Seems to me that we can never ensure that payment etc. to participants does not introduce bias; there will always be the possibility that some respondents will tell us 'what we want to hear'---even if not paid in some way. all we can do is constantly check collected data against all the rest.

Famous case apparently was 'Patient Zero', the alleged locus of the AIDS outbreak; later research concluded that this respondent was not PZ and rather, just liked talking to researchers.