Why two students with the same number correct get different scores

If you and a friend both got, say, 42 questions right in Reading and Writing and came away with different section scores, nothing went wrong. That is the digital SAT working as designed. Your score is not a count of correct answers. It is a statistical estimate that takes into account which specific questions you answered correctly, and the two of you almost certainly did not answer the same set.

College Board says this in plain language on its scoring page:

Two students who answer the same number of questions correctly in a test section may earn differing section scores based on the characteristics, including difficulty level, of the particular questions they answered correctly.

The Technical Manual puts the same point more bluntly: two students may have the same number of correct answers but have different reported scale scores. Below are the three mechanisms that produce that outcome, in the order they matter.

Mechanism 1: questions are not worth the same amount

The digital SAT is scored with Item Response Theory. Before a question ever counts, College Board has already measured how it behaves: how often students at each ability level answer it correctly, how sharply it separates stronger from weaker performers, and how likely a correct answer is by chance. Those measured properties are attached to the question permanently.

When the model estimates your score, it asks a different question than a raw count does. Instead of “how many did you get right”, it asks “what ability level best explains this exact pattern of right and wrong answers on these exact questions”.

A correct answer on a question that most students miss shifts that estimate more than a correct answer on a question almost everyone gets. Not because College Board decided harder questions deserve more points, but because the harder question carries more information about where you sit on the scale.

This is why the arithmetic you want to do does not work. Missing 4 easy questions and missing 4 hard questions are different events to the model, even though both are “4 wrong”.

What this looks like between two students

Suppose you and your friend each miss 6 Math questions. You missed 6 of the hardest items in the module. Your friend missed a scattered mix, including two items that most test takers answer correctly.

The model reads your pattern as consistent: strong through the easier and mid-range material, running out of room at the top. It reads your friend’s pattern as less consistent. Consistency at the level below your errors is exactly what raises confidence in a higher estimate. That difference can show up as a gap of a few 10-point steps even though the raw counts match.

The size of that gap is not something anyone outside College Board can compute, because the item parameters are not public. Any site that tells you a hard question is worth a specific number of points is inventing the number.

Mechanism 2: the model looks at guessing probability

The same College Board page names a second input:

The scores students receive are a product of several factors, characteristics of the questions they answered right or wrong (e.g., the questions’ difficulty levels), and the probability that the pattern of answers suggests they were guessing.

The 3-parameter logistic model used for multiple-choice questions includes a parameter for the chance of a correct answer from a student well below the question’s difficulty. A student who misses several mid-range questions and then answers a cluster of very hard ones correctly produces a pattern that the model can partly attribute to chance rather than ability.

Two clarifications, because this gets distorted online. There is no penalty for wrong answers, so guessing is never worse than leaving a question blank. And you are not being accused of anything. This is a property of the statistical model, applied to everyone identically, not a cheating flag.

Mechanism 3: you did not take the same test

This is the mechanism students underestimate most. The digital SAT is adaptive within each section. Everyone’s module 1 draws from a broad mix of difficulty, but your performance there routes you to either a lower-difficulty or a higher-difficulty module 2.

What can differ between two students	Why
The second module you received	Routing is based on your own module 1 performance
Which 8 questions were unscored	2 pretest items per module, randomly placed
The specific items in your form	Multiple forms are administered on the same date

So when you compare “42 correct” with a friend, you may be comparing 42 correct out of a harder second module with 42 correct out of an easier one. Those are not the same 42 questions, and the scale accounts for that. Routing happens separately in Reading and Writing and in Math, so a friend can be above you in one section and below you in the other for reasons that have nothing to do with either section’s difficulty.

The adaptive modules guide covers routing in full, including why the score ceilings people quote for the easier module are not published by anyone.

What this does not mean

It does not mean the scale is arbitrary or that scores are unreliable. The whole purpose of calibrating items in advance is that a 1300 means the same thing on every test date and every form. Difficulty differences are absorbed by the model rather than by a curve applied after the fact.

It also does not mean you can improve your score by attempting questions in a clever order or by targeting hard items. Every question in a module is available to you, there is no bonus for the order you answer in, and there is no way to see which items are unscored. The only variable you control is how many questions you answer correctly, and which ones follow from your actual skill level.

What you can compare instead

Number correct is a poor comparison unit across two students. Two things travel better.

Scaled section scores are directly comparable, because that is what the scale is for. Percentiles are comparable in a different way, telling you what share of a reference group scored at or below you. Both are more informative than a raw count, and both are what colleges see. You can look yours up with the percentile calculator or read the percentiles reference to understand which percentile system your report is actually using.

Practice tests behave differently, on purpose

If you are scoring a paper practice test, you will notice you do get a raw-score lookup table, and it returns a range rather than a single score. That range exists precisely because the simplified paper method cannot reproduce what IRT does. College Board says so in each scoring guide: the lower and upper values “establish the range of scores you might expect to receive had this been an actual test”.

Our practice test calculator applies the official procedure and shows the range rather than pretending to a single number. For the full scoring mechanism from the beginning, start with how the digital SAT is scored.