Measures

The key variables, namely, teacher mindset climate, stereotype threat, and mathematics anxiety, were all measured–2–13 weeks after the semester began. The preferred target date of the survey completion was 2–7 weeks after the semester began. Students’ grade point averages (GPAs) were collected via institutional records at the end of each semester (see Yeager, 2019, for details).

The current research focused on adolescents’ perceptions of the teacher mindset climate based on previous research that suggests classroom climates that influence students’ experience are the students’ subjective experiences of their classrooms, rather than some objective reality (Patall et al., 2018). Students’ perceptions of the teacher mindset climate were measured using three² items ( “My math teacher thinks that some kids are smart and others are not,” “My math teacher seems to believe that only a few students will understand the hardest problems,” and “My math teacher seems to like you better if you are good at math”). These items reflect the climate-level entity theory of intelligence and encompass teachers’ attitudes and behaviors that value inherent ability (Murphy et al., 2018). All items had standardized factor loadings above .60. Students answered these questions based on a five-point scale ranging from not at all true (1) to extremely true (5). Values were coded so that higher scores indicated stronger perceptions of a fixed mindset climate. The overall reliability of the three items was α_total = .75 (α_{Black girls} = .74, α_{Black boys} = .75, α_Latina = .75, α_Latino= .75, α_{White girls} = .76, α_{White boys} = .75).

To ensure the measurement invariance of the latent variable (i.e., teacher mindset climate), configural (invariant form), metric (invariant factor loadings), and scalar (invariant intercepts) invariance tests were performed using model constraints and the Satorra and Bentler (2001) scaled chi-square difference test. The results of measurement invariance showed that the three items of the teacher mindset climate showed adequate measurement invariance across (1) Black girls and Latinas; (2) Black boys and Latinos; (3) White girls; and (4) White boys (configural vs. metric: Satorra & Bentler Δχ²(6) = 11.63, p = .07; metric vs. scalar: Satorra & Bentler Δχ² (6) = 10.80, p = .10). The final scalar invariance measurement model’s fit indices were χ²(12) = 22.37, p = .03, comparative fit index (CFI) = .98, root mean square error of approximation (RMSEA) = .03 [90% confidence interval, CI: .01, .04].

Students self-rated their worries and concerns about mathematics stereotypes, using two items: “In math class, how much do you worry that people’s judgments of you will be affected by your gender?” In math class, how much do you worry that people’s judgments of you will be affected by your race or ethnicity?” The items were similar to those used in prior research (e.g., “In testing situations, I worry that people will draw conclusions about me based on what they think about my ethnic group,” Chung et al., 2010; see also Marx et al., 2005). These items were asked based on a five-point scale ranging from 1 (not at all) to 5 (an extreme amount). The current research used the average scores of the two items to assess the overall experience of stereotype threat in mathematics classrooms. It is important to consider both gender and racial/ethnic stereotypes when examining Black and Latinx students’ experience in classrooms, because being a “boy” or “girl” has a significantly different meaning for Black and Latinx students as compared to White students (Cole, 2009). In support of the decision to use the aggregated scores, the two items were highly correlated (r_total = .67; r_{Black boys} = .71, r_{Black girls} = .66, r_Latino= .70, r_Latina = .65, r_{White boys} = .72, r_{White girls} = .56), and the reliability across the two items was adequate (α_total = .79; α_{Black boys} = .79, α_{Black girls} = .81, α_Latino= .83, α_Latina = .77, α_{White boys} = .84, α_{White girls} = .71). Supplementary analyses utilized the scores for each item separately.

Students’ mathematics anxiety was assessed using the item, “In general, how much does the subject of mathematics in high school make you feel nervous, worried, or full of anxiety?” based on a five-point scale from 1 (not at all) to 5 (an extreme amount). Anxiety has often been measured with a single item with evidence of adequate validity and reliability in the extant literature (e.g., Goetz et al., 2016; Seo & Patall, 2020). Research has also suggested that asking participants to respond to a longer measure of emotion could induce unnecessary reflections and result in the measurement of emotions toward answering the questionnaire (Goetz et al., 2010). Accordingly, the NSLM used a single item to assess students’ mathematics anxiety.

Students’ mathematics achievement was measured using the institutional records of students’ mathematics course GPA. GPA in the entire 9th grade was included as an outcome for students who participated in the study in the fall semester. Some schools completed the stereotype threat survey in the fall semester of the 9th grade, and other schools completed the survey in the spring semester. For students who participated in the study during their spring semester, only the spring semester mathematics course GPA was included as an outcome to ensure temporal precedence of the focal variables. This is in line with the pre-registration of a previous study that utilized the same data (Yeager et al., 2019).

To control for potential confounding effects, previous mathematics achievement, mathematics course level, and family socioeconomic status were included as covariates. Previous mathematics achievement was measured based on the institutional records of students’ mathematics grades in the 8th grade. Mathematics course level was also assessed based on the institutional records of mathematics courses students took in the 9th grade. Course level was coded so that higher values represented more challenging courses (1 = basic mathematics, 2 = pre-algebra, 3 = algebra 1, 4 = geometry, 5 = algebra 2, 6 = other advanced math, 7 = pre-calculous/calculus). Finally, family socioeconomic status was measured based on the mother’s highest education level (1 = did not finish high school, 2 = finished high school, 3 = took some college courses, 4 = Associate’s degree, 5 = Bachelor’s degree, 6 = Master’s degree, 7 = Doctorate degree) and free or reduced lunch status (0 = not receiving free or reduced lunch, 1 = receiving free or reduced lunch).