Media manipulating same data to drive different narratives

You can find the original version of this article in The Hill

We pollsters frequently make claims about the beliefs of “most Americans.” We make these generalizations to help make complicated patterns in data easier to understand.

However, in the reporting of our work, we often see these broad simplifications stretched to almost caricature-levels, leading to extreme declarations like: “Americans are racist,” or “Refugees are terrorists.”

Like all “truths” in this age, public opinion has become a weapon for partisans on either side. But like all political combat, this tends to obscure more than it reveals. The reality is that Americans are not just simple statistics that can be easily inserted into a bar graph or Venn diagram.

For example, on Feb. 1, Ipsos released polling numbers regarding American attitudes toward President Trump’s executive order halting immigration from certain nations. For our part, the numbers were surprising from a scientific standpoint; but nowhere near as surprising as how they were used to advance various narratives in the press.

Breitbart’s headline screamed: “Poll: Public Overwhelmingly Supports Trump Push to Limit Migration.” Alternatively, TIME’s headline professed: “Most Americans Don’t Think President Trump’s Immigration Ban Will Make Them Safer.”

Is there any way you could have known both articles are about the same poll from just reading the headlines? Should we conclude that Americans like the immigration ban because it won’t make them safer?

Poetic license notwithstanding, it is clear that the same data set is being used to advance a message, rather than to inform. Which leads to the key issue: Is American opinion guided by a shared set of facts? Or is the news written to engage a pre-existing point of view of the audience?

According to a recent Ipsos survey on behalf of NPR, party lines clearly determine what news people are watching. A majority from both parties get most of their news from television, followed by the internet. But the similarities end there — 39 percent of Republicans getting their news from FOX News, while Democrats spread out their consumption among CNN (15 percent), ABC (14 percent) and CBS (15 percent).

A simple Google search on the sites of and is illustrative of these separate bubbles driving public opinion in different directions. On the weekend of Jan. 27-29, Fox News returns 923 hits for the word “terrorist” and 642 for “refugee”.

At the same time, on CNN, a search for “terrorist” returns 453 hits and “refugee” returns 692.That means that while Fox is saying “terrorist’” four times for every three times it says “refugee”. CNN is doing the reverse, mentioning “refugee” three times for every two mentions of “terrorist”.

Despite the creation of separate “facts” through slanted reporting, the reality is that Americans hold complicated, sometimes contradictory opinions. Americans can both support a refugee ban and believe in civil liberties. Conversely, they can oppose the ban and support measures to prevent terrorism.

Public opinion is complex and must be examined as such. Clinging to binary and polarizing assumptions based on polls does a disservice to the American public. Opinions are driven by our peers, the press we choose to consume, our own experiences, and many other factors. Coverage of hot-button issues should immerse itself in this complexity, not run away from it.