WASHINGTON _ Donald Trump is making the same mistake as President Mitt Romney.
He insists that public polls showing him trailing Hillary Clinton are biased results because they "oversample" Democrats. But that's exactly the same argument that a lot of conservatives used in 2012, wrongly convincing themselves that Romney would win easily and waking up shocked to find they were wrong and the polls were right.
The problem is, they claim something about the way polls are conducted that simply isn't true.
That isn't stopping Trump from railing against "oversampling."
"When the polls are even, when they leave them alone and do them properly, I'm leading," he said. "But you see these polls where they're polling Democrats. How's Trump doing? Oh, he's down. They're polling Democrats. The system is corrupt and it's rigged and it's broken."
Other conservatives were also solidly convinced in 2012 that pollsters' "oversampling" of Democrats was hiding the truth that Republican presidential nominee Romney was leading incumbent President Barack Obama _ by a lot.
The conservative Breitbart website, for example, stated flatly in September 2012 that news media polls showing Obama ahead were deliberately skewed to help the Democrat. "As is becoming routine, these new polls again oversample Democrats," the site said.
Another website called unskewedpolls.com doctored the news media polls to give Republican voters more weight. "The results have been staggering," said a column on Fox News. "The re-weighted polls all put Romney ahead of Obama with margins of between 3 and 11 points."
Fox News commentator Dick Morris was so sure the polls were skewed toward Democrats and Romney was actually far ahead that he predicted that the former Massachusetts governor would win the White House with 325 electoral votes to Obama's 213.
"It will be the biggest surprise in recent American political history," he told Fox's Greta van Susteren in late October 2012. "It will rekindle the whole question on why the media played this race as a nail-biter where in fact Romney's going to win by quite a bit."
Obama beat Romney by 51-48 percent, earning 332 electoral votes to Romney's 206.
Some still say the media are deliberately sampling too many Democrats.
"If we can put up that ABC/Washington Post poll once again, Donald Trump is trailing Hillary Clinton by 12," "Fox & Friends" co-host Steve Doocy said in during a June telecast.
"But what they don't tell you ... they actually talked to 12 percent more Democrats than Republicans. According to the Gallup poll, there are 3 percent more Democrats in the country than Republicans, so it looks like they've got a favorite in it."
The key to their mistake then _ and Trump's apparent mistake now _ was thinking that the number of Democrats and Republicans in a poll should match party registration. They argued that any poll with more Democrats than the number they thought it should be was a biased "oversampling" of Democrats.
But pollsters often use samples of Democrats and Republicans based on a participant's political self-identification at the moment, not registration. Some states don't even have party registration statistics available.
Brad Coker, managing director at Mason-Dixon Polling & Research, said Trump was "playing fast and loose with the word 'oversample.' "
"Most states don't register by party, so you don't know if 45 percent of the country is Democrat, 30 percent of the country is Democrat," Coker said. "And it's fluid. I know people who are lifelong Republicans who don't like the Republican Party now, so they say they're independents. It's always changing."
Some conservative commentators are wary of accepting Trump's claim of rigged polls.
"I'm not gonna (get) caught in a trap again of automatically rejecting what's out there but all the while understanding, this is a different election," Rush Limbaugh said on his nationally syndicated radio show Monday.
"Look, there's nobody that wants these polls to be more wrong than I do. And in 2012, I sat here every day and told you they were wrong because they were using the wrong turnout model, that they were not factoring turnout in 2010. All those tea party people."
In addition, Trump mischaracterizes a common polling practice that really is known as oversampling.
Trump seized on a 2008 email from the account of John Podesta, then and now a top Clinton adviser.
"WikiLeaks also shows how John Podesta rigged the polls by oversampling Democrats, a voter suppression technique," Trump said. "And that's happening to me all the time."
In the email, another operative wrote, "I also want to get your Atlas folks to recommend oversamples for our polling before we start in February. By market, regions, etc. I want to get this all compiled into one set of recommendations so we can maximize what we get out of our media polling."
However, that was describing the common polling practice of surveying more members of one group to get a statistically meaningful deeper look at that one bloc. Polls often, for example, will get only a sliver of Hispanics in the final total, too few to draw any conclusions about their thoughts. So a poll will also seek a broader sample, an "oversample," of Hispanics. But those numbers are not traditionally used in the overall poll.
"He's saying oversampling is having too many of a particular group when actually it's done because you have too few," said Lee Miringoff, the director of the Marist Institute for Public Opinion in New York. "Oversample for pollsters is that you're interested in finding out more about a group you typically don't get enough of. You interview more of the group, but it doesn't enter into your overall totals."
Trump hasn't sworn off all polls. Campaigning in Tampa, Fla., earlier this week, he touted an Investor's Business Daily poll Sunday that showed him leading Clinton by 43-41 percent in a four-way race that includes Libertarian Party candidate Gary Johnson and Green Party nominee Jill Stein.
"The most accurate poll for the last three presidential elections, by far, has us up 2 points nationwide," Trump told the crowd.
But his lead faded into a statistical tie in an Investor's Business Daily tracking poll the next day.
