Explore California ballot measures from 1911 to the present.
Filter by region, topic, year, or status. Click any measure for details, AI-generated summaries, and related measures.
—Measures
—Pass Rate
—Avg. Win Margin
—Avg. Turnout
—Year Range
Analysis
What California ballot measures reveal
A reported-analysis view of the full CalBallot dataset: what voters approve,
where the ballot is busiest, which rules change outcomes, and where campaign
money reshapes statewide fights.
Ballot activity rises and falls with election cycles
High confidence
Annual counts, decade composition, and election-cycle patterns show when California voters faced the heaviest measure load.
Annual volume and pass rate
Local vs. statewide by decade
Election-cycle load
Method: active records by election year. Trend fits are descriptive and not coverage-adjusted.
Topics
The issue mix is broad, but unevenly classified
CEDA supplies many local records with reliable measure types but sparse topical text.
Overall topic mix
Topic share by decade
Method: consolidated display topics. Treat large Other counts as a data limitation, not a substantive topic.
Measure Types
Fiscal tools dominate the local ballot
Measure type is the cleaner lens for CEDA-heavy local records.
Largest measure types
Fiscal share over time
Method: normalized category type from source records.
Geography
The map of ballot activity is not evenly distributed
Raw counts
County totals reflect local-government density and source coverage. They are not population adjusted.
Busiest counties
Method: active local records grouped by normalized county name. Map geometry loads from the public us-atlas county topology.
Rules
Thresholds can turn majorities into losses
Simple-majority elections are not the same world as 55% or two-thirds contests.
Method: pass/fail codes and vote-threshold fields; known edge cases remain under review.
Close Calls
Many outcomes are decided near 50%
The closest races are a useful quality check and an editorial trailhead.
Method: records with valid percent yes; margin shown around 50%, not legal threshold.
Campaign Finance
Statewide propositions can become billion-dollar fights
Statewide only
Finance coverage is intentionally scoped to matched statewide propositions from CalAccess.
Method: pre-aggregated finance_statewide.db summaries by measure and stance.
How these insights were calculated
1 / 10
Filter & Explore
Filter by statewide propositions or local/county measures
🏛️Statewide(1,315)
📍Local(10,997)
Click to select one or more regions (click again to deselect)
Click to select one or more topics (click again to deselect)
Click to select one or more years (click again to deselect)
Filter by measure outcome
✓Passed(8,188)
✗Failed(4,064)
⏳Pending/Unknown(60)
Filter by measure type (GO Bond, Property Tax, Sales Tax, etc.)
Active filters
0measures found
🗳️ Upcoming 2026 Ballot Measures
Get informed about California's upcoming ballot measures before you vote.
📋 Full official details will be available as the election approaches.
🎯
Ballot Measure Trivia
Loading question...
Question 1 of 15
About This Project
CalBallot is a tool for exploring over 12,000 ballot measures
from across California, spanning local school bonds to statewide propositions.
Background
This project grew out of the UCLA Voter Guide Project,
a volunteer-driven initiative I started to research and publish ballot measure summaries for California voters
in areas with limited local news coverage. While that project focused on writing new summaries with student volunteers,
I wanted to build something that could make historical ballot measure data more accessible and explorable.
Features
Filter by region, topic, year, and outcome
AI-generated plain-language summaries
Related measures recommendations using semantic similarity
Vote results and historical trends
Data Pipeline
Building this database required substantial data engineering work:
Data collection: Aggregated records from multiple sources including the CA Secretary of State,
NCSL, ICPSR, and CEDA research databases
Deduplication: Merged overlapping records using fingerprinting algorithms to identify
the same measure across different sources
Standardization: Normalized county names, vote percentages, and date formats
across inconsistent source data
Topic classification: Used K-means clustering on sentence embeddings to automatically
categorize measures into ~20 topic clusters
AI summaries: Generated plain-language summaries using LLMs for measures
with sufficient ballot text
Similarity matching: Computed semantic embeddings (all-MiniLM-L6-v2) to find
related measures based on content similarity
Author
Built by Igor Geyn, a data scientist and researcher
based in the Bay Area. My background is in political economy and causal inference, with a PhD from UCLA.