A school dashboard can show “92% graduation,” “41% math proficiency,” “average growth,” and a “7/10” rating on the same screen.

Those numbers do not share a denominator, time window or purpose. One may describe a graduating cohort, another tested students in selected grades, another a state model and another a private formula whose weights change when data are missing.

Before deciding whether a number is good or bad, decode what it actually counts.

Start with the official state report card

The Every Student Succeeds Act requires state report cards with information that includes academic results and other indicators. The U.S. Department of Education maintains a directory linking each state's report-card site. Use it as the starting point for a public school.

Also gather:

  • the district and school profile;
  • the definitions, data notes or methodology;
  • the assessment and accountability year;
  • the Civil Rights Data Collection profile for opportunity and discipline fields;
  • the school's current course catalog, handbook and improvement plan where available.

Third-party sites can make discovery easier. They are not substitutes for the source year, definitions and missing-data rules. Private schools are not subject to the same uniform public reporting, so a comparison may contain genuinely unequal evidence.

Decode every metric with eight fields

Copy this note before interpreting a chart:

Metric label:
School year / release date:
Numerator (who or what counted in the result):
Denominator (eligible cohort/population):
Definition / threshold:
Level (school, program, district, state):
Comparison group:
Missing, suppressed, rounded, estimated, or revised data:

Add the source URL and access date. If any of the eight fields is unclear, the number is a question—not yet evidence for a ranking.

Example: “90%” can mean several things

It might be:

  • 90% of tested students met a state proficiency threshold;
  • 90% of a ninth-grade cohort earned a regular diploma within four years after permitted cohort adjustments;
  • 90% of surveyed respondents chose a positive response;
  • 90% of enrolled students attended on a particular day;
  • 90% of graduates, rather than all students, completed an advanced-course requirement.

The percentage sign does not make these comparable.

Proficiency: who crossed a state threshold?

Proficiency usually reports the share of tested students at or above a state-defined achievement level for a particular assessment, subject and grade.

Ask:

  • Which assessment and performance standard?
  • Which grades or end-of-course tests?
  • How many students were expected to test and how many participated?
  • Is the result schoolwide or for one grade, subject or student group?
  • Did the assessment or cut score change?
  • Are students in the compared schools taking the same test?

A high proficiency rate can reflect strong instruction, students' prior preparation, selection, mobility, access to resources and other factors. It does not isolate what the school caused.

Do not directly compare proficiency rates from two states. States use different assessments and thresholds. Begin with same-state, same-year, same-subject comparisons and read each state's notes.

Growth: how did performance change for students?

Growth measures try to describe change over time, often by comparing a student's progress with academically similar students under a state model. The exact model matters.

Growth and proficiency can tell different stories:

  • high proficiency, low growth: many students meet the threshold, but current progress may be slower than the model expects;
  • low proficiency, high growth: many remain below the threshold, but students may be gaining relatively quickly;
  • both high: strong current attainment and progress under the reported measures;
  • both low: a serious signal that needs subgroup, trend and context review.

At high school, growth data may be limited because statewide tests occur in fewer grades or courses. Do not let a dashboard silently replace missing growth with a different measure without noticing.

When a third-party rating uses “progress,” read its methodology. GreatSchools, for example, explains that state data availability changes which progress measure and weights can be used. That transparency is useful; the composite still remains a synthesis, not a fact about fit.

Graduation: which cohort finished, when, and with what credential?

NCES defines the four-year adjusted cohort graduation rate for public schools using first-time ninth graders, adding permitted transfers in and subtracting qualifying transfers out, emigrations and deaths. The numerator is students earning a qualifying diploma within four years.

Check:

  • whether the metric is the four-year adjusted cohort graduation rate;
  • whether extended five- or six-year rates are also shown;
  • which diplomas count under the reporting definition;
  • whether the display refers to all cohort students, graduates only or seniors;
  • subgroup results and privacy suppression;
  • cohort size and at least a three-year trend.

“Percent of graduates who enroll in college” has a denominator of graduates and may exclude students who did not finish. It cannot replace the graduation rate. “Completion” may include credentials that the federal adjusted-cohort metric treats differently. Read the label.

A 95% school rate does not mean a particular student has a 95% personal probability of graduating. Aggregate data describe a cohort, not an individual forecast.

Attendance: chronic absence is not daily attendance

The U.S. Department of Education commonly defines chronic absenteeism as missing 10% or more of school days for any reason, excused or unexcused. A chronic-absence rate is therefore the share of students crossing that threshold.

It is not the same as:

  • average daily attendance;
  • truancy, which depends on unexcused absence and local rules;
  • number of absence events;
  • a single student's days missed.

A high rate is a meaningful warning, but it does not reveal the cause. Transportation, health, housing instability, disengagement, caregiving, work, school climate and data practices can all matter. Ask what changed, which groups are affected and what response the school has tested.

If School A has 94% average daily attendance and School B has 18% chronic absence, you cannot subtract or rank those figures as though they were the same metric.

Discipline and safety: counts need policy context

Data may report students suspended, suspension instances, days lost, referrals, expulsions, arrests or law-enforcement contact. Those units are not interchangeable.

Decode:

  • students versus incidents;
  • in-school versus out-of-school exclusion;
  • unduplicated students versus repeated actions;
  • rate per enrollment versus raw count;
  • schoolwide versus subgroup;
  • policy or reporting changes across years.

A lower reported count does not by itself prove a safer or fairer climate. Reporting access, definitions and school responses influence the number. A higher suspension rate can reflect more incidents, a more exclusionary response to similar behavior, or both. Bring the pattern to the school as a precise question.

The Civil Rights Data Collection is valuable for examining access and disparities, but it is periodic and can lag current practice. Confirm the collection year.

Advanced courses: offered is not accessible

A profile may say a school offers AP, IB, dual enrollment, Algebra I, calculus, computer science or career/technical education. For a student's path, ask five separate questions:

  1. Is the course in the current catalog?
  2. Did it actually run in the most recent year?
  3. What prerequisites or recommendations apply?
  4. How many eligible students requested it and received seats?
  5. Which student groups participate and complete it successfully?

Participation among graduates is different from participation among all enrolled students. Exam participation is different from passing. Course availability is different from completing a sequence.

Use administrative data to identify the access question, then use current school documents and tour questions to resolve it.

Teacher and staffing data: read the exact unit

Report cards may include inexperienced teachers, emergency or provisional credentials, or teachers working outside their licensed field. A separate source may show student–teacher ratio.

These do not automatically reveal:

  • the size of the student's classes;
  • turnover in the desired department;
  • vacancies at the start of the new year;
  • whether one teacher covers several course sections;
  • the quality of instruction or feedback.

Ask whether the metric counts people or full-time-equivalent positions, which year it covers and whether it is schoolwide. Then ask the school a current, department-specific question without requesting private personnel information.

Subgroups: the average can hide the relevant experience

Inspect results for student groups required or available in the state system, such as disability status, English learner status, economic disadvantage, race/ethnicity, homelessness, foster care or military-connected status where reported.

Do not use subgroup data to stereotype an individual student. Use it to ask whether opportunity and outcomes are consistent and what the school is doing about a persistent difference.

Compare:

  • the same subgroup over time;
  • subgroup with schoolwide result;
  • subgroup with same-state reference where meaningful;
  • access measures alongside outcome measures.

If a group has high graduation but low advanced-course participation, the combined question may be more useful than either metric alone.

Missing and suppressed data are not scores

Small-group results may be hidden or altered to protect student privacy. NCES guidance explains why aggregate reporting can still risk identification and why protection creates some loss of detail.

You may see:

  • a dash or blank;
  • N/A or “not reported”;
  • a range such as <5%;
  • top- or bottom-coded values;
  • rounded results;
  • “too few students”;
  • “not applicable.”

Do not convert any of these to zero, 100 or the school average. Find the legend and data notes. Ask whether the value is unavailable, not applicable, not yet released, invalid under reporting rules or privacy-suppressed.

Small groups also create volatility. A change involving a few students can move a percentage sharply. Look for counts or ranges when safely published and avoid dramatic stories from one year.

Put three or more years in a table, but first mark changes in:

  • test or achievement standard;
  • accountability formula;
  • course or grade configuration;
  • school boundary or admissions policy;
  • denominator definition;
  • reporting interruptions or exceptional conditions;
  • data revision.

A trend break may reflect a changed measure rather than changed performance. Compare like with like. Keep release date separate from school year; “2026 dashboard” can contain older school-year data.

Composite ratings: unpack before using

A letter grade, star count or 1–10 score compresses choices about:

  • included measures;
  • weights;
  • missing-data treatment;
  • reference group;
  • rounding;
  • update cycle.

Read the methodology and write the components into your notes. If the score compares schools within a state, do not compare a 7 in one state with a 7 in another. If weights rebalance when a component is missing, two schools' identical totals may summarize different evidence.

Use a composite to find questions, never as the only row in a high-school comparison scorecard.

What performance data cannot tell you

Aggregate dashboards cannot establish:

  • whether this student's exact course sequence will fit;
  • whether an advertised class will run next year;
  • how a particular teacher will teach;
  • whether the student will feel known or ask for help;
  • whether services in an individual plan will be appropriate;
  • whether the commute and schedule are sustainable;
  • what caused a difference between schools;
  • what will happen to an individual student.

They also cannot choose the family's priorities. Choosing a high school remains a fit, feasibility and evidence decision.

Create a one-page data memo

For each finalist, complete:

School + program / entry year:
Official state profile URL / data year:

Proficiency: definition, grades/subjects, participation, subgroups, trend
Growth: model, availability, subjects/grades, trend
Graduation: cohort definition, 4-year/extended, subgroups, trend
Attendance: chronic-absence definition, subgroups, trend
Discipline: students/incidents/days, rate denominator, subgroups, trend
Course access: offered / ran / eligibility / seats / participation / completion
Staffing: exact metric and year

Missing or suppressed data:
Definition or system changes:
Three strongest signals:
Three things the data cannot answer:
Questions / owner / due date:

Ask better follow-up questions

Replace “Why are your scores low?” with:

  • “The state profile shows ninth-grade math growth improving over three comparable years while proficiency remains below the state reference. How does the school interpret both measures, and what has changed in instruction?”
  • “The reported chronic-absence rate differs by student group. Which barriers did the school identify, what response was implemented and what interim measure shows whether it is working?”
  • “The catalog lists computer science, but the opportunity data are older. Did the sequence run last year, what prerequisites apply and how many ninth graders could enter?”
  • “This subgroup result is suppressed. Which public, privacy-safe evidence can help us understand access and support without requesting individual data?”

The goal is not to become an accountability statistician. It is to prevent a label from outrunning its evidence—and to leave every dashboard with a shorter, sharper list of questions for the actual school and student.

Sources

  1. U.S. Department of Education — State Report Card Directory (2026-07-23)
  2. U.S. Department of Education — Parent Guide to State and Local Report Cards (2026-07-23)
  3. Civil Rights Data Collection (2026-07-23)
  4. NCES — High School Graduation Rates (2026-07-23)
  5. NCES — Statistical Methods for Protecting PII in Aggregate Reporting (2026-07-23)

Frequently asked questions

What does a school proficiency rate mean?

It is usually the percentage of tested students whose scores met or exceeded a state-defined performance threshold in a subject and grade. Check the tested population, participation rate, assessment year, state standard, and student groups. Proficiency is not the same as individual growth and is not directly comparable across states.

Is a high school graduation rate a good measure of school quality?

A four-year adjusted cohort graduation rate is an important outcome, but it reflects one cohort and depends on who enters, transfers, and earns a qualifying diploma within four years. Inspect the definition, subgroups, longer-term trends, course access, and other outcomes; do not use it alone to predict an individual student's experience.

Why is some school data missing or shown as a range?

Small-group results may be suppressed, rounded, top-coded, bottom-coded, or displayed as ranges to protect student privacy or because reporting standards were not met. Missing does not mean zero or perfect. Read the state's data notes and ask whether the metric was not applicable, unavailable, delayed, or privacy-protected.

Keep reading