SES Stratification and Math Self-Concept in Japan and the U.S.: Rethinking the Big Fish Little Pond Effect

• In Event: Opportunities for social justice and equity in the teaching of mathematics

Proposal

Globally, STEM careers can be a source of economic mobility. However, whether students from lower socioeconomic (SES) backgrounds achieve social mobility through STEM employment depends partially on motivation and requisite mathematics skills, and research has found math self-concept to be an important predictor of both (Goldman & Penner, 2016; Musu-Gillette et al., 2015). Understanding how SES relates to math self-concept in middle school (MS) students may help educators expand students’ future career opportunities.

Theoretical Framework
Math self-concept has a variety of definitions. The present study uses Marsh et al.’s (1999) and Arens et al.’s (2011) conceptualization of math self-concept having two distinct aspects, math competence beliefs (MCB), a students’ belief in their ability to do well in math, and affective math self-concept (AMSC), math enjoyment and interest.

SES may relate to math self-concept differently in different country contexts. Japan and U.S. Japan and the United States differ in classroom-level SES distribution. With delayed (tenth grade) academic tracking, Japanese MS classrooms tend to be relatively academically and socioeconomically heterogeneous, while American MS classrooms reflect greater academic and SES homogeneity and more between-classroom, within-school SES variation, partially due to early (often by the start of MS) academic tracking (Alam & Mohanty, 2023; Chudgar et al., 2013; Matsuoka, 2014; Oakes, 2005).

While the connection between classroom-level SES and math self-concept is not clear, research has consistently found a negative relationship between classroom achievement and student academic self-concept (Marsh & Parker, 1984). Dubbed the big-fish-little-pond effect (BFLPE), this theory has found that students attending classrooms with higher achieving peers have lower academic self-concepts, regardless of subject or country context, and even after accounting for individual-level and school-level SES (Seaton et al., 2010). Classroom-level SES may also play a role in the BFLPE, given that classroom-level SES appears stronger than school-level SES (van Ewijk & Sleegers, 2009).

This study explores the relationship between MS students’ math self-concept and SES, and variations in two countries with different levels of classroom SES stratification. This study asks: (1) How does student and classroom SES relate to math self-concept for middle school students in Japan and the U.S.? (2) In Japan and the U.S., what role does MS classroom SES play in the big-fish-little-pond effect?

Methods
The present study utilizes data from Trends in Mathematics and Science Study (TIMSS) 2019 (Martin et al., 2020), including eighth grade math achievement tests and student, math teacher, and school survey data from Japan and the U.S. (N = 11,555; Japan: n = 4,362; U.S.: n = 7,193). The nested nature of the TIMSS data was addressed by using multiple group multilevel structural equation modeling (SEM; Preacher et al., 2010) in Mplus. MCB and AMSC were latent variables with variance modeled at two levels. Country-mean centering was used for achievement and SES, with level one data aggregated for level two. Unstandardized coefficients were used to compare between countries and across levels.

Findings
To answer the first research question about the relationship between individual and classroom level SES and math self-concept, a multilevel SEM was developed, controlling for gender, math achievement, and teacher gender and experience (RMSEA: 0.041; CFI: 0.979; TLI: 0.966; SRMR [within]: 0.037; SRMR [between]: 0.145). Individual SES was found to have a small positive relationship with MCB in Japan (0.014) and a small negative relationship with AMSC in the U.S. (-0.025). Neither classroom nor school SES were directly related to math self-concept in either country.

To answer the second research question about the role classroom SES plays in BFLPE, after accounting for school SES, classroom SES was found to be indirectly related to MCB through classroom math achievement in both countries (Japan: 85.5%; classroom SES => MCB: -0.165; classroom SES => classroom math achievement: -0.141; SES => MCB: -0.024; US: inconsistent: classroom SES => MCB: -0.093; classroom SES => classroom math achievement: -0.104; SES => MCB: 0.011). No indirect effect was observed with AMSC.

Contributions & Discussion
The present study points to different ways that SES interacts with two aspects of math self-concept in different country contexts. In Japan, a country with less SES stratification, higher student SES has a slight positive correlation with MCB. Japanese MS students with lower SES are slightly less likely to think that they do well at math, over and above actual math achievement. In the U.S., a country with more SES stratification, a small individual-level effect of SES is seen with AMSC. American students with lower SES are slightly more likely to enjoy math, over and above actual math achievement. This suggests that Japanese teachers might find that students from low-SES backgrounds may need additional encouragement to support their MCB, while American educators may find it helpful to leverage AMSC in students from low-SES backgrounds.

At the between level, this study contributes an important finding to the literature. A wealth of research has found a negative relationship between classroom level math achievement and student math self-concept (Seaton et al., 2010). However, the findings of the present study suggest that between-classroom MCB variance might be more accurately described as an effect of classroom SES, observed indirectly through classroom math achievement, regardless of country-level differences in SES stratification. In addition, classroom-level math achievement does not appear to correlate with AMSC.

Additional research is needed to confirm these findings and examine differences across subjects and contexts. Policy makers should consider the classroom SES implications of policies linked to classroom achievement stratification, including the timing of math tracking. Considering that Japan and the U.S. do eventually track in math, albeit at different times, it may be helpful for educators and parents to know that lower MCB may not be directly related to placement in classes with higher math achievement and that AMSC may be unrelated.

Limitations to this study include that the data is cross-sectional. A longitudinal study would more fully address the bidirectional relationship between math achievement and math self-concept. Survey items were self-reported and may reflect social desirability bias, and inaccuracies may be reflected in the self-reported TIMSS 2019 SES composite.
[word count: 997]

Author

• Kim Megyesi-Brem, California State University, Fullerton