Skip to content

Percentiles, Stanines and Grade Equivalents Explained

Last Updated on 18 November 2026 by Editorial Team

A grade equivalent of 7.2 for a fourth grader does not mean they can do seventh grade work. It means they scored what a typical seventh grader would score on the fourth grade test, which is a different claim entirely.

Every score report mixes at least two of these numbers and explains none of them properly. This page is the reference for what each one actually says, because the differences are not cosmetic and acting on the wrong one leads to real mistakes.

The seven numbers

The seven numbers that appear on school score reports
Number What it answers What it cannot tell you
Raw score How many questions were answered correctly Anything at all, without knowing the test
Scale score Where the student sits on a continuous scale spanning grades How that compares to other children
Percentile rank What share of the norm group scored at or below this student How far apart two students actually are
Stanine The same thing as a percentile, compressed to one digit Any detail within a band
Standard score How many standard deviations from the mean, rescaled Anything, without knowing the mean and SD used
Grade equivalent Which grade’s median matches this raw score What grade level work the student can do
Growth percentile How this student’s progress compares to similar starters Where the student stands overall

The third column is the one worth reading twice. Most misreadings come from asking a number a question it cannot answer, and the commonest example is the grade equivalent.

Grade equivalents, and why they mislead

A grade equivalent of 7.2 means: this student’s raw score matches the median raw score of students in the second month of grade seven, had those students taken this same test. The test was written for fourth graders. It contained no seventh grade content.

So the score says the child answered the fourth grade questions unusually well. It says nothing about whether they could handle seventh grade material, because the test never asked. A fourth grader with a 7.2 in mathematics has very likely never seen algebra.

Test publishers have been cautioning against grade equivalents for decades for this reason, and several have stopped reporting them. If your child’s report carries one, read it as an unusually high score for their grade and nothing more.

Percentiles, and the group they compare against

A percentile rank says what share of the norm group scored at or below your child. The number is meaningless without knowing who the norm group is, and reports often bury that in small print.

The difference is large. ISEE and SSAT percentiles compare a child against other private school applicants, a selective group, so a strong student lands mid scale. National norms on NWEA MAP or i-Ready compare against students generally. The same child can be at the 50th percentile on one and the 80th on the other with no contradiction.

The second trap is that percentiles are not evenly spaced. The gap in actual ability between the 50th and 55th percentile is small, because that is where most children are. The gap between the 94th and 99th is much larger, because the distribution thins out. Percentile differences near the middle mean less than the same difference near the edges.

Stanines, and what they throw away

The theoretical stanine distribution
Stanine Percentile range Band width in percentile points
1 1 to 4 4
2 5 to 11 7
3 12 to 23 12
4 24 to 40 17
5 41 to 60 20
6 61 to 77 17
7 78 to 89 12
8 90 to 96 7
9 97 to 100 4

Derived from the normal curve. Publishers round the edges differently: ERB’s ISEE table gives the bottom stanine three percentile points rather than four, because its percentile ranks run 1 to 99.

Stanines compress a percentile to one digit, which makes a report readable and loses information. A child at the 41st percentile and one at the 60th are both stanine 5, and in a borderline decision that difference matters.

When a report gives both, read the percentile. Stanines are for scanning a class list, not for deciding about an individual child. The ability tests comparison covers where they are used, since the CogAT, NNAT, OLSAT and CCAT all report them.

Standard scores and the 15 or 16 problem

A standard score expresses performance as distance from the mean, rescaled to convenient numbers. A mean of 100 is almost universal. The standard deviation is not: some tests use 15, others use 16, and reports rarely say which.

At one deviation above the mean that is 115 against 116, which nobody notices. At the two deviation mark used by many gifted thresholds it is 130 against 132. Comparing a score from a 15 point scale against a cutoff set on a 16 point one is a real error with real consequences, and it happens.

The CogAT uses 16. The NNAT3 uses 15. If a district sets a single numeric cutoff and accepts either test, those children are not being held to the same standard.

Growth percentiles, the most useful and least noticed

A conditional growth percentile compares a student’s progress against students who started at the same score. It answers a question none of the other numbers do: is this child moving faster or slower than similar children?

That makes it the fairest number on a report. A child at the 30th percentile overall but the 80th for growth is doing well, and reading only the first number would miss it entirely. The reverse is also true and more worrying: a high achiever with low growth is coasting.

MAP reports this prominently and it is still the number parents most often skip. The i-Ready and MAP comparison explains how each test handles growth, and the autumn, winter and spring chart shows the typical term to term change these percentiles are built from.

Which number to read first

For where a child stands: the percentile, having checked which norm group it uses.

For whether they are progressing: the growth percentile, if the test reports one.

For what to do next: neither. The domain or subtest breakdown underneath the headline, which is where the actionable information lives and which almost nobody reads. An overall score is an average, and averages hide exactly the specific weakness that would be worth addressing.

And for a grade equivalent: treat it as decoration. A reading level chart answers the question people are actually asking when they look at one, which is what their child should be reading now.

Screeners are a category of their own and are read differently again. DIBELS and Acadience report against a benchmark that moves three times a year rather than a percentile, so the same raw score can meet the goal in September and miss it in May. A screener result is a routing decision, not a measure of standing.

Common questions

What does a grade equivalent of 7.2 mean for a 4th grader?

That the child scored what a typical seventh grader in their second month would have scored on the fourth grade test. It does not mean the child can do seventh grade work, because the test contained no seventh grade material.

What is the difference between a percentile and a percentage?

A percentage is how much a student got right. A percentile is what share of the comparison group they beat. A child can answer 60 per cent of questions correctly and be at the 90th percentile, if the test was hard.

What is a stanine?

A percentile compressed into a single digit from 1 to 9. Stanine 5 covers the middle, from the 41st to the 60th percentile. It is easy to scan and loses detail, so read the percentile when both are shown.

Is a standard score of 130 always two standard deviations above the mean?

No. It depends on the scale. On a test with a standard deviation of 15 it is exactly two, but on one using 16 the two deviation point is 132. Check which the test uses before comparing to a cutoff.

What is a conditional growth percentile?

A comparison of how much a student grew against students who began at the same score. It is the fairest number on most reports and the one parents most often overlook.

Which number on a score report matters most?

The percentile for standing, the growth percentile for progress, and the domain breakdown for deciding what to do. The headline score on its own is the least useful figure on the page.

Sources

ERB (2021). What to Expect on the ISEE, for the published stanine conversion table.
NWEA (2025). MAP Growth 2025 Norms, for conditional growth percentiles and their interpretation.

Leave a Reply

Your email address will not be published. Required fields are marked *