Pre-employment Assessment Tests: What They Measure, and What You Can Ask For

The email arrives an hour after you apply, and it is not from a person. There is a link, a time limit, and a sentence telling you to find somewhere quiet. Forty minutes later you have answered sixty questions about shape rotation and rated how strongly you agree that you prefer to finish one task before starting another, and nobody has told you what any of it was for.

Assessments are the least explained part of hiring and the part people prepare for worst, mostly because the advice available is written by the companies that sell the tests or by the companies that sell practice packs for them. So this piece is built from sources that have no product in the fight: the federal regulator that polices these tests in the United States, the rule that sets the standard, and the 2022 meta-analysis that quietly rewrote what the industry believed about how well any of them work.

The four things you are actually being sent

Vendors give these products names that reveal nothing. Underneath, almost every assessment you will meet is one of four things, and the four behave completely differently.

1. A cognitive ability test. Numerical, verbal or abstract reasoning, worked at speed against a clock. Sometimes called an aptitude test, a general mental ability test, or by a brand name. It is the only one of the four where practice reliably moves your score, because a large part of what it measures early on is familiarity with the question format.
2. A personality questionnaire. Long lists of statements you rate, or forced-choice pairs where you pick which of two unflattering descriptions fits you better. It measures what you say about yourself, scored against a profile the employer chose. Nothing is timed and nothing is right.
3. A situational judgement test. A workplace scenario, four or five possible responses, and you rank them or pick the best and worst. It scores whether you pick the response the employer decided was correct, which is a measure of their house style as much as of your judgement.
4. A work sample or job knowledge test. Do a piece of the actual job: a coding exercise, a written brief, a spreadsheet task, an in-tray. Or answer questions about the field you claim to work in. These are the ones where your experience genuinely shows.

One practical consequence before anything else: if the invitation does not say which you are getting, the time limit usually does. A hard per-question clock means cognitive. No clock, and a progress bar with a hundred-odd statements, means personality. A scenario with named colleagues means situational judgement.

What the research says these tests predict

For about twenty-five years, everyone in hiring worked from one table. Schmidt and Hunter published a summary in 1998 that put cognitive ability tests at the top of the pile, and a generation of selection systems was built on that ranking. In 2022, Paul Sackett, Charlene Zhang, Christopher Berry and Filip Lievens published a reanalysis in the Journal of Applied Psychology arguing that the statistical correction underneath those figures had been applied wrongly for decades, and that most of them were too high.

The revised numbers are below, next to the ones they replaced. They are correlations between the method and later job performance, so higher means the method tells an employer more. A perfect predictor would be 1.0, and nothing in hiring is close.

Three things in that chart are worth sitting with, because they contradict what a lot of people in hiring still say out loud.

• Cognitive ability is no longer the standout. It moved from .51 to .31, and the same four authors put it plainly in their 2023 follow-up: “Cognitive ability is no longer the stand-out predictor that it was in the prior work. Structured interviews emerged as the predictor with the highest mean validity.”
• Work sample tests fell furthest. From .54 to .33, the largest single drop in the table. The test that looks most like the job turns out to predict it less well than a well-run interview does.
• Years of experience predicts almost nothing. .07, down from .18. The number most résumés are organised around, and the thing most job adverts ask for first, sits at the bottom of the list.

Why the numbers got smaller

This part is technical, and it is worth two paragraphs because it is the reason to trust the new figures over the old ones.

To measure whether a test predicts performance, you need people who took the test and then did the job. But the only people who did the job are the ones the employer hired, and the employer hired them partly because of the test. So the group you can study is narrower than the group the test would be used on, and a correlation measured in a narrow group comes out lower than the real one. Researchers correct for this, and the correction is large.

Sackett and colleagues showed that the correction had been applied to studies where it did not belong. Where the studies supplying the correction factor were not representative of the studies being corrected, the adjustment inflated the result. Strip out the inappropriate corrections and the whole table comes down, unevenly, which is why the ranking changed rather than just the scale. Their 2023 follow-up gives the size of it: the difference was “.21 lower for work sample tests, .20 lower for cognitive ability, .19 lower for unstructured interviews”.

What an employer has to be able to show

In the United States, an employer cannot simply buy a test and use it. The Equal Employment Opportunity Commission’s guidance on employment tests and selection procedures, published in December 2007, sets the frame: Title VII, the Americans with Disabilities Act and the Age Discrimination in Employment Act all apply to tests, and Title VII “prohibits employers from using neutral tests or selection procedures that have the effect of disproportionately excluding persons based on race, color, religion, sex, or national origin”.

The threshold question is whether the test screens one group out at a lower rate, and there is a number attached to it. The Uniform Guidelines on Employee Selection Procedures, at 29 CFR § 1607.4(D), state that “a selection rate for any race, sex, or ethnic group which is less than four-fifths (4/5) (or eighty percent) of the rate for the group with the highest rate will generally be regarded by the Federal enforcement agencies as evidence of adverse impact”. If a test trips that, the employer has to show it is job-related and consistent with business necessity. The EEOC’s wording for meeting that standard is that the employer shows the test “is necessary to the safe and efficient performance of the job”.

That is not theoretical. The same EEOC guidance describes EEOC v. Dial Corp., where a physical strength test introduced at a plant sent the share of women hired from 46% down to 15%, and the Eighth Circuit upheld the finding that using it violated Title VII. It also describes a cognitive test used at Ford Motor Company that had a disparate impact on African American applicants and was settled with $8.55 million in relief.

The adjustment you can ask for, and when to ask

This is the most useful thing on this page and the least known. If a disability affects how you take a test, the employer has to give the test differently, and the EEOC states the principle in its guidance for job applicants and the ADA: “the ADA requires that employers give application tests in a format or manner that does not require use of your impaired skill, unless the test is designed to measure that skill”.

Read the exception carefully, because it is narrow and it is fair. A timed numerical reasoning test is designed to measure numerical reasoning under time pressure, not your ability to read a screen, so a screen reader or a different format sits squarely inside the rule. A typing speed test is designed to measure typing speed. The question is always what the test is for.

Timing matters more than people expect. The same guidance says to let an employer know “as soon as you realize that you will need a reasonable accommodation” for the hiring process. An adjustment asked for before you sit the test is an administrative matter; one raised after a low score is a much harder conversation, and by then the result is usually already recorded.

• Ask in writing, to the recruiter or the address on the invitation, before you open the test link. One sentence naming the adjustment is enough, and you do not have to volunteer a diagnosis in order to ask.
• Name the adjustment, not the condition. Extra time, a screen reader, a paper version, a break between sections, larger text, a keyboard alternative, a room without a time-limited camera.
• Do not open the test first. Many platforms start the clock on first load and will not reset it.
• In the UK the duty is the same in substance. GOV.UK’s guidance on reasonable adjustments lists “changing the recruitment process so a candidate can be considered for a job” as an example of what an employer must do.

When a personality test crosses a line

There is a limit on what a personality questionnaire may ask before you have a job offer, and it is more interesting than most candidates realise. The EEOC’s enforcement guidance on pre-employment disability-related questions and medical examinations, issued in October 1995, draws the distinction: a psychological test is a medical examination where it “provide[s] evidence that would lead to identifying a mental disorder or impairment”.

A questionnaire measuring habits, tastes and how you prefer to work is not a medical examination. One built to surface anxiety, depression or a compulsive disorder is, and a medical examination cannot be required before a conditional offer. That rule sits underneath every long personality inventory sent out with an application, and it is why the carefully built ones stay firmly on the side of working style.

So if a pre-offer questionnaire is asking about your mental health history rather than how you like to work, that is worth noticing. It does not mean you should refuse and lose the application. It is a legitimate thing to raise with the employer, and it tells you something about how carefully the rest of their process was built.

Preparing for each type, honestly

Preparation has wildly different value depending on which of the four you are sitting, and the practice-pack industry is not going to tell you which is which.

Cognitive and aptitude tests: practise, and it pays

This is the one place where a few hours genuinely helps. Most of the early gain is format familiarity rather than any change in ability: knowing that the numerical section gives you about a minute a question, that the abstract section rewards spotting the rule before counting features, and that most of these tests do not penalise wrong answers, so a blank is strictly worse than a guess. Do two or three full timed papers rather than fifty loose questions, because pacing is the part that transfers.

Personality questionnaires: answer as the person, consistently

The instinct is to answer as the ideal candidate in the job advert. Resist it for two reasons. These instruments are long and repetitive by design, asking the same underlying thing several ways, so a profile assembled from what you think they want tends to contradict itself. And the trait most commonly screened for is conscientiousness, which on the revised figures predicts performance at .21 anyway. Answering honestly and quickly produces a cleaner profile than answering strategically and slowly.

Situational judgement: read for the employer, not for yourself

An SJT has a scoring key written by someone at that organisation, so the answer marked correct is the one their culture rewards. Before you start, read the company’s values page and the job advert again, and notice which way they lean: escalate to a manager or solve it yourself, tell the customer immediately or check with the team first. Then answer for that organisation. Most SJT items come down to whether the employer prizes autonomy or process, and they usually say which somewhere on their careers site.

Work samples: the one your résumé was promising

Treat a work sample as the job, because it is. Read the brief twice, do what was asked before doing anything clever, leave time to check it, and say what you would do differently with more time. An unfinished but correct submission with a clear note beats a rushed complete one, and the note is itself evidence of judgement.

If you are told you did not pass

Two practical things, then one that matters more.

1. Ask for feedback once, briefly. Some employers, particularly in the UK and Australia, will send a candidate report from the test provider on request. Many will not. Ask, accept the answer, and move on.
2. Ask how long the result is held. A number of employers apply a re-sit window and will not let you re-apply for six or twelve months, and knowing the length changes whether it is worth applying to a sister role this month.
3. Do not rebuild your résumé around a test result. The assessment sits after the screening stage, so a rejection there means your document worked. Rewriting it now would fix the part that was not broken.

If you are being rejected before any assessment arrives, that is a different problem with a different fix, and it is usually the résumé rather than you. Our guide to pulling keywords out of the job description covers the most common cause, and the free ATS score check will show you which of a posting’s terms your résumé is missing before you send it. It also helps to know what the status in the employer’s portal actually means, because “in progress” and “under consideration” are not the same thing.