Is There an Idealized Target of Sexual Harassment in the MeToo Era?

Abstract

Evidence suggests that Americans became more sympathetic toward people who experienced sexual harassment as the MeToo movement surged. Yet how comprehensive these shifts in public opinion have been remains unclear. I hypothesize that women who experience workplace sexual harassment are judged against the archetype of an idealized target of sexual harassment and deemed less credible when they fall short. Using data from a novel multifactorial survey experiment, I find that net of other factors, a Black woman is deemed less credible than a white woman. A woman is also deemed less credible when she does not assertively confront the harassment in the moment and when she does not report it to her organization. Further, she is deemed less credible when there are no witnesses and when her alleged harasser has not been publicly accused of harassment by others. Her credibility is not affected by a power disparity with the harasser, the presence of alcohol, or a prior romantic relationship with the harasser. Finally, the more facets of the archetype a target conforms to, the more credible she is perceived to be. These results demonstrate a hierarchy of sexual harassment targets, in which some are deemed more credible than others.

sexual harassment, perfect victim, bias, workplace inequality, credibility

Accounts of sexual harassment were once reported only intermittently in the media, but the MeToo movement has brought a swell of sexual harassment stories to the American public in recent years. This new attention has felt like a breakthrough to some sexual harassment experts. For example, Catherine MacKinnon, a prominent legal scholar whose work has shaped sexual harassment law and theory, argued that the MeToo movement “is eroding the two biggest barriers to ending sexual harassment in law and in life: the disbelief and trivializing dehumanization of its victims” (MacKinnon 2018).

There is some evidence that the MeToo movement has indeed shifted cultural beliefs about people who experience sexual harassment. Although the proportion of Americans who have experienced sexual harassment does not appear to have meaningfully decreased in recent decades (Kearl 2018; Livingston 1982), after the popularization of MeToo, Americans became more likely to agree that sexual harassment is a major problem (Saad 2017). Evidence also suggests a decline in stigmatization of people who report sexual harassment: a series of experiments run in 2017 and 2018 initially identified bias against women who report sexual harassment in the workplace, but that bias faded as the MeToo movement gained traction (Hart 2019). Moreover, people who experience sexual harassment have evinced shifting beliefs: women who experienced sexual harassment following the popularization of MeToo were less likely to experience self-doubt or low self-esteem following harassment (Keplinger et al. 2019).

Yet the MeToo movement’s reach may be limited. Some argue that the movement could generate backlash toward women in the workplace (e.g., Bower 2019). Its impact on public opinion appears to have peaked: Americans’ support for the idea that sexual harassment is a major problem, though still higher than it was in 1998, has already begun to decline, particularly among men (Brenan 2019). Meanwhile, others have pointed to narrow framings of sexual harassment that the movement left unchallenged or may even have reified (e.g., Schultz 2018). Tarana Burke, who used the phrase “Me Too” in activist work against sexual violence for a decade before the MeToo hashtag was popularized by women in Hollywood in 2017, has pointed out that Black women “didn’t see themselves represented in the faces of those who were bravely sharing their [MeToo] stories” and posits that this was because “the stakes were higher for them” (Burke 2021:243). MeToo cases reported in the media also appear formulaic in other ways: they focus on powerful men accused of sexual harassment by multiple subordinate women, often with witnesses who can attest that the harassment took place. These patterns suggest that Americans may be accustomed to a certain kind of person who tells a certain kind of account of sexual harassment, while people whose experiences of sexual harassment deviate from this model may be less legible to the public as plausible targets.

In this paper, I test the critique that the MeToo movement has not overturned a status quo in which some targets of sexual harassment are deemed more credible than others. I hypothesize that targets of workplace sexual harassment continue to be judged against an archetype of what I term the “idealized target” of sexual harassment—meaning the target who embodies cultural beliefs about how sexual harassment plays out. Drawing on insights from the rape and sexual harassment myth literature and patterns from the MeToo movement, I hypothesize eight features of the idealized target archetype: a target who is white, whose harasser holds more power than them in the workplace, who was harassed in a context without alcohol, who had witnesses, who responded assertively to their harasser, who promptly reported the harassment to the company, who was not formerly in a romantic relationship with the harasser, and whose harasser has been publicly accused of sexual harassment by others. I predict that those who do not align with a given feature, compared with those who do, will be less likely to be viewed as credible by others when they share an experience of workplace sexual harassment. In addition, I predict that there is a cumulative effect: with each feature of the idealized target archetype that a sexual harassment target embodies, the target will be perceived as increasingly credible.

To test these hypotheses, I employ a novel multifactorial survey experiment on a YouGov panel constructed from a subset of the Census Bureau’s American Community Survey (ACS) to be representative of the U.S. population across key demographics. Respondents read a brief vignette about unwanted sexual behavior in the workplace in which eight features of the account were randomly and independently varied, and then rated the harassment target’s credibility. I find that five of the eight features indeed shape the target’s perceived credibility in the predicted direction. This provides evidence that, just as there is a “perfect victim” archetype of rape, there is an idealized target archetype of sexual harassment, and those who fall short are deemed less credible.

In what follows, I first outline what is known about experiences of sexual harassment in American workplaces. I then examine Americans’ beliefs about how sexual harassment plays out, and the disjuncture between reality and beliefs. Next, I detail a similar disjuncture between the reality of, and cultural beliefs about, rape, and describe how scholars have developed an archetype of the “perfect victim” of rape to describe the narrowly defined, idealized version of a rape victim that people are most willing to deem credible. I propose that this archetype, and many of the cultural beliefs about rape that uphold it, can be extended to an archetype of the “idealized target” of workplace sexual harassment. Finally, I formalize my predictions in a set of hypotheses.

SEXUAL HARASSMENT IN AMERICAN WORKPLACES

Social scientists define sexual harassment as a three-part construct, comprising sexual coercion, where conditions of employment are made contingent on sexual activity; unwanted sexual attention, meaning unwelcome verbal or physical sexual advances; and gender harassment, meaning verbal or nonverbal behaviors that convey objectification, hostility, exclusion, or second-class status toward members of a gender (NASEM 2018). Lay understandings of sexual harassment vary across time and cultures and tend to be much narrower than the scholarly definition (Marshall 2003; Saguy 2003). Indeed, in surveys that measure the prevalence of sexual harassment, about twice as many people report having experienced behaviors that social scientists classify as sexual harassment compared to those who explicitly say they have experienced sexual harassment, suggesting that about half of people who experience what social scientists consider sexual harassment do not label their experiences that way (Ilies et al. 2003). Because of this discrepancy, the most reliable estimates of sexual harassment—detailed below—come from anonymous survey instruments that ask respondents whether they have experienced a range of behaviors that fit within the social scientific definition of sexual harassment (NASEM 2018).

Workplace sexual harassment is relatively widespread: approximately 40 percent of women and 15 percent of men in the United States report on anonymous surveys that they have experienced unwanted sexual behaviors at work (Kearl 2018), and these estimates likely undercount sexual harassment because they do not capture nonsexual forms of sexual harassment i.e., gender harassment. There are scant data on the rates of workplace sexual harassment for gender minorities, but rates are likely high given that gender minorities experience high rates of sexual harassment in other contexts (e.g., Cantor et al. 2019).

Surveys examining workplace sexual harassment in the United States, which largely focus on women, commonly do not find that rates of sexual harassment vary along racial or ethnic lines (Berdahl and Moore 2006; Bergman and Henning 2008; Kearl 2018), although in some studies racial minority women report experiencing sexual harassment at higher (McLaughlin, Uggen, and Blackstone 2012) or lower (Thomas et al. 2019) rates than white women. Racial minority women often face racial harassment alongside sexual harassment, meaning that, overall, they face greater rates of workplace harassment than white women (Berdahl and Moore 2006).

People who experience sexual harassment manage it with a wide range of strategies. Common responses are to ignore the behavior, avoid the person, or tell the person to stop; only a minority of people who experience sexual harassment—an estimated 7 to 25 percent, depending on industry—report it to their organizations (Firestone and Harris 2003; Fitzgerald, Swan, and Fischer 1995; NASEM 2018). There is no individual strategy that reliably addresses sexual harassment: although people who have experienced sexual harassment sometimes find that strategies such as ignoring, avoiding, confronting, or reporting the person to their organization improve the situation, very often such strategies lead to no change or make the situation worse, and assertive strategies, such as confronting or reporting sexual harassment, carry the risk of retaliation (Diekmann et al. 2013; Firestone and Harris 2003; Marshall 2005). In sum, people are vulnerable to sexual harassment across gender and race and ethnicity, and people who experience sexual harassment manage the situation in a variety of ways.

CULTURAL BELIEFS ABOUT SEXUAL HARASSMENT

Americans have long viewed people who say they have experienced sexual harassment skeptically, endorsing beliefs that people often fabricate or exaggerate sexual harassment, have ulterior motives for reporting it, should individually put an end to the behavior, or should be flattered by the attention (Fitzgerald et al. 1995; Lonsway, Cortina, and Magley 2008). Although some evidence suggests that Americans began to view sexual harassment targets more positively in the MeToo era (Hart 2019), it is plausible that public perceptions of sexual harassment remained narrowly defined. A disjuncture between limited cultural narratives about how sexual harassment plays out and its highly variable reality, I argue, has implications for who is believed when they recount an experience of sexual harassment.

It is well-documented that perceptions of people who experience sexual harassment are shaped by how they managed the harassment. Although ignoring the behavior or avoiding rather than confronting the person responsible are very common responses (Firestone and Harris 2003; Fitzgerald et al. 1995), when people speculate about how they might respond if harassed, they tend to imagine that they would angrily confront unwanted sexual behavior and, therefore, view those who do not as somehow complicit in “allowing” the harassment to happen (Diekmann et al. 2013; Woodzicka and LaFrance 2001). Similarly, although only a minority of sexual harassment targets report it to their organizations, prior research finds that people who do not immediately report are viewed as less credible and more blameworthy (Balogh et al. 2003). Notably, however, this research precedes the MeToo movement, which gave voice to many people who did not immediately confront or report the sexual harassment they experienced. It is possible, then, that greater visibility of the varied ways that people manage sexual harassment may have decreased the extent to which sexual harassment targets’ credibility is judged by their responses to the harassment.

While the MeToo movement may have tempered judgements of sexual harassment targets in some respects, it may also have played a role in further shaping which sexual harassment targets are legible to Americans. In media coverage throughout the MeToo movement, stories of sexual harassment have largely focused on powerful men accused of sexual harassment by multiple women, often with witnesses who could attest that the harassment occurred. Given that the media that people consume about sexual violence shapes their understanding of how sexual violence plays out in daily life (Hust et al. 2015; McDonald and Charlesworth 2013), prominent themes in stories told throughout the MeToo movement may have shaped or reified cultural beliefs about sexual harassment. Specifically, media coverage in the MeToo era may have lent relatively greater credibility to people who were sexually harassed by those who outranked them, who had witnesses, and whose harasser faced other allegations, although many who experience sexual harassment do not fit this description.

EXTENDING THE CONCEPT OF THE “PERFECT VICTIM” OF RAPE TO THE SEXUAL HARASSMENT LITERATURE

In this study, I suggest that consequential gaps remain between the wide-ranging and variable experiences of workplace sexual harassment and the more narrowly defined cultural ideas of how sexual harassment is thought to play out, despite heightened public attention to sexual harassment in the MeToo era. In doing so, I draw from the rape literature, in which scholars have theorized an archetypal “perfect victim” or “ideal victim” of rape who is lent credibility for fitting cultural stereotypes of how rape plays out (Armstrong, Gleckman-Krut, and Johnson 2018; Christie 1986; Crenshaw 1991; Freedman 2013). I extend this concept to the sexual harassment literature, terming the hypothesized archetype who fits cultural narratives of sexual harassment the “idealized target” of sexual harassment. I hypothesize that when a person experiences workplace sexual harassment that falls outside the bounds of cultural narratives about how sexual harassment transpires, that person is deemed less credible.

Rape and sexual harassment are related phenomena, and scholars have found that the cultural beliefs that serve to downplay the prevalence and severity of sexual harassment in the United States are very similar to those that downplay the issue of rape (Lonsway et al. 2008). Rape and sexual harassment myths—both of which center around women being victimized by men—contend that women often falsely report rape or sexual harassment, that women are motivated to report rape or sexual harassment for personal gain, that women actually enjoy rape or sexual harassment, and that women are at fault for rape or sexual harassment because they should have prevented it from occurring (Lonsway et al. 2008; Payne, Lonsway, and Fitzgerald 1999). It is logical that the cultural beliefs that delegitimize individual targets of sexual harassment would likewise overlap with those that delegitimize victims of rape.¹ Therefore, in addition to predictions drawn from the sexual harassment literature that I lay out above, I draw on scholarship about the perfect victim of rape to make three further predictions about the idealized target of sexual harassment.

Racialized stereotypes about women’s sexual restraint have shaped the extent to which women of different races are seen as sympathetic victims of sexual violence. Historically, white women have been typecast as sexually “pure” and restrained, while Black women have been assumed to be hypersexual “bad girls,” and these stereotypes have carried through to the present day (Collins 2004; Freedman 2013; Malone Gonzalez 2022; Ritchie 2017). Latina women are similarly typecast as sexually unrestrained “spitfires” (Khan 2019), such that Latina teenage girls are viewed as “perpetually at risk for pregnancy because of a ‘Latino culture’” (García 2009:536). These stereotypes of hypersexuality are consequential, because when women are thought to be sexually voracious and always desirous of sex, the sexual violence they experience tends to be construed as consensual. As Kimberlé Crenshaw describes, “Black women are essentially prepackaged as bad women within cultural narratives about good women who can be raped and bad women who cannot” (1991:1271). Indeed, this view of Black women as women who cannot be raped was once enshrined in American law: in some Southern states in the 19^th century, laws specified that only white women could legally be recognized as victims of rape (Freedman 2013). Even after the repeal of such laws, Black women who experience rape have continued to face greater skepticism and less sympathy from the police and courts relative to white women (Armstrong et al. 2018; Powell and Phelps 2021).

There is reason to believe that racial minority women are trivialized when they experience sexual harassment, as is the case when they experience rape. Cultural beliefs about sexual harassment, like rape, often position women who are perceived as hypersexual as having invited or enjoyed the harassment they experience (Lonsway et al. 2008; Payne et al. 1999); therefore, it follows that a sexual harassment target perceived as hypersexual because of racial stereotypes would be doubted.

In addition, I posit that if alcohol was involved in a target’s sexual harassment experience, this might shape the target’s perceived credibility. Cases of rape involving alcohol are often construed as misunderstandings, distortions, or instances where consent was implicitly given, rather than “real” rape: studies have shown that rape victims who were intoxicated at the time of their assault are viewed as more to blame for the rape, and police are often apathetic about making arrests in such cases (Bieneck and Krahé 2011; Krahé, Temkin, and Bieneck 2007). I predict that a similar pattern can be found with cases of sexual harassment: that a target of harassment will be perceived with greater skepticism if alcohol was present during the incident.

Finally, I suggest that the target’s prior relationship with their harasser is consequential. In the rape literature, it is well-documented that an account of rape is viewed skeptically if the rapist was an ex-partner or had otherwise previously been sexually intimate with the victim; in these cases, the victim may either be penalized for not having been sufficiently chaste prior to the attack, or delegitimized through the view that sexual consent cannot truly be revoked once it has been given (Armstrong et al. 2018; Bieneck and Krahé 2011; Krahé et al. 2007). The effect of being harassed by a former romantic or sexual partner in the domain of workplace sexual harassment has received less scholarly attention. I hypothesize that the same pattern may apply: that a person who has experienced sexual harassment will be perceived as less credible if they have a romantic history with their harasser.

IS THERE AN IDEALIZED TARGET OF SEXUAL HARASSMENT?

I have formed eight predictions about how targets of sexual harassment are perceived relative to the idealized target of sexual harassment. The predictions are as follows:

 

H1a: A Black or Latina sexual harassment target will be perceived as less credible than a white sexual harassment target.

 

H1b: A sexual harassment target who was harassed by a peer will be perceived as less credible than a sexual harassment target who was harassed by a superordinate.

 

H1c: A sexual harassment target who was harassed at an event with alcohol will be perceived as less credible than a sexual harassment target who was harassed at an event without alcohol.

 

H1d: A sexual harassment target without witnesses will be perceived as less credible than a sexual harassment target with witnesses.

 

H1e: A sexual harassment target who responded passively will be perceived as less credible than a sexual harassment target who responded assertively.

 

H1f: A sexual harassment target who did not report the incident to HR will be perceived as less credible than a sexual harassment target who did report.

 

H1g: A sexual harassment target who previously dated, or went on one date with, the harasser will be perceived as less credible than a sexual harassment target who had never met the harasser before.

 

H1h: A sexual harassment target whose harasser is not facing any other harassment allegations will be perceived as less credible than a sexual harassment target whose harasser is facing other allegations.

Importantly, these predictions should not be interpreted as a complete set of all features that might characterize the idealized target. Rather, I use this broad set of features, drawn from sexual harassment myths, rape myths, and patterns from MeToo media accounts, to test for evidence of a hierarchy of sexual harassment targets in the contemporary United States who are deemed less credible relative to the multifaceted archetype of the idealized target of sexual harassment.

I predict that failure to align with the eight hypothesized features of the idealized target of sexual harassment will each individually bring a credibility penalty to a person who has experienced sexual harassment, compared to a person who does align with each feature; however, I also predict that these features will have a cumulative effect. Specifically, I predict that:

 

H2: A sexual harassment target will be perceived as more credible with each additional feature of their experience that aligns with the idealized target archetype.

The present experiment builds on prior experimental vignette studies that test perceptions of sexual harassment targets. Some of these prior studies test perceptions that are beyond the purview of this paper, such as the likelihood that people conceptualize an incident as sexual harassment (Weinberg and Nelson 2017) or blame a target (Klein, Apple, and Khan 2011). But some prior vignette studies have taken up the focus of this paper: whether there are features that shape the extent to which a person is viewed as a credible target of sexual harassment. These studies examined two individual facets of what I conceptualize as the idealized target archetype. They find that responding assertively, rather than passively, is seen as the most plausible response to sexual harassment (Diekmann et al. 2013, Woodzicka and LaFrance 2001), and that targets are indeed deemed more credible when they immediately report harassment to their organization (Balogh et al. 2003). Thus, there is support for the idea that prior to the MeToo movement, targets were perceived negatively when their experiences did not align with cultural beliefs about sexual harassment.

Although I draw methodological inspiration from prior vignette studies about perceptions of sexual harassment targets, I depart from them in four ways. First, I simultaneously vary an array of features that are hypothesized to shape the perceived credibility of the idealized target. This allows me to test the archetype of the idealized target as a multifaceted construct, and offers a direct comparison of the magnitude of effects of different hypothesized features. Second, features such as the target’s race/ethnicity and the presence of alcohol have not previously been manipulated in experimental vignettes examining targets’ perceived credibility. Third, this study was conducted after the MeToo movement became mainstream, allowing me to test whether cultural beliefs about sexual harassment continue to shape perceptions of targets’ credibility, and whether patterns found in media stories in the MeToo era may reflect how targets are perceived. Finally, where prior studies have tested how sexual harassment targets are perceived by university students or nonrepresentative online panels, I employ a high-quality survey panel that is representative of the U.S. population across key demographics, allowing me to broadly generalize about how Americans perceive sexual harassment targets after the popularization of MeToo.

DATA AND METHODS

I use an original survey experiment to investigate the characteristics of the idealized target of sexual harassment. Survey experiments are an increasingly popular methodological strategy in sociological research because they allow the researcher to make causal claims. By randomly assigning respondents to view different experimental treatments while holding everything other than the treatments constant, I am able to show the causal effect of those treatments: in this case, how aspects of an alleged sexual harassment incident affect how credible the harassment target is perceived to be.

This study uses data from a survey of 1,000 respondents conducted in July of 2020 by YouGov. The sample is nationally representative of the U.S. population on the characteristics of gender, age, race, and education, using stratified sampling from the 2017 American Community Survey 1-year sample. The data are thus well-suited to evaluating whether Americans favor a sexual harassment target who embodies the idealized target archetype I theorize. Respondents’ demographic characteristics are reported in the online supplemental materials.

Experimental Design

I employed a multifactorial vignette experiment to simultaneously test whether each of the eight hypothesized features of the idealized target shapes how she is perceived by respondents. An advantage of experiments that vary numerous factors simultaneously is that they are well-suited to capturing forms of discrimination that respondents may otherwise be motivated to conceal due to social desirability bias, such as racial animus. Multifactorial experiments offer many varied characteristics, so the respondent can often attribute an unfavorable impression to something other than a trait that they may feel inclined to discriminate on when doing so is socially undesirable (Schacter 2016; Wallander 2009). Each respondent saw only one vignette.

I embed the experimental treatments—the eight hypothesized features of the idealized target—within a brief vignette written to mimic a journalistic account (see supplemental materials). I hold constant the fictitious organization, the role of the person being harassed, and, importantly, the sexually harassing behavior that occurred. Each respondent read that Richard Thompson “placed his hand on [the target’s] buttocks, and suggested they leave the event to go to a hotel room together.” Thus, all respondents were reacting to the same behavior. However, although this continuity allows me to isolate the effects of the features of the idealized target while holding the harassing behavior constant, it also means that I can evaluate perceptions of only one harassment type: unwanted sexual advances. Further research should examine whether the patterns found here hold for other forms of sexual harassment, such as gender harassment.

Another feature held constant across conditions is that the vignette describes a woman who has experienced sexual harassment from a man; thus, this experiment can speak only to how women targets of harassment are perceived. Men and nonbinary people also experience harassment; I opted to focus on women in this study because the cultural beliefs about sexual harassment on which the hypothesized features of the idealized target are based center on women being harassed by men (Lonsway et al. 2008), by far the most common configuration of sexual harassment (e.g., Kearl 2018). Notably, recent research has not found significant differences in how women versus men targets are perceived, nor women versus men harassers (Pica, Sheahan, and Pozzulo 2019).

Each experimental treatment was randomly and independently varied within the vignette; possible values for each of the eight treatments are presented in Table 1. Race/ethnicity is manipulated by using names that are widely perceived to belong to Black, Latina, and white women (Gaddis 2017a and Gaddis 2017b); for brevity, I refer to each target as the Black target, Latina target, and white target. Most respondents perceived Tanisha Washington as Black (74 percent), Mariana Velazquez as Latina (82 percent), and Katelyn Walsh as white (88 percent).

Table 1.

Features of the Sexual Harassment Target Presented in the Vignette

	Feature	Categories	Operationalization used in the vignette
1	Target’s race/ethnicity	Black	Tanisha Washington
		Latina	Mariana Velazquez
		White	Katelyn Walsh
2	Power disparity	No power disparity	[Harasser is] an accountant
		Harasser outranks harassee	[Harasser is] the director of accounting
3	Presence of alcohol	Alcohol present	The alleged incident occurred at a company happy hour
		No alcohol present	The alleged incident occurred at a company lunch
4	Witnesses	No witnesses	Other employees present at the event did not witness the incident, so her account could not be corroborated.
		Witnesses	Two other employees also described observing the incident, corroborating her account.
5	Target’s immediate response	Passive response	She does not recall saying anything in response.
		Active response	She said she pushed his hand away and told him she had no interest in doing so [going to a hotel room].
6	Target’s decision to report	Not reported	[She] did not subsequently report the incident to her company’s HR department.
		Reported	[She] reported the incident to her company’s HR department the next day.
7	Prior romantic relationship	Target and harasser briefly dated	She said that she had known Thompson before the incident: they had briefly dated in the past, but their relationship had been over for six months when the incident occurred.
		Target and harasser went on one date	She said that she had met Thompson before the incident: they had matched on a dating app six months previously and gone on one date but had not interacted since.
		No prior relationship	She said that she had never interacted with Thompson before.
8	Other allegations	No other allegations	No other employees have come forward with allegations against Thompson.
		Two other allegations	This is the third allegation of sexual harassment against Thompson.

Key Variables

After reading the vignette, respondents were asked to evaluate the sexual harassment target’s credibility with the question: “To what extent do you believe [name]’s account?” using a seven-point scale ranging from (1) “Doubt a great deal” to (7) “Believe a great deal” (see supplemental materials). I also captured two other perceptions of the harassment incident: the extent to which respondents perceived the harassment incident to be serious and the target to be a desirable employee. Because they are less central to my focus on the legibility of sexual harassment targets, I present analyses of these constructs in the supplemental materials.

The key predictor variables used to test Hypotheses 1a-1h are the eight hypothesized features of the idealized target that were experimentally manipulated (see Table 1). The key predictor variable used test Hypothesis 2 is a continuous variable that captures the number of hypothesized features of the idealized target that the protagonist of each randomized sexual harassment vignette aligns with. This variable ranges from 0 (no attributes of the idealized target archetype) to 8 (all the attributes of the idealized target archetype). I constructed the variable this way to directly test Hypothesis 2, that a sexual harassment target will be perceived as more credible with each additional feature of their experience that aligns with the idealized target archetype. However, in the supplemental materials, I explore the relationship between credibility and increasing alignment with the idealized target archetype when the variable is constructed in alternate ways—using only those features found to independently shape the target’s perceived credibility, and splitting the variable into the three dimensions of target characteristics, harasser characteristics, and contextual features.

Analytic Approach

The survey was taken by 1,000 respondents. I limit the analytic sample to the 926 respondents who indicated that they had read the vignette by correctly answering the manipulation check, but results are very similar when I estimate the same regressions using the full sample (see supplemental materials).

Multifactorial experiments that vary multiple conditions simultaneously have many possible configurations: in this study, there are 576 possible configurations of the eight treatments. Effects are computed with the average marginal component effect (AMCE), which estimates the effect of each treatment as an average over all possible combinations of the other treatments (Hainmueller, Hopkins, and Yamamoto 2014). This allows me to isolate the independent effect of any single treatment net of the others.

Because the dependent variable credibility was measured on an ordinal scale, I use ordered logistic regression to regress the outcome on the eight treatments. I tested the proportional odds assumption underlying the ordered logistic models using the Brant test, which indicated that I did not violate the assumption in either set of analyses. I present results in odds ratios, obtained by exponentiating the regression coefficients, for ease of interpretation. I selected reference categories for my predictor variables based on my theoretical predictions (Johfre and Freese 2021): the conditions I use as reference categories are those that I predict are farthest from the archetype of the idealized target.

RESULTS

This empirical analysis first examines the extent to which each of the eight hypothesized features of the idealized target individually shape a sexual harassment target’s perceived credibility. In Table 2, ordered logistic regression results demonstrate that five hypothesized features of the idealized target do significantly impact target’s perceived credibility in the predicted direction. First, the race/ethnicity signaled by the sexual harassment target’s name significantly impacts her perceived credibility: the white sexual harassment target has 35 percent greater odds of being viewed as more credible relative to the Black target.²(Odds ratios for ordered logistic regressions can be interpreted as the odds of selecting higher-value outcomes of an ordinal dependent variable relative to selecting lower-value outcomes (Long and Freese 2014).)

Table 2:

Ordered Logistic Regression Model of the Effect of the Hypothesized Features of the Idealized Target on Perceived Credibility, in Odds Ratios

	Model 1: Credibility of target
Name of target (ref = Tanisha Washington)
Mariana Velazquez	1.13
	(0.17)
Katelyn Walsh	1.35*
	(0.20)
Alleged harasser’s job (ref = accountant)
Director of accounting	1.14
	(0.14)
Presence of alcohol (ref = happy hour)
Lunch	0.85
	(0.10)
Witnesses (ref = no witnesses)
Two witnesses	2.90***
	(0.36)
Immediate response (ref = passive response)
Assertive response	1.66***
	(0.20)
Decision to report (ref = not reported)
Reported	1.51***
	(0.18)
Prior romantic relationship (ref = dated in the past)
Went on one date in the past	0.98
	(0.15)
Never met	0.89
	(0.13)
Other allegations (ref = none)
2 other women also accused him	1.52***
	(0.18)
Observations	926

Standard errors in parentheses. *** p<0.001, ** p<0.01, * p<0.05. Cut points not shown. Credibility is measured on a scale from 1-7 where 1 is doubt a great deal and 7 is believe a great deal.

This effect, like all AMCEs, is obtained while holding the seven other features constant, indicating that the lower credibility ascribed to the Black target holds net of other variations in the account of sexual harassment. There is no statistically significant difference between the perceived credibility of the Latina target and the Black target. When I change the reference category to directly compare the difference in the perceived credibility between the Latina target and the white target, I also find no statistically significant difference.

Figure 1 displays the probability, as predicted by the model in Table 2, that respondents will select each of the points on the seven-point credibility scale, depending on the target’s perceived race/ethnicity. All three targets have a similar likelihood of having their credibility doubted (i.e., of having respondents select the options “doubt a great deal,” “doubt a moderate amount,” or “doubt a little more than believe”). The white target has a greater predicted probability of respondents selecting “believe a great deal” relative to the Black target (a predicted probability of 0.36 versus 0.30), whereas the Black target has a greater probability of respondents selecting “somewhat believe and somewhat doubt” or “believe a little more than doubt” relative to the white target. The evaluations of the Latina target’s credibility fall in between those of the Black and white targets for each of these categories.

Figure 1.

Predicted probability of being perceived as credible, by race/ethnicity. Note: N=926. Data were collected on YouGov.

Four other features also significantly predict the target’s perceived credibility as hypothesized. The largest effect is that of the presence of witnesses: when there are witnesses to a sexual harassment incident, the target of sexual harassment has 190 percent greater odds of being viewed as more credible relative to one when the incident was not witnessed. The target has 66 percent greater odds of being viewed as more credible when her immediate response was to react assertively, rather than passively, to the harassment. Likewise, the decision to report significantly shaped the target’s perceived credibility: she has 51 percent greater odds of being viewed as more credible when she reported the harassment to the human resources department, rather than not reporting it. Finally, she has 52 percent greater odds of being viewed as more credible when there are other allegations against the man she has accused of sexual harassment, rather than when there are no other allegations.

Three of the hypothesized features of the idealized target of sexual harassment do not significantly impact the target’s perceived credibility. The power disparity between the harasser and target of harassment (with the harasser as either a higher-status director of accounting or a similar-status accountant), the presence of alcohol (presumably present at a happy hour versus not at a company lunch), and whether the harasser and target had a prior romantic relationship (had previously dated, gone on one date, or had never met) did not significantly impact the extent to which respondents rated the target credible.

In results presented in the supplemental materials, I also test whether the eight hypothesized features of the idealized target significantly shape two other perceptions of the idealized target. Whether the target had a prior romantic relationship with the harasser significantly shaped the perceived seriousness of the incident, as did the presence of witnesses or other allegations. The target was perceived as a more desirable employee when she responded assertively, when there were witnesses, and when there were other allegations.

The next analysis evaluates Hypothesis 2, whether the eight features matter in combination. I constructed a continuous variable representing the number of idealized target features that the protagonist of a given vignette matched.³ I used all eight of the hypothesized features to construct this variable (but see the supplemental materials for analyses using alternate constructions of this variable). For example, in a vignette in which the target was white, the incident occurred at a happy hour, the harasser did not outrank the target, there were no witnesses, the target responded assertively, the target did not report the incident, the target previously dated the harasser, and no one else had publicly accused the harasser, the number of idealized target features would be two, because two features of the hypothesized idealized target—being white and responding assertively—are present.

In Table 3, 3I regressed the target’s perceived credibility on the number of idealized target features. There was, indeed, a significant relationship: for each feature of the idealized target archetype that the protagonist of the vignette aligns with, she has 39 percent greater odds of being viewed as more credible.⁴ Thus, the evidence supports Hypothesis 2. In results presented in the supplemental analyses, I show that this same pattern holds for the perceived seriousness of the incident and perceived employability of the target.

Table 3:

Ordered Logistic Regression Models of the Number of Hypothesized Features Aligning with the Idealized Target Archetype on Perceived Credibility, in Odds Ratios

	Model 1: Credibility of target
Number of idealized target features	1.39***
	(0.06)
Observations	926

Standard errors in parentheses. *** p<0.001, ** p<0.01, * p<0.05. Cut points not shown. Credibility is measured on a scale from 1-7 where 1 is doubt a great deal and 7 is believe a great deal. Number of idealized target features is a continuous variable measured from 0 to 8.

DISCUSSION AND CONCLUSION

In this study, I hypothesized that there is an idealized target of workplace sexual harassment who is privileged relative to other sexual harassment targets, and that those who fall short of the archetype are deemed less credible. I find evidence that five of the hypothesized features of the idealized target impact perceived credibility as predicted, indicating that there is a multifaceted archetype of the idealized target of sexual harassment, even in the MeToo era. Furthermore, people whose experiences conform to the archetype along an increasing number of dimensions are viewed as increasingly believable.

Interpreting the Features of the Idealized Target of Sexual Harassment

Scholars have long observed that Black women who experience sexual violence are treated more skeptically than white women (Collins 2004; Crenshaw 1991; Freedman 2013). This study provides causal evidence of the claim: even when all other aspects of a sexual harassment incident are held constant, a woman whose name reads as Black (Tanisha Washington) is less likely than a woman whose name reads as white (Katelyn Walsh) to be viewed as credible. U.S. laws no longer make it legally impossible for Black women to be viewed as victims of sexual violence, yet this finding indicates that in public perception, Black women remain less legible as victims of sexual violence than white women. Thus, as Tarana Burke posited, the stakes are indeed higher for Black women sharing accounts of sexual harassment, even in the MeToo era. It is particularly ironic that Black women face this credibility penalty because Black women in the United States—like Burke herself—have historically been at the forefront of efforts to address sexual violence (McGuire 2010).

In post-hoc analyses found in the supplemental materials, I find that respondents’ own race shaped their perceptions of the Black, Latina, and white targets of sexual harassment. First, the credibility bias favoring a white woman over a Black woman is driven primarily by racial minority respondents. In addition, whereas in the full sample the Latina woman’s perceived credibility was not significantly different from that of either the Black woman or the white woman, I find that racial minority respondents—but not white respondents—viewed the Latina woman as significantly less credible than the white woman. It is possible that white respondents may have been more sensitive than racial minority respondents to concerns about being seen as racially biased, such that social desirability bias disproportionately impacted white respondents’ reported perceptions of the Black and Latina target. Alternately, it may be the case that racial minority respondents hold the Black and Latina targets to a higher standard than the white target when assessing credibility. Prior experimental work has occasionally found that marginalized groups more strongly express biases that uphold inequality: for example, that women respondents are more likely than men respondents to express gender bias toward women who are parents (Benard and Correll 2010).

Consistent with prior research conducted before the MeToo era (Balogh et al. 2003; Diekmann et al. 2013; Woodzicka and LaFrance 2001), I find that the way a person responds to sexual harassment remains influential in shaping their perceived credibility. The target of sexual harassment was viewed as significantly less credible when she did not respond assertively to sexual harassment. Likewise, she was viewed as significantly less credible when she did not report the harassment to HR. It is notable that these features continue to significantly shape sexual harassment targets’ credibility, given the visibility of accounts told throughout the MeToo movement in which targets did not respond assertively or report the incident. Confronting sexual harassment in the moment and reporting it to the organization may emphasize the unwanted nature of the incident, thereby heightening the person’s credibility as a target of sexual behavior that is clearly unwanted. Nevertheless, most sexual harassment targets neither report harassment nor respond assertively (Diekmann et al. 2013; NASEM 2018). Instead, when people experience sexual harassment, they often use strategies that minimize the risk of retaliation, such as ignoring or deflecting the behavior or avoiding the person responsible (Firestone and Harris 2003), responses that, though common, appear to jeopardize their perceived credibility.

Finally, I find that two features that were prominent in media stories throughout the MeToo movement significantly shaped the target’s credibility: the presence of witnesses and other allegations against the harasser. It is unsurprising that the presence of witnesses or other allegations shape the target’s perceived credibility, given that they provide additional evidentiary support beyond the account of the target. In addition, cases with additional witnesses or additional allegations have been represented heavily in media reports about sexual harassment in recent years; cases that mimic those found in media reports may have become familiar to Americans, and thus seen as more credible. In other words, the MeToo movement may have reified these characteristics of the idealized target.

Three hypothesized features of the idealized target of sexual harassment did not significantly shape perceptions of the target’s credibility. Respondents were no more likely to view the target as credible when her harasser was a lower rank (accountant) versus a higher rank (director of accounting). It may be that rank is simply not very important when people assess whether they believe the incident truly occurred; consistent with this, some prior work suggests that the harasser’s rank does not impact how sexual harassment cases are resolved (Welsh, Dawson, and Nierobisz 2002). Or, it may be that absolute rank matters less than rank within the chain of command. The harassment target in the vignette was in a different department (sales), and so respondents may have assumed that despite the power differential, the director of accounting could not use his position to coerce her. Alternately, it may be the case that the manipulation did not convey the rank difference clearly enough: respondents may have thought that a sales associate was lower-ranking than either an accountant or a director of accounting, thus perceiving a power differential in both cases.

Whether or not the harassment occurred at an event that implied the presence of alcohol (a happy hour) or did not (a lunch) also did not significantly shape respondents’ perceptions of the target’s credibility. One explanation is that while there is a familiar narrative that alcohol makes the accounts of rape victims unreliable (particularly at universities, where binge drinking tends to be widespread), the idea that workplace sexual harassment is a result of drunken misunderstandings is not widely invoked, and, therefore, consuming alcohol may not undermine a sexual harassment target’s credibility in the same way. Additionally, people may not view an office happy hour as an event at which employees are likely to get very intoxicated.

Finally, respondents did not view the target as any less credible when she had previously been romantically involved with the harasser. However, in analyses presented in the supplemental materials, I find that respondents were less likely to view the sexual harassment incident as serious when the target had previously dated the harasser. This nuance indicates that although a person is just as likely to be believed when they describe experiencing sexual harassment from someone with whom they were previously romantically entangled, the significance of the incident may nonetheless be trivialized.

This paper focuses on the fundamental question of how sexual harassment targets’ credibility is perceived, but I also asked how serious respondents perceived the incident to be, and how hirable they perceived the target to be. I find some continuities across all three variables—in each case, the target was more likely to be perceived as more credible if there were witnesses or other allegations. However, responding assertively shaped only the target’s perceived credibility and hireability, and the target’s having previously dated the harasser shaped only the perceived seriousness of the incident. These heterogeneous effects indicate that different facets of the idealized target archetype are relevant for different judgments about the target.

Broader Contributions and Future Directions

This study extends scholarship on cultural beliefs about victims in the United States by illustrating that, although there is a general stigma against people who are victims (Hegtvedt and Johnson 2000; Hunzaker 2014), the stigma associated with being a victim may be gradational rather than absolute. Using the case of workplace sexual harassment, I demonstrate a hierarchy in how people who experience sexual harassment are perceived, showing that targets of sexual harassment who do not match an idealized target archetype are viewed as less credible. The idealized target of sexual harassment closely parallels the archetype of the perfect victim of rape that has already been well-documented in the literature (Armstrong et al. 2018; Christie 1986; Crenshaw 1991; Freedman 2013). This continuity indicates that across different types of sexual violence, some people who are victimized are viewed as more credible than others.

The credibility penalty experienced by women who differ from the idealized target archetype most obviously impacts their experiences interacting with others: they are more likely to be greeted with skepticism by others when recounting an experience of sexual harassment. Yet this finding also has implications for how sexual harassment cases will be adjudicated by employers and courts, because both are comprised of human decision-makers whose evaluations are shaped by the cultural beliefs they hold (see, for example, Weinberg and Nielson 2017). Just as legal and therapeutic institutions demand that survivors of domestic violence fit a culturally legible version of a domestic violence survivor (Sweet 2021), the results of this study suggest that institutions that adjudicate sexual harassment will expect people who have experienced sexual harassment to adhere to the idealized target archetype and penalize those who do not. Thus, targets who stray from the archetype—such as Black women or women who do not immediately report their harassment—will face longer odds of convincing institutions that their harassment occurred.

That reporting sexual harassment to one’s organization is itself a facet of the idealized target archetype creates a particular bind for targets of sexual harassment who do not align with the archetype in other ways. Although the act of reporting would, in theory, bring the person closer in alignment with the archetype, reporting sexual harassment that is not deemed credible—which, as I demonstrate, is more likely for those who stray from the archetype in other ways—is likely to generate backlash and retaliation. Indeed, prior research has found that organizations see declines in the share of women—and particularly racial minority women—in management after implementing grievance procedures for discriminatory treatment (Dobbin, Schrage, and Kalev 2015), perhaps because of such ensuing backlash and retaliation.

The greater skepticism that targets of sexual harassment who stray from the idealized target archetype face may ultimately reinforce the archetype, such that the cultural ideal is self-sustaining. That is, people who stray from the idealized target archetype may be less likely to publicly speak out against sexual harassment they have experienced and receive less media attention, or unflattering media attention, when they do. In Tarana Burke’s words, “the stakes [are] higher” for people who are not already legible targets of sexual harassment. Minimal public attention toward people who stray from the archetype, in turn, may reinforce cultural narratives that such cases are not credible.

The findings of this study can be tied to theoretical work on how femininities are performed, evaluated, and rewarded. The gender order, in which women are subordinated relative to men in society, cannot be conceptualized as a freestanding system; instead, it interlocks with identities such as race, class, and sexuality to comprise a “matrix of domination” (Collins 2004; see also Crenshaw 1991). Those who occupy locations in the matrix of domination that are disadvantaged in terms of gender but advantaged in other respects—such as white women, who are disadvantaged by gender and advantaged by race—are incentivized to hew to a gender performance that emphasizes their advantaged identities along other dimensions. Doing so offers a femininity premium: “a set of individual benefits that accrue to those who can approximate these [advantaged identity] ideals” (Hamilton et al. 2019:316). The greater credibility afforded to white women relative to Black women recounting sexual harassment might be conceptualized as one facet of the femininity premium awarded to women who embody white femininity. Indeed, in this study, respondents did not receive cues about the harassment target’s racialized gender performance, and appeared to award white women a femininity premium—enhanced credibility—based on their whiteness alone.

In this study, I test whether eight features of a sexual harassment incident shape how the person being harassed is perceived, but these features should not be conceptualized as a complete picture of the idealized target of sexual harassment; further work is needed to build out our understanding of the archetype in the MeToo era. For example, I did not vary the perceived race of the harasser, and the race of the target and harasser may jointly shape how sexual harassment is perceived (see Freedman 2013).⁴ Moreover, research has not yet explored how gender nonbinary targets and harassers are perceived. A sexual harassment target’s class position or sector of work may also be consequential. These nuances in how race, gender, and class may shape perceptions of sexual violence warrant further research.

The scope of this research could also be expanded in other ways. The case of harassment that I portrayed in the vignette was sexual in nature; further research should expand the scope to other types of sexual harassment, for example by examining how targets are perceived when they have experienced nonsexual gender harassment, which though widespread, is often trivialized (Saguy and Rees 2021; Schultz 2018). In addition, in this study I focus on perceptions of sexual harassment targets in the U.S. context; future research should examine this issue in other cultural contexts.

The MeToo movement has in some ways altered the perceptions of sexual harassment in the United States and has ushered in new laws and policies to address sexual harassment (Beitsch 2018; Brenan 2019; Hart 2019; Keplinger et al. 2019; McGregor 2018). Yet this study indicates that despite progress, cultural ideals still support a hierarchy of sexual harassment targets in which some are deemed more credible than others. There remains an archetype of the idealized target of sexual harassment, yet the majority of people who experience sexual harassment do not embody this ideal, and are deemed less credible as a result.

Acknowledgements

This work was supported by the National Science Foundation [Award Number 2001736]. I would like to thank Shelley Correll, Cecilia Ridgeway, Aliya Saperstein, David Pedulla, Christine Schwartz, and Chaeyoon Lim for comments on this article.

Footnotes

1

How to describe people who have experienced sexual violence is a fraught issue. Some identify as victims, some as survivors, and some do not identify with either term (Boyle and Rogers 2020). For continuity with the terminology of the “perfect victim,” I use the term “victim” in this article when referring to people who have experienced rape, but it is an imperfect term.

2

When restricted to respondents who perceived the race manipulation as intended, the race effect becomes slightly more pronounced (see supplemental materials).

3

For the two manipulations comprised of three categories—race and prior relationship—I categorized a target as aligning with the idealized target archetype only when she was white and had never met the harasser. Results are nearly identical when I count the middle categories (Latina and previously went on one date) as aligning with the idealized target archetype.

4

I did not ask respondents in this study what race they perceived the harasser—Richard Thompson—to be, but when I asked respondents in a separate study posted on Amazon Mechanical Turk (n=50), 92 percent perceived him to be white.