Nonresponse Bias

The distortion that occurs when people who don't respond to surveys, studies, or feedback requests differ systematically from those who do respond—making the collected data unrepresentative of the full population. Customer satisfaction surveys typically have 10-30% response rates; the 70-90% who don't respond may be systematically different (too busy, too apathetic, too frustrated to bother). Employee engagement surveys over-represent engaged employees (who care enough to respond) and under-represent disengaged ones (who've already checked out). Every dataset with less than 100% response rate is potentially distorted by nonresponse bias.

When to use it

When survey results seem too positive or too uniform (nonresponse may be hiding the negative); when response rates are low and conclusions are being drawn from a small fraction; when feedback mechanisms are being used for important decisions; when customer, employee, or market research needs to be representative.

How it can help

For any survey or feedback mechanism: estimate how nonrespondents might differ from respondents, and adjust conclusions accordingly. Boost response rates through brevity, incentives, and follow-up. Track response rates by segment to identify where nonresponse is concentrated. When possible, compare respondent characteristics to known population characteristics to quantify the bias. The most important practice: never present survey results without the response rate, and never interpret results as representing the full population when only a fraction responded.