Are you ready to see how easy it is to publish a lie and make it sound like it is actual fact?
We present today’s latest example—a New York Post article saying Republican presidential candidate Donald Trump is “winning over Latino Republicans” in a new national poll.
The headline, “Trump winning over Latino Republicans, poll says,” was tweeted out earlier this afternoon by the profile of Corey Lewandowski, Trump’s campaign manager. As you might expect, it got some attention, because as you know by now, Trump has been having this problem with Latinos ever since he launched his presidential bid.
Too good to be true?
First, let’s examine how the Post’s Marisa Schultz begins her story:
WASHINGTON — Despite his controversial comments about Mexican immigrants, GOP frontrunner Donald Trump is the favorite among Latino Republicans, according to new polling results revealed to The Post.
Thirty eight percent favor Trump, followed by Cuban American Ted Cruz (15 percent), Jeb Bush (14 percent) and Cuban American Marco Rubio (8 percent), according to the national poll conducted by the Beck Research for the American Federation for Children.
“If you’re trying to stop Trump – this poll should trouble you,” said pollster Deborah Beck.
Later in the article, Schultz writes this:
The poll was conducted Jan. 19-24 and surveyed 1,100 likely voters. The survey of general election voters was conducted to survey school choice, and found 70 percent of Americans favor it, with even stronger support for education options among Hispanics and young people.
So, if you are reading the Shultz story, you would think that the 1,100 likely voters were all Latino, right?
In fact, if you go to the actual poll memo, you will come across this nugget:
The survey fielded January 19-24, 2016 among a total of 1,100 likely November 2016 voters. The national survey among 800 voters was supplemented with 200 additional Latino voters and 100 additional Millennials. Interviews were conducted via live callers on landline and cell phones. The base sample margin of error is +/- 3.5%.
(Brief aside: a poll about school choice is also discussing presidential candidates?)
You don’t believe us?
Here’s the screen grab (red box annotation is ours):
In other words, the poll interviewed 200 Latinos or were there more outside of the 200 Latinos the poll supplemented? (We reached out to Beck for clarification.)
200 Latino voters is a Facebook poll. You can’t conclude anything about anything with a sample that small.
But let’s continue to examine this lie.
If all those 200 Latinos in the poll supplement were Republicans (a BIG IF, by the way because the the NYPost story included poll information on Hillary Clinton and Bernie Sanders), then Trump’s 38% would be (wait for it) 76 Latino voters.
And then there is this—even though the poll memo does not identify which of the 200 Latinos were Republican, if we use 2014 data from Pew Research about Latinos and party affiliation, we can make an educated guess as to how many Latino Republicans might have been part of this latest Beck poll.
Here is what Pew reported in 2014:
…Republicans have made some progress among Hispanic voters. About one-quarter (27%) today say they identify with or lean toward the Republican Party. In 2012, 22% said the same.
But Republican Party gains among Hispanic registered voters in terms of party identification do not extend to Hispanic voters’ views of the party. Just 10% say the Republican Party has more concern than the Democratic Party for the Hispanic community, unchanged since 2012. Instead, the share of Hispanic registered voters who say there is no difference between the two parties is up, to 35% today from 23% in 2012.
Let’s use 27% as a number for the Beck poll. That would mean that 54 Latino Republicans were part of the Beck poll.
54 Latino Republicans.
And if we continue with that math, 38% of 54 is 20.52.
We can round that up to 21 and suggest that Trump’s 38% support by Latino Republicans was based on 21 Latino Republicans.
21 Latino Republicans.
But according to the NY Post and Trump’s campaign manager, Trump is “winning over Latino Republicans” because 21 Latino Republicans told you so?
Let’s tweet it out and make people believe it!
And that is how a media lie is created.
One more point: despite what 21 Latino Republicans tell you, Trump’s popularity with Latino voters is awful.
Hit it, Gollum:
Ok, one last point, we promise: let’s say that there were more Latino voters in the initial number of 800 people. And for argument’s sake, let’s say that within those 800 people, 17.4% of those called were Latino (that is the US Census average). With all the math based on the 27% of Latino Republicans, that would bump Trump from 21 Latino Republican voters to 35 Latino Republicans who picked him. That’s assuming that the initial number of 800 people had Latinos in that initial sample, based on national averages.
In short, the #NoMames label still sticks.
PS: The Beck people should share how it conducted this poll, because it is clear to us that the methodology presented and reported is not even in the public memo it released. We did reach out to them.