How does nonresponse affect what a survey sample can represent?

It tells you that your answers came from 30 of the 100 people you selected; the other 70 did not provide answers. In this hypothetical example, 30 out of 100 is a 30% response rate. Keep those groups distinct: the 100 selected people are not the same as the 30 respondents.

Your results describe the answers you received. Whether they also represent the selected group depends in part on how respondents’ views compare with the nonrespondents’ views. Nonresponse can make collected answers less representative, for example if people with especially strong positive or negative opinions are more likely to answer. [S1]

Not from the response rate alone. A low rate tells you that many selected people are missing, but it does not tell you whether those people would have answered more positively, more negatively, or much like the respondents. So it cannot establish the direction or size of possible nonresponse bias. [S1]

For example, suppose the 30 respondents mostly say they enjoy the club. That answer pattern alone cannot show whether the 70 nonrespondents enjoy it less, enjoy it just as much, or enjoy it more. Avoid presenting an adjustment as a known correction unless you have relevant information about respondents and nonrespondents; the response count by itself does not provide that information.

Start with a transparent accounting, then compare only information you actually have. For the hypothetical survey, your notes could say:

“Of 100 people selected for the hobby-club survey, 30 responded and 70 did not. The responses show what those 30 people reported. We do not know from the response count alone whether the 70 nonrespondents would have answered differently, so the results may not represent everyone selected.”

If you have appropriate, relevant information for both groups, you can describe what it shows and where it is incomplete. Do not assume that a difference in one known characteristic proves how nonrespondents would answer the survey question. The response rate supports caution, but it does not establish a method for correcting this particular survey or predict its bias.

Use this checklist before sharing the findings:

1. Record how many people were selected and how many responded.
2. Keep the respondent group distinct from the people originally selected.
3. Describe the results as answers from respondents, not automatically as answers from everyone selected.
4. Note whether you have relevant information about nonrespondents, and state its limits.
5. Explain that the response rate alone does not show the direction or size of possible nonresponse bias.

Copy and adapt this template: “We selected [number] people for a survey about [topic]. [number] responded. The results describe the responses received. We [do/do not] have relevant information comparing respondents with nonrespondents: [brief description or ‘none’]. The response count alone cannot show whether nonrespondents would have answered differently, so representativeness remains uncertain.” Keep the bracketed parts as placeholders until you can fill them with your actual survey information.