How to talk to someone with different political views

How do you talk to someone with different political views?

Start by asking and listening, and save your case for later. That order has more support than any clever phrase. In three field experiments, Kalla and Broockman (2020) had 230 canvassers talk with 6,869 voters across 7 U.S. locations about unauthorized immigrants and transgender people. In the first experiment, conversations built on arguments alone had no detectable effect. Otherwise identical conversations that added what the authors call “the non-judgmental exchange of narratives” shifted attitudes for at least four months.

Two caveats. The effect was small (d = 0.08). And those canvassers were trained, and were trying to move people in one direction on those issues. You are not a canvasser, and this post has no opinion on where your relative should land. What may carry over is the method: people seem to hear you better once you have asked about their life and left their answer ungraded.

Here is a version you could try with someone you know:

1. Ask first, and let them say no.
2. Ask what in their life made this issue matter to them.
3. Say back what you heard, and let them correct you.
4. Offer one experience of your own, then stop talking.
5. Set a small goal: leave knowing one thing you did not know. Then end it while it is still going fine.

Steps 2 to 4 borrow from the research in this post. Steps 1 and 5 are common sense. We found no study testing them.

Will the conversation go as badly as you think?

Possibly not, though the evidence comes from people hearing opposing views, not from talking to your brother. In a 2023 Pew survey, 61% of U.S. adults called political conversations with people they disagree with “stressful and frustrating,” and 36% called them “interesting and informative.” Pew reports no partisan difference. So the dread is shared.

The dread may also run ahead of the facts. Dorison, Minson and Rogers (2019) ran five studies with 2,455 people in total and found that partisans overestimated how negative they would feel after exposure to opposing views. Democrats and Republicans both did this with speeches by senators of the other party. The authors trace the error partly to underestimating how much people agree with their opponents. These were speeches and written statements, so treat this as a hint about your own forecasting and not a promise about Thanksgiving.

Avoidance is real, and it is not a one-sided habit. In Frimer, Skitka and Motyl (2017), 63% of 202 Americans gave up a chance at $3 more to avoid reading the other side’s view on same-sex marriage: 64% of supporters and 61% of opponents. A second sample of 245 came out at 62%. For why people misjudge each other in the first place, see what the perception gap is.

What should you say to someone with different political views?

Say less than you planned, and ask more. The wording below is hypothetical, written for this post. It comes from no study or real conversation, so change it until it sounds like you.

Moment	You might say	What backs it
Opening	“I’d like to understand how you see this. No debate. Is now okay?”	Common sense
Asking	“What happened in your life that made this matter to you?”	Kalla and Broockman (2020), with canvassers
Checking	“So what worries you most is the cost to your family. Did I get that right?”	Small lab evidence, covered in how to disagree respectfully
Sharing	“Here’s where mine comes from. Can I tell you what happened?”	Kubin et al. (2021)
Disagreeing	“I see that part differently, and I may be missing something.”	Yeomans et al. (2020), in writing
Leaving	“I need a break from this. Can we pick it up another time?”	Common sense

The sharing row rests on Kubin and colleagues (2021), who report 15 studies finding that political opponents respect moral beliefs more when they are backed by personal experience than by facts. Experiences involving harm did best. You might say, if you own a gun: “My dad taught me to shoot, and our house got broken into when I was twelve.” If you want stricter laws, you might say: “A kid in my class was shot, and I still think about his mother.” Both are hypothetical, and both are harder to wave away than a statistic.

Limits matter here. The authors say the studies often used self-report data from online samples, and that the advantage appears only in moral disagreement. On other disputes, facts earned respect just as well. The journal also ran a commentary titled “Speaking my truth: Why personal experiences can bridge divides but mislead.” Respect is the outcome being measured. A moving story can earn it and still be a poor guide to what is true, so keep the facts for later.

How do you talk to family with different political views?

Pick a moment with one person and no audience, such as a drive or a walk the day before. That is common sense. We found no experiment comparing settings, and the family data we have only measures time.

That data is worth knowing, because it is less alarming than the headlines it produced. Chen and Rohla (2018) used phone location data from more than 10 million users and found that visits across politically different precincts in 2016 ran 30 to 50 minutes shorter. Frimer and Skitka (2020) then asked people directly who was at the table. Pooling their two preregistered studies (4,013 people) and controlling for other factors, they found politically diverse dinners were 24 minutes shorter (95% interval 9 to 39), about 6% of the meal. Without those controls the difference was not statistically significant. They read this as “favoring the conclusion that politics is not straining family ties as much as previously thought.” Neither study recorded what anyone said.

For a holiday table, a few things seem sensible, none of them tested:

• Decide before you arrive whether this visit is for the conversation at all. Usually it is not.
• If you are hosting, you might say at the start: “Politics gets the day off. The dishes do not.”
• If someone wants to talk, move it: “I’d like to hear this properly. Walk with me after dinner?”

How do you avoid talking politics with family?

Say it early and briefly, then change the subject and give no speech about why. You have company in this. In a Quinnipiac poll of 1,574 registered voters in November 2023, 61% hoped to avoid politics with family or friends at Thanksgiving and 29% looked forward to it. Answers were similar across groups, so the person baiting you may be outnumbered.

You might say: “I’m keeping today politics-free. Tell me about the new job.” If they push: “I hear you. Not today.” If they push again, get up and do something useful. A sink full of pans has ended more arguments than logic, though we have no data on that. All of this is common sense.

When should you not have the conversation?

Skip it when the goal is to win, or when the setting makes listening impossible. Some clear cases, all judgment calls and none from a study:

• Someone has been drinking, including you.
• There are other people watching, because few people change position in front of a crowd.
• They are grieving or in a crisis, or they have told you they would rather not.
• You have ten minutes and a car to catch.

Staying on other subjects also has real support behind it. Santoro and Broockman (2022) paired strangers from opposing parties on video calls. Talking about their perfect day raised warmth toward the other party by 9 points on a 100-point scale (d = 0.34), an effect that was gone three months later. Talking about why each supported their own party had no effect. Other work disagrees on that second point, which the post on disagreeing covers. So a good talk about fishing with your uncle may do more for the relationship than a tense one about tax policy.

One line that matters more than the rest. If a disagreement comes with threats, control or fear, this advice does not apply. That is a different problem, and better conversation is not the fix. In the U.S., the National Domestic Violence Hotline is at 1.800.799.SAFE (7233). If the person is a partner, the post on relationships goes further.

How do you talk to friends with different political views?

The same way, with more care about the channel, since a text thread has no face or tone to soften it. That part is common sense too. Friendships may also be the easier ones to lose. In a 2025 YouGov survey by Güngör and Ditto, 37% of Americans reported a political breakup at some point in their lives. Of those, 62% had fallen out with a friend and 40% with a family member. So if a friend matters to you, you might say: “I don’t want to do this by text. Coffee on Saturday?”

Can you be friends with someone with different political views?

Yes, and many people are, though friendships across the line are usually fewer and thinner. In a 2016 Pew report, 84% of Republicans and 74% of Democrats said they had at least a few close friends in the other party, while only 7% and 6% said they had “a lot.” That data is ten years old and we would not assume it still holds. In a 2021 Survey Center on American Life poll of 2,019 adults, 15% said they had ended a friendship over politics. These surveys ask different questions, so read them side by side and do not add them up. Most people in the 2021 poll had not ended one.

What does the research not know?

It knows very little about your family. Nearly everything above involves strangers, canvassers or lab tasks. Santoro and Broockman said so directly:

However, our findings may not generalize to naturally occurring conversations between friends, co-workers, or relatives, and future research should explore this.

Erik Santoro and David Broockman, Science Advances, 2022

A relative has decades of history with you, which could make listening easier or much harder. We found no strong experiment that tells you which. If you want one modest test, try step 2 once, on one person, and see what you learn. For what comes after a first talk, the skill of disagreeing is the next step.

I’m The People is not open yet. If it opens as planned, it would offer a version of this between two people who have never met and have both chosen to be there. How it would work is written up separately.