I'm not quite sure why, but Congressional polling is generally quite poor.
This is from an article on the U.S House race part of the poll:
Two years ago at about this same stage, our poll showed Mills with an eight-point margin. Three weeks later in lost to Nolan by 1.4%. Schier says that was likely the result of the Democratic Party having a far superior get-out-the-vote operation. "I think the lead in the polls evaporated two years ago because the Democrats had a ground game that far outstripped the Republican ground game in that district," Schier says.
When the polls turn out to be wrong, a superior get-out-the-vote operation is a standard excuse of the pollster.
I think that a 9.4% difference between a late poll and the actual result being explained by a superior get-out-the-vote effort is ridiculous.
http://kstp.com/politics/stewart-mills-rick-nolan-8th-congressional-race-kstp-survey-usa-poll/4299301/I'm not claiming this poll is wrong just providing evidence of how poor polling by Congressional District whether for the House race or the Presidential race frequently is.