Frequently Asked Questions

The State Assessment Data Repository (SADR) is a comprehensive, U.S.-wide database that makes state assessment data more widely accessible. It includes publicly available assessment results for students in grades 3-8 from all 50 states, the District of Columbia, Puerto Rico, and the U.S. Virgin Islands.

Proficiency results come directly from State Education Agencies - from their websites and data portals or by data request - and never include personally identifiable information. District and school characteristics come from the National Center for Education Statistics (NCES). A complete list of sources by state is in the Technical Documentation.

The SADR grew out of a project at Brown University and is now part of the Education Data Center. The eddatacenter.org website, including its design and AI features, was built in collaboration with external design and engineering partners.

The EDC is not the only site with state assessment data, but it stands apart in a few ways:

The most up-to-date data. The EDC posts new assessment data promptly after State Education Agencies release each spring’s results, so you can explore current trends across schools and districts by grade or student group.

Built to be user-friendly. The SADR and its AI assistant, Zelma, let anyone ask questions in plain language - no technical skills required.

Easy to access. Every state’s data can be downloaded as a CSV from the Data & Downloads page, or pulled programmatically from the API.

Broader subject coverage. Many sites focus on reading/ELA and math because that is what federal reporting requires. States often publish results for other subjects too - such as science and social studies - and the EDC includes them to capture more of what students are learning.

Zelma is the AI research assistant built into the Education Data Center. Ask a question about state test scores in plain English and Zelma returns a chart drawn directly from the State Assessment Data Repository - no spreadsheets required.

Zelma is meant to make these data accessible to anyone who can ask a question, which is everyone: educators and education leaders, journalists and policymakers, parents, researchers, and students. Data files are also available for anyone who wants to work with the underlying data.

Zelma answers questions about U.S. state assessment data and returns a line, bar, or pie chart. You can switch any result to a table view using the Graph / Table toggle.

The more specific you are, the better the answer. Try to include:

• The state, district(s), or school(s) of interest
• The subject(s)
• The timeframe - a trend over several years, or a single year
• A student group, if you want a breakdown

For example:

• "Show me math scores in Minnesota over time."
• "How do reading scores differ by race/ethnicity in New Jersey?"
• "What were the top 5 performing districts in math in Mississippi in 2024?"

If Zelma doesn't quite get it the first time, rephrase and try again - it's always improving.

When a question could mean more than one thing, Zelma asks a quick follow-up before answering - for example, which "Springfield" you mean, which subject, or which years. It only offers choices the data actually supports for that place, so you won't pick a combination that returns nothing.

Yes. Ask Zelma to break a result out by gender, race/ethnicity, income, disability status, English-learner status, or grade level - for example, "ELA and math proficiency by gender in Houston." Each group appears as its own bar or line so you can compare them directly.

For a district chart, Zelma automatically overlays the state average as a faint dashed line so you have context for how the district compares. Subjects are always shown separately and never averaged together.

Every state’s assessment files have been reformatted by the EDC team using a standardized set of variable names. Zelma draws only from these files to answer questions - it cannot use any outside data. For full transparency, each result also displays the underlying SQL code used to generate it.

A few things can cause this:

• Zelma may need more detail. Try naming a specific year, subject, or place.
• The state may not report the data in a usable form. Zelma relies on results broken out by grade level and on counts of students tested; when a state only reports grades 3-8 combined, or omits how many students were tested, those results can’t be charted. The proficiency rates may still be in the raw files on the Data & Downloads page.
• The data may not exist for that subject and year - the state may have run only a field test, received a 2021 federal testing waiver, stopped giving the assessment, or simply not published it.

Still stuck? Email us at info@eddatacenter.org and we’re happy to look into it.

To protect student privacy, states withhold results for small groups of students (many omit a group of about 10 or fewer). In the raw files these cases are marked with an asterisk (*) and missing data as --. Zelma leaves suppressed cases out of its charts rather than estimating a value.

Where a state reports a range instead of an exact number, the raw files preserve the range; for charts, Zelma uses the midpoint (for example, "20-30 students tested" becomes 25).

Every result has a Share button in the upper-right corner. Use it to copy the result’s link or post it - anyone with the link sees the same chart.

You can create an account from the sign-in page. An account lets you save your queries so you can return to them later.

The interactive map is built directly from the data. Start with a nationwide view - either by coverage (how many years of data each state has) or by proficiency - then click a state to see its districts, and a district to open a profile with a subject-by-grade breakdown and a multi-year trend.

From a district, open its school view to see every school plotted by location and colored by proficiency. School-level detail is available nationwide - all 50 states, the District of Columbia, and the U.S. Virgin Islands - for the most recent year of data.

To protect student privacy under FERPA, states withhold results for small groups of students. A suppressed school shows no percentage for the selected filters - it is not zero, and we never estimate a value in its place. Switching to "All students" or a different grade often reveals more schools.

Data are available for all 50 U.S. states, the District of Columbia, Puerto Rico, and the U.S. Virgin Islands. States vary in the years and data components that are publicly available - see the Data Availability page for the details. If there are years or data you’d like added, email info@eddatacenter.org.

Every state includes English language arts (ELA) and math. Where states publish them, data are also included for science, reading (when reported separately from ELA), writing, social studies, STEM, Spanish language arts, and English as a second language (Puerto Rico only). Not all states post results for subjects beyond ELA and math; see the Data Availability page for what each state publishes.

Version 3.1, released in June 2026, includes Spring 2025 assessment data for all states. How far back the data go varies by state, with the earliest year ranging from 1998 to 2018. No state administered assessments in Spring 2020 due to the COVID-19 pandemic, so no data are included for that year. You can see the available years for each state and subject on the Data Availability page.

Assessment data are currently limited to grades 3-8.

Where states have published or shared subgroup information, data are available for the following student groups and subgroups:

• All students
• Race/ethnicity - American Indian or Alaska Native; Asian; Black or African American; Native Hawaiian or Pacific Islander; Puerto Rican; Two or More; White; Hispanic or Latino; Filipino; Other Indigenous Peoples; Not Hispanic or Latino; Unknown
• Gender - Male; Female; Gender X (for all non-binary values); Unknown
• English-language status - English learner; English proficient; EL exited; EL monitored or recently exited; long-term EL; ever EL; English only; never EL; IFEP; recently-arrived and non-recently-arrived ELs
• Economic status - economically disadvantaged; not economically disadvantaged
• Disability status - students with disabilities; students without disabilities
• Migrant status - migrant; non-migrant
• Homeless status - homeless; non-homeless
• Foster-care status - foster care; non-foster care
• Military-connected status - military; non-military
• Gifted & talented status - GT; non-GT
• SLIFE status - SLIFE; non-SLIFE
• High-needs status - high needs; non-high needs

Every state dataset includes an identical set of variables, even where a state has no data for a particular one. NCES variables include district and school NCES IDs, district and school state-assigned IDs, district and school types, indicators for virtual schools and charter districts, county names and IDs, and school-level information. A complete list and description of all variables is in the Codebook.

No. The dataset contains no student-level data or personally identifiable information (PII). All results are aggregated to the state, district, or school level, depending on the state.

The datasets include all U.S. public school districts (including charter schools) that have data available from State Education Agencies for students in grades 3-8 on the general summative assessments.

All state data are available on the Export Data page - select the state and year of interest and export a CSV. The data are also available programmatically through the API.

You can subscribe to EDC data updates to be notified whenever new data are released. You’ll also be offered the option to subscribe when you download a data file.

Yes. The EDC adds data to the repository in June and December each year, reflecting newly released results from State Education Agencies and any additional components received by data request. Planned releases are listed on the Release Calendar. To suggest data you’d like to see, email info@eddatacenter.org.

The SADR includes data from the annual statewide summative assessments. It does not include other assessments states administer, such as alternate assessments or English-language-proficiency assessments. Currently the SADR contains assessments given in English, except for the Texas STAAR-Spanish assessment, Puerto Rico’s standardized assessments, and the Colorado Spanish language arts assessment.

No. Each state defines proficiency according to its own grade-level and subject-area standards and administers its own distinct assessment, so a given proficiency rate does not mean the same thing from one state to the next. Compare results within a state - over time or across its districts and schools - rather than state to state.

Changes that affect comparability over time are described beneath the relevant charts as Notable Events - for example, a new assessment, revised proficiency cut scores, a year the test wasn’t administered (such as field testing), lower-than-usual participation, or data that aren’t currently available. In the raw files, variables flag when a subject’s assessment name or proficiency criteria changed for a given year.

Yes. The SADR includes all available results for all available school years, regardless of assessment changes. Those changes are surfaced two ways: variables in the raw files note when a subject’s assessment name or definition of proficiency changed, and Zelma’s charts include Notable Events describing when a state changed or introduced an assessment - so you can interpret trends accordingly.

Each state uses its own criteria to define subgroups such as race/ethnicity, gender, English-learner status, and economic status, and the SADR reflects each state’s own classifications. This is another reason results are not comparable across states - definitions as well as tests differ.

Cite the Education Data Center as the source, whether you used the raw data or a Zelma result.

Raw data

Education Data Center. (2026). State Assessment Data Repository (Version 3.1) [Data set]. Retrieved Month DD, YYYY, from https://www.eddatacenter.org/data

Example: Education Data Center. (2026). State Assessment Data Repository (Version 3.1) [Data set: Minnesota (2018-2025); Wisconsin (2018-2025)]. Retrieved June 2, 2026, from https://www.eddatacenter.org/data

Zelma query

Education Data Center. (2026). State Assessment Data Repository (Version 3.1) [Data set]. Zelma query: "[Query title]." Retrieved Month DD, YYYY, from [query link]

Example: Education Data Center. (2026). State Assessment Data Repository (Version 3.1) [Data set]. Zelma query: "Math scores by race in New York over time." Retrieved June 2, 2026, from https://www.eddatacenter.org/q/JTS8BN29Y8

Our Terms of Service are available here.

Questions, comments, and corrections can be sent to the EDC team at info@eddatacenter.org. Thank you for your feedback!