Science
Pre-employment testing: how to set the score that decides
The test doesn't tell you where to draw the line. Here's how to set a cut score you can defend.
November 16, 2026
Aiden Loe

The short version
A pre-employment testing cut score is the point on an assessment’s scale that decides who moves forward. Set it per role, using the evidence behind the test, how real applicants score, how many candidates your process can advance, and an adverse impact check at that threshold.
Pre-employment testing can help structure a hiring decision around the requirements of a role. But the test itself does not tell you where to draw the line.
A score threshold has to be chosen deliberately, set alongside what the assessment is actually built to measure, not picked as a round number that feels safe.
This piece walks through how to choose a threshold, the trade-offs it carries, and what to check before you rely on it.
What is a cut score in pre-employment testing?
A cut score is the point on an assessment’s scale above which a candidate moves forward, and below which they do not. It is a decision rule, not a scientific constant.
The same assessment can justify different cut scores for different roles, different hiring volumes, and different labor markets, because a cut score answers how many people a process can afford to advance at this stage, not what counts as an objectively good score.
This is why a cut score should never be borrowed wholesale from another role or another company. It is set per role family and reviewed as conditions change.
What evidence supports the score?
Before setting a number, check what evidence supports the assessment for the role: what it covers, what it measures, and what outcomes it has been linked to so far.
- Content validity: does the test cover the skills and knowledge that matter for the role?
- Construct validity: does it measure the trait it claims to measure?
- Reliability and fairness: are scores consistent over time and across raters, and do they mean the same thing for different groups?
- Criterion validity: does the score relate to a real outcome, either now (concurrent) or later (predictive)? Ask what outcome data exists, how much, and which outcome it is, such as progression to a later stage, training completion or job performance.
A threshold set on a test with weak criterion validity is just a guess with extra steps.
A threshold backed by evidence that scores relate to real outcomes is far easier to defend, and the closer those outcomes are to job performance, the better. (If you are still choosing a test, see Pre-employment assessment: 7 test types and how to choose.)
Ask which outcome the evidence uses, how large the sample was, and whether the result is as observed or statistically corrected.
Expect modest numbers: observed correlations for a single selection method often fall around .2 to .3.
How do your applicants actually score?
A threshold only means something in the context of how real applicants actually score, not a theoretical bell curve. Maki’s own benchmarking approach uses percentile bands of applicants’ scores as a starting point, then adjusts. As an example:
Below 30th percentile
Usually filtered out
Scores low relative to other applicants.
31st to 69th percentile
Hold
Revisited if more candidates are needed later.
70th percentile and above
Progress
Moves on to the next stage.
Percentiles depend on whose scores they come from, whether applicants, current staff or the general population, so always state the comparison group.
A threshold set before enough real candidates have taken the assessment is provisional. Until a few hundred candidates have completed it, the distribution is too thin to read reliably, so treat early cut-offs as a starting point and keep reviewing them as the data builds.
What does the cut score cost you in each direction?
Every cut score trades two kinds of error against each other.
Set it too high
You reject candidates who would have succeeded, shrinking your pipeline for no gain in quality.
Set it too low
You let through candidates the assessment was telling you not to, which erodes the point of testing in the first place.
There is no threshold that eliminates both risks. Choosing one means deciding which error your process can tolerate less, given how many roles you need to fill and how expensive a bad hire is for that role.
Behind the score
See how Maki’s science team builds, validates and audits every assessment.
Explore Maki’s science →Does the threshold hold up across groups?
A cut score that looks reasonable on paper can still produce adverse impact once it is applied to a real applicant pool.
A common first check is the four-fifths rule: compare each group’s pass rate with that of the group that passes most often.
A ratio below 0.80 is a flag to investigate, not proof of unfairness, and a ratio above it doesn’t prove fairness either. The check only means something when each group is large enough to compare.
The ratio depends on where you set the line. Here is a hypothetical example using the same assessment with two different cut scores:
| Cut score | Group A pass rate (highest) | Group B pass rate | Ratio | Above 0.80? |
|---|---|---|---|---|
| Lower | 55% | 50% | 50 ÷ 55 = 0.91 | Yes |
| Higher | 30% | 21% | 21 ÷ 30 = 0.70 | No |
The assessment hasn’t changed, but the higher cut score pushes the ratio below 0.80. That’s why the check has to be run at the threshold you actually use, and again whenever the threshold or the applicant pool changes.
At Maki, we run adverse impact checks on live candidate data as standard.
How do you put these four checks together?
Setting a cut score well means answering four questions, roughly in this order:
Each question is simple to ask, even when answering it takes work. Asking all four is what separates a threshold you can defend from one that just felt right at the time.
The answers change as data builds, so revisit them whenever the role, the applicant pool or the threshold changes. (The same discipline applies to scoring rubrics: see Behaviorally anchored rating scales: here is one of ours.)
Want assessments with the evidence to back your cut score?
See how Maki’s science team validates, benchmarks and checks every assessment for adverse impact.
Request a demoFAQ
Is a higher cut score always safer?
No. If the assessment predicts performance, a higher cut score reduces the risk of a bad hire. But it also rejects more people who would have done well, shrinks your pipeline and can widen the gap between groups’ pass rates. The right level depends on how costly each type of error is for the role.
How often should a cut score be reviewed?
Whenever the role, the applicant pool, hiring volume or the assessment itself changes, since each changes what the data behind the threshold means. Even without a trigger, check it at least once a year. Treat any threshold as provisional until a few hundred candidates have completed the assessment.
Is the four-fifths rule a legal requirement?
No. It’s a rule of thumb for flagging possible adverse impact. A ratio below 0.80 is a reason to investigate, and a ratio above it doesn’t prove that a threshold is fair. The check also only means something when each group is large enough to compare.
Sources
Reviewed by Maki’s science team, November 2026.
- AERA, APA and NCME (2014). Standards for Educational and Psychological Testing.
- Society for Industrial and Organizational Psychology (2018). Principles for the Validation and Use of Personnel Selection Procedures, 5th edition.
- Uniform Guidelines on Employee Selection Procedures (1978), 29 CFR §1607.4(D).
- Sackett, P. R., Zhang, C., Berry, C. M. and Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection. Journal of Applied Psychology, 107(11), 2040–2068.
- Maki internal benchmarking and adverse impact guidance.
See what Maki Agents can do for you



