Formative Assessment with Analytic Rubrics in Problem Solving

Abstract

This study examined the impact of formative assessment with analytic rubrics on problem solving in upper‑primary education (EGB‑S) in Quito, Ecuador. We used a quasi‑experimental nonequivalent groups design with pretest–posttest and follow‑up across 48 classrooms in grades 8–10 (N = 1,396). The intervention entailed a five‑dimension analytic rubric (comprehension, planning, modelling, justification, verification) embedded in short feedback cycles (≤ 1 week) that included self‑ and peer‑assessment. The comparison condition completed similar tasks without explicit formative rubrics. Outcome measures comprised task performance, implementation fidelity indicators, and a brief self‑regulated learning inventory. ANCOVA with robust errors and baseline controls indicated a significant, moderate intervention effect on posttest scores (β = 6.08; p < .001; g = 0.45; 95% CI [0.36, 0.54]), corroborated by multilevel models (γ = 5.61; p < .001). Gains were larger in justification (g = 0.51) and verification (g = 0.54). The rubric displayed adequate reliability (ICC ≥ .75; α/ω = .78–.86) and construct validity (CFI = .957; RMSEA = .045). A mediation analysis suggested that approximately 17% of the total effect operated through increases in self‑regulation. Effects were larger in 10th grade and in classrooms with high implementation fidelity. We conclude that analytic rubrics used for formative purposes enhance learning on high‑cognitive‑demand tasks, strengthen actionable feedback and self‑regulation processes, and provide a scalable, low‑cost route to operationalize the curriculum in Quito. Future work should prioritise cluster‑randomised trials, cost‑effectiveness analyses, and cross‑disciplinary extensions to estimate effect persistence and transfer.