WorldNetDaily has gotten interested in presidential poills over the last week. But not the ones -- that would be nearly every one -- that show Donald Trump losing; it cares only about the ones that show Trump winning, which would be about three of them.
WND's Bob Unruh uses an Oct. 19 article to tout the Investor's Business Daily/TIPP poll, "described as the 'most accurate poll in recent presidential elections,'" which showed Trump ahead of Hillary Clinton by one percentage point.
Unruh doesn't mention that those calling the IBD/TIPP poll the "most accurate" are pretty much limited to IBD and TIPP (FiveThirtyEight.com noted that the poll was among the most accurate in the 2012 election, but it doesn't equate to accurate performance this year), or that IBD is a right-wing newspaper with an interest in showing the Republican winning the election.
While the subhead of Unruh's article mentions "Wildly different results in establishment media surveys," the article itself doesn't address the issue, obscuring the fact that thte IBD/TIPP poll is an outlier in presidential polling this year; most other polls show Clinton ahead by various lengths.
The Huffington Post notes that IBD/TIPP"s outlier numbers show that "seems to be the victim of survey error."
Unruh's article touted two other polls that showed Trump ahead of or tied with Clinton: Rasmussen and Los Angeles Times/USC Dornsife. But those have issues as well, according to the Huffington Post: Rasmussen historically leans Republican, and LA Times/Dornsife's results "can be easily explained by its unique methodology ― it asks 'What is the percent chance that if you were to vote, you will vote for Clinton, Trump, or someone else?' instead of a traditional 'Who would you vote for?' question."
Unruh followed that with an Oct. 23 article touting a "shock poll" with Trump tied with Clinton -- but he's again citing IBD/TIPP, Rasmussen and LA Times/Dornsife.Unruh admits these are "outlier" polls, but also cites the Drudge Report, from which WND apparently stole its "shock poll" headline.
WND's Garth Kant, in an attempt to boost Trump's claim that polls not showing him winning are "phony," tried to throw shade on those polls in an article claiming that they "all sampled significantly more Democrats than Republicans. But, according to the most recent national survey on voter allegiances, conducted by Gallup just after the 2014 midterm elections, more Americans actually side with the Republicans than the Democratic."
In fact, as the Federalist explains, Democrats have traditionally outnumbered Republicans, and "it’s unwise to assume a pollster is biased because its sample included more Democrats than Republicans."
Kant also claimed that "another answer to the mysterious difference in the results of the ABC / Washington Post and IBD polls" is the contents of a stolen WikiLeaks email from the Clinotn campaign that discussed how to oversample polls, to which he added, "The ABC / Washington Post poll showing a 12 point Clinton lead did, in fact, sample 9 percent more Democrats than Republicans."
But that email was not discussing media election polls; as the Washington Post explains, it was an attempt to poll the Clinton campaign's messaging, and the call to oversample certain groups was an attempt to get a large enough sample to more accurately gauge a response to the messaging.
The Federalist notes that "by and large, it’s usually better to assume that results averaged across multiple polls from a variety of polling organizations are probably pointed in the right direction." But those polls don't show Trump winning, so Unruh, Kant and WND would rather attack them and promote the outliers.
UPDATE: Talking Points Memo's Josh Marshall debunks the idea that IBD/TIPP was the most accurate pollster in 2012.