Sample frame
A sample frame, or sampling frame, is the actual list or source a study draws its participants from, such as a customer database, a panel or a phone register. It stands in for the population, and gaps between the two cause coverage error.
How it works
Researchers define a target population, all the people the findings should describe, but can only sample from what they can list or reach. The frame is that practical list. When the frame leaves out part of the population (undercoverage) or includes people outside it (overcoverage), the results are biased before a single question is asked, and a larger sample does not fix it.
The classic failure is the 1936 Literary Digest poll, which predicted a win for Alf Landon from about 2.4 million responses. It polled its own readers and lists of car owners and telephone users, groups far wealthier than the average Depression-era voter, and Franklin Roosevelt won in a landslide. Later research blamed nonresponse even more than the frame. Researchers check a frame against the population before sampling and supplement it where it falls short.
A worked example
Only 78 percent of a regional rail operator's season-ticket holders have an email address on file, so a satisfaction survey sampled from the email list misses the rest, who skew older and renew at station ticket counters. A second frame, recent counter customers, adds 300 of them, and satisfaction with counter service drops 9 points in the combined results.
Related
Automate this in Pulse
When a customer list misses a group, say prospects who never bought, Pulse can recruit to that group's profile and close the coverage gap in the frame.