What Is a Sampling Frame?
A sampling frame is the list, map or other device from which you select your sample. A sampling unit is the thing you select from it. If the frame misses people, no amount of randomization will bring them back, so the frame sets the limit on what your survey can claim. Which frame you need depends on the design: sampling methods lists the options, and random sampling shows what the frame must allow.
Sampling Frame, Sampling Unit and Unit of Analysis
These three often get mixed up in a plan. They can be different things in the same survey.
| Term | Question it answers | Example |
|---|---|---|
| Target population | Who do we want to describe? | All farming households in three districts |
| Sampling frame | What do we select from? | The district agriculture office's farmer register |
| Sampling unit | What is picked at each stage? | Villages first, then households |
| Unit of analysis | What does the indicator describe? | Households, or women farmers within them |
In a two-stage survey the primary sampling unit might be a village and the secondary sampling unit a household. Each stage needs its own frame: a list of villages, then a list of households in each chosen village. Listing households only in the villages you picked is one reason cluster surveys are practical; see cluster sampling.
Make the unit explicit. "Households" and "adults in households" give different sample sizes and different frames, and a selected household does not tell you which adult to interview.
Types of Frame
- List frame: A register, roster, enrollment list or customer file. Cheap, but only as good as the last update.
- Area frame: Maps divided into areas, such as census enumeration areas. Used when no household list exists. Households are listed on the ground after areas are selected.
- Multiple frames: Two or more lists combined. Watch for people who appear on more than one.
- Event or flow frame: A visit log or queue, used for exit surveys. The frame is built as people arrive, so select with a rule fixed in advance; see simple random vs systematic sampling.
A Frame Checklist
Test any frame against three questions before you draw from it.
| Check | What to ask | How to test it | What to do about it |
|---|---|---|---|
| Coverage | Is everyone in the target population on the list? | Check a small independent sample of eligible people, such as ones found through a field walk, against the list; compare list totals with an outside count | Add an area frame for missing groups, or narrow the stated population |
| Coverage (the other way) | Is anyone on the list who should not be? | Check eligibility fields on a sample of entries | Screen for eligibility at selection |
| Duplicates | Does anyone appear twice? | Sort by name, phone or ID and look for repeats; spot-check near matches | Deduplicate before selection, or record multiplicity and weight later |
| Updates | How old is the list, and who maintains it? | Check the last update date; trace a sample of entries to see how many moved or died | Refresh the list, or plan for out-of-date entries |
| Identifiers and contact data | Can each selected unit be found? | Test a sample of addresses or phone numbers | Budget for tracing; do not replace units casually |
| Documentation | Can someone else repeat the selection? | Record the file version and date the list was taken | Keep a copy with the sampling log |
A Worked Frame Check
A fictional youth skills program keeps a register of 4,000 trainees and plans a satisfaction survey of 385 completes (see choosing a sample size).
- Coverage: Staff trace 100 trainees they know from class attendance records. 91 appear on the register, so about 9% of eligible trainees are missing, probably recent enrolments. The survey can claim to describe registered trainees, not all trainees, unless the gap is closed.
- Duplicates: Sorting by phone number in a check of 200 entries finds 6 repeats (3%). Remove them.
- Updates: The register was last updated 14 months ago. Tracing a sample finds that 14 of 200 entries (7%) have left the program and are not eligible.
- Effect on the plan: After removing duplicates and ineligible entries, about 90% of the 4,000 are usable, so the frame holds roughly 3,600 eligible trainees. To draw 385 completes you must also plan for 10% ineligible entries among those you select: 385 / 0.90 = 427.8, so select 428 and follow the replacement rules in the field protocol.
The numbers are for illustration. What matters is that each problem was measured, written down, and linked to a decision. Keep the audit with the sampling log.
Common Frame Problems
- Using a beneficiary list to describe a wider population: The list covers only people who already reached the program.
- Replacing hard-to-find units with easy ones: This turns a probability sample into a convenience sample and ignores the people the frame was meant to reach.
- Outdated lists in mobile populations: Displaced households, migrant workers and new arrivals are the first to be missing.
- Frame and unit mismatch: A list of phone numbers is a frame of phones, not people.
- Unequal chances from multiple entries: A household with two phones has twice the chance of selection unless you adjust.
For an overview of the problems that follow from these, see common sampling mistakes and, for what a missing group does to results, bias.
Frequently Asked Questions
What Is the Difference Between a Sampling Frame and a Population?
The population is who you want to describe. The frame is the list you can actually reach. The gap between them is coverage error, and you should name it in the report.
Do I Need a Frame for Every Survey?
For probability sampling, yes: you need some way to select with known chances, whether a list or a set of maps you can sample areas from. Without one, you are limited to non-probability methods; see sampling methods.
Is a Sampling Unit the Same as a Respondent?
Not always. The unit may be a village, school or household; the respondent is the person who answers for it. State both in the plan.
Last updated October 31, 20264 sources checked
M&E Studio also does this work for clients.
The team that writes and reviews this library takes on M&E system design, evaluation quality reviews, AI advisory and workflow automation.
sample frame, sampling frame definition, sampling unit, survey frame, list frame, area frame