Syntes: Kremer, Gray-Lobe, de Laat & van der Haar (2026) — a year of interleaved maths practice in Nigeria

Källa: Kremer, M., Gray-Lobe, G., de Laat, J., & van der Haar, L. (2026). A Year of Desirable Difficulties: The Impact of Interleaving Math Practice in Nigeria (NBER Working Paper 31853; November 2023, revised February 2026). https://doi.org/10.3386/w31853 — working paper, not peer reviewed (the paper says so on its title page).

Verified numbers (full NBER PDF read 2026-10-15; not stored)

• Setting: Bridge Nigeria low-cost private schools (Lagos and Osun states); scripted lessons on teacher tablets; two 45-minute maths periods a day ("my turn, your turn": teacher worked examples, then pupils' practice problems); weekly revision; mid- and end-term tests in both arms.
• Design: 63 schools, one excluded; 62 randomised (one grade-5 classroom each), stratified; 28 interleaved, 34 blocked; two single-school strata dropped; one interleaved school had no eligible pupils. Analysis sample 687 incumbent pupils, mean age 9.81, about half girls. Standard errors clustered by school.
• Treatment: same worked examples and the same number of practice problems in both arms; in interleaved classrooms about half of the practice problems came from earlier lessons in the same term (mostly the preceding three lessons or the current 3–5-lesson unit). The authors note this changes spacing as well as interleaving; effects cannot be attributed to interleaving alone.
• Tests: 30 open-ended items each, graded by a different teacher; termly posttests (average retention intervals 39, 38 and 29 days) and a cumulative end-of-year test five days after the program ended (items mirroring the three posttests; content introduced on average about 145 days earlier).
• Short-term (posttest index): +0.28 SD (Table 3, Panel B); similar across the three posttests (largest 0.31 SD on the third). Follow-up 97 %.
• Cumulative test: +0.03 SD, 95 % CI −0.28 to 0.34; equality with the posttest effect rejected (p = 0.02). Surrogate analysis predicted 0.24 SD from the posttest gains. Follow-up about 78 % in comparison schools, 4 points lower in interleaved schools (ns).
• Heterogeneity: posttests +0.54 SD for pupils one SD below the mean, +0.16 SD one SD above; cumulative test 20th percentile +0.49 SD, 90th percentile −0.33 SD (p < .05).
• Authors' reading: short-term gains similar to high-income-country pilots; no evidence of a cumulative year-end gain; inconclusive on cumulative performance (moderate effects either way not ruled out); frequent testing in both arms may have given blocked classrooms spaced, mixed practice too.

Role for C11

Weakened C11 2026-10-15 (durability bound): the advantage is shown on tests about a month after the material was taught; over a whole school year, on a cumulative test, it was not shown in this year-long classroom trial. Kill withheld (same trial supports the claim at about a month; imprecise year-end estimate; one setting; working paper). Preprint flag and recheck item in (the preprint rule).