Tag Integration — Catalog & Sign-Off

This folds the review into one place: the proposed tag system as it would sit on the Party Affiliation sheets, and the eight points where the new definitions differ from what's already built. Nothing is built yet — tick the boxes and add comments, export to MD, and that's the build order.

1 — Proposed tags & counts (for reference)

Tags layer onto the seven Party Affiliation categories as columns. Counts use the year-range reading (see decision 6 — the date-based reading shifts a few slightly). Active voters only.

2 — Sign-off decisions

Tag	Sub-value	Formula	Count
Recency started voting 2020+	Recent Democrat	`no D/R 2008–2018; ≥1 D 2020/22/24/26; no R 2020/22/24/26`	4,883
Recent Republican	`no D/R 2008–2018; ≥1 R 2020/22/24/26; no D 2020/22/24/26`	12,622
Recent Independent	`no D/R/X 2008–2018; ≥1 X 2020/22/24/26; no D/R 2020/22/24/26`	1,838
Recent General-Only	`no primary ever; no general before 2020; ≥1 general 2020/22/24`	21,840
Drop-off stopped 2020+	Drop-off Democrat	`≥1 D 2008–2018; no R 2008–2018; no D/R 2020/22/24/26`	7,606
Drop-off Republican	`≥1 R 2008–2018; no D 2008–2018; no D/R 2020/22/24/26`	12,674
Drop-off Independent	`no D/R 2008–2018; ≥1 X 2008–2018; no D/R/X 2020/22/24/26`	725
Drop-off General-Only	`no primary ever; ≥1 general 2008–2018; no general 2020/22/24`	1,547
Swapper	Heimlich	`≥1 D before 2024; no R before 2024; R in 2024; D in 2026`	205
Anti-Trump	`≥1 R 2008–2016; no D 2008–2016; ≥1 D 2018–2026; no R 2018–2026`	1,237
Engaged	Engaged	`voted in every general 2008–2024 (all 9)`	32,721

1. Drop-off Dem/Rep now exclude past-mixed voters changes counts

The "no opposite party in 2008–2018" clause you added is the tie-breaker we'd flagged. It removes 2,236 voters from each pull (Dem 9,842→7,606, Rep 14,910→12,674). Those 2,236 — who pulled both D and R in 2008–2018 and then went quiet — now get no drop-off tag; they sit untagged in Mixed.

Approve — leave the 2,236 untagged in Mixed Instead, assign them by majority party (more D ballots → Drop-off Dem)

2. Engaged broadened from niche pull to turnout tag meaning change

Old Engaged Independents (440) + Engaged Unaffiliated (286) = 726, both restricted to non-primary voters. Your new Engaged = anyone who voted all 9 generals = 32,721, spanning every category (Strong Dems, Solid Reps, all of them). The old two become just Engaged ∩ Independent and Engaged ∩ NonPartisan.

Approve — broad turnout tag (32,721) across all categories Keep the old restricted Engaged Independents / Unaffiliated instead

3. Swappers redefined into named patterns replaces old pull

Old Swappers (D-2022 / R-2024 / D-2026; ~230–270) → your Heimlich (≥1 D ever before 2024, no R before 2024, R-2024, D-2026; 205). Different rule — Heimlich forbids any prior R and accepts any prior D, not only 2022. Plus the new Anti-Trump (1,237). The two don't overlap (0).

Approve — replace old Swapper with Heimlich + Anti-Trump Also keep the original Swapper as a third pattern

4. Renames and two brand-new sub-tags

• Old "Recent Unaffiliated" = your Recent General-Only (identical logic, 21,840). Pure rename.
• Old "Drop-Off Unaffiliated" → Drop-off General-Only, but the past-activity window moved from 2008–2016 to 2008–2018 (1,479→1,547).
• Recent Independent (1,838) and Drop-off Independent (725) are new — issue-only voters never had recency/drop-off treatment before.

Approve the renames (Recent/Drop-off General-Only) Approve the 2008–2018 window change for Drop-off General-Only Approve the new Independent sub-tags

5. Typo confirmation

Your spec for Drop-off General-Only reads "no general in 2022/22/24." I read that as 2020/22/24.

Confirm — it's 2020/22/24 No — I meant something else (see comment)

6. Dates vs. year-ranges affects all counts

Existing pulls were built on the ten specific primary dates (e.g., 3/17/2020). Your new wording reads as year ranges. In presidential years and special-primary years, "by year" sweeps in primaries beyond the ten dates — which is why Recent Democrat moves 4,800 (dates) → 4,883 (years). Both workbooks should use one convention.

Ten specific dates — match what the categories already use (recommended) Full year-ranges — count every primary in the year, incl. specials

7. The 2018 boundary isn't used consistently

For Recency and Drop-off, 2018 sits in the "before" window (2008–2018). But Anti-Trump puts 2018 on the "after" / flip side ("≥1 D in 2018–2026"). That may be intentional (Trump-era flip starting 2018), but 2018 is doing opposite duty in different tags.

Keep as designed (intentional) Reconcile — make 2018 consistent (see comment)

8. "Started voting in 2020" vs. the formula

The Recency tag's stated intent is "started voting in primaries/generals in 2020+," but Recent Dem/Rep/Ind only look at primary history. So a voter who's voted every general since 2008 yet first pulled a D primary in 2020 gets tagged Recent Democrat and also qualifies as Engaged — odd against "new since 2020." Recent General-Only already requires "no general before 2020"; the three partisan/independent recency tags don't.

"Recent" = recent partisan-primary participation (ignore generals — current logic) "Recent" = genuinely new voter — add "no general before 2020" to the partisan recency tags too

3 — Consistency that already works (no action needed)

The tags nest cleanly under the categories: Recent/Drop-off Dem ⊂ Solid Dems · Rep ⊂ Solid Reps · Independent ⊂ Independent · General-Only ⊂ NonPartisan · both Swappers ⊂ Mixed · Engaged spans all. Within a voter, Recency / Drop-off / Swapper are mutually exclusive by construction (you can't both start and stop; swappers stay active on both sides) — only Engaged stacks, so the four tag columns never collide. One consequence: on the Strong Dems and Strong Reps sheets only the Engaged column can ever fill in; Recency/Drop-off/Swapper are structurally impossible there. (That's why the screenshot's Strong Reps + Anti-Trump row can't actually occur — placeholder, as you noted.)

4 — Anything else

Overall comments / additional swapper patterns / build-order notes

You mentioned the Swapper tag can grow if there's a meaningful pattern — note any here, plus anything else for the build.