How to read a psychometric test report: percentiles, stens and score bands

A psychometric test report packs a lot into a few numbers: a percentile, a sten, a band, perhaps a speed or accuracy figure. Read well, they tell you how a candidate compares with a relevant group and where to probe at interview. Read badly, they invite false precision. Here is how to read a psychometric test report with confidence.

Raw scores and why reports convert them

A raw score is simply the number of questions answered correctly. On its own it means very little: 24 out of 30 could be excellent on a hard test and ordinary on an easy one. To make scores interpretable, reports compare them with the scores of a norm group, a large sample of people who have taken the same test, and convert them into a standard score. The choice of comparison group matters a great deal, which is why reports should always say which norm group was used and how large it is; our article on norm groups explains how to judge whether a group fits your candidates. As a benchmark, the European test review model rates norm samples of fewer than 200 people as inadequate even for low-stakes use, and norms 20 or more years old as inadequate.^[1]

Percentiles, stens and T-scores

Percentiles are the most intuitive. A candidate at the 70th percentile scored as well as or better than 70% of the norm group. The catch is that percentiles are not equal steps. Most people score near the middle, so a small change in raw score can move someone from the 45th to the 60th percentile, while at the extremes the same change barely moves the percentile at all. Differences in the middle of the range are easy to over-read.

Stens (standard tens) place scores on a ten-point scale with a mean of 5.5 and a standard deviation of 2. Stens of 5 and 6 represent the broad average, and each step is the same distance on the underlying scale, which makes stens easier to compare than percentiles. T-scores do the same job on a finer scale, with a mean of 50 and a standard deviation of 10; a T-score of 60 is one standard deviation above the norm group average. The European review model lists these alongside other formats such as stanines and z-scores; they are different ways of expressing the same comparison, not different results.^[1]

Why scores are bands, not points

No test measures perfectly. Every score contains some measurement error, quantified by the standard error of measurement (SEM), which depends on the test's reliability. Harvill's classic explanation is that an observed score is best read as a range within which the person's true score is fairly likely to fall, rather than as an exact value.^[2] The Standards for Educational and Psychological Testing likewise expect test users to take measurement error into account when interpreting individual scores.^[3]

In practice this means two candidates a few percentile points apart may have no meaningful difference in ability. It is also why good reports group scores into descriptive bands, such as Very Low, Low, Average, High and Very High, rather than inviting a ranking to the nearest point. Treat neighbouring bands as overlapping and look for consistent patterns across several measures.

This is also the strongest argument against pass marks. A single cut-off turns a score with a margin of error into a hard yes or no, so candidates either side of the line are treated as different when the test cannot reliably tell them apart. The better practice is to interpret bands in the light of job analysis and alongside other evidence, such as interview, work samples and experience, rather than letting one number decide.

Speed and accuracy indicators

Many ability test reports now show how many questions the candidate attempted, how accurate they were on the questions they answered and how much of the time they used. These add useful context to the main score. A candidate with a middling score who answered few questions with high accuracy has a different profile from one who answered everything quickly with many errors. The first may be careful but slow under time pressure; the second may be fast but less precise. Neither pattern is good or bad in itself; what matters is which matters more in the role. Low completion can also signal anxiety, unfamiliarity with online testing or a need for adjustments, so ask before drawing conclusions.

Using the results to structure interview questions

The most valuable use of a report is often to plan the interview. Structured interviews, in which every candidate is asked the same job-related questions and answers are scored against agreed criteria, are among the most valid selection methods.^[4] Test results tell you where to focus those questions.

Where a candidate's numerical reasoning is in the Low band for a role that involves budgeting, for example, a behavioural question can explore how they have handled figures at work: describe a time you had to check or analyse financial information under pressure; what was the situation, what did you do and what was the result? Where a score is in the High band, ask for evidence that the strength shows up in practice. Either way, the test opens the line of enquiry and the interview tests it. Our guide to ability tests in recruitment covers how these scores relate to job performance.