Brief us

WP-34: Persuasion

Nineteen years of school to read the ballot

The average statewide ballot title in 2022 read at the level of a nineteenth year of schooling. Voters decide anyway, and the record of how they do it explains why the side that says no starts ahead.

By Brushworks research deskNovember 2026Working paper

Abstract. Ballot measures ask voters to make law from text few of them can read. In 2022, the average statewide ballot title scored 19 on the Flesch-Kincaid grade scale, and citizen initiatives have been approved about a third of the time over a century. We set out the record on how voters decide measures they cannot read in full, why uncertainty favors no, what the most expensive measure fight in American history shows about money and text, and how a campaign on either side should write for the voter who will decide on one sentence.

In brief

The finding
Voters cannot read most ballot measures and decide them anyway, on cues and on one sentence. When the sentence they hear is uncertain, they vote no.
For a campaign to pass a measure
Write the one sentence before the text is final, test it against a control, and say it early. Every unanswered doubt is a vote for the status quo.
For a campaign to defeat one
You do not need to win the argument. You need each voter to hold one true doubt that the other side has not answered.

A ballot measure asks a voter to write law. The text that reaches her usually reads like it. In 2022, Ballotpedia scored the titles of all 140 statewide measures with the Flesch-Kincaid formula, which estimates the years of schooling a reader needs.9 The average was 19: a graduate-school level. Titles drafted by citizens averaged 17 and those referred by legislatures averaged 20.1, 2 Voters decided all 140 anyway. This paper sets out how they do it, why the side that says no starts ahead, and what that asks of a campaign on either side.

The record

Over a century the citizen initiative has been a hard road. From 1912 through 2019, 36 percent of citizen initiatives were approved, against 57 percent of all statewide measures, including those referred by legislatures.3 Few even reach the ballot: of all initiatives filed for 2022, 3.52 percent qualified.1

The reading level of the question, by who wrote it. Flesch-Kincaid grade of 2022 statewide ballot titles; a high school diploma is grade 12.

Approval rates have risen in recent decades. Citizen-initiated measures passed 41 percent of the time in the 2000s, 49 percent in the 2010s and 57 percent from 2020 through 2024.4 In 2022, voters approved 81 percent of statutes referred by legislatures and 59 percent of statutes proposed by citizens.2

Approval of citizen-initiated measures by decade, and over the long run.
Approval by type of measure, 2022 statewide ballots.

Why uncertainty votes no

The best-established finding about direct democracy is that support for a measure tends to fall as the campaign goes on, and that voters who are uncertain about what a measure does tend to vote against it.5 A no vote keeps a world the voter knows. A yes vote buys one she has to imagine.

Voters who cannot read the text decide on cues: who is for it, who is against it, and what one sentence about it they believe. Lupia’s study of California’s 1988 insurance initiatives found that voters who knew little about the details but knew where the insurance industry stood voted much like voters who knew a great deal.6 The cue did the work of the text.

A measure is decided on the sentence a voter can hold, said by someone she believes.

What money can and cannot buy

California’s two sports-betting measures of 2022 were the most expensive ballot measure contest in American history, with about $463 million raised: $301 million in support and $162 million in opposition.7 Proposition 27, the online measure, received 17.7 percent of the vote, 1.9 million yes votes against 8.8 million no.8

Two thirds of the money, under a fifth of the vote. Share of the money raised on the two measures that supported them, and the certified yes vote on Proposition 27.

Money buys reach. It does not buy belief in a sentence the voter has reason to doubt, and in a measure campaign each side’s money also amplifies the other side’s question. The record suggests that opposition campaigns win by planting a specific, true doubt in each persuadable voter and that support campaigns win by answering it before it is asked.5

Writing for the voter who decides on one sentence

  • Decompose the measure. List every consequence the text implies, and check each one. A campaign that has not done this will learn its weaknesses from the opponent’s mail.
  • Write one sentence per consequence, per cohort. The family that worries about money hears a cost sentence. The household that cares about its town hears a local one. Every sentence must be true and checkable (WP-03).
  • Test against a control. Flight each sentence to a slice of each cohort and measure the result against a sealed holdout. Keep what moves people. Retire what backfires (WP-26).
  • Say it early. Support falls as the campaign goes on. The side that frames the measure first sets the doubt the other side has to answer.
The road to the ballot. Of every hundred initiatives filed for 2022, the number that qualified.

A century of direct democracy

The citizen initiative arrived in the United States with the Progressive movement. South Dakota adopted it in 1898 and Oregon first used it in 1902, and a wave of Western and Midwestern states followed in the next two decades. Its promise was that voters could act where legislatures would not. Its record has been mixed from the start: the decade with the highest approval rate was the first, when 70 percent of 33 citizen-initiated measures passed, and the lowest was the 1920s, when 26 percent of 277 passed.4

Approval of citizen-initiated measures by decade, the extremes and the recent rise.

Two things about the long record matter for a campaign today. The first is volume: in the decade from 2000 to 2009, 410 citizen-initiated measures reached statewide ballots, more than in any other decade.4 A voter in an initiative state may face a dozen questions on one ballot, each written in the language of a statute, which makes the cue and the single sentence more important, not less. The second is that approval has risen since: 49 percent in the 2010s and 57 percent from 2020 through 2024.4 Part of the rise reflects measures drafted to poll well before they are filed, and part reflects campaigns that have learned to answer the predictable doubt early. Neither changes the starting point. The status quo starts ahead, and the question for every campaign is who speaks to the undecided voter first.

The cues a voter uses

A voter deciding a measure she has not read relies on a small number of cues, and a campaign can know in advance which ones it will be judged on.

  • Who is for it and who is against it. The strongest cue is the identity of the sponsors and opponents the voter recognizes.6
  • What it costs her. A specific cost to her household, stated by someone she believes.
  • What it changes near her. A local consequence she can picture.
  • The title. For many voters the ballot title is the only text they read, which is why its reading level matters.
  • Uncertainty itself. A voter who cannot tell what a measure does, or hears two credible accounts, tends to vote no.5
The cues a voter uses on a measure she has not read, and which side each tends to help. A qualitative reading of the literature.

A worked example: two campaigns for one measure

The figures in this example are illustrative; the shape is the one the literature describes. A measure launches at 62 percent support. On a conventional schedule, the support campaign speaks late and the opposition plants its doubt in September. Support falls steadily and finishes at 47 percent. On a schedule built on the one-sentence method, the support campaign answers the predictable doubt in each cohort before it is raised and holds 54 percent at the close.

Support for an invented measure from launch to Election Day under two campaigns. Illustrative.

The difference between the two campaigns is not the amount of money. It is the order in which sentences reach voters, and whether each sentence was tested against a control before it was paid for.

Objections

Voters read the guide. Some do, and they decide on the guide’s summary and on the arguments printed beside it. Even for them, the cue of who signed which argument often matters more than the text.6

The side with more money wins. The most expensive measure contest on record ended with the better-funded side at 17.7 percent. Money buys reach; it does not buy belief.7, 8

No always wins. It does not. Citizen-initiated measures passed 57 percent of the time from 2020 through 2024, the highest rate in decades.4 The status quo starts ahead. It can be overtaken by a campaign that answers its doubts first.

Questions to ask before a measure is filed

  • What does the title say, and at what grade level?
  • What is the single most damaging true doubt about this measure, and who will raise it?
  • Which cohorts will hold that doubt, and what one sentence answers it for each?
  • Who are the sponsors and opponents voters will recognize, and what does each signal?
  • When will the support campaign speak first, and against what control?

Further reading

  • Bowler, S., and Donovan, T. (1998), cited above: the standard account of how voters decide ballot measures.
  • Lupia, A. (1994), cited above: how cues let voters decide like experts.
  • Gerber, E. R. (1999). The Populist Paradox: Interest Group Influence and the Promise of Direct Legislation. Princeton University Press: who wins and loses in direct democracy, and why money is not the whole story.

Implications

For a campaign to pass a measure. Start before the text is final. A sentence that cannot survive a skeptical reader should change the text, not the mail.

For a campaign to defeat one. Find the specific, true doubt each cohort will hold, and say it first. The status quo is already on your side.

For a government or an institution that has to put a question to voters. Write the title at a level the electorate can read. A measure that cannot be understood will be decided on someone else’s summary of it.

Notes and references

  1. Ballotpedia, year-end analysis, 2022, including signature qualification rates.
  2. Ballotpedia, Analysis of the 2022 statewide ballot measures: approval by type and readability of ballot titles.
  3. Ballotpedia, October 7, 2019.
  4. Ballotpedia, Initiative frequency and success throughout the decades.
  5. Bowler, S., and Donovan, T. (1998). Demanding Choices: Opinion, Voting, and Direct Democracy. University of Michigan Press.
  6. Lupia, A. (1994). Shortcuts versus encyclopedias: information and voting behavior in California insurance reform elections. American Political Science Review, 88(1), 63–76.
  7. Ballotpedia, the most expensive ballot measures in California: Propositions 26 and 27 (2022), $463,378,417 raised in total.
  8. Ballotpedia, California Proposition 27 (2022): certified result, 1,906,339 yes (17.72 percent) and 8,849,200 no.
  9. Kincaid, J. P., Fishburne, R. P., Rogers, R. L., and Chissom, B. S. (1975). Derivation of New Readability Formulas for Navy Enlisted Personnel. Naval Technical Training Command, Research Branch Report 8-75.

Basis of figures. Ballot-measure counts, approval rates, readability scores and spending from Ballotpedia’s published analyses; the Proposition 27 result is certified. The cue table is a qualitative reading and the drift example is illustrative. No Brushworks program figures appear in this paper.

Next paperThe Review

Start a conversation

Brief us.

Tell us the race, the calendar and the margin. Within a week we will tell you whether we can move it and what it would take. Candidate work and independent-expenditure work are taken by separate teams; say which you are. Governments and institutions: tell us the mandate and the jurisdiction, and we will tell you what registration it requires before we say anything else.

Send a briefbrief@brwrks.com

  • Candidate practice
  • Independent-expenditure practice
  • International practice

Brushworks Strategies Inc., a Wyoming corporation · Washington, New York, Ottawa, London, Munich, Dubai, Tokyo