Elite Analytics Dashboard

AI-powered visualizations impossible before modern data infrastructure — built for Chief Data Officers of Democratic campaigns.

Sprint 2.7 — 11 Elite Visualizations
17
Ridgeline Plot — Polling Distribution Shifts
Distribution of recent polls across 6 key Senate races
SOURCED GA-SEN: Emerson/Nexstar Feb 28–Mar 2, 2026 (n=1,000 LV, ±3.0%) · Other races: CDO backend API + Cook Political as of …
18
Beeswarm Poll Chart HONEST GAP
Individual public polls sized by sample size, colored by pollster rating
Honest gap (2026-05-17 audit): A prior version of this chart rendered 12 hardcoded polls with fabricated pollster names and dates. Those entries have been removed pending real public-poll ingestion. As public 2026 polls land (currently a handful for GA-SEN, NC-SEN, OH-SEN — see ridgeline above), they will appear here. Per M3 (honest about gaps): “we don't know” beats “made it up.”
A+ Pollster
B Pollster
C Pollster
Methodology — Beeswarm Polls SOURCED (when populated) as of …
Real-polls-only render. Source pollsters: Emerson College / Nexstar Media, Opinion Diagnostics, High Point University, Catawba / YouGov, On Message Inc. Pollster grades from 538 pollster ratings. Bubble area ∝ sample size; vertical jitter is collision-avoidance only.
GA-SEN Emerson as of … OH-SEN On Message as of … NC-SEN Opinion Diagnostics as of … NC-SEN HPU as of … NC-SEN Catawba/YouGov as of …
Sources linked per bubble on hover. Roadmap: live public-poll ingestion (DEV_PLAN Phase 8) will auto-populate this chart.
19
Sankey Voter Flow Diagram ESTIMATED
2020 → 2024 → 2026 voter alignment shifts (real endpoints, Pew-validated retention interpolations)
Methodology — Voter Flow SOURCED endpoints + ESTIMATED retention as of …
Endpoints (the corner totals): 2020 Biden 81.3M / Trump 74.2M, 2024 Harris ~75.0M / Trump ~77.3M (AP final certified). 2026 projection from RCP Senate-race composites. Did-Not-Vote baseline from U.S. Elections Project 2022 midterm. The 18 cross-tab link values are ESTIMATED — not measured retention rates. They apply the Pew 2024 validated-voter retention rule-set (~85% party loyalty, ~5% switching, ~8% dropoff) to the real endpoints. Tooltip values are read as percentage of source-period electorate, not raw vote counts — rendering a raw-count label would imply a measured cross-tab we do not have.
AP final 2024 endpoints as of … Pew 2024 validated-voter retention rules as of …
Sources: Pew 2024 validated-voter report for retention rules; AP/Dave Leip final certified results for endpoints. Roadmap: when Catalist 2024 validated cross-tabs publish, swap rule-based estimates for measured rates.
20
GA-SEN Poll Results by GOP Matchup — Ossoff Two-Party Share
Real data: Emerson/Nexstar Feb 28–Mar 2, 2026 · n=1,000 LV · MOE ±3.0% · Error bars shown
SOURCED
Converted from prior placeholder candlestick → real poll bar chart. Ossoff leads all GOP matchups within current polling. Carter is most likely nominee based on primary polling average (Collins +16.7 pts, per RCP/270toWin Mar 9 2026).
21
Narrative Contagion Network
Force-directed story spread across news sources
★ Origin
National
Critical coverage
Counter-narrative
SOURCED
Methodology: Real story: "Republicans Unveil $342M Senate Battle Plan" (NYT, Apr 6 2026 14:00 ET). 11 outlets confirmed via CDO backend /narrative/contagion RSS scraping. Gold node (★) = origin outlet. Edges animate in real chronological order.
22
Scenario Fan Chart — Ossoff GA-SEN Projection ESTIMATED
50/80/95% confidence bands to election day, anchored on a single sourced poll (center is FLAT at the anchor; only bands widen)
Methodology — Fan Chart ESTIMATED as of …
Single sourced anchor: Ossoff vs. average GOP matchup share ≈ 52.9% on Mar 2 2026 (Emerson/Nexstar; range 51.6–54.4% across three GOP candidates). Trajectory to Nov 2026: the center line is FLAT at the 52.9 anchor (a uniform-prior random walk has constant expectation — only the uncertainty bands widen with horizon). 2026-05-17 Phase 8.2.2 Pod 3 fix: prior version rendered a rising 48.5→52.1 trajectory under the same caption, which contradicts the random-walk-prior claim (a flat curve is the correct representation). Bands widen symmetrically around the anchor; the displayed range at Nov 2026 reflects uniform-prior pp/mo growth. This is a scenario, not a forecast. If no new polls land between now and Nov, the FLAT 52.9 is the prior, not the posterior. Anchor source: Emerson College / Nexstar Feb 28–Mar 2 2026. Roadmap: live anchor refresh from public-poll watcher (DEV_PLAN Phase 8).
23
Adaptive Cartogram — Senate Battleground SOURCED
Toggle metrics · Click state for county-level detail
Methodology — Cartogram Metrics SOURCED as of …
Persuadability: 60% inverse |2024 2-party margin| + 40% 2020→2024 swing magnitude. Sources: Brookings Jan 3 2025, Sabato Crystal Ball Dec 2024, AP/Dave Leip final certified 2024. Fundraising Density: FEC Q4 2025 cycle totals per state, normalized 0–100. Turnout Propensity: 2022 VEP % from U.S. Elections Project (Michael McDonald, UF) · Pew Mar 10 2023; national avg 45.1%.
Brookings 2024 recap as of … U.S. Elections Project as of … MIT Election Lab as of …
Per-state values displayed on hover. County-level click-through: 2020 actuals SOURCED from MIT Election Data & Science Lab; 2026 predictions ANALYST (linear extrapolation of 2020→2024 swing, replace with VAN/PDI forecast when ingested).
24
Horizon Chart — Multi-Race Tracker ANALYST
10 competitive races · time scrubber replays Jan–Apr 2026
Honest gap (2026-05-17 audit): The 10×12 month time-series displayed are analyst projections, not measured polls. Only 3 of 10 races (GA-SEN, NC-SEN, OH-SEN) have ≥3 public 2026 polls; the other 7 are extrapolated from Cook Political ratings + 2024 baseline + Catalyst projection model. As more public polls land, real time-series will replace the projections row-by-row.
April 2026
Methodology — Horizon Chart ANALYST + SOURCED anchors as of …
Anchor polls available (terminal-month only): GA-SEN (Emerson Mar 2 2026 D+3), NC-SEN (HPU/Catawba/Opinion Diagnostics Apr 2026 D+9), OH-SEN (Emerson Aug 18–19 2025 D−6 + On Message Inc. Mar 3–8 2026 D+2; Emerson + Cook-verified), TX-SEN (TPOR Apr 17–20 2026 D+3, single anchor; Cook still Likely R) — only the right-edge (April 2026) value of each of these four rows is anchored on a real poll; the preceding 11 monthly values per row are analyst interpolation from the Catalyst projection model. Unanchored rows (fully ANALYST): MT-SEN, MI-SEN, FL-SEN — Catalyst projection from Cook Political rating midbands + 2024 cycle baseline. Roadmap: replace each ANALYST row with measured polling as 2026 cycle polls land. Methodology corrected 2026-05-17 Phase 8.2.3 Pod 3 audit (prior version said the three anchored rows were “SOURCED” while the honest-gap banner directly above said all month time-series are analyst projections — banner-vs-card contradiction resolved by making the methodology card match the banner). Phase 8.2.4 Sweep I correction 2026-05-17: REMOVED AZ-SEN, NV-SEN, PA-SEN, WI-SEN from ANALYST roster — these states have no 2026 Senate seat in cycle, so projecting a 2026 race for them violated M1 (no fake data). Sweep J correction 2026-05-17: TX-SEN moved from ANALYST roster to anchored row — TPOR Apr 17–20 poll falsified the prior “no public H2H poll in Jan–Apr 2026 window” claim that lived on index.html. Sources linked in ridgeline methodology block above + Cook Political Senate Ratings.
25
Cross-Signal Synthesis Timeline — Ossoff GA-SEN ANALYST
Normalized signal blend (poll · news sentiment · fundraising) — social removed (no verifiable feed)
Methodology — Cross-Signal Synthesis ANALYST as of …
Each signal is normalized 0–100 and shown as a separate series. Real anchors: Poll (Emerson Mar 2 2026), Fundraising (FEC Q1 2026 normalized), News Sentiment (CDO backend /narrative/sentiment?race=GA-SEN when reachable). The interstitial month values are analyst interpolation — not measured monthly polls. Social-sentiment series previously shown has been removed (no verifiable feed). Read as a strategic-overview composite, not a forecast. Sources linked in respective methodology cards above. Roadmap: full real-signal synthesis available only for races where all three signal types have ≥3 measured monthly observations.
DATA SOURCES & REFERENCES
Polling Data
RealClearPolitics Senate 2026
Senate race poll averages across all competitive states
538 / ABC News Polls
Individual poll database with A–F pollster grades
Sabato's Crystal Ball
UVA Center for Politics election analysis and ratings
Election Results
MIT Election Data & Science Lab
Historical vote share and margin data by race and state
Cook Political Report
Official 2026 Senate race ratings and PVI scores
Campaign Finance
FEC Public Records
Federal Election Commission — Q4 2025 / Q1 2026 filings
OpenSecrets 2026 Senate
Compiled fundraising totals, cash on hand, outside spending
ProPublica Campaign Finance API
Programmatic FEC data access
Census & Demographics
Census ACS (American Community Survey)
State-level demographic, income, and housing data
PPIC Statewide Survey
California public opinion, crosstabs by race/age/region
Turnout Data
U.S. Elections Project
VAP and VEP turnout by state — 2022 midterm baseline
MIT Election Lab — Turnout
Voter turnout data across election cycles
Platform Intelligence
CDO Analytics API
Live competitive index, emotional sentiment, narrative trending
Congress.gov
Official legislative data and member voting records
Methodology
Turnout: 2022 VEP (Voting Eligible Population) % by state — U.S. Elections Project (Michael McDonald, Univ. of Florida) + Pew Research Center analysis Mar 10 2023 (pewresearch.org). National avg: 45.1%. Persuadability: SOURCED — derived 60% from |2024 final 2-party margin| (closer=higher) + 40% from 2020→2024 presidential swing magnitude (Brookings Jan 3 2025; Sabato Crystal Ball Dec 2024; AP/Dave Leip final certified results). Higher = more competitive AND more volatile electorate. Fundraising: FEC Q4 2025 cycle totals (filed Jan 31 2026) normalized 0–100 per candidate competition intensity. Ridgeline means: GA-SEN sourced from Emerson/Nexstar Feb 28–Mar 2 2026 (D+3). All other races: CDO backend API or Cook Political rating band. Where <3 polls exist, Gaussian bandwidth widened to reflect uncertainty.