationship to academic success have had mixed results. Purpose: In this multiple manuscript dissertation, the first study aims to (1) examine the latent factor structures of the original (Grit-O) 12-item and short (Grit-S) 8-item grit scales, (2) determine whether a new short grit scale based on a different modification of Grit-O and the conceptual definitions of the grit factors creates a better measure than the existing one, (3) analyze the correlations between grit questions and conscientiousness (from BFI), and (4) investigate the relationship between grit and feelings of success in calculus. The second study then examines (1) whether a more generally worded single-factor 4-item scale for grit is reliable, (2) whether this new general grit scale predicts success in a calculus course, and (3) whether the general grit scale is generalizable across gender and ethnicity. Methods: The purpose of the first article is to test the validity of the Grit-O and Grit-S scales developed by Duckworth and colleagues (2007, 2009) using confirmatory factor analysis. It also examines the correlations of the constructs of various grit scales with the personality measure of conscientiousness. This study also presents and validates a new “general grit” scale based on Grit-O and Grit-S, which are conceptually targeted toward the constructs of consistency of interest and perseverance of effort in an academic setting. The second article focuses on the development and validation of general grit items using exploratory and confirmatory factor analysis. Using structural equation modeling, the study tests whether grit predicts student success in calculus. These scales are examined across different genders and races by using Mplus to test for group invariance in the factor structure. Results: The results of the first paper show that alternate models for Grit-O and Grit-S fit better than the current ones in the literature (Grit-O: x2 = 304.882, d f = 53, CFI = .828, Grit-S: x2 = 65.749, d f = 19, CFI = .941, alternative models: x2 = 115.556, d f = 42, CFI = .950, x2 = 34.540, d f = 19, CFI = .979, and x2 = 2.534, d f = 8, CFI = .987). New questions representing the facets of the Grit-O scale showed good model fit on their own in a two-factor correlated model and had lower correlations with conscientiousness than Grit-O. Different models with grit items and students’ perceived feelings of academic success in their calculus class had inconsistent results. Paper 2 created a single factor grit scale, general grit, in order to overcome the inconsistencies in reporting of grit scores. The general grit scale was first tested with EFA and confirmed with CFA. Next, general grit and the new questions from paper 1 were tested with final exam score as an outcome variable. General grit showed good prediction of final exam grade and mediated effects of consistency of interest and perseverance of effort in the model. Last, the structures were tested across gender and ethnicity, and invariance for the structures was confirmed. Conclusion: Both papers have shown that better items can be created to measure grit in an academic setting. Verification of all scales showed good fit and ability to predict academic success. Current calls to abandon grit as a construct may be premature now that measurement issues have been overcome.Psychological, Health, and Learning Sciences, Department o