Pre-Employment Assessment: Types, Validity Numbers, and When to Skip One

A pre-employment assessment is a standardized instrument an employer administers before making a hire, scored the same way for every candidate, to predict how that person will perform on the job. Three features make something an assessment rather than an interview: every candidate gets the same stimulus, the result is a score, and the score feeds a decision rule. Cognitive ability tests, personality inventories, typing and data-entry tests, and situational judgment tests are all pre-employment assessments. A background check is not one, because it verifies history instead of measuring a capability. Neither is a resume review, because nothing about it is standardized.

In hourly hiring the word covers two jobs that are not the same job, and telling them apart decides whether the step is worth what it costs you. Either you are measuring an ability the role requires that a candidate cannot credibly self-report, or you are confirming facts: a current certification, which shifts they can actually work, how they will get to a 5 a.m. open. The first is an assessment problem. The second is a pre-screening problem, and a test battery is an expensive, slow way to solve it.

The Four Types

TypeWhat it measures2022 validity estimateWhere it earns its place
Job knowledge testFacts and procedures specific to the trade.40Licensed and technical roles where a knowledge gap is expensive
Work samplePerformance on a real task from the job.33Roles where the work can be sampled in under an hour
Cognitive ability testReasoning, problem solving, speed of learning.31Long training ramps and real judgment under novelty
Personality inventoryTraits, most usefully conscientiousnessBelow the three aboveRarely strong enough to decide anything on its own
Structured interview, for comparisonJob-relevant behavior and specifics, scored to a rubric.42Nearly every role, and the cheapest stage to run first

Those numbers are correlations with later job performance, from Sackett, Zhang, Berry, and Lievens in the Journal of Applied Psychology in 2022, which corrected the Schmidt and Hunter estimates the industry had quoted since 1998. Every method came down. The structured interview finished at the top. The full before-and-after table is in what the meta-analyses actually found.

Two readings matter for a hiring manager. Nothing on that list is close to accurate about an individual: at .42, a structured interview accounts for under a fifth of the variation in performance across many hires, so a vendor claiming more should be asked what they measured it against. And the ranking is not the one assessment pricing implies. The cheapest, fastest stage outperforms the test batteries, and it is the stage most hourly employers run worst.

What the Step Costs You

Validity is half the decision. The other half is who never finishes.

An assessment adds a timed task between applying and hearing back, commonly 20 to 40 minutes for a cognitive and personality battery. In high-volume hourly hiring that is exactly where the funnel leaks. A warehouse or care applicant is applying to five employers from a phone, often on cell data, often between shifts, and the employer who responds first usually wins. Accommodation and food services runs one of the highest quits rates of any sector in the Bureau of Labor Statistics Job Openings and Labor Turnover Survey, which also means your applicants sit in several live pipelines at once.

Run the arithmetic before you buy. Say a posting draws 200 applicants and the assessment takes 30 minutes. Assume 45 percent of everyone you invited finishes it. You now have 90 scored candidates, and the 110 who left were not a random sample. They skewed toward the people with the least spare time, which in hourly hiring means the people currently working somebody else's shift. Measure completion against everyone invited rather than everyone who started, because the second number is the one the tool can see and the first is the one that decides your pipeline. More on where this happens: candidate drop-off.

When an Assessment Is Worth It

Four conditions. The more of them hold, the better the case:

  • The ability is hard to self-report. Forklift safety rules, pharmacy math, code, a dispatcher's judgment call. Anyone can claim it, few can produce it on demand.
  • Training is long and washouts are expensive. A six-week ramp justifies a step that a two-day ramp does not.
  • Volume is high relative to openings. Losing half the pool to an extra step is survivable at four times the applicants you need. It is fatal at nine applicants for a line cook.
  • You can show the test relates to the job. See the next section.

Skip it when availability, credentials, reliability, and transportation decide the hire, which describes most frontline roles. A forklift knowledge test tells you nothing about whether the candidate can work nights, and nights are what you are trying to cover. For the roles where that is clearest, see the warehouse and logistics hiring guide and the shift lead screen.

The Legal Floor

An assessment is a selection procedure, and in the United States it falls under the Uniform Guidelines on Employee Selection Procedures, 29 CFR Part 1607. Three obligations that bite in practice:

Job-relatedness. If a test screens people out, you need evidence that what it measures relates to the job. A vendor saying the instrument is validated means the vendor ran a study on the vendor's sample, not on your roles.

Adverse impact. Track selection rates by group. The Guidelines' rule of thumb treats a rate below four-fifths of the highest group's rate as evidence of adverse impact worth investigating. Cognitive ability tests carry the largest mean score differences by race of any common method, which is why they draw the most litigation. Hiring bias is measurable at this stage in a way it never is in an improvised conversation.

Disability and medical inquiry. Timed tests can require accommodation. A personality inventory designed to reveal a mental impairment functions as a medical examination and cannot be given before a conditional offer under the ADA. Ask counsel which side of that line a specific instrument sits on, and ask the vendor in writing.

None of this makes testing a bad idea. It makes an unvalidated test a documented liability that a sloppy phone screen, for all its faults, does not create in the same way.

Assessment or Screen First

This is a sequence question, not an either-or. The screen goes first because it is cheap and fast and removes the candidates who fail on a fact. The assessment, if you run one, goes to the shortlist, where a 30-minute task costs you a few candidates instead of half the pool.

That first stage is what Prelim does, and the boundary is worth stating plainly: it is not an assessment platform. No cognitive test, no personality inventory, no proctoring. Paste the job description, edit the screening questions the AI drafts, and share one link that opens in any phone browser with no app and no account. Every candidate answers the same questions in the same order by text, on their own schedule, and you get scored and ranked results with full transcripts, so you can read the answer behind any score instead of trusting the number. If you need a validated ability test, buy one from a company that builds them. Conversational screening compared with a test battery lays out where that line falls.

Key Takeaways

A pre-employment assessment is a scored, standardized instrument used to predict performance. The four common families are job knowledge, work samples, cognitive ability, and personality, and the corrected 2022 estimates put all of them at or below a structured interview. Weigh validity against completion: a 30-minute test that loses half an hourly applicant pool is a worse screen than six questions almost everyone finishes, whatever its correlation. Run the cheap structured screen on everyone, reserve testing for the shortlist and for abilities nobody can self-report, and document job-relatedness before a test rejects a single applicant. Related: screening accuracy for measuring whether any of it works, and the interview scorecard for scoring the stage you already run.

Try AI-powered screening interviews

Screen candidates in minutes with structured scoring and instant results.

Start free