Numerical Reasoning Test

Every sum on this page is one division or one multiplication. Nothing needs long division, algebra or a formula you have to remember.

What takes the time is finding the two cells the question is about, then deciding what the bottom of the fraction should be. A percentage is one number divided by a base, and choosing the wrong base is the commonest way to lose a mark here.

So the tables are deliberately small. Three rows and at most three columns, which keeps the looking quick and leaves the deciding as the whole job.

Employers use this format for the same reason. A job that involves numbers usually hands them over already collected, and the work is what you do with them.

The three wrong answers on each question are not padding. Each one is a real figure, produced by one particular slip, and you can usually recognise your own.

One of the three is always what you get by reading the wrong row or column. The other two come from the right cells and the wrong sum.

• Wrong row or column. The figure is real and it answers a question nobody asked.
• Wrong base. Dividing by the other two depots rather than by all three of them.
• Percentage points read as per cent. A rate of 24 against a rate of 12 is 12 points apart and twice as high.
• A scale slip. A month read as a year, or a whole shift read as an hour.

Real sittings allow a calculator and most candidates use one. This page cannot tell whether you did, and it will not pretend otherwise.

So the number is a reading and reasoning score with the arithmetic left open. A calculator moves it less than you would expect, because the arithmetic was never the slow part.

Paper is fine too. The clock on each question is what keeps the pace honest, and 75 seconds is close to what a real sitting allows.

Pace is not the point here, and the site does measure it elsewhere. The figures in how processing speed shifts with age come from tasks where speed is the entire score.

Numerical reasoning is among the most sat tests in the world, and its distributions are private. The publishers own the item sets, and they own the norms that go with them.

A published curve would not transfer anyway. A percentile describes the people who sat those particular items, and these twelve tables were written for this page.

There is real research nearby. Lipkus and colleagues measured how well educated adults handle ratios and percentages, and found them worse at it than anybody expected. Shah and Hoeffner reviewed what actually makes a table or a chart hard to read.

Neither of those hands over a curve for this item set. So the page reports your figure and offers the arithmetic instead. Four answers mean a guess comes off one time in four.

It is not a maths qualification and it is not an intelligence test. Nothing on this page has been checked against either of those.

It is not a forecast of how you would do in a job. Employers use this format widely, and that is a fact about hiring rather than evidence about you.

The pool is twelve tables, so a second run repeats most of what you have already seen. What improves on the second attempt is your memory of these questions.

Speed and accuracy also trade against each other. A run at 45 seconds a question and a run with no clock are two different tests. That is why the card carries your pace beside your score.

Holding a total in your head while you divide by it is working memory doing its job. There is more on that in how working memory holds a number while you use it.