Exploring variables underlying non-response in a survey of nurses and nurses' aides in practice.

This paper demonstrates that distinguishing between the different sources of unit response failure (i.e. refusals and non-contact) in survey research is methodologically important since minimizing non-contact and refusals requires different approaches. The findings of an interview survey show that unit response failure can be contained successfully through careful study design and management. The potential value of using resampling (i.e. obtaining some information from non-respondents) for identifying: (i) which direction, if any, the data from respondents are biased and; (ii) influencing variables on the incidence of non-response is also demonstrated.