Time costs of mastery learning (Arlin & Webster 1983; Arlin AERJ & RER 1984)

Arlin, M.; Arlin, M., & Webster, J. · 1984

grade Cquasi-experimentindependentreplicated

Sample

Randomized 4-day experiment + 4 years of time-to-mastery records

Population

Elementary students; longitudinal mastery arithmetic program grades 1-4.

Design

Quantifies the equal-time confound and the time-inequality dynamics Bloom's model predicts should shrink.

Key findings

Mastery learning does not reduce individual differences; it CONVERTS them into time inequality. Huge aligned-test effects evaporate into negative learning-per-hour once time is counted, and four years of records show the slowest:fastest time ratio GROWING (2.5:1 → 4.2:1). The policy choice is unavoidable: hold time constant and let achievement vary, or hold achievement constant and tax fast learners' time.

Genetic confound

The clearest hereditarian-consistent result in the cluster: stable ability differences do not shrink with cumulative mastery of prerequisites — they re-express as GROWING time inequality.

Replication notes

Time-cost finding replicated in medical education; no long-term study shows corrective-time convergence.

DOI / URL

10.3102/00028312021001103

Cite

BibTeX RIS

Effects

Outcome	Metric	Value	Measure	Timing	Vs	Horizon	Class
Immediate achievement, mastery vs non-mastery	ES	>3.0 on aligned daily tests — but mastery students spent MORE THAN TWICE the learning time	researcher-designed	immediate	business-as-usual	end-of-treatment	domain-skill
Retention PER HOUR of instruction	ES	-1.17 — learning-per-unit-time FAVORED CONTROLS	researcher-designed	4 days post	business-as-usual	end-of-treatment	domain-skill
Time-to-mastery inequality, slowest vs fastest quartile	ratio	2.5:1 in grade 1 RISING to 4.2:1 by grade 4 — no convergence	unknown	4-year longitudinal	none	not-applicable	domain-skill

Cited by

Mastery learning (teach → test → reteach to criterion → advance)mixedconf: highgc: low