"Now, I want to have a chance to tell the story about my friend Ady Barkan," Sen. Elizabeth Warren (D-Mass.) began her answer to a question about health care at the latest Democratic debate. "Ady is 35 years old. He has a wife, Rachael, he has a cute little boy named Carl. He also has ALS, and it's killing him. Ady has health insurance, good health insurance, and it's not nearly enough."

Warren used Ady's desperate situation to memorably encapsulate her support for Medicare-for-all, and for those who agree with her already or are persuadable in that direction, it was undoubtedly a compelling strategy. Ady can be understood as emblematic of a larger problem, his suffering a microcosm of the deep dysfunction of American health care.

But for viewers with substantially different prescriptions for our health-care woes, Warren's story won't be so obviously archetypal. No one need doubt the facts to reject the meaning Warren assigns them. As The New York Times' Ross Douthat observed in his last weekend column of July, "the heart of polarization [in America today] is often not a disagreement about the facts of a particular narrative, but about whether that story is somehow representative — or whether it's just one tale among many in our teeming society, and doesn't stand for anything larger than itself." Much of our political debate runs on story, and we have a dangerous habit of inflating anecdote into argument.

The impulse to use a story like Warren did is understandable. A personal account helps readers or, for Warren, voters who may be unfamiliar with a complex topic engage at a human scale. It forges an initial connection, often through empathy, that triggers a more sustained interest in the wonkier details to follow.

The problem is that the stories we use this way aren't always representative, and there's a big difference between linking an anecdote to an issue using actual data and making the link simply because the story confirms our prior assumptions. Take Ady's story: About 90 percent of Americans have health insurance in 2019, but a rising number, around 44 million, are underinsured, meaning their access to care is limited in practice by a high cost to income ratio. Of course, most of those people do not, like Ady, have ALS or the $9,000 in uncovered monthly medical bills Warren said comes with it. So is Ady's story representative enough to inspire major policy changes? Maybe, but that's not universally obvious. It's a case which needs to be made.

To be fair to Warren, the inane format of our presidential debates, where candidates are expected to cogently treat enormous topics in mere seconds, doesn't permit discussion of much data, even if she'd wanted it. But most opinionating is not thus limited, and anecdotes are too often assumed to be representational regardless of real evidence. Consider, for example, a recent totally bonkers Twitter thread from communications strategist and professor Carol Blymire. The gist of it is that Blymire happened to overhear a conversation between a young woman and her boss. The boss had edited something the young woman wrote, and among her edits was a note that "hamster" is not spelled "hampster." "But you don't know that!" came the response. "I learned to spell it with a P in it so that's how I spell it." After the conversation ended, the boss left the area and the young woman, in tears, talked to her mother on speakerphone, concluding — with the mom's total sympathy and support — "I mean, I always spell hamster with a P, she has no right to criticize me."

The account went viral, and though Blymire was extremely careful to allow that there could be some unknown mental health factor in play, both she and many of her respondents entertained it as a representative case. Blymire included in her thread a story about another young worker who responded to edits by quitting his job and having "his parents call to tell [Blymire's colleague] what a terrible boss he was for 'correcting work that didn't need corrected.'"

Others were more sweeping in their assessments. One announced the young woman's behavior is the due result "of a culture that insists on 'no such thing as objective truth.'" Another declared it representative of an "emerging trend in post-collegiate employment." Reply after reply used this one woman's actions to condemn millennials en masse: "We have a whole generation like that."

Of course, this story demonstrates nothing of the sort. It might be representative, but nothing in the thread proves that, and neither do the isolated experiences (or, more often, rumors and urban legends) shared in the replies. What data we do have on millennials in the workplace, on the contrary, shows the generation is not markedly different on the job than generations prior.

The temptation to assume without evidence that an anecdote is representational is particularly strong when the topic is polarizing, which is to say these unfounded interpretations both feed into and feed off our division. And — though I'll be careful to note this observation is my own — it seems the speed and spread of modern communication makes such detrimental assumptions easier than ever.

Personal accounts and small news stories that once would have stayed unknown beyond the local level now, via social media, travel up the press food chain to carry the weight of national news in a matter of hours. That was not possible even a couple decades ago, when the time it would take for such anecdotes to reach a wide audience would also have rendered them too old to make headlines. The delay could function as a sieve.

Today we are served an endless supply of anecdotes hot, fresh, and ready to be held up as representative cases. That practice won't stop — nor, necessarily, should it. Personal connection is rhetorically useful and can foster thoughtful and needed debate. But the stretch of anecdotes as something with meaning beyond a specific situation must always be met with a certain skepticism. Whenever a story is deemed representative, even (or especially) if that take feels right, we should ever be ready to demand a broader proof.