2

Sexual Harassment Research

This chapter reviews the information gathered through decades of sexual harassment research. It provides definitions of key terms that will be used throughout the report, establishing a common framework from the research literature and the law for discussing these issues. In reviewing what sexual harassment research has learned over time, the chapter also examines the research methods for studying sexual harassment and the appropriate methods for conducting this research in a reliable way. The chapter provides information on the prevalence of sexual harassment and common characteristics of how sexual harassment is perpetrated and experienced across lines of industry, occupation, and social class. It concludes with common characteristics of environments where sexual harassment is more likely to occur.

DEFINITIONS OF KEY TERMS

The Equal Employment Opportunity Commission guidelines define sexual harassment as the following (USEEOC n.d.a.):

Sexual harassment was first recognized in cases in which women lost their jobs because they rejected sexual overtures from their employers (e.g., Barnes v.

Costle 1977¹). This type of sexual harassment became defined as quid pro quo sexual harassment (Latin for “this for that,” meaning that a job or educational opportunity is conditioned on some kind of sexual performance). Such coercive behavior was judged to constitute a violation of Title VII of the 1964 Civil Rights Act. Soon it was recognized in employment law that pervasive sexist behavior from coworkers can create odious conditions of employment—what became known as a hostile work environment—and also constitute illegal discrimination (Farley 1978; MacKinnon 1979; Williams v. Saxbe 1976²). These two basic forms of sexual harassment, quid pro quo and hostile environment harassment, were summarized in guidelines issued by the Equal Employment Opportunity Commission in 1980 (USEEOC 1980).

Hostile work or educational environments can be created by behaviors such as addressing women in crude or objectifying terms, posting pornographic images in the office, and by making demeaning or derogatory statements about women, such as telling anti-female jokes. Hostile environment harassment also encompasses unwanted sexual overtures such as exposing one’s genitals, stroking and kissing someone, and pressuring a person for dates even if no quid pro quo is involved (Bundy v. Jackson 1981;³Meritor Savings Bank v. Vinson 1986⁴).

An important distinction between quid pro quo and hostile environment harassment is that the former usually involves a one-on-one relationship in which the perpetrator has control of employment- or educational-related rewards or punishments over the target. In contrast, the latter can involve many perpetrators and many targets. In the hostile environment form of sexual harassment, coworkers often exhibit a pattern of hostile sexist behavior toward multiple targets over an extended period of time (Holland and Cortina 2016). For hostile sex-related or gender-related behavior to be considered illegal sexual harassment, it must be pervasive or severe enough to be judged as having had a negative impact upon the work or educational environment. Therefore, isolated or single instances of such behavior typically qualify only when they are judged to be sufficiently severe. Legal scholars and judges continue to use the two subtype definitions of quid pro quo and hostile environment to define sexual harassment.

Illegal sexual harassment falls under the umbrella of a more comprehensive category, discriminatory behavior. Illegal discrimination can occur on the basis of any legally protected category: race, ethnicity, religious creed, age, sex, gender identity, marital status, national origin, ancestry, sexual orientation, genetic information, physical or mental disabilities, veteran status, prior conviction of a crime, gender identity or expression, or membership in other protected classes set forth in state or federal law. Regarding sexual harassment, the focus of this report, this includes gender harassment, a term designed to emphasize that harmful or

___________________

¹Barnes v. Costle, 561 F.2d 983, 987 (D.C. Cir. 1977).

²Williams v. Saxbe, 413 F. Supp. 654 D.D.C. (1976).

³Bundy v. Jackson, 641 F.2d 934 (D.C. Cir. 1981).

⁴Meritor Savings Bank v. Vinson, 477 U.S. 57 (1986).

illegal sexual harassment does not have to be about sexual activity (USEEOC n.d.b.). Sexual harassment constitutes discrimination because it is harmful and it is based on gender—it is not necessarily motivated by sexual desire nor does it need to involve sexual activity.

Both legal doctrine and social science research recognize gender as encompassing both one’s biological sex and gender-based stereotypes and expectations, such as heterosexuality and proper performance of gender roles. Sexual harassment in the form of gender harassment can be based on the violation of cultural gender stereotypes. For example, a man may experience gender harassment for being a “sissy” or being easily embarrassed by pornography (violating stereotypes that men should be strong, heterosexual, and sexually bold). While a woman may be gender harassed for taking a job traditionally held by a man or in a traditionally male field. Gender harassment in such a situation might consist of actions to sabotage the woman’s tools, machinery, or equipment, or telling the woman she is not smart enough for scientific work. Subsequent sections of this report discuss gender harassment in greater detail.

Psychologists who study gender-related behavior have developed more nuanced terms to describe sexual harassment in order to more precisely measure and account for the behaviors that constitute sexual harassment and to describe how targets experience those behaviors. A three-part classification system divides sexual harassment into distinct but related categories: sexual coercion, unwanted sexual attention, and gender harassment (see Figure 2-1; Fitzgerald et al. 1988;⁵Fitzgerald, Gelfand, and Drasgow 1995; Gelfand, Fitzgerald, and Drasgow 1995).

Sexual coercion entails sexual advances, and makes the conditions of employment (or education, for students) contingent upon sexual cooperation.

Unwanted sexual attention also entails sexual advances, but it does not add professional rewards or threats to force compliance. In this category are expressions of romantic or sexual interest that are unwelcome, unreciprocated, and offensive to the target; examples include unwanted touching, hugging, stroking, and persistent requests for dates or sexual behavior despite discouragement, and can include assault (Cortina, Koss, and Cook 2018; Fitzgerald, Gelfand, and Drasgow 1995; Fitzgerald, Swan, and Magley 1997).

Gender harassment is by far the most common type of sexual harassment. It refers to ‘‘a broad range of verbal and nonverbal behaviors not aimed at sexual cooperation but that convey insulting, hostile, and degrading attitudes about” members of one gender (Fitzgerald, Gelfand, and Drasgow 1995, 430). Gender harassment is further defined as two types: sexist hostility and crude harassment. Examples of the sexist hostility form of gender harassment for women include

___________________

⁵ The empirical record on sexual harassment goes back over 30 years, and important studies were conducted in that first decade. Members of this committee thought carefully about whether to cite “older” articles (e.g., from the 1980s). We opted to retain those references when, in our expert opinion, their methods were rigorous and their conclusions would still apply in today’s world.

demeaning jokes or comments about women, comments that women do not belong in leadership positions or are not smart enough to succeed in a scientific career, and sabotaging women. The crude harassment form of gender harassment is defined as the use of sexually crude terms that denigrate people based on their gender (e.g., using insults such as “slut” to refer to a female coworker or “pussy” to refer to a male coworker; Fitzgerald, Gelfand, and Drasgow 1995).

Both women and men can and do experience all three forms of sexual harassment, but some subgroups face higher rates than others. For example,

women who are lesbian or bisexual (Cortina et al. 1998; Konik and Cortina 2008), women who endorse gender-egalitarian beliefs (Dall’Ara and Maass 1999; Siebler, Sabelus, and Bohner 2008), and women who are stereotypically masculine in behavior, appearance, or personality (Berdahl 2007b; Leskinen, Rabelo, and Cortina 2015) experience sexual harassment at higher rates than other women. Likewise, men who are gay, transgender, petite, or in some way perceived as “not man enough” encounter more harassment than other men (Berdahl 2007b; Fitzgerald and Cortina 2017; Rabelo and Cortina 2014).

Interestingly, the motivation underlying sexual coercion and unwanted sexual attention behaviors appears different from the motivation underlying gender harassment. Whereas the first two categories suggest sexual advances (the goal being sexual exploitation of women), the third category is expressing hostility toward women (the goals being insult, humiliation, or ostracism) (Holland and Cortina 2016). In other words, sexual coercion and unwanted sexual attention can be viewed as “come-ons,” while gender harassment is, for all intents and purposes, a “put-down” (Fitzgerald, Gelfand, and Drasgow 1995; Leskinen, Cortina, and Kabat 2011). However, it is important to note that these come-on behaviors are not necessarily about attraction to women; more often than not, they are instead motivated by the desire to devalue women or punish those who violate gender norms (Berdahl 2007b; Cortina and Berdahl 2008).

Some researchers further define the verbal insults associated with gender harassment, along with accompanying nonverbal affronts, as microaggressions. This term refers to “brief and commonplace daily verbal, behavioral, or environmental indignities, whether intentional or unintentional, that communicate hostile, derogatory, or negative” messages (Sue et al. 2007, 271) to or about historically stigmatized groups. This term can also be broken down into three categories: microassaults, microinsults, and microinvalidations (Sue et al. 2007). There is some concern that microaggression remains a poorly defined construct, with porous boundaries. Additionally, the use of the term micro is misleading, as it implies all these experiences are minor or imperceptible acts. Yet some microaggressions, such as referring to people by using offensive names, are obviously offensive and can be deeply damaging. Similarly the root word aggression is also misleading, as most experts reserve this term for behavior that carries intent to harm (Lilienfeld 2017). For these reasons, our committee chose to focus on incivility, a term in greater use in the workplace aggression literature.

Incivility refers to “low-intensity deviant behavior with ambiguous intent to harm the target, in violation of workplace norms for mutual respect. Uncivil behaviors are characteristically rude and discourteous, displaying a lack of regard for others” (Andersson and Pearson 1999, 457). Lim and Cortina’s 2005 study on two female populations in public-sector organizations (Ns = 833 and 1,425) revealed that sexual harassment often takes place against a backdrop of incivility, or in other words, in an environment of generalized disrespect. The authors argue

that, based on their findings, the same perpetrator “may instigate multiple forms of mistreatment—both sexualized and generalized—in efforts to debase women and reinforce or raise their own social advantage” (492). Lim and Cortina point out that if sexual harassment is tolerated in an organization or not seen as a deviant behavior, incidents of general incivility would be expected to be even less likely to receive attention from management. Based on these findings, it could be argued that generalized incivility should be a red flag for leadership or management in work and education environments, because when gender harassment occurs, it is virtually always in environments with high rates of uncivil conduct (Cortina et al. 2002; Lim and Cortina 2005).

Note that sexual harassment is often ambient, meaning it is “not clearly targeted at any individual or group of individuals” (Parker 2008, 947) in the work or education environment or behavior that goes beyond the direct target of the harassment (Glomb et al. 1997). Ambient sexual harassment is determined by a general “frequency of sexually harassing behavior experienced by others” and can include all types of sexually harassing behavior (309). For example, it can include pornography being displayed in a common area or sexually abusive language being used publicly in the work or education environment (Parker 2008). Ambient unwanted sexual attention and sexual coercion refer to observed instances of unwanted sexual pursuit, targeted at a fellow employee. In other words, one need not be personally targeted to feel the effects of sexual harassment (much like second-hand smoke).

Despite refined definitions and terms to describe sexual harassment and gender discrimination, documenting the degree of these behaviors in work and education environments remains challenging. This is in part because individuals experiencing these behaviors rarely label them as such. Numerous studies have demonstrated that more than half of working women report experiencing sexually harassing behavior at work, but less than 20 percent of those women actually describe the experience as “sexual harassment” (Ellis, Barak, and Pinto 1991; Ilies et al. 2003; Magley, Hulin, et al. 1999; Magley and Shupe 2005).

Considering these sources, the report uses the following definition of sexual harassment:

Box 2-1 provides a quick review of the key terms introduced in this chapter.

RESEARCH METHODS USED TO EXAMINE SEXUAL HARASSMENT

The goal of providing recommendations for preventing sexual harassment and mitigating its effects in academic science, engineering, and medicine requires evidence-based research. Different studies have different strengths and weaknesses, and these should be kept in mind when reviewing their findings, particularly if leaders in academic institutions, legislators, and researchers hope to design meaningful and effective interventions and policies. The two most commonly used study methods are surveys and laboratory experiments. Important findings have also emerged using in-depth interviews, case studies, sociolegal analyses, and other methods. When conducting or reviewing research examining sexual harassment, it is crucial that the methods used to conduct the research match the goals for the research. It is crucial to note that the prevalence of sexual harassment in a population is best estimated using representative surveys and not by relying on the invariably lower number of official reports of sexual harassment made to an organization (see the discussion in Chapter 4 about how rare it is for women to formally report their experience). The next sections discuss these various research methods and the kind of information they provide.

Survey Methods

Surveys, containing well-validated instruments, can be useful in estimating the prevalence (how common sexual harassment experiences or behaviors are among people in a given population) and determining correlates, antecedents, outcomes, and factors that attenuate or amplify outcomes from sexual harassment. For instance, they can assess links between harassment and different aspects of targets’ well-being, targets’ understanding of the resources available to them, and the strategies they use to cope. Basing a survey on a defined population accessible from a comprehensive list, or sample frame, can be helpful. Sometimes, too, using multiple instruments and data sources can be a highly effective approach. Though surveys have often focused on the targets of sexually harassing behavior (e.g., Fitzgerald, Drasgow, and Magley 1999), some work has also been done examining self-descriptions by perpetrators (e.g., Dekker and Barling 1998) and bystanders (e.g., Hitlan, Schneider, and Walsh 2006; Richman-Hirsch and Glomb 2002; Miner-Rubino and Cortina 2004, 2007).

Conducting surveys on sexual harassment is challenging, but fortunately researchers have addressed many of these challenges. Those wishing to conduct a survey on sexual harassment ought to follow the scientific methods described below and the ethical and safety guidelines for this type of research (WHO 2001). Poorly conducting surveys on sexual harassment is unethical because responding to the survey could needlessly retraumatize the respondent. Additionally, the resulting inaccurate data from such a survey could be used to question the importance and legitimacy of such an important and sensitive topic (WHO 2001).

An initial challenge in conducting survey research on sexual harassment is

that many women are not likely to label their experiences as sexual harassment. Additionally, women who experience the gender harassment type of sexual harassment are more than 7 times less likely to label their experiences as “sexual harassment” than women who experience unwanted sexual attention or sexual coercion (Holland and Cortina 2013). This illustrates what other research has shown: that in both the law and the lay public, the dominant understandings of sexual harassment overemphasize two forms of sexual harassment, sexual coercion and unwanted sexual attention, while downplaying the third (most common) type—gender harassment (see Figure 2-2; Leskinen, Cortina, and Kabat 2011; Schultz 1998). Regardless of whether women self-label their experiences as sexual harassment or not, they all have similar negative psychological and professional outcomes (Magley, Hulin, et al. 1999; Woodzicka and LaFrance 2005).

This labeling issue was first identified in research on rape and sexual violence. Surveys conducted by Koss (1992) revealed that when respondents were asked simply, “Have you been raped?” estimates of the number of people raped in the college population were very low, yet when asked whether they had experienced a series of specific behaviors that would meet legal criteria for rape, estimates of the number of people raped were much higher. Subsequent studies of sexual harassment found similar results (Ilies et al. 2003; Schneider, Pryor, and Fitzgerald 2011), and Fitzgerald and colleagues (1988) established the Sexual Experiences Questionnaire (SEQ) to standardize questions about specific sexual harassment behaviors rather than asking about “sexual harassment” generally. With extensive psychometric evidence supporting it, the SEQ has become the gold standard in the assessment of sexual harassment experiences in both work and school settings (Cortina and Berdahl 2008). Unfortunately, some recent studies attempting to measure the prevalence of sexual harassment have not followed this good practice and are thus likely to have low prevalence rates, be missing data about those who have experienced gender harassment, and as a result be unreliable for evaluating the prevalence of sexual harassment.

Another hurdle faced by surveys on sexual harassment is that women who have experienced sexual harassment may be reluctant to respond to a survey on the topic or to admit being a target or victim because sexual harassment can be stigmatizing, humiliating, and traumatizing (Greco, O’Boyle, and Walter 2015; Bumiller 1987, 1992). To encourage open self-reports, it is important that survey responses are confidential, if not anonymous, and to reassure survey participants that this is the case. Additionally, to help avoid a nonresponse bias (i.e., some segments of a population selectively declining to participate), sexual harassment experts do not use the term sexual harassment or sexual misconduct in the survey title and instead situate their questions about sexual harassment within a broader survey that asks about social concerns such as gender issues, civility, or culture. In a meta-analytic review of the incidence of sexual harassment in the United States, Ilies and colleagues (2003) found that directly asking respondents whether they had experienced sexual harassment (as opposed to using questionnaires that

list behaviors that constitute sexual harassment) led to substantially lower estimates of sexual harassment incidence.

When determining prevalence estimates, attention must be given to minimizing nonresponse biases in the survey sample. Nonresponse biases include attitudes and other characteristics that disincline people from survey participation (Krosnick et al. 2015). A reluctance to answer questions about sexually harassing experiences could represent a nonresponse bias. While low response rates are not synonymous with low levels of nonresponse bias, generally low response rates should be interpreted with caution and will raise limitations on what conclusions can be drawn because of the representativeness of the survey sample (Dillman, Smyth, and Christian 2008; Ilies et al. 2003). Just as it is important to be cautious about deriving prevalence estimates from samples with lower response rates, researchers and leaders in academic institutions must also be judicious when deriving such estimates from nonprobability samples (see Yeager, Krosnick, and Javitz [2009] for a discussion of the problems with opt-in internet surveys).⁶

A challenge for any survey that is particularly important for sexual harassment surveys is their ability to gather information about nonmajority members of a given workplace or campus. Often women of color and sexual- and gender-minority women have been underrepresented among survey respondents, resulting in unreliable prevalence rates for these specific populations. Recent research is beginning to address this by looking at sexual harassment through the lens of intersectionality and by working to oversample these underrepresented populations when conducting surveys.

Convenience sampling (in which participants are recruited from social media or specialized groups with a specific target group in mind) and snowball sampling (recruiting additional subjects by asking participants who else they know in their networks who would also know about the topic) are useful means of recruiting hard-to-reach or underrepresented populations (e.g., lesbians who are not “out” at work, minority groups for whom no lists are available) (Meyer and Wilson 2009). These studies can yield critical insights, even though the samples cannot be considered representative of a particular population. A good example of this approach is the recent study about the experiences of women of color in the fields of astronomy and planetary science, identified via convenience sampling. The researchers found that women of color were more likely to report hearing sexist remarks from supervisors or peers in the workplace than did white women, white men, or men of color. Women of color were also more likely to feel unsafe at work because of their gender (Clancy et al. 2017). This study shows how survey data can be used to test relationships among important variables such as race,

___________________

⁶ Nonprobability samples are samples that are not representative of the whole population and are often used when a defined population is not possible to specify or when it is not necessary to have a representative dataset to achieve the goals of the research. These samples can include convenience samples and snowball samples.

gender, sexual harassment, and sense of safety, yielding conclusions about who is most likely to be targeted for sexually harassing behaviors, and with what effects.

When determining and comparing prevalence rates, it is important to distinguish the prevalence rates for women separate from men and not to rely on a combined prevalence for both genders. Relying on combined rates will result in a lower rate because women are much more likely to experience sexual harassment than men (USMSPB 1995; Magley, Waldo, et al. 1999; Ilies et al. 2003; Kabat-Farr and Cortina 2014).

Another methodological feature to be particularly attentive to when estimating and comparing prevalence rates is the time period respondents are asked about. In some studies, no time limit is given, while others may limit it to the last 12 or 24 months. The longer the time period, the more likely the rates will be skewed and not assess current incidence. Longer time periods can result in higher incidence rates because more time means more women are likely to have experienced such behavior. However, after long enough periods, memory deterioration sets in, leaving behind only those sexual harassment experiences that left a lasting memory, and leaving out everyday sexist comments or ambient harassment. Additionally, longer time periods can also introduce the risk that the incident could have occurred at a past environment, not the current one under investigation.

Lastly, a key obstacle to obtaining accurate prevalence numbers across academia and between fields or workplaces is the number of surveys available that do not always use a standardized method for measuring or defining sexual harassment. Unfortunately, when institutions make their decisions about which survey or questions to use, they often do not seem to be aware of good practices in sexual harassment research or to have consulted with a sexual harassment researcher, because different methodologies and measurement approaches have been used (Wood et al. 2017). As a result, the surveys not only produce unreliable prevalence numbers but also pose a risk of “comparing apples to oranges” when analyzing the data across institutions. The largest concern when comparing prevalence rates is differences in how sexual harassment is defined in the survey and during the analysis of the responses. A meta-analysis of sexual harassment surveys demonstrates that the prevalence rate is 24 percent when women are asked whether they have experienced “sexual harassment” versus 58 percent when they are asked whether they experienced harassing behaviors that meet the definition of sexual harassment (and are then classified as such in the analysis) (Ilies et al. 2003). In other words, the direct query method gives an estimate of prevalence based on the respondent’s perception, while the behavioral experiences method estimates the extent to which potentially harassing incidents happen in an organization. This research also demonstrates that these differences were not due to differences in work environments or to sampling method (Ilies et al. 2003).

To try to present the most accurate information on the prevalence of sexual harassment, the report references surveys that follow good practices in both

sexual harassment research and survey research and that clearly identify differences in time period and definitions.

Experimental Methods

Another way that information has been gathered about sexual harassment has been through laboratory experiments, in which researchers examine the occurrence of sexually harassing behaviors by manipulating variables under controlled conditions. The advantage of this approach is that researchers can directly observe sexually harassing behavior. This approach, however, does not provide information on the prevalence of sexual harassment.

Some of the behaviors that have been directly observed in experiments include the following:

• Unsolicited sexual touching by someone in a supervisory role (Pryor, LaVite, and Stoller 1993);
• Unsolicited touching from peers (Pryor 1987);
• Nonverbal dominance behaviors (Murphy, Driscoll, and Kelly 1999);
• Sending unsolicited pornographic materials electronically (Dall’Ara and Maass 1999; Maass et al. 2003);
• Sending sexist jokes electronically (Galdi, Maass, and Cadinu 2014);
• Sending sexual come-ons electronically (Diehl, Rees, and Bohner 2012);
• Asking sexist questions in an interview (Hitlan et al. 2009); and
• Sexualized behavior, such as staring at a woman’s body, during an interview (Rudman and Borgida 1995).

Laboratory experiments can help uncover situational factors that encourage or discourage potential perpetrators from engaging in sexually harassing behavior. For instance, experiments show that sexual harassment is less likely to occur if those behaviors are not accepted by authority figures (Pryor, LaVite, and Stoller 1993). Another experiment found that men exposed to sexist television portrayals of women were more likely to send sexist jokes to women in an online interaction (Galdi, Maass, and Cadinu 2014).

Laboratory experiments can also provide a snapshot of how women might respond in a sexually harassing situation. For example, research by Woodzicka and LaFrance (2001) reveals the difference between how women think they would respond and how they do respond. In the first study, college women were asked to imagine how they would respond to being asked sexist questions during a job interview. In the second study, women participated in what they thought to be an actual job interview where such questions were asked. Results showed a disconnect between what women thought they would do (get angry, confront, and complain) and what they actually did (become fearful, neither confront nor complain).

On the other hand, there are also limitations to laboratory experiments. While they can reveal responses to actual behaviors, those reactions occur in an artificial laboratory setting (not a real professional or educational setting, with people who have real relationships, interdependencies, status hierarchies, etc.). Participants in experiments are often college students who have limited work experience and diversity (primarily white, middle class, under the age of 20). Also, experiments provide a snapshot of only one moment of time, providing a single look at behaviors and responses. Surveys and accounts from litigants in sexual harassment cases suggest that the worst cases of sexual harassment are not isolated incidents, but something that takes place over a period of time (Cantalupo and Kidder 2017a, 2017b), which experiments cannot assess.

Interviews, Case Studies, and Other Qualitative Methods

Qualitative research offers a wide range of methodologies that can be useful in understanding sexual harassment, though it is best known for individual, semi-structured interviews (Bazeley 2003). Qualitative research can also be conducted in focus groups, bringing together similar constituencies in order to facilitate conversations among participants. Several social science disciplines also use ethnographic or autoethnographic methods. Ethnography is a systematic way of participating and observing in particular settings or cultures to answer research questions about the intersection of culture and lived experience, where autoethnography invites researchers to reflect on their personal experiences, and connect those experiences to a wider research question. For instance, much of the early work on sexual harassment in the field sciences was either interviews or autoethnography, particularly among cultural anthropologists, who often conduct their field work alone (e.g., Sharp and Kremer 2006). Qualitative approaches also include textual analysis of existing primary sources (e.g., studying science syllabi or job postings for gendered language), and case studies or narratives, where a single story is followed in depth. Case study data is often collected via interview, the difference being that rather than interviewing a large enough number to achieve saturation, a researcher will go for greater depth with each participant to construct a more detailed narrative (e.g., Banerjee and Pawley 2013).

Qualitative approaches are widely recognized as the method of choice for generating insight into complex phenomena, the contexts in which they occur, and their consequences (Cho, Crenshaw, and McCall 2013). Such methods are thought to be particularly well suited to providing key background information and highlighting the experiences and perceptions of targets of oppression, such as those who have experienced sexual harassment. The approach also gives a voice to perspectives that tend not to be heard or to those with experiences that have few precedents in prior research (Sofaer 1999).

Sociolegal Methods

Sociolegal studies is an interdisciplinary field in which scholars use all the research methods described above (surveys, experiments, interviews, case studies, ethnography) to study a wide range of topics about formal laws, law-like systems of rules, and the social and political relationships that help constitute what law is (Banakar and Travers 2005). Legal research methods are also a part of sociolegal methods, and these include doctrinal analysis, legal history and doctrinal development studies, and answering questions about exactly what formal legal rules exist across jurisdictions and interrelated areas of law, where there is often ambiguity and conflict. Sociolegal scholars are, of course, attentive to what formal rules and laws actually exist (with sexual harassment, it is Title VII and Title IX doctrines), but a starting approach is to presume that what law is and how it works is much more complex than doctrinal study alone can reveal.

Sociolegal research methods tend to be based in the empirical, observational social sciences supported by legal research. Classic studies using these methods have documented how ordinary people generally resolve their disputes using local customs and norms rather than formal law (Macaulay 1963; Ellickson 1991); how bringing a personal injury claim in a small community is a mark of outsider, subordinated status (Engel 1984); and how difficult it can be for people who have experienced discrimination to use legal protections, because doing so causes them to feel victimized again (Bumiller 1992). These types of sociolegal studies share the strengths and limitations of ethnographic and qualitative research methods generally: on the one hand, they can capture the rich contextual detail of a particular setting, group of people, and set of relationships, but on the other hand, they are limited in time and location, and do not yield broadly generalizable claims. Nonetheless, decades of research using these methods have yielded a considerable body of research that strongly suggests that what the formal law is and what people understand it to be are often quite far apart; that using formal systems to make claims about wrongs done to them is a very difficult thing for most people to do, though it can be empowering and produce social change; and that laws and the legal system typically support existing power structures rather than fundamentally reshape them (Freeman 1978; Edelman 2016; Berrey, Nelson, and Nielsen 2017).

A sociolegal research method requires study of the law at many levels of experience to approach sexual harassment, for example, because it matters just as much what women think they deserve or will likely get as what the law formally offers them. Anna-Maria Marshall’s study of sexual harassment experiences among female staff members at a midwestern university in 1997–1998, for example, combined in-depth interviewing of 25 female staff members with legal analysis at the national level, policy analysis at the university level, and a survey sent to 1,000 female employees selected at random from a university workplace to understand what counted as sexual harassment from their perspectives (Marshall 2005). Whether something in a science, engineering, and medicine

educational or workplace setting is sexual harassment is a category of experience for everyone involved, in other words, that must be assigned meaning, obligations, rights, duties, and processes.

Sociolegal scholars can also bridge between the social science methodologies and the law through research on what they call the “iceberg” or the “tip-of-the-iceberg” problem. The tip-of-the-iceberg problem is the recognition by researchers that published legal disputes are a very skewed and systematically unrepresentative sample from the universe of disputes. As Peter Siegelman and John Donohue (1990) describe the problem, “Most potential disputes never get defined by the actors as such, most actual disputes don’t go to court, most court cases are settled rather than adjudicated, and most adjudicated cases are not appealed” (1133). Their analysis of published and unpublished district court opinions suggests that cases that reach the stage of a published judicial opinion may concern newer areas of case law or more dramatic or unusual circumstances that help explain why these cases were not disposed of earlier and before they appear for researchers to find. Publication as a legal outcome is one of the only ways a sexual harassment case could come to be known and studied, but there are many more legally protected routes to keeping cases and their outcomes from view. Confidential settlements, nondisclosure agreements, confidential notations in an academic or employment record, and dispositions of complaints that are not written down are all outcomes that cannot be studied, tracked, counted, or assessed.

Even when legal scholars attempt to collect samples of hundreds of sexual harassment claims, such as Ann Juliano and Stewart J. Schwab’s 2000 survey of every reported federal district and appellate court ruling on sexual harassment between 1986 and 1995, totaling nearly 650, they concede that these cases are not representative of the universe of incidents. Juliano and Schwab found that the most successful cases involved sexual conduct directed at a specific target in a mostly male workplace that the target had complained about but which the employer had failed to respond to with any formal process (Juliano and Schwab 2000, 593). Another study, Nancy Chi Cantalupo and William Kidder’s (2017b) recent study of sexual harassment in the academic context, attempts to pull cases from as far down the iceberg as possible, drawing in incidents recorded in more venues than the usual publication sources for judicial opinions, including media reports, administrative civil rights investigations at the Departments of Education and Justice, published lawsuits by students, and lawsuits over reinstatement for faculty members fired for sexual harassment. Cantalupo and Kidder find more physical (as opposed to verbal) harassment conduct and more evidence of serial harassers in documented complaints than survey researchers have found, for example. Even if they are not based in representative samples of cases and thus cannot be used to generalize about harassment rates, studies such as these can still yield important research conclusions about sexual harassment adjudications and judicial attitudes toward them.

PREVALENCE OF SEXUAL HARASSMENT

Studies on sexual harassment from the 1980s through today continue to show that sexual harassment of women is widespread in workplaces and that the rates of sexual harassment have not significantly decreased. Studies have also identified common characteristics of sexual harassment in different workplaces and uncovered characteristics of workplaces that are associated with higher rates of sexual harassment. This section and the next one review what research can tell us about the trends in sexual harassment rates over time and what the common characteristics are of sexual harassment and sexually harassing environments.

Wherever possible, the report cites the most recent scientific studies of a topic. That said, the empirical research into sexual harassment, using rigorous scientific methods, dates back to the 1980s. This report cites conclusions from the earlier work when those results reveal historical trends or patterns over time. It also cites results from earlier studies when there is no theoretical reason to expect findings to have changed with the passage of time. For example, the inverse relationship between sexual harassment and job satisfaction is a robust one: the more an individual is harassed on the job, the less she or he likes that job. That basic finding has not changed over the course of 30 years, and there is no reason to expect that it will.

To access the trends in prevalence for sexual harassment, ideally we would examine longitudinal data that uses a well-validated behavior-based instrument for different workplaces and industries; unfortunately, this data is not available. The U.S. Merit System Protection Board (USMSPB) was one of the first organizations to study sexual harassment, with a focus on the federal workforce, which includes a variety of job types and workplace environments. The USMSPB surveys, conducted in 1980, 1987, 1994, and 2016, asked scientifically selected samples of federal workers about their experiences of specific forms of sexual harassment⁷ at work in the past 24 months. These surveys used behavioral questions; however, they did not use the SEQ, and in earlier years the survey did not ask about nonsexualized forms of gender harassment such as sexist comments, which are known to be the most common form of sexual harassment (Kabat-Farr and Cortina 2014). As a result, this is not a good source of longitudinal data covering all three forms of sexual harassment.

This survey does, however, provide an opportunity to assess a population’s understanding of the term sexual harassment. The USMSPB conducted surveys that asked respondents whether they would classify certain behaviors as “sexual harassment.” The results showed that from 1980 to 2016 the proportion of respondents who classify the behaviors as sexual harassment rose, demonstrating

___________________

⁷ The 1980 survey used 6 forms of “unwanted, uninvited sexual harassment,” the 1987 survey used 7 (adding rape and sexual assault), the 1994 survey used 8 (adding rape and stalking), and the 2016 survey used 12 forms (adding gender harassment types). The original six categories remained consistent throughout the years.

an improvement in the population’s understanding of that term. The percentage of men who believe that pressuring a female coworker for sexual favors is sexual harassment rose from 65 percent in 1980 to 93 percent in 1994 and to 97 percent in 2016. Likewise, the percentage of men who perceived unwanted sexual remarks in the workplace as being sexual harassment rose from 42 percent in 1980 to 64 percent in 1994 and to 94 percent in 2016. There was also an increase seen in the perceptions of women—the percentage of women who considered a coworker’s sexual remarks as sexual harassment rose from 54 percent in 1980 to 77 percent in 1994 and to 95 percent in 2016. It is also significant to note that of respondents experiencing sexual harassing behaviors in the 2016 survey, only about 11 percent took any kind of formal action, such as filing a complaint or report with their organization (USMSPB 2018). As the results just discussed demonstrate, this lack of reporting was not due to respondents inaccurately defining sexual harassment; rather, it reflects a reluctance by people to take formal action, which will be discussed in more detail in Chapter 4.

The U.S. military is the other organization to study sexual harassment through large surveys early on and over multiple years. Starting in 1995 and going to 2012⁸ the Defense Manpower Data Center (DMDC) has used an SEQ-format survey that asked about more than 20 specific sex- or gender-related behaviors experienced in the past 12 months. As shown in the results in Table 2-1, the data demonstrate that the prevalence of all three types of sexual harassment has been consistent. It also demonstrates that the gender-harassing form of sexual harassment (broken out into crude and offensive behavior and sexist behavior) is by far the most prevalent type of sexually harassing behavior, a finding that is consistent with research in other workplace settings (Kabat-Farr and Cortina 2014).

Given that there is limited longitudinal data on the prevalence of sexual harassment that uses a well-validated behavior-based instrument, the best analysis of the prevalence of sexual harassment across workplaces and time comes from a meta-analysis by Ilies and colleagues (2003). Based on more than 86,000 respondents from 55 probability samples, Illies and colleagues demonstrate that on average, 58 percent of women experience sexually harassing behaviors at work.

___________________

⁸ After the 2012 survey, the military asked the RAND Corporation to conduct a new survey revising the methodology as needed. The result was a significant change in how sexual harassment was defined in the analysis, and thus the prevalence numbers cannot easily be compared with the previous series of surveys. Whereas previous surveys assessed the prevalence of sexually harassing behaviors, the RAND survey used behavior-based questions to determine the prevalence rate of legally defined sexual harassment, meaning that they asked questions and grouped results based on hostile work environment and quid pro quo harassment. While quid pro quo harassment maps cleanly to sexual coercion, hostile work environment requires the condition that the sexually harassing behaviors (such as gender harassment and unwanted sexual attention) be considered by the respondent to be pervasive or severe—essentially requiring a frequency or severity assessment that had not been previously used. With this much narrower definition of “what counts” as harassing behavior, the 2016 survey yielded a lower overall rate of sexual harassment for women over a 12-month time period: 21.4 percent (RAND 2016).

Looking further into the different workplace sectors, the researchers found that there was some variation between sectors, with the prevalence ranging from 43 to 69 percent (this is discussed further in Chapter 3 when comparing the academic environment to other sectors). Their analysis of trends over time revealed that over the 25 years examined, women who responded to surveys with behavioral-based instruments (and which used a probability sample) reported increasingly more experiences of sexual harassment. The authors note that their data cannot investigate the reasons for this change, and that only a time-trend analysis of data obtained from the same instruments can truly answer the question of what is the trend in prevalence rates.

CHARACTERISTICS OF SEXUAL HARASSMENT AND SEXUALLY HARASSING ENVIRONMENTS

Rigorous survey research has identified common characteristics of sexual harassment. This work pushes against some of the main assumptions made on what it is, as well as how sexual harassment affects the targets, the bystanders, and the atmosphere of work and education settings. Here the chapter describes some of the aspects of sexual harassment that are strongly supported by the literature. However, we note that the data on varying experiences of sexual harassment of women of color, sexual minorities, and gender minorities is sparse, so these characteristics are likely to reflect the experience of majority women.

Characteristics of Sexual Harassment

Women are more likely to be sexually harassed than men and to experience sexual harassment at higher frequencies (USMSPB 1995; Magley, Hulin et al. 1999; Ilies et al. 2003; Kabat-Farr and Cortina 2014). The 2012 DMDC survey results shown in Table 2-2 demonstrate that across all three types of sexual harassment, female personnel, compared with their male counterparts, were more

TABLE 2-2 Rate of Active Duty Military Women and Men Experiencing Sexually Harassing Behaviors at Least Once in the Past 12 Months

SOURCE: DMDC 2013.

likely to have experienced at least one instance of sexually harassing conduct over the prior 12 months. Likewise, in the 1994 USMSPB study of federal workers, it found more women (44 percent) than men (19 percent) describing experiences of any of seven types of sexually harassing behavior in the past 2 years at work (USMSPB 1995). In a more recent study using the SEQ, Rosenthal, Smidt, and Freyd (2016) surveyed 525 graduate students regarding their exposure to sexual harassment while in graduate school. Female students were 1.64 times more likely to have experienced sexually harassing behavior from faculty or staff (38 percent) compared with male students (23 percent). Though the occasional survey reports no significant gender difference (e.g., Konik and Cortina 2008) in a specific group, many studies have found women encountering more sexually harassing conduct than men encounter.

The overwhelming majority of sexual harassment involves some form of gender harassment (the put-downs of sexual harassment that include sexist hostility and crude behavior). Unwanted sexual attention is the next most common form of sexual harassment, and only a small minority of women experience sexual coercion. For instance, Schneider, Swan, and Fitzgerald (1997) analyzed data from two samples of women: factory workers and university faculty/staff. In both samples, gender harassment was by far the most common experience: 54–60 percent of women described some encounter with gender harassment, either with or without unwanted sexual attention. In contrast, sexual coercion was rare, described by approximately 4 percent of women in each sample. Moreover, sexual coercion never took place without unwanted sexual attention and gender harassment. When analyzing the sexual harassment of graduate students, Rosenthal, Smidt, and Freyd (2016) found that 59 percent of harassment incidents involved some form of gender harassment, while only 5 percent included unwanted touching, and less than 4 percent entailed sexual coercion. In another study, Leskinen, Cortina, and Kabat (2011) analyzed survey data from two samples of women who work in highly male-dominated sectors: the military and the law. Focusing only on data from women who had encountered at least one sexually harassing behavior in the prior year, they found that 9 of every 10 people who experienced sexual harassment had encountered gender harassment with little or no unwanted

sexual attention or coercion. While a recent national survey of 615 working men found that of the 25 percent of male respondents that admitted they had done at least one sexually harassing behavior in the last year, the most common form was gender harassment and the least common was sexual coercion (Patel, Griggs, and Miller 2017).

That gender harassment is the most common type of sexual harassment is an unexpected finding in terms of what constitutes sexual harassment because unwanted sexual advances and sexual coercion are the most commonly reported both in official Title IX/Human Resources documentation (Cantalupo and Kidder 2017a, 2017b) and in the media.⁹ This is in part why the misguided idea that sexual harassment is about sex has persisted.

In the vast majority of incidents of sexual harassment of women, men are the perpetrators. For instance, in the 1994 USMSPB study, 93 percent of sexually harassed women reported their perpetrators to be male (USMSPB 1995). The DMDC’s 1995 study turned up remarkably similar results, with 92 percent of sexually harassed women describing male perpetrators (Magley, Waldo et al. 1999). In Rosenthal, Smidt, and Freyd’s (2016) study of the sexual harassment of graduate students, among those who had been sexually harassed by faculty/staff, 86 percent of women described their harassers as male. Even when men are the targets of sexually harassing conduct, more often than not the perpetrator is also male (see also Kabat-Farr and Cortina 2014; Magley, Waldo et al. 1999).

Women are frequently harassed by coworkers and other employees (for students, it is fellow peers); superiors are not the most common perpetrators¹⁰ (USMSPB 1995, 2018; AAUW 2005; Schneider, Pryor, and Fitzgerald 2011; Rosenthal, Smidt, and Freyd 2016). For example, in Rosenthal, Smidt, and Freyd’s (2016) study of graduate students, 38 percent of female participants self-reported that they had experienced sexual harassment from faculty or staff, while 58 percent described sexual harassment from other students. In a study by Huerta and colleagues (2006), student targets of sexual harassment described the harassing experience that bothered them the most. Fully three-quarters of these targets indicated the perpetrator of this “most bothersome” incident to be a peer (fellow student), whereas only one-quarter had perpetrators who were higher-status individuals (staff, faculty, or administrators).

___________________

⁹ See, for example, https://www.nytimes.com/2017/10/05/us/harvey-weinstein-harassment-allegations.html?rref=collection%2Fbyline%2Fjodi-kantor; https://www.nytimes.com/2017/10/10/us/gwyneth-paltrow-angelina-jolie-harvey-weinstein.html?rref=collection%2Fbyline%2Fjodi-kantor&action=click&contentCollection=undefined&region=stream&module=stream_unit&version=latest&contentPlacement=10&pgtype=collection; https://www.buzzfeed.com/azeenghorayshi/geoff-marcy-at-sfsu?utm_term=.phP5anr0n#.kprpq6Gj6; https://www.buzzfeed.com/azeenghorayshi/ott-harassment-investigation?utm_term=.vi3ByvlNv#.wm83947r4; and https://www.reuters.com/article/us-foxnews-lawsuit/ex-fox-news-anchor-accuses-former-boss-ailes-of-sexual-harassment-idUSKCN0ZM21I.

¹⁰ One obvious factor that contributes to this difference is that there are most often more coworkers or peers than there are superiors.

Targets of sexual harassment often face repeated sexually harassing behaviors rather than one single incident. Rosenthal, Smidt, and Freyd’s 2016 study of graduate students, in which 38 percent of women had encountered sexual harassment from faculty/staff and 58 percent had faced sexual harassment from students, only a small fraction (one-third or less) of these women described their harassment experience as being limited to a single incident. This confirms earlier research using data from the 1987 USMSPB survey, in which researchers found that “75 percent of those experiencing sexual teasing and jokes reported that it was not a one-time occurrence, and 54 percent of those pressured for sexual favors reported that it had occurred more than once (USMSPB 1988). For most women, the harassment lasted more than a week, and often as long as 6 months” (Schneider, Swan, and Fitzgerald 1997, 402).

Sexual Harassment Among Women of Color and Sexual- and Gender-Minority Women

What is known about women’s experiences is that those who have multiple marginalities—for instance women of color and sexual- and gender-minority women—experience certain kinds of harassment at greater rates than other women (e.g., Buchanan, Settles, and Woods 2008; Clancy et al. 2017; Cortina 2004; Cortina et al. 1998; Konik and Cortina 2008; Rabelo and Cortina 2014). Additionally, the cultural context in which people from different racial and ethnic backgrounds operate, as well as when they are numerically less represented in a workplace, can have effects on how they experience sexual harassment (Cortina et al. 2002; Welsh et al. 2006). Thus, there is a wide spectrum of vulnerabilities, experiences, and consequences for women of color and gender minorities who are sexually harassed in the workplace.

As a field of study and as an analytical lens, intersectionality provides a framework to make visible the mutually constitutive relationship among race, ethnicity, sexuality, class, and other social positions that affect targets’ experiences of harassment (Collins 2015). It is rooted in Black feminism and Critical Race Theory and also makes visible intersecting axes of oppression that contribute to power hierarchies within a social structure related to race, ethnicity, gender, sexuality, and class. Addressing the legacy of exclusions of black women, legal scholar Kimberlé Williams Crenshaw used the concept of intersectionality to highlight the intersection of race and gender discrimination and how treating them as exclusive, and not intertwined, rendered the discrimination and multiple marginalities faced by black women invisible to antidiscrimination law (Crenshaw 1989, 1991). More recently, Crenshaw described intersectionality as a work in progress to denote the movement in and broadening of its use across disciplines and to a wider range of social locations (Carbado 2013; Crenshaw 2014).

Some scholars have applied an intersectional lens to examine the sexual harassment experiences of women of color, though research in this area is still

very limited. It is important to prioritize the study of sexual harassment among noncisgender (cisgender means feeling aligned with the gender you were assigned at birth), nonstraight, nonwhite women when considering the impact of sexual harassment within an organization. Recent research that has begun to look at sexual harassment through the lens of intersectionality reveals how the experiences of women of color compare with that of white women, white men, and men of color. This research demonstrates that women of color and sexual- and gender-minority women sometimes experience sexual harassment differently from other populations. Women of color often experience sexual harassment as a manifestation of both gender and race discrimination (Cortina et al. 2002; Murrell 1996), which combined can lead to higher rates of overall harassment (Berdahl and Moore 2006; Woods, Buchanan, and Settles 2009).

The RTI International interviews¹¹ were able to glean complexities of intersectionality and sexually harassing behavior. Respondents noted that the issues of sexual- and gender-based harassment are often overpowered by how other issues such as race and sexual orientation intersect with their lived experience as women. These women noted an inability to disentangle discrimination and biases as stemming either from gender or their intersecting identities (RTI 2018).

These studies demonstrate that an individual’s identity can affect how sexual harassment is perpetrated.

Likewise, lesbian, gay, and bisexual women encounter forms of harassment that reflect a combination of sexism and heterosexism (Konik and Cortina 2008; Rabelo and Cortina 2014). Nonbinary individuals, on the other hand, must negotiate their identities within the constructs of the gender binary that is still prevalent today (Dietert and Dentice 2009). A study by Irwin (2002) examined workplace discrimination in the education sector in Australia among gay men, lesbians, and transgender individuals. Irwin found that greater than 60 percent of teachers,

___________________

¹¹ This research was commissioned by the committee and the full report on this research is available in Appendix C.

academics, and educators who identified as lesbian, gay, or transgender have experienced homophobic behavior and/or harassment, and have been discriminated against in the workplace. The study also found that 16 percent of the individuals who identified as lesbian, gay, or transgender have been sexually harassed, and one participant was sexually assaulted.

The research on sexual minorities has shown that this population experiences more sexual harassment than heterosexual individuals. In a study of 629 employees in higher education, nearly 76.9 percent of sexual minorities (of both genders) experienced gender harassment, whereas only 30 percent of heterosexuals (of both genders) experienced gender harassment (Konik and Cortina 2008). This trend continued for the other forms of sexual harassment (unwanted sexual attention and sexual coercion): 39.7 percent of sexual minorities experienced these types, whereas only 15.5 percent of heterosexuals experienced these types. In another study the prevalence and impact of heterosexist harassment, which is insensitive verbal and symbolic (but nonassaultive) behaviors that convey animosity toward nonheterosexuality, was examined among students. The study specifically looked at how experiences of this type of harassment affected sexual minorities and heterosexuals differently and found that sexual minorities were more likely to experience heterosexist harassment than heterosexuals (58 percent and 39 percent, respectively), and when sexual minorities experienced the harassment, they were equally likely to experience it directed at them as in an ambient form (53 percent and 47 percent, respectively) (Silverschanz et al. 2008).

Characteristics of Sexually Harassing Environments

By far, the greatest predictors of the occurrence of sexual harassment are organizational. Individual-level factors (e.g., sexist attitudes, beliefs that rationalize or justify harassment, etc.) that might make someone decide to harass a work colleague, student, or peer are surely important. However, a person that has proclivities for sexual harassment will have those behaviors greatly inhibited when exposed to role models who behave in a professional way as compared with role models who behave in a harassing way, or when in an environment that does not support harassing behaviors and/or has strong consequences for these behaviors. Thus, this section considers some of the organizational and environmental variables that increase the risk of sexual harassment perpetration.

Women working in environments where men outnumber women, leadership is male-dominated, and/or jobs or occupations are considered atypical for women experience more frequent incidents of sexual harassment (USMSPB 1995; Fitzgerald et al. 1997; Berdahl 2007b; Willness, Steel, and Lee 2007; Schneider, Pryor, and Fitzgerald 2011). In particular, the more male-dominated the work environment, the more women experience the gender harassment form of sexual harassment. For example, in one study looking at the effect of workplace gender balance, the researchers analyzed data from women employees of the federal

courts. When comparing women who work in gender-balanced workgroups (i.e., equal numbers of men and women in the workgroup) with those who work with almost all men, the researchers reported women in the latter category were 1.68 times more likely to encounter gender harassment (Kabat-Farr and Cortina 2014).

The historical and cultural context of a work or education environment is of high relevance to the study of sexual harassment as well, since environments that are no longer male dominated in gender ratio may still be male dominated in their work practices, culture, or behavioral expectations.

The perceived absence of organizational sanctions increases the risk of sexual harassment perpetration. Perceptions of organizational tolerance for sexual harassment (also referred to as organizational climate for sexual harassment), are broken down into three categories: (1) the perceived risk to targets for complaining, (2) a perceived lack of sanctions against offenders, and (3) the perception that one’s complaints will not be taken seriously (Hulin, Fitzgerald, and Drasgow 1996). Research has shown that perceptions of an organization’s tolerance for all three forms of sexually harassing behavior are significantly related to both direct and ambient sexual harassment. In environments that are perceived as more tolerant or permissive of sexual harassment, women are more likely to be directly harassed (Fitzgerald et al. 1997; Williams, Fitzgerald, and Drasgow 1999) and to witness harassment of others (Glomb et al. 1997). In fact, one meta-analysis that combined data from 41 studies with a total sample size of nearly 70,000 respondents found perception of organizational tolerance to be the most potent predictor of sexual harassment in work organizations (Willness, Steel, and Lee 2007). In a recent national survey of 615 working men (Patel, Griggs, and Miller 2017), sexually harassing behavior was more commonly reported “among men who say their company does not have guidelines against harassment, hotlines to report it or punishment for perpetrators, or who say their managers don’t care.”

Social situations in which sexist views and sexually harassing behavior are modeled can enable, facilitate, or even encourage sexually harassing behaviors, while, conversely, positive role models can inhibit sexually harassing behavior (Dekker and Barling 1998; Perry, Schmidtke, and Kulik 1998; Pryor, LaVite, and Stoller 1993). In one study, college men who had professed a willingness to sexually coerce were found to be more likely to sexually exploit a female trainee when they were exposed to an authority figure who acted in a sexually exploitive way (Pryor, LaVite, and Stoller 1993). Hitlan and colleagues (2009) found that viewing a sexist film enhanced the tendency among the less sexist men to perform acts of gender harassment. In another experiment, men who viewed sexist TV clips were more likely to send women unsolicited sexist jokes and more likely to profess a willingness to engage in sexual coercion than men who watched programs portraying young, successful women in domains such as science, culture, and business (Maass, Cadinu, and Galdi 2013). Conversely, experiments show that sexual harassment is less likely to occur if those behaviors are not accepted by authority figures (Pryor, LaVite, and Stoller 1993). So, while social situations

do not necessarily function as triggers for existing predilections to sexually harass, they can act as a force encouraging or discouraging men to sexually harass, demonstrating the power of practiced social norms (e.g., the social norms communicated by the actions of the people in an environment rather than their words or the words from official policy for an organization).

Other factors that research suggests increase the chances of sexual harassment perpetration are significant power differentials within hierarchical organizations and organizational tolerance of alcohol use. Hierarchical work environments like the military, where there is a large power differential between organizational levels and an expectation is not to question those higher up, tend to have higher rates of sexual harassment than organizations that have less power differential between the organizational levels, like the private sector and government (Ilies et al. 2003; Schneider, Pryor, and Fitzgerald 2011). Environments that allow drinking during work breaks and have permissive norms related to drinking are positively associated with higher levels of gender harassment of women (Bacharach, Bamberger, and McKinney 2007). Culturally, these are, again, patterns more common in currently or historically male-dominated workplaces.

FINDINGS AND CONCLUSIONS

1. Sexual harassment is a form of discrimination that consists of three types of harassing behavior: (1) gender harassment (verbal and nonverbal behaviors that convey hostility, objectification, exclusion, or second-class status about members of one gender); (2) unwanted sexual attention (unwelcome verbal or physical sexual advances, which can include assault); and (3) sexual coercion (when favorable professional or educational treatment is conditioned on sexual activity). The distinctions between the types of harassment are important, particularly because many people do not realize that gender harassment is a form of sexual harassment.
2. Sexually harassing behavior can be either direct (targeted at an individual) or ambient (a general level of sexual harassment in an environment) and is harmful in both cases. It is considered illegal when it creates a hostile environment (gender harassment or unwanted sexual attention that is “severe or pervasive” enough to alter the conditions of employment, interfere with one’s work performance, or impede one’s ability to get an education) or when it is quid pro quo sexual harassment (when favorable professional or educational treatment is conditioned on sexual activity).
3. There are reliable scientific methods for determining the prevalence of sexual harassment. To measure the incidence of sexual harassment, surveys should follow the best practices that have emerged from the science of sexual harassment. This includes use of the Sexual Experiences

4. Questionnaire, the most widely used and well-validated instrument available for measuring sexual harassment; assessment of specific behaviors without requiring the respondent to label the behaviors “sexual harassment”; focus on first-hand experience or observation of behavior (rather than rumor or hearsay); and focus on the recent past (1–2 years, to avoid problems of memory decay). Relying on the number of official reports of sexual harassment made to an organization is not an accurate method for determining the prevalence.

5. Some surveys underreport the incidence of sexual harassment because they have not followed standard and valid practices for survey research and sexual harassment research.
6. While properly conducted surveys are the best methods for estimating the prevalence of sexual harassment, other salient aspects of sexual harassment and its consequences can be examined using other research methods, such as behavioral laboratory experiments, interviews, case studies, ethnographies, and legal research. Such studies can provide information about the presence and nature of sexually harassing behavior in an organization, how it develops and continues (and influences the organizational climate), and how it attenuates or amplifies outcomes from sexual harassment.
7. Sexual harassment remains a persistent problem in the workplace at large. Across workplaces, five common characteristics emerge:
   1. Women experience sexual harassment more often than men do.
   2. Gender harassment (e.g., behaviors that communicate that women do not belong or do not merit respect) is by far the most common type of sexual harassment. When an environment is pervaded by gender harassment, unwanted sexual attention and sexual coercion become more likely to occur—in part because unwanted sexual attention and sexual coercion are almost never experienced by women without simultaneously experiencing gender harassment.
   3. Men are more likely than women to commit sexual harassment.
   4. Coworkers and peers more often commit sexual harassment than do superiors.
   5. Sexually harassing behaviors are not typically isolated incidents; rather, they are a series or pattern of sometimes escalating incidents and behaviors.
8. Research that does not include the study of women of color and sexual- and gender-minority women presents an incomplete picture of women’s experiences of sexual harassment. The preliminary research on the experiences of women of color, and sexual- and gender-minority women reveals that their experiences of sexual harassment can differ from the larger population of cisgender, straight, white women.

8. 1. Women of color experience more harassment (sexual, racial/ethnic, or combination of the two) than white women, white men, and men of color do. Women of color often experience sexual harassment that includes racial harassment.
   2. Sexual- and gender-minority people experience more sexual harassment than heterosexual women do.
9. The two characteristics of environments most associated with higher rates of sexual harassment are (a) male-dominated gender ratios and leadership and (b) an organizational climate that communicates tolerance of sexual harassment (e.g., leadership that fails to take complaints seriously, fails to sanction perpetrators, or fails to protect complainants from retaliation).
10. Organizational climate is, by far, the greatest predictor of the occurrence of sexual harassment, and ameliorating it can prevent people from sexually harassing others. A person more likely to engage in harassing behaviors is significantly less likely to do so in an environment that does not support harassing behaviors and/or has strong, clear, transparent consequences for these behaviors.