AI, Tutoring, and Higher Education · Reviewed 2026-10-01
In a randomized trial across a large public university, access to a course-integrated AI study assistant reduced final grades and participation in existing online course activities within matched sections of the same course.
Read the research →Student Well-Being and Academic Development · Reviewed 2026-10-01
A randomized trial with 2,246 middle school students found no immediate academic effect from a parent-directed empathy program, but treatment students later became more likely to enter elite high schools and less likely to miss the entrance exam.
Read the research →School Schedules, Staffing, and Finance · Reviewed 2026-10-01
A new Missouri study found that adopting a four-day school week did not meaningfully improve non-teacher staffing or retention and did not reduce schoolwide expenditures overall.
Read the research →College Success and Peer Influence · Reviewed 2026-09-30
At an elite Ugandan university, students assigned to more scholarship peers earned stronger grades even when those peers were not the most academically prepared.
Read the research →College Access and Financial Aid · Reviewed 2026-09-30
In a large public college system, receiving a Federal Work-Study offer increased program participation by 27 percentage points, but most offered students still did not take a work-study job.
Read the research →Assessment and Learning Science · Reviewed 2026-09-30
Across more than two decades of Massachusetts English language arts exams, researchers found that statewide scores usually could not cleanly separate higher-order thinking from foundational skills.
Read the research →AI in Higher Education · Reviewed 2026-09-29
A four-year randomized trial found large short-term effects on selected registration tasks but no detectable effect on grades, credits, persistence, or four-year graduation.
Read the research →College Learning and Motivation · Reviewed 2026-09-29
Across four experiments with nearly 12,000 Canadian university students, brief growth-mindset and social-belonging interventions produced no detectable grade or persistence gains.
Read the research →Whole-Student Support · Reviewed 2026-09-29
Three years of coordinated support was associated with higher math scores, graduation, and early adult earnings, with only about half of the graduation effect explained by test scores.
Read the research →College Success · Reviewed 2026-09-28
Embedding tutors and advisers in community college courses raised support use and improved course outcomes, even though every student already had access to the same central services.
Read the research →College Access · Reviewed 2026-09-28
An early four-year tuition guarantee moved high-achieving, low-income students toward the University of Michigan and accelerated completion, but did not raise six-year degree attainment.
Read the research →Learning Technology · Reviewed 2026-09-28
In 83 Indian middle schools, dedicated implementation staff raised weekly Khan Academy use from 7.2 to 47.4 minutes and produced large math gains over 31 weeks.
Read the research →School Leadership · Reviewed 2026-09-27
Assistant principals encouraged by their principals were far more likely to aspire to the principalship, while evaluation scores were not related to aspiration.
Read the research →AI in Education · Reviewed 2026-09-27
Students found meanings in their classroom talk that an LLM and an adult-designed coding scheme missed.
Read the research →Higher Education · Reviewed 2026-09-27
A university's global ranking did not predict how deeply it had embedded AI into governance, curriculum, policy, and support.
Read the research →Special Education · Reviewed 2026-09-26
A nationwide Swedish special-education training program showed no detectable test-score gain overall or for students with individualized education programs.
Read the research →College Readiness · Reviewed 2026-09-26
Top-ranked Texas students entered public flagships with very different academic profiles depending on the resources available at their high schools.
Read the research →Tutoring · Reviewed 2026-09-26
In one online literacy program, familiarity with the tutoring system was more predictive of student growth than degrees or prior classroom experience.
Read the research →Tutoring · Reviewed 2026-09-25
A government-run online math tutoring program still improved learning at scale, but its effects were only about one-third of the original pilot.
Read the research →Student Support · Reviewed 2026-09-25
A North Carolina program pairing a school nurse with a social worker reduced absences among high-risk elementary students.
Read the research →Mathematics Education · Reviewed 2026-09-25
Los Angeles students with access to advanced seventh-grade math later completed more rigorous courses and enrolled in four-year college more often.
Read the research →Student Engagement · Reviewed 2026-09-24
Classroom phone restrictions modestly reduced recreational use among Chilean eighth graders but did not improve test scores and were linked to lower academic self-belief.
Read the research →Learning Science · Reviewed 2026-09-24
A 25-lesson goal-setting course improved reported soft skills in Kenya, especially among girls, but did not produce detectable exam-score gains.
Read the research →College Access · Reviewed 2026-09-24
An updated study of U.S. public universities found that international enrollment brought tuition revenue and was linked to more in-state freshmen, not fewer.
Read the research →Families and Communities · Reviewed 2026-09-23
Across more than 11,000 U.S. districts, parent-teacher and fundraising groups were consistently associated with higher test scores, while professional education nonprofits were not consistently associated.
Read the research →College Success · Reviewed 2026-09-23
Adults who connected with Tennessee's Navigate Reconnect advisers earned more post-secondary credentials than matched workers, while their earnings dipped modestly during the three-year study window.
Read the research →Special Education · Reviewed 2026-09-23
A 13-year Michigan study found that estimated additional school spending varied from about $2,000 to more than $23,000 depending on disability category, while stressing that a category does not define a student's services.
Read the research →Early Learning · Reviewed 2026-09-22
A new analysis of the national Head Start experiment found persistence within skill domains, but no strong evidence that cognitive gains automatically transferred to social-emotional gains or vice versa.
Read the research →College Transition · Reviewed 2026-09-22
A statewide Tennessee advising program increased immediate college enrollment, but researchers did not detect significant gains in persistence, degree completion, or employment.
Read the research →Teacher Pipeline · Reviewed 2026-09-22
Texas high-school education courses attracted students interested in teaching, but expanding access did not measurably increase entry into the public-school teacher workforce.
Read the research →Parent Partnership · Reviewed 2026-09-21
Weekly messages encouraging caregivers in Kenyan private primary schools produced a small average test-score gain among students who remained, while more families also left the school network.
Read the research →College Success · Reviewed 2026-09-21
In a randomized Spanish university study, showing students where their midterm score ranked in the class improved final-exam performance, with the strongest gains among students who had overestimated their standing.
Read the research →Educator Quality · Reviewed 2026-09-21
Maryland data linked teachers' local ties to stronger retention, but formal preparation and licensure pathway predicted early-career math effectiveness more strongly than localness itself.
Read the research →Student Support · Reviewed 2026-09-20
A New York City policy reduced the probability of a juvenile crime record near schools, but diversion expanded much more than arrests fell and suspensions increased even though measured behavior did not change.
Read the research →Assessment · Reviewed 2026-09-20
Using the same eight years of school data, researchers showed that changing how growth, proficiency, and time are weighted changes which schools are labeled for improvement.
Read the research →School Improvement · Reviewed 2026-09-20
In Michigan turnaround schools, school-level teacher turnover showed little immediate improvement, then fell substantially in the third implementation year and remained lower afterward.
Read the research →College Success · Reviewed 2026-09-19
At one public university, standard academic probation increased dropout and reduced degree completion among low-income students near the GPA cutoff, while a version paired with coaching removed those negative effects and improved grades.
Read the research →Attendance and Student Well-Being · Reviewed 2026-09-19
In one immigrant-receiving district, foreign-born students became 2.2 percentage points more likely to be absent after immigration enforcement intensified, a 37 percent increase from baseline.
Read the research →Literacy and Assessment · Reviewed 2026-09-19
Ohio third graders who narrowly fell below a fall promotion cutoff made larger reading gains than peers just above it, and the gains persisted through Grade 7, within a system that also triggered targeted literacy support and school accountability.
Read the research →Early Learning · Reviewed 2026-09-18
In the Head Start Impact Study, Black children retained a 0.23 standard deviation cognitive gain through kindergarten while the comparable gain among non-Black children faded.
Read the research →Learning Technology · Reviewed 2026-09-18
Among 10,075 third graders, more than 60 minutes of digital practice aligned with classroom instruction was associated with a 0.061 standard deviation gain in foundational literacy in the study's fixed-effects model.
Read the research →Learning Differences · Reviewed 2026-09-18
Across all 50 states and the District of Columbia, adopting Response to Intervention produced no detectable aggregate change in specific learning disability, speech or language impairment, or overall disability identification.
Read the research →Admissions and Assessment · Reviewed 2026-09-17
In a Mexico City structural model, placement-score differences explained 83 percent of the socioeconomic gap in elite public high-school assignments.
Read the research →Early Learning · Reviewed 2026-09-17
Across 1.58 million U.S. children, the center-based early-education enrollment gap narrowed only one to two percentage points per decade for 4-year-olds and widened for 3-year-olds.
Read the research →College Success · Reviewed 2026-09-17
At one large public university, reported bereavement and illness absences were associated with lower next-semester retention, especially among first-year students.
Read the research →Student Engagement · Reviewed 2026-09-16
Average daily attendance fell from 93.04 percent in 2018 to 2019 to 92.32 percent in 2023 to 2024, roughly two additional missed days per student, while national averages concealed far more severe local and individual patterns.
Read the research →Learning Science · Reviewed 2026-09-16
Across 25 randomized education studies, average intervention-level effects declined from 0.14 standard deviations at posttest to 0.06 after one to two years, yet a small 0.04 standard deviation adult effect remained.
Read the research →Mathematics · Reviewed 2026-09-16
In quasi-randomly assigned middle-school classrooms, stronger math beliefs among girls predicted higher later math scores for female classmates, while stronger math beliefs among boys lowered girls' own math beliefs.
Read the research →Education Investment · Reviewed 2026-09-15
A 2026 Virginia analysis found mostly null spending associations and mostly negative significant associations, while emphasizing that funding categories, timing, pass-rate measurement, and implementation sharply limit what the pattern can mean.
Read the research →Academic Recovery · Reviewed 2026-09-15
In North Carolina data covering 1.46 million students, the youngest cohort closed 58 percent of the reading gap and 87 percent of the mathematics gap by Grade 5, while average scores still remained below pre-pandemic benchmarks.
Read the research →Leadership and Learning · Reviewed 2026-09-15
Across more than 37,000 district-year records, cumulative superintendent instability was associated with lower achievement, with larger penalties after contentious departures and in the highest-poverty districts.
Read the research →Student Support & Implementation · Reviewed 2026-09-14
In a randomized trial with 2,528 students across two disadvantaged urban school districts, a mature comprehensive student-support program produced small, statistically insignificant overall effects on high-school graduation, while exploratory results varied by site and over time in ways that tracked treatment timing and implementation differences.
Read the research →College Readiness & Dual Enrollment · Reviewed 2026-09-14
Among roughly 1.74 million California public high-school students, dual-enrollment failure was associated with a 12.2-percentage-point lower college-enrollment rate than passing, but among students without AP access, those who attempted dual enrollment and failed were still 6.4 percentage points more likely to enroll in college than matched students who never participated.
Read the research →Gifted Education & Opportunity · Reviewed 2026-09-14
Across New York City public schools from 2010 to 2024, gifted-program presence varied more strongly between neighborhoods than between community school districts, and neighborhood homeownership independently predicted whether a program was present even after accounting for poverty, education, and school demographics.
Read the research →Assessment, Grading & Student Outcomes · Reviewed 2026-09-13
A 2026 NBER study using high-school records from Los Angeles and Maryland found that exposure to teachers who broadly gave higher grades than student performance predicted reduced future test scores, high-school graduation, college enrollment, and earnings; by contrast, leniency specifically at the pass/fail margin reduced retention and increased graduation.
Read the research →College Readiness & Academic Beliefs · Reviewed 2026-09-13
A 2026 nationally representative U.S. study found that student effort, parental investment, and beliefs about earning a college degree responded predominantly to local class rank rather than broader academic rank, even though both were strong predictors of eventual college completion.
Read the research →Families, Tutoring & Decision-Making · Reviewed 2026-09-13
In a 2026 field experiment in China, truthful information about peer families' tutoring hours shifted parents' tutoring investments by more than 30%; one month later, treated participants had increased actual tutoring by 1.14 hours per week, a 51% increase relative to the average baseline.
Read the research →College Admissions & Math Access · Reviewed 2026-09-12
In a lab-in-the-field experiment with 292 admissions officers from 69 selective institutions, otherwise high-achieving engineering applicants who had not taken calculus because it was unavailable at their high school were 16 percentage points less likely to receive an admit recommendation.
Read the research →Curriculum & Emotional Learning · Reviewed 2026-09-12
Across public-school textbooks and influential children's literature, researchers found that text exposed children to a broad range of emotions, while about 90% of visually depicted emotions were happiness or calm; nearly half of pages showed no overlap between emotions described in text and those shown in images.
Read the research →Assessment & Learning Conditions · Reviewed 2026-09-12
Using six years of Tulsa MAP testing and high-resolution weather data, researchers estimate that a test day reaching the 90s, compared with a day in the 70s, lowers math performance by more than 1.5 percentile ranks, or roughly 0.06 standard deviations, with smaller effects in reading.
Read the research →Educational Technology & Student Outcomes · Reviewed 2026-09-11
Across 7,825 U.S. school districts, greater uptake of a federal broadband subsidy was associated with lower math and reading/language-arts achievement among economically disadvantaged students, despite the program dramatically expanding internet access.
Read the research →Higher Education, Curriculum & Student Success · Reviewed 2026-09-11
A 2026 NBER study linking 459,415 syllabi from seven Texas public universities to student records found that exposure to more frontier-oriented course content was associated with higher completion, GPA, graduate-school attendance, earnings, and faster time-to-degree.
Read the research →Tutoring, STEM & Student Motivation · Reviewed 2026-09-11
In a randomized study of 422 ninth-grade Algebra 1 students, girls assigned to female tutors reported 0.73 SD higher STEM interest and were 3.9 percentage points more likely to earn at least a C- than girls assigned to male tutors.
Read the research →Learning Technology & Evidence Quality · Reviewed 2026-09-10
A September 2026 meta-analysis synthesized 841 effect sizes from 141 studies of educational apps for children ages 3 to 9. The pooled effect was positive, but effects were smaller in randomized trials and larger samples, and were larger on researcher-developed measures than on more independent outcomes.
Read the research →Expectations, Assessment & Opportunity · Reviewed 2026-09-10
Using nationwide Italian administrative data, randomized vignettes, belief elicitation, and a field experiment, researchers found that teachers were less likely to recommend demanding academic tracks for disadvantaged students even when academic performance and interests were held constant; personalized feedback substantially reduced recommendation gaps for high-achieving students.
Read the research →Educator Quality & Measurement · Reviewed 2026-09-10
A September 2026 study of Chile's national teacher-evaluation system found that teachers who completed re-evaluation improved their own evaluation scores by about one standard deviation, yet difference-in-differences estimates ruled out student-achievement effects above 0.04 standard deviations and found no meaningful changes in reported teaching practice or caring.
Read the research →Educator Quality & Instruction · Reviewed 2026-09-09
In 48 public primary schools in Pune, college STEM fellows with less formal education and teaching experience than regular teachers scored 1.56 standard deviations higher on a content-and-pedagogy assessment; when paired with training, coaching, and lesson scripts, the intervention raised student achievement by 0.32 SD in math, 0.23 SD in science, and 0.16 SD in language after one year.
Read the research →Learning Science & Study Skills · Reviewed 2026-09-09
In a 2026 randomized intervention with more than 2,000 undergraduates, general feedback had little effect, while personalized feedback shifted students toward more effective study practices and raised exam scores by about 0.05 to 0.08 standard deviations.
Read the research →Tutoring & Student Outcomes · Reviewed 2026-09-08
An August 2026 NBER study estimates that the opening of a district's first major private tutoring-chain center increased district-wide math and reading scores by roughly 0.02 to 0.03 standard deviations, while the gains were unevenly distributed across student groups.
Read the research →Tutoring & Implementation · Reviewed 2026-09-07
Harvard CEPR examined 22 math and reading recovery interventions across eight districts. Most reached fewer students and delivered less instructional time than planned, while three tutoring programs showed positive estimated effects.
Read the research →Tutoring & Literacy · Reviewed 2026-09-07
In a four-year randomized trial, students assigned to an adaptive early-literacy support model scored 0.12 to 0.17 standard deviations higher on state reading assessments across Grades 1 through 3, even though the model was less intensive than many high-dosage tutoring programs.
Read the research →Tutoring & Literacy · Reviewed 2026-09-07
In 81 Danish schools, targeted early-literacy tutoring improved decoding for the intended students and the study also detected gains among peers who were never directly tutored.
Read the research →Learning Science · Reviewed 2026-09-07
A major review of ten common learning techniques rated practice testing and distributed practice as high-utility approaches because their benefits generalized across a broad range of learners, materials, and outcomes.
Read the research →AI in Education · Reviewed 2026-09-07
In a randomized experiment with more than 6,000 middle-school students, AI support improved next-attempt correctness after errors, while delayed-learning gains remained more limited and dependent on structure.
Read the research →Tutoring Research · Reviewed 2026-09-07
A peer-reviewed meta-analysis of randomized PreK–12 tutoring field experiments reported an overall pooled effect of 0.288 standard deviations, with meaningful variation across program designs and settings.
Read the research →Learning Science · Reviewed 2026-09-07
In two experiments, repeated study could produce better recall after five minutes, but prior retrieval testing produced substantially better retention after two days or one week.
Read the research →AI in Education · Reviewed 2026-09-07
In a randomized trial involving more than 700 tutors and 1,000 students, students whose tutors had access to Tutor CoPilot were 4 percentage points more likely to master math topics, with larger gains among students of lower-rated tutors.
Read the research →Tutoring Research · Reviewed 2026-09-07
A 2026 meta-analysis of 263 randomized tutoring trials found pooled effects of 0.16 to 0.22 SD in studies most similar to large-scale programs using independent standardized outcomes, versus 0.40 SD in the full sample.
Read the research →Tutoring & Implementation · Reviewed 2026-09-07
Harvard CEPR evaluated 30 minutes of virtual ELA tutoring three times per week for 12 weeks. The study did not detect effects on ELA, math, or attendance in the studied implementation.
Read the research →Tutoring & Implementation · Reviewed 2026-09-07
A two-year randomized trial found that simplifying enrollment raised initial tutoring take-up sharply, yet weekly attendance remained below half and the main math assessment estimate was statistically imprecise.
Read the research →AI in Education · Reviewed 2026-09-06
A two-year randomized school experiment found modest math gains, but the most revealing result was behavioral: access to AI tutoring did not guarantee meaningful engagement with it.
Read the research →
Noor Lyra Tutoring