What the Grapevine Gets Right

Part 5 of 8 in a series on lies and deception at work.

Office gossip is usually right, and it’s how most lies at work come out. It can also rewrite memories and wreck reputations. So how far should you trust it?

In a study published in 2012, volunteers at the University of California, Berkeley, were wired up to heart-rate monitors and asked to watch the scores from an economics game. One player kept cheating, taking everything and giving nothing back. As the watchers realised this, their heart rates went up.

Then they were offered a slip of paper. They could write a note warning the next person due to play the cheat. Most did, and once they’d passed the warning on, their heart rates settled. In another of the experiments, people gave up some of their own payment for the chance to send a note.

The researchers, Matthew Feinberg, Robb Willer and colleagues, called this prosocial gossip. Most offices would call it telling tales.

Gossip has a terrible reputation for something nearly everybody does every day. In a series about lies, it’s also awkward, because the grapevine plays both parts. It’s how most lies at work are found out, and it’s how plenty of them get around.

So this essay asks a plain question: when office gossip tells you something, how far should you trust it? The short answer is more than its reputation suggests, and less than it feels like at the time. The long answer depends on how the story reached you.

At one place I worked, everyone knew a manager was about to be sacked. His department’s figures were terrible and his own staff complained about him all the time, so I believed it too. He wasn’t sacked. He was simply moved to a role in another department that suited his skills better, and someone from his old team was promoted into his job. The grapevine got the shape of the story right and the ending wrong.

What counts as gossip

Researchers use the word more kindly than the rest of us. To them, gossip is simply talk about someone who isn’t there. It can be warm (“she was brilliant in that meeting”), neutral (“he’s moving to the Leeds office”) or nasty.

Rumour is a close cousin. Nicholas DiFonzo and Prashant Bordia, who have spent their careers studying both, separate them by what they’re for. Gossip is evaluative talk about people, and it builds and polices social networks. Rumour is unconfirmed information passed around to make sense of something uncertain or threatening: redundancies, a merger, why the finance director left on a Tuesday. Office talk blends the two all the time. “Apparently they’re restructuring, and apparently it’s all down to the new director” is both at once.

How much of it we do

Plenty, but not in the way the stereotype suggests.

In 1997 Robin Dunbar and colleagues analysed conversations in relaxed social settings. About two-thirds of conversation time went on social topics. That figure is often quoted as “two-thirds of what we say is gossip about people who aren’t there”. It isn’t quite. Much of it was people talking about their own lives and relationships, and critical comments about others were rare.

Megan Robbins and Alexander Karan came at it more directly. They pooled five studies in which 467 people wore a small recorder that switched on at random over two to five days, catching snippets of everyday conversation. Their analysis, published in 2020, found that about 14% of conversation was gossip, which works out at roughly 52 minutes a day. Around three-quarters of it was neutral. Negative gossip was about 15%.

Extraverts gossiped more. Women gossiped more than men, but only the neutral, information-sharing kind. Younger people were more negative than older people. People with less money or education didn’t gossip more than anyone else, whatever the books about the habits of the rich might claim.

So the typical piece of office gossip isn’t a knife in the back. It’s who’s moving desks and whose leaving do is on Friday.

Why we do it

Monkeys and apes maintain their friendships by grooming each other, which takes hours and only works one at a time. Dunbar’s argument, set out in his 1996 book Grooming, Gossip and the Evolution of Language, is that talk does the same job for humans far more efficiently. You can’t pick the fleas off three colleagues at once. You can chat to all three by the kettle.

Dunbar also argued that gossip has a policing job: keeping track of free riders, the people who take more than they give. The Berkeley heart-rate study fits that idea. So does a follow-up in 2014, in which Feinberg, Willer and Michael Schultz put 216 people into groups playing a game where they could either pool money for everyone’s benefit or keep it. When players could gossip about each other and leave selfish players out of the next group, cooperation held up. People who had been left out often came back and behaved better.

Bianca Beersma and Gerben van Kleef found something similar in 2011. People contributed more to a group when they knew others might gossip about them afterwards.

At work, this is the unofficial system that manages reputations. Nobody writes it down, but most people are honest most of the time, and the few who aren’t, the prolific liars in the fourth essay, tend to become known. Word gets round.

How lies really come out

In the first essay I looked at research by Hee Sun Park, Tim Levine and colleagues that asked people how they had actually caught someone lying. Hardly anyone had spotted it from the liar’s face at the time. The commonest routes were physical evidence and information from somebody else. Often, someone simply told them.

Levine’s truth-default theory, also covered in the first essay, says we believe people until something jolts us out of it. He counts information from a third party among those jolts. The grapevine is usually how that jolt arrives.

The clearest evidence of how much this matters comes from a lie that depended on stopping people talking. For years, sub-postmasters who rang the Post Office about unexplained shortfalls in the Horizon accounting system were told they were the only one having problems. According to the Criminal Cases Review Commission, call-centre staff were instructed to say so. When Computer Weekly interviewed seven former sub-postmasters in 2009, every one of them had been told the same thing.

A lie like that only works while people can’t compare notes. Later that year a group of them met in a village hall in Fenny Compton, Warwickshire, and founded the Justice for Subpostmasters Alliance. The institutional lie at the centre of that story belongs in the sixth essay. The point here is simpler. Keeping people apart was the defence. People talking to each other was what cracked it.

Why the office grapevine is usually right

Studies of workplace grapevines going back to the 1940s tell an unfashionable story: most of what travels along them about working life turns out to be true. Soldiers’ rumours about their own units, and co-workers’ rumours about their jobs, were found to be accurate nearly all the time. DiFonzo and Bordia’s review of the research reached the same conclusion. Rumours inside organisations tend to be highly accurate, especially compared with rumours after natural disasters, which can be wildly wrong.

The difference is who is talking to whom. In an office, people know each other, and rumours arrive with a name attached. The example researchers quote is “John told me that Mary was going to be sacked.” You know John. You know whether he’s in a position to know and whether he’s been wrong before. If he turns out to be wrong, you know where to find him. That accountability discourages people from starting stories they can’t stand behind, and it gives everyone else something to check.

A disaster zone is the opposite: frightened strangers, no names and nothing anyone can verify.

Why it goes wrong: the chain

In the mid-1940s at Harvard, Gordon Allport and Leo Postman ran a laboratory version of the party game where a message gets whispered down a line. One person described a picture to someone who couldn’t see it, who described it to the next, and so on down a chain of six or seven. Most of the detail had gone within five or six retellings.

What survived was simplified, which they called levelling. A few details were exaggerated, which they called sharpening. And the story bent towards what people expected to see, which they called assimilation.

One picture showed a scene on a New York subway train: a white man holding a razor, arguing with a well-dressed Black man. In more than half the chains, at some point, someone said the Black man was holding the razor. The prejudices of 1940s America rewrote the picture as it travelled.

There’s a postscript. In 1987 Molly Treadway and Michael McCloskey checked how textbooks and legal writing had described the study since. They found it had been distorted again and again, often turned into an eyewitness experiment it never was, with the razor said to end up in the Black man’s hand by the end of most chains. They called their paper “Cite Unseen”. The most famous study of how stories change in the retelling had changed in the retelling.

Their own trial run also suggested that all sorts of objects wander in these chains, hats and newspapers included. The razor’s journey may owe something to ordinary memory as well as prejudice.

Laboratory chains probably exaggerate how badly real rumours decay. Out in the world, people hear a story from several directions, question it and check it, and some rumours studied in the field didn’t shrink at all. But the chain is a fair model of something offices do every day: the story that reaches you third- or fourth-hand, from people who weren’t there.

Assimilation has everyday versions too. The manager at the start of this essay is a small example. The ending everyone expected was the sack, so that was the story that went round, and I believed it. It left out the less dramatic option, which is what actually happened: a move to a job that suited him better.

The friend of a friend

In 2020 Sacha Altay, Nicolas Claidière and Hugo Mercier published four studies, involving 1,824 people, on what happens to a rumour’s source as it’s passed on. People shortened the chain. Told by a friend that “a friend of mine said...”, they didn’t say “a friend of a friend of mine said”. They said “a friend told me”. Each teller quietly moved the story one step closer to its origin than it really was.

It’s a tiny lie, and nearly everyone told it. It matters because it works. Rumours credited to a credible friend of a friend (“my mate’s dad is a police officer, and he says...”) were rated as more plausible, and people were keener to pass them on.

The researchers point out that inaccurate rumours tend to carry either vague sourcing (“everyone’s saying”) or exactly that credible friend of a friend. In May 1969 a rumour swept Orléans in France that young women were being kidnapped from the fitting rooms of clothes shops owned by Jewish families. No woman had gone missing. The sociologist Edgar Morin, who studied it, found it was usually credited to someone like a friend’s father who was a police officer, or a cousin’s wife who was a nurse. There was never a victim, but there was always a credible witness, one step away.

At work, listen for the shape of the source. “Mike in payroll told me” is a claim you can test. “Someone in HR apparently said” is built, deliberately or not, so that nobody can.

Five people, one source

There’s a trap here that catches clever people. When you hear the same story from five colleagues, it feels confirmed. It might not be. If all five heard it from the same person at the same leaving drinks, you have one source and five echoes.

A 2007 study by Ralf Sommerfeld, Manfred Milinski and colleagues showed how strongly gossip steers people. Working with 14 groups of nine students, they found that gossip changed how people treated others even when they could see the other person’s actual record for themselves. The researchers concluded that gossip has a strong potential for manipulation. A follow-up the next year suggested the remedy: hearing several separate pieces of gossip, rather than one, blunted the effect.

Separate is the word that matters. Several independent accounts that agree are genuinely stronger than one. Several repeats of one account are not. The courts understand this better than most offices.

Why courts keep witnesses apart

When I worked in the criminal courts, witnesses waited outside until they were called. That isn’t just tidiness. A witness who sits through everyone else’s evidence can’t help absorbing it.

Research shows how easily that happens. In a 2003 study at the University of Aberdeen, Fiona Gabbert, Amina Memon and Kevin Allan showed pairs of people what they believed was the same video of a crime. In fact each saw it from a different angle, and only one of each pair could see a woman take money from a wallet. Some pairs talked it over before being questioned separately. Of those who talked, 71% went on to recall details they had never seen and had only heard about from their partner. Many who never saw the theft said she was guilty.

Nobody in that study was lying. Their memories had simply absorbed the conversation.

The criminal courts have a second guard against second-hand knowledge. What someone was told by somebody else, offered as proof that it’s true, is hearsay. Under the Criminal Justice Act 2003 it can be admitted only in defined circumstances. “I heard that...” isn’t evidence of the thing heard.

Offices do the reverse. Something happens, and within the hour everyone involved has talked it through, compared versions and settled on a story. By the time anyone asks formally, there may be one account in five mouths. That doesn’t mean anyone lied. It means that if you ever need to find out what happened, talk to people separately and soon, and ask what they saw rather than what they’ve heard.

Who gets hurt

Most gossip is harmless. The damage is in the negative kind, and the research on how it travels explains why it’s so hard to shake off.

In 2010 Travis Grosser, Virginie Lopez-Kidwell and Giuseppe Labianca mapped who gossiped with whom inside one organisation. Colleagues who only worked together shared positive gossip. Negative gossip was shared between friends, and more so when the pair had other friends in common.

That makes sense. Running someone down is risky, so you only do it with people you trust. It also means the damaging version of a story tends to reach you from someone you like and believe, which is what makes it persuasive and hard to challenge. The same study found that heavy gossipers were rated by their peers as more influential, and by their bosses as poorer performers.

On the receiving end, Elena Martinescu and colleagues found in 2019 that people who heard negative gossip about their own work felt two things at once. Guilt and shame made them want to put things right. Anger at the gossiper made them want to get even. A later study by Martinescu’s team found that negative gossip left its targets feeling less included, and less willing to do anything for the organisation beyond their job description.

A minority use gossip deliberately. In a 2019 study of why people gossip, Freda-Marie Hartung and colleagues found that damaging someone’s reputation was the least common reason given. People high in Machiavellianism, the trait psychologists associate with manipulation and ends-justify-the-means thinking, were the exception. They were more likely to bad-mouth others specifically in professional settings.

“I’m only repeating what I heard”

The law has a view on that phrase. I’m not a lawyer, and office gossip rarely ends in court, but the principle is worth knowing. In English defamation law, the repetition rule means you can’t escape responsibility for a damaging claim by putting “I’ve been told that” or “it’s rumoured that” in front of it. In Lewis v Daily Telegraph, decided by the House of Lords in 1963, Lord Reid said that repeating someone else’s libel is as bad as making it yourself.

The best-known modern example is a single tweet. In November 2012, after BBC Newsnight wrongly linked an unnamed senior Conservative to child abuse, Sally Bercow asked on Twitter why Lord McAlpine was trending, adding “*innocent face*”. She hadn’t accused him of anything in so many words. In 2013 the High Court held that, in context, the tweet meant he was the man in the story, and treated her as if she had made the allegation herself. He was entirely innocent.

The grapevine grows where the official story isn’t

Allport and Postman’s best-known idea is a formula: the amount of rumour in circulation depends on how much a subject matters to people, multiplied by how ambiguous the evidence is. If either is zero, there’s no rumour. Nobody has managed to measure it that precisely, but the logic has lasted. Rumours flourish when people care about something and official sources aren’t telling them enough, or aren’t believed.

That leaves managers who complain about gossip in an awkward spot. In a Dutch childcare organisation, Lea Ellwardt, Rafael Wittek and Rudi Wielers found in 2012 that negative gossip about managers rose when staff didn’t trust them, weren’t on friendly terms with them and rarely had contact with them. It rose further when staff trusted each other. DiFonzo and Bordia drew a similar conclusion from their research: where employees trust management, there is less rumour.

A thriving rumour mill is often a symptom. In the fourth essay I argued that working somewhere people are afraid to tell the truth does more damage than a few prolific liars. The grapevine is where the truth goes when it can’t come in through the front door.

Denials have their own lesson. In a 1998 experiment, Bordia, DiFonzo and Verity Travers tested who is best placed to deny a rumour about organisational change. It wasn’t the most senior person. The denial that did most to reduce belief in the rumour, and the anxiety around it, came from someone whose level matched the rumour’s scope. A rumour about one department is best answered by that department’s head, not by a chief executive who plainly doesn’t know the details.

I’ve been on the other side of this. In IT, you tend to know first. New accounts and removed access tell you who’s coming and going, and when someone left suddenly, senior staff wanted access to their account. Colleagues often asked me what I knew. I always kept quiet.

How to listen to it

None of this makes gossip good or bad. It makes it information of varying quality, and the quality can be judged.

Ask how they know. “Who told you?” isn’t an accusation; it’s the question the whole of rumour research points to. Count sources, not repetitions. Treat “everyone’s saying” as a warning, and notice when you shorten the chain yourself. If you heard it from someone who heard it from someone, say so.

Be slowest with the negative kind, especially about someone’s work. It travels between friends, it’s believed because of that, and “I only passed it on” is no defence in law or in decency.

If you run a team, fill the gap. Say what you know, what you don’t and when you’ll know more. Silence doesn’t stop the conversation. It just means you’re not part of it.

And when several people who don’t know each other start describing the same problem, listen. That isn’t gossip getting out of hand. It’s the system working.

I did my best not to get involved in gossip. But sometimes something was causing problems in my job or putting the company at risk, and rumours were already going round. In IT, it could be obvious what was going on, and inside the IT team, people did get talked about. The researchers would call that gossip. At its best, it was the Berkeley kind, with the warning note.

The Post Office told sub-postmaster after sub-postmaster that they were the only one. It was never true. The lie began to come apart when enough of them found each other and compared notes, first in the pages of a computer magazine and then in a village hall in Warwickshire.

Sources and further reading

Allport, G. W. & Postman, L. (1947). The Psychology of Rumor. Henry Holt.

Altay, S., Claidière, N. & Mercier, H. (2020). It happened to a friend of a friend: inaccurate source reporting in rumour diffusion. Evolutionary Human Sciences, 2, e49.

Beersma, B. & Van Kleef, G. A. (2011). How the grapevine keeps you in line: gossip increases contributions to the group. Social Psychological and Personality Science, 2, 642–649.

Bordia, P., DiFonzo, N. & Travers, V. (1998). Denying rumors of organizational change: a higher source is not always better. Communication Research Reports, 15, 188–197.

Criminal Cases Review Commission. Post Office Horizon cases (ccrc.gov.uk).

Criminal Justice Act 2003, s.114.

DiFonzo, N. & Bordia, P. (2007). Rumor Psychology: Social and Organizational Approaches. American Psychological Association.

Dunbar, R. I. M. (1996). Grooming, Gossip and the Evolution of Language. Faber & Faber.

Dunbar, R. I. M., Marriott, A. & Duncan, N. D. C. (1997). Human conversational behavior. Human Nature, 8, 231–246.

Dunbar, R. I. M. (2004). Gossip in evolutionary perspective. Review of General Psychology, 8, 100–110.

Ellwardt, L., Wittek, R. & Wielers, R. (2012). Talking about the boss: effects of generalized and interpersonal trust on workplace gossip. Group & Organization Management, 37, 521–549.

Feinberg, M., Willer, R., Stellar, J. & Keltner, D. (2012). The virtues of gossip: reputational information sharing as prosocial behavior. Journal of Personality and Social Psychology, 102, 1015–1030.

Feinberg, M., Willer, R. & Schultz, M. (2014). Gossip and ostracism promote cooperation in groups. Psychological Science, 25, 656–664.

Gabbert, F., Memon, A. & Allan, K. (2003). Memory conformity: can eyewitnesses influence each other’s memories for an event? Applied Cognitive Psychology, 17, 533–543.

Grosser, T. J., Lopez-Kidwell, V. & Labianca, G. (2010). A social network analysis of positive and negative gossip in organizational life. Group & Organization Management, 35, 177–212.

Hartung, F.-M., Krohn, C. & Pirschtat, M. (2019). Better than its reputation? Gossip and the reasons why we and individuals with “dark” personalities talk about others. Frontiers in Psychology, 10, 1162.

Levine, T. R. (2020). Duped: Truth-Default Theory and the Social Science of Lying and Deception. University of Alabama Press.

Lewis v Daily Telegraph Ltd [1964] AC 234.

Martinescu, E., Janssen, O. & Nijstad, B. A. (2019). Self-evaluative and other-directed emotional and behavioral responses to gossip about the self. Frontiers in Psychology, 9, 2603.

Martinescu, E., Jansen, W. & Beersma, B. (2021). Negative gossip decreases targets’ organizational citizenship behavior by decreasing social inclusion. Group & Organization Management, 46, 463–497.

McAlpine v Bercow [2013] EWHC 1342 (QB).

Morin, E. (1969). La Rumeur d’Orléans. Seuil.

Park, H. S. et al. (2002). How people really detect lies. Communication Monographs, 69, 144–157.

Robbins, M. L. & Karan, A. (2020). Who gossips and how in everyday life? Social Psychological and Personality Science, 11, 185–195.

Sommerfeld, R. D., Krambeck, H.-J., Semmann, D. & Milinski, M. (2007). Gossip as an alternative for direct observation in games of indirect reciprocity. PNAS, 104, 17435–17440.

Sommerfeld, R. D., Krambeck, H.-J. & Milinski, M. (2008). Multiple gossip statements and their effect on reputation and trustworthiness. Proceedings of the Royal Society B, 275, 2529–2536.

Treadway, M. & McCloskey, M. (1987). Cite unseen: distortions of the Allport and Postman rumor study in the eyewitness testimony literature. Law and Human Behavior, 11, 19–25.