The Penalties For Self-Reporting Sexual Harassment

Abstract

Although sexual harassment in the workplace is illegal, it often goes unreported. This study employs causal evidence to evaluate one deterrent to reporting: bias against women known to be sexual harassment targets. I theorize about the form this bias takes and test the argument with a national survey experiment run in five waves from October 2017 to February 2018, where participants were asked to propose employment outcomes for an employee with one of four harassment experiences. Participants were less likely to recommend a woman for promotion if she self-reported sexual harassment relative to otherwise identical women who experienced nonsexual harassment or whose sexual harassment was reported by a coworker. The woman who self-reported sexual harassment experienced normative discrimination: that is, the promotion bias was significantly mediated by perceptions that she was less moral, warm, and socially skilled than the woman whose coworker reported her sexual harassment. These results indicate that women may hesitate to report sexual harassment because they rightly perceive that doing so could cause them to experience bias. Yet they also suggest that bias can be avoided if a bystander reports the harassment. Finally, exploratory analyses suggest that in the wake of #MeToo this bias may be fading.

Keywords

sexual violence workplace bias normative discrimination #MeToo gender

A month before the 2016 presidential election, a number of women accused then-presidential candidate Donald J. Trump of sexual harassment. At a campaign event in North Carolina, Trump commented on the accusations of one woman, saying, “Well, did you hear about it 12 years ago? Did you hear about five years? No, no, we never—we just heard about it recently. . . . And the reason she didn’t write it 12 years ago is very simple. It never happened, it never happened. It’s a lie.” (Beckwith 2016). In the wake of a surge of attention to stories of sexual harassment, this point has been raised frequently. Why would a person who experienced sexual harassment choose not to report it when it occurred?

This study seeks to provide the first causal evidence for a possible deterrent to reporting sexual harassment: that women who are known sexual harassment targets, especially those who self-report, are negatively stereotyped and therefore face bias. I utilize a series of novel survey experiments to test whether participants offer less favorable employment outcomes to women employees who experienced sexual harassment relative to identical employees who experienced nonsexual harassment. I then test whether a woman who self-reported her sexual harassment faces greater bias than a woman whose sexual harassment was reported by a third party. Additionally, I extend theory on gender, sexual harassment, and stereotyping by identifying the type of bias sexually harassed women face. Specifically, I examine whether status discrimination (perceptions of the employee’s competence) or normative discrimination (perceptions of the employee’s interpersonal qualities) explain differences in employment-related outcomes. Finally, at an exploratory level, I repeat the experiment over five months during a surge of media attention toward sexual harassment and attend to differences in bias and perceptions over this period. In full, I show that relative to women employees who experience nonsexual harassment or sexual harassment reported by a coworker, a woman employee who self-reports sexual harassment is perceived as less worthy of career advancement.

Sexual Harassment: Ubiquitous and Unchecked

Sexual harassment is both widespread and consequential. Studies estimate that approximately one in two U.S. women have experienced unwanted sexual interactions at work (Ilies et al. 2003; Pina and Gannon 2012). Experiences of sexual harassment put its targets at a clear disadvantage. Most immediately, sexual harassment is challenging to react to effectively (Collinson and Collinson 1996; Woodzicka and LaFrance 2001), while in the longer term, sexual harassment is associated with declines in job satisfaction and workplace productivity (Pina and Gannon 2012). Moreover, longitudinal data demonstrate that many women who are sexually harassed subsequently experience financial strain and obstacles to their desired career trajectories (McLaughlin, Uggen, and Blackstone 2017). Given its clear negative impact on workplace outcomes and its disproportionate incidence among women (Welsh 1999), sexual harassment likely contributes to the stalled progress toward gender equality in the labor market (England 2010) more than is frequently acknowledged.

The sexual harassment literature presents a puzzle: sexual harassment is prohibited by both the law and, typically, company policies (Dobbin 2009), yet it remains severely underreported. Only 5-30 percent of those who experience sexual harassment file formal complaints within their organizations, and less than 1 percent pursue legal action (McDonald 2012). Why do sexual harassment targets rarely pursue redress when the experience is clearly damaging?

Scholars have identified several reasons that people do not report sexual harassment to authorities. Lay understandings of sexual harassment are narrow, so people may not label sexual harassment experiences as such (Charlesworth, McDonald, and Cerise 2011) or else blame themselves if they do not react against harassment assertively (Fitzgerald, Swan, and Fischer 1995; Woodzicka and LaFrance 2001). Even when they do recognize an experience as sexual harassment, targets may opt not to report for other reasons: they do not believe that anything will be done, anticipate being blamed, or fear retaliation (McDonald 2012; Pina and Gannon 2012). Bleak expectations of organizations’ responses to sexual harassment reports are not baseless: Supervisors sometimes manufacture obstacles or narrow the company policy to avoid penalizing a favored employee (Marshall 2005; McLaughlin, Uggen, and Blackstone 2012). Worse, many of those who report their sexual harassment experience retaliation, such as being fired (Hesson-McInnis and Fitzgerald 1997).

Yet although instances of retaliation documented in the workplace might be attributed to interpersonal alliances in the workplace (i.e., motivation to protect the harasser), it is possible that there is also a general cultural bias against sexual harassment targets. That is, people who know neither the perpetrator nor target may still view a sexual harassment target as less worthy of career advancement. Indeed, women often choose not to report or even label unwanted sexual interactions as sexual harassment in part to avoid a perceived stigma that would come with identifying oneself as a target of sexual harassment (Fitzgerald, Swan, and Fischer 1995; Quinn 2000). However, causal evidence has not yet been supplied to test this idea. In the following section, I draw on literature about stereotyping in labor markets to develop two theories about how bias against sexual harassment targets might operate.

How Do Stereotypes about Sexual Harassment Targets Lead to Bias?

People use stereotypes as shortcuts to make inferences about people by applying traits that are thought to characterize certain groups to the individuals belonging to those groups (Fiske et al. 2002). These stereotypes provide information, including the expected competence, warmth, and morality of a person (Fiske et al. 2002; Goodwin, Piazza, and Rozin 2014). Once these group stereotypes are applied to individuals, people then orient their emotions and behaviors toward those individuals based on their inferred traits (Fiske, Cuddy, and Glick 2007). Stereotypes are consequential because they can result in inequitable treatment through either status discrimination or normative discrimination (Benard and Correll 2010). Status discrimination occurs when people are considered ill-suited for a task because a stereotype casts them as less competent at it. For example, because men are stereotyped to be more competent than women at tasks that are not stereotyped as feminine, they are favored for employment opportunities in neutral- and masculine-typed fields (Ridgeway 2011). Similarly, because mothers are perceived to be less committed and competent than women not presumed to be mothers, they are less likely to be hired (Correll, Benard, and Paik 2007). By contrast, normative discrimination occurs when a person fails to enact stereotypically expected behavior and is therefore viewed less positively on interpersonal qualities. For example, when women take on agentic leadership qualities, they violate the stereotype that women should be communal and group-oriented and are therefore viewed as less likable and hirable than men who do the same (Rudman et al. 2012). Likewise, when working mothers are shown to be competent and committed to their work, violating societal expectations that they focus on childcare, they are seen as less likable than identical working fathers and offered less favorable employment outcomes (Benard and Correll 2010). These dual forms of discrimination serve to maintain social hierarchies by casting subordinated groups as less competent and punishing them when they violate this expectation.

I propose that either experiencing or reporting sexual harassment may likewise prime stereotypes that negatively affect perceptions of women in ways that can be detrimental to their career advancement. Importantly, while previous literature has shown that women often choose not to report or even label unwanted sexual interactions as sexual harassment in part to avoid a perceived stigma, it has not been clear whether the stigma is the result of status discrimination, meaning sexual harassment targets are perceived as incompetent, or normative discrimination, meaning sexual harassment targets are seen as violating social norms about how they should act. By testing for multiple mediating pathways, I explore whether possible bias against sexual harassment targets can be explained by status discrimination, normative discrimination, or both.

In developing hypotheses of how stereotypes about sexual harassment targets may lead to bias, I restrict my focus to women who are sexually harassed by men. Sexual harassment is most commonly perpetrated by men against women (McDonald 2012), which scholars have argued is not coincidental: sexual harassment often serves as a means of maintaining the gender status quo in workplaces by undermining women’s ability to do their jobs (e.g., MacKinnon 1979; McLaughlin, Uggen, and Blackstone 2012). Indeed, the concept of sexual harassment first came into being with this gendered focus (Saguy 2000). Further, although sexual harassment can involve other gender configurations, these cases are less numerous and rarely reported in the media (Berdahl, Magley, and Waldo 1996; McDonald and Charlesworth 2013). Therefore, I focus on the man-perpetrator/woman-target configuration, as it is most common and familiar and thus most likely to elicit widely held stereotypes.

Sexual Harassment versus Nonsexual Harassment

One possibility is that knowledge that a woman has been sexually harassed elicits effects similar to a woman being sexually objectified. The sexual objectification of women is disadvantageous in that it dehumanizes women, such that they are treated as objects rather than people (Bird 1996). Indeed, studies show that when people are instructed to focus on the physical appearance rather than performance of a woman, they rate her lower on basic human characteristics such as competence, warmth, and morality (Heflick et al. 2011; Heflick and Goldenberg 2009). When a woman is objectified, she therefore experiences both status discrimination (meaning her perceived competence is diminished) and normative discrimination (meaning she is viewed more negatively on interpersonal qualities).

Such responses to objectified women are relevant to sexual harassment in that when a woman is sexually harassed, typically she also is objectified (i.e., her physical appearance rather than personality is made salient). Whereas Heflick and colleagues (2011) find that objectification itself diminishes a woman’s perceived competence, warmth, and morality, I explore whether mere knowledge that a woman has been objectified lowers these perceptions of her. By contrast, experiencing nonsexual harassment should not diminish perceptions of these traits, because it should not elicit objectification responses. This leads me to the following hypotheses:

Hypothesis 1a: Women who experience sexual harassment will face bias relative to women who experience nonsexual harassment.

Hypothesis 1b: The bias associated with experiencing sexual harassment, compared with experiencing nonsexual harassment, will be explained by status discrimination (perceptions of sexual harassment targets as less competent) and normative discrimination (perceptions of sexual harassment targets as less warm, socially skilled, and moral).

The Source of the Sexual Harassment Report

Another possibility is that it is not having been sexually harassed that affects the general viewer’s judgment, but rather objecting to it. Importantly, research suggests that there may be particularly negative stereotypes about sexual harassment targets who report their harassment (Fitzgerald, Swan, and Fischer 1995; Quinn 2000). The Illinois Sexual Harassment Myth Acceptance Scale, which identifies prevalent falsely held beliefs about sexual harassment, finds that these beliefs frequently center on the legitimacy of reporting harassment (Lonsway, Cortina, and Magley 2008). The authors of the scale identify four categories of sexual harassment myths: the view that women fabricate or exaggerate sexual harassment complaints; that women have ulterior motives for reporting it (such as the desire to ruin careers or extort money); that receiving sexual interest from a man at work is normal and flattering behavior; and that it is a woman’s responsibility to put an end to unwanted sexual attention when it occurs. From this perspective, it is not the experience of sexual harassment that compromises how a woman is perceived, but rather the perceived illegitimacy of reporting such behaviors. These beliefs suggest that when women report sexual harassment, it is because they are immoral, cold, and/or lacking social skills to diffuse the situation. If this is the case, women who report sexual harassment should experience normative discrimination because they violate social beliefs about how women who experience sexual harassment should behave.

I distinguish between social skills and competence because sexual harassment myths suggest that a woman who reports sexual harassment will be negatively perceived only on the former. Sexual harassment myths suggest that a woman who is sexually harassed should have the social skills to resolve the harassment without assistance but not that a woman who self-reports sexual harassment is less competent in other domains. Therefore, in contrast to objectification theory, the hypotheses generated by sexual harassment myth stereotypes predict that a woman who self-reports sexual harassment will experience normative discrimination (i.e., she will be seen as lacking in interpersonal qualities) but not status discrimination (i.e., she will not be perceived as lacking general competence).

The negative perceptions of a woman who is sexually harassed should not extend to cases in which a coworker, acting as a bystander, reports the harassment. Having a witness dispels the notion that the account is fabricated or exaggerated and suggests that the incident was worthy of bringing to the company’s attention. Thus, when sexual harassment is reported by another person, it should not elicit the same negative stereotypes against the target as when it is self-reported. I therefore hypothesize the following:

Hypothesis 2a: A woman who self-reports sexual harassment will face bias relative to a woman whose coworker reports her sexual harassment.

Hypothesis 2b: The bias associated with self-reporting sexual harassment, relative to having a coworker report sexual harassment, will be explained by normative discrimination (perceptions of the woman who self-reports sexual harassment as less moral, warm, and socially skilled).

Methods

The data used to evaluate these arguments come from a novel set of survey experiments with 924 U.S. residents (50 percent men, 49.5 percent women, 0.5 percent another gender), ages 18 or older. I recruited participants through Amazon Mechanical Turk (MTurk). Relative to the U.S. population, this sample is slightly younger (median age = 33 years), and it overrepresents Asians (7.7 percent of the sample) and underrepresents African Americans (6.4 percent of the sample) (CIA 2018). Though this sample is not fully representative of the U.S. population, it is diverse across demographics, and studies have found that MTurk participants provide data that is of comparable or higher quality than data from traditional population-based samples (Weinberg, Freese, and McEllhattan 2014). Therefore, these data are well suited for evaluating whether cultural biases against sexual harassment targets occur in the general population.

The experiment was fielded five times in late 2017 and early 2018. The first, coincidentally, was on October 5, 2017, the day that sexual harassment allegations against the film producer Harvey Weinstein broke, precipitating a surge of public attention to sexual harassment that has been termed the #MeToo movement. The sudden, sustained increase in media reports and social activism about sexual harassment that followed suggests the potential for cultural stereotypes about sexual harassment to shift. Since this study investigates stereotypes and biases that people hold about sexual harassment targets, I reran the experiment to track potential changes over the following four months (on November 3 and December 8, 2017; and January 9 and February 9, 2018).¹ While these experiments provide causal evidence about whether sexual harassment targets experience bias and what form that bias takes, evidence about the nature of trends over time is exploratory and not causal.

I utilized a 2×2 between-subject design in which participants are randomly assigned to one of four conditions in the experiment, varying the type of harassment experienced (sexual/nonsexual) and whether the harassment was self-reported or reported by a coworker. I used nonsexual harassment as a control because it allowed me to elucidate the specific effect of sexual harassment, holding constant biases that may be associated with being a target of harassment in general.² An experiment presents an ideal way to test the effect of experiencing sexual harassment on perceptions and workplace outcomes, because by varying only information about harassment I am able to hold potentially confounding information constant. Moreover, by randomly assigning individuals to condition, other factors that may shape attitudes toward sexual harassment targets are randomly distributed across conditions, allowing me to isolate the causal effect of the harassment type and harassment reporter on workplace outcomes.

Furthermore, the design minimized suspicion about the study hypotheses in two ways. First, I used an employee file rather than a resume or job application so that a report of sexual harassment can be conveyed relatively inconspicuously and therefore not explicitly signal what the study was about. Second, because participants only saw one employee file, they were not able to easily discern the purpose of the study by comparing multiple employees.

Experimental Procedures

The experiment was presented as follows: participants in each condition were asked to imagine that they were the manager of a company considering an employee for promotion, and they were shown the performance review of Sarah Carter, a woman sales associate. The performance reviews were identical across conditions except for information about the harassment the employee experienced and who had reported the harassment, which was conveyed through a section at the end of the review entitled “challenges.” In all conditions, Sarah experienced harassment from Brian Miller, a coworker. Depending upon condition, the harassment was sexual (he “repeatedly made sexual references to her body”) or nonsexual (he “repeatedly shouted and swore at her”). The reporter of harassment also was varied across condition: the harassment was either self-reported (“Sarah reported that . . . ”), or reported by a colleague (“A co-worker observed that . . . ”). In all conditions the harassment was described as currently being examined by the Human Resources department, which signaled that it might merit consequences but that a verdict had not been reached. I strategically limited the information so that participants were not able to make inferences about the harassment based on additional detail, as previous research has found that information such as the sexual harassment target’s reaction can alter its perceived legitimacy (e.g., Smirles 2004).

All other information about the employee presented her as average across conditions. In each case, she was a sales associate who had worked at the company for two years, had satisfactory but not exemplary ratings, and was described as dedicated and enthusiastic but unfocused and overcommitted. She was intentionally made to sound unextraordinary, as stereotypes are most readily drawn on in the absence of other qualifying information (Dovidio and Gaertner 2000). I selected “Sarah” and “Carter” because both names are common in the United States; however, given that the name “Sarah” is perceived as a white name, respondents likely envisioned a white woman (Gaddis 2017).

Measures

After reviewing the employee file, participants were asked a series of questions to assess whether they exhibited bias toward the employee and, if so, what mechanism(s) led to bias. They first answered two questions about employment outcomes (with randomly assigned ordering) that served as the primary dependent variables: their likelihood of promoting her on a Likert scale from one-to-seven (with poles labeled “extremely unlikely” and “extremely likely”), and their recommendation of a raise of $0 to $10,000 given her current salary of $60,000. (See online supplemental materials for means and differences across months.)

Because I theorized that bias against women who report sexual harassment is at least partially explained by negative stereotypes about them, I then measured how participants perceived the employee. I asked participants to rate on a Likert scale from one-to-seven (with poles labeled “not at all accurate” and “extremely accurate”) how well various traits represented the employee. Perceptions were measured with multiple traits that were pooled into composites averaged along a one-to-seven scale to measure the relevant constructs. I used preestablished traits to measure competence, morality, and warmth (Fiske et al. 2002; Goodwin, Piazza, and Rozin 2014) and created three statements to represent social skill. To assess status discrimination, I measured competence as a composite of competent, good at her job, and confident (Cronbach’s alpha = 0.80). To assess normative discrimination, I measured morality as a composite of principled, untrustworthy, and selfish (Cronbach’s alpha = 0.64); warmth as a composite of easygoing and agreeable (Cronbach’s alpha = 0.72); and social skill as “interacts well with customers and clients,” “fits in well at the company,” and “makes coworkers feel uncomfortable” (Cronbach’s alpha = 0.72).³ The means and standard deviations for all trait measures are provided in the supplemental materials.

After collecting this information, I asked participants in November onward to respond to the question “What do you think about the #metoo campaign?” at the end of the survey. This allowed me to assess participants’ familiarity with, and reactions to, the movement. In my analysis below, I include the variable month, which is conceptualized as time elapsed since the beginning of the #MeToo movement (where October is 0 months, November is 1 month, and so on).

Analytic Sample and Strategy

After participants answered questions about the employee, I asked them to recall the workplace altercation described in the employee file (with the options “absenteeism,” “harassment, not sexual in nature,” “violation of safety rules,” “sexual harassment,” and “none”). Eleven percent of participants did not correctly recall the type of harassment they had read about. I limited the analytic sample to those who correctly recalled the experimental manipulation, in line with previous research (e.g., Pedulla 2014). Limiting the sample this way does not drive the results: Although effect sizes are slightly smaller when participants who failed the manipulation checks are included in the models, the direction and statistical significance of effects remain comparable.

I modeled the effect of type of harassment (sexual /nonsexual) and who reported the harassment (self/coworker) on each of my dependent measures using ordinary least squares regression. Although ordered logistic regression is often favored for dependent variables measured on a Likert scale, there is some concern that significant interaction terms in latent-variable models may not necessarily represent true interactions (Ai and Norton 2003). In addition, I am not able to compute holistic measures of mediation analysis, such as the overall average causal mediation effect, when using ordered logistic regression. I therefore present results using OLS regression but produce parallel models using ordered logistic regression in the supplemental materials for comparison (results are substantively similar). For causal mediation analysis, I used the framework proposed by Imai, Keele, and Tingley (2010).

Investigating Bias Against Sexual Harassment Targets

I present analyses in four parts. First, I examine the effect of sexual harassment compared to nonsexual harassment on promotion likelihood and proposed raise. Second, I isolate the effect of self-reporting sexual harassment, rather than a coworker reporting sexual harassment, on workplace outcomes. Third, I look at the perceptions of sexual harassment targets to explore whether status discrimination or normative discrimination can explain why self-reporting was disadvantageous to sexual harassment targets’ promotion likelihood. Finally, I examine changes in bias over the course of the five-month study.

The Effect of Experiencing Sexual Harassment on Workplace Outcomes

I begin the empirical analysis by examining whether participants express more bias against women who experience sexual harassment relative to women who experience nonsexual harassment (see Table 1). If this were the case, the main effect of sexual harassment should be negative and significant. Instead, however, I find no significant effect of sexual harassment for promotion likelihood (model 1a) or proposed raise (model 1b). Thus, I do not find evidence to support hypothesis 1a, that women who experience sexual harassment, no matter the source of the report, are less likely to be promoted than women who experience nonsexual harassment. Without finding support for hypothesis 1a, there is no mechanism to test in hypothesis 1b. I therefore proceed to the next set of hypotheses.

Table 1:

OLS Regression Models of Differences in Employment Outcomes by Harassment Type, Harassment Reporter, and Month

Variables	Model 1a:Promotion Likelihood(n=813)	Model 2a:Promotion Likelihood(n=813)	Model 3a:Promotion Likelihood(n=813)	Model 1b:Proposed Raise(n=812)	Model 2b:Proposed Raise(n=812)	Model 3b:Proposed Raise(n=812)
Sexual harassment (ref = nonsexual harassment)	0.04 (0.10)	0.25† (0.14)	0.07 (0.25)	$182 ($155)	$195 ($221)	-$1 ($416)
Self-reported (ref = coworker-reported)		0.04(0.14)	0.16(0.26)		-$116($225)	-$186($420)
Sexual harassment × self-reported		−0.41* (0.19)	−0.92** (0.35)		-$24($311)	-$69($575)
Month			0.02(0.07)			-$63($115)
Sexual harassment × month			0.09(0.10)			$87($159)
Self-reported × month			−0.06(0.10)			$28($161)
Sexual harassment × self-reported × month			0.23†(0.14)			$24($221)

Constant	4.67*** (0.07)	4.65*** (0.10)	4.61*** (0.19)	$3,009*** (113)	3,067*** (159)	3,210*** (308)

NOTE: Values in parentheses are standard errors. Data were collected on MTurk in October 2017, November 2017, December 2017, January 2018, and February 2018. Month is coded as linear, with October beginning at 0. OLS = ordinary least squares.

***

p < 0.001, **p < 0.01, *p < 0.05, †p < 0.10.

The Effect of Self-Reporting Sexual Harassment on Workplace Outcomes

Hypothesis 2a posits that among women who have been sexually harassed, those who self-report will experience less positive employment outcomes. To test this hypothesis, I interact harassment type and harassment reporter. If those who self-report sexual harassment face bias not faced by those whose sexual harassment was reported by a coworker, the interaction term should be negative and significant.

I find that there is indeed a significant penalty in promotion likelihood against the employee who self-reported sexual harassment. The significant and negative interaction term in Table 1, model 2a indicates that a woman who self-reports sexual harassment is least likely to be recommended for promotion, and that this penalty is unique to targets of sexual harassment. When I compare the outcome of the woman self-reporting sexual harassment to the closest comparison categories, I find that she is 0.37 points less likely to be recommended for promotion than the woman whose coworker reported her sexual harassment, and 0.16 points less likely to be recommended for promotion than the woman who self-reported nonsexual harassment.⁴ The positive coefficient associated with sexual harassment in model 2a also indicates that the woman whose sexual harassment was reported by a coworker had a higher chance of promotion than the other employees. However, model 3a indicates that bias against the woman who self-reported sexual harassment was strongest at the beginning of the study, and a preference for the woman whose co-worker reported her sexual harassment occurred at the end of the study, as I discuss in more detail below.

Given women’s greater likelihood of experiencing sexual harassment, one might expect women participants to exhibit less bias against promoting the employee self-reporting sexual harassment. This is not the case. When I interact gender with the other terms in the models in Table 1, the interactions are not statistically significant, and when I separate each of the models by gender and compare estimates, Wald tests do not reject the null hypothesis of no difference between the coefficients.⁵ Thus, women participants were just as likely as men participants to penalize women employees who were sexually harassed. This is consistent with other research on gender bias, which tends to find that men and women are equally affected by gender stereotypes (Wynn and Correll 2018).

In contrast to the promotion analyses, I find that participants did not exhibit a bias in the raise they proposed for the woman who self-reported sexual harassment, as seen in model 2b. I also do not find that proposed raise varied meaningfully over time (model 3b). Thus, in response to hypothesis 2a, I find that a woman who self-reports sexual harassment is indeed penalized relative to a woman whose sexual harassment is reported by a coworker, but this penalty applies only to her recommended advancement within the company. This may be due to differing logics for promoting versus compensating sexual harassment targets, as I describe below.

The Mediating Effect of Sexual Harassment Target Perceptions in Explaining the Promotion Penalty

Having demonstrated a specific promotion penalty against women who self-report sexual harassment, I now turn to hypothesis 2b: that negative employment consequences experienced by the employee who self-reports sexual harassment relative to the employee whose sexual harassment was reported by a coworker are explained by normative discrimination and not status discrimination. To do this, I re-create the results for sexual harassment self-reporting penalty (Table 1, model 2a), this time restricting the sample to only sexual harassment targets. As before, participants were significantly more likely to promote the sexual harassment target whose harassment was reported by a coworker (predicted value=4.90) than the target who self-reported (predicted value=4.53) (Table 2, model 1).

Table 2:

OLS Regressions of the Role of Perceived Morality, Warmth, Social Skills, and Competence in Mediating the Promotion Penalty of Self-Reporting for Sexual Harassment Targets

Variables	Model 1:Promotion Likelihood(n=426)	Model 2:Morality(n=426)	Model 3:Warmth(n=426)	Model 4:Social Skills(n=426)	Model 5Competence(n=426)	Model 6:Promotion Likelihood(n=426)	Model 7:Promotion Likelihood(n=426)	Model 8:Promotion Likelihood(n=426)	Model 9:PromotionLikelihood(n=426)
Self-reported (ref=coworker reported)	−0.36** (0.14)	−0.22* (0.09)	−0.50*** (0.10)	−0.44*** (0.09)	−0.10(0.09)	−0.22† (0.12)	−0.06(0.12)	−0.04(0.12)	−0.26* (0.10)
Morality						0.68*** (0.07)
Warmth							0.62*** (0.06)
Social skills								0.75*** (0.06)
Competence									1.00*** (0.06)
Constant	4.90*** (0.10)	5.85*** (0.06)	5.28*** (0.07)	5.79*** (0.06)	5.62*** (0.06)	0.90*(0.40)	1.62*** (0.32)	0.57(0.37)	−0.72* (0.33)
Average causal mediation effect						−0.15*	−0.31***	−0.33***	−0.11
Proportion of total effect mediated						0.40	0.83	0.90	0.29

***

p < 0.001, **p < 0.01, *p < 0.05, †p < 0.10.

Hypothesis 2b posits that between the two women who have been sexually harassed, negative employment consequences experienced by the employee who self-reports are explained by normative discrimination, meaning that she is perceived as less moral, warm, and socially skilled (but not less competent) than the employee whose sexual harassment is reported by a coworker. In models 2-5, the dependent variables are morality, warmth, social skills, and competence, respectively. As the self-reported dummy variables show, relative to the employee whose coworker reported her sexual harassment, participants perceived the employee who self-reported sexual harassment to be significantly less moral, warm, and socially skilled, but not significantly less competent. I therefore find initial evidence of normative discrimination (the employee who self-reported sexual harassment was rated lower on interpersonal qualities) but not status discrimination (she was not seen as less competent).

Hypothesis 2b further holds that the penalty in workplace outcomes that a self-reporter of sexual harassment experiences can be explained by negative perceptions of her. To show support for hypothesis 2b, perceptions of the employee’s morality, warmth, and social skills, but not competence, should reduce the size and statistical significance of the coefficient for self-reporting harassment (Baron and Kenny 1986). Models 6-9 show that this is indeed the case. When morality is included in the model (model 6), the coefficient for self-reporting is reduced by more than one third and becomes marginally significant, whereas the coefficient for morality remains positive and statistically significant. The effect is even stronger for warmth and for social skills: when each of these variables are included in the model, the coefficient for self-reporting harassment is less than one fifth of its original size, whereas the coefficients for warmth and social skills remain statistically significant and positive (models 7 and 8). Finally, when competence is included in the model, the coefficient for competence is significant and positive, meaning that, not surprisingly, employees who are seen as more competent are more likely to be recommended for promotion (model 9). However, even controlling for competence ratings, women who self-report sexual harassment are still significantly less likely to be recommended for promotion.

I further test the mediation pathway of each of these perception variables following the causal mediation analysis framework proposed by Imai, Keele, and Tingley (2010). For perceived morality, I find that the average causal mediation effect (ACME), or estimate of how much of the total effect between treatment and outcome is explained by mediator, is -0.15 points for perceived morality (p<.05), meaning that 40 percent of the total effect of self-reporting on promotion likelihood is mediated by the employee’s perceived morality.⁶ The ACMEs of perceived warmth and social skills are 0.31 points (p<.001) and 0.33 points (p<.001), respectively, meaning that 83 percent and 90 percent of the total effect is mediated, respectively. This provides additional support for the mediating role of perceived morality, warmth, and social skills in the promotion penalty. By contrast, I find that the ACME of perceived competence is −0.11, meaning that just 29 percent of the total effect is mediated, and the effect is nonsignificant, indicating that the competence is not a statistically significant mediator.⁵ Thus, in line with hypothesis 2b, lower perceptions of the self-reporter’s perceived morality, warmth, and social skills all explain a significant portion of the penalty against the woman who self-reports sexual harassment, relative to the woman whose sexual harassment is reported by a coworker. These results are consistent with theories of normative discrimination and not status discrimination. Self-reporting sexual harassment does not affect the employee’s competence ratings, but it does affect how she is viewed in terms of interpersonal qualities, which in turn shapes her perceived fitness for promotion.

The nature of the negative characteristics attributed to the woman self-reporting sexual harassment also may explain why there was a penalty in promotion likelihood but not raise. It may be the case that information about a person’s character is weighed differently when determining these outcomes. In deciding whether to promote a person, and thereby give them greater authority, that person’s moral character, warmth, and social skills may be considered more relevant, whereas in deciding a person’s compensation, their competence may be the primary consideration. This could account for the specific bias against promoting, but not giving a raise to, the woman who self-reported sexual harassment. The lack of bias in compensation notwithstanding, the promotion bias inhibits the ability of a woman who reports sexual harassment to advance within a company, thereby maintaining the status quo in which women are severely underrepresented in the highest levels of organizational hierarchies (Purcell, MacArthur, and Samblanet 2010).

Stereotypes about Sexual Harassment in the Midst of #Metoo: An Exploratory Analysis

In the wake of a flood of media attention to sexual harassment, there is reason to believe that stereotypes about sexual harassment may have been shifting. This has not been the first time that social activism against sexual harassment has occurred in the United States: activists highlighted the issue as a plank of women’s rights activism in the 1970s, and it again punctuated the public’s consciousness in the 1990s in Anita Hill’s nationally televised testimony of her sexual harassment by then Supreme Court nominee Clarence Thomas (Saguy 2000). While public awareness about sexual harassment has been heightened before, the #MeToo movement may have been particularly effective in illustrating its prevalence, as women have stepped forward in large numbers to describe the sexual harassment to which they have been subjected. By encouraging people to share their experiences of sexual harassment, the #MeToo movement has exposed its continued pervasiveness in the workplace. Could this sudden public attention to sexual harassment have an effect on the stereotypes people apply to sexual harassment targets? While some research would suggest that a surge in the visibility of marginalized groups may shift perceptions (Rosenfeld 2017), it also might be the case that increased attention to women’s sexual harassment may be perceived as threating to men and thus elicit backlash, leading to increased bias. Indeed, the media reports that with increasing attention to sexual harassment some men are now refusing to mentor women employees (McGregor 2017).

I therefore test whether the strength of bias against promoting the woman who self-reported sexual harassment increased or decreased over the course of the study by adding the term month to the interaction between harassment type and harassment reporter (Table 1, model 3a). I find that bias was substantially greater at the beginning of the study: In October, the woman self-reporting sexual harassment was 0.76 points less likely to be promoted than the woman whose coworker reported her harassment, and 0.85 points less likely to be promoted than the woman who self-reported nonsexual harassment.⁷ These differences are consequential not only in magnitude but also in their position along the seven-point promotion likelihood scale: Although participants reported being more likely than not to promote either of the women who experienced nonsexual harassment and the woman whose sexual harassment was reported by a coworker, they were slightly less likely than not to promote the woman who self-reported sexual harassment (Figure 1). This represents an obstacle in the workplace advancement of women who self-report sexual harassment relative to other women who are known targets of harassment. However, it also shows that such bias can be mitigated if a bystander coworker reports the sexual harassment.

Figure 1.

Predicted values of likelihood of promoting the employee

Although participants exhibited substantial bias against self-reporters of sexual harassment in October, I find evidence that this bias declined over time, as represented in the positive three-way interaction term in model 3a (p=0.09).⁸ Although participants reported a lower likelihood of promoting the woman self-reporting sexual harassment in October, November, and December, the magnitude of this penalty decreased with each month, and in January and February participants no longer expressed bias. In addition, although participants did not express a preference for either employee who had experienced sexual harassment in the first months of the study, in the final months they were more willing to promote the employees who had experienced sexual, rather than nonsexual, harassment.

While this analysis cannot pinpoint the cause of the declining bias, the fact that these changes are specific to women who experienced sexual harassment, and particularly the woman who self-reported sexual harassment, suggests that social activism emphasizing the prevalence of sexual harassment may have impacted perceptions of sexual harassment targets. Most study participants were aware of the contemporaneous activism: When asked about the #MeToo movement, two-thirds of participants were familiar with it in November and more than three-quarters were familiar with it in the following three months. Indeed, some participants even suggested that the movement may be reducing stigma associated with sexual harassment. A participant in November mused, “I think it’s a really good thing that women are now less likely to be stigmatized by sexual harassment,” and a participant in January said, “I think that the people that have been sexually harassed are courageous and brave for speaking up and helping erase the stigma associated with it.” However, while the results suggest that the contemporaneous social movement may have altered cultural stereotypes, future work is needed to establish causality.

Conclusion

Nearly one in two American women have experienced sexual harassment in the workplace, yet very few have gone on to report the harassment. In this article, I address this puzzle with two contributions. First, I employ causal evidence to illustrate one reason why women may be reluctant to report, namely, that there is a cultural bias against women who self-report sexual harassment compared to other harassment targets. Second, I deepen our theoretical understanding of the nature of this bias, showing that it operates through normative, rather than status discrimination: that women who self-report sexual harassment are seen as less moral, warm, and socially skilled (but not as less competent), and these negative perceptions lead them to be perceived as less suitable for career advancement.

I find evidence of bias against the woman who self-reported sexual harassment, as predicted by sexual harassment stereotype myths, but I do not find evidence of bias against the woman who merely experienced sexual harassment, as predicted by objectification theory. It may be the case that the experimental manipulation—mere knowledge that an unfamiliar woman had been sexually harassed—was too indirect to trigger an objectification response in participants’ minds. Future research should explore other manipulations to further test the theory, such as having participants witness the sexual harassment. However, it may be the case that in a culture in which the workplace sexual harassment of women is commonplace and normalized (McDonald 2012; Quinn 2000), women are not penalized for experiencing sexual harassment, but only for objecting to it.

An experimental design gives this study strong internal validity: By changing only harassment experience and reporter, I am able to isolate the causal effects of each. However, there are a few limitations to the study’s external validity. First, many participants recruited on Amazon Mechanical Turk likely are not in the position to be making actual promotion and raise decisions in the workplace, making their decisions in this context somewhat artificial. The effects I detect here therefore measure not bias that is expressed by managers but bias that is “in the water” in U.S. society, namely, that expressed by a diverse pool of study participants, including both women and men. It may be the case that managers respond differently to employees reporting sexual harassment than the general population. Yet sexual harassment targets are more likely to be aware of a generalized bias, rather than a specific bias amongst managers, and to make decisions about whether to report their harassment accordingly. Second, participants may be sensitive to social desirability bias, rating the sexual harassment targets more favorably than they believe them to truly be, so as not to appear sexist or insensitive. If so, the bias I identify in the experiment may be a conservative estimate.

Future research should extend this study by examining how aspects of the employee and the harassment experienced shape stereotypes about her. For example, how do a sexual harassment target’s gender, race, or workplace performance shape the workplace outcomes people deem them worthy of? Likewise, do the type of harassment experienced (e.g., sexual assault versus verbal remarks), the power differential between employees, different organizational policies about harassment, or the gender of the harasser shape the stereotypes people apply to a sexual harassment target, and thereby the extent to which they are deemed worthy of promotion? Such research could draw out the nuances of extant stereotypes about sexual harassment targets and their consequences.

The results from this study suggest that sexual harassment targets’ reluctance to report their harassment may be strategic. The finding that self-reporting sexual harassment leads to a lower likelihood of perceived suitability for workplace advancement suggests that choosing not to report sexual harassment may be a tactic through which to avoid discrimination. Yet this tactic leaves sexual harassment targets in a double bind: If they report sexual harassment, they risk being perceived as less worthy of promotion, but if they do not report sexual harassment they must manage it alone, a challenging situation for which there is no dependable response (Collinson and Collinson 1996). Because women experience sexual harassment at substantially higher rates than do men (McDonald 2012), more women than men must contend with the career damage that it presents, meaning that the net effect of sexual harassment is to preserve the gender status quo.

However, the results of this study indicate that bystander support offers a way out of the double bind presented by sexual harassment. My findings illustrate that when a coworker reported a woman’s sexual harassment, she was no more negatively stereotyped nor penalized in workplace advancement than women who experienced nonsexual harassment. Although not all sexual harassment targets have a witness who is willing to attest to their harassment, when such a bystander is available, having them report the harassment may offer a way to evade the negative stereotyping associated with self-reporting sexual harassment. However, although this is an encouraging finding, it does not challenge the gender hierarchy whereby women are penalized if they speak out against their own sexual harassment.

Although I am not the first to posit that reporting sexual harassment may be stigmatizing (e.g., Fitzgerald, Swan, and Fischer 1995), these data allow me to add nuance to this observation by illustrating the form of discrimination this stigma takes. I do not find that a woman who self-reports sexual harassment is perceived to be less competent; rather, she is perceived to be less moral, warm, and socially skilled than an equivalent woman whose sexual harassment is reported by a coworker. This indicates that bias against women who report sexual harassment operates through normative discrimination rather than status discrimination, meaning that they are punished for violating social expectations about how women are supposed to act (i.e., that they are supposed to tolerate, rather than object to, sexual harassment). Women who self-report sexual harassment thus experience the same kind of bias as do women who challenge the gender status quo in other ways (Benard and Correll 2010; Rudman et al. 2012).

Finally, though I cannot attribute the diminishing bias against the self-reporter of sexual harassment to the #MeToo movement with certainty, the coincidence of surging media attention to sexual harassment and decreased bias is at least consistent with the idea that women speaking out against sexual harassment in large numbers may help to reduce stereotyping and bias. Though unflattering myths about women who report sexual harassment have been long-lasting and pervasive, these results suggest that they are also malleable.

Supplemental Material

GandS842147_SM – Supplemental material for The Penalties for Self-Reporting Sexual Harassment

Supplemental material, GandS842147_SM for The Penalties for Self-Reporting Sexual Harassment by Chloe Grace Hart in Gender & Society

Footnotes

Author’s Note:

This research was made possible by funding through the Lab for Social Research at Stanford University. I would like to thank Shelley Correll, David Pedulla, Jeremy Freese, Cecilia Ridgeway, Emily Carian, the Social Psychology and Gender Workshop at Stanford University, and the anonymous reviewers for comments, and Chrystal Redekopp and Medina Husakovic for research assistance.

Notes

Chloe Grace Hart is a PhD candidate in sociology at Stanford University and a research assistant at the Clayman Institute for Gender Research. Her dissertation research explores gender, sexuality, and sexual violence in organizations.

References

Chunrong

Norton

Edward C.

2003. Interaction terms in logit and probit models. Economics Letters 80 (1): 123-29.

Baron

Reuben M.

Kenny

David A.

1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51 (6): 1173-82.

Beckwith

Ryan Teague

. 2016. Read Donald Trump’s speech attacking his accusers. Time Magazine, 14 October. http://time.com/4532181/donald-trump-north-carolina-accusers-speech-transcript/.

Benard

Stephen

Correll

Shelley J.

2010. Normative discrimination and the motherhood penalty. Gender & Society 24 (5): 616-46.

Berdahl

Jennifer L.

Magley

Vicki J.

Waldo

Craig R.

1996. The sexual harassment of men? Exploring the concept with theory and data. Psychology of Women Quarterly 20 (4): 527-47.

Bird

Sharon R.

1996. Welcome to the men’s club: Homosociality and the maintenance of hegemonic masculinity. Gender & Society 10 (2): 120-32.

Charlesworth

Sara

McDonald

Paula

Cerise

Somali

. 2011. Naming and claiming workplace sexual harassment in Australia. Australian Journal of Social Issues 46 (2): 141-61.

CIA (Central Intelligence Agency). 2018. The world factbook: North America: United States. https://www.cia.gov/library/publications/resources/the-world-factbook/geos/us.html.

Collinson

Margaret

Collinson

David

. 1996. “It’s only Dick”: The sexual harassment of women managers in insurance sales. Work, Employment & Society 10 (1): 29-56.

10.

Correll

Shelley J.

Benard

Stephen

Paik

. 2007. Getting a Job: Is there a motherhood penalty? American Journal of Sociology 112 (5): 1297-339.

11.

Dobbin

Frank

. 2009. Inventing equal opportunity. Princeton, NJ: Princeton University Press.

12.

Dovidio

John F.

Gaertner

Samuel L.

2000. Aversive racism and selection decisions: 1989 and 1999. Psychological Science 11 (4): 315-19.

13.

England

Paula

. 2010. The gender revolution: Uneven and stalled. Gender & Society 24 (2): 149-66.

14.

Fiske

Susan T.

Cuddy

Amy J. C.

Glick

Peter

. 2007. Universal dimensions of social cognition: Warmth and competence. Trends in Cognitive Sciences 11 (2): 77-83.

15.

Fiske

Susan T.

Cuddy

Amy J. C.

Glick

Peter

Jun

. 2002. A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology 82 (6): 878-902.

16.

Fitzgerald

Louise F.

Swan

Suzanne

Fischer

Karla

. 1995. Why didn’t she just report him? The psychological and legal implications of women’s responses to sexual harassment. Journal of Social Issues 51 (1): 117-38.

17.

Gaddis

S. Michael

. 2017. How black are Lakisha and Jamal? Racial perceptions from names used in correspondence audit studies. Sociological Science 4:469-89.

18.

Goodwin

Geoffrey P.

Piazza

Jared

Rozin

Paul

. 2014. Moral character predominates in person perception and evaluation. Journal of Personality and Social Psychology 106 (1): 148-68.

19.

Heflick

Nathan A.

Goldenberg

Jamie L.

2009. Objectifying Sarah Palin: Evidence that objectification causes women to be perceived as less competent and less fully human. Journal of Experimental Social Psychology 45 (3): 598-601.

20.

Heflick

Nathan A.

Goldenberg

Jamie L.

Cooper

Douglas P.

Puvia

Elisa

. 2011. From women to objects: Appearance focus, target gender, and perceptions of warmth, morality and competence. Journal of Experimental Social Psychology 47 (3): 572-81.

21.

Hesson-McInnis

Matthew S.

Fitzgerald

Louise F.

1997. Sexual harassment: A preliminary test of an integrative model. Journal of Applied Social Psychology 27 (10): 877-901.

22.

Ilies

Remus

Hauserman

Nancy

Schwochau

Susan

Stibal

John

. 2003. Reported incidence rates of work-related sexual harassment in the United States: Using meta-analysis to explain reported rate disparities. Personnel Psychology 56 (3): 607-31.

23.

Imai

Kosuke

Keele

Luke

Tingley

Dustin

. 2010. A general approach to causal mediation analysis. Psychological Methods 15 (4): 309-34.

24.

Imai

Kosuke

Keele

Luke

Tingley

Dustin

Yamamoto

Teppei

. 2011. Unpacking the black box of causality: Learning about causal mechanisms from experimental and observational studies. American Political Science Review 105 (4): 765-89.

25.

Lonsway

Kimberly A.

Cortina

Lilia M.

Magley

Vicki J.

2008. Sexual harassment mythology: Definition, conceptualization, and measurement. Sex Roles 58 (9/10): 599-615.

26.

MacKinnon

Catherine

. 1979. The sexual harassment of working women. New Haven, CT: Yale University Press.

27.

Marshall

Anna-Maria

. 2005. Idle rights: Employees’ rights consciousness and the construction of sexual harassment policies. Law & Society Review 39 (1): 83-124.

28.

McDonald

Paula

. 2012. Workplace sexual harassment 30 years on: A review of the literature. International Journal of Management Reviews 14 (1): 1-17.

29.

McDonald

Paula

Charlesworth

Sara

. 2013. Framing sexual harassment through media representations. Women’s Studies International Forum 37:95-103.

30.

McGregor

Jena

. 2017. Will the #MeToo movement speed up the number of women in leadership or slow it down? The Washington Post, 26 December.

31.

McLaughlin

Heather

Uggen

Christopher

Blackstone

Amy

. 2012. Sexual harassment, workplace authority, and the paradox of power. American Sociological Review 77 (4): 625-47.

32.

McLaughlin

Heather

Uggen

Christopher

Blackstone

Amy

. 2017. The economic and career effects of sexual harassment on working women. Gender & Society 31 (3): 333-58.

33.

Pedulla

David S.

2014. The positive consequences of negative stereotypes: Race, sexual orientation, and the job application process. Social Psychology Quarterly 77 (1): 75-94.

34.

Pina

Afroditi

Gannon

Theresa A.

2012. An overview of the literature on antecedents, perceptions and behavioural consequences of sexual harassment. Journal of Sexual Aggression 18 (2): 209-32.

35.

Purcell

David

MacArthur

Kelly Rhea

Samblanet

Sarah

. 2010. Gender and the glass ceiling at work. Sociology Compass 4 (9): 705-17.

36.

Quinn

Beth A.

2000. The paradox of complaining: Law, humor, and harassment in the everyday work world. Law & Social Inquiry 25 (4): 1151-85.

37.

Ridgeway

Cecilia L.

2011. Framed by gender: How gender inequality persists in the modern world. New York: Oxford University Press.

38.

Rosenfeld

Michael J.

2017. Moving a mountain: The extraordinary trajectory of same-sex marriage approval in the United States. Socius 3:1-22.

39.

Rudman

Laurie A.

Moss-Racusin

Corinne A.

Phelan

Julie E.

Nauts

Sanne

. 2012. Status incongruity and backlash effects: Defending the gender hierarchy motivates prejudice against female leaders. Journal of Experimental Social Psychology 48 (1): 165-79.

40.

Saguy

Abigail C.

2000. Employment discrimination or sexual violence? Defining sexual harassment in American and French law. Law & Society Review 34 (4): 1091-128.

41.

Smirles

Kimberly E.

2004. Attributions of responsibility in cases of sexual harassment: The person and the situation. Journal of Applied Social Psychology 34 (2): 342-65.

42.

Weinberg

Jill

Freese

Jeremy

McElhattan

David

. 2014. Comparing data characteristics and results of an online factorial survey between a population-based and a crowdsource-recruited sample. Sociological Science 1:292-310.

43.

Welsh

Sandy

. 1999. Gender and sexual harassment. Annual Review of Sociology 25:169-90.

44.

Woodzicka

Julie A.

LaFrance

Marianne

. 2001. Real versus imagined gender harassment. Journal of Social Issues 57 (1): 15-30.

45.

Wynn

Alison T.

Correll

Shelley J.

2018. Combating gender bias in modern workplaces. In Handbook of the sociology of gender, edited by Risman

Barbara

Froyum

Carissa

Scarborough

William J.

Cham, Switzerland: Springer.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.29 MB