Fake news and unsubstantiated rumors have been all over the internet for years. But until recently, a very basic but important question about fake news remained unanswered: Do people believe this stuff, or do they merely support it because it is critical of the political figures they don't like?

This question is at the heart of a new paper from Adam Berinsky, a political scientist at Massachusetts Institute of Technology. And the answer is troubling.

Political rumors and false "news" that circulates on social media can be so far-fetched or ridiculous that it is hard to fathom anyone believing it. But some people do. An especially egregious example is the fabricated "Pizzagate" story. It claimed that Hillary Clinton and other prominent Democrats were running a child sex slavery ring out of a popular pizza restaurant in D.C. Despite being totally ludicrous on its face and lacking any supporting evidence, hundreds of people exposed to the "story" online believed it and harassed the business. It culminated in December of 2016 with an anti-Pizzagate zealot firing three bullets into the restaurant during business hours.

Some people really do believe even the most implausible rumors and lies, it seems. What hasn't been clear is how common this is. Is the problem limited to a fringe of borderline unstable people, or is belief in wild political rumors common? Survey data reveals that many Americans support conspiracy theories. For example, almost half express support for 9/11 conspiracy theories. But surveys are tricky things; support is not the same as believe.

In his new paper, Berinsky tests two political rumors that have no basis in fact: that Barack Obama is secretly a Muslim, and that the George W. Bush administration knew about the 9/11 attacks but intentionally declined to stop them. Predictably, Republicans are more likely to say they believe the Obama-Muslim rumor and Democrats the Bush-9/11 rumor.

But what does it mean when they say they believe it? That is the heart of Berinsky's study. Survey participants may be expressing a true belief, or perhaps they agree with the statement simply because they dislike the political figure it is aimed at. In other words, he sought to find out whether Republicans say they think Barack Obama is a Muslim because they really believe it, or because they just don't like Barack Obama. This latter kind of response is called "expressive."

Religion is a useful metaphor for understanding true beliefs versus expressive thinking. Some religious people consider scriptures like the Bible literally true, word for word. Others believe in the Bible but consider it allegorical or figurative, not factual. In the Obama-Muslim scenario, this is like the difference between a person believing Barack Obama literally is a closet Muslim and a person who agrees because he finds Obama inauthentic, dishonest, or not a "real" American. Such a response doesn't hinge on the factual accuracy of the Obama-Muslim claim.

The bad news is that Berinsky's research offers persuasive evidence that expressive responses are rare. When people say they think Obama is a Muslim, they mean it. They believe that despite documented decades of Christian church attendance (including his often-criticized association with United Church of Christ Reverend Jeremiah Wright) Obama worships Allah and says his prayers to Mecca daily behind closed doors.

This presents a serious problem given the prominence of fake and misleading (intentionally or otherwise) information on the internet and social media, where a growing share of Americans get most of their news.

How do people so readily internalize rumors and falsehoods as facts?

Partisanship or ideology filter and shape political information as we receive it. They are not objective forces, and they promote two common logical fallacies in thinking. First, the classic "Argument from Ignorance" can be applied readily to these rumors; you cannot prove Obama is a Muslim, but I cannot prove he is not. Therefore, no one can answer this definitively so either belief is equally valid. That I can provide supporting evidence that he is a Christian is irrelevant, since I cannot prove conclusively that in secret or in his mind he is not a devotee of Islam.

Motivated reasoning contributes as well. People who do not like Obama (or Trump, or any other politico) are motivated to find information that justified and confirms their existing beliefs. In our eagerness to prove that we are correct, we lower the bar for supporting evidence. "Facts" that make little sense or collapse under scrutiny are often accepted as true when they tell us what we want to hear.

If Berinsky's new research is correct, internet rumors and fake news are a bigger problem in politics than we recognize. Even with a mountain of evidence available to disprove a rumor, some people not only express support for it in polls but also believe it to be literally true. As more Americans turn to social media, consciously or otherwise, to get their news, exposure to the least reputable and most outlandish "news" sources will continue to rise.

Fact-checking after these wild rumors circulate is only minimally effective. Social media companies have devised creative solutions to many problems over the years. If they cannot find ways to stem the flow of unverified, blatantly false information disguised as news, political discourse is likely to get worse before it improves. This isn't harmless fun and games. People believe this stuff.