When the Body That Writes the Question Also Writes the Booklet: A Comparative Analysis of Pre-Referendum Information Mandates

Every referendum commission hits a structural fork in the road that few ever debate explicitly. Should the body drafting the ballot question also run the official information campaign explaining it? It sounds administrative. The consequences are constitutional. When one institution writes the question, designs the public information booklet, and regulates campaign conduct, it concentrates interpretive authority in a way no single legislative body would be permitted to hold. Split that authority—let one body draft the question and another produce voter materials—and you trade that concentration for a different risk set: inconsistency between ballot language and explanatory text, conflicting timelines, and the real possibility that the information body will quietly undermine or reinterpret the question the drafting body worked to keep neutral.

Three jurisdictions made three different choices. Ireland’s Referendum Commission historically had no control over question wording but produced the official information booklet. New Zealand’s Electoral Commission both drafted the question and authored voter information materials for the 2011 mixed-member proportional (MMP) referendum. The United Kingdom saw the Electoral Commission test and publish question recommendations while the government retained final wording control and separately commissioned public information. Each model carries its own failure mode. Together they generate a decision tree for institutional designers who need to locate information authority before a referendum is called—not after the first comprehension crisis.

Ireland: Separated Drafting and Information, With Alignment Costs

Ireland’s referendum architecture stands apart in comparative terms. The ballot question is a constitutional amendment bill—drafted by the government, approved by parliament, and formulated by the Attorney General’s office into the specific wording voters see. The Referendum Commission, a temporary body reconstituted for each vote, had no authority over that wording. Its mandate came from the Referendum Act 1998, later narrowed by the Referendum Act 2001 from advocacy to neutral information. Its job: explain the subject matter to the electorate. In the 2015 marriage equality referendum, the Commission produced a booklet on the proposed Thirty-fourth Amendment of the Constitution (Marriage Equality) Bill 2015 without having shaped the question: “Marriage may be contracted in accordance with law by two persons without distinction as to their sex.”

The structural separation had obvious advantages. Nobody could accuse the Commission of writing a question to favor a particular information frame. It treated the ballot text as fixed and built explanatory materials around it. The 2015 booklet ran 16 pages, went to every household, and included a plain-language summary, the exact constitutional text to be inserted, and a neutral statement of what Yes and No meant. No arguments for either side.

But separation produced an alignment problem the Commission could not solve. The ballot question was a constitutional insertion clause, not a policy question. Voters who read the booklet understood the policy context. Voters who only saw the ballot paper saw a sentence about contracting marriage “in accordance by law” without the surrounding explanation. The Commission could not rewrite the question to make it self-explanatory—that was not its role. It could only build an information bridge between technical constitutional language and the voter’s actual decision task.

The 2018 referendum on the Thirty-sixth Amendment—repealing the prohibition on abortion—sharpened the problem. The ballot asked voters to approve the provision “Provision may be made by law for the regulation of termination of pregnancy.” A meta-provision: it does not legislate abortion access but authorizes the Oireachtas to do so. The Commission’s booklet had to explain both what the amendment said and what it did not say—that the referendum was not a vote on specific abortion legislation but on whether parliament would be permitted to legislate. Post-referendum comprehension surveys showed a significant minority of voters believed they were voting on specific policy rather than a constitutional authorization. The booklet mitigated the confusion. It could not eliminate it, because the confusion originated in the question design—outside the Commission’s control.

New Zealand: Integrated Authority, With Interpretive Monopoly Risks

New Zealand’s 2011 electoral system referendum took the opposite approach. The Electoral Commission handled both question wording and the official information campaign. The referendum asked: “Should New Zealand keep the Mixed Member Proportional (MMP) voting system as used now for the House of Representatives?” If a majority voted No, a second question followed: “If New Zealand were to change to another voting system, which voting system would you choose?” The Commission drafted both questions, designed the voter information booklet, and ran the public awareness campaign.

Integration produced tight alignment between ballot language and explanatory materials. The booklet could reference the exact question structure and explain the two-stage logic without the awkward gap Ireland’s Commission faced. The 2011 booklet ran 12 pages, distributed via newspapers and online, with a side-by-side comparison of MMP against four alternatives: First Past the Post, Preferential Voting, Single Transferable Vote, and Supplementary Member. The explanatory text matched the ballot language because the same institution produced both.

But integration also concentrated interpretive authority in ways that carried their own risks. The Commission decided not only what the question meant but how each alternative system would be characterized in official materials. Those descriptions of four alternative systems were necessarily simplifications—each reduced to a paragraph of plain-language explanation that inevitably emphasized some features and omitted others. Critics noted that the Commission’s description of Supplementary Member framed it as a “hybrid” system without explaining that the term was contested among electoral scholars. The Commission’s framing became the authoritative public interpretation. No other body had been tasked with producing a competing or supplementary account.

This is the core risk of integrated authority: the information body’s characterization of the options becomes the de facto official meaning of the referendum, even when the question is ambiguous or when reasonable people disagree about how to describe the alternatives. In Ireland’s model, the information body’s interpretation is constrained by the question it receives. In New Zealand’s model, the information body’s interpretation is constrained only by its own internal editorial standards.

The United Kingdom: A Hybrid That Created Conflicting Accountability

The United Kingdom’s approach to referendum information has evolved through ad hoc decisions rather than coherent institutional design. The Electoral Commission, established under the Political Parties, Elections and Referendums Act 2000 (PPERA), has statutory authority to assess the “intelligibility” of referendum questions under Section 104. For the 2011 Alternative Vote referendum, the Commission tested proposed question wording through cognitive interviews and recommended changes the government accepted. But the Commission did not produce the official voter information leaflet. That responsibility fell to the government, which produced a separate booklet mailed to households.

This created a three-way split: the government drafted the initial question, the Commission tested and recommended changes, and the government produced the information booklet. The result was a booklet broadly consistent with the Commission’s recommended wording but not formally controlled by the Commission. The Commission published its own guidance online, but the mailed booklet—produced by the Cabinet Office—was the document most voters actually received.

The accountability problem was subtle but real. If voters found the booklet confusing, they could not easily identify which institution was responsible. The Commission had tested the question. The government had written the booklet. The government had accepted the Commission’s question recommendations, but the Commission had no authority over the booklet’s content. No single body was accountable for the end-to-end voter experience.

The 2016 EU referendum preserved the same structural split. The Electoral Commission assessed question intelligibility and recommended revising the initial wording from “Should the United Kingdom remain a member of the European Union?” to “Should the United Kingdom remain a member of the European Union or leave the European Union?” with Yes/No options replaced by “Remain a member of the European Union” and “Leave the European Union.” The government accepted the recommendation. But the official information leaflet was again produced by the government, not the Commission—and the government’s decision to produce a single leaflet rather than commissioning balanced materials from an independent body drew post-referendum criticism.

The Drafting Pipeline: Why Tooling Matters for Multi-Author Booklet Production

Regardless of where institutional authority sits, producing a referendum information booklet involves a multi-author drafting pipeline structurally similar to other high-stakes institutional writing. A typical booklet passes through at least four stages: initial plain-language summary by a subject-matter expert, readability review by a communications team, legal accuracy check by counsel, and translation or adaptation for additional language editions. Each stage can introduce inconsistencies—new vocabulary, shifted emphasis, unintended omissions—that compound across versions.

Ireland’s Referendum Commission managed this pipeline manually for each referendum. Version control meant tracked changes in word-processing documents. Review cycles ran through email. The Commission’s 2015 marriage equality booklet went through at least seven full drafts, according to its post-referendum report. The most significant revisions came at the legal accuracy stage, when counsel flagged that the plain-language summary had inadvertently described the amendment as “legalizing same-sex marriage” rather than “providing that marriage may be contracted by two persons without distinction as to their sex.” The correction was substantive. The original framing implied a policy outcome; the corrected framing tracked the constitutional text. A less disciplined process might have missed that distinction.

The lesson extends beyond referendum commissions. Any institution producing high-stakes written materials at scale—voter information, legal guidance, policy documentation—faces the same pipeline challenge. Referendum commissions have increasingly turned to collaborative drafting tools and structured editorial pipelines to manage multi-author booklet production, maintain version control across language editions, and enforce plain-language review cycles. Some have adopted platforms originally built for long-form writing and editorial planning; just as a novelist might use AI novel writing software that supports structured drafting and revision workflows, a referendum commission’s editorial team benefits from tooling that enforces review stages, tracks changes across language editions, and flags inconsistencies between the ballot question text and the explanatory booklet. The Authors Guild has documented how professional writing organizations are establishing best practices for maintaining editorial standards and human oversight when using collaborative or AI-assisted drafting tools, emphasizing that the quality of written outputs depends on preserving human editorial judgment throughout the drafting and revision pipeline—a principle that applies with particular force when the output is a voter information document with legal consequences.

Comprehension Testing: What Each Model Did and Did Not Do

None of the three jurisdictions conducted systematic post-referendum comprehension surveys rigorous enough to measure whether voters actually understood the question as the drafting body intended. Ireland’s Referendum Commission commissioned some post-referendum research, but it was not designed to isolate the booklet’s effect on comprehension. New Zealand’s Electoral Commission ran post-referendum evaluation focused on process satisfaction, not comprehension accuracy. The UK Electoral Commission’s question intelligibility testing was pre-referendum—conducted on draft wording, not on final information materials in combination with the question.

This is a gap comparative research could close. Work from the Pew Research Center has shown that public confidence in the fairness and accuracy of elections is a measurable variable that correlates with institutional design choices. The same methodological framework could apply to referendum comprehension: post-vote surveys testing whether voters can accurately describe what they voted on, cross-referenced with whether they received and read the official information booklet. Without that data, the debate over whether to integrate or separate information authority stays speculative.

A Decision Tree for Institutional Designers

The comparative analysis yields a decision tree built around two variables: whether the jurisdiction has an independent electoral management body with a track record of public trust, and whether the referendum question involves a single binary choice or a multi-option or sequential choice.

Step 1: Does the jurisdiction have an independent electoral management body (EMB) with established public trust?

If yes, proceed to Step 2. If no, the question of where to locate information authority is secondary to the more urgent problem of establishing institutional credibility. In jurisdictions without an independent EMB, information authority should be vested in an ad hoc commission that includes representatives from civil society organizations, legal professionals, and academic institutions—rather than in a government department. The risk of government-authored information being perceived as partisan is too high to justify direct governmental control, even if the government is technically neutral on the referendum question.

Step 2: Is the referendum question a single binary choice (Yes/No on a clearly defined proposition)?

If yes, integrated authority—same body drafts question and information—is acceptable, provided the EMB conducts pre-referendum cognitive testing of both the question and the draft information materials with a representative voter sample. The New Zealand model works for binary choices because the interpretive scope is narrower. The question asks whether to keep or change a system. The information booklet explains the current system and the proposed change. The risk of interpretive monopoly is lower because the choice structure itself constrains the range of reasonable descriptions.

If no—the question is multi-option, sequential, or involves a meta-provision as in Ireland’s abortion referendum—proceed to Step 3.

Step 3: For multi-option, sequential, or meta-provision questions, should information authority be separated from question-drafting authority?

Separation is recommended, but with a mandatory alignment protocol. The information body should receive the final question text no later than 12 weeks before the referendum date and should have statutory authority to publish a booklet explaining the question, including any reasonable interpretation of what each option means. The question-drafting body should not have veto authority over the information body’s explanatory text. But the information body should be required to submit its draft booklet to the question-drafting body for factual accuracy review—a review limited to verifying that the booklet does not misstate the legal effect of the question, not to approving the booklet’s framing or emphasis.

This protocol addresses the core weakness of Ireland’s model—the information body could not shape the question to be more comprehensible—while avoiding the core weakness of New Zealand’s model—the integrated body’s interpretive authority is unconstrained. The information body cannot rewrite the question. But it can build the bridge between question and voter understanding without seeking approval from the body that wrote the question.

Step 4: Does the referendum involve more than one question on the same ballot?

If yes, a single information body should produce a unified booklet covering all questions, regardless of which bodies drafted the individual questions. This prevents voters from receiving multiple booklets from different bodies with inconsistent formatting, vocabulary, or explanatory depth. New Zealand’s 2011 referendum, with its two linked questions, demonstrated the value of unified information production: the Commission’s booklet explained both questions in a single document, helping voters understand the sequential logic. If questions are drafted by different bodies—as can happen when multiple citizen initiatives appear on the same ballot—the information body should have statutory authority to produce the official booklet for all questions, with factual accuracy review by each drafting body.

The Accountability Principle

Underlying this decision tree is a principle all three jurisdictions struggled with: a single identifiable body should be accountable for the end-to-end voter information experience. Ireland achieved separation but left voters uncertain about who was responsible for the gap between question and explanation. New Zealand achieved integration but concentrated interpretive authority without an external check. The United Kingdom split authority in a way that made accountability diffused and post-referendum criticism difficult to direct.

The principle does not require integrated authority. It requires that wherever authority is divided, the division is governed by a protocol that makes the handoffs visible. The question-drafting body publishes its final text. The information body publishes its booklet with a documented factual accuracy review. Voters know who wrote the question and who wrote the explanation. If the explanation is inadequate, the accountability is traceable. If the question is poorly worded, the information body can say so in its booklet—a power Ireland’s Commission did not formally have but should.

Design Takeaway

For referendum process designers, the core lesson is this: the decision to integrate or separate question-drafting and information authority should be a function of question complexity, not a default institutional habit. Binary questions can tolerate integrated authority with cognitive testing. Multi-option, sequential, or meta-provision questions require separated authority with a mandatory alignment protocol. In all cases, a single body should be accountable for the end-to-end voter information experience, and that accountability should be visible to voters through documented publication of both the question-drafting process and the information review process. The decision tree above provides a starting framework. It should be adapted to the specific institutional context of each jurisdiction—including the maturity of its EMB, the complexity of its referendum questions, and its legal tradition regarding judicial review of electoral materials.