The typical school is smaller than the school a typical student attends

The median high school in our matched records enrolls 423 students. Weight the same records by their enrollment, and the midpoint rises to 1,395 students. Neither calculation is wrong. They answer different questions.

The first asks what size school sits in the middle of a list of schools. The second asks what size school sits at the midpoint of the reported student population. Large schools appear once in the first list, but represent many more students in the second. That difference matters when a family tries to reconcile a national benchmark with the large campuses around them.

Start with three schools

Imagine schools enrolling 100, 100, and 800 students. The middle school, when ordered by enrollment, has 100 students. Yet 800 of the 1,000 reported students attend the largest school. If you choose an enrollment record at random from this example, an 800-student school is much more likely than a 100-student school.

The school median is therefore 100; the enrollment-weighted median is 800. This is a difference in weighting, not an argument about whether either school is better. Change the three enrollments below to see when the two midpoints converge or separate.

What the national directory shows

We calculated both medians separately by school level, using positive NCES CCD 2024–25 enrollment values matched to our 2025–26 directory. Each row uses exactly the same schools for its two medians. A zero or missing enrollment is excluded, rather than treated as a tiny operating school.

The high-school difference is particularly large. Many small records influence the school median, while large comprehensive campuses account for much more enrollment. The combined/other group is especially heterogeneous: its schools do not share one grade span or instructional model. It should not become a benchmark for an ordinary neighborhood elementary school.

What this does not prove

A larger enrollment-weighted median does not establish that large schools cause better outcomes, that small schools lack opportunities, or that a child will feel anonymous at a large campus. Enrollment is an administrative count. Class groups, advisory systems, buildings, schedules, and student support determine how that scale is experienced.

Nor is this a survey of individual children. We weight school records by published enrollment; we do not possess student-level records or independently deduplicate children across institutions. The analysis includes the school types retained in our directory, including virtual and specialized programs. An administrative school record is not always equivalent to one conventional physical campus.

Choose the benchmark for the question

Use the school median when asking whether a campus is unusually large among schools. Use an enrollment-weighted statistic when asking how the published student population is distributed across school sizes. For a family comparing two campuses, the more useful first step is often to narrow the comparison to the same school level, state, and relevant grade span.

Then ask about the scale your child actually encounters: students in the grade, sections of required courses, typical class sizes, lunch periods, and how students get to know an adult who follows their progress. A 1,400-student high school with small learning communities can feel different from another institution of the same size.

How we calculated the weighted median

We sort each level’s included schools by enrollment, sum those enrollments, and find the first school size at which cumulative enrollment reaches at least half of the total. We use that school’s enrollment, without interpolating between school sizes. The ordinary median uses the middle school, or the mean of the two middle schools when the count is even.

This is a matched-directory analysis, not an official NCES national estimate. Schools that closed, changed identifiers, or failed to match the current directory can be absent. All enrollment weights use 2024–25; older CRDC counts never fill gaps. The tables recalculate from the same validated data as the profiles.

Where the high-school students are concentrated

The two medians become easier to understand when the high-school records are divided into enrollment bands. The table below uses the same positive 2024–25 enrollment values as the high-school row above. One percentage counts schools equally; the other counts their reported students.

In this matched group, schools with fewer than 250 students make up 34.7% of high-school records but only 5.2% of reported high-school enrollment. At the other end, schools with at least 2,000 students account for 7.4% of records and 26.9% of enrollment. This unequal weighting explains why the student midpoint sits much higher than the school midpoint.

A band can contain many school records and a much smaller share of reported students. Conversely, a smaller number of large schools can account for a substantial part of the student population. Neither column is a correction of the other: each supplies the denominator needed for a different question. Rounded percentages may not sum to exactly 100%.

These bands are descriptive choices, not official definitions of small, medium, or large schools. Moving a boundary from 499 to 599 would change the band counts without changing the underlying enrollments. The median comparison does not depend on those bands; we calculate it directly from individual school records.

A weighted median is not the same as a weighted mean

Return to the invented schools of 100, 100, and 800 students. Their ordinary mean enrollment is about 333, their ordinary median is 100, and their enrollment-weighted median is 800. An enrollment-weighted mean would be 660: the sum of 100 squared, 100 squared, and 800 squared, divided by the combined enrollment of 1,000.

That weighted mean answers another question: the average school size attached to a reported student in the example. It gives large schools a large influence because each of their students is associated with the school's full enrollment. The weighted median instead locates the point where cumulative enrollment reaches half the total.

We use medians in the main table to make those two midpoint questions explicit. A median still has limits: it does not describe the full spread and can move abruptly when the cumulative weight crosses a boundary. That is why the band distribution is useful alongside it. No single “typical size” can reconstruct the entire distribution.

The directory date determines which institutions enter the analysis

Our enrollment year and directory year differ. We begin with schools present in the 2025–26 directory and use their matched 2024–25 membership values. This keeps the analysis aligned with the profiles readers can open, but it is not identical to analyzing every school in the original 2024–25 membership universe.

A school that closed before the directory year can be absent. A newly opened school may have no prior-year enrollment. An identifier change can prevent a match even when a community considers the institution a continuation. Those cases can affect both which records remain and how much enrollment is represented.

We exclude missing and zero enrollment values from both medians rather than filling them with historical CRDC counts. Adding older counts would create a mixture of reporting years. Excluding unavailable values is transparent, but it does not establish that excluded schools have the same size distribution as included schools. The counts in the table show the actual scope of each calculation.

School enrollment does not determine the size of a class

A 1,600-student high school might divide a grade across several teaching teams, while a smaller school might have fewer sections of a required subject. Those are possibilities to investigate, not conclusions supplied by enrollment. A school total does not reveal how many students sit in one mathematics class or how consistently an adviser follows an individual student.

Ask for the grade-level enrollment and the current number of sections in the courses your child would take. Find out how the timetable handles full classes, schedule conflicts, and transitions between buildings or programs. For a student entering a large school, ask how the school helps new students establish a reliable adult contact.

For a small school, ask how it sustains course sequences and activities with fewer students, including any partnerships or shared instruction. A small total can support close relationships while also requiring particular scheduling arrangements; a large total can support a broad menu while requiring deliberate coordination. The directory count helps frame those questions but cannot settle the tradeoffs.

Sources and next steps

• NCES Common Core of Data files — directory and membership source files.
• Our methodology and coverage report — joins, missing values, and source years.
• School enrollment benchmarks — another way to compare size by school level.
• Compare individual schools — inspect the actual campuses on your shortlist.

For a local comparison, retain the same denominator discipline used here. Comparing an elementary school with the median for all high schools would answer little about whether its size is unusual. If a campus serves a broad grade span, inspect the actual grades before selecting a peer group. A familiar label can hide very different distributions of students across years.