Small talk has been the villain of the happy life for sixteen years. The charge comes from a study published in 2010, in which seventy-nine students wore a small recorder for four days: the happier ones, it found, had more deep conversations and less small talk. In 2018 two of the study’s authors went back with six new colleagues and roughly six times as many people. Three of the four associations the 2010 study had reported were still there; the small-talk one was gone, and the correction had come from inside.
Ambient sound sorted by hand
The 2010 study, published in Psychological Science by Matthias Mehl, Simine Vazire, Shannon Holleran and C. Shelby Clark, used a device called the Electronically Activated Recorder. Worn through the day, it recorded thirty seconds of ambient sound every twelve and a half minutes, without the wearer knowing exactly when it was listening. Seventy-nine undergraduates wore it for four days, and it collected 23,689 recordings from their waking hours, about three hundred each.
A single coder listened to every snippet and sorted it. Was the person alone, or talking with someone? And if there was a conversation, was it small talk, defined as an uninvolved exchange of trivial information, or a substantive conversation, in which something meaningful was exchanged? The paper gives an example of each. Small talk: “What do you have there? Popcorn? Yummy!” Substantive: “She fell in love with your dad? So, did they get divorced soon after?” Across the whole group, about eighteen percent of conversations were small talk and about thirty-six percent were substantive; the rest fell into neither category.
Well-being was measured separately, and with some care. Twice, three weeks apart, each student rated their own satisfaction with life and the single statement “I see myself as someone who is happy, satisfied with life.” Two or three people who knew them rated the same statement about them. The researchers combined all of it into one index, so that the people being counted as happy were not simply the people saying so.
The students who scored higher on that index had spent less of their four days alone and more of them talking with other people. They also had a smaller share of small talk in their conversations and a larger share of substantive ones. To make the size of the pattern visible, the authors compared the happiest and unhappiest ends of their sample: the happiest students spent about a quarter less of their day alone, talked about seventy percent more, had roughly a third as much small talk, and had twice as many substantive conversations. Controlling for personality left the substantive-conversation link intact and shrank the small-talk link.
The paper summarised all this in one sentence, that the happy life is “social rather than solitary, and conversationally deep rather than superficial,” and closed on the thought that people may find their lives more worth living when examined together. In the same three pages it also said that the findings were correlational and causally ambiguous, and that whether deep conversation makes people happier was something future research should test experimentally.
Advice without its qualifiers
The 2018 paper notes that the small-talk result had “garnered considerable scientific and lay interest,” which is a dry way of saying it became advice. Happy people avoid small talk; if you want to be happier, talk about something real.
A study of seventy-nine people is, by the standards of the field, too small to expect its result to repeat, however many recordings the coder had to sit through, and all seventy-nine were undergraduates, average age under nineteen, living in an environment built for meeting new people. The study and analysis plan were not preregistered, and with a sample that small the original study might, in the 2018 paper’s phrase, have been capitalising on chance.
Whether satisfied people draw deeper conversations or deeper conversations leave people more satisfied is, in the 2018 paper’s words, “still to be clarified” by experiments it says have yet to be done. Both studies watched and neither intervened in anyone’s conversations, and the advice to cut small talk grew out of a result that a larger sample could not find.
Nothing in either paper can tell a particular person whether a Tuesday’s chat with the barista counted, and the people writing this are readers of research rather than anyone trained to treat a person. Someone who has spent most of a month alone, or whose satisfaction with life has sunk and stayed there, has a reason to see a counsellor.
The second look
The replication pooled the original data with three other groups who had worn the same recorder: a study of fifty women being treated for breast cancer and fifty-one of their partners, 184 healthy adults in Atlanta recorded at the start of a meditation trial, before any intervention, and 122 adults in the Tucson area who had recently separated from a spouse. With the students, that came to 486 people, with average ages ranging from under nineteen to nearly sixty. The outcome this time was satisfaction with life alone, the considered judgement rather than the day’s mood. Anne Milek led the paper, with Mehl and Vazire among its eight authors, and the replication ran in the same journal, with the data and code posted openly.
Most of the original findings held. In the pooled estimate, more satisfied people spent less time alone and more time talking with others, though the alone-time association came out about half the size of the original, and they had more substantive conversations, at a size the paper calls similar to the one first reported, a link only slightly diminished by controlling for personality. Not every sample showed it on its own: the quantity and substantive links were reliably detected in two of the three new groups, and the pooled estimate is what the paper stands on. Those pooled estimates leave out one of the five samples, the fifty-one partners of the cancer patients, whose answers were not independent of their spouses’; the authors ran the same models with that sample swapped in instead and report that the results did not differ substantially, apart from two of the personality-moderation models. There was no convincing sign that being introverted or extraverted changed any of these links, which the authors say somewhat surprised them, though they add that the study may not have had the power to detect such an effect.
Small talk was the exception. In the pooled data the association between how much small talk people had and how satisfied they were was close to zero, and the credible interval around it included zero whichever way small talk was counted, as a share of all recordings or as a share of a person’s conversations. The paper’s introduction says the small-talk link had been expected to replicate; its verdict is that the negative correlation from the original study “seems to have been a false-positive finding.”
Satisfaction with life was the only outcome measured, not how anyone felt, the emotional side of well-being, where the authors speculate that small talk might behave differently for introverts and extraverts. The researchers did not code who the conversations were with, so a snippet of chat with a barista and a snippet of chat with a spouse fell into the same category; other work, cited there but not read by us, reports that more daily interactions with weak ties, a barista or a colleague or a classmate, predicted greater average well-being. The replication’s own design and analysis plans were not preregistered either. In the recently separated sample the two coders agreed poorly on what counted as small talk. The substantive-conversation link, though it held, is moderate, which leaves most of the difference between one person’s satisfaction and another’s unaccounted for, and even four deliberately varied samples, the paper concedes, may not stand for the general population.
A pool six times the size of the original study kept three of its four associations. Pooled across samples of very different ages and circumstances, the people more satisfied with their lives talked more, and more of what they talked about was something. The chat about the popcorn, on the evidence so far, was just a chat.
