I believe these media outlets hire polling companies to do the polls. These companies have no vested interest in being proved wrong. After elections, they get bragging rights - or do damage control - over who was most correct. This is important as these companies compete for private marketing clients when there are no elections.
The polls have different results. That is more a function of how they forecast the electorate than any bias in the outcome.
Trump is behind, as of now. Whether he can make up ground depends upon whether he can convince people they want him as POTUS, as well as factors beyond either candidate’s control, such as major terrorists attacks.
How can you be sure? The polling models try to figure out what voting patterns will be, based on previous elections. Trump is appealing to a lot of people who aren't always motivated voters. Add to that, there are groups from which his support is massively high, and thus will be under-reported by these models.
My guess is that's why the dems, media and disgusting nevertrumpista backstabber Republicans went all-in with the Khan rigged story. It was an attempt to cut into Trump's support in Veterans and the military.
Add to that, Trump's past two days' rallies have been really, really good. He seems focused on the message, and has Pence and Flynn with him. The "insiders" who gave the media damaging info seem to be gone; I'm suspecting there were one or two high-profile infiltrators.
For all those reasons, I can't understand why you say "Trump is behind". We need polls with targeted foci because of the uniqueness of this election. Saying "he's behind" amounts to pushing the inadequate polls, and pushing the narrative can sometimes make something so.
Nonsense. The corruption is so obvious a child can see it.
I took a statistics course in graduate school years ago, working on my MBA. It was one of the best classes I have ever taken and I remain skeptical of poll results to this day because of what I learned in that class.
A poll is made up of numbers and statistics are just a way to analyze those numbers. And one thing I learned in that class is that one can make those numbers say anything you want, depending on your viewpoint or agenda. Completely different outcomes arrived at from the same exact numbers.
BTW, the title of the textbook for that class was "How To Lie With Statistics."