There can be little doubt that sexual misconduct is a system-wide criminal justice concern. Public outrage and scrutiny has forced attention to these acts of official oppression which indicate serious personnel and perhaps agency and institutional level problems. Police officer sexual assault is pointedly disturbing and marks the polar opposite behavior the citizenry expects from sworn law enforcement agents. In addition to undermining public confidence and deteriorating police-community relations, officer sexual misconduct shakes the very foundation of democratic society by challenging trust in the rule of law and civil liberty.

The problem is also a current concern in corrections. The U.S. Congress formed the National Prison Rape Elimination (PREA) Commission in 2005 to analyze and reduce sexual violence in correctional settings. The PREA Commission implemented a national research agenda which examined both inmate and staff perpetrators and found several accounts of correctional officer sexual assault, an understandably sensitive, controversial, and highly confidential subject.

The topic of sexual misconduct is so sensitive, in fact, that researchers typically are met with near categorical resistance. A range of barriers exist to studying the topic. Data are embedded within internal affairs or agency investigator offices (a vault within a vault of sorts), are potentially litigious, and usually only made available to researchers in heavily redacted form. The potential findings from such analyses carry implications for administrators and frontline practitioners alike. The former are on tilt between the needs of minimizing negative public relations, honest disclosure, and proactively addressing problematic situations. The bulk of practitioners, in turn, rally around unions and officer associations who attempt to block and hamper legal and scientific scrutiny, giving reaffirmation to the suspicions of the dark side of police subculture.

This paper considers the indicators and scope of sexual harassment and sexual assault between 2002 and 2006 in the United States military. After reviewing extant literature on sexism, generally, and sexual harassment and sexual assault more specifically, the multivariate models and analytic techniques used to examine survey reported misconduct data are described. Findings inform discussion of policy and practice implications with particular attention to criminal justice best practices.

Background

As emphasized by Firestone and Harris (1994, 1999, 2003, 2008, Firestone 2007) the U.S. military provides an interesting context for analyzing sexual harassment (and sexual assault) behaviors. The military, including the reserve component, is large enough to provide an adequate sample of individuals across various demographic group memberships (sex, race, ethnicity, age) for meaningful comparisons. The active duty component, at least, claims to be the largest equal opportunity employer in the U.S. Following orders and invoking hierarchical decision making is ingrained in the military culture so that personal opinions/prejudices are supposedly irrelevant to duty requirements. In addition, within the military system, cohesion is very highly valued, and divulging negative information about fellow soldiers or about the organization in general is taboo.

Interestingly, cohesion has been used to exclude rather than include individuals seen as “outsiders” or “troublemakers” in militaristic settings (e.g., women, race/ethnic minority members; see Harris and Firestone 1997; Shields 1998). Furthermore, since harassment in general is part of military culture, and sexual harassment may be a subset of those incidents, the military environment may be less open to receiving complaints (either formal or informal). In addition, sexual assault has been used as a technique of control by those with power over those with less power, especially in closed organizational settings such as prisons and law enforcement agencies. Below, we review the primary forms of sexual misconduct in order of ascending severity from sexism to sexual harassment and sexual assault.

Sexism

Normative assumptions about how women and men differ with regard to work-related skills, attitudes and knowledge typically suggest that women are deficient compared to male peers (Heilman 1997). For example, the pop-psychology literature is replete with self-help advice for women on overcoming deficiencies by “succeeding at corporate gamesmanship” (Harragan 1977), “breaking into the boys’ club” (Jardim and Hennig 1990), “improving communication styles and supervising skills” (Feuer 1988; Fierman 1990), and how to be “feminine and still succeed in the workplace” (Foley 2007; Trunk 2007; Wish 2008). These attitudes often translate, whether intentionally or not, into sexist behaviors.

Sometimes labeled gender harassment, sexism includes generalized sexual or sexist comments or behaviors that insult, degrade or embarrass based on gender, typically women. Sexist attitudes originate from stereotypical views of gender appropriate behavior (De Judicibus and McCabe 2001; Bem 1974), such as masculine traits of rationality, risk taking, and aggression. Feminine traits include nurturance, emotional expressiveness, and self-subordination. These attitudes result in the stereotypical beliefs that women are inferior to men (particularly in the paid workplace) and that men have the prerogative to initiate sexual behavior—a risky mantra for an environment where pressure and invitation are often blurred (Bartling and Eisenman 1993; Walker et al. 1993. McElroy et al. (1996; see also, Saal and Moore 1993; McEnrue 1989) found that blacks and women are more likely than white males to perceive promotions are based on unfair criteria (e.g. “she slept her way to the top” or if a black is promoted it constitutes “reverse discrimination”). Such perceived inequities are associated with reduced job satisfaction, increased work attendance and organizational commitment (McElroy et al. (1996). Thus, an environment can be sexist, based on perceptions of inequity, although the behaviors creating that situation may not constitute the legal definition of sexual harassment.

Sexism relates to both sexual harassment and sexual assault because people with sexist attitudes are unlikely to believe a target who says the behavior was unwanted and may blame the target for having in some way encouraged the perpetrator (Valentine-French and Radtke 1993). Glick and Fiske (1996) proposed that sexism may not only be a single concept; rather attitudes toward women may be ambivalent, comprising “hostile sexism” and “benevolent sexism.” Hostile sexism can be described as the negative attitude toward women that is commonly associated with sexist prejudices (e.g., Tougas et al. 1995). In contrast, benevolent sexism can be characterized as a set of attitudes that are sexist in their manifestation of stereotypical roles for women but are subtly positive and affectionate towards women (Harris and Firestone 1997; Glick and Fiske 1996). According to Glick and Fiske (1996), ambivalent sexists reconcile their hostile and benevolent attitudes by differentiating between “good” and “bad” women. Thus, benevolence is targeted at those women that conform to traditional roles (“good girls”), whereas hostility is reserved for women in nontraditional roles (“bad girls”) (Glick et al. 1997). This differentiation between “good” and “bad” subcategories of women appears to provide a means for men to justify and excuse aggressive behaviors towards some women. Such behaviors may include sexual harassment and sexual assault.

Sexual Harassment

Sexual harassment in the workplace has been the focus of much academic research across disciplines as well as much media attention. Research shows that sexual harassment is a widespread phenomenon with negative consequences for both individuals and organizations such as career interruptions, lowered productivity, lessened job satisfaction, lowered self confidence, loss of motivation, physical health ailments, and loss of commitment to work and employer (Crull 1982; DiTomaso 1989; Gutek 1985; Gutek and Koss 1993, USMSPB 1981, 1987, 1995; Dansky and Kilpatrick 1997; Faley 1991; Niebuhr 1997).

The original definition of sexual harassment for the military was “deliberate or repeated unsolicited verbal comments, gestures, or physical contact of a sexual nature which are unwelcome” (USMSPB 1981). The initial definition was expanded to include any conduct of a sexual nature which created “an intimidating, hostile, or offensive working environment” (USMSPB 1995). Even the expanded definition is so broad, however, that empirical and theoretical inconsistencies arising from specific studies remain (Schneider 1982).

For instance, definitions are sometimes inconsistent and often discipline-specific, which further confounds clear conceptualizations (Terpstra and Baker 1986). Recognizing that considerable overlap in conceptualizations exists, most researchers use the definitions specific to their discipline. Military sociologists and criminologists focus on organizational and societal level environmental variables (e.g., power/status differences); psychologists focus on individual variables (e.g., sexist attitudes); economists look at labor market issues (e.g., who benefits?); while organizational/business studies use work structures (e.g., formal/informal hierarchies, power dynamic, organization culture). As a result, the body of literature available may be so restricted that it is only useful within a specific discipline or for a single explicit purpose.

Because the defining criteria for identifying sexual harassment have been “uninvited and unwanted,” other complicating factors lie in the perceptions and evaluations of being “unwanted.” Definitions of “acceptable” versus “unwanted” as well as their visions of effective policies are likely to differ vastly between the perpetrators and the targets as well as by gender (Baker et al. 1990; Dougherty 1999, 2006; Fitzgerald and Ormerod 1991; Loredo et al. 1995; Saal 1996; Sev’er and Ungar 1997).

Perhaps most problematic is that virtually any behavior, including requests for dates, pressure for sexual activities, comments, jokes, and aggression up to rape can constitute sexual harassment. Many argue that definitions of these behaviors as sexual harassment could vary systematically depending on individual characteristics as well as the specific contexts in which the behavior occurred. In other words, some argue that sexual harassment appears highly subjective and the experiences of women and men are variable and open to alternative explanations (Dougherty 1999, 2006; Gordon 1981).

Sexual harassment in the workplace has typically been characterized as consisting of two forms, both of which are defined legally. The quid pro quo type is the easiest to identify and, although frequencies are low, it is the most likely to be challenged. This form includes the exchange of work-related benefits or consequences for sexual favors through bribes, threats or even physical force (see Firestone and Harris 1994). The second form, environmental harassment, includes unwanted sexualized actions to alter, interfere with or affect one’s work performance by creating a hostile and offensive work climate (Firestone and Harris 1994; Sev’er 1999). The definition of this type is blurred regarding how to ascertain whether an act is “unwanted” and on deciding on whom the burden of proof should fall that the action was against the individual’s will. Expectations of economic losses and/or psychological pain due to the harassment have also been an issue as some courts demand that targets have proof of both before claims of environmental harassment can proceed.

Two Supreme Court rulings inform environmental harassment. First, a “reasonable” woman standard grants any woman classified as reasonable to assess whether she is being subject to harassment or to acceptable behaviors (e.g., teasing, fun jokes, etc., Greenhouse 1993; Wells and Kracher 1993). Second, the ruling that “psychological stress” does not have to be documented by medical professionals establishes precedent for allowing women to interpret their experiences within the boundaries of the organization (Wells and Kracher 1993).

An organization’s culture includes the value and belief system, including regularities, norms, rules for working and getting along, and the organizational climate (Schein 1990, 1996). Bastien et al. (1995) demonstrated the ways in which culture is structured to have considerable impact on how people behave in an organizational setting. How sexual harassment policies (reporting procedures, training) are exacted is impetus for the stories employees tell about an organization’s willingness to eradicate or perpetuate sexual harassment (Conrad and Taylor 1994; Hulin et al. 1996).

One aspect of organizational culture derives from the gender balance of the employees. It seems clear that sexual harassment is more prevalent in male-dominated occupations (e.g. police, professional sports, military). It has been suggested that work settings that place a high value on “masculine” qualities such as power, toughness, dominance, aggressiveness, and competitiveness may contribute to negative attitudes toward women (Firestone and Harris 2008). In addition, there is some evidence that fields such as the military may attract individuals who possess more traditional gender-role attitudes. In such a setting, women may be seen as disrupting the masculine camaraderie that infuses the culture of the occupation and weans a “warrior culture” traditionally deemed necessary to maintain a ready and effective fighting force (Fitzgerald et al. 1995).

Sexual Assault

The term sexual assault has been used to describe a large range of nonconsensual sexual behaviors from kissing and/or touching to coerced penetration by physical force or threat of force. The question of force is balanced against conceptualizations of “bad girls” who deserve bad things and “good girls” who need protection strongly impact how most script rape (Tendayi et al. 2004; Conly 2004). If a victim is considered incapable of giving consent (due to age, mental/physical status, intoxication), the act may be considered rape or sexual assault and attempted rape is often considered the equivalent of actual rape. Whether rape is subsumed under sexual harassment or sexual harassment is considered a form of rape, conceptual distinctions between the two become clouded and provide some with the evidence to contend that sexually wrong behaviors are in the eye of the beholder.

While most people believe that rape in the workplace is uncommon, Lee and Kleiner (2003) contended that at the time of their research, 51,000 rapes/sexual assaults occurred in the workplace each year. Being sexually assaulted in the workplace not only leads to physical injuries and psychological trauma, many victims develop symptoms similar to posttraumatic stress disorder (PTSD). This reaction may be compounded when the environment where the assault occurred is a military workplace where women are sometimes viewed as “outsiders” or as “bad girls.” In either case, such women are sometimes perceived as deserving whatever happens to them (Lee and Kleiner 2003; Tendayi et al. 2004).

The definition of sexual assault and rape has evolved from one designed to control “competing male interests in controlling sexual access to females, rather than protecting women’s interests in controlling their own bodies and sexuality” (Greenberg et al. 2004: 776; Hasday 2000) to a code focused on the use of force and lack of consent (Lyon 2004). The Uniform code of Military Justice (UCMJ) originally defined sexual assault as:

a crime…; intentional sexual contact, characterized by use of force, physical threat or abuse of authority or when the victim does not or cannot consent. Sexual assault includes rape, nonconsensual sodomy (oral or anal sex), indecent assault (unwanted, inappropriate sexual contact or fondling), or attempt to commit these acts.…Consent” shall not be deemed or construed to mean the failure by the victim to offer physical resistance (DoD 2004).

Revised Uniform Code of Military Justice Provisions

In Section 522 of the NDAA for FY 2006, Congress amended the UCMJ regarding sex offenses to consolidate and reorganize the array of military sex offenses under Article 120, UCMJ, “Rape, Sexual Assault, and Sexual Misconduct.” These revised provisions took effect October 1, 2007. As amended, rape is defined in the UCMJ as a situation where any person causes another person of any age to engage in a sexual act by: (1) using force; (2) causing grievous bodily harm; (3) threatening or placing that other person in fear that any person will be subjected to death, grievous bodily harm, or kidnapping; (4) rendering the person unconscious; or (5) administering a substance, drug, intoxicant or similar substance that substantially impairs the ability of that person to appraise or control conduct. The revised Article 120 of the UCMJ defines “consent” as “words or overt acts indicating a freely given agreement to the sexual act at issue by a competent person.”

Military crime statistics for 2007 indicated that 2,085 total sexual assaults were reported by or against service members (DoD 2007). However, past research suggests that few individuals (the range of reported incidences is from 15%–25%) report sexual assault to authorities (Clay-Warner and Burt 2005; Firestone and Harris 2003; 2008; Harned et al. 2002). Past research also indicated that while both men and women can experience sexual assault, the risk of workplace assault may be higher for women, especially those in male-dominated occupations (Dekker and Barling 1998; Frank et al. 1998; Haavio-Mannila et al. 1998; Sadler et al. 2003). This is reinforced by data from a recent DoD survey. A single-item measure of unwanted sexual contact (asking whether someone, without their consent or against their will, sexually touched them, had (attempted or completed) sexual intercourse with them, oral sex with them, anal sex with them, or penetrated them with a finger or object), as reported in the 2006 Gender Relations Survey of Active Duty Members report, indicated that 6.8% of women and 1.8% of men indicated experiencing unwanted sexual contact. Women in the Army were more likely than women in the other Services to indicate experiencing unwanted sexual contact, whereas women in the Air Force were less likely. Also, junior enlisted members were more likely than senior members, junior officers, and senior officers to indicate unwanted sexual contact (Lipari et al. 2008: iv).

Research Methods

Data Sources—2002 Sexual Harassment Survey

A sample of respondents from the “Armed Forces 2002 Sexual Harassment Survey” generated data for the Office of the Secretary of Defense by the Defense Manpower Data Center. This was a “worldwide scientific survey of how men and women work together in the Active-duty Military Services.” The stated purpose of the survey was “to assess the prevalence of sexual harassment and other unprofessional, gender-related behaviors”.

A single-stage, stratified random sample of 60,415 respondents was drawn for the survey by mail and the Web, representing male and female enlisted personnel and officers in the Army, Navy, Marines, Air Force and Coast Guard. A total of 19,960 usable surveys were returned for a response rate of 36% and includes 10,235 males and 9,725 females, illustrating the oversampling of women. The sampling frame was stratified by service branch, sex, paygrade, race/ethnicity, likelihood of deployment and geographic location.

In this analysis, cross-tabulation is used to assess the extent to which men and women report various types of behaviors that might be construed as harassment. Logistic regression is used to test the impact of different forms of sexual harassment on the likelihood of reporting sexual assault. In addition we investigate whether men or women, different race and ethnic groups, and/or different ranks are more or less likely to label an event as sexual harassment. Results should support past research indicating that the presence or absence of environmental sexual harassment is highly predictive of both individual harassment and sexual assault.

Variable Construction 2002 Sexual Harassment Survey

Among the items in the “Gender Related Experiences in the Military in the Past 12 Months” section of the survey, respondents were asked about sex/gender related talk and/or behavior that was unwanted, uninvited, and non-consensual. Respondents were provided a list of 19 items and asked whether that item had occurred “very often,” “often,” “sometimes,” “once or twice,” or “never.” We recoded the first four responses in an “ever” occurred category with a value of 1. “Never” was coded 0. Based on the original statements, we identified individualistic forms of sexual harassment that are personal, directly physical in nature, and leave little room for misinterpretation by either the victim or the perpetrator (sexual assault, touching, sexual phone calls). This form can be differentiated from a broader category of more public, environmental harassment (jokes, whistles, suggestive looks). The latter actions can be experienced even if directed at another individual and are ambiguous enough to leave their interpretation dependent on the environmental context. Respondents were initially classified as having experienced individualistic or environmental unwanted, uninvited sexual behavior, or any form (individualistic, environmental, or both).

Respondents were then asked whether they considered “ANY of the behaviors…which YOU MARKED AS HAPPENING TO YOU … to have been sexual harassment [emphases part of original survey]”. Responses included “none were sexual harassment,” some were sexual harassment; some were not sexual harassment,” and “all were sexual harassment.” This variable was dichotomized to indicate whether “any” events were labeled as sexual harassment, or none were labeled as harassment. Another question asked “Did you report this situation to any of the following installation/Service/DoD individuals or organizations.” The responses included references to the various official channels for reporting. Individuals who responded “yes” to any of the categories were classified as having used official channels to report the incident. Independent variables utilized include sex of respondent, rank (junior enlisted, senior enlisted, junior officer, senior officer), whether respondent was married, and service branch.

With response categories options of “very often,” “often,” “sometimes,” “once or twice,” or “never”, sexism was operationalized by the following four questions:

  • How frequently have you heard people of your gender referred to in negative or insulting terms?

  • How frequently were you treated you “differently” because of your gender (for example, mistreated, slighted or ignored you)?

  • How frequently did you hear offensive sexist remarks (for example, suggesting that people of your gender are not suited for the kind of work you do).

  • How frequently did someone put you down or was condescending to you because of your gender?

Variable Construction 2006 Gender Relations Survey

Sexist behavior involves unwanted actions that refer to an individual’s sex and are directed toward all persons of that sex. Experiences of sexist behavior include verbal and/or nonverbal behaviors that convey insulting, offensive, or condescending attitudes based on the sex of the respondent. To be included in the calculation of the sexist behavior rate, members must have experienced at least one of the four behaviorally stated items defining sexist behavior. Sex discrimination is unfair or unequal access to professional development resources and opportunities due to a Service member’s gender. A new baseline measure of sex discrimination was introduced in 2006 where members were asked if they had experienced, within the 12 months preceding the survey, any discriminatory behaviors related to evaluations, career development, or assignments where their gender was factor and whether they considered at least one of the behaviors to be sex discrimination.

Sexual harassment is comprised of three component measures (each measured by four of the 12 items in Question 35 that measures sexual harassment): crude/offensive behavior (verbal/nonverbal behaviors of a sexual nature that were offensive or embarrassing), unwanted sexual attention (attempts to establish a sexual relationship), and sexual coercion (classic quid pro quo instances of specific treatment or favoritism conditioned on sexual cooperation).

To insure valid comparisons for the analysis, Respondents were then provided a list of 19 items and asked whether that item had occurred “very often,” “often,” “sometimes,” “once or twice,” or “never.” We recoded the first four responses in an “ever” occurred category with a value of 1. “Never” was coded 0. Based on the original statements, we identified individualistic forms of sexual harassment that are personal and frequently directly physical in nature, and leave little room for misinterpretation by either the victim or the perpetrator (sexual assault, touching, sexual phone calls). This form can be differentiated from a broader category of more public, environmental harassment (jokes, whistles, suggestive looks). The latter actions can be experienced even if directed at another individual, and are ambiguous enough to leave their interpretation dependent on the environmental context. Respondents were initially classified as having experienced individualistic or environmental unwanted, uninvited sexual behavior, or any form, (individualistic, environmental, or both). We focus on the separate categories of environmental and individual harassment for this research.

In this survey, unwanted sexual contact includes rape, non-consensual sodomy (oral or anal sex), or indecent assault (unwanted, inappropriate sexual contact or fondling) and can occur regardless of gender, age, or spousal relationship. Incident rates of unwanted sexual contact used two measures, 1) A two-item measure based on the Sexual Experiences Questionnaire (SEQ), allowing results to be compared to the 1995 and 2002 results, and 2) A new baseline measure designed for the WGRA2006 to be consistent with the definition in the amended Article 120 of the Uniform Code of Military Justice (UCMJ) effective in October 2007 (See Lipari et al. 2008: iv).

Analysis

Figure 1 presents the basic information on the extent to which respondents reported sexual harassment behaviors in 2002 and 2006. More than half of the females identified at least one unwanted, uninvited, did not participate willing type of incident in both surveys and the percent increased from 52.55% in 2002 to 56.25% in 2006. The percent of women experiencing harassing behaviors was more than twice that of men in 2002 and nearly twice that of men in 2006. This comparative change is due to the fact that the increase in reported harassment was greater for the men (increase by 5.99%) than for the women (increase by 3.69%).

Fig. 1
figure 1

Percent reporting some form of sexual harassment, 2002 and 2006

Full size image

Figures 2, 3, 4, 5 show percentages reporting experienced harassment or assault. It is noteworthy that when percentages are translated into numbers, the data suggest that 475,913 members of the active duty services reported one or more harassing incidents (34.27%). Of these 360,946 are estimated for males and 114,967 are estimated for females. Similar patterns of results are displayed for environmental and individual harassment. The results for sexual assault are striking. With 2.17% of the males reporting attempted or actual sexual assault, this translates into 25,702 incidents. Over 5% of women (5.02%) reported attempted or actual assault, reflecting 10,185 incidents. The percentages reporting experiencing harassment or assault are displayed visually in Figs. 2 through 5.

Fig. 2
figure 2

Percent reporting sexual harassment behaviors, 2006

Full size image
Fig. 3
figure 3

Percent reporting environmental harassment 2006

Full size image
Fig. 4
figure 4

Percent reporting invidual harassment 2006

Full size image
Fig. 5
figure 5

Percent reporting sexual assault, 2006

Full size image

In an attempt to show the impact of environmental harassment on individualized experiences, Table 1 focuses on attempted or actual sexual assault for men and women by whether or not environmental harassment is reported. Results are provided both for 2006 and 2002. The most striking observation is that assault reports are very rare when no environmental harassment is claimed and much more prevalent when environmental harassment is reported. In a proportional sense, the impact is greater for the men than the women. With well under 1% of men reporting assault but no environmental harassment, but nearly 9% when environmental harassment is reported, the odds of assault are increased by nearly 35 times.

Table 1 Assault or attempted assault by environmental harassment
Full size table

(34.53). There is also a substantial increase in the odds of assault for women when environmental harassment is present, nearly twelve times higher (11.84). The same pattern of increased odds of assault is evident for 2002, even though the reported incidents were lower.

Figure 6 provides a clear display of the differing experiences of men and women and the impact of environmental harassment.

Fig. 6
figure 6

Sexual assault by reported environmental harassment, 2006

Full size image

Table 2 breaks out the data on sexual assault by sex and race/ethnicity of the respondents. As shown in the top portion of the table, of the males African American respondents have the highest percent reporting assault at 3.36%. Next are the “others” (2.72%), followed by the Hispanics (2.44%), and white non-Hispanic respondents have the lowest incidence (1.77%). Females have higher rates than males for all race and ethnic categories, but Hispanic women have the highest rate (6.45%) followed by Black women (5.23%), then others (4.76%) and finally white non-Hispanic women (4.52%).

Table 2 Sexual assault by race, ethnicity and sex and by whether or not respondents reported environmental harassment
Full size table

The middle and lower portions of Table 4 add a control for whether or not the respondents reported any environmental harassment. The differences are striking. For both males and females, when no environmental harassment is reported the percent reporting sexual assault is under 1% for all race and ethnic groups. The percentages are substantially higher for all categories when environmental harassment is reported. Among the males, nearly 15% (14.63%) of the African American respondents report attempted or actual assault. This is followed by others (9.34%), then Hispanics (8.83%) and finally white, non-Hispanics (7.65%).

Though the percentages reporting assault are higher in all categories for the women than the men, the link to environmental harassment is very clear. Over 12% (12.20%) of Hispanic women report sexual assault if they also reported environmental harassment, compared to 0.66% reporting assault if they did not experience environmental harassment. Nearly 11% (10.95%) of Black women reported assault if they also reported environmental harassment, followed by 8.99% for others and 8.63% for whites.

Table 3 provides the results of logistic regression models designed to predict the probability of reporting attempted or actual sexual assault. The first model is for males, the second for females and the third for the total sample. In all three models the dominant variables increasing the likelihood of assault are individual harassment, followed by sexist behavior and then environmental harassment. Controlling for these three factors three other variables are statistically significant for the males. Officers are about half as likely, junior enlisted men are more than twice as likely and Black males are nearly twice as likely to report sexual assault. The non-significant variables are also of interest. There are no meaningful differences by branch, deployment status or being stationed outside the U.S.

Table 3 Logistic regression model’s predicting the probability of reporting attempted or actual sexual assault 2006
Full size table

Again controlling for the three major predictors, two other variables are statistically significant for the females. Being a junior enlisted member increases the odds of reporting assault by more than three times, though note that being an officer is not statistically significant in reducing the likelihood of assault. Being deployed outside of the U.S. more than doubles the likelihood reporting assault. Again, branch and deployment status display no statistically meaningful differences and Black women are not significantly different from others in this model.

The results for the total sample roughly parallel those for the separate analyses, with one major surprise. When the other variables are controlled the coefficient for “Female” is negative (−0.20) and statistically significant at the 0.04 level. This suggests that if sexist context, environmental harassment and, consequently, individualized harassment did not occur women would actually have a lower probability of reporting sexual assault than men in the armed services.

The roles of individual and environmental harassment in the logistic regression analyses are interesting. Our conceptual model (Fig. 7) suggests that environmental harassment along with sexist behavior create a context in which individual harassment is viewed as acceptable by potential perpetrators, and this context in turn increases the likelihood of sexual assault.

Fig. 7
figure 7

Explanatory model to predict attempted and/or actual sexual assault

Full size image

Figure 8 captures the linkage between individual and environmental harassment in a clear and powerful display. When no environmental harassment is reported, individual harassment is very rarely reported. For males, of those reporting no environmental 89.11% also report an absence of individualized harassment. For females, of those reporting no environmental 81.61% also report an absence of individualized harassment. When environment is reported the probability that there will also be individualized harassment is extremely high: 98.09% of the time for males and 99.09% of the time for females. The reporting of individualized harassment has by far the greatest increase in the probability of sexual assault.

Fig. 8
figure 8

Individual by environmental harassment

Full size image

The results in Fig. 7 are also displayed in Table 4, capturing the statistical strength of this relationship with gamma. Gamma is an interesting statistic that can attain a value of 1.0 in circumstances where a relationship is “conditionally perfect.” This means that if one condition is present one could perfectly predict the presence of another condition. This is very close to the case here, with gammas of 0.995 for males and 0.996 for females. When environmental harassment is reported, individualized harassment also is almost always reported.

Table 4 Environmental harrassment reporting by gender
Full size table

Discussion

The above analyses support past research indicating that sexism, sexual harassment and sexual assault in the workplace are overlapping, yet distinct concepts. Findings suggest that the context in which men and women perform work duties is a key factor in whether or not individuals reported sexual harassment or rape. For large social institutions such as the various branches of the military and criminal justice system, survey efforts that provide reporting opportunities are valuable in multiple respects. Beyond providing victims a collective voice, there is a strong correlation between increased reports of sexual misconduct, regardless of form, and reported rape and attempted rape. Ostensibly, sexual assault is more likely to occur in environments where sexism and sexual harassment are more prevalent—thus, environmental harassment manifests in individual harassment and, in extreme cases, sexual assault. The anticipatory model illustrated above (Fig. 8) captures this escalation trajectory and suggests that attention to environment can better prevention. Gender ratio, degree of supervision, minor form tolerance, and nature of job duties are all variable elements that, when conceptualized as risk factors, are subject to policy manipulation and daily practice discretion.

Results highlight how attempting to remedy the problem of harassment by focusing on changing individual behaviors, rather than on altering a militaristic culture in which sexism may still be unofficially condoned and institutionally supported are unlikely to succeed. Though necessary for punishment and individual deterrence, it is uncertain whether punishment examples yield any general deterrence effect. It seems likely that an organizational context in which environmental harassment may still be unofficially condoned and institutionally supported as a process for excluding women (and men considered “outsiders,” e.g. race/ethnic minorities, sexual minorities) from becoming part of an organization which values cohesion and espirit d’corp, sends a message of permissiveness to those individuals inclined to engage in the more egregious individualized forms of harassment and sexual assault.

There are multiple research queries still to be addressed necessary for a thorough prevention approach. Regarding deterrence, the question of whether discretionary leniency by “sentencing” authorities, such as ranking officers in the armed services and police chiefs and sheriffs in law enforcement, communicates tolerance and amnesty has not been empirically addressed. This issue is particularly important in administrative sanctioning decisions for non-criminal policy violations as leniency here may well suggest unimportance assigned by supervisors. In that minor sexual misconduct forms are more prevalent, readily addressable through administrative measures, and vital in terms of victimization severity escalation, lower tolerance policies may affect deterrence.

Another largely unexamined issue is the effectiveness and best practices nature of awareness and prevention training. It is not known, for example, the extent that sexual misconduct prevention training is represented across different levels of law enforcement training academy curricula and, related, whether continuing education is addressing workplace sexual misconduct.