i didnt mean to say there is any bias in sense of random sampling, but rather, in sense of the behavior being measured, i.e., they measured the behavior when it's free money but draw conclusion as if it's one's own money.