Data Analysis for PolicyMaking in Georgia
Module 6 · Writing the Policy Data Brief › 6.3 Writing the finding section
Subsection 6.3

Writing the finding section

~3 min

Reading

The finding section is the heart of the Policy Data Brief. It is what everything else serves. If your finding is vague, hedged beyond meaning, or unsupported by the data, the brief fails. If it is overstated, it becomes a liability. The target is a finding that is precise, honest, and usable.

The 2–3 sentence rule

The finding section of a one-page brief should be 2–3 sentences. The first sentence states the main result with a number. The second sentence adds context or the comparison. The third, if needed, notes a methodological limit or specifies the most important sub-finding.

One number that matters

Choose the single most meaningful statistic. If you have a finding about mail ballot rejection rates, the number is not "we found several interesting patterns across 159 counties"—it is "2.3× higher rejection rate in high-Black-share counties." Lead with that.

No jargon

Do not use: ecological inference, CVAP, RPV, ACS 5-year estimate, VTD, or any acronym without spelling it out. Write for a reader who has not taken this course. "The number of eligible voters (citizens 18 and older) in each district" is better than "CVAP."

Georgia redistricting example

Research question: "Did the 2021 Georgia General Assembly's congressional redistricting plan reduce the number of districts where Black voters constituted a majority of eligible voters, compared to the 2011 plan?"

Finding section example:

"Under Georgia's 2021 congressional district plan, Black citizens of voting age (18+) constituted a majority of the eligible voter population in one of fourteen congressional districts. Under the 2011 plan, Black citizens constituted a majority in two of thirteen districts. This comparison is based on American Community Survey five-year estimates for eligible voter populations; these estimates carry margins of error of approximately ±2–3 percentage points for individual districts, and should be interpreted as approximate. Whether the change from two to one majority-Black district resulted from neutral redistricting criteria or from intentional discrimination would require examination of the legislative record and district simulation analysis beyond the scope of this brief."

What this example does well

It leads with the number (two to one). It names the metric (Black citizens 18+ as a share of eligible voters). It acknowledges the data source and its margin of error. It explicitly states what the finding does not establish (intent), which is critical post-Callais.

Anatomy of a Strong Finding Section — Annotated "Under Georgia's 2021 congressional plan, Black citizens of voting age (18+) constituted a majority in one of fourteen districts. Under the 2011 plan, two of thirteen districts were majority-Black by eligible voters. This comparison is based on ACS five-year estimates; margins of error are approximately ±2–3 percentage points per district. Whether the change from two to one majority-Black district resulted from intentional discrimination would require additional evidence — beyond the scope of this brief." Sentence 1: The number first. Lead with the result, specific metric. Sentence 2: The comparison. Before/after, explicit baseline. Sentence 3: Methodology limit. Data source + margin of error noted. Hedge on intent (post-Callais). Effects ≠ intent. This is critical after Callais v. Landry.
Diagram 6.3 · Annotated finding section. Each sentence serves a specific purpose. The post-Callais hedge on intent is required for redistricting analysis.