Aptitude test practice pays when it is done one question type at a time. A timed reasoning test, which recruiters also call a psychometric test, draws on a small stock of question types. Each has a method, and a wrong turn that one of the options is waiting to catch.
This guide works one question of each of twelve types, numerical, verbal and logical, with every step shown, then sets a pace in seconds, a rule for guessing, a way to review and a one-week plan. For what each kind of employer test is for, and for test day, begin with how to prepare for a pre-employment test.
EdProcto publishes this guide, and its page for candidates opens practice sets for most of these types. No method here promises a pass or a job, and a practice score is never an IQ.
How do you practise numerical reasoning questions?
Practise numerical reasoning by reading the table's headings and units before the question, then writing each calculation as one line before you work it out. The first three questions use this table.
| Office | Staff | Sales last year (£000) | Sales this year (£000) |
|---|---|---|---|
| Leeds | 12 | 400 | 432 |
| Cardiff | 8 | 250 | 210 |
| Glasgow | 10 | 350 | 318 |
Percentage change from a table
By what percentage did Cardiff's sales change? New minus old: 210 − 250 = −40. Over old: −40 ÷ 250 = −0.16, a fall of 16%. The wrong turn is dividing by this year's 210, which gives 19%. Say "new minus old, over old" on every question until it is a habit, and keep the sign.
Share of a total
What share of this year's sales came from Leeds? Add the right column first: 432 + 210 + 318 = 960. Then 432 ÷ 960 = 0.45, so 45%. The wrong turn is a total from the wrong column: last year's 1,000 gives 43.2%. Put a finger on the column heading before you add.
Ratio
A training budget of £7,500 is shared between the offices in the ratio of their staff. How much does Cardiff receive? The ratio is 12:8:10, which makes 30 parts. One part is 7,500 ÷ 30 = £250, so Cardiff's 8 parts come to £2,000. The wrong turn is dividing by the three offices, which gives £2,500. Check by adding the shares: 3,000 + 2,000 + 2,500 = 7,500.
Rate and unit conversion
A van travels at 54 km/h. How many metres does it cover in 20 seconds? Put both figures into the same units before multiplying. 54 km/h is 54,000 metres in 3,600 seconds, or 15 metres a second. In 20 seconds that is 15 × 20 = 300 metres. The wrong turn is 54 × 20 = 1,080, which multiplies kilometres an hour by seconds. Write the unit beside every number and the mismatch shows.
EdProcto's Numerical reasoning set, one of the kits, asks for the percentage change, the share and the total from a sales table. The figures are new at each opening and the answer is typed, in the maths expression shape.
How do you practise verbal reasoning questions?
Practise verbal reasoning by answering from the words on the screen and nothing else.
True, false or cannot say
The depot opens at 6 am on weekdays. Every delivery booked before noon leaves the same day. Since March, drivers have used handheld scanners in place of paper dockets.
- A delivery booked at 11 am on Tuesday leaves on Tuesday. True: 11 am is before noon, and the rule covers every such delivery.
- Drivers still use paper dockets. False: the scanners replaced them.
- A delivery booked at 2 pm leaves the next day. Cannot say: the passage has a rule for bookings before noon and none for later ones.
The wrong turn is answering "false" where the passage is silent, or bringing in what you know about depots. "False" needs a sentence that contradicts the statement. Underline the sentence that settles each answer. If there is none, the answer is "cannot say".
Analogy
Thermostat is to temperature as dimmer is to: bulb, brightness, switch or electricity? Turn the first pair into a sentence: a thermostat controls temperature. Run that sentence on the second pair: a dimmer controls brightness. The wrong turn is a word that merely goes with dimmer, such as bulb or switch. Write the linking sentence before you look at the options.
Odd one out
Increase, expand, enlarge, reduce, extend. Find the description that fits the most words. Four mean "make larger"; "reduce" means the opposite, so it is the odd one out. The wrong turn is hunting for a clever rule before reading the meanings. Name the group in three words or fewer, and the stray word is plain.
How do you practise logical reasoning questions?
Practise logical reasoning by putting the structure on paper: gaps under a series, letters in place of the nouns in an argument, a row of boxes for a seating puzzle.
Number series
2, 6, 5, 15, 14, 42, and then? The gaps are +4, −1, +10, −1 and +28, which follow no single step. When gaps jump about, test for two operations taking turns: ×3, −1, ×3, −1, ×3. The next move is −1, so the answer is 41. The wrong turn is staring at the gaps alone. Work through a fixed list of checks: a constant gap, a growing gap, a multiplier, two rules in turn, two series interleaved.
Letter series
AZ, CX, EV, GT, and then? Turn letters into positions, A as 1 and Z as 26. The first letters run 1, 3, 5, 7, so the next is 9, I. The second run 26, 24, 22, 20, so the next is 18, R. The answer is IR. The wrong turn is counting the alphabet in your head and slipping by one. If paper is allowed, write A to Z with its numbers before the clock starts.
Syllogism
All accountants are careful with figures. Some people who are careful with figures are auditors. Therefore some accountants are auditors. Does the conclusion follow? It does not. Swap the nouns for letters: all A are B, some B are C, so some A are C. The B that are C may all sit outside A, so nothing forces the conclusion.
This wrong turn has a name, belief bias. In three experiments, Evans et al. (1983) found that people's beliefs about a conclusion swayed their verdict on the logic, and that the bias was more marked on invalid arguments than on valid ones. Here the conclusion sounds true of real offices, and the argument is still invalid. Use letters or circles every time, so the content has no vote.
Seating and ordering
Asha, Ben, Chloe, Dev and Emma sit in a row at desks 1 to 5, numbered from the left. Dev is immediately to the right of Emma. Ben is at an end. Exactly two people sit between Emma and Ben. Chloe is somewhere to the right of Dev. Who is at desk 3?
Start with the tightest clues. Ben is at desk 1 or desk 5, so Emma, three desks away, is at 4 or 2. If Emma is at 4, Dev is at 5 and Chloe has no desk to his right, so that case fails. Ben is therefore at 5, Emma at 2 and Dev at 3. Chloe must take 4, which leaves desk 1 for Asha. The answer is Dev. Of the 120 possible orders, this is the only one that fits all four clues.
The wrong turn is starting from the loosest clue, "somewhere to the right", or reading it as "immediately to the right". Draw the boxes and follow each case to its end.
How do you practise data checking under time?
Practise data checking by comparing in chunks of three or four characters, left to right, against a clock.
KT-48217-M36 and KT-48271-M36: identical or different? Chunk them: KT, then 482, then 17 against 71. They differ, because the seventh and eighth characters have changed places. The wrong turn is reading each record as one shape, which lets a swap of neighbouring digits through.
Checking is largely a test of speed, and speed responds to repetition. Pooling 36 studies of mental speed tasks with 21,810 people, Scharfen, Blum and Holling (2018) found that response times fell with each sitting, up to four, without reaching a plateau, and fell more on complex tasks than on simple ones. EdProcto's Data checking set gives 20 pairs of records at eight seconds each, 160 seconds in all.
How many seconds should you spend on each question?
Divide the time by the number of questions, after holding some time back. Take a test of 30 questions in 24 minutes. Flat, that is 48 seconds each. Hold back three minutes and the pace becomes 21 minutes ÷ 30 = 42 seconds. That sets the checkpoints: 10 questions done by 7 minutes, 20 by 14, all 30 by 21, and three minutes for the ones you skipped. Do the same sum with the figures in your own invitation.
The clock is part of the difficulty. Kellogg et al. (1999) gave thirty people arithmetic problems both timed and untimed, and timing lowered performance in both the high and the low maths-anxiety groups. Practise at your pace figure, not only at leisure.
When should you guess?
Guess whenever a wrong answer costs nothing; where it costs marks, guess once you can rule out an option. With four options, a blind guess is worth a quarter of a mark on average. If each wrong answer loses a third of a mark, the blind guess is worth nothing (1/4 − 3/4 × 1/3 = 0). Rule out one option and it is worth 1/9 of a mark; rule out two and it is worth a third. The instruction screen says which rule applies.
If time is left, go back. Kruger et al. (2005) report that in nearly all the research, most changed answers go from wrong to right and people who change answers usually gain. Change an answer when you find a reason.
How should you review a practice set?
Review a practice set by type, not by score. The total says how the set went, and part of it is luck: Burton (2001) measured how far chance moves multiple-choice scores and showed that the effect depends on the length of the test and the number of options for each question. A tally of misses by type says what to do tomorrow.
Arendasy and Sommer (2017) had 960 test takers sit four ability tests twice, about a month apart, and the gains at the second sitting fitted a model of learning during the test itself. On that reading, what carries over is how a kind of question works.
- Tally the misses by type. Percentage change, share, ratio, series and the rest.
- Name each miss. A slip, a method you did not know, or the clock.
- Write the rule in a few words. "Total of this year's column." "Letters into numbers."
- Redo the misses on new questions. In EdProcto's kits, study mode shows the answer and the reason at once and lets you retry only what you missed.
- Keep an error log. One line per miss, as in the guide on what to do with the questions you got wrong.
What does a week of aptitude test practice look like?
A week is enough for three timed mixed sets and three days on your weakest types: 190 questions in all.
| Day | Work | Questions |
|---|---|---|
| 1 | Mixed set, timed, cold. Tally the misses by type | 30 |
| 2 | Weakest type: three short sets in study mode, in EdProcto's reasoning kits | 30 |
| 3 | Second weakest type: three short sets | 30 |
| 4 | Mixed set, timed. Compare the tally with day 1 | 30 |
| 5 | Types still missed: three short sets at your pace figure | 30 |
| 6 | Mixed set, timed | 30 |
| 7 | Ten questions from the error log, then stop | 10 |
At 42 seconds a question, 190 questions are 133 minutes of answering. Allow as long again for reading the working, and the week costs about four and a half hours. If your test includes data checking, add 20 pairs on days 2 to 6: 100 pairs, a little over 13 minutes at eight seconds each.
Three mixed sets is deliberate. Scharfen, Peters and Holling (2018) pooled 122 studies of repeated ability tests, 153,185 people in all, and found gains from retesting but none after the third sitting. That is a finding about formal sittings, not home practice, yet it argues against a plan made of one mock after another.
How does EdProcto help you practise these question types?
EdProcto is the best place to practise the types that can be made by rule, because those sets arrive with new figures every time they open, and all that is left to learn is the method. Not every set works that way, and this list says which.
- Made by rule, new at each opening. Numerical reasoning (12 questions), Number series (12, in six kinds), Letter series (10), Deduction: who sits where (8 puzzles of four people and three clues), Coding and decoding (10), Number matrices (10), Mechanical reasoning (10), Reasoning word problems (10), Engineering basics (12) and Data checking (20 pairs).
- Four forms, renamed. Syllogisms (10) uses four argument forms, two valid and two not, with new groups each time.
- From a fixed bank. Verbal analogies (12, from sixty pairs), Odd one out (12) and Situational judgement (8, from thirty situations) draw on curated banks, so a question can come round again. For the situational kind, see the guide on best and worst questions.
The battery has no passage with true, false or cannot say. For that type, paste a report into the quiz maker on the home page and Procto AI writes questions from the text. Ratios and speeds have rule-made sets among the school maths kits.
Most of the battery is multiple choice, and every question shows its working afterwards. The product's own words for them are "reasoning tests in the style of the standard aptitude batteries", with no IQ given and no norm claimed. The hiring page shows the same battery from the employer's side.
Every kit opens on the Free plan, which also has Procto AI write 20 questions a month from material you paste. Learner takes the cap off self-tests and lifts that allowance to 100. The pricing page sets out both plans.
Quick questions
How many practice tests should I take before an aptitude test?
Three timed mixed sets in a week, with short sets on your weakest types between them. Scharfen, Peters and Holling (2018) found no further gain after a third formal sitting.
How long should I spend on each question in a reasoning test?
Take the minutes, less a reserve, and divide by the questions. For 30 questions in 24 minutes with three minutes held back, the pace is 42 seconds.
Should I guess on an aptitude test?
Yes where wrong answers cost nothing. Where they cost marks, guess once you can rule out one option.
Are practice questions the same as the real test?
No. A practice set teaches the method for a type, not the paper you will sit. Do the employer's own sample questions first if the invitation links to any.
Where can I practise numerical, verbal and logical reasoning questions?
The reasoning battery is on EdProcto's page for candidates, and the Free plan covers it. The rule-made sets give new figures at each opening; the analogies, odd one out and situational sets draw on fixed banks.
References
- Arendasy, M. E., & Sommer, M. (2017). Reducing the effect size of the retest effect: Examining different approaches. Intelligence, 62, 89–98. https://doi.org/10.1016/j.intell.2017.03.003
- Burton, R. F. (2001). Quantifying the effects of chance in multiple choice and true/false tests: Question selection and guessing of answers. Assessment & Evaluation in Higher Education, 26(1), 41–50. https://doi.org/10.1080/02602930020022273
- Evans, J. St. B. T., Barston, J. L., & Pollard, P. (1983). On the conflict between logic and belief in syllogistic reasoning. Memory & Cognition, 11(3), 295–306. https://doi.org/10.3758/BF03196976
- Kellogg, J. S., Hopko, D. R., & Ashcraft, M. H. (1999). The effects of time pressure on arithmetic performance. Journal of Anxiety Disorders, 13(6), 591–600. https://doi.org/10.1016/S0887-6185(99)00025-0
- Kruger, J., Wirtz, D., & Miller, D. T. (2005). Counterfactual thinking and the first instinct fallacy. Journal of Personality and Social Psychology, 88(5), 725–735. https://doi.org/10.1037/0022-3514.88.5.725
- Scharfen, J., Blum, D., & Holling, H. (2018). Response time reduction due to retesting in mental speed tests: A meta-analysis. Journal of Intelligence, 6(1), Article 6. https://doi.org/10.3390/jintelligence6010006
- Scharfen, J., Peters, J. M., & Holling, H. (2018). Retest effects in cognitive ability tests: A meta-analysis. Intelligence, 67, 44–66. https://doi.org/10.1016/j.intell.2018.01.003