Response rate

The response rate is the proportion of sampled and contacted individuals who complete a survey, or complete it fully enough to count, out of everyone eligible and selected to be asked. In simplest form, response rate = completed interviews / eligible sample, though several standardized formulas exist, such as those published by AAPOR, that differ in how partial completions and cases of unknown eligibility are treated, so published rates are not always directly comparable across surveys.

A low response rate is a symptom, not itself proof of bias: results are only distorted if the people who respond differ systematically from those who do not, on the specific thing being measured, which is nonresponse bias. A survey can have a high response rate and still be biased if its sampling frame was flawed to begin with, and it can have a low response rate with little real bias if nonrespondents resemble respondents closely enough.

Response rates for telephone and mail surveys have fallen for decades across the industry, pushing researchers toward alternative modes, online panels and text surveys among them, and toward heavier reliance on survey weighting to compensate for who was missed. The main pitfall is treating response rate alone as a quality score: a lower-response survey with careful weighting and a solid frame can outperform a higher-response survey built on a poor sampling frame.