Why Not Let High Ability Students Start School in January? The Curriculum Compacting Study

27 school districts randomly assigned to three treatment groups plus control; ~436 teachers in grades 2-6; 783 high-ability students

US elementary students (grades 2-6) identified by their teachers as high-ability with advanced prior content knowledge, in rural, suburban and urban heterogeneous classrooms. Median pretest percentiles on out-of-level ITBS were 93 (reading and maths concepts) and 90 (maths computation).

Randomised at the DISTRICT level across 27 districts - a genuine cluster-randomised design and the strongest evidence in this topic. Two features do a lot of work. First, the outcome tests were administered OUT OF LEVEL (one grade up) precisely to avoid the ceiling effects that make grade-level tests useless for this population - the above-level method used as an instrument. Second, the result is a null, and it is developer-led: Reis and Renzulli invented compacting, so a null from them is substantially more credible than a positive would be. Limitations: outcomes remain ceiling-limited even one grade up (medians at or above the 90th percentile), and the replacement activities were 95% enrichment and only 18% acceleration, so the study tests "removing redundant content is safe" and not "acceleration raises achievement."

Teachers eliminated 40-50% of the curriculum for these students - most often in mathematics, then language arts - and achievement did not fall. On out-of-level ITBS post-tests there were no treatment-control differences in reading, maths computation, social studies or spelling; one treatment group scored significantly HIGHER than control in science; and students whose mathematics was compacted scored significantly HIGHER on maths concepts in every treatment group. Between two fifths and one half of the school year, for a child who already knew the content, was buying nothing measurable - and in maths it was costing something. The finding names the price of not diagnosing prior mastery, and it is the only randomised K-12 evidence on pretest-based content skipping the archive could locate.

Low. Districts were randomly assigned, so the comparison is between equivalent populations of teacher-identified high-ability students; selection into "high ability" limits generalisation beyond that band but not the internal contrast.

No independent replication in more than thirty years, which is conspicuous for a finding this consequential and is the main reason confidence in this topic is capped. The journal version (Reis et al. 1998, Gifted Child Quarterly) reports the same study on a 336-student analysis subsample.