Poster #77 - Grit-S, Long-Term Grit and Related Constructs: Measurement Invariance and Mean Differences across Gender and Ethnicity

• In Event: 1-115 - Poster Session 03
  In Poster Session: PS 03 Section - Education, Schooling

Integrative Statement

Muenks et al. (2017a, 2017b) have examined overlap in grit’s factor structure with conceptually similar self-regulatory variables, or its relations to important motivational beliefs. These previous studies had relatively small samples that did not allow the latent constructs to be examined all together in the same model. Therefore, the objective of this study is to examine how grit overlaps, at the latent variable level, with conceptually similar constructs that include self-regulation (SR), effort-regulation (ER), task value (TV), and self-efficacy (SE), in a large sample of adolescents. Further, the research interest is whether there are differences in latent means and correlations among the constructs across gender and ethnic groups (African American and non-African American). Findings regarding a new long-term measure of grit will be presented.
Duckworth et al. (2007) defined grit as a personality characteristic consisting of two components, perseverance of effort (PE) and consistency of interest (CI). They and others have found that it predicts some outcomes, although findings are mixed. In addition, Muenks et al. (2017a, 2017b) examined the relation of grit to other constructs as a bivariate relation rather than examining the correlations among all latent variables together. To date there have been few studies examining possible gender differences and ethnic group differences in grit, and extant findings are mixed (Credé et al., 2016). We used a multiple group six factor simple structure, using either Grit-S or long-term grit, model with different estimation options.
1,147 high school students completed the Grit-S, a long-term grit scale and the SR, ER, TV, and SE subscales of the Motivated Strategies for Learning Questionnaire (MSLQ). We conducted two multiple group analyses with full information maximum likelihood estimation for each construct to examine differential item functioning (DIF) across gender (518 females, 623 males) and ethnicity (398 African Americans, 749 non-African Americans, 25% not European American). Then, we fit a six factor simple structure model with weighted least square estimation to all of the response data to see the relations among the latent variables, then conducted two multiple group analyses to examine if means and correlations are different between sub-groups of interest.
There was no DIF for female and male groups, but one item from the ER subscale and one item from the CI subscale demonstrated DIF for African American and Non-African American groups. Model fit was acceptable and all six factors significantly positively related to each other. Males had higher scores than females on CI and lower scores on SR, ER, and TV. Males and females did not differ on PE and SE. We found no differences between the two ethnic groups. Findings regarding models using long-term grit will also be presented.
With the growing interest in grit, it is imperative to understand whether individuals’ respond to the Grit items and the MSLQ in similar ways, and that those items function in the same manner across groups to establish measurement invariance. The gender differences we found have potential implications for interventions that have been developed to enhance grit or other relevant constructs.

Authors

• Jessica R Gladstone, University of Maryland, College Park
  Presenting Author

• Monica Morell, University of Maryland, College Park
  Non-Presenting Author

• Ji Seung Yang, University of Maryland, College Park
  Non-Presenting Author

• Annette Ponnock, University of Maryland, College Park
  Non-Presenting Author

• Lara Turci, University of Maryland, College Park
  Non-Presenting Author

• Allan Wigfield, University of Maryland, College Park
  Non-Presenting Author