Why Do Leaders Fail? Hubris, Narcissism, Groupthink, Power, and Bad Decisions
Updated: 3 days ago
Author: Ukrainian Psychological Hub · Published: September 25, 2026 · Editorial Policy
Leaders fail for many different reasons, and the strongest psychological explanation is not a single defective trait. Leadership failure usually develops when the person, the role, the decision process, the surrounding group, and the distribution of power stop correcting one another. A leader may have been selected for confidence, visibility, or charisma rather than for the capacities needed to make difficult decisions. Success can increase discretion while weakening feedback. Teams can suppress dissent or fail to share unique information. Personality tendencies such as narcissism can help someone emerge as a leader without making that person more effective once in the role. Organizational incentives can reward short-term action while hiding long-term damage.
The evidence therefore points toward a system of failure rather than a universal profile of the “failed leader.” Recent meta-analytic research explicitly distinguishes leader emergence from leader effectiveness, while an updated meta-analysis of personality and leadership shows that different traits predict different leadership criteria and that leader behavior can mediate some trait–effectiveness relationships (Javalagi et al., 2024). That distinction is central: what helps a person become recognized as a leader is not identical to what helps that person lead well.
This article explains why leaders fail at the level of psychological mechanisms and organizational processes. It treats hubris, narcissistic traits, groupthink, power, information failure, destructive behavior, and bad decisions as related but distinct constructs. It also separates failure from simple unpopularity, a single bad outcome, a clinical diagnosis, and the broader question of what makes a good leader.
Quick Answer: Why Do Leaders Fail?
Leaders most often fail when several vulnerabilities combine: selection for signals that look leader-like but do not guarantee effectiveness; overconfidence or hubristic judgment; poor use of advice; weak information sharing; conformity and suppressed dissent; excessive discretion without accountability; destructive or abusive behavior; context–role mismatch; and organizational systems that do not detect or correct errors early. The same leader can perform well in one environment and fail in another because leadership effectiveness is an outcome produced by a person-in-context, not a permanent personal property.
Some of these mechanisms have stronger evidence than others. Personality–leadership relationships are supported by large meta-analyses, but their effects depend on the criterion and context. Narcissism has a clearer relationship with leader emergence than with effectiveness (Grijalva et al., 2015). Power has experimentally supported effects on confidence and advice taking, yet the broader psychology of power is context-sensitive rather than uniformly corrupting (Guinote, 2017; See et al., 2011). Groupthink remains influential as a theory of defective group decision-making, although reviews show that its classic package of causes and symptoms has not been uniformly confirmed (Esser, 1998).
What Counts as Leadership Failure?
Leadership failure is best treated as an outcome category: the leadership process persistently fails to provide the direction, coordination, adaptation, decision quality, relationships, or protection from harm that the role requires. In organizations, failure may appear as chronic underperformance, poor strategic decisions, loss of follower trust, inability to adapt, preventable ethical breaches, destructive treatment of subordinates, repeated information failures, or derailment from a role. These outcomes can overlap, but they should not be collapsed into one score.
A leader can fail without being destructive. A technically weak manager may be promoted beyond their current capabilities, misread a new market, or lack the skills required by a more complex role. That is incompetence or role mismatch, not automatically toxic leadership. Conversely, a leader can produce short-term performance while creating serious follower harm. Performance and harm are different criteria, which is why leadership research increasingly uses multiple outcomes rather than a single global judgment.
Leader emergence is not leadership effectiveness
Leader emergence asks who becomes influential, is nominated, or comes to be seen as a leader. Leadership effectiveness asks how well the leader performs against relevant outcomes. The two constructs overlap imperfectly. A 2026 meta-analysis of 203 data sets found substantial variation in how personality facets relate to emergence versus effectiveness (Baker et al., 2026). The practical implication is simple: selection systems can systematically reward qualities that help a candidate win a leadership role without selecting equally well for qualities needed after appointment.
Narcissism illustrates the problem. A major meta-analysis found a positive association between narcissism and leadership emergence, but no overall linear relationship with leadership effectiveness. The emergence relationship was explained by extraversion, and self-ratings of effectiveness were more favorable than observer ratings (Grijalva et al., 2015). A leader can therefore look unusually leader-like before appointment while offering little evidence of superior effectiveness afterward.
A bad outcome is not always proof of a bad decision
Leadership outcomes are partly shaped by uncertainty, competition, timing, resources, and events outside a leader’s control. Psychology also documents outcome bias: people evaluate the quality of a decision differently after learning whether the result was good or bad, even when the information available at the time of decision was the same (Baron & Hershey, 1988). Evaluating leaders only by hindsight can therefore confuse decision quality with luck.
A stronger evaluation asks whether the decision used the best available information, considered credible alternatives, handled uncertainty explicitly, invited relevant expertise, established safeguards, and updated when new evidence appeared. These process criteria do not guarantee success, but they provide a better basis for learning than judging every failure as proof of personal incompetence.
Leadership Failure Is Not One Psychological Construct
Several neighboring ideas must remain separate. Leadership style refers to a pattern of leader behavior; leadership theory is an explanatory framework about how leadership works. Personality traits describe relatively stable individual differences; leadership behaviors describe what leaders do. Management includes planning, coordination, resource allocation, and formal organizational functions; leadership centers more specifically on influence and collective direction, although real jobs often combine both. Supervision is a formal relationship in which one person oversees another’s work and can occur with strong, weak, or harmful leadership.
Power, authority, status, dominance, and prestige also differ. Power concerns asymmetric capacity to affect outcomes. Authority concerns socially or institutionally recognized legitimacy to direct action. Status is socially conferred rank or respect. Dominance obtains deference through threat, intimidation, or coercive capacity; prestige obtains deference through valued competence, knowledge, or respect. These distinctions are developed in the Hub’s guides to power and leadership, power versus status, and dominance versus prestige.
Keeping the constructs separate prevents a familiar error: explaining every failure through a fashionable label. A domineering leader is not necessarily narcissistic. A narcissistic leader is not necessarily abusive. A supervisor can be abusive without meeting any clinical diagnosis. A leader can make poor decisions because of information architecture, incentives, or role overload even when personality is unremarkable.
1. Organizations Can Select for Leadership Signals Instead of Leadership Effectiveness
Selection is the first place leadership can fail. People infer leadership from observable signals: confidence, verbal fluency, decisiveness, social energy, status cues, certainty, and the ability to command attention. Some of these cues are genuinely useful. The problem begins when the cue becomes a substitute for the criterion. Being convincing during selection is not the same task as integrating weak signals, revising a strategy, building a team, or making consequential decisions under uncertainty.
Meta-analytic evidence supports meaningful relationships between personality and leadership while also showing that those relationships depend on the outcome. Javalagi, Newman, and Li’s 2024 meta-analysis included 120 samples and 32,579 participants for leader emergence and 116 samples and 42,487 participants for leader effectiveness. It found cross-cultural moderation, behavioral mediation for some traits, and incremental prediction from Honesty–Humility (Javalagi et al., 2024). The 2026 facet-level meta-analysis similarly found that narrow facets vary in their relationships with emergence and effectiveness (Baker et al., 2026).
Traits therefore matter, but they are not a leadership destiny. An earlier integrative meta-analysis found that leader traits and behaviors both predicted effectiveness criteria and that behavior tended to explain more variance than traits, supporting models in which traits partly operate through what leaders actually do (DeRue et al., 2011). A selection process that treats a personality score or a strong interview performance as a complete model of leadership is asking one measurement to answer too many questions.
2. Hubris and Overconfidence Can Distort Judgment
Hubris in leadership research describes an exaggerated belief in one’s own abilities, judgments, or exceptional standing, often accompanied by overambitious decisions and resistance to criticism. A major review describes hubristic leadership in relation to overconfidence, narcissism, pride, and the consequences of significant power, while also emphasizing conceptual and measurement problems in the field (Sadler-Smith et al., 2017). Hubris is used here as a leadership and decision-making construct, not as a clinical diagnosis.
Overconfidence is narrower. It can involve excessive certainty in one’s judgments, overestimation of performance, or overplacement relative to others. The distinction matters because confidence is not uniformly harmful. It can support action under ambiguity, persistence, strategic risk taking, and social influence. A meta-analysis of CEO overconfidence found links among overconfidence, strategic risk taking, and performance that depended on the pathway and context, illustrating why “overconfidence always causes failure” is too simple (Burkhard et al., 2023).
The most dangerous configuration is confidence without correction. When a leader can make consequential choices, receives weak feedback, discounts disconfirming expertise, and interprets prior success as proof of superior judgment, uncertainty becomes progressively less visible inside the decision process. The leader may still look decisive from outside. Internally, the system has lost calibration.
A systematic review of overconfidence and narcissism among senior executives also warns that the constructs overlap without being interchangeable and that their measurement varies considerably across studies (Brunzel, 2021). That methodological point is important for popular discussions of “hubris”: observed boldness, a large acquisition, an ambitious strategy, or a confident public statement is not enough by itself to establish a stable psychological trait.
3. Narcissistic Traits Can Help Leaders Emerge Without Making Them More Effective
Narcissism is one of the clearest examples of why emergence and effectiveness must be separated. Leadership research usually examines narcissistic personality traits, often grandiose narcissism, rather than diagnosing Narcissistic Personality Disorder. A behavioral description of a leader cannot establish a clinical diagnosis, and organizational studies of narcissism should not be read as if they had done so.
Grijalva and colleagues’ meta-analysis found that narcissism was positively associated with leadership emergence but not with leadership effectiveness overall. Observer-rated effectiveness was unrelated to narcissism, whereas self-rated effectiveness was positively related, and the linear null relationship concealed evidence consistent with a curvilinear pattern (Grijalva et al., 2015). The study is especially useful because it shows how a broad statement such as “narcissists make good leaders” or “narcissists make bad leaders” erases the criterion, the rater, and the shape of the relationship.
Later work likewise shows that follower responses to narcissistic leaders can vary with visibility and context rather than forming a single universally negative effect (Nevicka et al., 2018). For leadership failure, the key mechanisms are more specific than the trait label: self-enhancement can distort self-assessment; entitlement can make correction aversive; admiration-seeking can make symbolic success more attractive than operational learning; and sensitivity to ego threat can change how dissent is interpreted.
This is why the planned dedicated English Hub article on narcissistic leadership owns the narrower intent around narcissism, leader emergence, and effectiveness. The present article uses narcissism only as one pathway into leadership failure and does not turn it into a general explanation for difficult bosses.
4. Power Changes the Decision Environment
Power gives leaders greater capacity to act, allocate resources, set agendas, reward, punish, and define which information reaches a decision. Those capacities can improve coordination when time is limited and responsibility is clear. They can also weaken the corrective environment around the decision maker.
A major review of the psychology of power describes power as an intensifier of goal-directed action. Power can increase confidence, action orientation, self-expression, and prioritization while also reducing social attention or perspective taking under some conditions (Guinote, 2017). The pattern is conditional. Power does not produce one invariant psychological state in every person and every situation.
Advice taking provides a more specific causal mechanism. Across a field study and experiments, See and colleagues found that greater power could increase confidence, reduce reliance on advice, and lower accuracy when outside input would have improved judgment (See et al., 2011). This does not mean leaders should outsource decisions. It means that the authority to decide and the epistemic ability to know when someone else has better information are different capacities.
The broader question of whether power corrupts is treated separately in Does Power Corrupt? Psychology, Morality, Behavior, and the Power Paradox. For leadership failure, the useful conclusion is narrower: power changes what leaders can do and can alter how they process goals, confidence, other people, and advice. Responsibility, norms, accountability, role expectations, and institutional constraints help determine which effects become behavior.
5. Groupthink Can Suppress Dissent, but the Classic Theory Is Often Overstated
Groupthink is Irving Janis’s influential account of defective group decision-making in highly cohesive groups under conditions that can include insulation, directive leadership, stress, and poor decision procedures. In popular language the term is often reduced to “people agreeing with each other.” That loses the theory’s structure and exaggerates how securely all of its components have been established.
Reviews of the evidence have been more cautious. Esser’s review concluded that groupthink retained heuristic value while noting that the empirical literature remained limited and that many theoretical ideas had not been adequately tested (Esser, 1998). A meta-analytic integration by Mullen and colleagues found no overall significant effect of cohesiveness on decision quality by itself; poorer decisions appeared under certain additional conditions, including directive leadership, and the form of cohesion mattered (Mullen et al., 1994).
The practical target is therefore not “cohesion” in the abstract. High-trust teams can coordinate extremely well. Failure becomes more plausible when agreement becomes socially rewarded, dissent carries interpersonal or career costs, the leader announces a preferred answer before discussion, alternatives are not genuinely compared, and the group lacks a procedure for surfacing disconfirming information.
6. Teams Often Fail to Pool the Information They Already Have
A group can contain enough information to make a good decision and still fail because the information is distributed unevenly. Hidden-profile research studies exactly this problem: information known to everyone favors one option, while different members privately hold pieces that collectively favor a better option.
A meta-analysis covering 65 studies and 3,189 groups found that groups discussed far more shared than unique information and that hidden-profile groups were dramatically less likely to discover the correct solution than groups with fully shared information. Better coverage of unique information was associated with better decision quality (Lu et al., 2012). Broader reviews of group performance likewise show that information exchange, task structure, member knowledge, and decision procedures interact in ways that make group performance more complex than simply “more minds are better” (Kerr & Tindale, 2004).
Leadership can worsen this problem unintentionally. A leader who states an early preference creates an anchor for discussion. Members may repeat information everyone already accepts, interpret silence as consensus, or decide that raising a contrary fact is socially costly. Failure then comes from the information architecture of the group, even if every participant is intelligent and conscientious.
7. Low Psychological Safety Can Hide Errors Until They Become Expensive
Psychological safety concerns whether people believe they can take interpersonal risks such as asking questions, admitting uncertainty, reporting mistakes, or challenging an idea. It is not the same as interpersonal trust, engagement, job satisfaction, inclusion, or a generally pleasant workplace climate. Those constructs can correlate, but they answer different questions.
A major meta-analysis synthesized 136 independent samples representing more than 22,000 individuals and nearly 5,000 groups and found meaningful relationships between psychological safety, its antecedents, and work outcomes beyond several related constructs (Frazier et al., 2017). For leadership failure, psychological safety matters because leaders are structurally dependent on other people to tell them what they cannot see.
A leader can be personally intelligent and still operate inside an epistemically poor system. If subordinates learn that bad news is punished, uncertainty is interpreted as weakness, or contradiction is treated as disloyalty, the leader’s information environment becomes increasingly curated. Decision quality may deteriorate before the leader receives any signal strong enough to force revision.
8. Charisma Can Create Influence Without Guaranteeing Correct Judgment
Charisma is especially easy to confuse with effectiveness because both can be measured through follower perceptions. A leader who communicates a compelling vision, uses emotionally resonant language, and creates a strong sense of confidence can legitimately improve motivation and coordination. Those effects do not make every charismatic judgment accurate.
A meta-analytic review of 76 independent studies and 36,031 individuals found that charismatic leadership dimensions were associated with important objective and supervisor-rated outcomes, while also identifying persistent problems in conceptualization, measurement, and causal inference (Banks et al., 2017). The evidence therefore supports charisma as a meaningful leadership construct while resisting the shortcut that treats perceived charisma as a direct measure of leader competence.
For failure analysis, the relevant distinction is between influence and epistemic quality. A leader can be exceptionally effective at getting people to commit to a decision and still be wrong about the decision itself. The more persuasive the leader, the more important independent information channels and dissent procedures become.
9. Destructive Leadership Is More Specific Than Ordinary Failure
Destructive leadership refers to patterns of leader behavior that harm followers, teams, or organizations. It is a research family containing overlapping constructs such as abusive supervision, petty tyranny, despotic leadership, and other forms of hostile or undermining leadership. It should not be used as a synonym for every disappointing manager.
A systematic review and meta-analysis integrating 418 studies found a substantial empirical literature linking destructive leadership constructs to harmful workplace outcomes, while also documenting definitional overlap across the field (Mackey et al., 2021). An earlier meta-analysis likewise found destructive leadership associated with poorer follower outcomes and stronger withdrawal, resistance, and counterproductive responses (Schyns & Schilling, 2013).
Abusive supervision is narrower still. Tepper’s foundational work defined it around subordinates’ perceptions of sustained hostile verbal and nonverbal supervisory behavior, excluding physical contact, and linked it to multiple adverse employee outcomes (Tepper, 2000). A later meta-analysis consolidated relationships with justice, attitudes, stress-related outcomes, and behavior (Mackey et al., 2017). The dedicated future articles on toxic leadership and abusive supervision will own those search intents; here they function as specific routes by which leadership can fail.
10. Dark Traits Are Research Constructs, Not Internet Diagnoses
Dark Triad research examines narcissism, Machiavellianism, and psychopathy as personality constructs. In organizational studies these are typically measured dimensionally with research scales. They should not be converted into clinical diagnoses from anecdotes, management style, public behavior, or a list of disliked traits.
Psychopathy is a useful example. A meta-analysis of psychopathic personality characteristics and leadership found a weak positive relationship with leader emergence, a weak negative relationship with effectiveness, and a more substantial negative relationship with transformational leadership (Landay et al., 2019). Those are population-level associations in a research literature. They do not imply that an ineffective, cold, manipulative, or aggressive leader can be clinically labeled “a psychopath.”
The same caution applies to narcissism. Trait narcissism, grandiose narcissism, narcissistic leadership, and Narcissistic Personality Disorder are not interchangeable terms. A science-based explanation identifies the measured construct and the outcome instead of using diagnostic language as a moral insult.
11. Context and Role Fit Can Turn Strengths Into Liabilities
Leadership is contingent on what the situation requires. A fast-moving emergency may reward rapid centralization and clear direction. A research team facing an uncertain technical problem may need distributed expertise and open disagreement. A mature operational unit may need reliability and process discipline. A turnaround may require disruption. A leader can therefore fail because the behavioral repertoire that worked in one environment is carried unchanged into another.
This is one reason the Hub separates leadership styles from leadership theories. A style describes a pattern of behavior; a theory proposes relationships among leaders, followers, tasks, situations, and outcomes. Treating a style label as a universal prescription removes exactly the contextual variables that often explain success and failure.
Role transitions amplify the problem. Promotion changes task complexity, time horizon, stakeholder diversity, information quality, and the degree to which results depend on other leaders. What made someone an excellent specialist or first-line supervisor may not be enough for a strategic role. Failure can emerge from a lag between the leader’s existing capabilities and the new architecture of the job rather than from a newly acquired personality flaw.
12. Organizational Systems Can Manufacture Leadership Failure
Leadership psychology becomes misleading when every organizational outcome is attributed to the person at the top. Incentives, reporting structures, governance, resource constraints, norms, staffing, market conditions, and the design of decision rights can all create predictable error. A system that rewards only quarterly output may rationally produce decisions that damage long-term capacity. A board that receives information only through the chief executive has created an information bottleneck. A culture that punishes dissent can produce apparent consensus around weak strategies.
This does not dissolve personal responsibility. It locates responsibility more precisely. Leaders are accountable for the choices they control, including the systems they design, but they are not omnipotent causes of every outcome. Leadership failure is often a coupled process in which personal tendencies and institutional arrangements reinforce one another.
The distinction also clarifies the relationship between leadership and power. The Hub’s canonical bridge on Power and Leadership: Influence, Status, Authority, and Followership owns the general question of power in leadership. The present article focuses on how power can become one component in a failure pathway when it weakens correction, concentrates discretion, or changes the social cost of disagreement.
Hubris, Narcissism, and Overconfidence Are Related but Different
Hubris, narcissism, and overconfidence are often bundled together because all three can involve elevated self-appraisal. Their psychological meanings differ. Overconfidence concerns miscalibration or overestimation in judgment. Narcissism is a broader personality construct involving patterns of grandiosity, entitlement, admiration-seeking, and self-focus, with different forms and measurement traditions. Hubris in leadership research emphasizes excessive pride, exceptionalism, overambition, and resistance to correction, often in connection with power and success.
The constructs can reinforce one another without being identical. An overconfident leader may genuinely care about the team and remain open to criticism in domains where uncertainty is visible. A narcissistic leader may make some well-calibrated technical judgments. A hubristic pattern may develop or intensify after repeated success and expanded discretion. A systematic review of upper-echelon research specifically argues for disentangling overconfidence and narcissism because the literatures use different measures and produce different organizational implications (Brunzel, 2021).
Does Power Corrupt Leaders?
The most accurate answer is conditional. Power increases capacity and discretion. It can help leaders act decisively, coordinate resources, protect a team, and pursue long-term goals. It can also reduce dependence on others, increase confidence, intensify existing goals, and change attention to subordinate perspectives. Which pathway dominates depends on motives, norms, role expectations, accountability, responsibility, institutions, and the leader’s active goals (Guinote, 2017).
Leadership failure becomes more likely when power is paired with weak constraints and poor information. The issue is not simply that the leader has authority. It is that the same person may control the decision, the evaluation of the decision, access to dissenting information, and the consequences faced by people who challenge it. Separating those functions creates correction.
For a broader treatment of the evidence, see Does Power Corrupt? and Power and Morality. The Hub also distinguishes power over others from the subjective and practical experience of control in Power and Control.
Why Smart and Previously Successful Leaders Still Fail
Intelligence and prior success can improve leadership while also creating specific vulnerabilities. Success produces genuine information: a strategy worked, the person handled difficult problems, and others trusted their judgment. Repeated success can also reduce the number of situations in which the leader receives unambiguous correction. Senior leaders often operate farther from frontline data, receive more filtered communication, and make decisions whose feedback arrives slowly.
This creates a paradox of learning. The more powerful the role, the more the leader needs other people to supply inconvenient information; the more consequential the hierarchy, the more costly those people may perceive dissent to be. The leader’s previous success can make both sides of the exchange less likely to challenge established assumptions. Over time, an adaptive confidence built from real achievement can drift into poor calibration.
The solution is institutional as well as personal: independent data, multiple advice channels, explicit red-team or dissent roles where appropriate, post-decision review, and separation between proposing a strategy and evaluating its outcome. These mechanisms protect learning without requiring the leader to become indecisive.
Leadership Failure and Bad Decisions
Not every leadership failure is a decision error, but high-impact failures often contain a decision architecture problem. Four questions are especially useful: What information reached the leader? Which information did not? How were alternatives generated and challenged? What happened to people who disagreed?
A high-quality process makes uncertainty visible. It distinguishes forecasts from commitments, assumptions from facts, and confidence from evidence. It records why an option was chosen so later evaluation can compare reasoning with outcomes without reconstructing the past. It also asks what evidence would trigger revision before the organization becomes psychologically or financially committed to one path.
This is where groupthink, advice discounting, psychological safety, and hidden-profile research meet. Each studies a different mechanism, yet all show why decision quality depends on the route information takes through a social system. Leadership is partly the design of that route.
Warning Patterns That Deserve Attention
No single behavior diagnoses a failing leader, and occasional mistakes are inevitable. Risk rises when patterns accumulate across time and sources. Warning signs include repeated rejection of disconfirming evidence; increasing reliance on loyal insiders; punishment or ridicule of dissent; decisions announced before expertise is heard; metrics changed when they become unfavorable; persistent self-credit and externalized blame; turnover among people who previously supplied contrary information; escalating commitments without new evidence; and a widening gap between the leader’s self-assessment and credible multi-source feedback.
Other patterns point to different mechanisms: chronic indecision may reflect overload, weak role clarity, or low competence rather than hubris. Micromanagement may arise from low trust, anxiety, poorly designed delegation, or incentives that make the supervisor personally accountable for every error. Hostility toward subordinates may meet research definitions of destructive leadership or abusive supervision when it is sustained, but ordinary conflict, strict performance standards, and unpopular decisions are not automatically abuse.
How Leaders Can Reduce Their Own Failure Risk
Build correction into the role
The most reliable defense against leadership failure is not confidence in one’s humility; it is a system in which correction can occur even when the leader is tired, defensive, certain, or socially dominant. Independent reporting channels, structured challenge, protected dissent, and periodic external review make feedback less dependent on mood and interpersonal courage.
Separate advocacy from evaluation
Leaders who originate an idea have psychological and reputational investment in it. Assigning independent people to test assumptions, define failure criteria, and evaluate evidence reduces the chance that the proposal’s strongest advocate also becomes its only judge.
Delay public commitment when information is incomplete
Public commitment can turn updating into a status threat. When uncertainty is high, leaders can state a provisional hypothesis, specify what would change their mind, and distinguish a decision deadline from a need to appear certain. This preserves decisiveness while keeping revision legitimate.
Ask for information before asking for opinions
Hidden-profile research suggests that groups can fail by discussing what everyone already knows. A leader can first ask each relevant participant for unique facts, risks, constraints, or observations before the group debates preferred options. The sequence matters because early preference statements can reorganize the discussion around agreement.
Use multi-source feedback carefully
Self-perception is one data source, not the final standard. The narcissism meta-analysis is a useful reminder that self-rated and observer-rated leadership effectiveness can diverge (Grijalva et al., 2015). Useful feedback systems define behaviors, collect information across relevant observers, protect confidentiality where appropriate, and connect feedback to specific changes rather than to global judgments of character.
How Organizations Can Prevent Leadership Failure
Prevention begins before promotion. Selection should match predictors to the outcomes the organization actually needs: judgment under uncertainty, learning, coordination, integrity, follower development, execution, safety, innovation, or other criteria. The same evidence should not be used to claim that someone will emerge as a leader, be admired as a leader, and be effective across every outcome.
Organizations should also monitor role transitions. A promotion increases complexity and changes the meaning of delegation, expertise, and time. Early support can include clear decision rights, access to domain experts, feedback from multiple levels, and explicit expectations for how disagreement is handled. Waiting until performance visibly collapses is an expensive way to discover that a role and a leader no longer fit.
Governance should preserve independent information. Boards, senior teams, and oversight structures need routes to data that do not depend entirely on the person being evaluated. The principle applies at smaller scales too: a team leader should not be the sole source of information about team morale, risk, or project status.
Finally, organizations should distinguish development from containment. Some problems can be improved through feedback, training, coaching, clearer roles, or better decision procedures. Sustained abusive behavior, manipulation of controls, retaliation, or severe ethical violations require governance responses that protect other people and the organization rather than treating every problem as a coaching opportunity.
What the Evidence Can and Cannot Establish
Leadership research contains meta-analyses, longitudinal studies, field studies, experiments, and cross-sectional surveys. These designs answer different causal questions. A correlation between a leader trait and an outcome does not show that the trait caused the outcome. Mediation models can be theoretically useful without proving causal sequence when the underlying data are cross-sectional. Longitudinal evidence strengthens temporal inference but still may contain confounding. Experiments offer stronger causal leverage for specific mechanisms, yet laboratory manipulations may simplify the complexity of real leadership roles.
The power-and-advice literature provides experimental evidence for a specific mechanism: power can alter confidence and advice use under studied conditions (See et al., 2011). By comparison, much CEO narcissism and hubris research relies on archival proxies, unobtrusive indicators, executive samples, or observational designs, which makes measurement validity and endogeneity central concerns (Brunzel, 2021).
This is why the article avoids one-cause explanations. Leadership failure is a multilevel outcome. The strongest conclusion from the evidence is that failure risk rises when selection, personality, power, decision procedures, team dynamics, incentives, and accountability align in ways that suppress correction.
Frequently Asked Questions
Why do good leaders fail?
A leader can have many effective qualities and still fail when the role changes, information is incomplete, feedback becomes filtered, a strategy is poorly matched to context, or a team cannot challenge assumptions. Effectiveness is criterion- and context-dependent rather than a permanent badge.
Do narcissistic leaders always fail?
No. Meta-analytic evidence does not support a simple linear claim that greater narcissism always means lower leadership effectiveness. Narcissism is more consistently associated with leader emergence than with effectiveness, and results depend on rater, context, and measurement (Grijalva et al., 2015).
Is hubris the same as narcissism?
No. They overlap in self-enhancing tendencies but are distinct constructs. Hubris emphasizes excessive pride, exceptionalism, overambitious judgment, and resistance to correction; narcissism is a broader personality construct. Overconfidence is narrower still and refers to miscalibrated certainty or self-estimation.
Does power make leaders corrupt?
Power changes action capacity, motivation, confidence, and social cognition, but its effects depend on goals, norms, responsibility, accountability, and context. Research does not justify treating corruption as an automatic psychological consequence of holding power (Guinote, 2017).
Is groupthink caused by team cohesion?
Cohesion alone is not a sufficient explanation. Meta-analytic evidence found no overall significant effect of cohesiveness on decision quality, although poorer decisions appeared when cohesion was combined with other conditions such as directive leadership, and different forms of cohesion behaved differently (Mullen et al., 1994).
Are charismatic leaders more effective?
Charismatic leadership is associated with several meaningful outcomes, but charisma is a construct with measurement and causal-inference limitations. Perceived charisma should not be used as a synonym for accurate judgment or global leadership effectiveness (Banks et al., 2017).
What is the difference between toxic leadership and abusive supervision?
Toxic leadership is a broad popular and research-adjacent label whose definitions vary. Destructive leadership is a broader scholarly family of harmful leader behaviors. Abusive supervision is a more specific research construct centered on sustained hostile verbal and nonverbal supervisory behavior as perceived by subordinates. The concepts overlap but are not interchangeable.
Can an incompetent leader be ethical?
Yes. Competence and ethics are different dimensions. A leader can act in good faith while lacking the knowledge, judgment, coordination skill, or contextual fit required by the role. Conversely, a highly competent leader can behave destructively or unethically. Leadership evaluation should measure both capability and conduct.
How can a leader know whether confidence has become overconfidence?
Useful signals include persistent gaps between forecasts and outcomes, a pattern of dismissing qualified advice, unusually narrow uncertainty ranges, repeated surprise at negative evidence, and resistance to specifying what would change a decision. Calibration improves when predictions and assumptions are recorded before outcomes are known.
Can leaders recover after failure?
Often, depending on the mechanism and the consequences. Recovery is more plausible when the leader can identify the failure process, accept credible feedback, change decision routines, repair damaged relationships, and operate within a governance system that verifies improvement. Severe misconduct or sustained abuse raises different questions because protection and accountability take priority over developmental experimentation.
The Core Principle
Leadership fails when correction fails. Personality matters because it shapes attention, motivation, social behavior, and response to feedback. Power matters because it changes discretion and dependence. Group processes matter because leaders rarely possess all relevant information. Organizational design matters because it determines who can challenge a decision and what happens when they do. Context matters because the same behavior can be adaptive in one task and damaging in another.
The practical aim is therefore not to identify a mythical personality that can never fail. It is to build leadership systems in which emergence is not confused with effectiveness, confidence is calibrated by evidence, charisma is not mistaken for competence, power is paired with responsibility, dissent remains possible, and decisions can be revised before error becomes identity.
Related Articles
References
Baker, N., Scott, W., Nye, C. D., Chernyshenko, O. S., Park, H. W., & Omori, C. L. (2026). The many facets of leadership: A meta-analysis of personality facets, leader effectiveness, and emergence. Journal of Applied Psychology, 111(9), 1097–1118. https://doi.org/10.1037/apl0001378
Banks, G. C., Engemann, K. N., Williams, C. E., Gooty, J., McCauley, K. D., & Medaugh, M. R. (2017). A meta-analytic review and future research agenda of charismatic leadership. The Leadership Quarterly, 28(4), 508–529. https://doi.org/10.1016/j.leaqua.2016.12.003
Baron, J., & Hershey, J. C. (1988). Outcome bias in decision evaluation. Journal of Personality and Social Psychology, 54(4), 569–579. https://doi.org/10.1037/0022-3514.54.4.569
Brunzel, J. (2021). Overconfidence and narcissism among the upper echelons: A systematic literature review. Management Review Quarterly, 71, 585–623. https://doi.org/10.1007/s11301-020-00194-6
Burkhard, B., Sirén, C., van Essen, M., Grichnik, D., & Shepherd, D. A. (2023). Nothing ventured, nothing gained: A meta-analysis of CEO overconfidence, strategic risk taking, and performance. Journal of Management, 49(8), 2629–2666. https://doi.org/10.1177/01492063221110203
DeRue, D. S., Nahrgang, J. D., Wellman, N., & Humphrey, S. E. (2011). Trait and behavioral theories of leadership: An integration and meta-analytic test of their relative validity. Personnel Psychology, 64(1), 7–52. https://doi.org/10.1111/j.1744-6570.2010.01201.x
Esser, J. K. (1998). Alive and well after 25 years: A review of groupthink research. Organizational Behavior and Human Decision Processes, 73(2–3), 116–141. https://doi.org/10.1006/obhd.1998.2758
Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A., & Vracheva, V. (2017). Psychological safety: A meta-analytic review and extension. Personnel Psychology, 70(1), 113–165. https://doi.org/10.1111/peps.12183
Grijalva, E., Harms, P. D., Newman, D. A., Gaddis, B. H., & Fraley, R. C. (2015). Narcissism and leadership: A meta-analytic review of linear and nonlinear relationships. Personnel Psychology, 68(1), 1–47. https://doi.org/10.1111/peps.12072
Guinote, A. (2017). How power affects people: Activating, wanting, and goal seeking. Annual Review of Psychology, 68, 353–381. https://doi.org/10.1146/annurev-psych-010416-044153
Javalagi, A. A., Newman, D. A., & Li, M. (2024). Personality and leadership: Meta-analytic review of cross-cultural moderation, behavioral mediation, and honesty-humility. Journal of Applied Psychology, 109(9), 1489–1511. https://doi.org/10.1037/apl0001182
Kerr, N. L., & Tindale, R. S. (2004). Group performance and decision making. Annual Review of Psychology, 55, 623–655. https://doi.org/10.1146/annurev.psych.55.090902.142009
Landay, K., Harms, P. D., & Credé, M. (2019). Shall we serve the dark lords? A meta-analytic review of psychopathy and leadership. Journal of Applied Psychology, 104(1), 183–196. https://doi.org/10.1037/apl0000357
Lu, L., Yuan, Y. C., & McLeod, P. L. (2012). Twenty-five years of hidden profiles in group decision making: A meta-analysis. Personality and Social Psychology Review, 16(1), 54–75. https://doi.org/10.1177/1088868311417243
Mackey, J. D., Frieder, R. E., Brees, J. R., & Martinko, M. J. (2017). Abusive supervision: A meta-analysis and empirical review. Journal of Management, 43(6), 1940–1965. https://doi.org/10.1177/0149206315573997
Mackey, J. D., Ellen, B. P., III, McAllister, C. P., & Alexander, K. C. (2021). The dark side of leadership: A systematic literature review and meta-analysis of destructive leadership research. Journal of Business Research, 132, 705–718. https://doi.org/10.1016/j.jbusres.2020.10.037
Mullen, B., Anthony, T., Salas, E., & Driskell, J. E. (1994). Group cohesiveness and quality of decision making: An integration of tests of the groupthink hypothesis. Small Group Research, 25(2), 189–204. https://doi.org/10.1177/1046496494252003
Nevicka, B., Van Vianen, A. E. M., De Hoogh, A. H. B., & Voorn, B. C. M. (2018). Narcissistic leaders: An asset or a liability? Leader visibility, follower responses, and group-level absenteeism. Journal of Applied Psychology, 103(7), 703–723. https://doi.org/10.1037/apl0000298
Sadler-Smith, E., Akstinaite, V., Robinson, G., & Wray, T. (2017). Hubristic leadership: A review. Leadership, 13(5), 525–548. https://doi.org/10.1177/1742715016680666
Schyns, B., & Schilling, J. (2013). How bad are the effects of bad leaders? A meta-analysis of destructive leadership and its outcomes. The Leadership Quarterly, 24(1), 138–158. https://doi.org/10.1016/j.leaqua.2012.09.001
See, K. E., Morrison, E. W., Rothman, N. B., & Soll, J. B. (2011). The detrimental effects of power on confidence, advice taking, and accuracy. Organizational Behavior and Human Decision Processes, 116(2), 272–285. https://doi.org/10.1016/j.obhdp.2011.07.006
Tepper, B. J. (2000). Consequences of abusive supervision. Academy of Management Journal, 43(2), 178–190. https://doi.org/10.5465/1556375