Abstract
The question this article answers is: when one colleague is consistently more difficult with one person than with everybody else, what does the evidence say about why, and what should that person, their manager, and the organisation do, given that none of them can see inside the colleague?
The article draws on four bodies of work. The first treats mistreatment as something that happens between two particular people rather than something one person carries everywhere, using the social relations model, a method that separates what a person does with everybody from what they do with one partner in particular (Back & Kenny, 2010), and a synthesis of seven studies finding moderate consistency in how people behave across partners and little evidence that anybody draws the same response from everybody (Kenny, Mohr & Levesque, 2001). A three-sample study of coworker pairs applies the same logic to incivility (Taylor, Locklear, Kluemper & Lu, 2022).
The second body of work asks who gets targeted. It covers envy of high performers (Lam, Van der Vegt, Walter & Huang, 2011; Kim & Glomb, 2014; Duffy, Scott, Shaw, Tepper & Aquino, 2012), supervisors who single out particular subordinates (Tepper, Moss & Duffy, 2011; Li, Law, Zhang & Yan, 2024; Xiao, Togola, Zhang & Yu, 2024) and incivility that falls along lines of sex and ethnicity (Cortina, Kabat-Farr, Leskinen, Huerta & Magley, 2013). The third concerns the dark traits and the conditions under which they are expressed (Wisse & Sleebos, 2016; Dåderman & Ragnestål-Impola, 2019; Feng, Keng-Highberger, Yam, Chen & Li, 2023). The fourth concerns reciprocity and perspective (Andersson & Pearson, 1999; Rosen, Koopman, Gabriel & Johnson, 2016; Baumeister, Stillwell & Wotman, 1990).
A late chapter sets out the people most often mislabelled in this territory, including the reader. The practical section proposes a sequence of checks and moves, labels the evidence status of each, and is candid that almost none of them have been tested as an intervention with the particular pattern this article describes.
Keywords: workplace incivility; targeted mistreatment; abusive supervision; envy; high performers; selective incivility; dark triad; Machiavellianism; reciprocity; organisational psychology.
PART I. WHAT ACTUALLY HAPPENS
Chapter 1: Fine With Everyone Except You
1.1 The Inbox That Tells You
The scenarios below are composite. They draw on recurring patterns in research and advisory practice, and they describe no real, identifiable person or organisation.
A newly appointed head of operations at an insurer notices it in her second month, and she notices it first in her inbox: the financial controller, whose sign-off she needs on every change to the claims process, answers the other heads of function within the hour, copies them politely into his replies and thanks them by name, while her requests sit for four or five days and come back with a single line, usually a question she has already answered in the original message. In the monthly leadership meeting, he is relaxed and funny with everybody else, and when she speaks, he looks at his laptop.
She raises it carefully with a peer she trusts, who is puzzled and says that the controller is one of the most helpful people in the building. She raises it more carefully still with her own director, who says that he has always found him straightforward, that the controller is under pressure from the audit committee, and that perhaps she could try picking up the phone. By the end of the quarter she has stopped mentioning it and started to wonder at about two in the morning, whether the problem is her.
Two other things are true, and neither of them is visible to anybody who is not her. Her first project has already cut the cost of a claims process the controller spent three years defending, and the board has praised it. And she has, without quite deciding to, begun writing to him in shorter, colder messages than she writes to anybody else, copying in her director, which he has certainly noticed.
1.2 Why Nobody Believes You
The defining feature of this pattern is that the person on the receiving end is the only reliable witness, and that is no coincidence. A colleague who is difficult with one person and courteous with the rest is, by construction, a colleague whom most people experience as courteous, so every conversation in which the target tries to describe the problem is a conversation with somebody whose own experience contradicts it.
That produces a second harm on top of the first. The target is not only treated worse than everybody else, they are also asked, implicitly and repeatedly, to accept that their reading of their own working life is less reliable than the readings of people who were not there; one study of incivility aimed at a particular person found it linked to a weaker sense of belonging and to embarrassment, and the embarrassment was greater when the person doing it had more power (Hershcovis, Ogunfowora, Reich & Christie, 2017).
The phenomenon can be stated without any technical vocabulary. Someone who is fine with everyone else is difficult with you, and the people you tell do not believe you because, to them, that person is fine.
1.3 The Two Conclusions That Come Too Early
People reach two conclusions in this situation, and in my experience, they reach one within a few weeks and rarely revisit it. The first is that the colleague is a particular kind of person: manipulative, narcissistic, the sort who is charming to anybody who matters and cruel to anybody who does not. The second is that the problem lies with the person on the receiving end: too sensitive, too abrupt, too new, or simply the kind of person who attracts this treatment wherever they go.
Both conclusions have something to them, and the evidence in this article supports each in specific circumstances. What neither of them survives is being reached first. Each turns a pattern between two people into a verdict about one of them, and each leaves the head of operations in 1.1 either fighting about what her colleague is or wondering what is wrong with her, when the most useful information in the scene is the pattern itself: who it is aimed at, when it started, what it follows, and what she is now doing back.
Chapter 2: Common, and Aimed
2.1 Three in Four
Start with how ordinary this is, because the target in 1.1 is likely to feel that it is rare. Incivility at work, which researchers define as low-intensity rude or discourteous treatment where it is unclear whether harm was intended, is widespread, and a 2026 review of the field reports a pooled figure of about three in four employees having experienced it (Magley, Kabat-Farr, Malcore & Walsh, 2026).
That figure describes incivility in general rather than incivility aimed consistently at one person, and nobody has measured how often the specific pattern in 1.1 occurs, so far as I could find. The figure does establish that the head of operations is not in an unusual situation. She is in one of the most common unpleasant situations at work, and most organisations have no process for it, because each instance is too small to report and too persistent to ignore.
2.2 The Pattern Is Aimed
The more important point is that mistreatment at work is rarely scattered evenly. Hershcovis and colleagues pooled 57 studies of workplace aggression, using a meta-analysis, which means combining the results of many earlier studies rather than running a new one, and concluded that the pattern of predictors is target-specific (Hershcovis et al., 2007): conflict between two people, for example, predicted aggression aimed at people more strongly than aggression aimed at the organisation, which is the statistical form of a very ordinary observation: people are difficult with particular others for particular reasons.
Supervisors show the same shape, and the research on them says so in terms. A 2024 study of 254 employees in 84 work groups builds on the observation that leaders tend not to abuse subordinates equally, and found that being treated worse than one’s peers lowered performance through a sense of being worse off than comparable colleagues, which researchers call relative deprivation (Xiao, Togola, Zhang & Yu, 2024). A larger pooling of research on leaders who form closer working relationships with some team members than others, covering 4,114 work groups and 21,745 people, found that this kind of uneven treatment harmed the harmony and solidarity of the group as a whole (Yu, Matta & Cornfield, 2018).
So when the head of operations says he is only like this with me, she is describing something the research treats as normal in its form, whatever its cause, and the next three chapters take the causes in turn.
2.3 Why It Costs More Than It Looks
Low-level mistreatment is easy to dismiss because each instance is so small. The evidence on what it adds up to is not small. A pooling of 66 samples compared the effects of aggression depending on who it came from, and found that aggression from a supervisor did the most damage to how people felt about their work, while aggression from coworkers did more to what people went on to do than aggression from outsiders (Hershcovis & Barling, 2010).
The selective incivility study in Chapter 5 found uncivil treatment linked to intentions to leave (Cortina, Kabat-Farr, Leskinen, Huerta & Magley, 2013). That is the cost an organisation can count, and the cost it cannot count is the one in 1.1: a capable person spending the small hours wondering whether the problem is her.
Neither figure tells you how often the pattern ends in a resignation, but together they tell you it is not trivial.
Evidence status: evidenced. Prevalence is from a 2026 review reporting a pooled figure; target specificity is from a 57-study meta-analysis and from studies of supervisors. No study was found that measures how often one person is consistently singled out by one colleague.
PART II. WHY YOU
Chapter 3: The Pair Is the Unit
3.1 Some of It Belongs to Neither of You
The most useful single idea in this literature is also the least intuitive, and it comes from a method rather than from a finding. The social relations model is a way of taking a group in which everybody rates or interacts with everybody else, and separating three things that ordinary observation runs together: how a person behaves with everybody, how everybody behaves with that person, and how two particular people behave with each other over and above both of those (Back & Kenny, 2010). The third part belongs to the pair, not to either member.
When David Kenny and colleagues brought together seven studies of how people actually behave in interactions with different partners, the picture was of moderate consistency in how a person behaves across partners, little or no evidence that anybody consistently draws the same response from everybody they meet, and preliminary evidence of responses unique to particular partners (Kenny, Mohr & Levesque, 2001). Put plainly, people carry some of their behaviour with them from one encounter to the next, almost nobody is treated the same way by everyone, and some of what happens between two people happens only between those two.
That second clause deserves a moment, because it bears directly on the fear in 1.1 that the problem is the target. The idea that some people are simply the kind whom others mistreat, wherever they go, found little support in that synthesis. It is a group-level finding from studies of social interaction rather than of work, and I would not stretch it to cover every case. It still runs opposite to what the head of operations was telling herself at two in the morning.
3.2 The Same Is True of Incivility at Work
Shannon Taylor, Lauren Locklear, Donald Kluemper and Xinxin Lu applied the same logic to workplace incivility, in three separate samples: employees of a restaurant chain in the United States, employees of a technology manufacturer in China, and employees drawn from many industries (Taylor, Locklear, Kluemper & Lu, 2022). They report that the incivility people experience and the incivility they give out both vary substantially at the level of the pair, and that within a pair the two are linked even after allowing for each person’s general tendency to receive or to give it.
That is the finding the chapter title rests on. How rude one colleague is to another is partly a feature of the first colleague, partly a feature of the second, and partly a feature of the two of them together, and the last part is large enough to matter; I have seen a figure for how large, in the model’s author’s own account of the study, and I have not been able to read it in the paper, so I am not quoting it; see publication status footnote 2.
Evidence status: evidenced as to the pair being a real unit of behaviour; the size of the pair’s share at work is not quoted. A synthesis of seven studies of social interaction and a three-sample study of coworker incivility. Neither was designed around one person singled out by one colleague.
3.3 What This Does and Does Not Say
It does not say that nobody is simply unpleasant. The same work finds moderate consistency in how people behave across partners, so a colleague who is curt with a great many people is showing you something about themselves, and Chapter 6 comes back to that. Nor does it say that the target has done something to deserve the treatment, which is a different claim and one the evidence in Chapter 8 argues against directly.
What it says is narrower and more practical. Where the difficulty is concentrated on one pair, a good part of the explanation may live in the history, the stakes and the habits of that pair, which is the one part of the problem the two people involved can actually change. It also says that the question What kind of person is he? is aimed at the smallest part of the answer, which is why the pair is where to look first.
Chapter 4: Why Good Work Attracts It
4.1 The High Performers Are the Ones It Finds
If the pattern is aimed, the obvious question is who, and the most counterintuitive answer in the literature is also one of the best supported. Eugene Kim and Theresa Glomb surveyed 4,874 university staff in 339 work groups and found that higher performers were more likely to report being targets of mistreatment by colleagues (Kim & Glomb, 2014). A second study of 217 employees in 67 work groups across three organisations, using data from more than one source and collected at two points in time, found that the route ran through envy in the rest of the group, and that it was weaker where people identified strongly with their work group.
A study of teams reached a similar conclusion from a different direction. Lam and colleagues found that better-performing team members attracted more harming from teammates when the team’s goals were less cooperative, and that more cooperative goals reduced it (Lam, Van der Vegt, Walter & Huang, 2011), and Duffy and colleagues, in a study of hospital employees and a study of student teams followed over four points in time, found that envy led to undermining, meaning behaviour intended over time to damage somebody’s reputation, relationships or success. The link was stronger where people identified weakly with their colleagues (Duffy, Scott, Shaw, Tepper & Aquino, 2012).
This is the finding I would most like a reader in the head-of-operations position to carry away. The difficult treatment she received is, on this evidence, at least as consistent with her having done something well as with her having done something wrong, and her first project cut the cost of a process her colleague had spent three years defending.
Evidence status: evidenced. A two-study paper with a combined sample of more than five thousand staff, and two further multi-study papers. The designs are field surveys and cannot establish that performance causes mistreatment; they establish that the two travel together and that envy is a plausible route.
4.2 Supervisors Do It at Both Ends
Supervisors are a special case, because they have more power over the target and more ways to make the treatment deniable. Tepper, Moss and Duffy studied supervisor and subordinate pairs at two points in time and found that when supervisors saw a subordinate as different from themselves in values and attitudes, they experienced more conflict with that person, rated their performance lower, and treated them with more of what researchers call abusive supervision, meaning sustained hostile behaviour short of physical contact (Tepper, Moss & Duffy, 2011).
A 2024 study added the other end. Across two scenario experiments and a field study, subordinates seen as performing poorly drew abusive treatment through the supervisor’s anger, and subordinates seen as performing well drew it through the supervisor’s envy, especially where the supervisor was inclined to compare themselves with others (Li, Law, Zhang & Yan, 2024). Two further studies, published in 2025, found supervisor envy towards competent or unusually innovative subordinates leading to hostile treatment, in one case in 198 supervisor and subordinate pairs and in the other in 237 (Azeem, Haq, Murtaza & Khan, 2025; Wang, De Pater, Yi & Wray-Bliss, 2025).
The practical implication is uncomfortable for anybody who has been told to keep their head down and perform. Performing well does not protect against this pattern and may, in some settings, even draw it.
4.3 Why Nobody Else Sees It
Put Chapter 3 and 4.1 together and the disbelief in 1.2 stops being mysterious. If a colleague’s difficult behaviour is aimed at the person who threatens their standing, it will be invisible to everybody who does not, which is nearly everyone. Nobody is lying when they say the controller has always been helpful. They are reporting their own pair, not hers, and both accounts are true.
It also suggests why the pattern can look like a personality problem to the target and like a personal problem to everybody else. From her chair, the behaviour is consistent and targeted, so it looks like a trait; from theirs, it is absent, so it looks like something she has brought with her.
Chapter 5: When It Follows a Line
5.1 Selective Incivility
Some aimed mistreatment has nothing to do with performance or envy, and it is the kind that does the most damage when you treat the problem as a difficult relationship. Cortina and colleagues surveyed employees of a city government (369 people), a law enforcement agency (653) and the United States military (15,497), and found that the target’s sex and ethnicity, though not their age, affected how much incivility they experienced; that women of colour reported the worst treatment, and that uncivil treatment was linked in turn to intentions to leave (Cortina, Kabat-Farr, Leskinen, Huerta & Magley, 2013). The authors call this selective incivility and describe it as a modern form of discrimination.
It matters here because selective incivility looks, from the inside, exactly like the pattern in 1.1. It is also the easiest to miss.
It is low-intensity, deniable in any single instance, and aimed at particular people while leaving others untouched. What distinguishes it is that it follows a line, and the line is visible only when you look at who else the same colleague is difficult with. One case shows nothing, while five may show a line.
Evidence status: evidenced. Three large survey samples, one of them more than fifteen thousand people. Self-report throughout. Whether a particular instance is discrimination in law is a question for a qualified adviser, not for this article.
5.2 Difference as a Trigger
The supervisor study in 4.2 is relevant here too, because the trigger it found was perceived difference in values and attitudes, not in any protected characteristic (Tepper, Moss & Duffy, 2011). That suggests a wider version of the same shape: a colleague may be difficult with the one person on a team who thinks differently, came from a different background or function, or was hired by somebody else, and courteous with everybody who resembles them.
A 2026 study asked directly why supervisors are uncivil to some subordinates and not others, and found incivility towards subordinates seen as poor performers was highest where supervisors cared about results at almost any cost and held flexible views about what is morally acceptable (Wu, Shen, Ma, Zhang & Ren, 2026). I note it because it is recent and asks the same question this article addresses. I have read its abstract, not its figures, so I report only its direction.
5.3 Why the Line Changes What You Do
A difficult relationship is something two people can work on. A pattern that follows sex, ethnicity, age, disability or any similar line is clearly not a relationship problem, and advising the target to build rapport with the person treating them worse for that reason asks them to manage somebody else’s conduct at their own expense. Chapter 10 treats this as a separate route for exactly that reason, and it is the one place in this article where I would move quickly from observation to formal process.
Chapter 6: The Dark Traits Question
6.1 Where the Traits Sit
This series is about dark personality at work, and the question most readers have been waiting for is whether the colleague who is difficult only with them is showing a dark trait. The honest answer has two halves. The half people usually leave out is the more useful.
The first half is that the dark traits do show up among people who mistreat others at work. In a study of 172 Swedish full-time employees, the people who reported bullying colleagues scored high on Machiavellianism, a pattern of cynical, strategic and manipulative dealing with other people, and on psychopathy, which covers callousness and impulsivity, and the people who reported being bullied did not (Dåderman & Ragnestål-Impola, 2019). A 2026 pooling of nine studies of the three traits most often grouped together, Machiavellianism, narcissism and psychopathy, found all three linked to bullying others, with psychopathy most strongly (Sui, Malik, Tiliopoulos & Yu, 2026). These three are often grouped together as the dark triad.
The second half is where the traits are not. In the Swedish study, they sat with the people doing the bullying rather than with the people it was done to, which is worth setting beside the target’s two-in-the-morning question in 1.1. And a pooling of 36 samples covering 13,896 people found that the personality of those on the receiving end of harassment was related to it only weakly (Nielsen, Glasø & Einarsen, 2017), while a larger pooling of 105 samples and 51,008 people found that features of the working context predicted who experienced incivility better than personality did (Yao, Lim, Guo, Ou & Ng, 2022).
Evidence status: evidenced, with a heavy caveat on design. The dark-trait studies are mostly single surveys relying on what people say about themselves, and the 2026 pooling reports very large differences between the studies it combined. They show that the traits and bullying travel together. They cannot show that any given colleague has either.
6.2 Conditional, Not Constant
The second piece of evidence about dark traits is more directly relevant to the only-with-you pattern because it suggests they are expressed selectively. Wisse and Sleebos studied supervisors in 225 teams and found that supervisors high in Machiavellianism were more abusive only when they believed their position gave them real power (Wisse & Sleebos, 2016). A 2023 paper, in a study of 355 leaders in the United States and a second of 273 leaders and 1,252 subordinates in China, described the abuse of Machiavellian leaders as cold and strategic, rising with their ties to their own superiors and when team members’ relationships became strong enough to threaten them (Feng, Keng-Highberger, Yam, Chen & Li, 2023).
An older experiment makes the same point outside work. People high in Machiavellianism behaved more generously when others were watching and pursued their own interests when they were not (Bereczkei, Birkás & Kerekes, 2010). A pattern of courtesy to the people who matter and hostility to those who do not is therefore consistent with what is known about this trait. It is also consistent with envy, selective incivility, and a two-sided spiral, and the pattern alone cannot tell you which.
I looked for a study showing directly that people high in any dark trait are courteous upwards and hostile downwards as a stable habit, the pattern sometimes called kiss up, kick down, and did not find one I could verify. The phrase is widely used, and the idea is plausible, and I would treat any confident claim that research has established it with caution. I would not repeat it as a finding.
6.3 Different Observers, Different People
Another finding helps explain why people around the target do not see what the target sees. In a study of people rated from several directions at work, narcissism was linked to worse ratings of interpersonal performance and integrity from supervisors and was unrelated to the ratings given by subordinates (Blair, Hoffman & Helland, 2008); whatever the direction, which in that study ran against the popular expectation that bosses are charmed and subordinates suffer, the point that matters here is that different observers of the same person saw different people.
A study of 203 corporate professionals in management development programmes found a related split within a single set of ratings: psychopathy was linked to higher ratings of charisma and presentation and lower ratings of responsibility and performance (Babiak, Neumann & Hare, 2010). Charm is visible to everybody, and what lies underneath it may be visible only to the people who deal with the consequences.
Neither of these is a study of one colleague singled out by another. They are reasons to expect that the people in 1.2 who find the controller helpful are not wrong about their experience, and that the target is not wrong about hers.
Chapter 7: How It Becomes Mutual
7.1 Tit for Tat
The part of this pattern that people on the receiving end are least likely to notice is their own part in it, and I say that with some care, because it is not an accusation. Andersson and Pearson argued in 1999 that incivility tends to spiral, with each discourtesy provoking a slightly sharper reply until the exchange becomes openly hostile (Andersson & Pearson, 1999). That was a theoretical argument, and the evidence since has largely supported its central mechanism. The mechanism is ordinary.
In a study that followed employees through two working weeks with several short surveys each day, incivility received earlier in a day reduced people’s self-control and was followed by more incivility given later the same day, and the drop in self-control was larger where the workplace was more political (Rosen, Koopman, Gabriel & Johnson, 2016). In a survey of 512 school employees in a Swedish municipality, the strongest single predictor of being uncivil to others was having been treated uncivilly by colleagues and supervisors (Torkelson, Holm, Bäckström & Schad, 2016). And in the three-sample pair study in 3.2, incivility given and received were linked within the pair itself (Taylor, Locklear, Kluemper & Lu, 2022).
7.2 The Short, Cold Emails
Return to the head of operations. By the end of the quarter, she was writing to the controller in shorter, colder messages than she wrote to anybody else and copying in her director, which is an understandable response and which, from his chair, reads as exactly the discourtesy and escalation he may believe he has been receiving from her since her project landed. Neither of them has to be a particularly difficult person for this pair to have become a difficult pair.
That is the practical meaning of the spiral finding. By the time the pattern is established, it is often no longer one person being difficult with another, and the one move entirely within the target’s control is to stop supplying the next turn of it, and Chapter 10 comes back to how.
Evidence status: evidenced as to reciprocity; inference as to the long-running pattern. A diary study across two working weeks, a municipal survey and a three-sample pair study. None followed a single pair over months, so the claim that a spiral sustains the pattern in 1.1 is my inference.
7.3 Two Accounts of the Same History
A classic study of how people remember conflict explains why both sides can feel wronged at once. When Baumeister, Stillwell and Wotman compared accounts written from the position of the person who caused anger and the person who felt it, those who had caused it tended to see their behaviour as meaningful and understandable and the incident as closed, while those on the receiving end tended to see it as arbitrary or incomprehensible and as part of a continuing series of harms (Baumeister, Stillwell & Wotman, 1990).
The accounts were not of the same events, so the study shows how the two positions tend to be remembered rather than who is right in any case. The implication for the pair in 1.1 is nevertheless direct. The target is likely responding to an accumulated history, and the colleague is likely seeing only the most recent message and viewing the response as out of proportion.
Chapter 8: Where the Evidence Is Thin
8.1 The Question Nobody Has Tested
The central question of this article is whether the target’s sense of being singled out is accurate, and I could not find a peer-reviewed study that tests it directly [NEEDS SOURCE: accuracy of perceived selective targeting at work]. Almost all of the research above relies on what targets say they experienced, and the handful of designs that also collect the other side’s account or an observer's account were not built to ask whether one person is treated worse than their peers by one particular colleague.
That matters because both directions are known. People overestimate how much others notice their own appearance and actions (Gilovich, Medvec & Savitsky, 2000), which is a finding about how conspicuous we feel rather than about being singled out, but which suggests caution about any feeling of being the focus of somebody else’s attention. By contrast, research on selective incivility and envy shows that some people really are singled out, for reasons they may be the last to suspect.
8.2 The Argument About Targets
A long-running, heated argument in this field is whether research on target characteristics amounts to blaming them. One study found that people lower in agreeableness, meaning less warm and cooperative, and higher in neuroticism, meaning more prone to anxiety and irritability, reported more incivility, partly through their own more provocative behaviour (Milam, Spitzmueller & Penney, 2009); Cortina, Rabelo and Holland argued that this approach invites victim-blaming, and proposed that research should look at the choices of the people doing the mistreating rather than at the people receiving it (Cortina, Rabelo & Holland, 2018).
Does the outcome of that argument change what the target in 1.1 should do? Mostly no, and for a reason I find reassuring. The pooled evidence in 6.1 finds the target’s personality only weakly related to what happens to them and the working context more closely related. So whichever side is right in principle, the practical advice points to the context and the pair rather than repairing the target.
Evidence status: contested as to target characteristics; no evidence base as to the accuracy of feeling singled out. The dispute is live and does not change the recommendations in Chapter 10. The absence of a direct test is itself the finding at 8.1.
PART III. BEFORE YOU USE ANY OF THIS
Chapter 9: The People You Will Mislabel
Everything up to here describes a pattern and several reasons for it. This chapter is about people who produce something that looks like the pattern for reasons that have nothing to do with them, and one of those people is the reader. If this article makes you more confident about what your colleague is, it has failed.
9.1 The One Whose Job Is to Say No to You
Some roles are structurally difficult with particular colleagues, and a financial controller whose sign-off is required on every change to a process is one of them. If the head of operations in 1.1 is the only person whose work regularly needs his approval to spend or change something, he may be slower and shorter with her because she generates the most work and risk for him, not because of anything about her or him. The test is whether his treatment of other people in her structural position, or who held it before her, looks the same.
9.2 The One Responding to Something You Did
The spiral in Chapter 7 means that by the time a target notices the pattern, the colleague may be responding to something real in the target’s own behaviour, and the target may not remember it or may regard it as trivial. A project that overturned somebody’s three years of work, a meeting in which they were corrected in front of their own director, a message copied to a senior person without warning: any of these can start a pair on its way, and the evidence in 7.3 suggests the person who started it will usually remember it less clearly than the person it was done to.
9.3 The One With a Grievance You Have Not Heard
Sometimes the colleague is behaving badly and also has a point. The pattern in 4.2 shows supervisors turning hostile towards subordinates they see as different or as performing poorly, and a peer can do the same towards somebody they believe was promoted over a better candidate or given a project that should have been theirs. A grievance does not justify the treatment. It does change what will end it, because a pattern driven by an unspoken complaint rarely ends until the complaint is heard.
9.4 The One Having a Hard Time
Somebody under heavy strain may have the capacity to be courteous with most people and not with the one colleague who adds the most to that strain. The previous article in this series made the same point about the anxious, reactive pattern of difficult behaviour, which in my experience gets worse under exactly the pressure organisations apply to fix it. Ask when it started, and whether it started with something in the colleague’s life rather than in yours.
9.5 The One Who Needs a Clinician
This boundary is categorical, and the whole series holds it. Machiavellianism, narcissism and psychopathy as studied at work are traits on which everybody sits somewhere, and none of them is a diagnosis. Telling a personality disorder or any other clinical condition apart from ordinary difficult behaviour is the work of a qualified clinician who has assessed the person, never of a colleague who has watched them in meetings. An observation is not a diagnosis, and nothing in this article can be turned into one.
9.6 And You
The last person worth considering is the reader, and I include myself in that. The feeling of being singled out is powerful and partly self-sustaining, because once it forms, every short reply confirms it and every courteous one is noticed less.
Nothing in 8.1 says the feeling is usually wrong. It says nobody has tested how often it is right, which is a reason to check it before acting on it rather than a reason to dismiss it. So check it.
PART IV. WHAT TO DO
Chapter 10: Check the Pattern Before You Name It
Evidence status: mixed and stated per item. Some items rest on field evidence about group conditions that reduce envy-driven mistreatment and uneven treatment. The rest are reasoned from the evidence or drawn from practice. None has been tested as an intervention with one person singled out by one colleague. Each item carries its own status.
10.1 Test Only With Me Before You Say It
The first move is to find out what the pattern feels like, and it costs nothing but a few weeks of attention. Keep a dated record of the specific behaviour rather than your reading of it: the request, the date, the reply deadline, what the reply said, who else was copied, and what happened in the meeting. Then, where you can see it without asking anybody, notice how the same colleague treats two or three people in comparable positions.
That record answers three questions the evidence says matter.
Whether the treatment really is concentrated on you (Chapter 3). Whether it follows a line of sex, ethnicity, age or similar, which changes the route entirely (Chapter 5). And when it started, which is the single best clue to whether it is about performance, a grievance, strain or your own half of a spiral (Chapters 4, 7 and 9). Practitioner judgement, reasoned from the evidence above.
A note on asking colleagues. Observers of mistreatment do not simply see it; they interpret it, and coworkers who believe a target deserved abusive treatment tend to feel contentment and to exclude the target, while those who believe it undeserved tend to feel anger and support them (Mitchell, Vogel & Folger, 2015), so asking a colleague whether they have noticed is also asking them to take a view about whether you deserved it, and it is worth knowing which view they are likely to take before you ask.
Observers are not neutral in the other direction either. In two small experiments, people who watched incivility tended to respond negatively to the person who instigated it, and that response was driven by their own negative feelings towards the instigator (Reich & Hershcovis, 2015). Witnesses judge in both directions, and that is one more reason the dated record is worth more than anybody’s impression, including yours.
10.2 Stop Supplying Your Half
The one move entirely within the target’s control is the one the spiral evidence points to most directly. If incivility received earlier in a day is followed by incivility given later the same day, partly through reduced self-control (Rosen, Koopman, Gabriel & Johnson, 2016), then the practical rule is to write the reply to a difficult colleague at a moment of your choosing rather than the one they chose, and to write it in exactly the register you use with everybody else.
For the head of operations, that means one specific thing: her messages to the controller match the length, warmth, and copy of her messages to every other head of function, even though his to her have not.
That is not appeasement, because it removes the one piece of the pattern he can currently point to and gives the record in 10.1 a clean baseline. In a scenario study of 296 employees, more agreeable people were less likely to return perceived incivility (Lee, Steigerwald, Jex, Rada-Bayne & Holden, 2024), which suggests the choice is at least partly available to anybody. Evidenced by reciprocity within a day; practitioner judgement on the remedy.
10.3 Change the Structure of Contact, Not the Temperature
Where the pattern is driven by envy or competition, the evidence points at the group rather than the pair. Envy-driven mistreatment of high performers was weaker where people identified strongly with their work group (Kim & Glomb, 2014), envy-driven undermining was weaker where people identified with their colleagues and where undermining was not the team norm (Duffy, Scott, Shaw, Tepper & Aquino, 2012), and harming of better performers fell when team goals were more cooperative (Lam, Van der Vegt, Walter & Huang, 2011).
Translated into what a person can actually change, that suggests shared credit rather than solo credit, a joint objective with the difficult colleague rather than a sequence of approvals, and making their contribution visible in work that is also yours. The head of operations could, for example, present the next phase of the claims project jointly with the controller, with his cost analysis named as the basis for it. Evidence of group identification and cooperative goals reduces envy-driven mistreatment; inference about the specific moves.
10.4 If You Manage Them Both
A manager told about this pattern tends to do one of two things: dismiss it because the colleague has always been fine with them, which 4.3 explains, or call both parties in to sort out what gets called a personality clash. The evidence suggests a different first move. Leaders who form visibly closer relationships with some team members than others harm the harmony of the whole group (Yu, Matta & Cornfield, 2018), and being treated worse than one’s peers lowers performance (Xiao, Togola, Zhang & Yu, 2024), so the first question is whether the manager’s own pattern of attention is feeding the comparison.
The second is whether the incentives put the two in competition, since 10.3 applies more to the manager’s design choices than to anyone else’s. The third is the record: the manager should review it before deciding what kind of problem this is, and specifically whether the treatment follows a line. Evidence of uneven treatment by leaders; practitioner judgement on the sequence.
10.5 When It Follows a Line, Change the Route
If the record in 10.1 shows the treatment falling on people of one sex, ethnicity, age or other protected characteristic, the relationship moves in 10.2 and 10.3 are no longer the right first moves. Selective incivility is a form of discrimination (Cortina, Kabat-Farr, Leskinen, Huerta & Magley, 2013), and asking a target to build rapport with the person treating them worse for that reason places the cost of somebody else’s conduct on the person least responsible for it. Take the record to whoever in the organisation handles conduct, and take a qualified adviser’s view on how the law applies to your situation, which this article does not give. Evidenced as to the pattern; the route is practitioner judgement.
10.6 Three Refusals
Do not tell anybody what they are. A colleague who is told they are manipulative, narcissistic or a bully will deny it with complete confidence, and the conversation moves from what happened to whether you were entitled to say it; the record in 10.1 supports a different sentence: on the fourth, the eleventh and the nineteenth my requests took five days, the same requests from other heads took under a day, and the claims change has now missed its window. That is true whatever is going on inside the person. Practitioner judgement.
Do not treat your own performance as the thing to fix. If the treatment is driven by envy of good work (Kim & Glomb, 2014; Li, Law, Zhang & Yan, 2024), doing worse will not end it and doing less visible work will cost you more than it costs them. Change how the credit is shared, never how good the work is. Evidence of envy of high performers; practitioner judgement on the remedy.
Do not frame a line as a clash. A manager who responds to selective incivility by mediating a personality clash has treated a conduct problem as a relationship problem and asked the target to carry half of it. Practitioner judgement, reasoned from Chapter 5.
10.7 A Worked Version
Here is what the sequence looks like for the head of operations over one quarter, and none of it requires her to decide what the controller is. For four weeks she keeps the record, and it shows that his replies to her average about five days and his replies to two peers who also need his sign-off average under one, and that the pattern began in the week her first project was praised at the board. It follows no line she can see.
That points at envy or a grievance rather than at a fixed habit. So she rewrites her own half first: normal length, normal warmth, normal copy list. Then she changes the structure, proposing that the next phase of the claims work carries a joint objective, with his cost analysis named as its basis and both names on the board paper.
If the record changes, the problem was largely the pair and the stakes around it. If it does not, she now has a dated, specific account of what happened, which her director can act on without anybody having to be labelled, so either outcome is information. Practitioner judgement throughout; the sequence has not been tested.
Chapter 11: What This Changes
11.1 The Short Version
If you read nothing else in this article, I would ask you to read this. Mistreatment at work is common and it is aimed. About three in four employees have experienced incivility, and across 57 studies what predicts aggression depended on who it was aimed at. Some of it belongs to the pair. In three samples, how uncivil coworkers were to each other varied substantially from pair to pair beyond each person’s general tendency, and across seven studies of social interaction, almost nobody drew the same response from everybody.
Good work is one of the things it finds. Among 4,874 university staff in 339 groups, higher performers were more likely to be mistreated, and a second study traced that to envy. The dark traits sit with the people doing it. In a study of 172 employees, the people who bullied scored high on Machiavellianism and psychopathy and the people bullied did not; in 225 teams, Machiavellian supervisors were more abusive only when they felt powerful. It becomes mutual. Incivility received in the morning was followed by incivility given in the afternoon, and by the time a pattern is noticed, both people may be supplying it.
11.2 So What Do You Do
Four things, and none of them requires you to decide what anybody is.
Check the pattern before you name it, with a dated record and a look at how the same colleague treats people in positions like yours. Stop supplying your half, by writing to the difficult colleague exactly as you write to everybody else. Change the structure of contact rather than the temperature, through shared goals and shared credit. And if it follows a line, change the route, because that is a conduct matter rather than a relationship.
11.3 Why Any of This Matters
Two reasons, and the second is the one I would leave you with. The first is practical. The two conclusions people usually reach, that the colleague is a type or that the target is the problem, both lead to actions that tend to make the pattern worse: a confrontation about character, which the target loses, or a retreat into doing less, which rewards the envy that may have started it.
The second is about who pays. This pattern is almost invisible to everybody except the person on the receiving end, and the evidence in Chapter 4 suggests that person is disproportionately likely to be doing good work. An organisation that treats every report as a personality clash or oversensitivity is, on this evidence, quietly taxing some of its best people for being good at their jobs.
11.4 What I Still Do Not Know
Four things remain open, and I would rather state them than resolve them by omission. Whether the feeling of being singled out is usually accurate, since no study I could find tests it directly. How large the share of workplace incivility that belongs to the pair actually is, since I have confirmed that it is substantial, not just its size.
Whether any of the moves in Chapter 10 ends the pattern once it is established, since none has been tested with one person singled out by one colleague. And whether the dark traits predict aiming hostility selectively, as distinct from predicting bullying in general, since the most direct evidence concerns supervisors and power rather than peers and targets.
A closing note on how I have handled the material. This is the pattern in which the temptation to name a colleague’s personality is strongest, because the behaviour is so consistent and so personal, and it is also the pattern in which that temptation is least useful, because the evidence is almost entirely about pairs, groups, and conditions. What I would most like to have made harder is the confident sentence about what your colleague is, and what I would most like to have made easier is the dated record of what they do.
