Week 9: Direct democracy: initiatives, referendums, and minority rights

In plain words

Week 9 of the My Vote Forward seminar studies direct democracy: citizens voting on laws themselves through initiatives and referendums. Students read four studies: one on how voters handled complicated insurance initiatives in California, one finding that measures restricting civil rights usually passed in the United States, a reply finding the opposite in Switzerland, and a review arguing that direct democracy often makes government work better.

Direct democracy raises two worries. Can ordinary voters understand complicated laws? And will majorities vote away minorities’ rights, as Madison feared in week 1? The research gives different answers in different places. This week asks why.

Defined terms

• Initiative: a process in which citizens propose a law or constitutional amendment by petition, and voters decide it. See initiative.
• Referendum: a vote by the electorate on a law or question, which may be referred by a legislature or triggered by petition. See referendum.
• Ballot measure: any proposition on a ballot for direct decision. See ballot measure.
• Recall: a vote on whether to remove an official before the end of a term. See recall.
• Information shortcut (heuristic): a cue, such as an endorsement, that lets a voter infer a position without studying the details.
• Majority tyranny: a majority using its power to deprive a minority of rights. See tyranny of the majority.

Learning objectives

By the end of week 9, a participant should be able to:

1. Distinguish initiatives, referendums, ballot measures, and recalls.
2. Explain Lupia’s finding about information shortcuts and its conditions.
3. State Gamble’s finding on civil rights measures and her methods.
4. Explain Frey and Goette’s reply and the conditions they identify for minority success.
5. Summarize Matsusaka’s review and the strongest objections to it.
6. Explain why a priorities channel is not direct democracy, and what risks it shares with it.

Core readings

1. Arthur Lupia, “Shortcuts Versus Encyclopedias: Information and Voting Behavior in California Insurance Reform Elections,” American Political Science Review 88(1), 1994, pp. 63 to 76.⁠¹
2. Barbara S. Gamble, “Putting Civil Rights to a Popular Vote,” American Journal of Political Science 41(1), 1997, pp. 245 to 269.⁠²
3. Bruno S. Frey and Lorenz Goette, “Does the Popular Vote Destroy Civil Rights?” American Journal of Political Science 42(4), 1998.⁠³
4. John G. Matsusaka, “Direct Democracy Works,” Journal of Economic Perspectives 19(2), 2005, pp. 185 to 206.⁠⁴
5. James Madison, The Federalist No. 10 (1787), reread for its argument about majority factions.⁠⁵

Lecture notes

Why direct democracy belongs in this course

Direct democracy is the subject of week 9 because it is the most established way that citizens act on public decisions between general elections. Switzerland, many American states, and other jurisdictions let voters decide laws directly. The site’s cases on Swiss direct democracy, California’s initiatives, the Brexit referendum, the Scottish independence referendum, and New Zealand’s 2020 referendums document how differently it works across places. The site’s chapter on referendum failures collects the failures. This lecture concentrates on two questions the research has tested: whether voters can decide competently, and whether minorities are safe.

The week also sharpens a distinction the course depends on. A priorities channel, such as the one My Vote Forward describes, is not direct democracy: it does not enact anything. But it shares some of direct democracy’s risks, especially the risk that a majority’s stated preferences are presented as a mandate on questions that affect minorities’ rights. Knowing the evidence on direct democracy helps judge how serious those shared risks are.

Lupia: shortcuts versus encyclopedias

The competence worry is old. Arthur Lupia’s 1994 study tested it on unusually difficult ballot questions. He notes that “voters in mass elections are notorious for their apparent lack of information about relevant political matters,” and that scholars divide between those who think “an electorate of well-informed voters is necessary for the production of responsive electoral outcomes” and those who think “apparently ignorant voters will suffice because they can adapt their behavior to the complexity of electoral choice.” Lupia surveyed California voters “who faced five complicated insurance reform ballot initiatives,” and found “that access to a particular class of widely available information shortcuts allowed badly informed voters to emulate the behavior of relatively well informed voters.” He concludes that the finding suggests “the conditions under which voters who lack encyclopedic information about the content of electoral debates can nevertheless use information shortcuts to vote as though they were well informed”.⁠¹

The finding is often cited as a general defense of direct democracy. Its scope is narrower. It shows that shortcuts worked in this case for voters who had access to them. Shortcuts depend on trustworthy cues, such as knowing which side an industry or a trusted group supports. Where cues are absent, misleading, or manipulated, shortcuts can mislead. The site’s chapters on misinformation and money in politics consider how cues can be distorted, and week 7’s discussion of agenda and framing applies here.

Gamble: civil rights at the ballot box

Barbara Gamble’s 1997 study tested Madison’s fear directly. Her hypothesis: “without the filtering mechanisms of the representative system, direct democracy promotes majority tyranny as the scope of civil rights conflicts expands and citizens vote on civil rights laws.” She analyzed “over three decades of initiatives and popular referenda from five major civil rights areas: housing and public accommodations for racial minorities, school desegregation, gay rights, English language laws, and AIDS policies.” Her result: “Citizen initiatives that restrict civil rights experience extraordinary electoral success: voters have approved over three-quarters of these, while endorsing only a third of all initiatives and popular referenda”.⁠²

Gamble’s study is the strongest empirical support for the Madisonian worry in the American context. It deserves to be stated without softening: in the measures and period she studied, proposals that restricted civil rights passed far more often than ballot measures generally.

Frey and Goette: the Swiss contrast

Bruno Frey and Lorenz Goette replied the following year. They describe Gamble’s data as covering seventy-four civil rights initiatives in state and local ballots between 1959 and 1993, and they do not question her methods. Instead, they examine Switzerland, “where three-quarters of the referenda held worldwide were undertaken,” in their words, considering the same type of civil rights issues at the Swiss federal, cantonal, and city level (for Zurich). Their conclusion is the opposite of Gamble’s: in their data, “direct democracy protects civil rights.” They also “identify conditions under which minority interests have a particularly high probability of success,” which they suggest may partly explain the difference between the two results.⁠³

The exchange is a model of how to read contested evidence. Neither study refutes the other. Each describes a different system over different periods. The lesson is that the effect of direct democracy on minorities depends on institutional context: what can be put to a vote, what thresholds apply, how courts review results, and how political culture treats minorities. That conclusion is less satisfying than a universal verdict, and more accurate.

Matsusaka: the performance case

John Matsusaka’s 2005 review takes a broader view. He describes the purpose of his essay as describing “the practice and theory of the increasingly important political phenomenon of direct democracy and the main lessons from the scholarly literature,” and reports that “many questions remain to be answered, but the emerging view is that direct democracy works, allowing the general public to participate in lawmaking often seems to improve the performance of government”.⁠⁴

The strongest version of Matsusaka’s argument holds that the initiative gives voters a check on legislators who may be captured by organized interests, and that even the threat of an initiative can move legislatures toward majority preferences. The strongest objections are that initiative campaigns can be dominated by well-funded interests, that complex measures can produce unintended consequences, and that majority preferences on some questions, as Gamble shows, include restricting others’ rights. Students should read the full review to see which outcomes it measures and how.

Madison’s worry, revisited

Federalist No. 10 argued that a large republic makes it “less probable that a majority of the whole will have a common motive to invade the rights of other citizens”.⁠⁵ Direct democracy removes part of the representative filter Madison relied on, though not the size of the electorate. Gamble’s findings suggest Madison’s concern was well founded in some settings. Frey and Goette’s findings suggest other safeguards, such as institutional design and political culture, can protect minorities even without the filter. The week’s practical lesson is that any channel connecting majority preference to public decision needs explicit protection for questions of rights. Week 5’s reading of Sen supports the same conclusion from the theory side, and the site’s page does a majority decide rights? applies it to My Vote Forward.

Priorities are not ballots, and why the difference matters

A priorities channel ranks subjects; it does not decide laws. That difference reduces some risks. No one’s rights change because “housing” ranks above “transport.” But two risks carry over. First, if a ranking is treated as a mandate, it can be used to justify decisions that its participants never voted on, a pattern discussed in can an institution ignore a result? from the opposite direction. Second, a channel that asked about specific measures affecting particular groups would begin to function like a ballot without a ballot’s legal protections. The framework’s chapter on proposals states how it treats official ballot items, and students should test it against this week’s evidence.

A worked example: shortcut or trap?

Hypothetical example: a statewide initiative would change how electricity rates are set. Most voters know little about rate regulation. Supporters include a consumer group; opponents include the state’s largest utility. Following Lupia, many voters could infer their interest from the endorsements: if they trust the consumer group and distrust the utility, they vote yes. Now change one fact: the consumer group is newly formed and funded, without disclosure, by a competing energy company. The same shortcut now misleads. Whether shortcuts help depends on whether cues are what they appear to be, which is why campaign finance disclosure and independent reporting matter to direct democracy’s competence. The site’s chapters on money in politics and local news collapse bear on both conditions.

Common misreadings this week guards against

Three errors recur. One treats Lupia’s study as proof that voters are always competent on ballot measures, when it shows conditions under which shortcuts work. A second treats Gamble’s findings as a universal law of direct democracy, when Frey and Goette’s Swiss evidence shows the outcome varies with context. A third treats Matsusaka’s “direct democracy works” as a claim about every measure, when it is a summary of an emerging scholarly view about average effects on government performance. Each reading is valuable within its scope and misleading outside it.

Design choices that shape referendum outcomes

Beyond competence and rights, the readings point to design choices that every referendum system makes, often without public debate. Who may place a question on the ballot: the legislature, the executive, or citizens by petition? What threshold applies: a simple majority, a supermajority, or a turnout minimum? Who writes the question wording, and who reviews it for neutrality, a problem week 6 showed can change answers? Is the result binding or advisory, and what happens if the government that called the vote disagrees with it? The site’s cases on the Brexit referendum, the Scottish independence referendum, and New Zealand’s 2020 referendums show different answers to these questions and their consequences. Students should notice that each choice is itself a form of agenda power, the subject of week 7.

Discussion questions

1. Lupia’s voters used shortcuts successfully. What shortcuts would a voter use on a measure about a subject they know nothing about? When would those shortcuts mislead?
2. Gamble and Frey and Goette reach opposite conclusions. List three institutional differences between American states and Switzerland that might explain the difference.
3. Should some subjects be off-limits for direct democracy? Who should decide which?
4. Matsusaka argues that direct democracy often improves government performance. What measure of “performance” would you use, and who might disagree?
5. Is a priority ranking safer than a ballot measure because it enacts nothing, or riskier because it carries no legal safeguards?
6. Madison relied on representation to filter majorities. Does direct democracy need a substitute filter? What could it be?

Assignment

Ballot measure case analysis (1,500 words). Select one ballot measure or referendum from the site’s cases or from an official election results page in a jurisdiction of your choice. Using official sources for the measure’s text, campaign finance disclosures where available, and certified results, analyze (a) what information shortcuts voters likely used, (b) whether the measure affected the rights of an identifiable minority, (c) which institutional safeguards applied, and (d) whether Lupia, Gamble, Frey and Goette, or Matsusaka best explains the outcome. Present the strongest argument of the measure’s supporters and opponents in terms each would accept.

Evidence

All descriptions of the four studies are quoted from their published abstracts, opened on the review date. Frey and Goette’s description of Gamble’s data (seventy-four initiatives, 1959 to 1993) and their statement about Switzerland’s share of referenda worldwide are their own claims, attributed to them. The Federalist quotation is from the Library of Congress text. Discussion of the strongest objections to Matsusaka is the seminar’s summary of common arguments, not a finding.

Competing interpretations

On competence, one camp argues that information shortcuts make direct democracy workable, while another argues that shortcuts are easily manipulated by funded campaigns. On minority rights, Gamble’s evidence supports the view that direct democracy endangers minorities, while Frey and Goette’s evidence supports the view that institutional context determines the outcome. On performance, Matsusaka’s review reports a favorable emerging view, which critics challenge on grounds of special-interest influence and unintended consequences. Each view has evidence; none has a universal verdict.

Application to this framework

My Vote Forward is not a direct democracy system: it collects priorities and optional opinions, and it does not enact laws. This week’s readings show that it should nonetheless avoid presenting majority priorities as mandates on questions of rights, should disclose who participated so that shortcuts and mobilization can be judged, and should keep clear the line between education and official ballots described in proposals.

Unresolved questions

• Which institutional features best explain the difference between Gamble’s American findings and Frey and Goette’s Swiss findings?
• Do information shortcuts work as well in today’s media environment as in the 1990s California elections Lupia studied?
• If a priority ranking were widely cited as a mandate, would it acquire the risks of direct democracy without its safeguards?