Employer guideRead at source, 9 September 2026

Pre-Employment Assessments: the Two Rules That Decide Whether You Can Use One

A pre-employment assessment is a selection procedure, and federal law treats it as one. Under the Uniform Guidelines on Employee Selection Procedures, any selection procedure with an adverse impact is considered discriminatory unless it has been validated — and adverse impact is generally presumed where one group's selection rate is less than four-fifths of the highest group's. Separately, the ADA forbids a medical examination before a conditional job offer, which is where personality inventories get employers into trouble. Two questions decide whether you can use a test: can you produce a validation study for roles like yours, and is the test a medical examination in disguise.

The guidelines do not care what you call it

A cognitive test, a situational judgment questionnaire, a typing test, a structured scoring rubric, a physical ability trial — all of them are selection procedures. Section 1607.3(A) is the sentence the whole subject hangs on:

Read the structure of it. Adverse impact does not make a test unlawful by itself; it shifts the burden. Once the impact is there, the question becomes whether you can produce the validation, and "the vendor says it's validated" is not the same as having a study that covers a job like yours in a workforce like yours.

The four-fifths rule, and how to run it on your own data

Section 1607.4(D) sets the threshold that federal enforcement agencies apply:

The arithmetic is straightforward and most employers have never done it. For each group, divide the number who passed by the number who took the test. Divide the lower rate by the highest rate. If the result is below 0.80, the guidelines treat that as evidence of adverse impact. It is a rule of thumb rather than a statistical test, and small samples make it jumpy — but it is the rule of thumb the agencies use, so it is the one to run first.

Three ways to validate, and only one of them is usually practical

Section 1607.5 recognizes three approaches, and they are not equally available to a frontline employer.

• Criterion-related validity — a study showing the test scores predict job performance. Statistically demanding and sample-hungry; realistic for a large employer with a stable role and good performance data, and rarely realistic otherwise.
• Content validity — a demonstration that the procedure is a representative sample of the job's duties. This is the practical route for most frontline roles, and it is also the one that improves the test: a work sample that mirrors the actual task is easier to defend and better at predicting than a generic aptitude battery.
• Construct validity — a demonstration that the procedure measures a characteristic shown to be necessary for successful performance. The most conceptually demanding of the three and the least commonly relied on.

The practical translation for a warehouse, a care home or a fleet is unglamorous: a short work sample beats a personality inventory on every dimension that matters. It is easier to defend under content validity, easier to explain to a candidate, harder to coach for, and it does not raise the ADA problem below.

The line the ADA draws, and where tests cross it

Section 12112(d) separates the process into before and after a conditional offer. Before an offer, a covered entity shall not conduct a medical examination or make inquiries of a job applicant as to whether such applicant is an individual with a disability or as to the nature or severity of a disability. What you may do at that stage is inquire into the ability of an applicant to perform job-related functions.

After a conditional offer and before work begins, an employer may require a medical examination and condition the offer on the result — provided all entering employees in that job category are subjected to the same examination regardless of disability, the information is collected and maintained on separate forms and treated as a confidential medical record, and the results are used only as the ADA permits.

Five questions for the vendor

• Show me the validation study — and specifically, does it cover a job like this one, in a workforce like ours? A study on office roles does not validate a test for a night-shift warehouse.
• Which validation strategy is it? Criterion, content or construct. The answer tells you how much of the argument you are inheriting and how much you would have to build.
• What adverse impact has the instrument shown, and in what populations? A vendor that has never measured it is telling you something.
• Has it been reviewed as a medical examination under the ADA? Ask for the reasoning, not the assurance.
• What is the accommodation process? For an applicant who needs an alternative format or extra time, there has to be one, and it has to be reachable before they abandon the application.

Where the answer to the first question is thin, the cheapest fix is usually not a different vendor. It is a shorter, job-specific work sample built and documented as content-valid — which is also the version a hiring manager will actually use. Our pre-screening questions guide covers the step before this one, and the FAQ below covers assessment centers, the heavier end.

Pre-hiring tests for hourly roles: which ones are worth the applicants they cost

Every pre-hire assessment has two prices: what the vendor charges, and the applicants who stop at it. For hourly roles the second is usually larger. The test for whether a pre-hiring test earns its place is simple: would the result change who you hire, for a skill you could not check in a ten-minute work sample after the screen? Most generic batteries fail it. A few specific tests pass it easily.

Two rules apply to every row. First, the test has to measure the job. A 60-words-a-minute cutoff for a role that types a few notes a shift is a filter with no reason behind it, and under the four-fifths rule an unnecessary filter is the kind you end up having to defend. Second, the test has to measure the skill rather than a disability. The EEOC’s ADA regulation makes it unlawful to fail to select and administer tests so that, for an applicant whose disability impairs sensory, manual or speaking skills, the results accurately reflect the skill the test purports to measure “rather than reflecting the impaired sensory, manual, or speaking skills,” except where those skills are what the test measures (29 CFR 1630.11, current as of September 10, 2026, read at source September 28, 2026). In practice that means a stated way to ask for an alternative format or more time, reachable before the applicant gives up.

Placement is the lever you control. The completion numbers in the next section (31% of caregiver clickers finishing an application against 5% for therapy roles) come from forms with no test on them at all. A test placed in the application flow compounds that loss; the same test given after the phone screen is taken by people with a reason to finish it.

What a test costs you in applicants

Legal defensibility is one axis. Volume is the other, and it is the one that shows up in the vacancy.

Across the 891 Boostpoint-managed Meta campaigns in our 2026 Social Job Advertising Benchmark, the share of clickers who completed an application ran from 31% for caregiver roles down to 11% for registered nurse and 5% for therapy roles. The pattern is consistent: the more credentialed the candidate, the less patience they extend to your process. An assessment placed before any human contact is another completion event applied to people you have already paid for, and its cost lands hardest on exactly the roles you are struggling to fill.

That is not an argument against assessments. It is an argument about placement. A test that runs after a recruiter has decided a candidate is worth an hour is applied to a smaller number of people with a reason to finish it, and it costs a fraction as much per completed candidate as the same test placed in front of the funnel.

Pre-hire testing, run properly: the five records a hiring assessment test needs

Pre-hire testing is usually bought as a product and run as an afterthought. The Uniform Guidelines treat it as a process, and most of what they ask for is paperwork an employer can keep without a consultant. Whether you use a vendor's hiring assessment test or a work sample you built yourself, five things should exist before the first applicant takes it.

• A written look at the job. The guidelines say any validity study "should be based upon a review of information about the job," normally a job analysis (29 CFR 1607.14(A)). For a frontline role that can be one page: the tasks, how often they happen, and which ones the test samples.
• The cutoff, and why. Write down the passing score and the job reason for it before anyone sits the test, so it cannot drift toward the candidates a manager already likes.
• Pass rates by group. Section 1607.4(A) says users should keep records that show the impact of their tests on employment opportunities by race, sex and ethnic group. Employers with 100 or fewer employees, and others not required to file EEO-1 reports, may use a simplified version under 29 CFR 1607.15(A)(1): each year, the number hired, promoted and terminated for each job and the number of applicants, by sex and where appropriate by race and national origin, plus the selection procedures used.
• A retest rule. The guidelines say users "should provide a reasonable opportunity for retesting and reconsideration" and may take reasonable steps to keep the test secure (29 CFR 1607.12). A simple policy, such as a retake after a set waiting period, meets that intent.
• The completed tests themselves. Test papers and application forms are hiring records, and 29 CFR 1602.14 requires hiring records to be kept for one year from the date they were made or the personnel action, whichever is later, and until final disposition if a charge is filed.

None of this makes a pre-hire test safe on its own; the four-fifths rule and the validation routes above still decide that. What the records do is let you run the four-fifths check at all, and show your working if anyone asks.

Personality tests for hiring employees: which kind you are buying

A personality test for hiring employees falls on one side or the other of a line the EEOC has drawn in its enforcement guidance on pre-employment disability-related questions and medical examinations. Its answer to whether an employer may give psychological examinations to applicants is "Yes, unless the particular examination is medical." A psychological test is medical if it provides evidence that would lead to identifying a mental disorder or impairment, such as those listed in the DSM. If a test "is designed and used to measure only things such as honesty, tastes, and habits, it is not medical."

The guidance adds a detail that matters when a vendor reassures you about intent. Its example is a test designed to reveal mental illness that an employer says it uses only for tastes and habits; because the test is interpreted by a psychologist and routinely used in clinical settings to provide evidence toward a diagnosis, the guidance treats it as medical anyway. What the instrument is built and used for decides it, not the label on the report.

For hourly roles the practical case against recruitment personality tests is rarely legal alone. A generic inventory is hard to link to a specific job, which is what content validity needs, and every added page in the application loses people you paid to reach. If a trait genuinely matters, such as reliability on an early shift, a structured interview question or a work sample usually measures it more directly; our skills-based hiring guide and interview scorecard template show how to score it.

Frequently asked questions

A scored assessment is one way to replace a proxy with a measure. The wider version of that argument, and the rules that govern the swap, are in skills-based hiring for frontline roles. If the tool doing the scoring is a model rather than a test, AI screening covers what attaches to it.