Researchers try callbacks, reminders, mixed contact methods, language access, and incentives to improve response. They may weight completed interviews so known characteristics better match the population. Those steps help, but weighting cannot directly recover an unmeasured attitude shared by people who remain absent. A low response rate raises concern, while the actual direction and size of bias require evidence.

Keep nonresponse separate from undercoverage. Undercoverage means the sampling frame did not offer some people a chance to be selected. Nonresponse means selected people did not complete the survey. On a CLEP item, identify where the loss occurred. That single distinction usually reveals which survey error the scenario describes.

Video lesson: Science Behind the News: Opinion Polls & Random Sampling