What the data actually shows
The clearest documented change is affective polarization — feelings toward the other side rather than the substance of policy. Research associated with Shanto Iyengar and colleagues finds that in the US, warmth toward one's own political group has stayed fairly stable while warmth toward the opposing group has fallen substantially, so the gap between how people feel about "us" and "them" has widened markedly. The same pattern appears, to varying degrees, in a number of other democracies.
Alongside this is partisan sorting: political identity has become more aligned with other identities — social, geographic, cultural — so that disagreement feels more total and more personal than when party lines cut across those other groupings. Sorting makes the divide feel deeper because more of life now lines up with it, even where positions on specific issues have not moved as much.
The counterweight comes from research on the "perception gap," notably work by the organization More in Common. Across studies, people on each side substantially overestimate how extreme, how hostile, and how monolithic the other side is — and those most engaged with political news often have the most distorted picture, not the most accurate one. So measured animosity is up, but a large share of the felt division rests on inaccurate impressions of the other side rather than on real differences of that size.
Why this feels different from how it actually is
Division feels overwhelming partly because the information environment selects for the most extreme and combative voices. What reaches you is disproportionately the angriest content from the loudest minority on each side, which is unrepresentative of how most people actually think — so your sense of "the other side" is built from an outlier sample, and registers as far more hostile than the median person.
Sorting compounds the feeling. Because political identity now aligns with so many other identities, a disagreement that is really about one issue can feel like a clash of entire ways of life. The divide feels more total than the underlying positions warrant, because more of life now lines up on the same axis.
And there is a self-confirming loop: when each side believes the other is more extreme and hostile than it is, both sides respond more defensively, which produces exactly the antagonism each expected. The perception gap doesn't just misread the division — it can quietly widen the real one by changing how people behave toward an opponent they have overestimated.
The imagined opponent is more extreme, more hostile, and more numerous than the real one.
What the research says to do about it
The most direct, evidence-aligned step is to calibrate against actual data rather than impressions: studies on the perception gap suggest that simply learning what the other side really believes — as opposed to the caricature — tends to shrink the perceived divide, because the imagined opponent was more extreme than the real one. Checking the caricature against survey reality is itself a corrective.
Because the loudest, most combative voices are unrepresentative, weighting them less is reasonable. Research on affective polarization suggests the most engaged news consumers often hold the most distorted picture of the other side, so treating the angriest online material as a sample of "what they think" reliably misleads. The median person on either side is less extreme than the feed implies.
And the broader social-trust and contact research points gently toward ordinary, non-political shared activity — being around people across lines in settings that aren't about politics — as associated with lower out-group animosity. The effects here are modest and the evidence is mixed in places, so this is offered as a direction the research leans, not a guaranteed remedy.
What the research says does not help
Inferring "how divided we are" from social media, comment sections, or the most heated coverage reliably misleads, because those environments amplify the most extreme voices on every side. The aggregate of ordinary people is consistently less extreme than the signal those channels produce — so reading them as a representative sample inflates the perceived divide.
Assuming the other side is uniformly extreme and hostile is precisely the perception-gap error the research documents, and it tends to be self-confirming: expecting hostility produces defensive behavior that generates the very antagonism that was assumed. Treating the caricature as the reality widens the real gap.
Concluding that disagreement itself proves we are hopelessly broken overstates the case. Robust disagreement has always existed in pluralistic societies; what has measurably changed is the animosity and sorting around it, plus a perception gap that exaggerates the rest. Mistaking normal disagreement for unprecedented division is itself part of the distortion.
Animosity has grown, while perceived division has grown even faster — leaving the gap between feeling and reality unusually wide.
What this looks like in real life
Judging 'the other side' from the loudest voices
Building an impression of the opposing group from comment sections and the most heated coverage means sampling the angriest minority, who are unrepresentative of how most people think. The median person on either side is less extreme than the feed implies — which is exactly the perception-gap error the research documents.
One disagreement that feels like a clash of worlds
Because political identity now aligns with so many other identities — social, geographic, cultural — a disagreement really about a single issue can feel like a clash of entire ways of life. Sorting makes the divide feel more total than the underlying positions warrant, even where views on specifics have moved less.
Real numbers in context
On the measured side, affective polarization research (Iyengar and colleagues) finds that in the US, warmth toward one's own party has stayed roughly stable while warmth toward the opposing party has fallen substantially over recent decades, widening the us-versus-them gap — a real, documented change, mirrored to varying degrees in other democracies. Partisan sorting has reinforced it by aligning political identity with other identities.
On the perception side, work by More in Common finds a sizable "perception gap": people on each side overestimate how extreme and hostile the other side is, often by a wide margin, with the most politically engaged frequently holding the least accurate picture. The honest synthesis is that animosity is genuinely up, but perceived division outruns actual division — the feeling has risen faster than the underlying difference. These figures are directional, drawn from specific surveys, and should be read as patterns rather than precise constants. This summary is descriptive and non-partisan by design.