Of course a polling org "polls everywhere." But if your purpose is to sample a race in Ohio, then you make sure that the attitudes reflect those of Ohio respondents. You ask about issues on the ballot in Ohio--- not Georgia or Massachusetts. You ask about Ohio races as well as the presidential race.
When, however, ALL of your questions deal with a totally different subject (I.e., opinions of speakers at the Dem Convention--and ONLY at the Dem convention) it's clear that you believe all of your respondents have seen or were at the Dem convention. Now, no pollster in his right mind would suppose that such a sample would include ANY Republicans in such a sample, let alone a legitimate sample size for a poll.
Here is what almost certainly happened: PPP had done a poll about convention speakers among DNC attendees, then added on a separate Ohio question (done who knows when) to their reall poll of Dems. But either way, it's totally bogus as a poll.
Is there any evidence for your conjecture?