Donald Trump Is “Disparaging Polls That Show Him Down,” Blasting Poll Oversampling As “Voter Suppression”

During Rally, Trump Calls Poll Oversampling A “Tactic Of Voter Suppression.” During an October 24 rally in St. Augustine, FL, Trump criticized polls “he once held in such high esteem” and “decr[ied] the polling practice of ‘oversampling’ calling it a tactic of voter suppression.” NBC News noted that in fact “oversampling is standard practice for pollsters and can give a deeper look into larger groups of voters.” The candidate cautioned supporters against believing the polls and underestimating him. From an October 24 NBC News article:

The Republican nominee promised that if elected he would be the voice of the people, a voice that would “boom through the halls of Washington” and prove that this election would be “bigger than Brexit.”

That is, of course, if Mr. Trump pulls off a win 15 days from now. The polls he once held in such high esteem and gleefully spouted from his podium during the primary have now drawn his ire and wary eye. In fact, the GOP nominee has spent much of his dwindling time on the trail disparaging polls that show him down. Of late, Trump has begun decrying the polling practice of “oversampling” calling it a tactic of voter suppression. “It's called voter suppression,” Trump extrapolated of the goals of oversampling. “Because people will say 'oh gee, Trump's out.' We're winning, we're winning.”

In actuality, oversampling is standard practice for pollsters and can give a deeper look into larger groups of voters.

But Trump cautioned of underestimating him, as some did during the primary process. “Remember what he said?” Trump reflected on President Obama's nay-saying in the early part of the this year. Mocking the president, Trump mimed, “Donald Trump will never win the Republican primary, he will never do it, sorry. Sorry, he will never win. The Republicans will never do that. Well, they did that. Sorry.” [NBCNews.com, 10/24/16]

Fox Figures Echo Trump’s Claim That “Oversampling” Is A Way Of Rigging Polls

Fox’s Steve Doocy: “If You Oversample Certain Demographics, Then You Wind Up With Better Results.” Fox & Friends co-host Steve Doocy claimed oversampling in polls is a way to “manipulate data” and give “better results.” Co-host Brian Kilmeade echoed Trump’s claim that oversampling is a form of voter suppression, suggesting that if Trump supporters see a “12-point advantage for Hillary Clinton,” they might not show up to support their candidate. From the October 24 edition of Fox News’ Fox & Friends:

STEVE DOOCY (CO-HOST): Speaking of rigged, thanks to Wikileaks we have just seen exactly how John Podesta and company over in the Clinton camp were going to -- they put out a multipage document in detail showing exactly how to manufacture detailed and desired data --

ABBY HUNTSMAN (CO-HOST): You don't say.

DOOCY: By oversampling. If you oversample certain demographics, then you wind up with better results for you.

[...]

BRIAN KILMEADE (CO-HOST): People are saying, OK, why are the polls the way they are, why are they sampled where they appear to be. And maybe they say that the Democrats want to make everybody feel like the die has been cast. If you look at the ABC poll and you believe what the ABC poll says today, it is a 12-point advantage for Hillary Clinton. Now somebody might be going, “I’m not even going to vote. You know, my guy is not going to have a chance. Not going to show up.”

DOOCY: Sure. But when you look at the ABC poll, where it's 12 points ahead for Hillary Clinton, if you look in the fine print, they asked 9 percent more Democrats. So, you know, surely, there is an edge in registration for Democrats. But what they're talking about is the way you manipulate the data. For instance, this thing goes on to say in Arizona, oversampling of Hispanics and Native Americans is highly recommended. In Florida, make sure the sample’s not too old -- because apparently they feel they vote Republican --, has enough African American and Hispanic voters, and on independents when it comes to the cities of Tampa and Orlando, those are better persuasion targets than North or South Florida. So include Tampa or Orlando first. So if you essentially bake the cake that way, you’re going to wind up with better numbers for Hillary. [Fox News, Fox & Friends, 10/24/16]

On Fox, Rudy Giuliani Claims If It Weren’t For Oversampling, “You Would Have Pretty Much An Even Race.” Trump surrogate and regular Fox News guest Rudy Giuliani claimed that the “oversampling of Democrats is apparent” in the latest polls and that if it weren’t for oversampling, polls would show that the race is “pretty much ... even.” When asked by Fox & Friends co-host Ainsley Earhardt why pollsters oversample instead of “count[ing] the vote as is,” Giuliani said pollsters are “making assumptions” about the turnout. From the October 25 edition of Fox News’ Fox & Friends:

RUDY GIULIANI: Well something is going on. I do think the oversampling of Democrats is apparent. Now, the pollsters have an explanation for that,  they’re oversampling by about 10 percent Democrats. I think if I got that sample down to about five percent oversampling for Democrats, you would have pretty much an even race. You do have the most expert poll, the Business Investors --

STEVE DOOCY (CO-HOST): Investor's Business Daily.

GIULIANI: Most accurate for the last three elections showing him ahead by two. So I don't know. I think this is a closer election than they think, but let them feel confident. You know, fine.

AINSLEY EARHARDT (CO-HOST): Why do they oversampling? Why don't they count the votes as is?

GIULIANI: They make assumptions as to what they think the turnout is going to be, and that's the danger of a poll because if the turn out turns to be different than the sample, you just get a Brexit or you get one of these things like the Colombia -- I don't know if you are familiar with the Colombia peace vote, but that was going to win by 10 percent. It lost by 8 percent. [Fox News, Fox & Friends, 10/25/16]

Fox’s Lou Dobbs: Trump’s Claim That Poll Oversampling Is Rigging The Polls “Resonates … Because Everyone Knows These Polls Aren’t Making A Lot Of Sense.” Fox Business’ Lou Dobbs defended the Republican presidential candidate, saying Trump's claims “resonate” with the Republican base and with independent voters because “everyone knows these polls aren’t making … sense.” From the October 25 edition of Fox News’ Outnumbered:

SANDRA SMITH (CO-HOST): He really keeps hammering on this point, Lou, that the polls are wrong and that he's actually winning.

LOU DOBBS: It's not an accident, as you know, that he’s doing that because this resonates with his base, it resonates with independents, because everyone knows these polls aren’t making a lot of sense. You’ve got some outliers at 12 point differentials and others with him in the lead -- national tracking polls, specifically IBD and the LA Times-USC polls primarily.

SMITH: But his critics are quick to point out that when this race started, and especially in the primary season, he was quick to point out those polls when they showed that he was winning

DOBBS: Quick is interesting word to use, because remember it took a long time for him to plow through the national liberal media attacking him as not being serious, not being able to get even to the primaries, and the next thing you know it was February 1 and he had a little bit of a setback in Iowa, and the next thing you know, he starts -- the Trump train, as they call it, starts rolling. So it has been, I know what you mean by quick, but this thing was a process that he had to overcome national media bias every step of the way. [Fox News, Outnumbered, 10/25/16]

Fox News’ Chris Stirewalt Debunks Trump And Fox's Claim That Oversampling Is “Rigging” The Polls

Fox News’ Stirewalt: Oversampling Not “Nefarious” And “Anybody Who Says Otherwise Either Does Not Understand How Polling Works Or Is Lying.” According to Fox News Channel’s digital politics editor, Trump’s claim that the polls are “rigged” because emails revealed the Clinton campaign was “oversampling” in their internal polls is incorrect and no “different than previous years.” From a October 24 Fox News.com piece:

If a pollster expects that Hispanic voters make up 10 percent of a state’s population, which means that out of 1,000 voters surveyed, only 100 would be Hispanic. If a campaign wanted to know more about how Hispanic voters were responding to a specific message or what the deeper trend lines would look like within a subpopulation, a campaign would “oversample” that group and do perhaps 500 interviews with Hispanic voters.



But that doesn’t have any effect on the final outcome of the polls. There would still only be 100 Hispanic voters included in the 1,000-voter, horserace survey. The sample of 500 Hispanics would only be used for assessing that subset.



After all, campaigns want real numbers, not intentionally skewed ones. The idea that the Clinton campaign would pay for bad polling makes no sense.



What’s really disheartening here is that some people seem to believe that the private polls conducted for a campaign are the same as polls conducted for news organizations. It’s just not the case.



Both campaigns have aggressive polling operations, and both presumably do oversampling for key demographic groups to test messages and track emerging trends. None of this is nefarious. None of this is different than previous years. And it doesn’t have anything to do with how many people identify as Democrats and Republicans in national public surveys.



Anybody who says otherwise either does not understand how polling works or is lying.



Adding to today’s Wiki-pollster stew is an alleged plan of action for going into Trump-friendly online communities to dishearten his backers over the state of the election. Trump supporters claim this is evidence of polling rigging. But all that we find in the memo is just Democrats discussing ways to torture Republicans online, something that GOPers do to Democrats as well.



So, here’s your takeaway: oversampling is a normal activity for campaign pollsters, good public pollsters are not “oversampling” Democrats and if there is a Clinton effort to sow anguish among Trump supporters online, this discussion alone would seem to be evidence that it’s working. [Foxnews.com, 10/24/16]

Other Media Point Out Oversampling Is A Standard, Statistically Sound Polling Method

Wash. Post’s The Fix: Claiming Oversampling Is Voter Suppression Is “Laughably Incorrect.” The Washington Post’s Philip Bump called Trump’s claim about oversampling “laughably incorrect,” explaining that oversampling is a standard polling method to “get robust enough sample sizes” for different subgroups. Bump highlighted that without a “statistically relevant sample size,” pollsters risk “a huge margin of error.” From an October 24 Post article:

The problem is that it can be hard to find enough people to get robust enough sample sizes to offer the necessary information. Normal polling in a state will usually have no problem getting enough white people in the mix to evaluate where they stand, but you may need to specifically target more black or Hispanic voters to get a statistically relevant sample size.

[...]

Small samples of poll respondents mean a huge margin of error. Until you get to about 400 people in your sample, the margin of error drops quickly; once you pass 400, though, it doesn't change a whole lot. (This is why a lot of polls use sample sizes of 400 to 600.) If you're trying to figure out how to craft a message to Hispanic voters in Colorado, for example, you're going to need to seek out more Hispanic voters in the state to include in the survey. This is called an oversample, since it's an intentional effort to include more people from a certain group in your sampling. [Washington Post, 10/24/16]

The Atlantic: “Oversampling? Its Benign.” The Atlantic’s Andrew McGill called out Trump’s misunderstanding of polls and clarified that oversampling is a “solution” when pollsters are “particularly interested in a small subgroup so they’ll end up with large enough samples to draw real conclusions.” McGill also explained that pollsters “rebalance the sample to bring it back in line with the overall demography of the population” before reporting their findings, “negating the inflationary effect of the oversample.” From the October 24 Atlantic article:  

This afternoon, Trump threw his support behind the idea. “When the polls are even, when they leave them alone and do them properly, I’m leading,” he said at a rally in Florida. “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.”

Let me cut this off at the head: Neither the writer nor Trump understand how polls work.

Both are citing a 2008 email from the hacked account of Clinton aide John Podesta, posted yesterday on WikiLeaks. In the email, a prominent Democratic strategy firm recommended “oversampling” certain voters when running polls, including blacks, Hispanics and young people.

[...]

When pollsters field a survey, they randomly call (or contact online) a representative sample of the population they’re studying. If you’re looking at the entire United States, maybe that’s 1,000 people. The magic of statistics means the researcher can be fairly confident, within a certain margin of error, that the opinions of that sample will match up with the population as a whole. If the sample is large enough—and 1,000 people almost always is—that margin of error will be minimal.

But if the pollster is particularly interested in a smaller subgroup—say, suburban housewives—they might run into trouble. What if they only contacted 50 suburban housewives during their random phone calls? That’s a lot less than 1,000. The margin of error for their responses will be a great deal higher—making it harder to accurately predict what suburban housewives really think.

Oversampling is the solution. When pollsters launch a survey, they’ll often try to interview more people from underrepresented groups so they’ll end up with a large enough samples to draw real conclusions. Before they report the results, they’ll rebalance the sample to bring it back in line with the overall demography of the population—negating the inflationary effect of the oversample.

[...]

But oversampling? It’s benign. The fact that a post detailing a media oversampling conspiracy has gotten more than 1 million page views says something about Trump supporters’ fears of losing the election, but even more about America’s statistical illiteracy. [The Atlantic, 10/24/16]

Mother Jones: “Oversampling Is A Normal And Longtime Practice.” Mother Jones’ Kevin Drum highlighted the need for oversampling if a candidate is particularly interested in a specific demographic of voters. Drum emphasized that the oversampling “wouldn’t affect the overall poll.” From an October 24 Mother Jones article:

In case you care, oversampling is a normal and longtime practice for folks who are running presidential campaigns—which is what John Podesta was doing. If you survey, say, a thousand people, you're likely to get a sample of only 130 African-Americans. This means that if you happen to be particularly interested in African-American voters, you need to deliberately oversample them in order to get a statistically reliable pool of respondents. The same is true for any smallish group of people. If, for some reason, you want to target Hispanic environmentalists or white women under age 30, you have to oversample them too.

Ordinary polls don't normally do this, though they do sometimes. For example, suppose everyone is obsessed with blue-collar white men and their alleged anger at the political system. A polling firm might want to oversample them in order to report how they really feel. That wouldn't affect the overall poll, though. It would be released as a separate survey on a matter of current interest. [Mother Jones, 10/24/16]

Huff. Post: “Oversampling Doesn’t Change … The Poll Overall.” The Huffington Post’s Ariel Edwards-Levy pointed out that oversampling doesn’t overstate “the size of one group relative to others,” but makes sure the results are “as accurate as possible.” Edwards-Levy said the practice doesn’t change the poll overall and is only used to look at specific demographics. From an October 24 Huffington Post article:

Oversampling, though, isn’t about overstating the size of one group relative to others ― it’s about making sure the results for that group are as accurate as possible.

Often, pollsters are looking to get a better picture of a group that ordinarily includes only a small fraction of survey-takers. They’ll make an effort to find extra people in that group, raising the sample size and allowing them to better measure what that slice of the population thinks. Crucially, though, oversampling doesn’t change the weight that any subgroup is given in the poll overall.

[...]

Oversampling, while useful, is expensive, which would make it a particularly baffling and ineffective way of biasing the polls. [Huffington Post, 10/24/16]

Polling Organization: Oversampling Actually Increases Reliability Of Polls

Pew Research Center: “Oversampling Allows For Estimates To Be Made WIth A Smaller Margin Of Error.” According to Pew Research Center’s methodology for its polling, oversampling decreases the margin of error, making polls more accurate because “the oversampled groups are weighted to their actual proportion in the population.” From Pew:

For some surveys, it is important to ensure that there are enough members of a certain subgroup in the population so that more reliable estimates can be reported for that group. To do this, we oversample members of the subgroup by selecting more people from this group than would typically be done if everyone in the sample had an equal chance of being selected. Because the margin of sampling error is related to the size of the sample, increasing the sample size for a particular subgroup through the use of oversampling allows for estimates to be made with a smaller margin of error. A survey that includes an oversample weights the results so that members in the oversampled group are weighted to their actual proportion in the population; this allows for the overall survey results to represent both the national population and the oversampled subgroup.



For example, African Americans make up 13.6% of the total U.S. population, according to the U.S. Census. A survey with a sample size of 1,000 would only include approximately 136 African Americans. The margin of sampling error for African Americans then would be around 10.5 percentage points, resulting in estimates that could fall within a 21-point range, which is often too imprecise for many detailed analyses surveyors want to perform. In contrast, oversampling African Americans so that there are roughly 500 interviews completed with people in this group reduces the margin of sampling error to about 5.5 percentage points and improves the reliability of estimates that can be made. Unless a listed sample is available or people can be selected from prior surveys, oversampling a particular group usually involves incurring the additional costs associated with screening for eligible respondents.



An alternative to oversampling certain groups is to increase the overall sample size for the survey. This option is especially desirable if there are multiple groups of interest that would need to be oversampled. However, this approach often increases costs because the overall number of completed interviews needs to be increased substantially to improve the representation of the subgroup(s) of interest. [Pew Research Center, accessed 10/25/16]