Why your arithmetic score will not match another site's

This one has the cleanest answer of the six. “Good at maths” is two different abilities that happen to share a name — and the reference kit we work from separates them, with different marker tests for each.

Version 1.0 · Published 7 September 2026 · Last reviewed 7 September 2026 · About the arithmetic test

1. Two factors, and this is the whole argument

Canonical reference · read at source

The kit separates them like this:

The third RG test is worth the name it was given. Necessary Arithmetic Operations asks you to say which calculation would answer a question, and does not ask you to do it. That is a whole test built to strip the arithmetic out of the reasoning, which is about as clear a statement as a reference kit can make that the two are different things.

So if one site called you quick and another called you weak, the most likely explanation is not that either was wrong. It is that they were measuring different factors and both reported the result as “maths”.

2. A rate has no natural maximum, so somebody chose one

Ours is a count in a fixed ninety-second window. There is no score at which the task says “that is everything” — you could always do one more.

So the top of our range is a declared number rather than a measured one: thirty correct, which is one every three seconds sustained for the whole window. It is not derived from anyone's performance, and it is marked as a decision on the test page. A site that gives you a percentage or a rating out of ten has made the same kind of choice — the difference is usually whether they tell you.

The window length matters just as much. Ninety seconds and three minutes are not the same test: a longer window measures more sustained concentration and less initial speed, and it lets a slow starter catch up. Neither is more correct. They are not comparable.

3. Typing is part of our score, and multiple choice would be worse

You type every answer here, which means some part of your score is your hands and your keyboard. That is a real cost and we would rather name it than bury it.

The alternative is worse for a reason worth understanding. A multiple-choice format removes the typing — and puts a floor under guessing. With four options, a quarter of your answers are right without any arithmetic at all, and the score stops being a count of what you can do.

We took the typing over the guessing, deliberately. It is why this is the only test here where chance is exactly zero, and why every correct answer on it is a correct answer.

4. Which sums, and how hard

Our multiplications use numbers from 2 to 12; our additions use numbers up to 49. Change either and the same person's score moves without anything about them changing. A test built on two-digit multiplication is a different test, and a test that stays under ten is a different test again.

Ours publishes every range in a table, along with the rule that every answer is a whole number and never negative — subtractions are built larger-minus-smaller, divisions are built from a multiplication so they come out exact. A test that does not tell you its ranges has not given you enough to interpret your own score.

5. Practice moves this more than it moves most things

This is the most practisable thing we measure, and that is worth saying plainly. Someone who does mental arithmetic daily will score higher than someone who has not since school — and that is a fact about practice, not about capacity.

It also means a second attempt tends to be higher for reasons that have nothing to do with you getting faster: you have got used to the rhythm of typing and pressing Enter. Our result page names that when it compares two runs, and refuses to call the difference meaningful.

What would have to be true for two scores to be comparable

• The same factor. Calculation or reasoning — this is the one most often got wrong, and it is not a fine distinction.
• The same window, or both untimed.
• The same response format. Typed answers and multiple choice have different floors — one is zero and one is not.
• The same number ranges, or at least published ones.
• The same amount of recent practice in the person, which is not a property of the test at all and moves this score more than it moves most.
• Both norm-referenced against a described sample, if either is going to say where you sit relative to other people. Ours is not, and says nothing.

So which number should you believe?

Neither, as a statement about whether you are good at maths. Ours is a count of sums completed correctly in ninety seconds, typed, on one screen, on one day. That is all it claims and all it can claim — our reference sample is 0 people.

The comparison that survives every objection on this page is the per-operation breakdown: not the total, but whether division cost you time that multiplication did not. The test reports that for you and stores nothing to do it.

Sources

• Educational Testing Service. Kit of Factor-Referenced Cognitive Tests (1976 Edition). ets.org — the two factors reproduced in the table above with their marker tests, read at source, including Necessary Arithmetic Operations under RG.