Index Framework | EZstates Elections

Index Framework

The mathematical definitions of every index used across EZstates Elections — P, CP, SP, V, and the Key–Campbell classification scalar W(t).

← Back to Elections Hub
Core Indices

P  Partisanship Index

P = (D − R) / 100

Where D = % of group voting Democratic, R = % voting Republican. Ranges from −1 (fully Republican) to +1 (fully Democratic). Zero means the group is evenly split.

  • P > 0 — group leans Democratic
  • P < 0 — group leans Republican
  • |P| > 0.4 — strongly partisan (reliable base)
  • |P| < 0.1 — centrist (persuasion target)

CP  Critical Power (Partisan Power)

CP = group_share × |P|

Measures how much a group contributes to the stable partisan margin of an election. A group must be both large and strongly partisan to score high. Groups with high CP anchor a candidate's base — the strategic imperative is mobilization.

  • group_share — fraction of the electorate (0–1)
  • |P| — absolute partisan lean
Example: Black voters in Georgia — large share (~30%), very high |P| (~0.85) → high CP. Decisive through mobilization.

SP  Swing Power (Volatility Power)

SP = group_share × V_total

Measures how much a group contributes to election-to-election volatility. A group must be both large and volatile across cycles to score high. Groups with high SP decide close elections — the strategic imperative is persuasion.

  • group_share — fraction of the electorate (0–1)
  • V_total — combined volatility score (see below)
Example: White voters in Michigan — very large share (~75%), centrist P (low CP), but high V_total → dominant SP. Deciding elections through defection, not loyalty.

V  Volatility Indices

V_time = std(P) across years, normalized V_space = std(P) across states, normalized V_total = √(V_time² + V_space²) / √2

Three complementary volatility measures computed across the full group × state × year dataset:

  • V_time — how much a group's P fluctuates across election years (temporal instability)
  • V_space — how much a group's P varies across states in the same year (geographic dispersion)
  • V_total — Euclidean combination; the primary volatility signal used in SP
Classification Scalar

W(t) — Key–Campbell Election Classification

W(t) = Σ [ CP_g(t) × |ΔP_g(t)| ] / Σ CP_g(t)

A CP-weighted mean of absolute partisan shift across all groups between election t−1 and t. Higher W(t) = more coalition movement = more transformative election.

  • ΔP_g(t) = P_g(t) − P_g(t−1) for group g
  • CP_g(t) = weight (larger, more partisan groups count more)
W(t) RangeElection TypeInterpretation
W < 0.04MaintainingNormal vote holds; no significant coalition movement
0.04 ≤ W < 0.08DeviatingTemporary shift; existing alignment largely intact
0.08 ≤ W < 0.14ConvertingSubstantial realignment of key groups
W ≥ 0.14RealigningDeep, durable shift in partisan coalition structure
Decisive Switch Criterion

cand_won_due_to_group — Binary Decisive Test

decisive = True if |margin_state| < dem_group_votes_grew_by AND vo_key_type = 'Switcher' AND vo_key_direction = 'Rep→Dem' (for Dem winner)

A switcher group is flagged as individually decisive when the candidate's margin of victory in the state is smaller than the estimated Democratic vote gain attributable to that group's switch. In other words: without this group's switch, the candidate would have lost the state.

★ marks on switcher maps and state profiles indicate decisive switches — groups whose movement alone was sufficient to flip the state's outcome.
📄 Read the full Methodology Paper →
Live Distribution Reference

Actual distribution of each index across all group × state observations for the selected year. Auto-updates when new election data is added. Use these figures to contextualise any individual value you see elsewhere in the system.

Election year
Loading distributions…