Timed assessments in recruitment: are you ranking ability or pace?

By Junaid Khalid, founder of PrepClubs

Timed aptitude tests are among the most common filters in early-stage hiring: quick, cheap at volume and easy to rank. But a timed score blends two things: how much of the paper a candidate reached, and how accurately they answered what they reached. When the clock is tight, our data suggests the first of those carries more weight.

Key figures

• 727 practice attempts analysed across nine assessment formats
• 57.4% of questions in the final fifth of speeded tests left blank
• 0.72 correlation between score and coverage on speeded tests, versus 0.56 for accuracy

What we measured

PrepClubs runs practice versions of widely used pre-employment tests, written to each test's published format. For the State of Timed Assessment 2026 study, we analysed 727 practice attempts from 575 candidates across nine assessment formats. Four are speeded, allowing 14 to 18 seconds per question: formats modelled on the CCAT, Wonderlic, PI Cognitive and Cubiks Logiks. The others give a more generous clock, such as critical thinking tests in the Watson-Glaser format at about 45 seconds per question.

For every attempt we split the score into two parts: coverage, the share of questions answered, and accuracy, the share of answered questions that were right.

Finding 1: on speeded tests, reaching questions matters more than getting them right

On the speeded formats, the raw score correlated more strongly with coverage (r = 0.72) than with accuracy (r = 0.56). On the generous formats the order reversed: accuracy 0.77, coverage 0.59.

So when time is generous, a high score mostly means a candidate answered well. When time is tight, it mostly means they got further.

“When time is tight, it mostly means they got further.”

Finding 2: the final fifth of a fast test often goes unanswered

Across 533 speeded attempts, 57.4% of the questions in the final fifth of the paper were left blank. On the generous formats the figure was 12.1%. Among candidates who used at least 90% of the time available, the final-fifth blank rate was 61.2%, so early quitting does not explain it.

A blank shows only that a candidate did not reach a question, not that they could not answer it.

“A blank shows only that a candidate did not reach a question, not that they could not answer it.”

Finding 3: willingness to guess moves scores

None of the formats we studied deducts marks for a wrong answer, yet most candidates on the speeded tests still left questions blank. On a 50-question speeded paper, a guess on every blank would have added about two marks on average, roughly 7% of the typical score.

That matters at the cut-off. On our CCAT-format test, the middle half of candidates spanned 44% to 68%, a range of about 12 marks. Two marks can move someone across a threshold set near the middle, so part of what a mid-range cut-off ranks is who knew that a blank earns nothing.

A finding we withdrew

We expected to show that rushing makes candidates answer worse, since accuracy falls steeply across a speeded paper. It did not hold up. Faster and slower responses to the same question differed by about three points, and the decline matched how these tests are built, with harder questions towards the end. In our data, the clock mainly limits how many questions get reached.

What recruitment teams can do

1. Find out whether your test is speeded. Ask your provider how many seconds per question it allows and what share of candidates finish. If most do not finish, the score carries a large pace component.
2. Ask for coverage alongside the score. Two candidates on 30 out of 50 can look very different: one answered 32 questions and got 30 right, the other answered all 50. Many platforms can report questions attempted.
3. Tell every candidate how the test is scored. If wrong answers carry no penalty, say so in the instructions. Otherwise you are partly ranking who already knew.
4. Treat mid-range cut-offs as bands, not lines. A difference of two or three marks near the middle of the distribution is within what guessing alone can produce. Read the score alongside other evidence rather than as a hard pass mark.
5. Match the clock to the job. If speed under pressure is a genuine requirement of the role, a speeded test measures something relevant. If not, a more generous limit gives a cleaner read of reasoning and narrows the gap for candidates given extra time as a reasonable adjustment.
6. Check thresholds against your own outcomes. The real test of a cut-off is whether it predicts performance in your roles. Compare the scores and coverage of past hires with how they went on to perform.

Method and limitations

• These are practice attempts on PrepClubs, not employer sittings. Candidates chose to practise, which likely skews the sample towards motivated job seekers.
• Our tests follow each publisher's published format but are not the official tests, and our scores are not normed or mapped to any employer's cut score.
• The findings are correlations. They show how parts of a score relate, not what causes them. A late question can be answered worse because it is harder, because the candidate is tired, or because it is rushed.
• We re-graded every attempt from its stored answers and kept only those where the result matched. Of 1,473 attempts on record, 556 failed that check, 26 were empty and 164 used less than half the clock, leaving 727.
• We do not collect demographic, device or country data, so we cannot say whether these effects differ between groups of candidates.

The full method, sample and discard ledger, with every figure downloadable, is published in the PrepClubs State of Timed Assessment 2026 report.

What this means for hiring

A timed test score combines two measurements. On a generous clock it mostly reflects how well candidates answer; on a tight one, how far they get. Knowing which kind of test you use, and setting thresholds with that in mind, is a simple way to avoid screening out capable people for the wrong reason.

About the author

Junaid Khalid is the founder of PrepClubs, a practice platform for pre-employment aptitude tests, where he leads research into how candidates perform under assessment time limits.