TL;DR. The idea of recognizing a manipulative person “by their face” is weak: in a meta-analysis combining 206 studies and 24,483 judges, people could tell lies from truths with only 54% accuracy. The sturdier route is to look at behavior: in a classic study of 59 couples, the most frequently reported influence tactic was reason; coercion and the silent treatment stayed at around 2 out of 7. On about 11,000 shopping websites, 1,818 dark patterns were found; in one experiment, mild patterns raised the acceptance rate for a dubious service from 11% to 26%, and aggressive ones to 42%. In a 900-person debate experiment, GPT-4 with access to personal information was more persuasive than the human in 64.4% of the debate pairs where AI and humans were not equally persuasive; the September 2026 correction reported that the added contribution of personalization is not settled. CEO move: do not decide under pressure; slow the decision down, verify the claim, and write down the boundary and the escalation rule in advance. Student move: learn the tactic catalog the way you would learn reasoning errors, and log the cases.
This article can be read alongside the pieces on first-conclusion bias, the attention economy’s new tools, trust calibration and the Barnum effect.
What is manipulation, and how does it differ from legitimate persuasion?
The Stanford Encyclopedia of Philosophy entry, updated on March 31, 2026, says manipulation is usually treated as a form of influence distinct from both coercion and rational persuasion, and sorts the definitions into roughly three clusters: influence that bypasses reason, a form of trickery, and a form of pressure. There is no single criterion.
Psychology uses the word more neutrally. Buss and colleagues (1987) define manipulation as the tactics people intentionally use to influence their environment, and state explicitly that this does not require malicious intent. Influencing others is ordinary in working life: Kipnis and colleagues (1980) extracted 370 tactics in 14 categories from incidents described by 165 lower-level managers, and found eight dimensions in a 58-item questionnaire (754 respondents in total, CEOtudent calculation): assertiveness, ingratiation, rationality, sanctions, exchange, upward appeals, blocking and coalitions. Yukl and Falbe (1990) reported that the most frequently used tactics were consultation and rational persuasion. In the meta-analysis by Higgins and colleagues (2003), ingratiation and rationality were positively related to work outcomes.
The right question is therefore not “is this person a manipulator?” but “by what route is this attempt changing my decision?” The article uses three criteria (CEOtudent editorial framework): would the method still work if it were stated out loud (transparency), is saying “no” as easy as saying “yes” (symmetry), and is there time to verify before deciding (time).
Which influence tactics does the research document?
Buss and colleagues (1987) first asked 67 students what they do to get their partners to do something or to stop doing something. Then 59 couples who had been together for at least six months (118 students) rated 35 behaviors from 1 (not at all likely) to 7 (extremely likely); each person rated both themselves and their partner. Factor analysis yielded six tactics.
Table 1. Buss and colleagues (1987): mean score per item for each tactic (1-7)
| Tactic | Sample item (summarized) | To get them to do it: self / partner | To get them to stop: self / partner |
|---|---|---|---|
| Reason | Explaining why they should do it | 5.31 / 5.38 | 5.28 / 5.05 |
| Charm | Paying compliments, giving a small gift before asking | 3.13 / 3.11 | 2.51 / 2.56 |
| Regression | Pouting, sulking | 2.17 / 2.27 | 2.39 / 2.28 |
| Coercion | Demanding, yelling, threatening | 1.81 / 1.71 | 2.40 / 2.35 |
| Silent treatment | Not responding until they do it, ignoring them | 1.79 / 1.89 | 2.04 / 2.47 |
| Debasement | Lowering oneself so that they will do it | 1.44 / 1.35 | 1.42 / 1.27 |
Source: Buss and colleagues (1987), Table 2; figures are as printed.
The most frequently reported tactic was the most innocent one: in every condition the reason score was more than twice the coercion score (CEOtudent calculation). The tactic varied with the goal: charm was used more to get someone to do something, coercion and the silent treatment more to get someone to stop. The person using a tactic and the person on the receiving end did not see the same thing: the average agreement between the two sides’ reports was 0.42 in the “do” condition and 0.25 in the “stop” condition. The sample consists of dating university students; it cannot be generalized directly to the workplace. Buss (1992) expanded the list to twelve tactics; one of them, “hardball”, is a mix of threats, lies and violence and falls under the scope note.
Table 2. Catalog of eight influence tactics (CEOtudent editorial framework)
| # | Tactic | How it sounds | What it leans on | Countermove |
|---|---|---|---|---|
| 1 | Charm and ingratiation | “Nobody would do this better than you.” | The desire to be liked | Separate the compliment from the request |
| 2 | Silent treatment | An unanswered message, being ignored in a meeting | Intolerance of uncertainty | Do not chase; restate the request in writing, with a date |
| 3 | Coercion and insistence | “Say yes right now, or else…” | Conflict avoidance | Respond to the request, not the tone; postpone the decision |
| 4 | Sequential requests | A big request after a small favor; “then at least this” after a refused request; terms that change after acceptance | Consistency, reciprocal concessions | Evaluate every request from scratch |
| 5 | Making reality debatable | “Nothing like that was discussed, you are exaggerating.” | Trust in someone who is close or senior | Keep a written record; compare with a third source |
| 6 | Artificial scarcity, time limits and crowds | “Only 3 left in stock”, a countdown, “24 people are viewing this listing” | Fear of missing out, herd effect | Check the offer again the next day |
| 7 | Hiding, obstruction, trick questions | A fee in small print, hard cancellation, a double-negative question | Limits of attention, sticking with the default | Find the total cost and the cancellation step before confirming |
| 8 | Personalized persuasion | An argument tailored to age, job, views | Personal data, fluent language | Test the argument independently of who it was written for |
Do sales and persuasion techniques really work?
Table 3. Three classic compliance techniques and the later evidence (published findings)
| Technique | First study | Later evidence |
|---|---|---|
| Foot-in-the-door: a small request first | Freedman and Fraser (1966), 2 experiments | Gamian-Wilk and Dolinski (2020): in meta-analyses the effect is small and the effect size of the first study was never obtained again; in replications it was effective in Ukraine and ineffective in Poland in 2003, and non-significant in both in 2013 |
| Door-in-the-face: a large request first | Cialdini and colleagues (1975), 3 experiments | Feeley and colleagues (2012): r = 0.126 for verbal compliance (k = 78), r = 0.052 for actual behavior (k = 39, non-significant). Genschow and colleagues (2021): in a direct replication, compliance rates were similar to those of 45 years earlier |
| Low-ball: raising the cost after acceptance | Cialdini and colleagues (1978), 3 experiments | Burger and Caputo (2015): reliable and effective; strongest when the initial acceptance is stated in front of others and the second request is only slightly more costly |
Source: abstracts of the publications named; figures are as printed.
The techniques are real, but the effects are conditional and mostly small: r = 0.126 corresponds to about 1.6% of the variance in compliance (CEOtudent calculation), and a verbal “yes” does not always turn into behavior. The frequently cited rates from the first experiments were not used; only the abstracts of those papers could be accessed.
The website of Robert Cialdini’s training company lists seven principles: reciprocity, scarcity, authority, consistency, liking, social proof and unity. The list is a useful classification; it is not seven separately measured effect sizes, and the source is commercial.
Can someone who behaves manipulatively be recognized by their face or character?
Most of the old article’s tips relied on “reading” the other person. Bond and DePaulo (2006) combined 206 studies and 24,483 judges: without special aids or training, people told lies from truths with an average accuracy of 54%, correctly classifying 47% of lies and 61% of truths. That is 4 points above a coin flip (CEOtudent calculation). According to the review by DePaulo and colleagues (2003), which combined 1,338 estimates for 158 cues, liars are less forthcoming in some respects, but many behaviors are unrelated or only weakly related to lying.
Personality research does not license labeling an individual either. Paulhus and Williams (2002) compared measures of Machiavellianism, subclinical narcissism and psychopathy in 245 students; the three correlated with one another between 0.25 and 0.50, and the authors treated them as overlapping but distinct constructs. These are group-level patterns in questionnaire scores; they are not a tool for diagnosing a manager, a colleague or someone close to you. In Buss’s study, those who used a tactic a lot were also on the receiving end of more tactics from their partners: the unit of analysis is the interaction, not the person.
What does the research say about gaslighting?
Sweet (2019, abstract) defines gaslighting as a type of psychological abuse aimed at making the victim seem and feel “crazy”, and argues that the phenomenon is rooted in social inequalities and power-laden intimate relationships. The theoretical framework by Klein, Wood and Bartz treats it as a form of manipulation that over time leads a person to doubt their sense of reality, and spells out two limits: there is very little empirical work on the topic, and whether the behavior is deliberate is debated. For working life, Kukreja and Pandey (2023) developed a 12-item scale in three samples with a total of 679 employees from India and the United States; the scale focuses on the supervisor-subordinate relationship and has two dimensions: trivialization (7 items) and affliction (5 items).
The practical upshot: instead of the diagnosis “I am being gaslit”, record observable behavior: what was said, when, and which written record it contradicts.
Scope note. Coercion that has become persistent, threats, violence or abuse are outside the scope of a thinking guide. In such a situation, the right route is the professional support services and emergency lines where you are. This article is for general information; it is not psychological or legal advice.
Which designs steer decisions in sales and on screen?
Mathur and colleagues (2019) crawled about 53,000 product pages on about 11,000 shopping websites: they found 1,818 dark patterns of 15 types in 7 categories on 1,254 sites (about 11.1%), deceptive practices on 183 sites, and 22 third parties offering these patterns as a ready-made service. The authors treat the figures as a lower bound; the method looked only at text-based interfaces.
Table 4. Distribution of the 1,818 dark patterns by category (CEOtudent analysis, Mathur and colleagues 2019, Table 1)
| Category | Example | Count | Share |
|---|---|---|---|
| Scarcity | Low-stock and high-demand messages | 679 | 37.3% |
| Urgency | Countdown timer, limited-time message with no clear deadline | 481 | 26.5% |
| Social proof | “Someone just bought this” notification | 325 | 17.9% |
| Misdirection | Button text that shames the refusal, preselecting the expensive option | 270 | 14.9% |
| Obstruction | Easy sign-up, hard cancellation | 31 | 1.7% |
| Sneaking | Adding items to the cart, hidden subscription | 26 | 1.4% |
| Forced action | Forcing account creation to complete the transaction | 6 | 0.3% |
Counts are the sums of the type counts in the paper’s Table 1; shares are a CEOtudent calculation.
Scarcity and urgency together account for 1,160 instances, or 63.8% of the total: most of the patterns produce time pressure, not information. Of 393 countdown timers, 157 (39.9%) were deceptive: when the time ran out, the timer reset or the offer stayed valid.
Luguri and Strahilevitz (2021) ran two experiments with samples weighted to be representative of US adults. According to the co-author’s presentation to California’s privacy agency, the acceptance rate for a dubious identity-protection service was 11% in the control group, 26% with mild patterns and 42% with aggressive patterns: 2.4 and 3.8 times as high (CEOtudent calculation). The aggressive design made 11.1% of participants abandon the survey, the mild design 1.5%: the mild version, which draws no backlash, is the more insidious one. In the second experiment, with 3,777 people, the price difference between $8.99 and $38.99 a month did not significantly change the acceptance rate (19.8% versus 17.3%). A double-negative question got 33.4% of respondents to accept the service; half of them did not realize they had accepted.
Putting the two datasets side by side reveals a mismatch (CEOtudent analysis): according to the paper’s abstract, the most effective patterns were hidden information, trick questions and obstruction, and “act now” messages did not increase purchases; yet the most common ones in the crawl were precisely the scarcity and urgency messages. Because the crawl looked only at text, the comparison should be read with caution; the practical conclusion is still clear: attention should go to the total cost and the cancellation path, not to the timer. The September 2022 staff report of the US Federal Trade Commission notes that the patterns have a stronger effect when they are used together.
How is AI changing persuasion?
Table 5. AI and persuasion: published findings and later records
| Study | Design | Published result | Limitation or correction |
|---|---|---|---|
| Matz and colleagues (2024) | 4 studies, 1,788 participants in total; messages ChatGPT wrote to match a psychological profile | Personalized messages more effective than non-personalized ones | Self-report measures only |
| Salvi and colleagues (2025) | 900 participants in the US, 10-minute debate; opponent a human or GPT-4, with or without personal information | GPT-4 with access to personal information more persuasive than the human in 64.4% of the debate pairs where AI and humans were not equally persuasive; an 81.2% increase in the odds of agreeing more with the opponent | Author Correction of September 3, 2026: the difference between personalized and non-personalized GPT-4 is not significant (48.7%; P = 0.07; 0.04 in the original publication). The difference between GPT-4 without personal information and the human is also not significant (P = 0.30) |
| Costello, Pennycook and Rand (2024) | 2,190 people who believed a conspiracy theory, evidence-based dialogue with GPT-4 Turbo | About a 20% reduction in belief; the effect persisted 2 months later | Science published an Editorial Expression of Concern on June 11, 2026; according to the announcement, the corrected analysis was being evaluated by the journal |
| Roozenbeek and colleagues (2022) | 5 videos of 1.5 minutes each; 6 experiments (6,464 people) and a YouTube field study (22,632 people) | Effect size for recognizing the technique between 0.28 and 0.68; in the field, correct recognition rose by about 5% | Measured within 24 hours; US samples; the videos were produced in partnership with Google Jigsaw |
Source: the publications named, the Author Correction and the AAAS announcement; figures are as printed.
In the Salvi study, the personal information given to the model was limited: gender, age, ethnicity, education level, employment status and political orientation. After the correction, the supported claim is narrower: GPT-4 with access to personal information was more persuasive than the human opponent, but how much of that advantage came from personalization is not certain. According to the AAAS announcement, the authors of the Costello paper reported an inconsistency in how the exclusion criteria were applied and extra rows in the published dataset caused by a code-merging error; according to the authors, the corrected analysis gives the same result in direction, significance and magnitude. The notice itself could not be accessed; the figures are therefore reported as printed, not as settled findings.
The same capability works in both directions: in one experiment it reduces a false belief, in another it argues for whatever position it is given. Fluency is not proof of accuracy. On the defensive side, short videos that introduce the tactic in advance improve the ability to recognize it, but the effect is modest and has only been measured in the short term: learning the catalog helps, but it is not enough on its own.
The Influence Check: which 6 questions should you ask before saying “yes” under pressure?
The protocol is a CEOtudent editorial framework; it is not a validated scale. It is applied at the moment of an offer, a request or a “decide right now”.
Table 6. The Influence Check (CEOtudent editorial framework)
| # | Question | Flag (1 point) |
|---|---|---|
| 1 | Deadline: is the deadline real, and can it be verified? | There is time pressure, and its justification cannot be verified |
| 2 | Sequence: did the request come after a small “yes”, after a large request that was refused, or after terms that changed later? | Yes |
| 3 | Transparency: would the method still work if it were explained out loud? | No |
| 4 | Symmetry: is saying “no” as easy as saying “yes”? | No; there is an extra step, shaming wording or a preselected box |
| 5 | Record: is the thing in dispute the issue itself, or your memory and perception? | Your perception is being questioned when there is no written record |
| 6 | Personalization: does the argument lean on personal information about you? | Yes, and the source of the information is unclear |
Scoring and stop rule. 0-1 points: evaluate as normal. 2-3 points: slow down; ask for the offer in writing, wait overnight, and verify a single claim with an independent source. 4-6 points: stop; do not decide in that conversation, and bring the matter to a second person. If the decision has a consequence that is hard to reverse, such as money, a contract or sharing personal data, 2 points are enough to stop. In situations that involve a threat, the score is not consulted; the scope note applies.
Ready-made boundary sentences for work. The sentences are neutral; they target the request, not the person.
- “I am not going to decide on this today; once I have it in writing, I will get back to you by noon tomorrow.”
- “The terms I agreed to have changed; I need to evaluate this from scratch under the new terms.”
- “Here are my notes from the meeting; if we remember it differently, let us work from the written record.”
- “I will discuss this request with my manager alongside my current priorities and respond after that.”
- “I cannot take this on; I can suggest this alternative instead.”
Write the outcome in a decision journal; to raise the objection in a meeting, the guide to disagreeing productively can be used.
What are the CEO move and the Student move?
CEO move. The thing to protect is decision quality. Four rules are enough: no decisions under pressure; for an important decision, at least one claim is verified with an independent source; the boundary sentence is written in advance; and it is clear whom to go to in which situation (manager, human resources, legal). Labeling the person does not improve decision quality; recording the behavior does.
Student move. Learn the eight tactics in the catalog the way you would learn reasoning errors. Log each example you encounter with the date, the context, the tactic and your response. Use the same list to audit the tactics you use yourself.
How did the old article’s 8 tips hold up under audit?
Table 7. Claims in the 2021 version and the sources opened (CEOtudent analysis)
| Old claim | What the opened source says | Status |
|---|---|---|
| Introduction: if it is your spouse, get a divorce; if it is your partner, break up | None of the sources opened gives such a directive | Removed |
| 1. They test your patience by not reacting | The silent treatment is a documented tactic (Buss 1987); no source for the intent to “test” or for the advice to “hide your anger” | Partly supported |
| 2. They never change; struggling with them can lead to depression | The consistency in Buss 1987 is across two contexts, not over time; no source for the depression claim | Not found, removed |
| 3. They are cunning and deceptive; do not try to explain | A label on the person; accuracy in telling lies from truths is 54% (Bond and DePaulo 2006) | Not found, removed |
| 4. They always have an excuse (“you are too emotional”) | Trivialization is one dimension of the workplace scale (Kukreja and Pandey 2023); the generalization “always” is not supported | Partly supported |
| 5. They first show interest and win your trust | Charm and ingratiation are documented tactics (Buss 1987; Kipnis 1980); no source for the “mistreatment” sequence that follows | Partly supported |
| 6. They gather information about your weak points | No source for interpersonal relationships; persuasion with personal information was tested in AI experiments | Partly, in a different context |
| 7. They are masters of the first impression | No finding in the sources opened tests this claim | Not found |
| 8. They give inconsistent, evasive answers to personal questions | Liars are less forthcoming in some respects, but most of the cues are weak (DePaulo 2003) | Not supported as a method of diagnosis |
| Darlene Ouimet quote | The opening sentences of a post dated January 25, 2011 on the author’s own blog (archived copy) | Attribution verified; not counted as evidence |
The quote comes from a personal post in which the author describes her experience of therapy: she writes that controlling, abusive and manipulative people do not question themselves, do not ask whether the problem lies with them, and say that the problem is always someone else. The earlier version rendered it loosely. This is testimony, not a research finding; that is why the new version does not use the quote as an epigraph.
Why was this article rewritten?
The 2021 version was an eight-item list that cited no sources: it labeled people as “manipulators”, recommended divorce and breakups, and said that such people never change and that struggling with them can lead to depression. In this version:
- Labels for people have been replaced by the language of observable behavior and tactics.
- The relationship directives and the health claim were removed; a scope note was added.
- The eight tips were audited against the sources opened; the quote was verified from the primary record and given with its context.
- Working life, sales techniques, dark patterns and AI-assisted persuasion were added; the replication, correction and expression-of-concern records for well-known findings were incorporated.
- Ad codes and broken images were removed.
Frequently asked questions
How can you tell that a person is manipulative?
Diagnosing the person is not reliable; people can tell lies from truths with only 54% accuracy. Look at the behavior: deadline, sequence, transparency, symmetry, record and personalization.
How do you deal with manipulative behavior at work?
Postpone the decision, ask for the request in writing, keep records, use a ready-made boundary sentence, and take it to a manager or human resources if needed. If there is a threat or abuse, the scope note applies.
How can you tell that you are being gaslit?
This article does not diagnose. The research describes the phenomenon in close relationships with an unequal balance of power and acknowledges that empirical evidence is scarce. Keep records; if it is persistent, professional support is the right route.
Can countdowns and “last item” warnings be trusted?
Of the 393 timers in the crawl, 157 were deceptive. Check the offer again the next day.
Does AI persuade people better than people do?
In one experiment, GPT-4 with access to personal information was more persuasive than the human in 64.4% of the debate pairs where AI and humans were not equally persuasive; the difference between GPT-4 without personal information and the human was not significant.
Sources
- Buss, Gomes, Higgins and Lauterbach (1987). Tactics of manipulation. Journal of Personality and Social Psychology, 52(6), 1219-1229. Full text (scan).
- Buss (1992). Manipulation in close relationships. Journal of Personality, 60(2), 477-499. Abstract only.
- Kipnis, Schmidt and Wilkinson (1980). Intraorganizational influence tactics. Journal of Applied Psychology, 65(4), 440-452. Abstract only.
- Yukl and Falbe (1990). Influence tactics and objectives in upward, downward, and lateral influence attempts. Journal of Applied Psychology, 75(2), 132-140. Abstract only.
- Higgins, Judge and Ferris (2003). Influence tactics and work outcomes. Journal of Organizational Behavior, 24(1), 89-106. Abstract only.
- Freedman and Fraser (1966). Compliance without pressure. Journal of Personality and Social Psychology, 4(2), 195-202. Abstract only.
- Cialdini et al. (1975). Reciprocal concessions procedure for inducing compliance. Journal of Personality and Social Psychology, 31(2), 206-215. Abstract only.
- Cialdini, Cacioppo, Bassett and Miller (1978). Low-ball procedure for producing compliance. Journal of Personality and Social Psychology, 36(5), 463-476. Abstract only.
- Gamian-Wilk and Dolinski (2020). The foot-in-the-door phenomenon 40 and 50 years later. Psychological Reports, 123(6), 2582-2596. Abstract only.
- Feeley, Anker and Aloe (2012). The door-in-the-face persuasive message strategy. Communication Monographs, 79(3), 316-343. Abstract only.
- Genschow et al. (2021). Does social psychology persist over half a century? Journal of Personality and Social Psychology, 120(2), e1-e7. Abstract only.
- Burger and Caputo (2015). The low-ball compliance procedure. Social Influence, 10(4), 214-220. Abstract only.
- Bond and DePaulo (2006). Accuracy of deception judgments. Personality and Social Psychology Review, 10(3), 214-234. Abstract only.
- DePaulo et al. (2003). Cues to deception. Psychological Bulletin, 129(1), 74-118. Abstract only.
- Paulhus and Williams (2002). The Dark Triad of personality. Journal of Research in Personality, 36(6), 556-563. Full text.
- Sweet (2019). The sociology of gaslighting. American Sociological Review, 84(5), 851-875. Abstract only.
- Klein, Wood and Bartz (2026; online 2025). A theoretical framework for studying the phenomenon of gaslighting. Personality and Social Psychology Review, 30(2), 195-215. Full text.
- Kukreja and Pandey (2023). Workplace gaslighting. Frontiers in Psychology, 14, 1099485. Full text.
- Mathur et al. (2019). Dark patterns at scale. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), 81. Full text (arXiv version).
- Luguri and Strahilevitz (2021). Shining a light on dark patterns. Journal of Legal Analysis, 13(1), 43-109. Abstract only; figures from the co-author’s presentation to the California Privacy Protection Agency.
- Federal Trade Commission (September 2022). Bringing Dark Patterns to Light. Staff Report. Archived copy.
- Salvi, Horta Ribeiro, Gallotti and West (2025). On the conversational persuasiveness of GPT-4. Nature Human Behaviour, 9(8), 1645-1653; Author Correction, September 3, 2026. Full text.
- Costello, Pennycook and Rand (2024). Durably reducing conspiracy beliefs through dialogues with AI. Science, 385(6714), eadq1814. Abstract only; AAAS announcement for the Editorial Expression of Concern (June 11, 2026).
- Matz et al. (2024). The potential of generative AI for personalized persuasion at scale. Scientific Reports, 14, 4692. Full text.
- Roozenbeek et al. (2022). Psychological inoculation improves resilience against misinformation on social media. Science Advances, 8(34), eabo6254. Full text.
- Noggle. The Ethics of Manipulation. Stanford Encyclopedia of Philosophy; first published March 30, 2018, substantive revision March 31, 2026.
- Ouimet (January 25, 2011). Controllers and Manipulative People don’t Question Themselves. Emerging From Broken (the author’s blog), Internet Archive copy.
- Influence at Work. Dr. Robert Cialdini’s Seven Principles of Persuasion. Website of the author’s training company (commercial source).
Tables 2, 4, 6 and 7 are CEOtudent analysis or editorial framework; Tables 1, 3 and 5 give the printed figures. The total of 754, the 1.6%, the 4 points, the category counts and shares, the 63.8%, the 39.9%, and the 2.4 and 3.8 times are CEOtudent calculations. Removed from the old article: labels for people, the divorce and breakup recommendations, the “they never change” and depression claims, and the unsourced tips. The attribution of the Darlene Ouimet quote was verified, but it was not counted as evidence. Not used because they could not be accessed: Jonason and Webster (2012), the full texts of Sweet and of Luguri and Strahilevitz, the text of Science’s Editorial Expression of Concern, and the frequently cited rates from the first compliance experiments. The overall charm mean in the text of Buss and colleagues (2.81) was not used because it does not match the simple average of the four cells in the table (2.83).
This content was compiled with the support of AI following in-depth research, then written and prepared for publication by the CEOtudent editorial team.
This post is also available in:















