Metrics

Prevalence

Definition

The proportion of a population that is responsive, often used interchangeably with richness.

Why it matters

It drives every sample-size and recall-uncertainty calculation.

In practice

Prevalence is the proportion of a population that has the trait you are looking for, here responsiveness. In eDiscovery it means the same thing as richness.

Prevalence drives sample design. The rarer responsive documents are, the larger a sample needs to be to say anything precise about them, and the more a few documents either way move the estimate.

Formula

Prevalence = responsive documents in a random sample ÷ sample size

Worked example

A random sample of 800 documents from 250,000 finds 96 responsive: 12% prevalence. The 95% interval is roughly 9.7% to 14.3%, which projects to about 24,000 to 36,000 responsive documents in the collection.

Illustrative numbers, not from a real matter.

Compare with

  • Richness: Same quantity, different vocabulary.
  • Elusion Rate: Elusion rate is prevalence measured only in the documents set aside after review.

Work it out

← Back to glossary