| Instrument design | Multi-item scales for every parameter, written to WCRC's definitions and pre-tested with a small group of your employees | A single question can be misread. Several items measuring the same thing can be checked against each other |
|---|
| Translation | Translation into up to three languages, checked for equivalence of meaning | Every employee answers the same question, whatever language they answer in |
|---|
| Sampling | Stratified by employee group and location, sized for the precision standard | Every part of the organisation is heard in proportion, and scores are accurate to ±3 points |
|---|
| Response quality | Screens out incomplete, patterned and implausibly fast responses before analysis | Only considered answers count |
|---|
| Representativeness | Weights responses to your actual workforce composition | A group that answers in large numbers can't outweigh one that answers less |
|---|
| Reliability | Tests the internal consistency of each parameter's items | Proves the items measure one thing consistently |
|---|
| Structural validity | Confirmatory factor analysis of the ten-parameter structure | Proves the ten parameters are distinct, not ten names for the same feeling |
|---|
| Measurement invariance | Tests that employee groups read items the same way | Equity gaps reflect real differences, not different readings of a question |
|---|
| Inter-rater reliability | Two analysts score the systems audit independently, against a written rubric | An audit score doesn't depend on who happened to score it |
|---|
| Data verification | Three checks on every outcome figure, using ISO 30414 human capital definitions where one applies | Every number in the score traces to a record |
|---|
| Normalisation | Three published rules convert every indicator to a 0–100 scale | Unlike units become comparable, and you are compared with your own industry |
|---|
| Triangulation | Compares the three evidence streams within each parameter | Where survey, audit and data disagree, the disagreement is reported as a finding |
|---|
| Significance testing | Every reported gap and change is tested at 95% confidence, with correction for multiple comparisons and effect sizes | Nothing is reported as a difference unless it is one |
|---|
| Qualitative coding | Two researchers code interviews and focus groups independently | Themes come from the evidence, not from one reader's impression |
|---|
| Value linkage (Signature) | Statistical analysis relating your parameter scores to your own business outcomes over the year | Shows which parameters move together with results in your organisation. It is reported as association, not as proof of cause |
|---|
| Governance | An independent Methodology Council reviews the framework every year; all thresholds are published | The rules are fixed in advance and are the same for every organisation |