SAT scoring explained: adaptive modules, IRT and percentiles

The numbers

Pretest questions look identical to scored ones and can appear anywhere in a module. Don’t try to spot them.

How the adaptive route works

Each section has two modules. Module 1 is a broad mix of easy, medium and hard questions, and it is the same kind of module for everyone. Your performance on it routes you to one of two versions of module 2:

• Harder module 2: more medium and hard questions. The whole 200–800 range is available.
• Easier module 2: more easy and medium questions. By most published estimates, section scores are effectively capped around the low 600s, however well you do.

The routing decision is made automatically between modules. Bluebook does not tell you which route you are on, and your feelings are an unreliable guide: a well-prepared student on the harder route will find module 2 hard. That is the intended experience.

IRT in plain English

College Board scores each section with an IRT model. Every question has been pretested on real students, so the model knows roughly:

• how difficult the question is,
• how sharply it separates stronger from weaker students, and
• how likely a guess is to land on the right answer.

The model then finds the ability level that best explains your pattern of right and wrong answers and converts it to the 200–800 scale. College Board does not publish item parameters or a raw-to-scaled table, so exact conversions are not public. Four consequences follow:

1. Equal raw scores can give different scaled scores. Route and question difficulty both matter.
2. Easy misses tend to hurt more than hard misses. Missing an easy question while getting harder ones right is unusual, and the model treats it as strong evidence against you.
3. Guessing never hurts. A wrong answer and a blank both earn nothing, so answer everything.
4. There is no “hard question bonus” to chase. You can’t choose which questions to see. The strategy is simply to maximize correct answers, starting with the ones you should never miss.

What each miss costs near the top

On the harder route, these approximate conversions come from published Bluebook practice-test outcomes. They are the same curves Courselo uses for full mocks. Real forms vary by about 10–20 points either way.

So 1550 ≈ about 2 misses per section, 1500 ≈ 3–4 per section, and 1400 ≈ 6–8 per section. Near the top, each question is worth 10–20 points. This is why careless errors are so expensive.

Section score bands

Percentiles

Your report shows two percentiles. The SAT User percentile compares you with recent SAT test takers. The Nationally Representative Sample percentile compares you with all US students in your grade, including those who don’t take the SAT, so it runs higher. The table uses approximate SAT User percentiles, which shift slightly each year.

College Board’s college-readiness benchmarks are 480 in Reading and Writing and 530 in Math.

Reading your score report

• Total and section scores, plus a score range for each section of about ±30–40 points. The range is the measurement error of a single sitting: your score would likely land within it on another day. A 1520 and a 1550 are closer than they look.
• Knowledge and skills: a bar for each of the eight content domains (four per section), showing the share of the section each domain represents and how you did. Use it to choose what to review before a retake, together with your Courselo error log.
• Percentiles and benchmarks, as above.

Scores for SAT Weekend tests usually appear 2–4 weeks after test day. For fall 2026 tests the schedule is about 13 days.

Superscoring, Score Choice and sending scores

• Superscoring. Many colleges combine your highest Reading and Writing score and your highest Math score across different test dates. For example, October 760 + 740 = 1500 and December 720 + 790 = 1510 give a superscore of 760 + 790 = 1550. When your target colleges superscore, you can focus your retake preparation on the weaker section. You still sit the whole test.
• Score Choice. In most cases you can choose which test dates to send. Some colleges require all scores, so check each college’s policy.
• Free score reports. You can usually send a small number of free score reports (traditionally four) when you register and for a short window after the test. Check your College Board account for the current allowance and deadline.
• Canceling. You can cancel a sitting’s scores only for a short time after the test (traditionally until late on the Thursday after test day). Because Score Choice exists, canceling rarely makes sense.

How Courselo scores your practice

Courselo full mocks are fixed forms equivalent to the harder route, and every question counts. Your fraction correct in each section maps to 200–800 through the curves above, and the total is the sum. The diagnostic uses the same curves over fewer questions, so treat its result as an estimate within about ±40 points per section. For a final calibration, take College Board’s official Bluebook practice tests in your last weeks.

Quick recap

• Total = R&W (200–800) + Math (200–800). 90 of the 98 questions are scored.
• Module 1 routes you. Only the harder module 2 opens scores above the low 600s per section.
• IRT weighs question difficulty. Easy misses hurt most, and guessing never hurts.
• 1550 ≈ 2 misses per section; each miss near the top costs 10–20 points.
• Superscoring and Score Choice make a planned second sitting low-risk.