Pre-Employment Personality Tests: How Hiring Uses Them
The CQ Team · 7 min read ·
Hiring personality tests predict job performance modestly, applicants inflate their scores, and US law bars clinical questions before a job offer.
A pre-employment personality test asks how you usually act, and the employer compares your answers with what the job needs. The traits that predict performance best are conscientiousness and emotional stability, and even those predict modestly: conscientiousness has a corrected validity of .19, against .42 for a structured interview (Sackett, Zhang, Berry & Lievens 2022). In the US the test is legal as long as it does not work as a medical exam before a job offer and does not screen out protected groups without a job-related reason (EEOC).
Habits, preferences and tendencies. The research that judges hiring tests sorts their scales into the Big Five traits: conscientiousness, emotional stability, extraversion, agreeableness and openness. Barrick and Mount did this across 117 studies and 23,994 people, and conscientiousness predicted performance at a similar level, .20 to .23, in all five job groups they checked, from police to sales to managers (Barrick & Mount 1991).
What a hiring test may not measure before an offer is mental illness. The EEOC says a psychological test is a medical exam if it gives evidence that would identify a mental disorder, and employers cannot require medical exams at the applicant stage. A test designed to measure only honesty, tastes and habits is not a medical exam (EEOC 1995).
If you want to see the trait language before a real test, our Big Five conscientiousness guide explains what high and low scores look like.
Modestly, and less well than older estimates claimed. In 2022 Sackett and colleagues corrected a long-standing overcorrection in the selection research, and most predictors fell by .10 to .20 points. Here is where personality tests now sit:
| Hiring method | Validity in 2022 | Earlier estimate (1998) |
|---|---|---|
| Structured interview | .42 | .51 |
| Job knowledge test | .40 | .48 |
| Cognitive ability test | .31 | .51 |
| Integrity test | .31 | .41 |
| Conscientiousness | .19 | .31 |
| Unstructured interview | .19 | .38 |
Corrected validities from Sackett, Zhang, Berry & Lievens (2022), Table 3. 0 means no link with rated job performance and 1 a perfect one. The other four Big Five traits came in between .05 (openness) and .10 (extraversion and agreeableness).
Two more details from the same analysis matter to you as an applicant. Items framed around work predict better than general ones: conscientiousness rises from .19 to .25 when the questions ask how you behave at work. And personality scales show small gaps between groups: the Black and White difference on conscientiousness was −.07 standard deviations, against .79 for cognitive ability tests (Sackett et al. 2022).
The critics read the same numbers as a reason for caution. In a paper by a panel of former journal editors, Michael Campion's summary of 12 meta-analyses found a median corrected validity of .18 for personality tests, against .40 for cognitive tests, and the panel argued that self-report personality tests in hiring should be reconsidered (Morgeson et al. 2007). The 2022 figures sit close to their median. Our guide to using your results at work covers what the scores are better for once you have the job.
You can shift your score, and real applicants do. A meta-analysis of 33 studies compared job applicants with people taking the same tests with nothing at stake. Applicants scored higher on conscientiousness by .45 standard deviations and on emotional stability by .44, and only .13 and .11 higher on openness and extraversion. The authors read the pattern as applicants inflating the traits they see as relevant to the job (Birkeland et al. 2006). Campion's review counted 39 studies on whether people can fake, and virtually all found that scores could be distorted (Morgeson et al. 2007).
The case for "it rarely changes the outcome" comes from a different design. Hogan, Barrett and Hogan followed 5,266 applicants who were rejected, reapplied for the same job six months later and took the same test again. On any one scale, 5.2% or fewer improved their score, and the changes were as likely to go down as up (Hogan, Barrett & Hogan 2007). Its first author worked for the test's publisher, Hogan Assessment Systems.
Which to trust? The meta-analysis is the broader evidence: 33 studies, independent authors, many tests. The retest study is a single test, covers only rejected applicants and comes from the publisher, so it shows that a second attempt rarely moves a score, not that the first attempt was unpolished. The Morgeson panel's view was that faking cannot be avoided and that the larger problem is the modest validity itself.
Yes in the US, within the limits federal law sets.
| Rule | What it means for the test | Source |
|---|---|---|
| ADA, before an offer | No medical exams, including psychological tests that would identify a mental disorder | EEOC 1995 |
| ADA, any stage | Tests that screen out people with disabilities must be job-related, and you can ask for a reasonable accommodation in how the test is given | EEOC 2007 |
| Title VII | A test that screens out a race, color, religion, sex or national origin group must be job-related and consistent with business necessity | EEOC 2007 |
| Four-fifths rule | A pass rate for any group below 80% of the highest group's rate is generally taken as evidence of adverse impact | 29 CFR 1607.4(D) |
The EEOC also places the duty on the employer: it is responsible for a test's validity, even when a vendor built it (EEOC 2007).
The leading case shows where the line falls. Rent-A-Center gave employees seeking promotion a battery that included 502 questions from the MMPI, a clinical inventory, and anyone with more than 12 "weighted deviations" was not considered. In 2005 the Seventh Circuit held that because the MMPI is designed in part to reveal mental illness, it is best treated as a medical examination, and its use violated the ADA (Karraker v. Rent-A-Center). This is a summary of federal rules, not legal advice.
As the version of you that shows up at work. Work-framed items are the ones that predict performance, .25 against .19 for general items, so read each question with your working week in mind. Inflation is already the norm: applicants as a group already score about half a standard deviation higher on conscientiousness than people with nothing at stake.
Three practical points from the sources above:
Before your next application, ask yourself: would the people who work with you describe you the way you are about to describe yourself? To practise the format, CQ has short tests in the style of the Hogan Personality Inventory, the Caliper Profile and the Predictive Index. They are not the official instruments, and an employer's version will use its own items and norms. For how much any personality test can tell you, see how accurate personality tests are, and for the employer's side, personality tests for teams. More on work and personality sits in the careers hub.
There is no standard pass or fail score. Each employer decides how to use the results, and some set a hard cut-off: Rent-A-Center rejected anyone above 12 "weighted deviations" on its promotion test (Karraker v. Rent-A-Center).
We found no recent primary survey that isolates personality tests. A 2011 SHRM poll of 495 HR professionals found 18% used one for hiring or promotion (SHRM), and a 2022 SHRM survey of about 1,700 found 56% used some form of pre-employment assessment, of any type (AP via WTOP).
Not by much. In the study of 5,266 rejected applicants who retook the same test six months later, 5.2% or fewer improved on any scale (Hogan, Barrett & Hogan 2007).
The International Personality Item Pool has over 3,000 items and over 250 scales in the public domain, free to use without permission (IPIP).
Validity figures come from Sackett, Zhang, Berry and Lievens (2022, Table 3), read in full, and Barrick and Mount (1991), read in full. Faking figures come from the Birkeland et al. (2006) and Hogan, Barrett and Hogan (2007) abstracts, and from Morgeson et al. (2007), read in full. Legal points come from the EEOC's 1995 and 2007 guidance, 29 CFR 1607.4 and the Karraker opinion, each read in full.
The most useful number here for an applicant is the smallest: a .19 validity is a weak link for any one person, so the test is one input beside interviews and work samples that predict as well or better.
Curious where you sit? The DISC test is free, takes about 2 minutes, and needs no sign-up.
Take the DISC test →