Check whether an instrument score crosswalk applies

Distinguish a link from a unit conversion

Two questionnaires can measure related concepts without having interchangeable raw scores. A score crosswalk is based on evidence linking particular instruments under particular conditions. It is not equivalent to converting centimetres to metres, and a shared score range does not make two measures comparable.

The instrument developers' HealthMeasures services page provides access to PROsetta Stone crosswalk resources and distinguishes free tables from optional services. Use the actual linking documentation for the instruments involved rather than inventing a linear transformation from their minimum and maximum scores.

Check the scope of the linking study

Record the exact instrument names, versions, language, respondent type and scoring method. Confirm that the crosswalk covers the relevant score range and population. Read whether the evidence supports group-level comparisons, individual interpretation or both, and what uncertainty remains after linking.

Check that the constructs genuinely align. A physical-function score and a depression score do not become the same quantity because both use a T-score metric. Preserve the original score, linked score and crosswalk version in separate fields so the transformation can be reversed or audited.

Hypothetical example

A fictional dataset contains scores from version A of a questionnaire, while an available crosswalk was developed for version B with different items. The analyst does not apply the table solely because both versions produce totals from zero to 40. The team checks the developers' evidence and either obtains a supported link or retains separate analyses.

This example illustrates a compatibility decision. It does not assert that every version change invalidates linking; the conclusion depends on the documented calibration and validation evidence.

Validate the implementation and interpretation

Test several supported scores against the published table, including boundary values and missing responses. Do not interpolate or extrapolate beyond the documented procedure without a justified method. Record how impossible raw scores and unsupported versions are rejected.

Report the source of the crosswalk, the original and target metrics and the limits of comparability. Use the instrument-scoring contract to preserve prerequisites and the analysis test-oracle worksheet for table checks. A crosswalk can make a comparison possible, but it does not erase measurement differences or establish validity in a population that was not adequately represented in the supporting evidence.

Source verification: 6 November 2026. Examples are hypothetical.