The Truth About Machiavellianism at Work: It's Not What You Think | Ep 8

I'm using it because researchers use it to describe the theory behind the behaviour. A person high in trait Machiavellianism is strategic, patient, willing to use other people instrumentally and, above all, controlled. They weigh what it costs to be caught and then they act accordingly. That last part is the entire trait and most of what follows in this episode comes out of it.

Now what it's not, because the term is often misused and misunderstood. It's not lying exactly. Somebody who lies constantly gets caught constantly. This trait is defined by the opposite calculation. It's not charm, which is a different trait doing a different job. And it's not a diagnosis, because there's no such clinical category.

Narcissism has one. Psychopathy has a relative. Machiavellianism has nothing. What it looks like, if it looks like anything, is somebody keeping their head down in a meeting situation they could have won, because winning this week would cost them something later. Nobody writes that down. There's no box on any form for a thing that a colleague chooses not to do, because Machiavellian is a skill that keeping things below the threshold that people notice.

That's why the rest of this is about measurement rather than about people. For about a decade, the serious argument in this field has not been about what Machiavellianism does. It's been about whether the trait exists separately from psychopathy. And this debate exists because of the related personality questionnaires.

Three studies compared what the standard measure of each trait actually predicts, across personality ratings and a spread of other outcomes. The two came out looking almost identical. The authors concluded that the measures appeared to be assessing the same thing, and that the Machiavellianism assessments failed to capture the trait their own theory describes.

The specific failure is worth noting. The theory stays controlled, whereas the questionnaire correlates with impulsiveness, which is the opposite and the signature of trait psychopathy. Here's the part that should have ended the argument. Two summaries of the earlier research suggested that the overlap between the standard measures were about half, which for two scales supposedly measuring different things is close to what you get for two versions of the same test.

Then somebody rebuilt both, across four and a half thousand people, keeping only the questions consistent with what the theory predicts. Rebuilt in this way, the overlap fell to about a quarter. So the theory was fine, the questions weren't. This means much of what you've read about Machiavellianism at work may actually be about psychopathy.

You can't fix that by reading more carefully, because the labels are the issue. There are some So somebody stopped asking people what they're like in general and started asking them what they'd done that day. 317 adults, an evening survey, every day for 30 consecutive days. Not, are you a manipulative person, which everybody answers generously.

Instead, today, did you do something manipulative? Measured the ordinary way, as traits, Machiavellianism and psychopathy overlapped by more than 70%, which is a finding that started the whole argument. Measured this alternative way across 30 days, the overlap fell to about 16%. What separated them was the content of the days.

The Machiavellian days were days of restraint, and the item that captured it best was keeping a low profile to get my way. The psychopathic days were days of exposure, and the item that captured that best was walking into something dangerous. One more thing from the same study, and it's a part almost nobody has picked up.

A manipulative day predicted a more impulsive day afterwards. The reverse never happened. This suggests the restraint state is a stable one, and the exposed state happens when the restraint has been running a while, and something gives. The limitations of this study were that it was done as a single study in one country, and both measures still rely on people describing themselves.

Nothing. If you've come across this subject in the last year, there's a fair chance you did so through a claim about AI and cheating, and the claim is half right. 812 students in Taiwan. Four traits rather than the usual three, with everyday sadism added. Then they measured how much each student had misused generative AI on their coursework.

Three of the four traits predicted it. Narcissism, psychopathy, and everyday sadism. All about the same strength. Machiavellianism predicted nothing at all. Then the interpretation attached itself. The calculating Machiavellians are too careful to risk it. That's not what the authors said, and I'd encourage anyone quoting this study to read what they did say, because it's more interesting.

They gave two reasons. Neither is about restraint. People under-report their own misconduct, and the people best at concealing things are the hardest group of all to measure honestly. In the version of the personality questionnaire they used, five of their own questions didn't hang together with the rest of the scale.

So, a result of nothing, from a measure that partly failed, asking people to describe behaviour they specialise in hiding. That's not evidence the behaviour didn't happen. It's a measure telling you it couldn't see. menus Now the awkward part. A separate study asked 283 people how willing they would be to pass off chatbot text as their own.

They found exactly the opposite. Machiavellianism predicted it. So did narcissism and psychopathy. Across 23 studies, a review including only work that measured at least two traits together found something in between. Machiavellianism was linked to plagiarism and cheating, but not consistently linked to paying somebody else to do the work.

Think about that last one. Paying somebody else is the version with the longest evidence trail and the most people who know, which is the version a careful calculator ought to avoid most, and it's the version the trait fails to predict. Another result in the middle study gets quoted far less than it should.

The strongest predictor was not a dark trait at all. It was honesty and humility, and specifically the fairness part of it, meaning how committed somebody is to treating people equitably when nobody is checking. Three competent teams, three different answers on the simplest question anybody could ask.

Hang on to that because it's about to explain everything. All right. Nothing. What happens to these people at work? Whether it shows depends on something else entirely, and that something else has a name and a measure of its own. It's called political skill. Reading a room accurately, influencing people, building relationships, and coming across as sincere while doing all three.

It's measured routinely, it's taught, and it's not a dark trait. 1,428 people were rated by their colleagues. High political skill and the ratings were high, whether or not the person was also high in Machiavellianism. Without the skill, the ratings fell, and they fell hardest among the people who'd been there longest.

Read that carefully because it's counterintuitive. Time doesn't expose this trait. Time exposes the people who lack the skill to carry it. The second study is the one that stops this from being a compliment. The same researchers looked at whether political skill also reduced the conduct. It did not.

The skilled ones were doing the harm anyway, early on, at exactly the point their ratings were most favourable. The mask didn't change the behaviour. It changed who could see the behaviour. It Where does the cost actually land? 550 employees rated by 1,100 colleagues. Reputation is the hole of the route.

People with higher Machiavellianism developed worse reputations. And reputation, not the trait, is what pulled their performance ratings down. Political skill broke the chain. One more finding that nobody likes. A German household panel of 8,500 records. The higher the score, the more likely they were to be managing somebody.

Rising steadily all the way up the scale rather than levelling off somewhere respectable. It's a modest effect on a small base. And most managers are not high in this trait. What it establishes is that nothing in the usual approach to selecting for promotion filters it out. Here's where that leaves the empty file.

A misconduct record is a record of detected misconduct. And detection is concentrated on people who act visibly, act impulsively, and lack the skill to manage the aftermath. The trait most likely to survive that record is the one defined by not doing any of those things. This does make a clean record suspicious.

It makes it a much narrower piece of evidence that people assume it is. And that's a feature of the system. This does make a special task. So I want to write it a little bit. This does make a So what do you actually do? Four things, and I'm going to be transparent about the evidence supporting each, which is a little thinner than anybody selling this would admit.

Three things not to do first. Don't set out to catch them. Three research teams tried to establish whether this trait predicts cheating on an essay with captive samples, an unambiguous offence with funding and consent. They got three different answers. You're no better placed than they were, and the attempt has a failure mode.

You become the person with a theory about a colleague, and the trait most reliably linked to that behaviour is one nobody will be measuring. Don't screen for it. The standard measure does not measure the thing it names, so a score is ambiguous before anybody answers a question. The measures are self-reports of concealment, which is the design problem you watched defeat a competent research team ten minutes ago.

On the one occasion in this whole literature where somebody measured an alternative, fairness beat every dark trait in the same study. Finally, don't read an empty personnel file in either direction. It's weak evidence that they behave well. It's not evidence that they didn't, and nothing on file proves how careful they are.

It's an untestable argument, especially when you're the one making the claim. What Now the honest framing. Nobody has ever trialled it against the problem. The four moves come from the mechanism and my own practice rather than a study, and I want to label them before I give them to you. Count episodes, not incidents.

Your organisation asks of each event whether that event on its own is serious enough to act on. That question has a correct answer and it's almost always no, which is while the file stays empty. The question to ask instead is how many times something small has happened, over what period, and whether the small things all point in the same direction.

In practice that means keeping detailed records of things that aren't worth raising individually. The commitment made in a meeting and not in writing. The decision that was already taken. The information that reached you late but reached somebody else on time. Dated, factual, nothing about the character.

And two caveats because this can go wrong. It's slow because a rate needs months before it's a rate, and it looks bad if it's found and can't be explained. So the note should contain only things you'd be happy to read aloud, publicly. I believe The second move is the one I'd actually bet on, and the evidence points somewhere specific.

Ask the colleagues, not the manager. The only channel through which this trait has ever been exposed is what colleagues come to think. Both studies found it in co-worker ratings. Neither found it in anything a manager could see alone, which stops being surprising the moment you notice that the manager is the audience the behaviour is organised around.

Any process that collects a view from several colleagues independently over time is looking in the one place the evidence says something is visible. Multi-rater feedback is the obvious instrument, and I'd make two adjustments. Ask about behaviour, not qualities, because is he collaborative is a question about character, whereas did he share the forecast before the meeting is a question about an event.

Compare across cycles rather than reading each round on its own, because a rate is the thing you're trying to see, not a single round. This is the only recommendation here with a finding behind it rather than an argument, and one of those easily beats four of the other kind. This Two more, and both are about the room rather than the person, so neither requires you to conclude anything about a person.

Reconnect the decision to its consequences. What produces this pattern is a structure where the benefit of a decision lands on whoever took it, and the costs land somewhere else, later, on somebody who wasn't there. Record who took it, carry commitments into the next meeting rather than letting them quietly expire, and put the person who carries it in the room when it's made.

None of that accuses anybody of anything, which is the whole point of it. The fourth is a personal version of that. Send a short note afterwards saying what was agreed and by whom. In two minutes, it turns an understanding into a dated record without being prejudicial to anybody. Here's where all this runs out.

None of these four have been tested against the problem. The mechanism is evidenced, the remedy is not. The research on why this happens is good. The research on what to do about it doesn't exist. And anybody offering you a validated method for handling a Machiavellian colleague is describing a product rather than evidence-based literature.

So In the next episode, I'm covering something that looks like anger, but isn't. Some cruelty at work has nothing to do with provocation and everything to do with somebody being bored. Look out for it. One last thing about this episode, as usual. Everything here is a statement about groups. None of these studies promotes a conclusion about the colleague you may have been thinking about while watching.

The central point of this episode is that you don't have the evidence. That cuts both ways. Research can change what you count and what you look at. It can't tell you what's going on inside a person. latches Three last things and then we're done for this episode. The full article supporting this video is on my site.

It carries every reference and what each one is actually worth and the four things I still don't know. I've put the link in the description below and it's on the screen now. If you found this episode interesting and you'd like the rest of the series please like and subscribe to my channel. It helps get my content out there and ensures the next one lands in front of you.

And a question for you to answer from your experience. Think about the last time your organisation decided somebody had done nothing wrong. Was that a finding or was it just a lack of evidence? Nothing had been written down. I'm Dr Nick Kecker. Thanks for watching. I'll see you next time. Bye.