‘Don’t ask me nothin’ about nothin’, I just might tell you the truth’. The interaction between unit non-response and item non-response

Abstract

Organisations conducting surveys usually concentrate far more of their efforts on minimising unit non-response than on minimising item non-response. This is partly because the potential damage to survey quality caused by unit non-response is far greater than that caused by item non-response
Also, although unit non-response is more prevalent, it is also generally easier to rectify. Reducing unit non-response is achieved in the long term by interviewer training, and in the short term by various response maximisation. If, despite this, the final response rate is deemed unsatisfactory, further steps can be taken to increase it. In the case of item non-response, the main means of reducing it is again interviewer training. Post-survey, there is virtually no chance of converting item non-response into response.
There is, however, a third component, which is less discussed, and this is the interaction between unit and item non-response. There is no doubt that there is some correlation - people who are more reluctant to take part in surveys will also be more reluctant to answer individual questions, and vice versa