Methodology
How the hrmforce Skills Framework is built: five layers, five skill types, a proficiency scale with behavioural anchors, and an assessment method with validity evidence per skill.
Five layers
| Layer | Unit | Count | What the layer is for |
|---|---|---|---|
| 1 | Domain | 10 | High-level reporting and the layout of a skill passport. |
| 2 | Cluster | 66 | The conversation level with the client: which clusters matter for this job. |
| 3 | Skill construct | 356 | The unit of measurement. Assessment method, instrument, weight and behavioural anchor attach here. |
| 4 | Skill | 3068 | The search layer: granular terms as they appear in vacancies and CVs. They inherit the assessment route of their construct. |
| 5 | Behavioural anchor | 1068 | Assessment and the development conversation. Anchors at N1, N3 and N5. |
Why two skill layers: you cannot provide behavioural anchors and validity evidence for thousands of terms. The constructs therefore remain the unit of measurement, while the granular layer makes the whole thing searchable and matches vacancy text.
Five skill types
| T | Name | What that means for measurement |
|---|---|---|
| K | Knowledge | Declarative knowledge. Measure with a knowledge test, not a questionnaire. |
| V | Skill | Applied and learnable. Measure with a work sample or structured assessment. |
| G | Behavioural competence | Dispositional behaviour. Links to Big Fifty and 360 feedback. Not developed in three months. |
| T | Tool or technology | A concrete tool or package. Expect rapid obsolescence, review annually. |
| A | Ability | Relatively stable capacity. Measure with the Ability Scan or a simulation. Not trainable to a higher level. |
The proficiency scale
| Level | Label | Autonomy | Complexity | Scope | Depth |
|---|---|---|---|---|---|
| N1 | Guided | works under supervision and follows instruction | routine, one variable at a time | own task | basic understanding of terminology |
| N2 | Applying | works independently within set frameworks | familiar situations with limited variation | own role | working knowledge, applies standard approach |
| N3 | Proficient | sets own approach and seeks input proactively | several variables, some ambiguity | own team or process | thorough knowledge, weighs alternatives |
| N4 | Advanced | directs others and decides on the approach | complex, with conflicting interests | several teams or a department | in-depth knowledge, advises others |
| N5 | Leading | sets the standard and the policy | strategic, under high uncertainty | organisation, value chain or profession | authority, advances the profession |
Five levels instead of seven or eight, because rater reliability barely improves above that while usability in a conversation declines. The four dimensions come from SFIA 9, and defining levels by cognitive demand, task complexity and autonomy follows DigComp 3.0.
Assessment methods and validity
Every skill has a primary assessment method. The validities are the corrected meta-analytic estimates from Sackett, Zhang, Berry and Lievens (2022) in the Journal of Applied Psychology. They replace the Schmidt and Hunter (1998) figures, which came out systematically too high because of an excessive range restriction correction.
| Assessment method | rho | SD | What it is |
|---|---|---|---|
| Structureerd interview | 0.42 | 0.19 | Behaviour-based interview with fixed questions and a rating scale per question. |
| Cognitieve capaciteitentest | 0.41 | 0.14 | Standardised test of verbal, numerical or abstract reasoning. |
| Kennistoets | 0.40 | 0.13 | Test of declarative job knowledge, usually assembled per client. |
| Persoonlijkheidsvragenlijst | 0.40 | 0.15 | Self-report of behavioural preferences. At hrmforce the Big Fifty and the 15PF. |
| Zelfrapportage met verificatie | 0.38 | 0.09 | Biodata and structured experience questions, empirically keyed against performance criteria. |
| Werkproef | 0.33 | 0.09 | The candidate performs a representative work sample under standardised conditions. |
| Assessment center | 0.29 | 0.09 | Multiple exercises, multiple assessors, dimension-wise scoring. |
| Situational Judgement Test | 0.12 | 0.11 | Scenario with response options, scored on knowledge or behavioural tendency. |
| Praktijkobservatie | - | - | Structured workplace observation with a form based on behavioural anchors. |
| 360 feedback | - | - | Multiple raters around the person rate behaviour against behavioural anchors. |
| Portfolio/bewijsstuk | - | - | A collection of prior work or results, assessed against fixed criteria. |
| Certificaat/diploma | - | - | Formal proof of completed education or examination. |
| Simulatie | - | - | Simulated work situation, usually digital, with standardised scoring. |
The spread matters as much as the mean. Above an SD of .18 validity varies strongly by context and local validation is needed. Think in incremental validity and method combination, not in rankings.
International standards
| Standard | Version | Date | Source |
|---|---|---|---|
| ESCO | v1.2.1 | 2025-12-10 | esco.ec.europa.eu |
| O*NET | 31.0 | 2026-08 | www.onetcenter.org |
| Lightcast Open Skills | rolling 2026 | 2026 | lightcast.io |
| SFIA | 9 | 2024-10-30 | sfia-online.org |
| EU JRC frameworks | DigComp 3.0, EntreComp, LifeComp, GreenComp, AILit | 2026 | joint-research-centre.ec.europa.eu |
| WEF Future of Jobs | 2025 | 2025-01 | www.weforum.org |
| ISCO-08 | 2008 | 2008 | www.ilo.org |
| Beroepenindeling ROA-CBS 2014 | editie 2025 | 2026-04-17 | www.cbs.nl |
The mapping is conceptual and not formally validated. Where a reference could not be justified the field was left empty rather than inventing a plausible code.
What this framework is not
- Not a standard. Every profile is a starting point to be tailored with the organisation before selecting on it.
- Not a decision rule. A fit score supports a conversation. Always combine it with a structured interview, the strongest single predictor.
- Not a normed measurement. There are no norm groups per skill yet: a level is a definition, not a percentile.
- Not a replacement for rater training. Behavioural anchors only work well alongside a calibration session for the assessors.
Request a demo
If you want to apply this framework to your own job architecture, we are happy to help.
Request a demo