A service list reaches service users
An assessment about unmet need that recruits only from current program records may miss people who could not access the program. An online link may miss people without suitable internet access or those who never encounter the invitation.
These are design issues before they become response-rate issues.
Compare the frame with the target population
- Define whom the assessment intends to describe.
- Identify who appears in the contact source.
- List groups likely to be absent or hard to reach.
- Choose additional methods or narrow the claim appropriately.
Changing recruitment methods can help, but each method has its own selection process.
Probability sampling needs actual probabilities
Randomly selecting names from an incomplete list is random within that list. It does not make the list cover everyone. Similarly, calling a voluntary online survey “random” because anyone can click it misdescribes the design.
CDC's CASPER methodology illustrates a defined household sampling approach with specific procedures. It should not be reduced to borrowing a sample size while ignoring the sampling design.
The most useful report explains who could have been selected, who participated, and where generalization remains limited.
