# Extraction Report: Białka 2022

## Source Crosswalk

- Thesis PDF: `pdf/paper.pdf`
- Text extraction: `out/text.txt`, `out/fulltext.md`
- Raw dataset: `data/daneZBlic.csv`
- Manual numeric analysis: `data/ZBiałka_Analiza danych_metaanalizy.rmd`
- Manual result workbook: `data/ZBiałka_tabele2.xlsx`
- Reproducible computation: `analysis/effect_sizes.qmd`

The thesis is treated as the source of truth for the research question, materials, procedure,
question wording, and moderator coding. The local dataset and manual analysis script are
treated as the source of truth for exclusions, group descriptives, and effect sizes.

## Extraction Decisions

- Included primary effects: three same-scenario high-vs-low stakes contrasts:
  `Wypracowanie`, `Karta`, and `Bank`.
- Excluded from primary extraction: the separate `dowód` / contrary-evidence condition.
- Excluded from primary extraction: the scenario-confounded paired comparisons in the
  manual Rmd, because the high and low conditions compare different scenarios.
- Effect sign convention: `d = mean(low) - mean(high)`.
- Effect computation: group means/SDs from `daneZBlic.csv`; SMD via
  `esc::esc_mean_sd`.

## Issues For Review

### HUMAN_CHECK_BIALKA_EXCLUSIONS

The thesis reports 186 participants and 51 attention-check exclusions. The local Rmd says
LimeSurvey exported 204 complete responses. The actual exclusion vector in the Rmd contains
52 IDs, because IDs `209` and `320` are included in the vector even though the nearby
`length(...)` check counts only 50 IDs. The QMD follows the actual exclusion vector and
uses final `N = 152`.

Required human answer: confirm whether the 52-ID exclusion vector is the intended source of
truth, or provide the corrected exclusion list.

### HUMAN_CHECK_BIALKA_PRIMARY_CONTRASTS

The extraction uses only same-scenario high-vs-low stakes contrasts. The `dowód` condition
is not used as a stakes contrast, and scenario-confounded paired comparisons are excluded.

Required human answer: confirm that this matches the intended meta-analytic unit.

## Validation Notes

The YAML contains local provenance for dataset-derived values because these values do not
come from PDF tables. Thesis page provenance is used for materials, sample description, and
question wording.