Would you cite this poll?
“Because the sample is based on those who initially self-selected for participation rather than a probability sample,...”
All this means is people who were willing to be polled, were polled as opposed to pollsters calling 15,000 people in order to get 1000 to respond as happens most times. There is as much chance those who agreed to be polled were pro as they were anti Trump.
What, exactly is your problem with this methodology?