Construct Validity of a Math Intelligence Mind-Set Scale for Third-Grade Students

• In Event: Investigating Academic Mind-Sets in New Contexts and With Novel Conceptual Approaches

Abstract

Intelligence mindset, specifically the belief that intelligence can be changed (growth mindset) or not changed (fixed mindset), is theorized to determine academic behaviors and achievement (Dweck & Leggett, 1988). However, inconsistencies in the research regarding the link between intelligence mindset and achievement indicates potential limitations in the current conceptualization and measurement of intelligence mindset. In the current study, we developed a new scale to assess growth, fixed, and a novel “mixed” mindset, which assessed combined aspects of both fixed and growth mindsets. Specifically, we evaluated three components of construct validity (i.e., substantive, structural, and external; Flake et al., 2017) of the newly-developed Math Intelligence Mindset Scale (MIMS) for elementary-school children.
To evaluate substantive validity, we developed MIMS items that aligned with the definition of intelligence mindset and removed language associated with other achievement-motivation factors. We then conducted cognitive interviews of the MIMS items with 17 children in January 2019 (10 boys; mean age=8.92 years). Two changes were made to MIMS items based on children’s responses: 1) “in general” was removed from two items because of variations in understanding what the phrase meant and 2) “you” was changed to “person” in two fixed-oriented mindset items that asked about the belief a person had with grouping people into categories. Table 1 presents the final, updated items.
To evaluate structural validity of the MIMS, we compared three models with theoretically-driven factor structures using data from 698 third-grade students who completed the updated MIMS, other achievement-motivation measures, and a math test in their school classrooms in spring 2019. We conducted confirmatory factor analyses and difference tests to compare among a unidimensional model, a two-dimensional model with separate growth and fixed factors and cross-loaded mixed mindset items on both factors, and a three-dimensional model with separate growth, fixed, and mixed mindset factors. Math intelligence mindset was best represented by the two-factor model with cross-loaded items. To improve model fit, we also dropped non-significant mixed mindset item factor loadings from the fixed mindset factor, leaving a two-factor model with a combined growth/mixed-mindset factor and separate fixed-mindset factor ([χ2(89)=124.46, p=.008, RMSEA=0.026 (90% CI=0.014−0.036], CFI=0.98, and TLI=0.97).
To evaluate external validity, we examined correlations between intelligence mindset, other achievement-related factors, and math achievement. Both intelligence mindset factors were strongly related with effort beliefs and perceptions of mistakes in math and moderately related with challenge-seeking behaviors (Table 2). Math fixed mindset had the strongest correlation with math achievement (r=-.34) relative to growth/mixed mindset.
As more schools encourage educators to adopt teaching practices based on intelligence mindset research, it is critical that the conceptualization and measurement of intelligence mindset is rigorously tested. This study advances our knowledge of intelligence mindset by providing validity evidence of a new scale to measure math intelligence mindset in children. Our analyses suggest the MIMS best represents a two-dimensional model of intelligence mindset, mixed mindset items are interpreted similarly to growth mindset items, and there is strong overlap in responses to intelligence mindset, effort beliefs, and perceptions of math mistakes items made by third-grade children.

Authors

• Connie Barroso Garcia, Texas A&M University - College Station

• Colleen M. Ganley, Florida State University

• Robert Schoen, Florida State University

• Chris Schatschneider, Florida State University