So first of all, that "survey" was hocked on an online college platform that collects your info and forwards you information, which means they're probably profiting off of it. It also means the sample was anything but representative, and was probably less refined than a GCS.
Secondly, based on the data points from the survey
outlined here, it appears that those two-way figures in the graph above are quite misleading:
Almost all of those crosstabs line up perfectly with the numbers in the above quote, which basically mean that they excluded the 30% of people who "chose not to vote", as well as the 9% who chose a third party; that's why all of those numbers add up to 100.
Trump got a whopping one-third of the vote among those surveyed; 20% went to Clinton. It's not as if teenagers like to be edgy and non-conformist - especially when answering questions about their political views.
In any other context, this would immediately be dismissed as a "JUNK POLL". Or, you know, you could just look at the claim that 80% of teenage whites are Trump supporters and save yourself the time.