Introduction

Sexual harassment (SH) is a common experience in the workplace, although estimates vary widely, ranging from 40% to 75% of women and from 13% to 31% among men [1]. Earlier research has defined the contours and estimated the incidence of SH [2, 3], while a second generation of studies has focused on the circumstances and sequela of SH [4–6]. At the intersection between legal and psychological constructs [7, 8], the definition of SH is still a matter of debate, despite conceptual and psychometric advances in the 1990s. Fitzgerald’s Sexual Experience Questionnaire distinguishes three dimensions: gender harassment, defined as “behaviors that convey hostile, offensive, and misogynist attitudes” [3], unwanted sexual attention defined as “incidents such as sexual imposition, touching, or repeated requests for dates” [9], and sexual coercion [3, 10, 11]. These domains have been refined overtime [12], but they still capture a continuum of severity and distinguish between legal entities of quid pro quo (sexual coercion) and hostile environment (gender harassment and unwanted sexual attention).

Research on the causes of SH indicates that it is prevalent across ages, socioeconomic groups, and cultures [13], suggestive of a “universal” phenomenon [14]. Fitzgerald and Drasgow’s conceptual model theorizing the causes and consequences of SH draws attention to the situational characteristics increasing SH, including the workplace climate and gender structure [3, 15]. Supporting this framework, a meta-analysis of 41 studies indicates a strong effect of organizational climate and job gender context on SH victimization [4]. Tolerance for SH and lack of sanctions are predisposing factors [16], while imbalanced sex ratio [4] and hegemonic masculinity norms valuing toughness and aggression as the “epitome of masculinity” [17, 18] work in tandem to increase SH [19]. These contextual elements are generally considered as individual perceptions rather than group-level measures, resulting in over-estimation of associations [15]. A recent study applied group level measures to show that sexist climate in US military units increased individual risks of SH [20]. The study however, did not consider gender differences in these contextual effects. Few studies in fact, consider a Person-X-Situation model interacting the social context with individual characteristics. Pryor et al. applied such a model showing that the environmental climate had different effects for males and females in the US military [21, 22].

The prominence of SH in the military has received much attention in the US [23, 24] in the wake of the feminization of the profession and higher SH prevalence compared to civilian populations [2, 6]. In the military context, SH is often assessed as a component of military sexual trauma (MST), which represents a continuum of harm from sexism to sexual assault (SA) [24, 25].The 2018 US estimates among active duty military personnel indicated that 24.2% of servicewomen and 6.3% of servicemen had experienced SH (excluding assault) in the past 12 months while 6.2% of women and 0.7% of men reported SA [26]. Cultural norms valuing hegemonic masculinity likely contribute to higher SH in military populations [19], while maculinized environments that tolerate SH behavior and lower sociocultural power contribute to heightened risk of SH within military populations [27]. Few studies have explored SH in other military settings. One Swedish study indicated that 1.9% of female cadets reported sexual quid-pro-quo [28] and 83% reported gender harassment and unwanted sexual attention in a 24 months span. A South Korean study reported a prevalence of SH of 5.7% among female military personnel [29].

The present study aims to extend our current knowledge of MST in military populations outside of the United States, by 1) evaluating the incidence of a spectrum of experiences from sexist remarks, unwanted verbal sexual attention, sexual coercion, to SA among women and men in the military and by 2) exploring individual and contextual factors that increase the risk of experiences these events according to gender.

Methods

Results

The sociodemographic and sexual health characteristics of the study population are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of French active duty servicemen and women included in the COSEMIL survey (n = 1,500).

https://doi.org/10.1371/journal.pone.0259182.t001

Altogether 14.6% of men and 30.3% of women reported repeated sexual comments that put them ill at ease in the workplace in the last 12 months (Table 2). Likewise, repeated unwanted verbal attention was more frequent among women (8.4%) than men (0.9%). Sexual coercion was less common, described by ten women (4.9%) and 4 men (0.4%). The gender gap was also evident in the case of SA, mostly in the form of unwanted sexual contact, reported by 12.6% women and 3.5% men. Four women (1.7%) and six men (0.5%) reported attempted or forced sex in the context of their work in the last 12 months. Altogether 36.7% of women and 17.5% of men experience MST, including any form of sexual stressor in the last 12 months.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Measures and prevalence of Sexual harassment, receipt of sexual comments, sexual coercion, unwanted sexual attention and assault in the military work environment in the last 12 months by gender, n (%).

https://doi.org/10.1371/journal.pone.0259182.t002

Thirty four percent of women and 16% of men who reported MST, experienced several forms of sexual stressors, mostly as a combination of repeated sexual comments and sexual assault (13% of women and 6% of men) or repeated comments and repeated unwanted verbal attention (8% of women and 3% of men) (Fig 1A and 1B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

a, b: Distribution of type of sexual stressors experiences among service men and service women who report any MST event.

https://doi.org/10.1371/journal.pone.0259182.g001

Frequency of sexual comments varied from 4.7% to 21.9%, across units while SA varied from 1.9% to 12.1%. Factors associated with these experiences varied by type of stressor and sex (Table 3). Younger, non-cohabitating and childless men were more likely to have experienced SA while older men and men who had sex with men were more likely to report repeated sexist comments contributing to greater risk of verbal SH. Younger women, on the other hand, were more likely to indicate instances of unwanted verbal sexual attention and SA while non-cohabitating women were more likely to be exposed to repeated sexual comments. Men from socially disadvantaged backgrounds, including enlisted servicemen and participants born overseas or abroad, were more likely to report SA. Socio-economic differences were more complex among women as the most socially disadvantaged (lower education and enlisted personnel) were more likely to experience SA while the most socially advantaged (higher education and officers) were more likely to report sexual coercion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Percentage of service members who report a repeated sexual comments, repeated unwanted verbal sexual attention, sexual coercion, or sexual assault in the military work environment in the last 12 months by gender and sociodemographic characteristics.

https://doi.org/10.1371/journal.pone.0259182.t003

Results also indicate a significant effect of the work environment. Men working in the Navy were more likely to report repeated SA while women in the Army were more likely to report repeat sexism contributing to an increased verbal SH. In addition, women who had been deployed in the last 12 months were more likely to report repeated unwanted verbal attention and sexual coercion.

Because of differential associations between sexual stressors and individual and work environement factors, many factors that were related to specific sexual stressors were not longer associated with MST. For men, only sexual orientation and military branch were associated with MST, while for women, less time serving in the military (less than 5 years), lower level of education and military rank were related to MST.

Multivariate analyses indicated greater likelihood of experiencing verbal SH among servicemen who have sex with men and among women officers. The odds of SA were elevated among younger men and reduced among more educated women. Finally, the odds of MST were increased among men who had same sex partners and female officers but lower among more educated women.

For men and women alike, the work environment was significantly related to their probability of experiencing verbal SH, SA as well as MST. Women in the Air Force were less likely than women in the Army to experience verbal SH while men in the Navy were more likely to have experienced SA (this variable was not included in the final model due to model instability). The odds of verbal SH were 80% and 20% higher among men who worked in units where sexual comments and sexual oppression (unwanted attention, coercion and assault) were more prevalent (above the median), and were 50% and 40% higher when men worked in units with low acceptance of increased feminization of the army and higher female representation. Likewise, the odds of SA were elevated among men working in units with higher female representation. Altogether, the odds of men experiencing MST were elevated in units with higher prevalence of sexual comments, and lower acceptance of female representation. For women, the odds of SA were particularly high (OR = 5.7) among women working in units with a higher prevalence of sexual oppression, while at the same time these women had reduced odds of verbal SH. Likewise, the odds of SA were lower when women worked in units with higher prevalence of sexism and low acceptance of military feminization. Altogether, the odds of experiencing any sexual stressor (MST) were lower among women working in units with low female representation and low acceptance of female representation (Table 4A and 4B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4.

a: Multivariate analysis of Verbal SH, Sexual Assault, and MST in the military work environment in the last 12 months among men. b: Multivariate analysis of Verbal SH, Sexual Assault Sexual Oppression and MST in the military work environment in the last 12 months among women.

https://doi.org/10.1371/journal.pone.0259182.t004

Discussion

In line with previous studies conducted in the US military [23, 24], we found that MST is a common experience for women in the French military. While patterns of sexual stressors were comparable between men and women, the gender gap was profound, with women experiencing multiple forms as well as the most severe forms of sexual stressors. Such results display the realities of MST in France, although the nature, meaning and consequences of these experiences are likely different for men and women [6, 12, 34].

While published population-based estimates are unavailable in France, our 12 months prevalence of SA (12.6% among women and 3.5% among men) appears to be higher than in the French civilian population, based on the 2016 Virage study reporting 12 month prevalence rates ranging from 2.9% to 5.5% among women 20–49 years and from 1.16% to 1.5% among men of the same age groups (Virage) [29]. Comparisons with other military populations suggest higher MST incidence in the French military (men and women alike), but similar patterns of sexual stressors [26–34]. For both sexes, our MST prevalence rates were close to 10 percentage points higher than US estimates, and far exceeded the 5.7% prevalence reported among Korean military women [29]. Conversely, our 12 months MST prevalence for women (36.7%) was lower than the 84% reported among women cadets and officers in the Swedish army, although this estimate was based on a 24-month time frame [28]. SA was also more prevalent in our study compared to the 2016–2018 US estimates for servicemen (0.6%-0.7%) but closely aligned with the experiences of US servicewomen (4.3%- 6.2%) [26]. Less than 2% of women in our study reported attempted or forced penetrative sex, versus 3% among US military women and 1.9% among Swedish army servicewomen (albeit over 24 months).

Differences in MST estimates across studies, likely reflect variations in definitions and measures [2, 8, 24], survey procedures [2, 8] and social context [28]. The COSEMIL MST metric explored several of the SEQ DoD dimensions but included fewer and modified items to adapt to the cultural context as suggested by Wasti and colleagues [35] and to allow comparisons with the national Virage estimates [32]. In particular, our measure of unwanted sexual attention only included verbal encounters, as we categorized all physical forms of unwanted sexual contact in a seperate category of SA, to explore the continuum of harm. In addition, unlike the revised SEQ DoD [36], the COSEMIL study did not evaluate the perceived severity of hostile environment events, which may have contributed to an overestimation of MST, although we only counted repeated events as MST. Comparability of study results may also be compromised by sample selection. Unlike most prior research which suffers high non-response rates [24], the COSEMIL study used a probability sampling design across military branches and ensured high response rate to reduce selection bias [30]. Beyond measurement and study procedures, differences in MST estimates across studies may reflect differences in social context and workplace environment [4]. Substantial variation in SH is observed across the 28 EU member States, as evidenced by Latcheva, showing higher SH prevalence in France and Northern Europe than other regions [37]. Such cross-cultural comparisons are interesting to consider within the military context, which shares a number of common features, including organizational structure and masculinist culture [19, 24] but are limited in scope given the paucity of research conducted outside of the United States. Using the SEQ DoD construct, a Swedish study substantiates the relevance of the SEQ DoD’s three dimensional SH construct in a European context but assesses MST over a 2 year period, limiting the potential for comparison [28].

Beyond overall prevalence rates, this study provides an opportunity to explore SH as a continuum of harm [24, 25] distinguishing experiences of verbal SH from SA. While we show significant overlap between these events, we also find differences in their predictors leading to null findings when examining the aggregated measure of MST as associations related to verbal SH and SA cancelled each other out.

The opportunity to explore MST in th French military context adds to existing research mostly conducted in the US, providing an opportunity to examine the extent to which predictors of SH are similar or different across military contexts. Consistent with a number of studies [4], we found that the workplace environment was the strongest predictor of verbal SH and SA in our military population [27], although these effects differed by gender and by sexual stressor. As previously described [38], deployment seemed to increase the risk of SA among women, although the association was no longer significant after multivariate adjustments. Supporting the Person Situation model, job gender context and group level exposures had different effects for men and women [4], and we also found these effects to differ by sexual stressor. While sexist and violent environments were related to individual experience of sexual stressors for men, low acceptance of female representation was associated with reduced MST among women, which was unexpected. These differential effects by sexual stressors and gender have not been reported in previous literature, although the small female sample size limits the interpretation of these findings. Willness’ systematic review indicates smaller job gender context effects in the military than in other workplace environments, which could be explained by differential effects by gender, or by different definitions of SH, combining different sexual stressors [4]. Other conditions of the workplace including social cohesion or unequal gender norms were not statistically associated with individual experiences of sexual stressors in our study.

Beyond situational factors, we also identified a number of individual and interpersonal characteristics associated with verbal SH and/or SA. Specifically, less educated women were more likely to report instances of SA, as was also the case for men born outside of mainland France (although non significant in multivariate analysis). These results corroborate Bell’s observation of sociocultural power conditions heightening the risk of sexual victimization in military populations [27]. On the contrary, women officers were more likely to report instances of sexual coercion and repeated unwanted verbal sexual attention. The greater exposure to SH among women who do not conform to the gender order has been noted in prior studies in the general population, including the FRA study among women across 28 EU states which indicates that women in the highest occupational groups and with higher levels of education are more likely to experience SH than others [37]. In line with the sanctioning of gender non-conformity, we also found that men who had sex with men were more likely to report verbal SH, which potentially echoes the nature of sexual comments, shown to be commonly homophobic in a US military study [39]. The increased risk of overall MST among men with male sexual partners is consistent with prior studies reporting increased risk of MST among LGBT individuals in the US military [40–42].

In addition to the SH measurement concerns discussed above, this study has a number of limitations, including the small female sample size and small sample size of LGBT individuals, resulting in large confidence intervals. Lack of information about acceptance of and retaliation against SH in the workplace, shown to be strongly predictive of MST [16, 23] also limits the interpretation of the environmental context condusive of MST. In addition, the cross-sectional nature of the study limits the understanding of the continuum of harm and prevents any causal interpretation although our conclusions are mostly consistent with findings of prospective studies.

Despite these limitations, the use of measures of MST along the continuum of harm, the use of a probability sample of the military population, and the systematic gender exploration of predictors of different dimensions of MST all constitute important strengths of this analysis. Given the public health implications of SH, especially in military populations, where servicewomen have been shown to suffer greater mental health sequealea following SH than in civilian populations, we suggest that the high prevalence of SH in the French military calls for programmatic action to prevent MST as well as screening programs and services for those who have experienced MST in the French military. The implementation of such programs have been effective in the US military [43], but have not been evaluated in the European context, including France, where a ministry of defense program (Themis program) was created in 2014 and revised in 2018. Such efforts not only address a human’s rights imperative but also have positive public health implications in reducing the health sequelae related to these events.

Acknowledgments

The authors would like to thank all the team who collected data on the field and contributed to this work. We would also like to thank all participants for their contribution to this study. We have no conflict of interests to declare.