# L7 — Country selection (data availability BEFORE outcomes)

**Rule:** Justify the pilot set by **source coverage / measurement infrastructure**, not by crisis history, regime transitions, or any performance look.  
**MODEL_CHANGE=NO.** No country risk scores.

## Recommended pilot set (5)

| Country | Region / system family | Why included (availability, not outcomes) | Substitute rule |
|---------|------------------------|-------------------------------------------|-----------------|
| **United States** | Presidential · federal · Anglo | Anchor: L1 live series, ANES/SCA/Voteview/CBS already landed in observatory | Required anchor — not substituted |
| **United Kingdom** | Parliamentary · unitary · Anglo | BES + ESS + OECD CCI/trust landings HTTP200; Manifesto/ParlGov party systems | If BES microdata license blocks: keep UK via ESS+OECD only for overlapping constructs |
| **Germany** | Parliamentary · federal · continental EU | GLES + ESS + CHES + ParlGov + OECD landings HTTP200; strong expert/party metadata | If GLES license blocks: ESS+CHES+OECD retain EU continental cell |
| **Brazil** | Presidential · federal · Global South | Distinct presidential/federal cell outside OECD-core surveys; ACLED coverage class; Latinobarómetro/LAPOP/CSES modules historically present for affect/trust analogs | **Landing caveat this pass:** CESOP host DNS NXDOMAIN; treat ESEB as **OPEN_PREFERRED** via CSES/LAPOP/Latinobarómetro — do **not** drop Brazil solely for one dead host |
| **South Korea** | Presidential · unitary · East Asia | Non-Western OECD democracy with national election-study tradition + OECD CCI; fills Asia cell | **Landing caveat:** knep.or.kr TLS self-signed this pass; prefer KGSS / Statistics Korea / CSES Korea module landings as substitutes **for the same constructs** — country kept |

## Rejected / deferred (this pilot)
- Full EU-27 panel (too large; ESS already represented via UK/DE)  
- China / Russia / non-competitive systems (construct definitions for elite party systems break without redefinition — out of scope)  
- India (election-study density uneven vs BES/GLES/ANES; defer)  
- France (redundant with DE for continental parliamentary once ESS/CHES present)

## Heterogeneity axes covered (design, not outcomes)
1. Presidential vs parliamentary  
2. Federal vs unitary  
3. Anglo vs continental EU vs Latin America vs East Asia  
4. OECD consumer-confidence coverage vs partial  
5. Election-study microdata traditions vs expert/party datasets  

## Non-justifications (forbidden here)
- “Interesting transitions,” coups, polarization episodes, forecast difficulty, media salience.

## Landing probe pass (metadata only)
See `data/lanes/L7/source_landings_v0.json`. HTTP200 landings: WVS, ESS, CSES, OECD trust, OECD CCI, ParlGov, CHES, Manifesto, ACLED, BES, GLES, V-Dem, EIU page.  
Open/blocked landings this pass: CESOP/ESEB DNS; KNEP TLS — **do not** reinterpret as construct failure.