How we measure progress

Where the questions come from

Every question in Scoreberry is written by us. None of them are copied or reworded from the College Board, from a released practice test, or from another prep company's book. They are written to match the real digital SAT in style, length and difficulty - never in wording.

No question reaches a student until it has been answered twice. After it is written, someone comes back to it later with the answer key, the explanation and the reasoning hidden, solves it cold, and rates its difficulty from scratch. Only if that second answer matches the recorded one does the question go live.

When the two answers disagree, the question is pulled rather than quietly corrected. A disagreement usually means the question itself is ambiguous, and the fix is to rewrite it - or to retire it, with a note saying what went wrong.

Where this falls short. Two people can still be wrong in the same way. Every question in the app carries a “Report a problem” link for exactly that reason, and a question that turns out to be ambiguous is rewritten or retired rather than patched.

How we measure a skill

Each skill - transitions, circles, words in context - gets a percentage from 0 to 100. It is not simply the share you got right: two things change how much a single answer counts.

Getting a hard question right says more about you than getting an easy one right, and missing an easy one costs more. That is the first adjustment.

The second adjustment is time. Every 10 answers you go back, an answer counts half as much - so what you did last week matters more than your first week, and the number moves when you improve instead of staying anchored to a bad start.

The percentage itself is then plain arithmetic: the weight of the questions you got right, divided by the weight of all of them, over your 25 most recent answers in that skill. Under 3 answers there is no percentage at all - it says “Not enough data” instead, because a percentage built on one or two questions would look precise and mean nothing.

The word beside a percentage is only a bracket for it: 0 to 49 is Focus here; 50 to 75 is Developing; 76 to 100 is Strong.

A whole area - Algebra, or Craft and Structure - is measured by running the same calculation over every answer in it, rather than by averaging its skills. That way a skill you have answered twice cannot swing an area as hard as one you have answered twenty times.

What the number cannot tell you. It measures how you answer our questions, in our proportions. A skill with 3 answers behind it gets a percentage, but a thin one - the more you practice a skill, the more its number is worth.

How we choose what you practice next

The first thing a new student does is a 20-question check-in: 10 Reading and Writing, 10 Math, untimed, and the same questions for everyone, so taking it again later is a fair comparison. It is deliberately too short to be a score. Its job is to find a starting point.

After that, the home page suggests 2 sets of 10 questions. A skill you have measurably struggled with always comes first. Only when there is no measured weakness left do we point you at an area you have barely touched - not having practiced something is not yet evidence that you are bad at it.

Separately, every question you get wrong outside a full-length test comes back. It returns no sooner than 2 days later - long enough that you are remembering it rather than repeating it - and it leaves the list only when you get it right.

Where this falls short. We can only recommend what we have questions for. If your weakest skill is one our question bank cannot fill a set from yet, we point you at the next thing instead of sending you to an empty session.

How a practice test is scored

A full-length test runs the way the real one does: two modules per section, and the second module depends on the first. Answer at least 60% of module one correctly and you get the harder second module; below that, the easier one.

That choice matters for your score, exactly as it does on the real test. A perfect paper on the easier path tops out at 650 for that section, where the harder path reaches 800.

A test scores 54 Reading and Writing questions and 44 Math questions, and your number of correct answers is looked up in a fixed table - the same table for everyone, written down before you sat the test and never adjusted afterwards. Nothing is graded on a curve against other students.

About those tables. Our scoring tables are a straight-line approximation, not the College Board's. They are built to be realistic and to move the way a real curve moves - a few points either way is normal, and the closer a score sits to the middle of the scale, the closer the two agree.

The score estimate

Your progress page shows an estimated score built from ordinary practice, so you do not have to sit a two-hour test to know roughly where you stand. It is put together like this.

Each area is weighted the way the test itself weights it. Information and ideas is 7 of the 27 questions in a Reading and Writing module, so it counts for that much of the estimate and no more.

Those weights are applied to your area percentages to get one expected share correct for the section. That share becomes a number of correct answers, is routed onto the easier or harder second module by the same 60% rule a real test uses, and is looked up in the same table a real test is scored with. There is no second formula for the estimate.

The answer is shown as a range 80 points wide, never a single number. A section is only estimated once the areas that have enough answers behind them cover at least 50% of the test, and areas you have not practiced are left out of the weighting rather than counted as zeros.

What an estimate is not. This is an estimate from practice, not a score. It assumes you answer the second module about as well as the first, which slightly understates the easier path - so if it is wrong, it is more likely to be low than high. For a real number, sit a full-length test.

What none of these numbers can tell you

• None of this predicts your SAT score. It describes how you have answered our questions, which is the best evidence we have and is not the same thing.
• Practice at home is not test day. Nothing here measures nerves, a short night's sleep, or an unfamiliar room.
• A percentage moves slowly on purpose. After a good session, expect the number to acknowledge it over a few days rather than jump.
• Every number is worked out again from your answers each time a page loads. We keep no separate record of your ability that could quietly drift away from what you have actually done.