docs(solutions): domain decomposition reveals hidden Overton variance

Aggregate centrist support surge masks two distinct mechanisms:
strategic moderation (85%) and migration acceptance expansion (15%).
Party-level: CDA+ChristenUnie drive the shift, D66 barely moves.
Compounding note for future Overton/domain analysis work.

docs/solutions/best-practices/domain-decomposition-overton-analysis.md

@@ -0,0 +1,64 @@
+---
+module: analysis/right_wing
+problem_type: best-practice
+date: 2026-05-25
+tags:
+  - overton-window
+  - domain-decomposition
+  - mp-level-analysis
+  - party-disaggregation
+  - migration-policy
+  - centrist-voting
+category: best-practices
+---
+
+# Domain Decomposition Reveals Hidden Variance in Overton Window Analysis
+
+## Context
+
+Aggregate centrist support for right-wing motions surged post-2024 (d=+0.65). But the aggregate masked two distinct stories: one of strategic moderation (85% of motions, non-migration domains) and one of genuine acceptance expansion (15%, migration domain). Without domain decomposition, the analysis would incorrectly claim either "no shift" or "complete shift" when both processes occurred simultaneously.
+
+## Guidance
+
+**Always decompose political shift analyses by policy domain before drawing conclusions.** The aggregate number answers a different question than each domain-specific signal.
+
+**What to check per domain:**
+
+1. **Volume change:** Did motion count increase or decrease?
+2. **Material impact change:** Did the policy substance intensify or moderate?
+3. **Centrist support change:** Did centrist parties accept more of these motions?
+4. **MP-level disaggregation:** Which specific centrist parties drive the shift?
+
+**The migration exception (2024 Dutch parliament):**
+- Material impact barely changed (3.26→3.13)
+- Yet centrist support more than doubled (0.153→0.369)
+- M=5 support went from 0.000 to 0.185
+- This is the one domain where acceptance genuinely expanded
+
+**All other domains (climate, nitrogen, economy, defense, etc.):**
+- Material impact significantly declined (2.72→2.30)
+- High-impact share collapsed (20.8%→8.0%)
+- M=5 motions: pre-2024 "exit Paris accord" (CS=0.0), post-2024 "build nuclear plants" (CS=1.0)
+- Strategic moderation explains the shift
+
+**Party-level disaggregation:**
+- CDA and ChristenUnie (Christian-conservative centrists) more than doubled migration vote share
+- D66 (secular-progressive) barely moved from single digits
+- The shift is not "centrists" generally — it's a specific faction within the center
+
+## Why This Matters
+
+Without domain decomposition, an analyst would report "centrist support for right-wing motions increased" and either: (a) claim the Overton window shifted right (wrong — 85% of motions show content moderation, not acceptance expansion), or (b) claim the Overton window didn't shift (wrong — migration is the 15% where it demonstrably did). Domain decomposition prevents both errors and reveals the more honest picture: strategic moderation was the dominant mechanism, with migration as the sole measurable exception.
+
+## When to Apply
+
+- Before drawing conclusions from aggregate voting trend data
+- When policy domains have different baseline extremity levels (migration already had M=5 motions; climate gained them post-Ukraine energy crisis)
+- When centrist parties are ideologically heterogeneous (Christian-democrats ≠ progressive-liberals)
+- When the target audience is academic or policy-oriented and will spot domain conflation
+
+## Examples
+
+**Bad:** "Centrist support for right-wing motions surged from 0.25 to 0.51 — the Overton window shifted."
+
+**Good:** "Centrist support for right-wing motions surged from 0.25 to 0.51. Domain decomposition reveals two mechanisms: strategic moderation across 85% of motions (content became milder, centrists rewarded moderation), and genuine acceptance expansion in the migration domain (content impact stable, centrist support doubled)."