Last updated: 7 October 2026
Quick Answer: A sampling frame is the actual list or source a sample is drawn from, such as a customer database, address file or phone list. It matters because a sample can only represent people on the frame. A landline list, for example, misses the 78.7% of US adults who lived in wireless-only households in late 2024.
A sampling frame is the working version of your population: the list, file or source from which you actually select people. University of Washington epidemiologist Stephen Mooney puts it simply in a 2019 review: the enumeration is called the sampling frame because it frames the sample selection process.
That framing has a hard consequence. Even a perfectly random draw estimates the population on the frame, not the population you care about, so the frame sets the ceiling on how representative any sample can be.
How Is a Sampling Frame Different From the Target Population?
The target population is everyone you want to describe, while the sampling frame is the list you can actually draw from. In a perfect study they match; in practice the frame is usually a subset of the target population, with some extra entries that do not belong.
Take a brand studying all US buyers of its product. Its target population includes store shoppers, gift buyers and guest checkouts. Its CRM list, the convenient sampling frame, holds only customers who registered an account. The gap between those two groups is where most survey error starts.
How Is a Sampling Frame Different From a Sample?
A sampling frame is the full list you select from, and the sample is the subset you actually select and contact. If the frame is 20,000 customer records and you draw 400 at random, the 20,000 are the frame and the 400 are the sample.
Errors arise at both steps. Frame errors decide who could ever be chosen; sampling and nonresponse decide who ends up in the data. A large sample cannot repair a frame that excluded a group from the start.
What Are Some Examples of Sampling Frames?
A sampling frame example can be a list, a file, a set of numbers or a flow of people, depending on how the population is reached. The table sorts common frames from broadest coverage to narrowest, with the group each one tends to miss.
| Target population | Typical sampling frame | Who it tends to miss |
|---|---|---|
| US households | Census Master Address File or USPS address files | Very new or unconventional housing |
| US adults by phone | Landline and mobile number ranges (dual frame) | People without a phone |
| A brand's customers | CRM or loyalty database | Guest checkouts, cash and store-only buyers |
| A company's employees | HR roster | Contractors and very recent hires |
| A clinic's patients | Appointment and visit logs | People who never booked |
| Website visitors | Live traffic, via an intercept | Mobile visitors who block pop-ups |
The US Census Bureau's American Community Survey draws from the Master Address File, the Bureau's official inventory of known housing units and group quarters, updated in part from the Postal Service's delivery file.
What Is Coverage Error in a Sampling Frame?
Coverage error is the gap between a sampling frame and the target population, and it comes in three forms. The American Association for Public Opinion Research (AAPOR) defines them in its 2016 task force report on address-based sampling:
- Undercoverage: Members of the target population missing from the frame.
- Overcoverage: Entries on the frame that are not in the target population.
- Duplicates: The same unit listed more than once, which raises its chance of selection.
Undercoverage is the dangerous one: nothing in the data flags it. The same AAPOR report cites an estimate that landline-only phone frames excluded 41% of households, and the share of wireless-only adults has kept rising since. The CDC's National Health Interview Survey put it at 78.7% for July to December 2024, and 88.9% among adults aged 25 to 29.
How Do You Build a Good Sampling Frame in Research?
To build a good sampling frame in research, start from the target population, find the source that covers the most of it, then clean the list before you draw. A usable frame is complete, current, free of duplicates and carries contact details that actually reach people.
Check five things before fielding:
- Coverage: Which groups in the target population are absent, and how large are they?
- Currency: When was the list last updated, and how many entries have gone stale?
- Duplicates: Can one person appear twice under different emails or numbers?
- Eligibility: Are ineligible records, such as staff or test accounts, removed?
- Contactability: Does each record include a channel the person actually answers?
Where one source misses a group, combine two, as a brand can by pairing its CRM with store-intercept recruitment.
Who Does Your Sampling Frame Leave Out?
Your sampling frame leaves out anyone who never made it onto the list, and anyone listed with a contact method they ignore. Email-based frames drop people who rarely check email, and online panel frames drop people who never sign up for panels, often the less connected or less fluent customers a brand most needs.
Phone numbers make a wider frame than email addresses for many customer bases, because most people keep one number for years. WhatsApp-native interviews turn a list of mobile numbers into reachable respondents inside a chat they already use, with no link and no app, and AI phone interviews call the same numbers by voice. Alchemic publishes 60+ languages including Spanish and Hindi, with managed fieldwork or bring your own sample across 14 markets including the USA and the UK.
The questions to ask any supplier about respondent reach start with the frame.
When Is an Imperfect Sampling Frame Good Enough?
An imperfect sampling frame is good enough when the people it misses are few, or unlikely to differ on what you measure, and when you report the gap. Every real frame has some coverage error, so the question is size and direction, not perfection.
Weighting to known population totals can reduce the damage, but it only works for groups that are on the frame in small numbers, not for groups that are absent. When a study must describe a whole national population, a probability panel recruited from an address-based frame is the better choice than any customer list, despite its cost and slower fieldwork.
Once it has the client's brief, Alchemic designs and tailors the sampling plan to handle these risks before fielding, rather than leaving that work to the buyer.