The Constituent Memo: Teaching Labs to Brief Local Decision-Makers Without Losing the Error Bars
Why a Policy Brief Belongs in Lab Training
Most researchers learn to present evidence to other researchers. Far fewer receive structured practice in presenting the same evidence to someone who controls a budget line, a permit, or a public comment period. That gap is not merely a writing problem. It is a translation problem with epistemic stakes: a brief that overstates a figure can travel farther and faster than the correction that follows it.
At NI Lab, we treat science communication as part of lab practice rather than as a separate publicity function. The workshop described here asks participants to take a single peer-reviewed finding — often one panel from one figure — and turn it into a one-page constituent memo for a local decision-maker. The goal is not to win an argument. The goal is to produce an accurate, useful, and candid account of what the evidence does and does not support.
The exercise works for undergraduate research groups, graduate seminars, and mixed lab teams. It takes roughly ninety minutes, plus a short peer-review block. It requires no advocacy training, only a published paper, a local policy context, and a willingness to revise.
Setting Up the Workshop
Participants work in teams of three or four. Each team selects one published figure with a clear, bounded claim. The figure can come from the team's own accepted paper, a lab-adjacent publication, or an open-access article assigned in advance. The only rule is that the team must be able to state the finding in one sentence that names the population, the comparison, the outcome, and the time frame.
A useful opening constraint: choose a finding that could plausibly matter to a county health department, a city council committee, a school board, or a state agency. The policy relevance should be modest and specific. A figure about neighborhood heat exposure may inform a tree-canopy budget discussion. A figure about commute times may inform a transit planning memo. The workshop is not designed to produce a lobbying document; it is designed to produce a disciplined briefing.
Each team receives a one-page worksheet with four sections: the claim, the stakeholder, the limits, and the plain-language draft. The facilitator introduces a simple distinction that governs the rest of the session: the figure shows a result, the analysis supports an inference, and the brief may recommend a consideration. Those are three different statements, and the workshop treats conflating them as the central error to avoid.
Mapping the Stakeholder Before Writing a Word
Teams begin with a stakeholder map rather than a draft. The map identifies the office or agency, the staff role most likely to read the brief, the decision lever that office actually controls, and the timing of the relevant public process. A city council member may vote on a budget; a planning department staffer may draft a memo; a county health officer may convene a working group. These are different audiences with different information needs.
The worksheet asks three questions. First, what decision or discussion could this evidence inform? Second, what would the staffer need to know to judge the claim? Third, what would be misleading to leave out? The third question is the most important. Teams often discover that the most policy-relevant fact is not the headline result but the boundary condition: the study's setting, sample, duration, or measured outcome.
This stage also establishes a neutrality norm. The brief should not tell an official what to believe about a contested value question. It should clarify what the evidence suggests, where it is silent, and what additional information would reduce uncertainty. That distinction keeps the exercise within the lab's competence and gives the writing a defensible purpose.
The Claim-Limit Check
Before drafting, each team completes a three-column claim-limit check. The first column records what the figure visibly shows. The second records what the study design and analysis can support. The third records what the brief may responsibly say. The columns rarely match exactly, and that is the point.
For example, an observational study may show that neighborhoods with more tree cover have lower summer surface temperatures. The analysis may support an association after adjustment for several covariates. The brief may say that the study found an association in a specific city and season, that the design cannot establish causation, and that the result may not transfer to other climates or housing stocks. A sentence that converts association into certainty fails the check, even if it reads more cleanly.
Teams also identify the single most likely misreading. If a figure compares two groups, readers may assume the groups were randomly assigned. If a result is a modeled projection, readers may treat it as an observation. The brief should preempt the most plausible misreading in plain language rather than bury a caveat in a footnote.
Plain-Language Revision Under a Word Limit
The drafting constraint is 250 words, excluding the header and a short source line. The limit is not arbitrary. It forces teams to decide what the reader most needs and to cut secondary findings that dilute the central claim. The draft uses four headings: What we know; What remains uncertain; What this could mean for [jurisdiction]; and What we are not saying.
The final heading is the workshop's honesty check. It gives teams a place to state limits without sounding defensive: the study did not measure cost, did not include rural counties, did not follow participants beyond one year, or did not test the policy in question. Reviewers report that this section is often the most useful part of the brief because it tells a busy staffer where the evidence stops.
Plain-language revision is done aloud. One team member reads the draft while another marks any term that would require a glossary. Jargon swaps are collected on a shared board: not 'statistically significant' but 'unlikely to be due to chance alone, given the model'; not 'heterogeneous treatment effects' but 'the program worked differently for different groups.' The aim is not to eliminate technical precision but to make the precision legible.
A Rubric for Accuracy, Usefulness, and Candor
Peer review uses a four-criterion rubric, each scored from one to four. The rubric is deliberately short so that teams can apply it in a single session.
| Criterion | What reviewers ask |
|---|---|
| Claim accuracy | Does the brief distinguish what the figure shows from what the study supports? |
| Stakeholder fit | Is the brief addressed to an office with a plausible decision lever and timeline? |
| Uncertainty candor | Are the main limits stated in plain language without being buried? |
| Plain-language clarity | Could an intelligent non-specialist follow the claim on one reading? |
A score of three or higher on every criterion is the threshold for a completed draft. Reviewers must also answer one open question: what would a skeptical staffer ask first? That question becomes the team's revision target. The workshop ends with a five-minute read-aloud of one revised brief and a short discussion of what changed between drafts.
What the Exercise Changes
The most common revision is not stylistic. It is epistemic. Teams cut claims they cannot defend, narrow the jurisdiction in the header, and add a sentence about who was not studied. They also learn that a policy brief is not a shortened paper. It is a different genre with a different reader, a different failure mode, and a different standard of usefulness.
That lesson belongs in lab training. Researchers who can brief a local official without exaggerating a figure are better prepared for press interviews, grant reporting, advisory boards, and public comment. They are also better readers of policy claims made by others. The one-page memo is a small artifact, but it makes the lab's relationship to public evidence visible — and accountable.
