# L7 — Construct comparability matrix (no redefinition)

**Question:** For each **existing** construct, can we measure it comparably across the pilot without changing the construct definition?  
**Fields per construct:** equivalent source · wording/measurement equivalence · temporal coverage · scale comparability · country-specific calibration needs · must-remain-UNKNOWN.

Dispositions are **measurement-feasibility only**. No scores. No transitions.

---

## L1 — Public / institutional trust (US production construct L1-US-v0.1)

| Field | Assessment |
|-------|------------|
| Equivalent source | US: Pew/OWiD trust series (landed). Pilot peers: **WVS** confidence-in-government / institutions; **OECD** trust-in-government indicators; **ESS** trust-in-parliament/politicians (UK/DE); Latinobarómetro/LAPOP for Brazil; KGSS/WVS for Korea |
| Wording / measurement equivalence | **PARTIAL.** “Trust in government/institutions” family is shared, but item batteries differ (single item vs multi-institution; 4-pt vs 11-pt). WVS/ESS not byte-equivalent to Pew US series |
| Temporal coverage | WVS waves sparse (multi-year). ESS biennial (EU). OECD trust indicators more recent / uneven start years. **Not** monthly like some US polls |
| Scale comparability | **Requires explicit crosswalk** (not a silent rescale). Do not claim 0–100 US L1 bands apply abroad |
| Country-specific calibration needs | **YES** — response-style / institutional-referent differences (presidential “government” vs parliamentary cabinet; federal vs unitary referents) |
| Must remain UNKNOWN | **International L1 score / Rigidity Index abroad** — production authorization is **L1-US-v0.1 only**. Cross-country L1 **levels** remain UNKNOWN |
| Portability disposition | **MODIFY** (comparable *family*, not drop-in series) |

---

## M1 — Macroeconomic / consumer perception (research-only; score_authorized=0)

| Field | Assessment |
|-------|------------|
| Equivalent source | US: Michigan SCA / FRED UMCSENT (landed). Peers: **OECD Consumer Confidence Index (CCI)** for US/UK/DE/KR (and BR where OECD covers); national stats offices as secondary |
| Wording / measurement equivalence | OECD CCI is a **harmonized** presentation of national CCIs — closer to “comparable” than trust items, but underlying national questionnaires still differ |
| Temporal coverage | Monthly OECD CCI for core OECD members — **best temporal match** in the pilot for a perception construct |
| Scale comparability | OECD long-term average = 100 style indices — **not** identical to ICS 1966=100 Michigan; needs documented transform, not construct redefine |
| Country-specific calibration needs | Mild — mainly index base and component weights; document, don’t retune to outcomes |
| Must remain UNKNOWN | Any M1 **score**, Phase coupling, or cross-country “sentiment risk” |
| Portability disposition | **KEEP_CANDIDATE** for *harmonized CCI-style measurement* · **MODIFY** for Michigan-micro identical items abroad |

---

## S1 — Affective tribalization / out-party affect (parked exploratory)

| Field | Assessment |
|-------|------------|
| Equivalent source | US: ANES out-party FT (landed). Peers: **CSES** like-dislike / FT analogs; BES/GLES party affect; ESS party closeness (weaker); LAPOP/Latinobarómetro for Brazil |
| Wording / measurement equivalence | Feeling-thermometer 0–100 is **sometimes** present (CSES/ANES family); elsewhere like-dislike 0–10 or ordinal favorability — **not equivalent without crosswalk** |
| Temporal coverage | Election-study cadence (irregular); poorer than monthly macro |
| Scale comparability | 0–100 vs 0–10 needs declared mapping; party-system multiparty → “out-party” definition **must not silently redefine** S1’s two-party US operationalization |
| Country-specific calibration needs | **HIGH** in multiparty systems (UK/DE/BR): which out-party? mean of non-preferred? largest rival? — choosing an aggregator is a **measurement decision**, risking construct drift |
| Must remain UNKNOWN | Multiparty “out-party FT gap” as if it were US two-party S1; any S1 score |
| Portability disposition | **MODIFY** (two-party FT gap) · **INSUFFICIENT** if forced identical two-party definition onto multiparty systems without a frozen mapping that preserves S1 meaning |

---

## E2 — Elite fragmentation / legislative separation (parked; N3/N4 US)

| Field | Assessment |
|-------|------------|
| Equivalent source | US: Voteview NOMINATE N3 + CBS N4 offer/attract (landed). Peers: **CHES** expert placements (EU); **Manifesto/CMP** left-right; **ParlGov** party metadata; no Voteview abroad |
| Wording / measurement equivalence | NOMINATE roll-call ideology ≠ CHES expert salience/placement ≠ CMP manifesto scores. **Same construct family only at high abstraction** (“elite ideological separation”) |
| Temporal coverage | CHES intermittent waves; CMP election-timed; not Congress-session identical |
| Scale comparability | Different latent scales — **no silent NOMINATE units abroad** |
| Country-specific calibration needs | **HIGH** — chamber rules, party discipline, and cosponsorship norms differ; N4 offer/attract CBS is **US procedural** and not portable as-is |
| Must remain UNKNOWN | E2 score; CBS offer/attract outside US House/Senate; NOMINATE-equivalent claims without independent estimation protocol |
| Portability disposition | **MODIFY** for CHES/CMP *distance-between-parties* research sensors · **INSUFFICIENT** for N4 CBS portability · **INSUFFICIENT** for Voteview drop-in |

---

## E3 — Elite defection velocity (L3 ontology; no score)

| Field | Assessment |
|-------|------------|
| Equivalent source | US: party-switching / Record events (L3 KEEP narrow). Peers: legislative biographies / party-switch datasets (sparse); news event coding |
| Wording / measurement equivalence | Party switching is relatively clear; other L3 types need dictionaries per country |
| Temporal coverage | Rare events everywhere — monthly rates remain dishonest (L3 verdict) |
| Scale comparability | Exposure-normalized annual counts may compare; rates fragile |
| Country-specific calibration needs | Party system fluidity (BR vs UK) changes base rates without meaning “more defection construct” |
| Must remain UNKNOWN | Monthly E3 velocity scores internationally |
| Portability disposition | **MODIFY** (narrow party-switching only) · **INSUFFICIENT** for full US L3 ontology abroad this pilot |

---

## L4 — Boundary permeability (elite legislative reciprocity ≠ S2)

| Field | Assessment |
|-------|------------|
| Equivalent source | US: CBS offer/attract (KEEP). Peers: cross-party cosponsorship / committee cooperation where recorded; mass bridging surveys (ESS social trust is **S2**, not L4) |
| Wording / measurement equivalence | Procedural “offer/attract” is US-centric. Cross-party amendment/cosponsor analogs need country codebooks |
| Temporal coverage | Sparse outside US CEL/CBS lineage |
| Scale comparability | Low without shared denominators |
| Country-specific calibration needs | **HIGH** |
| Must remain UNKNOWN | L4 score; treating ESS generalized trust as L4 permeability |
| Portability disposition | **INSUFFICIENT** for drop-in L4 · **MODIFY** only for carefully coded cross-party legislative traffic (separate sensors) |

---

## B1/B2 — Conflict / disorder (ACLED stub)

| Field | Assessment |
|-------|------------|
| Equivalent source | **ACLED** event data (landing HTTP200) — designed for cross-country event coding |
| Wording / measurement equivalence | ACLED codebook is shared; still needs country subsetting rules (admin1, event types) frozen in advance |
| Temporal coverage | Daily→monthly feasible where ACLED covers; US/UK/DE/BR/KR coverage must be verified per year **before** use (availability matrix, not outcomes) |
| Scale comparability | Event counts / population exposure — document; don’t redefine B1/B2 to fit sparse years |
| Country-specific calibration needs | Reporting density and protest law differences |
| Must remain UNKNOWN | B1/B2 scores; using ACLED to backdoor transition labels |
| Portability disposition | **KEEP_CANDIDATE** for *shared-codebook event counts* · coverage matrix still OPEN |

---

## PRD / CTH — Commitment threshold (backlog)

| Field | Assessment |
|-------|------------|
| Equivalent source | US leads: SCA/SCE. Peers: national expectation surveys uneven |
| Portability disposition | **INSUFFICIENT** this pilot (backlog) — remain UNKNOWN abroad |
| Must remain UNKNOWN | Any PRD/CTH international score |

---

## Exp001 / Exp002

| Construct | Portability |
|-----------|-------------|
| Exp001 Bloc Coupling | **INSUFFICIENT / CLOSED** — do not revive internationally |
| Exp002 Recovery Ratchet | **INSUFFICIENT — do not rescue** internationally or domestically |

---

## ELECTION-01

US-only PIT vote-margin protocol. **Not** in L7 pilot scope for portability scoring. Remain separate.