# Extraction report — `turridescartessschismlockes2017`

## What was processed
- Target PDF: `inbox/Turri-DESCARTESSSCHISMLOCKES-2017.pdf` (ingested/processed via `invoke ingest --only "Turri-DESCARTES"` then `invoke process turridescartessschismlockes2017`).

## Source hierarchy used (per instructions)
1) `papers/turridescartessschismlockes2017/out/bundle.json`
2) `papers/turridescartessschismlockes2017/out/tables/*` (none produced; see below)
3) `papers/turridescartessschismlockes2017/out/fulltext.md` and `papers/turridescartessschismlockes2017/out/text.txt`
4) `papers/turridescartessschismlockes2017/out/tei.xml`
5) Rendered page images (required to recover accurate numeric footnotes + bibliographic header where TEI/text were garbled):
   - `papers/turridescartessschismlockes2017/out/images/pages/p001.png` (bibliographic metadata)
   - `papers/turridescartessschismlockes2017/out/images/pages/p008.png` (Experiment 1 N/design)
   - `papers/turridescartessschismlockes2017/out/images/pages/p009.png` (scale + item examples)
   - `papers/turridescartessschismlockes2017/out/images/pages/p010.png` (Experiment 2 N + results context)
   - `papers/turridescartessschismlockes2017/out/images/pages/p011.png` (Experiment 2 item wording)
   - `papers/turridescartessschismlockes2017/out/images/pages/p016.png` (NOTES: sample/recruitment + Exp1 t/Beta for stakes→knowledge)
   - `papers/turridescartessschismlockes2017/out/images/pages/p017.png` (NOTES: Exp2 t/Beta for stakes→knowledge)

## Table extraction gap
- `out/tables/` is empty: both Camelot and Tabula outputs were fully filtered out (see `out/bundle.json`).
- Consequence: numeric results for NOTES and Appendix regression tables had to be recovered from `out/tei.xml` and/or rendered page images.

## Effect size limitations (why `d` and `v` are null)
The paper does **not** report low- vs high-stakes **knowledge** descriptive statistics (means/SDs, or split Ns) for either experiment, and the reported stakes→knowledge results are given only as regression/t statistics in NOTES (Exp1 Note 11; Exp2 Note 31). The **direction** of the low-vs-high knowledge difference cannot be unambiguously determined from the available outputs (coding of the condition variable is not stated, and knowledge means are not reported). Per extraction rules, I did not guess the sign or split Ns, so `effect_size.d` and `effect_size.v` are `null` with `needs_review: true`.

### Required manual fields

[FIELD: exp1_stakes_knowledge_direction]
- Needed to compute signed SMD for Experiment 1.
- Please provide **either**:
  - the mean knowledge score in low vs high stakes (plus SD/SE and group Ns), **or**
  - an explicit statement from the paper/supplement showing which direction the stakes manipulation moves knowledge scores (low > high or high > low) for the overall stakes contrast.

[FIELD: exp2_stakes_knowledge_direction]
- Needed to compute signed SMD for Experiment 2.
- Please provide **either**:
  - the mean knowledge score in low vs high stakes (plus SD/SE and group Ns), **or**
  - an explicit statement from the paper/supplement showing which direction the stakes manipulation moves knowledge scores (low > high or high > low) for the overall stakes contrast.

[FIELD: exp1_split_n_low_high]
- If you can locate split Ns for low vs high stakes in Experiment 1 (collapsed across other factors), provide `n_low` and `n_high`.
- This will enable computing `d` and `v` from the reported `t(600) = −2.85` (Note 11) in a reproducible `analysis/effect_sizes.qmd`.

[FIELD: exp2_split_n_low_high]
- If you can locate split Ns for low vs high stakes in Experiment 2, provide `n_low` and `n_high`.
- This will enable computing `d` and `v` from the reported `t(300) = 0.27` (Note 31) in a reproducible `analysis/effect_sizes.qmd`.