I think 'lying' is a little harsh when used in the context of research (or ethnographic research). For example, I have lost count of the number times there has been a difference between something a respondent did and said. They weren't lying as such. I'd prefer to call it 'contextual reality'. And our approach to truth is that there are many truths which need to be disentangled before an insight can be cut.
Here is an excellent, perhaps a little simplistic, article I read in the Guardian recently.