First, nobody cares about most instances. There's a bit of political bias in this, in that very often the physical sciences rely less on failure to attribute and more often on analysis of collected data and methodology, so the plagiarism is more likely in the literature review. And if it's not, the data are still valuable and useful--and if the graduate doesn't deliver in the lab or in the field then who cares, the person's a loser who gets bumped into management or education. But if you're into the social sciences or humanities, all that your rep stands on is words. Not many (C) in those fields, apart from econ and some outlier fields.
Tribes protect their own. You fake data, you're deep-6ed. You plagiarize your theses, and you're given a chance to correct them--or somebody says that it's really the school's fault for not making sufficiently explicit the expectations to avoid plagiarism. Or some other crap. The point is that there are different standards by field and since who goes into a field has a decided political/ideological/etc./etc. skew, those standards will necessarily reflect the same underlying skewness.
= new reply since forum marked as read