EDORA
Skip to content
∑

EDORA • Learn

Justice Index · Advocacy Lab · Field Guide

EDORA Learn — Methods

Small-N, Suppression, and Pooling Rules

When data slices become very small, both privacy risk and statistical volatility increase. Most research systems apply explicit “small-n” rules to protect identities and prevent misleading fluctuations. This page summarizes the main approaches found in the source material.

Why Suppression Exists

Small groups—such as a single county or program with only a handful of youth—can reveal identifying information when cross-tabulated by age, gender, or race. Even when identity protection is not the primary concern, a small denominator makes rates swing dramatically. Suppression, therefore, serves both ethical and analytic purposes.

Common Thresholds

  • Many agencies suppress data when the count of people or cases in a cell is less than 10 to 25, or when fewer than 5 events occur.
  • Some systems use complementary suppression—hiding adjacent cells that could reveal a suppressed value through subtraction.
  • Aggregated “statewide” or “regional” totals are often shown instead of small local counts.

Pooling and Smoothing

To keep data visible without violating thresholds, researchers may combine several years or similar counties into pooled groups. A two- or three-year rolling window can stabilize rates that would otherwise fluctuate from random variation alone.

  • Year pooling: Combines adjacent years (e.g., 2020–2022) to reach minimum sample size.
  • Geographic pooling: Groups neighboring small jurisdictions into a shared reporting area.
  • Weighted averaging: When pooling, totals are usually weighted by the number of people in each year or area, not by a simple arithmetic mean.

Annotation and Transparency

Suppression and pooling should never be silent. Good data practice labels any hidden or merged value with symbols such as “—” or “Pooled 2021–2022.” A note should describe the threshold used, the years included, and whether the series break affects comparability over time.

When to Apply Each Approach

  1. Suppress if counts are below the minimum and pooling cannot reach the threshold.
  2. Pool if adjacent years or nearby units raise reliability without distorting meaning.
  3. Annotate any resulting chart or table with explicit notes on how the change affects interpretation.

Data & Methods

The research text file shows that suppression and pooling practices vary by state and dataset. Some use automated thresholds baked into data dashboards; others rely on analyst judgment. Regardless of implementation, the consistent principle is that small counts should not mislead readers or expose individuals. Where values are hidden or averaged, the documentation should specify exactly how and why.

Related

Transparency note: Suppressed or pooled values should always be documented in metadata and labeled visibly in any public display to prevent misinterpretation or inadvertent disclosure.

Back to Learning Center