Understanding Assessment Percentiles and Comparison Groups: A Simple Employer Guide

An employer receives two assessment reports. One candidate answered 32 out of 40 questions correctly. Another report says a different candidate is at the 80th percentile. Which result is better?

The question cannot be answered from those numbers alone. 32 out of 40 tells us how many questions were answered correctly. The 80th percentile tells us where someone stands in relation to a particular group of people who took a comparable assessment. These numbers describe different things.

Understanding the difference matters. When employers confuse percentages and percentiles, they may reject capable candidates, exaggerate the meaning of small score differences or compare people who were tested under very different conditions.

This guide explains percentiles and comparison groups using everyday language. It also shows when they can help, when they can mislead and what employers should check before using them in recruitment or promotion.

What is a percentage score?

A percentage score is usually the share of available marks a person earned. If a test has 40 equally weighted questions and a candidate answers 32 correctly, the percentage score is:

32 ÷ 40 × 100 = 80%.

This means the candidate earned 80% of the available marks. It does not mean the candidate performed better than 80% of other test takers.

Percentage scores are easy to calculate, but their meaning still depends on the questions. An 80% result on a very easy test may represent a different level of skill from 80% on a demanding test. The employer must understand what the test covers and whether the questions match the job.

Also check whether all questions carry equal marks. If some tasks have different weights, simply dividing the number of correct answers by the total number of questions may produce the wrong score.

What is a percentile rank?

A percentile rank describes a person's relative position within a defined comparison group. A candidate at approximately the 80th percentile has performed at least as well as roughly 80% of that group, under the method used to calculate the rank.

It does not mean that the candidate answered 80% of questions correctly. Nor does it mean there is an 80% chance that the person will succeed in the job.

Imagine 100 people taking a comparable reasoning test. Their scores are arranged from lowest to highest. A candidate whose result falls around the 80th percentile is positioned above much of that group. But the exact calculation can differ depending on how equal scores are handled.

This is particularly important when many people receive the same score, as can happen in a 40-question test. Different published percentile methods can produce slightly different ranks. Employers should know which method the assessment provider uses.

Percentage versus percentile: a quick comparison

A useful rule for hiring managers is to ask: “80% of what?” If the answer is “the available marks,” it is a percentage. If the answer is “a defined group of people,” it is a percentile.

What is a comparison group?

A comparison group, sometimes called a norm group, is the group of test takers whose results are used as the reference for interpreting a person's score.

For example, a company might compare a candidate's numerical reasoning result with:

• People applying for similar entry-level jobs.
• Experienced accounting or finance professionals.
• Recent university graduates.
• Employees being considered for promotion.
• A broad sample of adults who took the same test.

These groups may have very different score patterns. A candidate can therefore have the same percentage score but a different percentile depending on the group used for comparison.

Worked example: the same score, different position

Suppose Ada scores 30 out of 40, or 75%, on a numerical reasoning assessment.

For illustration, assume the provider has appropriately collected comparable scores from two separate groups and applies the same ranking method:

Ada did not gain or lose any questions. Her result is still 30 out of 40. What changed is the reference group.

The numbers are fictional and are not based on actual Talentrenza users or validated norms. They demonstrate why the report must identify the comparison group before the percentile can be interpreted responsibly.

An employer hiring a junior accounts assistant might find the entry-level comparison more relevant. An employer assessing experienced finance managers might prefer evidence from an experienced, appropriately matched group. Neither choice should be made simply to produce a more flattering number.

How percentiles are calculated: a simple illustration

Percentile calculations can be more complicated than they first appear. One straightforward teaching method counts how many people scored below a candidate and adds half the number who received the same score.

Illustrative percentile rank = (number scoring below + 0.5 × number scoring the same) ÷ total people × 100.

Suppose 20 people completed the same test under comparable conditions:

• 14 scored below Tunde.
• 2 people, including Tunde, received Tunde's score.
• 4 scored above Tunde.

The counts add up: 14 + 2 + 4 = 20.

Using the illustrative midrank approach:

(14 + 0.5 × 2) ÷ 20 × 100 = 15 ÷ 20 × 100 = 75th percentile.

This tells us Tunde's approximate standing in this small group. It does not show whether he has the skills required for the job.

Other ranking conventions may place tied scores differently. A professional score report should identify its calculation approach, especially when the results influence important decisions.

Why a group of 20 is not enough for strong claims

The example above is useful for learning the arithmetic, but a small group can produce unstable rankings. If only a few people join or leave the comparison group, percentile positions may shift noticeably.

An employer should not treat a percentile from 20 local applicants as if it were a carefully developed national or international benchmark. The suitability of a reference group depends on its size, composition, how participants were selected, when they were tested and how closely the conditions and assessment versions match.

Choosing the right group for recruitment

Before using percentile ranks, ask whether the reference group resembles the people and decisions for which the assessment is intended.

1. Match the job level

An entry-level customer service applicant should not automatically be judged against experienced senior managers. The latter group may have more work experience, training and familiarity with business scenarios.

Conversely, using a very inexperienced reference group for a specialist position may make a modest score look exceptional. Neither comparison establishes that the person can perform the actual job.

2. Match the skill being assessed

A general reasoning benchmark may not be the best comparison for a role requiring detailed bookkeeping, spreadsheet modelling or technical writing. The employer should first establish the job's skill requirements and then consider whether a suitable comparison group exists.

A candidate's position on a broad reasoning test cannot replace direct evidence of a specialist skill that was never assessed.

3. Check the assessment version

Two tests with the same number of questions are not automatically equally difficult. If one group took an easier form, its score distribution may not be comparable with results from a harder form.

Comparing scores across forms may require evidence that the forms are appropriately linked or equivalent for the intended interpretation. Employers should not assume that the same raw score represents the same level of performance across different versions.

4. Consider language and testing conditions

Differences in language, device access, time limits, assistive technology and interruptions may influence scores. A comparison group drawn from one testing environment may not be appropriate for a different environment without supporting evidence.

The aim is not to make every candidate identical. It is to avoid interpreting differences caused by irrelevant testing barriers as differences in job skill.

5. Check how recent the data is

Job requirements and candidate populations can change. A comparison group assembled many years ago may no longer reflect the people currently applying for a role.

There is no single universal expiry date for a comparison group. The assessment owner should document when it was created, when it was last reviewed and what changes would trigger a fresh evaluation.

Five questions to ask an assessment provider

When a vendor offers percentile reports, ask for clear answers to these questions:

1. Who is in the comparison group? Request a description of relevant job levels, experience and other characteristics where appropriate and lawful.
2. How many people contributed? Ask about sample size and whether the group was assembled in a way that supports the intended comparison.
3. Which test version and conditions were used? Check whether results are genuinely comparable.
4. How were tied scores handled? Understand the ranking method, especially for shorter tests.
5. When was the group reviewed? Ask about updates, quality checks and known limitations.

If a provider cannot explain the source of its reference group, treat the percentile as an unverified claim rather than a reliable hiring benchmark.

Where percentile ranks stop

A percentile reports relative position, not whether a candidate has demonstrated the minimum capability required for a job. Employers should establish any job-related minimum separately and avoid interpreting a high percentile as proof of competence. For standard-setting procedures, see How Employers Can Set Fair Pass Marks for Assessments; for selection decisions that combine absolute and relative evidence, see Job Standards Versus Candidate Comparisons.

Worked hiring example: three candidates

Consider three applicants whose reported percentile ranks are 82, 79 and 61 against the same documented reference group. These figures describe relative standing in that group, not the percentage of questions answered correctly and not proof that any applicant meets the requirements of the vacancy. A three-point difference between 82 and 79 should not automatically be treated as a meaningful difference in job readiness.

Before interpreting the ranks, verify that the same test version, scoring procedure and relevant reference population were used, and request information about measurement precision and sample size. If the comparison groups differ, the three ranks are not directly comparable. If the employer has not established a defensible job-related standard, the ranks cannot substitute for one.

The practical decision is to use the percentile report as one qualified source of evidence, then refer to the separate guide on job standards versus candidate comparisons for how an employer might combine minimum requirements with other selection evidence. This example deliberately does not rank candidates into automatic hire or reject categories.

What about percentile differences of just a few points?

Suppose two candidates are at the 74th and 77th percentiles. It is tempting to treat the second as clearly superior. That conclusion may not be supported.

Assessment scores contain some measurement uncertainty. A small difference in percentile rank can reflect a small raw-score difference, ties or the shape of the comparison group's score distribution.

Percentile scales are also not equal-interval measures: the difference between the 50th and 60th percentiles does not necessarily represent the same score difference as the gap between the 80th and 90th percentiles.

Employers should avoid making fine distinctions between candidates without suitable evidence about score precision and decision consequences. A report that provides only percentile ranks may hide information needed for responsible interpretation.

Using percentiles for employee development and promotion

Percentiles can sometimes help identify where an employee's results sit relative to a relevant group. But the purpose matters.

For development, an employee may benefit more from knowing which tasks they handled well and where practice is needed than from learning that they rank at a particular percentile. A skills report should point to actions: training, coaching, practice or supervised work.

For promotion, employers must consider the demands of the higher-level role. A high percentile among current junior staff does not automatically establish readiness to manage a team. Leadership judgement, communication, planning and other role-specific requirements may need separate evidence.

Never convert a percentile into a promotion rule simply because the number is easy to sort.

Fairness and privacy considerations

Comparison groups can create fairness problems when their membership is poorly understood. A benchmark drawn mainly from one language group, region, device type or career stage may not support every intended use.

Employers should investigate whether the assessment measures the relevant skill and whether avoidable barriers influence performance. Unexpected differences between groups are a reason for careful review, not automatic proof of bias or proof that the assessment is fair.

Use personal information only when there is a lawful and appropriate reason. Access to candidate scores and any sensitive information should be limited to authorised staff. Explain how results will be used and retained, and apply relevant privacy and employment requirements in each jurisdiction.

A global employer should not assume that one reference group, legal explanation or decision rule works equally well in every country.

Common mistakes employers should avoid

Mistake 1: Saying an 80th-percentile candidate got 80% correct. These are different measures. Show the raw or percentage score separately where appropriate.

Mistake 2: Assuming a percentile is a pass mark. A relative ranking cannot by itself establish minimum job competence.

Mistake 3: Comparing different tests as if they were identical. A 75th percentile in numerical reasoning is not interchangeable with a 75th percentile in communication.

Mistake 4: Ignoring the reference group. A percentile without a clearly identified comparison group has little interpretive value.

Mistake 5: Treating tiny ranking differences as decisive. Consider score uncertainty and the consequences of an incorrect decision.

Mistake 6: Calling a local applicant ranking a global norm. A small, convenience-based group of applicants is not automatically representative of a profession or country.

Mistake 7: Using percentile reports to make claims about future performance. Evidence connecting assessment results to relevant outcomes is needed before making predictive claims.

A practical employer checklist

Before using percentile ranks in recruitment or promotion, confirm the following:

• We understand the difference between percentage scores and percentile ranks.
• We know exactly which assessment and version produced the result.
• The comparison group is clearly described.
• Its size, age and relevance have been reviewed.
• Testing conditions are sufficiently comparable for the intended use.
• We understand how tied scores are treated.
• We know the limits of score precision and avoid overinterpreting small differences.
• Our hiring or promotion standards are linked to job requirements.
• We do not use percentiles as automatic proof of competence or future performance.
• Candidate privacy, accessibility and local legal requirements have been considered.

If important answers are missing, pause the use of percentile-based decision rules until the limitations are understood.

A four-week plan for improving score interpretation

Instead of adopting a generic four-week recruitment plan, conduct a percentile-report verification exercise. First document the question the percentile is intended to answer; next obtain the provider’s reference-group definition, sample size, test version, calculation and treatment of ties; then compare those details with the role and population being assessed. Record any limitations before a hiring manager sees the report.

Recheck the documentation when the test form, reference population or provider methodology changes. A percentile is only as interpretable as its supporting comparison data; a timetable does not establish that those data are appropriate.

Frequently asked questions

Is the 90th percentile the same as scoring 90%?

No. The 90th percentile describes relative standing within a specified group. A 90% score describes marks earned on the test. One does not determine the other without additional information.

Is a candidate at the 60th percentile above average?

They are above the middle of the specified comparison group under many common ranking approaches. That does not necessarily mean they meet the job's required standard or are above average in every relevant skill.

Can two candidates with the same test score have different percentiles?

If they are compared with different reference groups, yes. Under the same test, comparison group and calculation method, tied scores should normally receive the same reported percentile.

Can an employer calculate percentiles from its own applicants?

It can calculate ranks within a clearly defined applicant group, but small or unusual groups may produce unstable and misleading comparisons. An applicant-pool rank should not be presented as a professionally established population norm without supporting evidence.

Does a high percentile mean the candidate will perform well at work?

Not necessarily. It describes relative test performance. Predicting workplace performance requires suitable evidence connecting the assessment and its use to job outcomes.

Should Talentrenza employers use percentile rankings to shortlist candidates?

Employers should use only the scoring and reporting features actually available in their Talentrenza product, understand what each reported figure means and apply approved, job-related selection rules. This article does not claim that Talentrenza currently provides validated percentile benchmarks.

Which matters more: a pass mark or a percentile?

They answer different questions. A justified minimum standard may help establish required competence; a relevant percentile may add context about relative standing. Neither should replace thoughtful review of the role and available evidence.

Conclusion

Percentiles can be useful when employers know who is being compared, how the scores were obtained and what the comparison can reasonably show. Without that information, a precise-looking ranking may create false confidence.

The safest starting point is simple: keep percentage scores and percentile ranks separate, use comparison groups that fit the intended purpose, and never let a ranking replace evidence of the skills a job actually requires.

For further reading, see How Employers Can Set Fair Pass Marks for Assessments, Job Analysis for Assessment Design: A Step-by-Step Guide, and Assessment Reliability vs Validity: What Employers Need to Understand.