The Homework Site

Research desk

Research & data

Original data roundups from The Homework Site: what districts, studies and public records say about snow days, grading, homework and school calendars — with every source linked and datasets you can download.

Printed bar charts without labels spread on a wooden desk with a magnifying glass

The research desk publishes original datasets about US schools: what districts actually write into their homework and AI policies, how grades and homework time have changed, and what it takes to close school for snow. Each page states its method, its gaps and its snapshot date, and five of the six come with a free CSV you can reuse.

Start here

How a dataset gets built

Every dataset starts with a question someone actually asks, such as “is there a homework limit in my district?” We then go to the primary record: a board policy, a federal survey table, a study’s own results section. Each row carries a link to that source. Where a document could not be found or opened, the row says so instead of filling the gap with a guess. That is why some tables show “unknown” more often than you might expect. It also means the numbers can be checked line by line, which matters if you plan to cite them.

How the research feeds the rest of the site

These pages are the evidence layer under our guides and tools. The district homework data sits behind how much homework is normal and homework statistics. The detector studies support do AI detectors actually work? and the teacher guide to detecting AI work honestly. The GPA trend gives context to the GPA scale and converter, and the snow thresholds inform the snow day calculator. State-level rules, such as kindergarten cutoffs and required school days, live on the school facts by state pages.

Using and citing the data

The CSV downloads are free under CC BY 4.0: credit The Homework Site and link to the page. Journalists, teachers and students are welcome to quote individual figures with the same credit. If you spot an error or a newer source, the contact section of our about page is the fastest route to a correction.

Original research

Easiest and Hardest AP Exams: Score Distributions 2019–2026

AP score distributions for all 40 exams, 2019–2026: the easiest and hardest AP exams by pass rate and share of 5s, the …

Original research

The Enrollment Cliff: High School Graduate Projections by State and What They Mean for Applicants

The enrollment cliff in numbers: US high school graduates peaked in 2025 and fall 13% by 2041 (WICHE). State table, regional …

Original research

Gaokao vs SAT: China's College Entrance Exam and the SAT Compared in Numbers

Gaokao vs SAT in numbers: 12.9 million gaokao candidates in 2026 vs 2.0 million SAT takers, 750 vs 1600 points, one sitting vs …

Original research

National Merit Semifinalist Cutoffs for the Class of 2027: All 50 States, Trends and Dates

National Merit semifinalist cutoffs for the Class of 2027 in all 50 states and DC: 208 to 223, every state lower than 2026. …

Original research

SAT and ACT Scores by State 2026 (and Why the Rankings Mislead)

Average SAT and ACT scores by state for the class of 2026, with participation rates and testing mandates. Why low-participation …

Original research

School Cellphone Bans in 2026: Which States Have Them and What the Research Shows

School cellphone bans in 2026: 25 state laws verified one by one, bell-to-bell vs class-time rules, 77% Pew support and what …

Original research

Does Screen Time Hurt Grades? What the Biggest Studies Actually Show

Screen time and grades: PISA ties 5-7 leisure hours a day to math scores 49 points lower, yet meta-analyses find small effects. …

Research · Learning

Is Students' Attention Span Really Shrinking? The 8-Second Myth vs the Research

Is the 8-second attention span real? No. We trace the goldfish myth and show what studies of screen switching, phones and short …

Original research

Can Kids Still Read? US Students' Reading Scores on NAEP and PISA (2026)

Students' reading scores in 2026: NAEP 2024 has 30% of 4th graders Proficient, PISA 2025 fell 14 points to 490. What "can't read" …

Research · Dataset

Teachers Are Warning Us: What the Data Says About Behavior, Absences and Burnout

Teachers say kids skip school and act out. We checked AEI, RAND, Pew and NCES data: chronic absenteeism is 22.6%, down from a …

Original research

Is Test-Optional Ending? Colleges That Require the SAT or ACT Again (2026–27 Tracker)

Colleges requiring the SAT or ACT again in 2026: 25 reversals with dates and sources, while 85% of colleges stay test-optional …

Original research

AI Policies in 32 of the Largest US School Districts (2026 Tracker)

What 32 of the largest US school districts allow students to do with AI, as of October 14, 2026: who has a policy in force, who …

Original research

Homework Policies in 32 of the Largest US School Districts

We looked for the official homework rules of 32 of the largest US school districts. Only 3 of the 15 documents we read set …

Research review

Do AI Detectors Flag Honest Students? What 6 Studies Found

A sourced review of six studies and vendor reports on AI-detector false positives: the Stanford finding that 61% of non-native …

Research · Dataset

Grade Inflation in US High Schools: Average GPA 1990–2019 and What It Means

The average US high school GPA rose from 2.68 in 1990 to 3.00 in 2009 (NCES) and 3.11 in 2019 (as reported by the Hechinger …

Research · Dataset

How Much Homework Do Students Get Around the World? Hours by Country (PISA)

Weekly homework hours for 15-year-olds in 65 countries and economies from OECD PISA 2012, with 2003 comparisons: Shanghai 13.85 …

Research · Dataset

How Much Snow Does It Take to Close School? What Districts and Data Say

A sourced dataset of 32 snow and cold closure thresholds from US districts, news reports, a crowdsourced map and an NBER study — …

Frequently asked questions

Where does The Homework Site's research data come from?
The Homework Site builds its datasets from primary sources: district policy documents and board pages, federal surveys such as the NCES High School Transcript Study, OECD PISA reports, peer-reviewed studies, vendor documentation and named news reports. Each row in a dataset links to the source it came from. When a figure could not be verified, it is left out or marked as unknown rather than estimated.
Can I use The Homework Site's data in my article or class?
Yes. The downloadable datasets on The Homework Site research pages are published under Creative Commons Attribution 4.0 (CC BY 4.0). You can share and adapt them for any purpose, including commercially, as long as you credit The Homework Site and link to the page the data came from. Each research page includes a suggested citation.
What is the average high school GPA in the US?
The average US high school GPA was 3.11 in 2019, according to the Hechinger Report’s account of the federal transcript study. It was 2.68 in 1990 and 3.00 in 2009, a rise of 0.43 points in 29 years. Over roughly the same period, NAEP math and reading scores for 12th graders barely moved, which is why researchers call the trend grade inflation.
How many US school districts have an AI policy?
In The Homework Site’s tracker of 32 of the largest US school districts, 14 had a published AI policy or guidance in force as of October 14, 2026: 6 board policies and 8 guidance documents. Four more were in draft, five restricted AI, and for nine no public policy was found. Most districts with rules allow student use only with conditions such as teacher permission.
How much snow does it take to close school?
There is no national number. Reported thresholds range from any snow at all in the Deep South to more than 2 feet along the Canadian border, according to a crowdsourced 2014 map. In between, sourced figures include about 3 inches in central Illinois and Indiana and 6 inches for New York City school buildings. In the Upper Midwest, extreme cold closes schools more often than snow depth.