Fox News polls from yesterday point to bad news for Trump with Clinton well up in Colorado and Virginia.
So now the question easily becomes, "Who is polling better?" Rasmussen has a republican bias, but so does Fox News. Rasmussen shows Trump up, Fox News shows Clinton up.
Independent polls show it all over the place. So it begs the question, is polling technique and analysis being tested here? The results could be greatly affected by who is getting polled (or who can actually be reached).
Let's look at the numbers from the polls taken after the email FBI announcement.
Gravis has Clinton up by 2 points. It was an auto-dial poll, no word on cell v land breakout. Breakout was 78% white and 16% Hispanic, 5% other.
Fox News has Clinton up by 10 points. It was an auto-dial poll with about 3 in 5 being cell and 2 in 5 being land. No breakout on race. However, their results on the Senate match up well with Monmouth. Monmouth's breakout was 83% white and 11% Hispanic, 3% black (three times that of Gravis), 2% other.
In 2012, the turnout was 78% white, 14% Hispanic, 3% black. You know under-representing blacks by even 2 pts will make a difference in the polling because blacks will likely not vote for Trump, perhaps at all. But even accounting for that, Fox News and Monmouth seem much higher than Gravis, meaning the white people being polled are from a little bit different pools.
All this number porking tells us, while we know polls are a bit off with accuracy, they may be impossible to judge from company to company.