Sexual Harassment

Abstract

Sexual harassment, which the Medical Subject Headings thesaurus classifies under both sexual behavior and social behavior, is unwelcome sexual attention, coercion, or gender-based hostility directed at a person, most often studied in workplaces and schools. Cognitive psychology treats it less as a legal category than as a problem in social perception and judgment: how an ambiguous interpersonal act is categorized as harassment, why observers disagree about the same behavior, and what disposition and situation jointly predict who harasses. This article reviews the measurement of harassment, Pryor's person-by-situation model of the propensity to harass, gender differences in how behavior is construed, and misperception and sex-role spillover accounts. A worked example traces why the same act is labeled harassment by one observer and dismissed by another.

Keywords: sexual harassment, social perception, person-situation interaction, gender differences, moral disengagement

Introduction

Sexual harassment is unwelcome conduct of a sexual or gender-based nature — ranging from sexist remarks and unwanted attention to coercion and assault — that a target experiences as offensive, threatening, or an obstacle to full participation in a workplace, school, or other setting (McDonald, 2012). It is a social phenomenon with legal definitions, but cognitive psychology approaches it through the mental processes that give those definitions traction: perception, categorization, judgment, and the interaction of a person's dispositions with the situation they occupy.

Two features make harassment a revealing object of cognitive study. First, the same behavior is frequently construed differently by the person who performs it, the person who receives it, and a third party who observes it, so harassment is in part a problem of divergent social perception rather than of fixed physical facts. Second, whether a given individual harasses is poorly predicted by disposition alone or by situation alone; it emerges from their interaction, which is why the field's most durable model is explicitly a person-by-situation account (Pryor et al., 1993). Both features place harassment squarely within the psychology of judgment under ambiguity.

The Medical Subject Headings thesaurus indexes Sexual Harassment beneath both Sexual Behavior and Social Behavior in its classification of behavior and behavior mechanisms. That dual placement is a retrieval convenience rather than a theoretical claim: it locates the topic among its neighbors in the biomedical literature but says nothing about the perceptual and judgmental machinery the sections below describe. This article uses the classification only to situate harassment among related behaviors and otherwise treats the psychology on its own terms.

Defining and Measuring Harassment

Before harassment can be studied it must be defined operationally, and the central cognitive difficulty is that the behaviors in question form a continuum from the plainly benign to the plainly abusive, with a wide ambiguous middle. Fitzgerald, Gelfand, and Drasgow (1995) addressed this by building the measurement of harassment on the target's experience of specific behaviors rather than on whether the target applied the label “harassment.” Their instrument distinguished three conceptually distinct dimensions: gender harassment (sexist or degrading remarks not aimed at cooperation), unwanted sexual attention, and sexual coercion (pressure for sexual cooperation linked to work outcomes). Separating the dimensions mattered because they load differently on perception: gender harassment is the most common and the most contested, while coercion is rarer and more consensually recognized.

That measurement strategy fed a broader theoretical account. Fitzgerald and colleagues (1997) proposed an integrated model in which harassment is a function of organizational climate and job gender context, and in which its effects on the target — psychological, health-related, and work-related — are mediated by how the experience is appraised. Appraisal is the cognitive hinge: the same objective behavior produces different outcomes depending on whether the target construes it as threatening, controllable, and attributable to the harasser rather than to themselves. The model thus embeds a social-cognitive process inside what might otherwise look like a purely situational phenomenon.

The definitional problem also has a legal counterpart that shaped the research. Because harassment law in several jurisdictions asks whether a “reasonable person” would find conduct hostile, and because a large literature suggested that men and women sometimes diverge on exactly that question, courts and researchers debated a “reasonable woman” standard (Blumenthal, 1998). The standard is, in effect, a claim about whose social perception should anchor the categorization — a question the cognitive evidence on perceptual differences, reviewed below, bears on directly.

Person-by-Situation: The Propensity to Harass

The most influential cognitive model of who harasses is explicitly interactionist. Pryor (1987) developed the Likelihood to Sexually Harass (LSH) scale, which presents men with scenarios offering an opportunity to exploit power for sexual ends without penalty and measures the reported likelihood of doing so. High-LSH men are not uniformly harassing; rather, they hold a latent disposition — a cognitive association between social power and sexuality, together with a lack of perspective-taking — that is expressed only under enabling conditions.

Pryor, LaVite, and Stoller (1993) made the conditionality explicit. In their experiments, high-LSH men harassed a female confederate when local cues signaled that harassment was tolerated — for example, when an authority figure modeled inappropriate behavior — but behaved indistinguishably from low-LSH men when the situation signaled disapproval. The disposition and the situation each did little alone; their product predicted behavior. This person-by-situation structure is the model's enduring contribution, and it reframes prevention: changing the local norms that license expression can suppress harassment even without changing the underlying dispositions of those present.

The interactionist account converges with organizational findings. The single strongest antecedent of harassment in a large meta-analysis was organizational climate — the perceived tolerance of harassment — followed by job gender context, both of which are exactly the situational multipliers the LSH model predicts (Willness et al., 2007). The first demonstration makes the interaction concrete, letting the reader set a latent propensity and a situational permissiveness independently and observe that predicted harassing behavior remains low unless both are high.

Perceiving Harassment: Gender and Construal

Because harassment is partly constituted by how a behavior is construed, a long research program has asked whether men and women perceive the same conduct differently. The careful answer is that they do, but modestly and unevenly. Rotundo, Nguyen, and Sackett (2001) meta-analyzed scores of studies and found a reliable gender difference: women were somewhat more likely than men to appraise a given behavior as harassment. Crucially, the difference was not uniform. It was small for behaviors at the extremes — sexual coercion was recognized as harassment by nearly everyone, and plainly innocuous conduct by almost no one — and largest in the ambiguous middle, where the same act supports more than one reasonable construal.

Blumenthal (1998) reached a convergent conclusion in a meta-analysis framed around the legal “reasonable woman” question, finding that the average gender difference in perception, though consistent in direction, was small relative to the variation within each gender. The practical upshot is that gender is a weak predictor of how any particular person will categorize a particular act; individual differences in attitudes, experience, and perspective-taking matter more. The perceptual gap is a genuine phenomenon that has been overstated when treated as a categorical divide rather than a modest shift in a shared distribution.

This pattern is itself a signature of categorization under ambiguity. When a stimulus is clearly on one side of a boundary, observers with different thresholds still agree; disagreement concentrates near the boundary, where small differences in threshold flip the judgment. The second demonstration renders this directly, letting the reader vary the severity of a behavior and watch two observers with different labeling thresholds agree at the extremes and diverge most in the ambiguous middle.

Misperception and Sex-Role Spillover

Not all harassment is driven by a propensity to exploit power; some arises from ordinary errors of social cognition. Stockdale (1993) developed the misperception account, which begins from a well-documented perceptual bias: men, on average, are more likely than women to read friendliness as sexual interest. When this over-sexualized construal of ambiguous social cues goes uncorrected — especially in a person who also lacks perspective-taking or who occupies a permissive setting — it can generate unwanted sexual attention that the actor does not recognize as unwelcome. Misperception does not excuse harassment, but it identifies a distinct cognitive route to it, one that prevention can address by targeting the construal rather than only the motive; reviews of the perpetrator literature treat misperception alongside propensity and situational licensing as complementary routes to the same behavior (Pina et al., 2009).

A complementary account locates the error at the level of social roles rather than momentary cues. Gutek and Morasch (1982) proposed sex-role spillover: the carryover of gender-based expectations into the workplace, where they are irrelevant to the work itself. Spillover is most pronounced where the sex ratio of a job is skewed, so that a person's gender becomes unusually salient — a woman in a heavily male occupation is perceived first through the lens of gender role and only secondarily through her work role, which invites the treatment of her as a sexual object rather than a colleague. The account predicts that harassment exposure tracks the gender composition of the setting, a prediction broadly consistent with the job-gender-context effects seen in later organizational research.

Berdahl (2007) reframed these strands around social status rather than sexual desire, arguing that much harassment functions to protect or assert status within a gender hierarchy — which explains why harassment often targets those who violate gender expectations and is not well explained by attraction alone. The third demonstration models the spillover prediction, letting the reader vary the gender composition of a work group and observe how the salience of gender role, and with it predicted harassment exposure, rises as one's own gender becomes a numerical minority.

Worked Example

Consider a single workplace incident: a manager repeatedly comments on a subordinate's appearance and lingers in her workspace. The cognitive account predicts that whether this is labeled harassment, and with what consequence, depends on perception and appraisal as much as on the acts themselves. Follow three observers.

The actor construes his own behavior through a self-serving and possibly misperceiving lens: he reads the subordinate's politeness as reciprocal interest (the misperception bias), and because the organization has never sanctioned such conduct, the situation signals tolerance — the permissive multiplier in the person-by-situation model (Pryor et al., 1993). The target appraises the same behavior as unwanted sexual attention that threatens her standing and is attributable to the manager; that appraisal, not the behavior alone, drives the psychological and work-related harm (Fitzgerald et al., 1997). A coworker observing the exchange falls somewhere between, and where he falls depends on the ambiguity of the specific acts: comments on appearance sit in the contested middle of the severity continuum, exactly where perceiver differences are largest (Rotundo et al., 2001). Table 1 lays the three construals side by side.

The example makes the section's point exact. The behavior is one set of acts, but there are three categorizations, and the disagreement is not evidence that someone is lying; it is the predictable output of distinct social-cognitive processes operating on an ambiguous stimulus. Prevention and adjudication both have to contend with the fact that harassment lives partly in perception — which is why measurement anchored in specific behaviors, rather than in the contested label, has proved more tractable than asking whether “harassment” occurred.

Discussion

The through-line of the cognitive psychology of sexual harassment is that the phenomenon is constituted jointly by behavior and by its perception, and that both the doing and the construing follow regular psychological principles. Who harasses is predicted by the interaction of a latent propensity with a permissive situation rather than by either alone (Pryor et al., 1993), and the strongest situational multiplier — organizational tolerance — is also the most actionable (Willness et al., 2007). How a behavior is categorized depends on where it falls on a severity continuum and on the perceiver's threshold, with gender a real but modest contributor concentrated in the ambiguous middle (Rotundo et al., 2001; Blumenthal, 1998). And distinct cognitive routes — misperception of social cues (Stockdale, 1993) and sex-role spillover (Gutek & Morasch, 1982) — show that harassment can arise from error as well as from intent. Two decade-spanning syntheses consolidate how these strands fit together, tracing the field's movement from cataloguing behaviors toward modeling the organizational and cognitive processes that produce them (Cortina & Berdahl, 2008; O'Leary-Kelly et al., 2009).

Integrating these pieces changes what prevention should target. Because disposition expresses itself only through permissive situations, altering the local norms that signal tolerance can suppress harassment without first changing the people present — the practical core of the interactionist model. Because appraisal mediates harm, and because categorization is hardest in the ambiguous middle, interventions that clarify norms and anchor judgment in specific behaviors reduce both the incidence and the contested aftermath (Fitzgerald et al., 1997). Berdahl's (2007) status-based reframing adds that harassment is often about enforcing a gender hierarchy rather than pursuing sex, which is why conduct aimed at those who defy gender expectations belongs in the same analysis.

None of this reduces harassment to a misunderstanding. The point of the cognitive analysis is the opposite: by specifying exactly which perceptual and judgmental processes are engaged — propensity, situational licensing, appraisal, categorization near a boundary, misperception, spillover — it identifies where intervention is possible and why disagreement about particular cases is the expected, rather than the anomalous, result.

Current Directions

The most recent synthesis of the field consolidates decades of work and marks a shift in emphasis. Cortina and Areguin (2021) reviewed the organizational literature and argued for understanding harassment primarily as an exercise of power and a mechanism of exclusion — “putting people down and pushing them out” — rather than as misdirected sexual interest, extending Berdahl's status account into a comprehensive framework and foregrounding the harms to targets and organizations alike. The review also documents how much of the phenomenon is gender harassment, the most common and least prosecuted form, which sharpens the earlier measurement insight that the contested middle carries most of the volume.

A second strand quantifies consequences with stronger designs. McLaughlin, Uggen, and Blackstone (2017) followed a cohort longitudinally and found that early-career sexual harassment predicted later financial stress and career disruption as targets changed jobs to escape it, giving the appraisal-mediated harms of the integrated model concrete economic form. A third strand develops the moral-cognition side of perpetration: Page, Pina, and Giner-Sorolla (2016) built and validated a scale of moral disengagement specific to harassment — the cognitive maneuvers, such as euphemistic labeling (“harmless banter”) and diffusion of responsibility, by which people license conduct they would otherwise condemn — linking harassment to the broader psychology of how ordinary moral self-regulation is switched off.

Finally, the field has turned reflexive about its own progress. Quick and McFadyen (2017) asked whether decades of research and policy had actually reduced harassment and concluded that measured prevalence had changed little, pressing researchers toward interventions that engage the situational and cognitive levers the basic science identifies rather than relying on training that targets neither. The open questions now concern which specific changes to climate, accountability, and construal actually move behavior, and how to measure that movement reliably.

Common Misconceptions

“Harassers are simply bad people who would harass anywhere.”

A personal propensity to harass expresses itself as behavior mainly where the situation signals tolerance; the same disposition produces little harassment under disapproving norms, which is why climate is the strongest antecedent (Pryor et al., 1993; Willness et al., 2007).

“Men and women live in completely different perceptual worlds about harassment.”

The gender difference in perceiving harassment is reliable but modest and is largest only for ambiguous behaviors; for clearly severe or clearly benign conduct, perceivers of both genders largely agree (Rotundo et al., 2001; Blumenthal, 1998).

“Sexual harassment is really about sexual desire.”

Much harassment functions to protect or assert status within a gender hierarchy rather than to pursue sex, which is why it frequently targets those who violate gender expectations (Berdahl, 2007; Cortina & Areguin, 2021).

“If the target did not call it harassment, nothing happened.”

Measurement based on specific experienced behaviors reveals harassment that targets do not label as such, and the harm is mediated by appraisal rather than by whether the word is used (Fitzgerald et al., 1995; Fitzgerald et al., 1997).

Glossary

Appraisal.

The cognitive evaluation of an event's threat, controllability, and cause; in the integrated model of harassment, appraisal mediates the behavior's effect on the target's health and work.

Categorization.

The judgment that places a behavior on one side of a boundary, such as harassment versus acceptable conduct; disagreement concentrates near the boundary where the stimulus is ambiguous.

Gender harassment.

Sexist or degrading conduct that conveys hostility based on gender without aiming at sexual cooperation; the most common and most contested dimension of harassment.

Likelihood to Sexually Harass (LSH).

A scale measuring a latent disposition to exploit power for sexual ends when it can be done without penalty; predictive of harassment only under permissive conditions.

Misperception bias.

The tendency, stronger on average in men, to read ambiguous friendliness as sexual interest; a distinct cognitive route to unwanted sexual attention.

Moral disengagement.

Cognitive maneuvers such as euphemistic labeling and diffusion of responsibility that let a person license conduct they would otherwise condemn; measured for harassment specifically.

Organizational climate.

The perceived tolerance of harassment within a setting; the single strongest antecedent of harassment in meta-analysis and the key situational multiplier of disposition.

Person-by-situation interaction.

The principle that behavior is predicted by the product of a personal disposition and an enabling situation rather than by either factor alone; the core of the LSH model.

Quid pro quo.

Harassment in which work outcomes are made contingent on sexual cooperation; corresponds to the sexual-coercion dimension and is the most consensually recognized form.

Reasonable woman standard.

A legal test asking whether a reasonable woman, rather than a generic reasonable person, would find conduct hostile; in effect a claim about whose perception anchors the categorization.

Severity continuum.

The range of harassing behaviors from plainly benign to plainly abusive; perceiver agreement is high at the extremes and lowest in the ambiguous middle.

Sex-role spillover.

The carryover of gender-based role expectations into a workplace where they are irrelevant to the work; most pronounced where the sex ratio of a job is skewed.

Social behavior.

The broad class of interpersonal activity, indexed in MeSH, within which harassment is classified alongside sexual behavior.

Unwanted sexual attention.

Unwelcome verbal or physical conduct of a sexual nature that is not linked to work outcomes; the dimension of harassment most affected by misperception.

Key Researchers

Jennifer L. Berdahl

is a professor at the University of British Columbia whose sex-based harassment theory recast much harassment as the protection of social status within a gender hierarchy rather than the pursuit of sexual interest.

Faculty page - Wikipedia

Lilia M. Cortina

is a professor of psychology and women's and gender studies at the University of Michigan whose recent syntheses frame harassment as an exercise of power and a mechanism of workplace exclusion.

Faculty page - Google Scholar

Louise F. Fitzgerald

is a professor emerita of psychology at the University of Illinois at Urbana-Champaign whose Sexual Experiences Questionnaire and integrated process model anchored the measurement and appraisal-based study of harassment.

Faculty page - Google Scholar

Barbara A. Gutek

is a professor emerita of management at the University of Arizona who originated the sex-role spillover theory, linking harassment exposure to the gender composition of the workplace.

Faculty page

Afroditi Pina

is a reader in forensic psychology at the University of Kent whose work reviews the theory and treatment of harassment and developed a scale of moral disengagement specific to it.

ORCID - Faculty page - Google Scholar

John B. Pryor

is a distinguished professor emeritus of psychology at Illinois State University whose Likelihood to Sexually Harass scale and person-by-situation experiments established the interactionist model of who harasses.

Faculty page

Margaret S. Stockdale

is a professor emeritus of psychology at Indiana University Indianapolis whose misperception model identified the over-sexualized reading of friendliness as a distinct cognitive route to harassment.

ORCID - Faculty page

Chelsea R. Willness

is a professor at the University of Calgary whose meta-analysis identified organizational climate and job gender context as the strongest antecedents of workplace harassment.

Faculty page - Google Scholar

Frequently Asked Questions

What is sexual harassment in psychology?

Sexual harassment is unwelcome conduct of a sexual or gender-based nature that a target experiences as offensive or as an obstacle to full participation in a setting such as a workplace or school. Cognitive psychology studies it as a problem in social perception and judgment — how an ambiguous behavior is categorized and appraised — rather than only as a legal category.

Who is likely to commit sexual harassment?

There is no single type. The best-supported model is interactionist: a latent personal propensity, measured by the Likelihood to Sexually Harass scale, predicts harassment mainly in situations that signal tolerance (Pryor et al., 1993). Either the disposition or the permissive situation alone predicts relatively little; their interaction predicts behavior.

Do men and women perceive harassment differently?

Yes, but the difference is modest. Women are on average somewhat more likely to label a given behavior as harassment, and the gap is largest for ambiguous behaviors and nearly absent for clearly severe or clearly benign ones (Rotundo et al., 2001; Blumenthal, 1998). Individual differences within each gender are larger than the average difference between genders.

What is the misperception theory of harassment?

It holds that some harassment arises from a perceptual bias — reading ambiguous friendliness as sexual interest — that is stronger on average in men (Stockdale, 1993). The account identifies a cognitive route to unwanted sexual attention distinct from the deliberate exploitation of power, and suggests prevention can target the construal itself.

What is sex-role spillover?

Sex-role spillover is the carryover of gender-based expectations into a workplace where they are irrelevant to the job (Gutek & Morasch, 1982). It is most pronounced where the sex ratio of an occupation is skewed, making a person's gender unusually salient and inviting treatment based on gender role rather than work role.

Is sexual harassment about sex or about power?

Contemporary accounts emphasize power and status. Berdahl (2007) argued that much harassment protects status within a gender hierarchy, and recent syntheses frame it as a mechanism of exclusion rather than misdirected desire (Cortina & Areguin, 2021). This explains why harassment often targets those who defy gender expectations.

How is sexual harassment measured?

The most widely used approach asks targets about specific experienced behaviors rather than whether they call the experience “harassment,” distinguishing gender harassment, unwanted sexual attention, and sexual coercion (Fitzgerald et al., 1995). Behavior-based measurement captures harassment that targets may not label as such and avoids the contested word.

Does sexual harassment cause lasting harm?

Yes. Longitudinal research links early-career harassment to later psychological distress and career disruption, with harm mediated by how the experience is appraised (Fitzgerald et al., 1997; McLaughlin et al., 2017). The effects extend beyond the incident to job changes made to escape it.

References