[cgbg]

This is an interesting (if lefty biased) discussion on problems with polling:

https://www.pewresearch.org/fact-tank/2016/11/09/why-2016-election-polls-missed-their-mark/

They are basically admitting that voodoo is more scientific than polling.

[MrChips]

Voodoo probably is more scientific. But, but look inside this poll. One major flaw here: this was an automated poll and it did not include the millions upon millions of people who have cell phones. It also does not tell you what specific breakdown, especially by party affiliation, was used. But, setting all of that aside, consider the fact that this poll shows a 10% gap but only 6% undecided. The methodology does not explain where the other 4% went. I seriously doubt that Kanye West has locked up that many votes. Does the Libertarian party even have a candidate? Let’s reduce that 4% to 1%, giving them some benefit of the doubt, and add 3% to the 6% undecided. Then, let’s assume President Trump gathers in 2/3 of that undecided vote. That puts the race at Biden 53%, Trump 46%. That means a flip of only 3.5% for Trump to win. And we’re talking popular vote, not electoral vote, which should come more easily. As we say down here in the south, “Hold on to the huckleberry bushes!“ Keep the faith.