Small talk is not the enemy. The recorder study that said so did not replicate

Happy people make less small talk, said a famous 2010 recorder study. Its own authors rechecked with 486 people: deep talk mattered, small talk did not.

In 2010 a short paper in Psychological Science gave the internet one of its favourite findings about happiness. Matthias Mehl of the University of Arizona and three colleagues had clipped a small audio recorder onto 79 undergraduates and let it run for four days. Every 12.5 minutes it switched itself on for thirty seconds. That produced 23,689 snippets of ordinary life: people alone, people eating, people talking about popcorn.

Coders then sorted the conversations. Small talk was the banal kind, where nothing much gets exchanged. Substantive talk was the kind where something does. The title said what they found: Well-being is related to having less small talk and more substantive conversations. Compared with the least happy students, the happiest ones had roughly a third as much small talk, 10.2 percent of their conversations against 28.3, and about twice as many substantive ones, 45.9 percent against 21.8.

The finding travelled faster than its caveat. Skip the weather, ask the real question, and you will be happier; that is how it has been retold in articles and talks ever since. The caveat was in the paper itself: the data were correlational, and nobody knew which way the arrow pointed.

Eight years later, the same team checked

The people who went back were largely the same people. In 2018 Anne Milek led a replication, and two of the original authors, Mehl and Simine Vazire, signed it. They pooled the original students with three very different groups who had worn the same kind of recorder for other studies: women being treated for breast cancer and their partners, healthy working adults in Atlanta enrolled in a meditation trial, and adults in Tucson who had recently separated from a spouse. 486 people in all.

Part of the 2010 result held. People who were more satisfied with their lives spent less time alone and more time talking, and they had more substantive conversations. The pooled correlation for substantive talk was 0.23, a moderate effect, close to the original.

The small-talk half did not hold. Across the pooled samples the correlation between small talk and life satisfaction was 0.05, with a credible interval that took in zero with room to spare, and the authors said so plainly: the negative correlation from the underpowered original study, they wrote, seems to have been a false positive.

So the widely shared version had it half right. More real conversation goes with a better life; small talk, on the best data available, was never the problem.

Then someone ran the experiment

A correlation from a recorder cannot tell you whether deep talk makes people happier or happy people simply talk more deeply. Michael Kardas, Amit Kumar and Nicholas Epley tried to get closer to cause. Their paper, Overly shallow?, came out online in 2021 and in print in the Journal of Personality and Social Psychology in 2022, a dozen experiments in all.

The most memorable of the twelve is the first. At a management conference, 25 pairs of financial-services executives who had never met were given questions to discuss. Not the weather. One of them was "Can you describe a time you cried in front of another person?" Before talking, each executive predicted how awkward it would be and how much the other person would care about their answers. Afterwards they said how it had actually gone.

They had overestimated the awkwardness by roughly three points on a ten-point scale, and underestimated both how connected they would feel and how interested the stranger would be in what they said. The later experiments ran the same test on master's students, on volunteers from university and community subject pools, and on 204 pairs of strangers recruited in public parks. They found the same pattern and one more thing: people guessed far better when the partner was a friend or family member. The error is about strangers: we expect them to care less than they do.

Two more results say something about how to use this. When participants had both a shallow and a deep conversation, with two different strangers, they felt more connected to the deep partner. And when they were simply told that people tend to underestimate how much others care, they chose deeper questions for their next conversation.

These were still arranged conversations between people who had agreed to take part, with the questions handed to them. Even a park bench with a researcher nearby is not the same as the bus stop. It is still the most direct evidence we have that the fear of going deeper is largely a forecasting mistake.

Back on the street

Put the three papers side by side and a gentler rule comes out than the one on social media. In the pooled recorder data from 2018, small talk showed no link to being less satisfied with life. It is often how two neighbours find out they are both willing to stop and talk. The mistake is to stay there because you are sure the other person does not want more.

The fix Kardas and his colleagues tested is almost embarrassingly small: know that you are probably wrong about the stranger. The same kind of miscalculation turns up when people predict how a message out of the blue will land, which we wrote about in Reaching out lands better than you predict.

Notice, too, who corrected the small-talk claim: the people who made it, with more data, in the same journal. We went looking for that kind of correction last time as well, with a far louder number, in Fifteen cigarettes a day: where the loneliness number comes from.

Sources

• M. R. Mehl, S. Vazire, S. E. Holleran, C. S. Clark, "Eavesdropping on Happiness: Well-being Is Related to Having Less Small Talk and More Substantive Conversations", Psychological Science 21(4), 539–541, 2010 — https://pmc.ncbi.nlm.nih.gov/articles/PMC2861779/
• A. Milek, E. A. Butler, A. M. Tackman, D. M. Kaplan, C. L. Raison, D. A. Sbarra, S. Vazire, M. R. Mehl, "'Eavesdropping on Happiness' Revisited: A Pooled, Multisample Replication of the Association Between Life Satisfaction and Observed Daily Conversation Quantity and Quality", Psychological Science 29(9), 1451–1462, 2018 — https://pmc.ncbi.nlm.nih.gov/articles/PMC6139582/
• M. Kardas, A. Kumar, N. Epley, "Overly Shallow?: Miscalibrated Expectations Create a Barrier to Deeper Conversation", Journal of Personality and Social Psychology 122(3), 367–398, 2022 — https://doi.org/10.1037/pspa0000281