Contribution to the handbook
Priority · Assessed 2026-10-28
The question and evidence provide a concrete testing method: compare proposal win rates in a sample with those in the full study data using R². A consequential limit is that matching the study sample does not establish representation of people outside it or reproduction of individual preferences. This is material guidance for evaluating preference methods, but it is one study’s example, not a validated governance test.
Does this challenge the goal?
In-scope gap
The question concerns how to evaluate preference methods, which fits the handbook’s existing testing and citizen-service goals; it does not challenge those goals.
This inquiry identifies a relevant evaluation detail, not proof that the handbook lacks a test or that the study’s measure is suitable for governance.
Why this category?
- Concrete evaluation method
- Consequential limits on interpreting a sample-match measure
The evidence concerns one study’s participant data and does not establish that this measure is sufficient for broader public representation or governance decisions.
This is an AI assessment of contribution to the goal, separate from whether the question has an answer.
Tests, stop conditions and unresolved work →
Authority, accountability and resilience →
Read the exploration goalSpecify what the authors compared between an augmented sample and the full participant data and how they quantified the match.
What we can say so far
The authors compared each policy proposal’s win rate in an augmented sample with that proposal’s win rate in the full participant sample. A win rate was the share of recorded pairwise choices in which the proposal was selected. They quantified the similarity of the two sets of proposal win rates using R²—the squared Pearson correlation—with 100% indicating identical win-rate patterns. This measures aggregate pattern similarity, not whether individual choices were reproduced or whether the participants represent people outside the study.
What remains open
No material gap is recorded within this part’s current scope.
Conditions and limits
- This is a report of the authors’ measure, based on the supplied interpretation of the preprint; it is not independent verification.
- The measure concerns aggregate proposal win-rate patterns in the study data; a high score does not establish individual-level matching or representativeness beyond the study.
Evidence and related ideas
Supporting evidence, qualifications and context play different roles in this answer.
context
The volunteer policy data The study uses volunteer choices between randomly paired policies from the two leading candidates’ 2022 Brazilian presidential programs. The choices came from an experiment involving collaborative construction of a government program.
Describes the volunteer policy-choice data but not the sample-to-full-data comparison measure.
Large Language Models (LLMs) as Agents for Augmented Democracy
Read full source: Large Language Models (LLMs) as Agents for Augmented Democracy (opens in a new tab) ↗Read source discussion: Large Language Models (LLMs) as Agents for Augmented Democracy →
context
Individual choice predictions The authors report that the models predicted individual policy choices more accurately than a rule assigning people the proposals of the candidate matching their self-reported political orientation.
Concerns individual choice prediction, rather than the aggregate sample-match measure.