http://www.reuters.com/article/us-usa-election-poll-idUSKCN0YP2EX
However, if the Reuters poll showed it a tie in mid May, and found Hillary double digits ahead now -- something is wrong with their poll. (The one in mid May, the one now, or both. IIRC Reuters was the first one to show Trump and Clinton essentially tied and I blew it off at the time.)
BTW, THIS is "Reuters Polling Explorer" methodology: http://thomsonreuters.com/en/articles/2013/new-interactive-tool-launched-reuters-polling-explorer.html:
"Instead of using telephones to gather results on a semi-regular basis, we partnered with the research firm Ipsos and opted to reach a huge number of mostly pre-screened respondents online. The result is that we have not only polled far more people than a traditional telephone survey can reach, but we've been polling continuously for more than a year and a half, asking hundreds of questions of about 11,000 people every month."
Caveat emptor. I'm not sure what this poll shows, or doesn't show.
This is exactly why Gallup exited the polling business.
You cannot get a scientifically-based random poll anymore that’s honest and reliable.
Reuters-Ipsos Internet poll is the exact opposite of a random poll.
The numbers simply aren’t believable.
JMHO, it most resembles an adage I heard around the halls of Grad School:
"If you can't dazzle 'em with brilliance, baffle 'em with bullsh**."
...also JMHO, the BS runs very, very deep in this one.