It’s easy to imagine that our fellow citizens see the same political news that we do, and yet many of them draw the opposite conclusions about the candidates. But this impression is only partly true. To a significant extent, prospective voters are seeing and hearing different things, depending on their parties and demographic groups. Specifically, the most inflammatory comments often reverberate most widely among a candidate’s opponents and hardly reach his supporters at all.

This point is well known, but it would be informative to quantify it for the 2024 campaign. CNN and several partners have been asking an online panel an open-ended question about what news they have heard lately. The results are published after a significant delay (presumably due to the work involved in the analysis), and the only reports that I have found are rather cursory so far. They leave me with a methodological concern: individuals’ reports of what they hear may not match what they were actually exposed to, because their attentions and memories may be selective. Still, these simple reports offer insights.

The graphic with this post shows the main topics that a sample of Americans say they heard regarding Donald Trump during several days in September. At that time, some of us were hearing his lies about Haitian immigrants in Springfield, OH–which echoed the inflammatory and hateful slanders that have preceded massacres of vulnerable minority groups here and around the world–and we wondered how anyone could consider voting for this man. But the people who heard about Trump and pets and immigrants had something in common: they were generally Democrats.

Republican and Independent voters remembered hearing one main news item about Donald Trump in late September: he had survived a second assassination attempt, this one on his own golf course. For them, Trump was a victim of crime, and the main events of this campaign were attacks on him. In general, many Americans get a diet of news about crime and unrest. According to Pew, crime is usually the second-most common news topic, after weather, and 77 percent of people see crime news. That is the context in which Republican and Independent voters processed the news that Trump was a crime-victim.

A more recent article about news consumption from Oct 11-14 doesn’t divide the data according to the respondents’ choice of candidates, which makes it less relevant for my purposes. But it is interesting that the word “assassination” continued to be prominent in news about Trump in mid-October.

It’s worth asking whether “the media” is responsible for our balkanized news environment. There are many competing news sources, and people can choose among them, so it’s possible that balkanization is inevitable.

It’s also worth asking whether individuals are responsible for choosing to follow–and remember–high-quality news, and if so, what that is. (I am far from perfect in that respect, spending too much time on polls and horserace news and not enough on troubling issues.)

In any case, it is an analytic mistake to assume that many people support the most awful things that one observes. To understand is not to forgive, but I can at least understand why people would feel differently about Trump if they didn’t hear what I hear about him.