Before doing your survey,
you should test it on at least 5 to 10 people who are not
familiar with the survey and who also are people
you would ask to participate in the survey later (i.e., if
you are doing a survey of restaurant workers then test the
survey on restaurant workers, etc.). Look at the questions
they answered, did not answer and how they answered them.
If it appeared that certain questions gave them problems,
were unclear, or did not get a response, then edit your survey.
If possible, ask them which questions didn't make sense or
were difficult to answer. You can also debrief with the interviewers
and see what their thoughts were about the how the survey
went. Ask yourselves if the responses get to the information
you are looking for. If not, then you need to refine your
questions to bring out the type of data you need. Keep track
of the survey time. If it is beyond 30 minutes for an individual
survey it is probably too long.