A few days ago, we waded into the debate over a conspiracy theory circulating among some supporters of Democratic presidential candidate Bernie Sanders -- the one that has CNN, among other media outlets, keeping favorable post-debate poll results for Sanders under wraps because the network, and other parts of the media establishment, are in the tank for Hillary Clinton.
After we gave a Pants on Fire rating to the claim that CNN deleted from its website an online poll favorable to Sanders, many readers attacked PolitiFact as being part of this conspiracy against Sanders.
Nonetheless, we’ve decided to wade in again after a reader sent us a Facebook post that added a new wrinkle to the alleged CNN conspiracy.
The reader sent us a post that showed an image of results from a CNN-ORC International poll taken between Oct. 14 and 17. On the image, two columns were circled in red -- two columns of sub-results by age group for a question about how favorably or unfavorably the respondent views Hillary Clinton.
The two columns show responses for survey participants ages 18 to 34 and ages 35 to 49. Every possible answer in those columns has been marked N/A -- "not applicable."
The creator of the image was dumbfounded, slapping this comment on the image in big red type: "CNN's Crap Polling: No one under the age of 50 has an opinion of Hillary Clinton?"
The reader who contacted PolitiFact wondered whether the image creator’s question was legitimate, so we took a closer look.
Bottom line: It’s all a misunderstanding about how CNN reports the results of its surveys. (After we emailed questions to the person who had posted it on their Facebook page, we didn’t receive a response, but the post was pulled down. We saved a screenshot, above.)
The easiest way to debunk the graphic’s assumption is to look a few columns to the right. There, one can find a heading that reads "under 50," with 74 percent of respondents in that age group saying they feel favorably about Clinton, 23 percent saying they feel unfavorably, 3 percent saying they have no opinion and 1 percent saying they’ve never heard of her. (Our real question: Who belongs to the 1 percent who have never heard of Hillary Clinton?)
So this column of poll results clearly shows that people under 50 did tell the survey takers about their opinions about Clinton.
What about those mysterious columns full of "N/A"?
They are there because of CNN’s polling policy. The policy is explained fully in the same document that the Facebook image was drawn from, though it appears on an earlier page not shown in the Facebook post. This section explains the CNN policy on "crosstabs" -- the polling industry’s term for results sorted by demographic subcategories such as age, race, gender or income level.
"Crosstabs on the following pages only include results for subgroups with enough unweighted cases to produce a sampling error of +/- 8.5 percentage points or less," CNN wrote. "Some subgroups represent too small a share of the national population to produce crosstabs with an acceptable sampling error. Interviews were conducted among these subgroups, but results for groups with a sampling error larger than +/-8.5 percentage points are not displayed and instead are denoted with ‘N/A.’ "
In other words, for the age ranges 18 to 34 and 35 to 49, the survey reached too few people to produce a statistically valid sample -- so to avoid publishing questionable numbers, CNN simply reported them as "N/A."
This pattern -- categories with larger sample sizes given specific figures, and those with smaller sample size marked "N/A" -- is echoed elsewhere in the full document.
"The CNN/ORC polls are highly regarded," said Karlyn Bowman, a polling analyst at the American Enterprise Institute. "They are one of the few partnerships to provide us subgroup data, which is invaluable. The Facebook question does a disservice to the poll."
The Internet graphic accused CNN of being such an incompetent pollster that it found that "no one under the age of 50 has an opinion of Hillary Clinton."
This is based on a misreading of the CNN-sponsored poll. Two age-range subcategories produced too few respondents to qualify as statistically valid results under CNN’s threshold, which was plainly spelled out in the survey-results document. The claim is even debunked on the graphic itself -- it showed a category with actual results for respondents under the age of 50. We rate the claim Pants on Fire.