How do you deal with passive aggressive people?

Someone in your life keeps agreeing to things and not doing them, or saying “fine” in a way that is clearly not fine, and you cannot work out whether you are being managed or whether you are inventing it. I co-founded Subtext, and messages people cannot decide about are most of what I look at.

The first thing worth knowing is how unstable the ground is. Passive aggression started as a wartime military description, spent decades as a formal psychiatric diagnosis, and was removed from the diagnostic manual altogether in 2013. There is no widely validated standalone scale for it in adults. And the confident advice you have read about how to handle it is almost entirely untested.

That does not mean the behaviour is not real. It means most of what gets written about it is opinion wearing a lab coat.

It was a diagnosis, and then it wasn’t

The term came out of a US War Department bulletin drafted in 1943, describing soldiers who resisted authority indirectly through inefficiency, procrastination and passive obstruction rather than open defiance¹. That document was the direct ancestor of the first diagnostic manual.

It survived as a diagnosis for decades. In the 1980 edition it carried a feature unique among the personality disorders: clinicians could only diagnose it if the patient met criteria for no other personality disorder. By 1994 it had been renamed negativistic personality disorder and moved to an appendix for further study. By 2013 it was gone entirely, surviving only as an unspecified personality disorder or through trait facets.

Theodore Millon, who championed the renaming, gave five reasons for the move²: insufficient acceptance in the clinical literature, content too narrow and behavioural, behaviour too situational to reflect a syndrome, a term implying a motivation at a time when criteria were meant to be purely descriptive, and too much overlap with other disorders.

The empirical case was strong. In 1,158 psychiatric outpatients, internal consistency of the seven diagnostic items was low at alpha 0.50, only 3 per cent met criteria, and almost 90 per cent of those had at least one other personality disorder³. A separate study of 1,215 undergraduates found the correlation between the older passive-aggressive construct and the newer negativistic one was only .52, lower than the correlation between negativistic and borderline personality disorder².

The fourth of Millon’s reasons is the one that matters for anyone trying to decide what is happening in their own life. Diagnostic criteria that omit motive cannot distinguish passive aggression from disorganisation or inattention, because the whole concept depends on the behaviour meaning something. Missing a deadline counts as passive aggression only if it was intended to communicate. There is no way to read that off the behaviour.

Which is exactly the position you are in when you look at the message. It is also why Subtext will tell you what a message conveys and refuse to tell you what someone meant by it. The second question is not answerable from the words, and the diagnostic manual gave up on it for the same reason.

What the research studies instead

Three replacement constructs, all self-report or peer-report, all measuring perceived behaviour rather than verified acts.

Negativistic personality, from Millon’s model, describing conflict between dependency on others and the desire for self-assertion. Millon proposed subtypes; these are theoretical taxonomy and were never validated as distinct types, so treat them as a framework rather than findings.

Covert hostility, the internalised side of unexpressed anger and cynicism. Note that the widely used Buss-Perry Aggression Questionnaire has no passive-aggression subscale. Its four factors are physical aggression, verbal aggression, anger and hostility.

Indirect aggression, meaning behaviour intended to harm while concealing intent, avoiding counterattack. Measured through several scales originally developed with Finnish schoolchildren using peer nomination.

The honest summary: there is no single widely validated modern scale for adult passive aggression. When you read a statistic about how many people are passive aggressive, ask what instrument produced it.

At work, one thing has actually been measured

Most of the workplace discourse rests on a vendor survey. One study measured the thing directly.

Yuan, Park and Sliter separated active email incivility, meaning put-downs and demeaning content, from passive email incivility, meaning being tuned out, ignored, having replies withheld⁴. A survey of roughly 233 working US employees supported the distinction empirically, and a diary study found passive incivility was positively associated with insomnia, which then predicted heightened negative affect at the start of the workday.

The finding that explains the whole phenomenon: active incivility drove a stronger emotionality appraisal, while passive incivility was appraised as more ambiguous. Participants reported not knowing what it meant.

That ambiguity is the harm. Because uncivil acts of this kind are deniable, targets burn cognitive resources decoding whether the person was malicious or merely careless, and that decoding cost is what you take home. It is also why an openly rude email is easier to recover from than three words that might mean nothing, and why resolving the ambiguity quickly matters more than resolving it correctly.

Correlational and diary-based, so it does not establish that a terse email causes insomnia. It does establish which category of message people cannot put down.

The broader incivility literature is real: a survey of 1,180 US public-sector employees found 71 per cent reported experiencing incivility over five years, with instigators disproportionately people holding organisational power⁵. Be sceptical of the widely quoted 98 per cent figure, which comes from a practitioner magazine article rather than peer review.

The markers will not settle it

Only one written marker has solid experimental backing, and it is the full stop on a one-word reply. Emoji have a documented ambiguity problem rather than a hostility problem. Everything else people list, meaning “noted”, “per my last email”, “sure”, “K”, delayed replies, ellipses and the thumbs-up, has no direct peer-reviewed evidence as a marker of hostility.

I have written that up properly, with the studies and their limits, in what a full stop actually signals and is that email passive aggressive, or just short. The short version for the purposes of this post is that the markers will not settle the question for you, which is why the rest of this article is about the person rather than the punctuation.

Are you over-reading it?

This is a live disagreement and I am not going to pretend otherwise.

The over-reading case. Kruger and colleagues found across five experiments that senders substantially overestimate their ability to convey tone in writing, because they hear their own intended prosody while composing and cannot decouple from it⁶. Sillars and Zorn found receivers rated emails more negatively than uninvolved observers did⁷. Hostile attribution bias and the illusion of transparency both predict over-attribution of hostile intent under ambiguity.

The counter-case, and it is strong. Pollmann and Roos asked both parties⁸. Participants uploaded a message they had actually received and rated its warmth; the person who wrote it was then asked what they had intended. Across 347 receivers with 171 matched senders for texts, and 361 receivers with 61 matched senders for work emails, sender and receiver ratings aligned well. Six moderators, including message length and emoji, changed nothing.

The plausible reconciliation, presented as a reading rather than a fact: the over-reading studies use lab settings, constructed stimuli, strangers, and tasks like detecting sarcasm in isolated items. Pollmann and Roos used real messages between people with existing relationships. Context and history may supply exactly the disambiguating information a lab strips out.

Two flags on the counter-case. The venue is a companion open-access title rather than the parent journal, and the study is recent and not yet replicated.

What this means for you. In lab conditions and between strangers, tone travels unreliably and skews negative. Between people who know each other, in real exchanges, it travels better than the folklore suggests. So if this is a colleague you barely know, your uncertainty is well founded and Subtext will be more use to you than it is to someone reading a partner’s message. If it is someone you have exchanged thousands of messages with, your reading is probably closer than you think, in either direction.

That gap is where a tone-checking tool is actually worth something, and where I would be careful about overclaiming. Subtext can tell you what readings a message supports and how far apart they are. It cannot tell you what someone meant, and the research above is the reason I will not pretend otherwise.

Free to start, on your phone or in the browser.

Does naming it help?

There is no controlled experiment testing whether calling someone’s behaviour passive aggressive reduces it. All three research passes behind this article agree. Anyone giving you a confident script here is giving you clinical practice, which is not nothing, but it is not evidence.

What is testable, and has been tested, is narrower.

Rogers, Howieson and Neame had participants read hypothetical conflict-opening statements varying on I-language against you-language, and on whether the speaker communicated awareness of the other person’s perspective⁹. Both I-language and communicating perspective reduced how defensively people expected the recipient to react.

The counterpoint matters. Korobov analysed young adult couples arguing in natural contexts and found arguments were often initiated with I-language feelings talk, and that rather than defusing conflict it was met with deflection¹⁰. Only when the I-language demonstrated awareness of the other person’s feelings did it promote alignment.

Read together: I-statements reduce anticipated hostility when people rate hypothetical statements, and perform less impressively in observed real arguments unless they carry genuine perspective-taking. That is a lab-versus-field gap, and it suggests the perspective-taking half is the active ingredient rather than the grammatical construction.

So the defensible version of the advice is: describe the specific behaviour rather than the trait, say what it did to you, make a concrete request, and show you have understood their side. Not because a trial proved it, but because the one thing that separated success from failure in real arguments was demonstrated awareness of the other person.

On assertiveness training generally, a review found a large but dated literature supporting it, mostly peaking in the 1970s and 80s, with modern high-quality research diminished drastically.

Where this stops being about style

An irritating communication style, a chronic withdrawal pattern, and a course of controlling behaviour are three different things, and the third one is not a communication problem.

The distinction turns on pattern, transparency, power and effect. Silence and indirection used as punishment, in a repeating cycle where the only thing that ends it is your capitulation, and which produces hypervigilance and changed behaviour in you, is a different phenomenon from someone who is bad at saying what they want.

In England and Wales, controlling or coercive behaviour in an intimate or family relationship has been a criminal offence since the Serious Crime Act 2015, where the behaviour is repeated or continuous between people who are personally connected and has a substantial adverse effect on day-to-day activities.

Do not pathologise the first thing while naming the second. Most passive aggression is exactly what it looks like: someone who finds direct conflict harder than indirect resistance. If what you are living with is the other thing, the wording of your next message is not the problem to solve, and someone outside the situation is worth talking to.

Sources

Numbered in the order they appear above. Where a figure could not be verified against the primary source, the text says so.

1. War Department Technical Bulletin, Medical 203 (drafted 1943, issued 1945). The direct ancestor of the first Diagnostic and Statistical Manual. Described here via the secondary account in source 2; primary text not directly consulted.
2. Hopwood, C. J., and Wright, A. G. C. (2012). A comparison of passive-aggressive and negativistic personality disorders. Journal of Personality Assessment, 94(3), 296 to 303. 1,453 undergraduates sampled, 1,215 retained. Funded in part by NIMH grant F31MH087053. https://www.tandfonline.com/doi/abs/10.1080/00223891.2012.655819
3. Rotenstein, O. H., and colleagues (2007). A cross-sectional evaluation of the DSM-IV criteria for passive-aggressive personality disorder. Journal of Psychiatric Research. 1,158 psychiatric outpatients. Funder not stated in the accessible record. https://www.sciencedirect.com/journal/journal-of-psychiatric-research
4. Yuan, Z., Park, Y., and Sliter, M. T. (2020). Put you down versus tune you out: Further understanding active and passive email incivility. Journal of Occupational Health Psychology, 25(5), 330 to 344. Support acknowledged in connection with CDC and NIOSH; exact grant line needs confirming. https://psycnet.apa.org/record/2020-30563-001
5. Cortina, L. M., Magley, V. J., Williams, J. H., and Langhout, R. D. (2001). Incivility in the workplace. Journal of Occupational Health Psychology, 6(1), 64 to 80. 1,180 US public-sector employees. The description of the exact population conflicts across secondary sources and needs verifying against the paper. https://psycnet.apa.org/record/2001-16943-006
6. Kruger, J., Epley, N., Parker, J., and Ng, Z.-W. (2005). Egocentrism Over E-Mail. Journal of Personality and Social Psychology, 89(6), 925 to 936. The specific accuracy percentages differ by study within the paper and conflict across secondary sources; none are quoted here. https://web-docs.stern.nyu.edu/pa/kruger_email_ego.pdf
7. Sillars, A., and Zorn, T. E. (2021). Hypernegative Interpretation of Negatively Perceived Email at Work. Management Communication Quarterly, 35(2), 171 to 200. https://journals.sagepub.com/doi/10.1177/0893318920979828
8. Pollmann, M. M. H., and Roos, C. A. (2025). “I get u”. People correctly interpret the tone of text messages and emails. Computers in Human Behavior Reports, 18, article 100689. A companion open-access title; recent and not yet replicated. https://doi.org/10.1016/j.chbr.2025.100689
9. Rogers, S. L., Howieson, J., and Neame, C. (2018). I understand you feel that way, but I feel this way: the benefits of I-language and communicating perspective during conflict. PeerJ, 6, e4831. Vignette study, 253 students, Edith Cowan University. No external funding. https://peerj.com/articles/4831/
10. Korobov, N. (2020). Failure of I-statements for Mitigating Interpersonal Conflict in Arguments Between Young Adult Couples. Studies in Media and Communication, 8(2), 49 to 60. Qualitative, 20 couples, roughly 140 hours of home recordings; the 61 per cent figure is a descriptive count without inferential statistics. Author-fee open-access venue. https://redfame.com/journal/index.php/smc