My own experience: When I have been on a University campus' wifi and needed to look at a paper, I could download and see the entire article, not just the abstracts. This has been true for all of the limited number of times I have tried this. I have also done the same at some, but not all, conferences; typically the very large ones only. I would then doubt that the numbers are being skewed towards the scihub and other 'rebellious' forms of paper acquisition. However, this is only my experience.