RPM fluency metrics predict standardized math achievement better than traditional raw scores, with common-problem RPM the strongest predictor¶

Claim · All claims
Evidence · 1 study (3 entries) · associational r2–r3 · q3 · i0 negligible
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i0 Common RPM correlated with SBAC achievement at r = 0.67, exceptional RPM at r = 0.59, and the traditional raw score at r = 0.42, all significantly different from each other. → Ethan Roy 2025 q3 i? In regression models controlling for grade, income, and executive function, RPM models explained more SBAC variance than the raw-score model (common RPM model R2adj = 0.538), and the full RPM model fit significantly better (Cox z = −17.470, p < 0.001). → Ethan Roy 2025 (2)

Evidence¶

Ethan Roy 2025¶

Ethan Roy, Mathieu Guillaume, Amandine Van Rinsveld, Project iLead Consortium & Bruce D. McCandliss. (2025). Tablet-based arithmetic fluency assessment reveals developments in math cognition and math achievement from childhood to adolescence. npj Science of Learning. https://doi.org/10.1038/s41539-025-00314-5

q3 · i0 · associational · r3

Correlation analyses relating fluency metrics to SBAC scores. The article reports "common RPM r =0 . 6 7" and exceptional RPM r = 0.59, and notes the traditional raw score showed the weakest correlation of all metrics (r = 0.42), with all three metrics significantly different by Pearson and Filon's z-test.

"Pearson ’s r showed that the two RPM metrics signi ficantly correlated with math achievement (common RPM r =0 . 6 7 ,p < 0.001); (exceptional RPM r =0 . 5 9 ,p < 0.001)."

Ethan Roy 2025 (2)¶

Ethan Roy, Mathieu Guillaume, Amandine Van Rinsveld, Project iLead Consortium & Bruce D. McCandliss. (2025). Tablet-based arithmetic fluency assessment reveals developments in math cognition and math achievement from childhood to adolescence. npj Science of Learning. https://doi.org/10.1038/s41539-025-00314-5

q3 · i? · associational · r3

Series of linear models predicting SBAC scores controlling for grade, household income, and executive function measures (599 observations per Table 1). The article reports the common RPM model explained 54% of variance (R2adj = 0.538) and "the full RPM model explained 8.5% more variance in standardized test scores than the raw score model".

"Additionally, when comparing the full RPM model and the traditional raw score model, we found that the full RPM model explained 8.5% more variance in standardized test scores than the raw score model (ΔR2 adj < 0.098)."

Ethan Roy 2025 (3)¶

Ethan Roy, Mathieu Guillaume, Amandine Van Rinsveld, Project iLead Consortium & Bruce D. McCandliss. (2025). Tablet-based arithmetic fluency assessment reveals developments in math cognition and math achievement from childhood to adolescence. npj Science of Learning. https://doi.org/10.1038/s41539-025-00314-5

q3 · i? · associational · r2

Standardized beta-weights from the full RPM regression model (Table 1). The article reports "RPM on common problems was by far the strongest predictor of SBAC scores" (beta = 0.47), while exceptional RPM's beta was near zero in the combined model (0.01).

"Examining the standardized beta-weights of this model revealed that RPM on common problems was by far the strongest predictor of SBAC scores, even when considering measures of executivefunction and parental income"

Discussion¶

Related Claims¶

• A 3-minute tablet-based single-digit arithmetic fluency assessment is highly reliable and explains roughly 36% of variance in standardized math test scores — related
• Students across all grades answer exceptional problems (ties, successor, identity) faster than common problems — related