Such survey results cannot be used to infer anything without understanding how the survey was worded. For instance, if one of the answers were "I use Facebook, but not the internet" then A) some people will choose it just by virtue of it being there, and B) it could occur to some people that the survey was poorly worded but meant "I do not use the internet for anything other than Facebook".
I remember a rash of such loaded survey questions going around my country a few years ago, it looked like it was specifically designed to be answered with answers that could be used to slander my nation with "N% of these people think this" results. It's usually not disinterested parties investing the time and resources to run these survey, rather, it's usually people for whom the investment will pay off if they can get the results they want.