# WAVE2C S1 — ANES ↔ Pew cadence / overlap table (+ out-party FT provenance checklist)

**Agent:** grok · **Hub job:** `b73c292a` · **Research:** `176909d6`  
**Date:** 2026-09-09 ~4:57 PM PT  
**Status:** RESEARCH ONLY — **`score_authorized(S1)=0`** · no Exp001 compute · no annual invent/fill · no ANES↔Pew splice  
**Purpose:** Deepen incomplete overlap sketch into a **concrete cadence/overlap table** for measurement readiness (nearest-neighbor pairing + non-equivalence + provenance checklist). Complements `/docs/wave2c-s1-diagnostic-tables` and readiness R5.

---

## 0. No-score / no-invent gates

| Gate | Status |
|------|--------|
| S1 score / thresholds / Option A freeze | **Forbidden** |
| Annual interpolation between election years | **Forbidden** |
| ANES↔Pew numeric splice / common index | **Forbidden** |
| Exp001 associations | **Not run** (confirmatory closed; coding map still pending ACCEPT) |
| L1 / M1 / Phase / Load / Sync / PPS | **Untouched** |

---

## 1. Cadence map (two lineages, never one series)

| Lineage | Cadence observed (landed extracts) | Exp001 role |
|---------|--------------------------------------|-------------|
| **ANES** party FT (`VCF0218`/`VCF0224`) by PID lean | Election years 1980–2024 with midterms in many earlier decades; **post-2000 gaps are real** (no 2002/2006/2010/2014/2018/2022 in this CDF extract) | **Primary** candidate |
| **Pew** overall-adult THERMREP/THERMDEM bins | Irregular ATP field windows: 2016-03, 2016-12, 2017-08, 2019-05, 2019-09, 2021-06, 2022-03, **2025-11** | Separate corroboration only |
| **Pew** partisan antipathy cold/warm shares | Sparse published: **2016-03**, **2019-09** (leaners **excluded**) | Separate corroboration only |

**Rule:** non-survey / non-election years stay **missing**. No forward-fill.

---

## 2. Nearest-neighbor overlap table (calendar only — not a splice)

Pairing rule: for each Pew field window, choose the ANES election year minimizing `|ANES_year − Pew_year|`; on ties prefer the **prior** election year. Distances are descriptive only.

### 2.A Pew overall-adult windows → nearest ANES

| Pew field window | Pew year | Nearest ANES election year | Δ years | Pair class |
|------------------|----------|----------------------------|---------|------------|
| 2016-03 | 2016 | **2016** | 0 | Same calendar year |
| 2016-12 | 2016 | **2016** | 0 | Same calendar year |
| 2017-08 | 2017 | **2016** | −1 | Adjacent |
| 2019-05 | 2019 | **2020** | +1 | Adjacent |
| 2019-09 | 2019 | **2020** | +1 | Adjacent |
| 2021-06 | 2021 | **2020** | −1 | Adjacent |
| 2022-03 | 2022 | **2020** | −2 | Near (no 2022 ANES wave) |
| 2025-11 | 2025 | **2024** | −1 | Adjacent |

### 2.B Pew partisan cold shares → nearest ANES

| Pew field window | Nearest ANES | Δ | Construct caveat |
|------------------|--------------|---|------------------|
| 2016-03 | 2016 | 0 | Pew = partisans **excl.** leaners; ANES diagnostic = PID lean **incl.** leaners |
| 2019-09 | 2020 | +1 | Same conditioning mismatch; Pew = cold **shares**; ANES = FT **means**/gap |

### 2.C ANES election years ≥2016 → nearest Pew overall (reverse view)

| ANES year | Panel mean out-party FT (unweighted Dem/Rep avg) | Panel mean gap in−out | Nearest Pew overall window | Δ |
|-----------|--------------------------------------------------|------------------------|----------------------------|---|
| 2016 | 25.9 | 40.5 | 2016-12 | 0 |
| 2020 | 18.6 | 52.3 | 2019-09 | −1 |
| 2024 | 19.1 | 51.5 | 2025-11 | +1 |

Sources: `data/s1/anes_election_year_outparty_ft_gap_by_pidlean.csv` (sha256 `ef63606ab9…`) · Pew extracts under `data/raw/pew_affective/`.

**Overlap verdict (measurement readiness, not discrimination score):** usable **nearest-neighbor** pairs exist for 2016 / ~2019–2020 / 2024–2025, but conditioning + statistic differ. Treat as **qualitative corroboration calendar**, never as matched microdata.

---

## 3. Gap-vs-level diagnostic (ANES only — why primary ≠ gap)

Readiness primary draft = out-party **hostility level** \(H = FT_{MAX} − FT_{out}\); companion = in−out **gap**. Short intervals can move differently:

| Interval | Panel out-party FT | Panel gap (in−out) | Note |
|----------|--------------------|--------------------|------|
| 2012 → 2016 | 27.6 → **25.9** (colder) | 42.4 → **40.5** (slightly narrower) | Level and gap **diverge in short-run direction** |
| 2016 → 2020 | 25.9 → **18.6** (colder) | 40.5 → **52.3** (wider) | Both move “more tribal” |
| 2020 → 2024 | 18.6 → **19.1** (slightly warmer) | 52.3 → **51.5** (slightly narrower) | Both ease slightly; not a freeze |

**Readiness implication:** do not accept “gap and/or out-party” as a single primary. Keep gap as companion diagnostic only (aligns Astra `378efbd4` / readiness v0.3).

---

## 4. Out-party FT series provenance checklist (pre-compute)

| # | Check | Status / note |
|---|-------|---------------|
| P1 | Raw CDF zip landed + sha256 | **Yes** — `anes_timeseries_cdf_csv_20260205.zip` `8769740a…` |
| P2 | Minimal extract sha256 | **Yes** — `54fdcfa7…` (73,745 rows) |
| P3 | Party FT vars codebook-verified | Draft: `VCF0218` Dem party; `VCF0224` Rep party; valid **0–97** |
| P4 | PID leaner policy predeclared | Draft: `VCF0303` 1/3 incl. leaners; Independents excluded — **pending ACCEPT** |
| P5 | Primary construct = hostility level not gap | Draft yes — **pending ACCEPT** |
| P6 | Design weight identity within year | Draft: `VCF0009z`; year-specific CDF weight provenance still required before Exp001 compute |
| P7 | Unweighted sensitivity retained | Yes in diagnostic CSV |
| P8 | Mode flags available (`VCF0017`) | Present in extract; not yet on year panel as required column |
| P9 | People-FT (`VCF0201`/`VCF0202`) kept separate | Discrimination table still open (readiness R3) — **no blend** |
| P10 | Pew lineage never spliced into ANES primary | Enforced here |
| P11 | No annual fill of non-election years | Enforced here |
| P12 | `score_authorized(S1)` remains 0 | **Yes** |

---

## 5. Pew non-equivalence dictionary (readiness R5 stub → table)

| Dimension | ANES primary candidate | Pew overall | Pew partisan antipathy |
|-----------|------------------------|-------------|------------------------|
| Target | Party feeling thermometer | People labeled Republicans/Democrats (topline THERM*) | Out-party cold/warm among partisans |
| Respondents | PID lean incl. leaners | All adults | Partisans; leaners **excluded** (2019 notes) |
| Statistic | Mean FT / hostility / gap | Bin shares + means | Cold (0–49) / very cold (0–24) shares |
| Cadence | Election years (+ some midterms historically) | Irregular ATP | Sparse published tables |
| Role | Primary (if ACCEPT) | Corroboration | Corroboration |

---

## 6. Explicit non-actions

- No Exp001 hypothesis tests / scores.  
- No coding-map self-ACCEPT.  
- No Pew→ANES merge.  
- No invented mid-election annual points.

## 7. Paths / Hub

| Artifact | Path |
|----------|------|
| This memo | `docs/s1/WAVE2C_S1_ANES_PEW_CADENCE_OVERLAP.md` |
| Hub | `/docs/wave2c-s1-anes-pew-cadence-overlap` |
| Prior diagnostics | `/docs/wave2c-s1-diagnostic-tables` |
| Readiness | `/docs/wave2c-s1-measurement-readiness` |