4.1.4. Non-Response errors

Non-response can be complete non-response (if a sampling unit, i.e. a beekeeper, did not answer at all) or partial non-response (some questions are not answered or only partially answered).

Keeping questions clear and simple can help to reduce the chance of missing data. Emphasising the importance of the survey and explaining how the results will be used may increase the response rate. If the participant can appreciate that there is some benefit to completing the survey, they will be more likely to take part. Use of rewards and incentives can be useful to increase participation of people selected already to take part in the survey.

Non-response bias occurs when there is something systematically different about participants who do not respond from those who do. Trying to minimise non-response is therefore very important. Reminders are useful in this regard. Non-response can in principle be estimated by randomly sampling some sample units after termination of the survey, approaching the non-respondents with another survey mode and comparing the main outcome with the survey response on these sample units. However such an effort is rarely regarded as a good use of scarce resources of time and money.