Rising Up recognized by Deeptech...
Psychometric test in 2026: complete definition, main use cases (recruitment, coaching, education), limits of classic psychometric tests and the scientific alternative based on behavioral assessment.
You have probably heard an HR director or a recruitment firm mention "running a psychometric test" before a hire. It has become almost a ritual in some executive hiring processes: the candidate takes an online personality test, receives a profile, and the recruiter draws conclusions about their fit for the role.
But what does a psychometric test actually measure? Is it reliable enough to inform a hiring decision in 2026? And more importantly, are there modern, scientific alternatives that avoid its main limitations?
At Rising Up, our scientific team in cognitive sciences (CNRS, ENS-PSL, EHESS, Université Paris Cité) has been studying and comparing the main behavioral assessment methods for more than 10 years. This article gives you a complete and honest overview of the topic:
The term psychometrics comes from the Greek psukhē (soul, mind) and metron (measure). Literally: the measurement of the mind.
A psychometric test is a standardized questionnaire designed to assess a psychological characteristic of a person: a personality trait, a cognitive ability, a communication style, motivation, or intelligence. It relies on statistical methods to compare an individual's answers to a reference population, or to their own answers (depending on the method used).
Modern psychometrics emerged at the end of the 19th century. Three landmark dates help understand its evolution:
Since then, psychometric tests have multiplied: PAPI, Hogan, MBTI, Big Five, HEXACO, SOSIE, Saville Wave, Eysenck, Process Communication Model. Each rests on different theoretical foundations, but they all share the same basic logic: ask the candidate to declare their traits, preferences or habits, then compare those answers statistically to a norm.
Depending on their objective, the psychometric tests used in business fall into three broad families.
Objective: describe the stable traits of a person (extraversion, openness, emotional stability, etc.).
Most commonly used in recruitment in France:
Average duration: 20 to 60 minutes. Cost per session: 50 EUR to 300 EUR.
Objective: measure a person's cognitive abilities (logical reasoning, working memory, information processing speed, problem solving).
Examples used in executive assessment:
Duration: 30 to 90 minutes. Used mainly for very specific profiles (senior management, R&D, demanding technical roles).
Objective: measure a precise skill (mental arithmetic, verbal reasoning, mechanical comprehension, situational judgment).
Examples: SHL Verify batteries, Talent Q Aspects, specific numerical or verbal aptitude tests.
These tests are shorter (10 to 30 minutes) and target one precise dimension rather than a full profile.
A classic psychometric test involves 3 main steps.
The recruiter or the firm sends a personalized link to the candidate. The test is taken online, generally in autonomy, from any device. Average announced duration: 20 to 60 minutes depending on the test.
The candidate answers a series of questions through forced choice or a Likert scale. Two common formats:
In theory, there is no right or wrong answer. In practice, the candidate often knows what the employer wants to hear.
The recruiter receives a detailed report in the form of a graph, a table or a narrative. They may be invited to a debrief with a certified consultant (Cubiks, Hogan Assessments, SHL, etc.) to interpret the results.
The report typically presents:
Used in the right context, psychometric tests do have legitimate use cases.
The classic psychometric test suffers from three biases that are well documented in scientific literature. These limits are particularly penalizing in a recruitment context.
The candidate ticks what they think they are, but also and above all what they think the employer wants to see. This is the social desirability bias.
On the behavioral skills assessed in business, the gap between declaration and actual behavior ranges between 20 and 40% according to psychometric meta-analyses (see our article on how much a bad hire really costs). In other words, in a psychometric test taken in a recruitment context, up to 40% of the answers are influenced by the image the candidate wants to project.
A savvy candidate who has done some research on the test and the role will adjust their answers. A naïve candidate will answer more honestly, and will sometimes be penalized by comparison. This bias makes the test poorly reliable to distinguish high-performing candidates from candidates who are simply skilled at answering.
A classic psychometric test does not measure what the person does. It measures what the person says about themselves. These are fundamentally different things.
In contemporary cognitive sciences, work performance is far better predicted from behavioral observations (how the person reacts to a situation) than from self-descriptions (how the person describes themselves in the abstract). This difference is massive and has been documented since John Campbell's work in the 1990s on the structure of job performance.
Most psychometric tests measure generic dimensions (extraversion, emotional stability, openness) that are not aligned with the specific competencies required by the role to fill.
A profile scored "very emotionally stable" on the Big Five can be excellent for a crisis management position, and completely unsuited for a creative role that demands high sensitivity to weak signals. The test does not say which profile fits which role.
Yet in recruitment, it is precisely the fit between role, profile and context that matters, not the generic score.
Faced with these limits, cognitive science research has been proposing a new approach since the early 2010s: contextualized behavioral assessment.
The idea is simple: instead of asking a candidate to describe themselves, we present them with concrete professional situations and observe their reactions. What they do, not what they say they would do.
At Rising Up, this approach has been formalized in the CoreSkills scientific framework, co-developed with a consortium of researchers (CNRS, ENS-PSL, EHESS) and published in open access on the Open Science Framework (DOI 10.17605/OSF.IO/RVEHU).
Rising Up's Soft Skill Scan, recognized as a breakthrough innovation by the European Innovation Council, combines two complementary dimensions:
The candidate takes the Scan in 20 minutes, from any device, with no possible preparation. The situational challenges have no obvious "right answer": they reveal the candidate's natural trade-offs under pressure, outside the "desirable" script.
You receive a behavioral profile on the 18 core skills of the Rising Up scientific framework, organized in 3 families:
The result is presented through strengths and areas of vigilance, with a role compatibility score on a simple scale:
It is a decision-support tool for the manager, not an oracle. The final decision always belongs to the manager, after reviewing the Scan and conducting the interview. The tool enriches and steers the interview, it does not replace it.
| Criterion | Classic psychometric test | Rising Up Soft Skill Scan |
|---|---|---|
| Type of measurement | Declarative self-assessment | Situational challenges + real habits |
| Social desirability bias | 20-40% variance | Strongly reduced (no "right" answer) |
| Average duration | 20-60 min | 20 min |
| Number of dimensions | 4 to 20 depending on the test | 18 core skills in 3 families |
| Scientific foundation | Trait theory (1940s-1960s) | Contemporary cognitive sciences |
| Contextualization to the role | Low | Crossed with CoreSkills AI (job description analysis) |
| Output | Narrative report to interpret | Strengths / vigilance + role compatibility score |
| Price per session | 50 EUR to 300 EUR | ~20 EUR in the 3,000 EUR pilot |
No. There is no legal obligation in France. Some recruitment firms use it as a systematic filter, other companies do without it entirely. It is a methodological choice, not a standard.
Yes, to some extent. Candidates who have researched the test and the role can steer their answers toward the desired image. This is precisely the social desirability bias. Modern tests include "control scales" to detect inconsistencies, but these can still be worked around.
Officially, psychometric tests claim to measure stable traits. Empirically, stability is moderate to low: people who evolve professionally often see their profile evolve too. The MBTI, for example, shows up to 50% different results on a retest 5 weeks apart. This is a signal that the assumed "stability" is closer to a contextual adaptation.
No. Our methodology, the Soft Skill Scan, is based on dynamic situational challenges and a questionnaire on real working habits. It is a fundamentally different approach, rooted in contemporary cognitive sciences and validated by the recognition of the European Innovation Council.
A classic psychometric test taken through a consultancy costs between 100 EUR and 300 EUR per candidate (with debrief). The Rising Up pilot program includes 150 assessments over 8 weeks for 3,000 EUR excl. tax, i.e. about 20 EUR per assessment, plus the CoreSkills AI analysis of your 5 job descriptions.
If you want to assess objectively the behavioral skills of your candidates before the interview, without the methodological limits of classic psychometric tests, discover the Rising Up recruitment pilot program:
🎯 150 behavioral assessments over 8 weeks, on the candidates of your 5 target roles
🎯 CoreSkills AI analysis of your 5 job descriptions (kept for life, even if you stop mid-way)
🎯 3,000 EUR excl. tax paid once at signature, valid until 30 September 2026 (then 4,500 EUR excl. tax)
🎯 Limited to 5 companies per month
The asymmetry is clear: a single bad hire avoided covers 10 times the cost of the pilot.
Send us a message 📩 at hello@risinguparis.com and we will set up a 15-minute demo for you within the week.
From recruitment At development skills, Rising Up provides reliable behavioral data to guide your HR and educational choices.
.svg){kind=small}
.svg)
.svg)
