Higher algebra scores in the trial’s second year

Cognitive Tutor Algebra I. [1] [2]

An ACT-R-based cognitive model tracked students’ solution steps and mastery of individual skills. [4]

The study does not identify a numbered model or software release. [1] [4]

Carnegie Learning. [1] [2]

The tutor checked students’ steps, provided hints, and selected practice for skills that needed work. [1] [2]

Classes generally used the textbook and classroom activities three days a week and tutoring software two days a week. [1] [2]

Students entered algebra steps. The tutor returned feedback and selected further exercises. [1] [2]

The tutor selected exercises and gave feedback. Teachers remained responsible for instruction. [1] [2]

Teachers taught the curriculum and supervised students’ work with the software. [1] [2]

Teachers received three days of training before implementation and another day during the school year. [1] [2]

The intelligent tutor used a cognitive model of students’ algebra skills. [1] [2]

The trial compared complete curricula. It did not separate the tutor’s effect from the textbook, activities, and teacher training. [1] [2]

Not assessed

High-school algebra scores improved in the second year of implementation. [1] [2]

Improved learning. [1] [2]

The second-year gain was positive, but it was not statistically significant after the Clearinghouse adjusted for multiple comparisons. [3]

Students. [1] [2]

Students who took Algebra I at participating high schools. [1] [2]

The revised analysis included 11,066 high-school students and 5,519 middle-school students across two annual cohorts. [1] [2]

The reported gain was in the second high-school cohort. First-year and middle-school estimates were not statistically significant. [1] [2]

Each cohort studied Algebra I for one school year. [1] [2]

Improved algebra learning was the intended benefit. [1] [2]

Schools incurred higher adoption costs and needed time for implementation. The trial did not show significant gains in every setting. [1] [2] [5]

AIBD-0214-E01 [1] [2] [3] [6]

Researchers assigned matched pairs of schools to Cognitive Tutor or their usual curriculum. [1] [2]

Students took standardized algebra readiness and proficiency tests. Researchers adjusted the end-of-year comparison for baseline scores and other characteristics. [1] [2]

The trial included 73 high schools and 74 middle schools. The revised analysis excluded students without observed baseline tests. [1] [2]

Algebra proficiency. [1] [2]

Schools continued the algebra curriculum that they normally used. [1] [2]

The revised high-school estimate was −0.07 standard deviations in year one and +0.21 in year two. [1] [2]

The software’s separate contribution was not estimated. [1] [2]

The second-year estimate depended on adjustment for student characteristics. Its p-value of .04 did not meet the Clearinghouse’s corrected threshold of .025. [2] [3]

Students generally used the tutor for about 40% of class time during the school year. [1] [2]

The 2024 reanalysis provides public code. Its full replication data are withheld to protect the participating schools’ identities. [6]

Annualized curriculum adoption cost. [5]

Excluding Houston, estimated annual cost was US$97.18 per student for Cognitive Tutor and US$27.88 for comparison curricula. [5]

U.S. dollars. The estimate used the vendor’s 2006 price guide to represent ordinary district adoption. [5]

Per student per year, with five-year adoption and equipment lives. [5]

These are estimated adoption costs, not the trial’s discounted invoices. [5]

School districts would pay for materials, software, equipment, and training. [5]

The estimates compare expenditure. They do not value learning gains or calculate net benefits. [5]

Costs include textbooks, software, computers, and curriculum-specific teacher training. [5]

The comparison is the cost of the curricula that control schools used. [5]

RAND combined vendor prices and district reports, annualized the costs, and weighted them by the number of schools. [5]

The analysis does not estimate financial returns or a payback period. [5]

Carnegie Learning supplied the curriculum. Participating schools taught it, and RAND researchers evaluated it. [1] [2]

The study covered 147 schools in 51 districts across Alabama, Connecticut, Kentucky, Louisiana, Michigan, New Jersey, and Texas. [1] [2]

Carnegie Learning supplied the commercial curriculum and tutoring software. [1] [2]

Schools needed computers for regular sessions, the curriculum materials, and trained algebra teachers. [1] [2]

Development and school deployment were in the United States. The study does not document an international flow of student data. [1] [4]

Not assessed

Schools needed regular computer sessions, trained teachers, and time to implement the curriculum. [1] [2]

The Clearinghouse reviewed five qualifying Algebra I studies and rated their combined results as mixed. [3]

The revised Clearinghouse review rated the wider algebra evidence as mixed and removed this trial’s statistical significance after correcting for multiple comparisons. [3]

Not assessed

Not assessed

Not assessed

Not assessed

Not assessed

AIBD-0214-source-0001, AIBD-0214-source-0002, AIBD-0214-source-0003, AIBD-0214-source-0004, AIBD-0214-source-0005, AIBD-0214-source-0006 [1] [2] [3] [4] [5] [6]

The original trial, its addendum, and the revised Clearinghouse review establish the result. Separate original papers describe the tutor and adoption costs. [1] [2] [3] [4] [5] [6]

English. [1] [2] [3] [4] [5] [6]

RAND conducted the trial and cost analysis. The Clearinghouse independently reviewed the evidence. The technical account was written by developers. [1] [2] [3] [4] [5] [6]

The trial and addendum provide the estimates. The Clearinghouse review supplies the correction for multiple comparisons. [1] [2] [3] [4] [5] [6]

The original manuscript, addendum, revised review, mechanism paper, and cost report were inspected. A later dataset paper supplies the school-year dates. [1] [2] [3] [4] [5] [6]

The second high-school cohort had higher algebra scores under the combined curriculum. [1] [2]

The sources are linked to the findings that they support. [1] [2] [3] [4] [5] [6]

A large randomized trial supports a positive second-year estimate. An independent review found that it did not survive correction for multiple comparisons. [1] [2] [3]

Not assessed

High-school students’ algebra learning. [1] [2]

Personal: 2 [2] [3]

The second high-school cohort had a measurable algebra gain, although the corrected significance test was inconclusive. [2] [3]

The second-year gain was positive, but it was not statistically significant after the Clearinghouse adjusted for multiple comparisons. [1]

Not assessed

Not assessed

Not assessed

Independently studied [1] [2] [3] [4] [5] [6]

A randomized evaluation of a full algebra curriculum. [1] [2]

Not assessed

Education. [1] [2]

Cognitive Tutor combined classroom algebra lessons with practice that adapted to each student. [1] [2]

The trial covered the 2007–08 and 2008–09 school years. [6]

Schools used the curriculum for two consecutive years. [1] [2]

SHA-256: d0d3b347b993…

Not assessed

Not assessed

Not assessed

5 revisions. Latest: 25 October 2026.