A poll Nate Silver once said 'has no idea what they're doing' helped determine who made the GOP debate stage
Fox Business, which is hosting the next series of GOP debates on Tuesday, took the average of the last four national polls to determine which candidates would make the main stage. Eight candidates qualified for the primetime event, which required at least a 2.5% average in those polls.
The four polls used were from Fox News, Quinnipiac University, The Wall Street Journal/NBC News, and Investor's Business Daily/TIPP.
The IBD/TIPP poll, in particular, has been subject to scrutiny over the past two election cycles. In 2008, the survey found Republican nominee John McCain crushing Democratic nominee Barack Obama among youth voters, 74-22. (Obama won the youth vote, 66-32, according to exit polls.)
This prompted polling guru Nate Silver, the editor in chief of FiveThirtyEight, to famously declare in 2009 that the pollster had "literally no idea what they're doing. I mean, literally."
The pollster performed better during the 2012 election. But so far in 2016, it has been subject to the same kinds of wild swings for which it has become known.
For example, it was the only pollster in September to find real-estate mogul Donald Trump trailing retired neurosurgeon Ben Carson - and by 7%. The next month, it found a collective 12-point swing where Trump was up by 5%.
To be fair, the IBD/TIPP poll likely didn't keep Christie off the main stage. Even if the survey had not been included, the most recent national poll after that was from CBS and The New York Times - which, like the IBD poll, found Christie at 1%.
However, it did doom Huckabee's chances of getting on the primetime stage. He received 1% in the IBD poll, compared to the 4% support level he achieved in the CBS/New York Times survey. Swapping the two polls would have given him an average of 3%.
On Friday, MSNBC host Joe Scarborough raged over the poll selection for the debate. He was particularly irate at the inclusion of the IBD/TIPP poll.
The grumbling about the usage of national polls to determine debate standing has increased this week, as Christie and Huckabee were booted off the main stage and Sen. Lindsey Graham (R-South Carolina) and former New York Gov. George Pataki (R) were kept out of even the lower-tier debate.
It intensified after the off-year Election Day on Tuesday, when polling leading up to the Kentucky gubernatorial election was wildly off base. Some GOP candidates have reportedly pushed for different criteria - some have suggested holding two debates with candidates chosen at random.
Pollsters themselves have also been critical of the format. Tom Jensen, the director of Public Policy Polling, explained the problems in an email to Business Insider. In a random sample of 500 voters, for example, "you're basically talking about a difference in 5 supporters within a poll making the difference between Rand Paul and John Kasich being on the main stage, and Chris Christie and Mike Huckabee not," he said.
"That's way too small of a difference to use to determine who really gets the chance to break out in the race and who doesn't. Polls simply aren't precise enough - and could never reasonably be expected to be - to use half-point differences in them to decide who stays and who goes