Every method we have available for turning prior election data into a target vote count, the formula each uses, and where the method comes from. The final section shows exactly how the live performance numbers for HD-62, HD-63, and the Commissioner race were computed — formula, variables, and source for each.
CORE
The core identity
Every method below is a way to estimate one of the two terms in a single equation. The win number is projected turnout times the share you need.
Win number = Projected turnout × Target share
Projected turnout = Registered voters × Turnout rate
Target share = 50% + 1 (raise to 52–55% for a safety margin)
Project turnout from elections of the same type and cycle. Do not average a presidential year with an off-year; the two behave nothing alike. For a 2026 midterm, the 2022 midterm is the anchor.
Projected turnout = average turnout of the last comparable elections
× (current registration ÷ registration then)
Use 2–3 most recent like-for-like elections (midterm↔midterm, presidential↔presidential).
Scale by registration change so growth since the comparable year is captured.
Source: NGP VAN recommends the last three similar elections; the same-cycle discipline is stated in the CallHub method. callhub.io/tools/win-number-calculator
METHOD 3
Vote deficit & the three levers
The method that matters most in a Republican-leaning seat. The deficit is what you must add on top of the Democratic baseline; it closes only three ways, and the math sizes each.
Vote deficit = Win number − Base Democratic vote
Persuasion universe = Vote deficit ÷ persuasion success rate
≈ 3 × deficit (at a 1-in-3 success rate)
Mobilization — turn out drop-off Democrats. Near 1-for-1; the cheapest votes.
Registration — add eligible-unregistered voters, then turn them out.
Because a mobilized base Democrat yields ~1 vote and a persuasion contact yields ~⅓, every drop-off Democrat you turn out replaces about three persuasion targets. The lever mix matters more than the headline gap.
“Base Democratic vote” is not one race; it is the average partisan performance across several comparable contests, which cancels out individual candidate effects. Two forms: a granular index (NCEC's DPI) and a macro lean (Cook PVI).
Democratic Performance Index (DPI)
For each contested partisan race i:
Dem two-way share sᵢ = Dᵢ ÷ (Rᵢ + Dᵢ)
DPI = mean(sᵢ) across the basket of comparable races
Use lower-profile, straight-partisan races; exclude top-of-ticket contests with strong candidate effects.
Build a separate index per turnout context (midterm vs. presidential).
Cook Partisan Voting Index (PVI) — macro lean
PVI = district presidential two-party share − national two-party share
(last two presidential elections, weighted 75% recent / 25% prior)
Tells you where the votes are cheapest. The win number says how many; this sorts precincts by how to get them.
Score each voter/precinct on two axes:
likelihood of support × likelihood to turn out
→ four quadrants: GOTV base · persuade · register/activate · skip
Source: Hal Malchow, The New Political Targeting (2008) — the canonical support×turnout quadrant framework, documented in the academic targeting literature. arxiv.org/pdf/1311.7326
ADJUSTMENT
Roll-off — the effective electorate is smaller
Not everyone who shows up finishes the ballot. Down-ballot races (state house, commissioner) lose votes to roll-off, so the real electorate for your race is smaller than total turnout.
Roll-off rate = 1 − (votes in your race ÷ total ballots cast)
Effective electorate = Total ballots × (1 − roll-off rate)
Sources: Clark & Peterson, Too Far to the Bottom? — predicts state-legislative lower-chamber races suffer the most roll-off, rising when a race is uncontested or lopsided. Voter-Rolloff.pdf. MIT Election Lab on ballot-level undervotes. electionlab.mit.edu
ADJUSTMENT
Candidate overperformance — the persuadability signal
Compare a past Democrat's actual result to the baseline. Precincts where a Democrat beat the index show the ceiling isn't fixed — they point to where persuasion has already worked.
Overperformance = candidate's actual share − DPI baseline (per precinct)
The baselines in the calculator are not estimates. Here is each one's formula, the exact races used (the variables), and the source.
House seats — single contested race
HD-62 and HD-63 were contested D-vs-R every cycle, so the baseline is that race's own Democratic two-way share — no index needed.
Base share = Dem votes ÷ (Rep votes + Dem votes)
Source: Clermont County BOE official canvass (HD-62, entirely in Clermont); state-certified full-district total (HD-63, Clermont + Brown).
Seat / cycle
Rep
Dem
Base share
HD-62 · 2022 (midterm)
31,810
17,755
35.8%
HD-62 · 2024 (pres.)
42,134
24,251
36.5%
HD-63 · 2022 (midterm)
29,195
9,100
23.8%
HD-63 · 2024 (pres.)
42,034
13,682
24.6%
Commissioner — performance index (the 2022 seat ran unopposed)
No Democrat ran for the seat in 2022, so there is no same-office midterm baseline. The baseline is a DPI of contested county-wide races. President, U.S. Senate, and Governor are excluded — strong candidate effects (Brown, Vance, DeWine) distort the partisan read.
sᵢ = Dᵢ ÷ (Rᵢ + Dᵢ) DPI = mean(sᵢ)
Midterm-turnout basket — Source: Clermont County BOE, 2022 General Election Official Results (Summary Group Detail, Nov 8 2022).
Contested race (2022)
Rep
Dem
Share
Attorney General
57,815
22,430
27.9%
Auditor
56,788
22,778
28.6%
Secretary of State
56,963
22,229
28.1%
Treasurer
56,897
22,627
28.5%
Chief Justice
55,485
24,465
30.6%
Supreme Court (Fischer)
55,690
24,106
30.2%
Supreme Court (DeWine)
55,130
24,541
30.8%
U.S. Congress (2nd)
57,353
22,842
28.5%
Midterm DPI (mean of 8)
29.1%
Presidential-turnout basket — Source: Clermont County BOE, 2024 General Election Official Results (Summary Group Detail, Nov 5 2024).
Contested race (2024)
Rep
Dem
Share
Supreme Court (Shanahan)
72,728
33,399
31.5%
Supreme Court (Deters)
77,231
33,373
30.2%
Supreme Court (Hawkins)
73,873
34,899
32.1%
U.S. Congress (2nd)
76,326
34,149
30.9%
State Senate (14th)
75,761
33,321
30.5%
Commissioner (Batchler/Combs)
74,062
34,611
31.8%
Commissioner (Painter/Mazzuckelli)
74,120
34,064
31.5%
Clerk of Courts
74,016
34,146
31.6%
County Recorder
74,600
33,777
31.2%
Presidential DPI (mean of 9)
31.3%
The two 2024 commissioner races alone average 31.7% — the closest same-office anchor. So the Commissioner baseline band is 29.1% (midterm DPI) → 31.7% (same-office), central ~30%.