Any poll of LIKELY VOTERS in a true battleground state that doesn't have the race within 5% either way should be discarded.
I don’t even like polls from “likely voters”. They should only be from registered voters. Nearly every self reporting “likely voter” winds up not voting. It’s an odd thing humans do when they lie to complete strangers asking for honest opinions to improve their image to that stranger. Whether it be scientific studies, doctors, pollsters, nobody wants to look bad even if it hurts their health or messes with data. This is also why what’s being called the “shy vote” is going to make an even bigger impact this year.
The topline data (who ya voting for, trump or Biden) isn’t all that useful. Pay attention to cross tabs and methodology. Cross tabs will show you the demographic break down. A lot of these polls are under sampling republican voters for whatever reason. May not be on purpose, may just be pubs don’t answer, idk. Either way many polls are clearly building in 5-10 pt advantages for Biden by under sampling Republicans. That’s just what happens when you poll 60% dems and 40% pubs in a county that went for trump in 2016. That poll is basically garbage. The fact that many of the pollsters aren’t coming out and announcing “hey we only managed to sample 40% pubs” like they should, tells me it may be on purpose in some of these cases.
Another thing I’m noticing is in methodology. I believe it was the recent Ramuessen that stated only some 70% of republicans say they planning on voting trump vs 94% of dems voting Biden. Even those on the left will admit there’s no way that number is true for republicans. So you wonder what type of questions they’re asking voters. Some of the questions I’ve seen get asked like this. “Do you agree with Biden’s message that America should be more united.” Dogshit loaded question. What if, like in the case of basically all right leaning voters, you don’t view Biden’s message as unifying at all. You’re probably all for a more unifying message, but don’t believe Biden is going to deliver. Better way to ask would be along the lines of “Do you believe Biden’s message is unifying.” Or the inverse that I’ve also seen is “Do you agree with trumps divisive rhetoric” is also a dogshit loaded question. No one wants to be viewed as being or agreeing with “divisive rhetoric”. If you want accurate data, you simply CANNOT frame questions these ways. There’s far to much of this type of this nonsense going on to give these polls much weight.