Coverage Error and Undercoverage
What You'll Learn
You'll learn how missing parts of your population from your sampling frame leads to systematic bias in your estimates.
The Core Problem
Coverage error occurs when your sampling frame—the actual list or method you use to select your sample—doesn't include everyone in the target population. This creates undercoverage, meaning some groups are systematically left out before sampling even begins.
Unlike sampling error (random variation from sample to sample), undercoverage is a form of sampling bias that systematically distorts your results in one direction.
A Real-World Analogy
Imagine you want to understand smartphone usage across all adults in a city, but you draw your sample exclusively from a list of landline telephone numbers. You've automatically excluded people who only use cell phones—likely younger adults and urban professionals. No matter how carefully you randomly sample from that landline list, your results will be biased toward older, more traditional phone users. The missing groups create a coverage gap.
Common Sources of Undercoverage
- Outdated lists: Phone directories missing new residents or unlisted numbers
- Online-only surveys: Excluding people without internet access
- Geographic limitations: Missing homeless populations or remote areas
- Administrative records: Lists that don't capture informal workers or undocumented individuals
Why It Matters
Undercoverage creates systematic estimation problems. Your sample statistics will consistently misrepresent population parameters because entire segments are absent. Unlike sampling error, you can't fix this by simply increasing sample size—you need a more complete frame.
Key Takeaway: Coverage error occurs when your sampling frame systematically excludes portions of the target population, creating biased estimates that no amount of random sampling can fix.