# Extraction report: mayetal2010practicalinterestsrelevant

## Summary
- Pipeline outputs used (preferred sources):
  - Paper metadata (title/DOI/date/abstract): `papers/mayetal2010practicalinterestsrelevant/out/tei.xml`
  - Study design + scale + reported tests: `papers/mayetal2010practicalinterestsrelevant/out/fulltext.md`
  - Extraction index/status: `papers/mayetal2010practicalinterestsrelevant/out/bundle.json`
- Extracted tables:
  - `papers/mayetal2010practicalinterestsrelevant/out/tables/` contained no retained numeric tables (table candidates were filtered out by the pipeline heuristics), so it was not used for numeric extraction.
- Output YAML: `papers/mayetal2010practicalinterestsrelevant/mayetal2010practicalinterestsrelevant.yaml`
- Effect sizes:
  - `s1_e1`, `s1_e2`, `s2_e1` computed from raw-data group summaries in `papers/mayetal2010practicalinterestsrelevant/analysis/effect_sizes.qmd` using established packages:
    - between-groups: `esc::esc_mean_sd`
    - within-subject: `metafor::escalc(measure="SMCRP")` with `r_within` estimated from data.

## Manual interventions / non-pipeline sources (required disclosure)
- External dataset used (provided in project drive):
  - Source directory: `/home/bartosz/Insync/b.mackiewicz@uw.edu.pl/Google Drive/Meta-analizy w filozofii eksperymentalnej/Metaanaliza - stakes, context-sensitivity/Epistemology: Anti-intellectualism and contextualism/DANE_DO_OPRACOWANIA/may`
  - Copied into paper directory:
    - `papers/mayetal2010practicalinterestsrelevant/data/May et al. KPI Data - Between-Subjects (Exp 1).xls`
    - `papers/mayetal2010practicalinterestsrelevant/data/May et al. KPI Data - Within-Subjects (Exp 2).xls`
  - Used to compute: split Ns, means, SDs, within-person correlation (Study 2), and all `d`/`v`.
- No direct PDF or rendered page-image consultation was needed for extraction (fulltext + TEI + data files were sufficient).

## Notes / potential limitations
- Study 2 contrast (LS-NA vs HS-A) changes both stakes and whether an alternative is mentioned (only HS-A mentions an alternative). This is documented in the effect’s `contrast.other_notes`.
- For Study 2 reported tests, p-values are reported as thresholds (p<.001); `reported_test.p` is set to `null` with the threshold preserved in `reported_test.notes`.

## Validation / rendering
- YAML schema validation: `papers/mayetal2010practicalinterestsrelevant/mayetal2010practicalinterestsrelevant.yaml` validates against `docs/stakes_meta_schema.json` (via `jsonschema`).
- Quarto computation record: `papers/mayetal2010practicalinterestsrelevant/analysis/effect_sizes.html` rendered from `papers/mayetal2010practicalinterestsrelevant/analysis/effect_sizes.qmd`.
- HTML extraction report: `papers/mayetal2010practicalinterestsrelevant/mayetal2010practicalinterestsrelevant.html` generated via `invoke render-extraction mayetal2010practicalinterestsrelevant`.

## Package recompute check
- Relative to the previous hand-written formulas:
  - `d` values did not change for any effect.
  - `v` changed slightly for between-subject effects only:
    - `s1_e1`: `0.033159390561 -> 0.033157749548` (delta `-0.000001641013`)
    - `s1_e2`: `0.033938271605 -> 0.033928189300` (delta `-0.000010082305`)
  - `s2_e1` (within-subject) is unchanged (`d=0.388549287526`, `v=0.002733530591`).