The cheap-talk point is really important. Most of the difficulty in designing economic experiments revolves around eliciting true preferences. Well designed experiments often have entirely different results to surveys that just ask what people believe about themselves.