How to prepare

Numerical reasoning practice > How to prepare

Preparation for this test has a low ceiling and a low floor. A few hours of the right practice makes a real difference; twenty hours of the wrong practice makes almost none.

1. The problem with fixed question banks

Most free practice material is a fixed set of questions. The first time through, it is useful. The third time through, you are recalling answers rather than working them out, and your score stops meaning anything.

You can tell this is happening when your times improve but your accuracy on new material does not.

The practice on this site generates the numbers for each question, so the same question is very unlikely to appear twice. That is the one thing it is built to do.

2. Practise pacing, not just accuracy

Untimed practice trains the wrong thing. Almost everyone can answer these questions given three minutes; the test does not give you three minutes.

• Always practise with the timer running
• Decide in advance what you do when a question is not coming — most people should move on at about the 45-second mark
• Leaving one question to answer two is usually the right trade

The single most common way to lose marks is spending four minutes on a hard question and running out of time on three easy ones.

3. The calculator question

Whether a calculator is allowed depends on the test, and you should check the instructions rather than assume.

Either way, practise the estimation habits below. They are faster than a calculator for most of these questions:

• Simplify before dividing — 630 ÷ 180 is easier as 7 ÷ 2
• Use multipliers for percentage change — × 1.3 rather than "find 30%, then add"
• Look at how far apart the options are before deciding whether you need an exact figure

4. Find the question type you are weak at

Scores are rarely uniformly low. Usually one question type is doing the damage, and it is often percentages.

Work through the six types separately rather than doing mixed tests only:

• Data tables — finding the right cell
• Charts — reading values off an axis
• Percentage change — the biggest source of lost marks
• Ratios and shares — which value divides which
• Per unit and rates — which way round to divide
• Forecasts — rate versus amount

5. How long to spend

There is no evidence-based number here, and anyone who gives you one precisely is guessing. What is observable is the shape of the returns:

• The first session or two removes most of the format shock
• A few more sessions fix the question types you are weak at
• Beyond that, improvement is slow, because the underlying skill is not what is changing

If you are short of time, do timed practice on your weakest type rather than more mixed tests.