Curriculum Matters: Math Growth Among Low-Performing Elementary Students in Grades 3–5

• In Event: Data-Driven Approaches to Supporting Striving Learners in the Post-Pandemic Era

Abstract

Introduction
Early mathematics skills, particularly those in middle and elementary school years, are linked to long-term academic and career success (Duncan et al., 2007; National Mathematics Advisory Panel, 2008). However, despite the increasing importance of foundational math in a tech-driven world, many U.S. students continue to struggle. According to the 2024 NAEP, only 40% of fourth graders reached proficiency in math, while nearly 25% fell below basic levels (National Center for Education Statistics, 2025). High-quality math curricula are one potential solution to improve student outcomes (Bhatt et al., 2013; Koedel et al., 2017). This study evaluates the impact of the i-Ready Classroom Mathematics (iRCL) curriculum on math achievement among low-performing students in grades 3–5.
Methods
Given the observational nature of the data, researchers used propensity score matching (PSM) to reduce selection bias and create comparable treatment and control groups based on 22 baseline covariates (see Figure 1; Austin, 2011; Caliendo & Kopeinig, 2008). Matching was only accepted if each group had at least 350 students and all standardized mean differences were below |0.1|, aligning with ESSA’s (n.d.) moderate evidence standards. Matching was conducted using R’s MatchIt package (Ho et al., 2011; R version 4.4.0). To account for the nested structure of the data, longitudinal Hierarchical Linear Modeling (HLM) was used to assess the curriculum’s impact on students who were at least two grade levels behind in math—referred to as striving learners. Models were estimated using the lme4 package in R (Bates et al., 2025), following Raudenbush and Bryk’s (2001) framework.
Data and Measures
The final matched sample included 10,000 students; 5000 in the iRCL group and 5,000 in a business-as-usual curriculum group. All students completed three administrations of the i-Ready Diagnostic, a widely used, independent, adaptive assessment. The Diagnostic is frequently used in educational research and is sufficiently independent from the iRCL curriculum to serve as a valid outcome measure.
Results
A one-way random effects ANOVA showed that 61.2% of the variance in math achievement was between students (ICC = 0.612), justifying the use of multilevel modeling. The second model estimated the average initial math score (B = 408.80, SE = 0.26, p < .001) and growth rate points per time unit (B = 12.58 SE = 0.09, p < .001). In the third model, curriculum type and an interaction term between time and iRCL use were added. Initial scores were similar between groups (iRCL: 408.75; control: 408.85; p > .05). However, the interaction between time and iRCL exposure was significant (b = 1.23, SE = 0.18, p < .001), indicating greater learning growth for iRCL students. This effect remained robust after controlling for all covariates in the final model (b = 1.23, SE = 0.18, p < .001). For more information, see Table 1 and Figure 2.
Conclusion
Students using the i-Ready Classroom Mathematics curriculum showed significantly greater growth in math achievement compared to peers using traditional curricula, suggesting that iRCL may be an effective tool for supporting low-performing elementary students.

Authors

• Preston Martin, University at Buffalo - SUNY

• Gregory King, Curriculum Associates