Foundation Term 1 Calibration (Archived)
Archival calibration reference — data quality, methodology, and instrument insights
Archival calibration reference. This page documents an earlier Foundation Term 1 joint IRT + RT calibration. It is retained for methodological background and data-quality evidence. It should not be read as the current operational recommendation. The current model-development summary is in IRT Development: State of Play; current response-process and speed branches are described separately in Response Process and Speed & Profiles.
1 Archival alignment notes
| Stage | N | Explanation |
|---|---|---|
| Raw Foundation Term 1 records | 2,558 | All Foundation A/B students reported in the original review. |
| Included in model | 2,517 | Students covered by the earlier joint IRT + RT calibration after inclusion filters. |
| Excluded / low evidence | 41 | Difference between raw records and model coverage; source-specific exclusion reasons should be checked before reuse. |
| Low-information flagged | 2.4% | Original review flag for sparse evidence / low response or RT counts. |
Older terminology on this page maps to the current A4 language as follows:
| Older term | Current interpretation |
|---|---|
| gap missingness | interior_unseen_or_blank_proxy |
| tail missingness | trailing_nonreach_proxy |
| attempted | valid/scorable response |
| timed-math speed | research interpretation only; not a live band adjustment |
Any wording in the archived calibration materials that sounds like speed is recommended for operational scoring should be read as superseded. For current governance, speed remains response-process evidence only and must not adjust live risk bands without further validation.
2 What this archival page contributes
The original Foundation Term 1 calibration contributed useful background evidence about item quality, response-time data quality, low-information cases, and the feasibility of joint IRT + RT modelling. It is now treated as historical evidence rather than the current model-development recommendation.
3 What has been superseded
- Current achievement-model governance is summarised in State of Play.
- Current row-state terminology is defined in Response Process.
- Current speed and profile governance is defined in Speed & Profiles.
- Operationally, speed remains response-process evidence only unless future validity, fairness, stability, and actionability gates are passed.
4 Original child report
The original child report is stored at analysis/foundation_t1_model_review/foundation_review_report.qmd. It requires optional interactive/plotting dependencies that are not installed in the active render environment. Render the original child report in a technical environment with those dependencies if the historical diagnostic plots are required.