Abstract
Sexual harassment (SH) in academia is a pervasive public health issue, underscoring the need for effective prevention initiatives. Bystander intervention (BI) models are increasingly used in workplace settings; however, one gap in the existing literature is understanding BI from the perspective of the target of harassment. The purpose of this study is to explore faculty and staff experiences with SH, whether bystanders were present, and the nature and impact of their actions. Data for this cross-sectional study were collected from 5,227 faculty and staff at a university in the Mid-Atlantic region of the United States. Of those reporting experiencing SH in the past year (n = 395), more than half (56.4%) indicated a witness was present. A significantly greater percentage of participants who experienced SH committed by a person with supervisory or influential impact over them reported that a witness was present (62.0%), compared to those whose harassment was committed by someone without such a role (50.2%). In cases where a witness was present, less than one-quarter of participants (21.4%) reported that the bystander took action to intervene in the situation; bystanders were more likely to take intervention action in cases where the SH was committed by someone who did not have supervisory or influential impact over the target. Just over half of the participants (53.4%) reporting bystanders took action indicated the intervention was helpful. Bystander action was positively linked to target disclosure. These findings can help guide the adaptation of initiatives that use this model for SH prevention in academic settings.
Introduction
The United Nations defines sexual harassment (SH) as “any unwelcome sexual advance, request for a sexual favor, verbal or physical conduct or gesture of a sexual nature, or any other sexual behavior that creates an intimidating, hostile, or offensive work environment, is made a condition of employment, or interferes with work” (United Nations, 2008, p. 1). Academia is identified as a workplace setting that is specifically primed for SH to occur due to its hierarchical structure, power imbalances, values around individualism and competition, presence of stereotypical masculinity, unclear SH policies, network protections, and lack of transparency in hiring and promotion (National Academies of Sciences, Engineering, and Medicine [NASEM], 2018; Täuber et al., 2022; Zara et al., 2024). A meta-analysis on workplace harassment found that 58% of women in academia experienced SH, a higher rate than other work settings, including the private sector or non-military government organizations (Ilies et al., 2003), and a recent systematic review concluded that academic settings have a high prevalence of SH (Zara et al., 2024). The occurrence of SH has been well-documented in science, technology, engineering, and medicine (STEM) disciplines as well as in non-STEM and humanities academic fields (Karami et al., 2020). In 2018, the NASEM released a report calling attention to the insidious nature of SH in higher education, revealing that 20% to 50% of students and over 50% of women faculty and staff experience SH victimization. Using foundational work by Fitzgerald et al. (1998), the NASEM (2018) report identified a spectrum of SH behavior that includes sexual coercion (e.g., expecting sexual activity or favors for professional or educational advances and opportunities), unwanted sexual attention (e.g., unwelcome sexual advances or requests), and gender harassment (e.g., verbal and nonverbal behaviors that demean or exclude one gender). Gender harassment is further characterized as including two types: sexist hostility (e.g., demeaning jokes or comments based on gender) and crude harassment (e.g., sexually crude and denigrating comments or terms based on gender).
SH is a major public health problem, and all of its forms negatively impact physical and psychological health and well-being, leading to anxiety, depression, post-traumatic stress disorder (see Bondestam et al., 2020; Thakur & Paul, 2017). In addition, SH is associated with poor occupational well-being, reduced job satisfaction, weakened organizational commitment, and poor job performance (Bondestam & Lundqvist, 2020; Chan et al., 2008; Langhout et al., 2005). The NASEM (2018) report emphasizes that gender harassment is the most experienced type of SH (as compared to sexual coercion and unwanted sexual attention) and may appear more covert, yet it has serious and negative impacts on the mental health and career trajectories of targets, supported by decades of research (e.g., Cortina et al., 1998; Raj et al., 2020).
Across studies and types of SH, gender remains a significant factor in all forms of SH, with women experiencing it at higher rates than men (Fitzgerald & Cortina, 2018). Importantly, binary notions of gender have been historically used in research in this area, though recent research accounts for diverse gender identities and reveals that those with minoritized gender identities, including nonbinary and trans-identified faculty and staff, are vulnerable to workplace incivility and SH (Jagsi et al., 2023; Krebs et al., 2023; Vargas et al., 2021). In addition, studies indicate that individuals with marginalized racial, ethnic, gender, and sexual identities experience more frequent harassment with poorer health and well-being outcomes and encounter greater negative implications for their careers compared to those without marginalized identities (Berdahl & Moore, 2006; Bondestam & Lundqvist, 2020; Buchanan & Fitzgerald, 2008; Jagsi et al., 2023; Konik & Cortina, 2008; Vargas et al., 2021). In addition, most research on SH in academic workplaces focuses on faculty experiences, neglecting the experience of staff. A small body of research, however, indicates that staff in certain positions—namely managerial, administrative, and professional personnel—are even more likely to experience workplace harassment than academic faculty (see review by Henning et al., 2017). The same review found that other staff, including support staff and service workers, also experience SH in their roles.
Given the widespread nature of SH in academia, there is an urgent need for prevention (NASEM, 2018). Because formal reporting of SH in academia remains low (Kirkner et al., 2022) other avenues outside of reporting are needed to address the issue. Bystanders, or third-party witnesses, provide a potentially powerful, yet often under tapped, resource to intervene to prevent and/or disrupt the harassing behavior (Banyard, 2015; Bond & Haynes-Baratz, 2022). Bystanders can intervene before, during, or after instances of harassment or harm and can take a range of actions, including direct approaches (e.g., confronting the transgressor), indirect methods (e.g., calling upon an authority or peer to help the situation), or delayed action (e.g., following up with the victim at a later time; Banyard, 2015; Bowes-Sperry & O’Leary-Kelly, 2005). The use of bystander intervention (BI) education as a prevention strategy to address campus sexual violence with high school and college students has grown steadily over the past two decades, with research demonstrating generally positive results including increased bystander efficacy, intentions to intervene, and prosocial bystander behavior (see Kettrey & Marx, 2019; Mujal et al., 2021; Park & Kim, 2023), as well decreases in victimization and perpetration among students (Coker et al., 2015, 2019), though research indicates that outcomes of BI are not always positive for victims (Hoxmeier & McMahon, 2021; McMahon, 2022; Moschella & Banyard, 2020). More recently, research on BI approaches has been applied to other settings beyond students in school contexts, including SH in academia (Bond & Haynes-Baratz, 2022; Collins et al., 2021; Elias-Lambert et al., 2022; Lee et al., 2019; Täuber et al., 2022). In fact, the National Academies has specifically called for the implementation of BI models to help prevent SH in academic settings (2018) and provided guidance on implementation (2023).
Research on BI related to SH in academia is limited but growing. A recent systematic review of SH prevention in academia found that bystander programs are promising, though they are typically focused on students and issues of sexual assault rather than SH among employees (Zara et al., 2024). Most of the work thus far has focused on student–faculty interactions (Heretick & Learn, 2020), examined the impact of observing incidents on the well-being from the perspective of the bystanders (Acquadro Maran et al., 2022), or evaluated the impact of a training program (Bond & Haynes-Baratz, 2021; Elias-Lambert et al., 2022). Less information is available about bystander action from the perspective of the actual target of SH, such as whether others were present, intervened, and whether the action was helpful. This is a critical gap in the field, as there is research suggesting a discrepancy between what bystanders and the targets of sexual violence perceive as helpful bystander action. For example, studies demonstrate that, although bystanders might believe that their intervention is beneficial in situations of sexual violence, victims sometimes find these actions unhelpful or even harmful, for example, making comments that are victim-blaming, intervening in ways that escalate violence or retaliation, or involving others in the situation when that was not wanted by the target (Hoxmeier & McMahon, 2021; McDonald et al., 2016; McMahon, 2022; Moschella & Banyard, 2020). As workplaces, including institutions of higher education, work to develop effective BI prevention education programs and health promotion initiatives to address SH, it is of paramount importance to incorporate the viewpoint of the targets and to better understand the landscape of bystander opportunities to intervene as well as the types of actions that were helpful or not. As such, this exploratory study focuses on data from those faculty and staff who experienced SH in academia, whether bystanders were present, whether they intervened, what they did and whether it was perceived as helpful or not.
BI and SH in academic settings
BI as a strategy for reducing sexual violence has most commonly been examined within college contexts to address sexual assault and to a lesser extent, dating violence (Banyard, 2015). BI related specifically to SH has been explored in the larger workplace literature, beyond just academia. For example, foundational work by Bowes-Sperry and O’Leary-Kelly (2005) discussed the potential role of “observer intervention” by bystanders in the workplace as an important avenue for addressing SH, given that most targets do not report their experiences formally within organizations. They helped explain the decisions that bystanders face when witnessing SH in the workplace are based on two dimensions: (a) the level of immediacy of intervention (e.g., is it a situation requiring immediate action as it unfolds, or can it be addressed later?) and (b) the level of involvement of the observer (which can range from choosing no involvement to becoming fully involved in the situation). Several researchers have used this framework to assess bystander decisions regarding intervention in the workplace. For example, McDonald et al. (2016) examined 74 descriptions of SH and found that there were often bystanders present, and that, most often, they engaged in low-immediacy, low-involvement interventions. They also found that observer actions were not always helpful and were quite limited in their effectiveness related to preventing continued harassment.
Other researchers have demonstrated the challenges to taking bystander action that are inherent in workplace settings, and different from student BI on college campuses. For example, research on BI in related fields of workplace microaggressions, incivility, bullying, and bias reveal barriers such as fear of retaliation, perceived threats to career advancement, and beliefs that organizations will not protect whistleblowers (Griffith et al., 2022; Haynes-Baratz et al., 2022; Täuber et al., 2022). Scholars in the field of workplace incivility have long emphasized the role of power in impacting observer’s decisions about intervention. For example, in a series of three studies that included both vignette and field data about actual workplace mistreatment situations, Hershcovis et al. (2017) found that those individuals with high levels of power were more likely to intervene than those with lower levels of power and were more likely to address the source of the behavior by confronting the transgressor, while less likely to provide support to the target or more likely to avoid the situation altogether. Those with less power or status in workplaces are also more likely to experience greater mistreatment and are more likely to experience social retaliation victimization (e.g., ostracization and blame) when they tell colleagues (Cortina & Magley, 2003; Fitzgerald & Cortina, 2018). These barriers to bystanders’ action in the workplace are relevant for academia, where power structures are often inherent (NASEM, 2018). Those with less power are subject to vulnerabilities unique to academia, such as needing support for tenure, opportunities for collaboration on grants and publications, mentoring, access to networks, letters of recommendation, and more (Bates et al., 2018; Binder et al., 2018; NASEM, 2018, 2023; Täuber et al., 2022).
The Current Study
The current study was designed to assess the perspectives of BI from faculty and staff who experienced SH, including whether witnesses were present at the time of the incident, and if so, whether the witness(es) did anything, and whether witness presence and action was associated with sociodemographic and incident-specific characteristics. Additionally, the study explored whether witness presence was associated with target’s disclosure of SH. In particular, we were interested in: (a) estimating the percentage of faculty and staff who experienced workplace SH in the past year and reported that at least one witness was present at the time of the incident, and if so, whether the witness(es) did anything, (b) identifying correlates of witness presence and action, which included sociodemographic and incident-specific characteristics (i.e., location of the incident, offender characteristics), and (c) exploring whether witness presence and action at the time of the SH incident was associated with disclosure of SH (both informal disclosure and formal reporting to the university).
Method
Participants and Procedures
Data for this study came from a larger, IRB-approved, campus climate survey collected in 2022 from multiple campuses at one university in the Mid-Atlantic region of the United States (of which no other papers have yet been published). The survey tool was based on the validated Administrator-Researcher Campus Climate Collaborative (ARC3) Campus Climate Survey Instrument, consistent with the recommendations of the NASEM’s (2018) report on the SH of women. The survey was further developed through an extensive design process involving the identification of validated items or scales from existing climate surveys, making necessary adaptations to reflect the university’s campuses, and developing new items where needed. All faculty, staff, postdoctoral associates, teaching assistants, and research assistants employed at the start of the 2022 academic year in the records of the University’s Office of Institutional Research were invited to participate in the Qualtrics-based survey via email. Student climate surveys are administered according to a separate cadence. The survey remained open for 6 weeks during the fall semester. Employees were reminded to take the survey through a range of outreach measures, including direct e-mails, social media posts, a dedicated webpage, listserv announcements, and flyers distributed and displayed on campus. Participants provided informed consent and were entered into raffles to receive several electronic gift card prizes. Access to the survey was provided through their employee portal, and the survey could not be shared nor taken multiple times. In total, 19.8% (n = 5,227) of the 26,400 invited responded to the survey invitation.
Measures
The measures below were part of the larger campus climate study and were asked in the order that they appear below.
Demographics
Participants were asked to identify their gender identity, age, and sexual orientation. Participants’ primary role at the university (e.g., faculty, staff, postdoctoral associate) was provided by the Office of Institutional Research during data matching to help shorten the length of the survey.
Age
Participants indicated their age by selecting from the following options: 23 or younger, 24 to 29, 30 to 39, 40 to 49, 50 to 59, 60, and older.
Gender
To assess gender identity, we used a two-part question (“How do you describe your gender?” followed by “Do you currently identify as transgender, genderqueer, or nonbinary?”). Any participants who indicated transgender, nonbinary, or genderqueer on either of the two questions were grouped with those who selected agender, genderfluid, or not listed/other. People who selected “No” to the second question and “woman” to the first question were classified as “non-TGQN women.” People who selected “No” to the second question and “man” to the first question were classified as “non-TGQN men.”
Sexual Orientation
Participants were asked, “How do you describe your sexual orientation? Please check all that apply” and were presented with following options: asexual, bisexual, gay, lesbian, pansexual, queer, straight/heterosexual, I describe myself in some other way and prefer not to answer. Participants who identified as asexual, bisexual, gay, lesbian, pansexual, queer were classified together as LGBTQA+.
Race/Ethnicity
Participants were asked a single question based on recommendations from the 2015 Census National Content Test-Race and Ethnicity Report (Jones & Bentley, 2017). Participants were asked, “How do you describe your race/ethnicity? Please check all that apply” and were presented with the following options: Asian/Asian American, Black/African American, Hispanic or Latinx, Middle Eastern or North African, Native American or Alaskan Native, Native Hawaiian or Pacific Islander, Multiracial, White/Caucasian, and a race/ethnicity not listed here. Participants who indicated more than one race/ethnicity, as well as those who selected “multiracial,” were classified as multiracial, due to cell sizes.
Primary Role
Data on participants’ primary role at the university were provided by the Office of Institutional Research and included the following categories: faculty, staff, postdoctoral associate, teaching assistant, and research assistant. For the purposes of this study, only participants categorized as faculty or staff were included.
SH Experiences
A modified version of the Sexual Experiences Questionnaire–Department of Defense (SEQ-DoD; Fitzgerald et al., 1999) asked participants how often they experienced 16 behaviors that capture four categories of SH (sexist gender harassment, crude gender harassment, unwanted sexual attention, and sexual coercion). Prior research has demonstrated that this measure has high reliability across its subscales for men and women samples, with reliability coefficients ranging between .78 and .94 (Stark et al., 2002), and the measure maintains its factor structure, convergent validity, discriminant validity, and good internal consistency across cultural contexts (Sánchez et al., 2022). Participants who experienced at least 1 of 16 sexually harassing behaviors were asked “Did any of the situations you just marked on the survey occur in the last 12 months?” Participants could respond “yes” or “no.” For the purposes of the current study, dichotomous variables indicating the presence or absence of one or more types of sexually harassing behaviors that occurred in the last year were constructed for SH overall and for each of the four categories (0 = no experience of past-year SH, 1 = experienced past-year SH). The Cronbach’s alpha for the SEQ-DoD was .82 for the total sample.
SH Incident Characteristics
Participants who experienced at least 1 of 16 sexually harassing behaviors were asked a series of follow-up questions about the one situation that had “the greatest impact” on them. This approach is commonly used in campus climate and workplace harassment research (Krebs et al., 2016) and was intended to focus respondents’ attention on the most significant incident they experienced, whether that impact was emotional, professional, psychological, or otherwise. Follow-up items asked about the gender of the person who committed the behavior, and whether the person who committed the behavior had any supervisory or influential impact (e.g., elevated social/academic status, decision-making power over resources or technology) over the participant. Last, participants who experienced at least 1 of 16 sexually harassing behaviors were asked, “to indicate whether the incident occurred in person or virtually” (e.g., online or through electronic communication). Those who selected “virtually” were then asked to describe the specific context, such as over text or messaging apps, phone or video calls, email, social media or direct messages, or via a conferencing platform (e.g., Zoom, WebEx).
Disclosure and Formal Reporting
Participants who reported experiencing at least 1 of 16 sexually harassing behaviors were asked to focus on the single incident that had “the greatest impact” on them. They were first asked, “Did you tell anyone about the situation you experienced (not counting this survey)?” Participants could respond “yes” or “no” (0 = did not disclose to anyone; 1 = did disclose to someone). Participants were then asked to respond “yes” or “no” to the item, “Did you formally report the incident to [name of University]?” (0 = did not formally report, 1 = formally reported). Formal reporting was defined as submitting a report to an official university office or entity (e.g., the Office of Employment Equity, the Title IX/ADA Compliance Office, Human Resources, campus police), and is distinct from informal disclosure to individuals such as colleagues, friends, or family members.
Presence of Witnesses for SH Experiences
Participants who experienced at least 1 of 16 sexually harassing behaviors were whether anyone else saw the situation that had “the greatest impact” on them. Participants could respond “yes” or “no.” Those who responded “yes” were asked to select how many people witnessed the incident and whether the person(s) who witnessed the incident did anything (yes or no).
Data Analysis
For the current study, 221 participants were removed from the sample for quality control. This included 218 participants who did not make it through the first section of the survey, the demographic items, and 3 participants who were removed for writing inappropriate, odd, or inconsistent responses for “other: please specify” for their gender item. Missingness was then assessed across all items and found that no more than 1.1% of individuals were missing data for any individual variable. Given the very low level of missing data, the absence of systematic patterns of missingness, and the focus on sociodemographic covariates that are not meaningfully recoverable through imputation, listwise deletion was deemed appropriate and unlikely to introduce bias or meaningfully reduce statistical power. Accordingly, participants with missing data on any of the included sociodemographic variables were deleted listwise resulting in a sample size of 4,242. Next, a subsample of participants who reported at least one experience of workplace SH in the last year (n = 459), and did not have missing data on any of the SH follow-up items including the outcome variable—whether or not a witness was present (n = 26)—was utilized for the current analysis (n = 433). Lastly, due to small cell sizes in this subsample, we removed Teaching Assistants (n = 13), Graduate Assistants (n = 13), and Postdoctoral Associates (n = 12) and focused on faculty and staff, for a final analytic sample of 395.
Additionally, the data used in the current study were adjusted using weights in a statistical raking procedure modeled after the Association of American Universities campus climate survey to help address the demographic representativeness of the sample (Cantor et al., 2020). The initial step of the weighting procedure was to create a base-weight for each participant. Because a census (the survey was sent to all eligible employees) was conducted, a base weight of one was assigned to each participant to reflect non-response. This adjustment consisted of a statistical raking procedure that adjusted the base weight to the demographic data available to the research team (Cantor et al., 2020; Deming & Stephan, 1940). The demographic variables used in the statistical raking procedure were based on employee’s institutional records which the Office of Institutional Research matched to participants’ responses and included the following: Sex (woman/man), age group (29 or younger, 30–39, 40–49, 50–59, 60, and over), race/ethnicity, and primary role at the university. As the weighting procedure used data provided by employee’s institutional records, the research team was not able to ask about gender in a more inclusive way, which was done on the survey instrument, for the raking procedure.
Following the statistical weighting procedure, descriptive statistics were used to examine the prevalence of witness presence during incidents, the characteristics of those incidents, and the types of harassment observed. Weighted cross-tabulations and chi-square tests were used to assess associations between witness presence, witness action, and a range of variables, including victim demographics (e.g., age, gender identity, race/ethnicity, sexual orientation, primary affiliation), characteristics of the person who committed the incident (e.g., gender, supervisory, or influential position), incident context (e.g., location, online vs. in-person), and type of harassment. Associations between bystander action and outcomes—including disclosure, formal reporting, job satisfaction, and perceived impact of the incident—were examined using weighted contingency tables and design-based F-tests derived from Rao-Scott adjusted chi-square statistics. Stata version 18 (StataCorp LLC) was used for all statistical analyses.
Results
Most of the sample, 395 participants, were over 30 years old (85.5%), a majority (64.1%) were White, and most of the sample identified as straight/heterosexual (78.2%) and women (78.7%). The sample was just over two-thirds staff (67.5%; see Table 1 for details and breakdown by faculty/staff).
Participant Descriptives
Presence of Bystanders
Figure 1 displays a diagram of bystander behavior for incidents of SH experienced by faculty or staff using a similar methodology as McMahon et al., 2023. Among those who experienced workplace SH in the past year, 56.4% (n = 227) reported that at least one other person was present (besides the person who committed the incident) to witness the incident (see Table 2). Rates were similar by role, with 58.7% (n = 83) of faculty who experienced SH and 55.1% (n = 144) of staff who experienced SH reporting the presence of a witness (see Table 3).
Diagram of Bystander Behavior in Faculty and Staff Sexual Harassment Incidents
Prevalence of Witness, by Full Sample Participants’ Sociodemographic Information
Note. Sociodemographic characteristics were examined for the full sample only; faculty–staff breakdowns were not presented because of small cell sizes in several categories. *p < .05
Prevalence of Witness for Incident Type by Role (Faculty and Staff) and Incident Type
Note. SH = sexual harassment. Percentages represent participants reporting at least one incident of workplace sexual harassment in the past year.
p < .05.
There were no statistically significant differences in witness presence based on the type of SH experienced. Among participants who experienced at least one incident of sexist gender harassment, 58.6% (n = 203) reported that a witness was present. This proportion was the same for both faculty (58.6%, n = 79) and staff (58.6%, n = 24). The presence of a witness was reported by 55.1% (n = 107) of participants who experienced at least one incident of crude gender harassment. Rates were similar across roles, with 63.2% of faculty (n = 37) and 51.5% of staff (n = 70) reporting a witness present. Among participants who experienced unwanted sexual attention, 50.5% (n = 34) reported that a witness was present. Specifically, 41.2% (n = 5) of faculty and 52.6% (n = 29) of staff who experienced at least one incident of unwanted sexual attention reported that a witness was present for these incidents. Finally, 44.6% (n = 4) of participants who experienced an incident of sexual coercion reported that a witness was present. Due to the small number of participants reporting sexual coercion in the past year, faculty and staff were combined for this analysis.
Among participants who experienced an incident of SH in-person (rather than virtually), 54.6% (n = 176) reported that a witness was present compared to 61.3% (n = 46) of participants who experienced virtual or online SH, though the difference was not statistically significant. When examining these patterns separately by role, staff who experienced in-person SH were significantly less likely to report the presence of a witness than staff who experienced SH virtually (51.3% vs. 70.2%, F(1, 261) = 5.22, p = .02). In contrast, for faculty, the percentage reporting a witness was present did not differ significantly by whether the incident occurred in person or virtually (60.8% vs. 48.2%, F(1, 128) = 1.18, p = .28).
A significantly greater percentage of participants who experienced past-year SH committed by a person who had a supervisory or influential impact over them reported that a witness was present compared to participants who experienced past-year SH committed by a person who did not have supervisory or influential impact over them (62.0% vs. 50.2%, F(1, 393) = 4.00, p < .05). Among staff, 56.0% (n = 81) of participants who experienced SH committed by someone with supervisory or influential impact reported that a witness was present, compared to 44.0% (n = 63) of those harassed by someone without such impact (F(1, 263) = 4.29, p = .04). In contrast, among faculty, rates of reported witness presence did not differ significantly by the status of the person who committed the incident: 61.6% (n = 51) of those who experienced SH committed by someone with supervisory or influential impact reported a witness present, compared to 38.4% (32) of those harassed by someone without such impact (F(1, 129) = 0.26, p = .61).
Bystander Action
Among participants who experienced past-year SH and reported that at least one witness was present, 21.4% (n = 48) indicated that the witness took some form of action (see Tables 4 and 5). Whether witness intervened varied by whether the harassment was committed by a person who had supervisory or influential impact over the victim. Specifically, 28.9% (n = 27) of witnesses took action when the person who committed the incident did not have supervisory or influential impact over the victim, compared to 16.1% (n = 21) when the person who committed the incident did hold such power (F(1, 392) = 4.20, p < .05). For faculty, 54.8% (n = 10) of witnesses took action when the person who committed the incident had supervisory or influential power over the victim, compared to 45.2% (n = 8) when the person who committed the incident had supervisory or influential impact over the victim, although the difference was not statistically significant. For staff, 62.6% (n = 19) of witnesses took action when the person who committed the incident did not have supervisory or influential impact over the victim compared to 37.4% (n = 11) when the person committing the incident did have supervisory or influential impact over the victim (F(1, 263) = 4.20, p < .03).
Bystander Action, by Participants’ Sociodemographic and Incident-Specific Characteristics
Note. SH = sexual harassment. Sociodemographic characteristics were examined for the full sample only; faculty–staff breakdowns were not presented because of small cell sizes in several categories.
p < .05.
Bystander Action by Role (Faculty and Staff) and Incident Type
Note. Percentages represent participants reporting at least one incident of workplace sexual harassment in the past year, where the witness took some form of bystander action.
p < .05.
Further analyses on bystander action produced interactions with cell sizes too small to separate faculty and staff responses, so they are reported collectively. Bystander action (vs. inaction) did not vary by victim’s sociodemographic characteristics, including age, gender identity, race/ethnicity, sexual orientation, and primary affiliation, or by any other incident-specific characteristics such as location of the incident, type of harassment, or gender of the person who committed the harassment. Of those who reported that the witness did something (n = 48), 53.4% (n = 26) reported that the action made things better, while 45.6% (n = 22) reported that the action had no effect at all. No respondents reported that the witness’ action made things worse. The outcome of the witness’ action did not vary by victim’s sociodemographic characteristic, including age, gender identity, race/ethnicity, sexual orientation, and primary affiliation, or by incident-specific characteristics such as location of the incident, type of harassment, supervisory role, and/or gender identity of the person who committed the harassment.
Disclosure and Reporting
Most faculty and staff who experienced an incident of SH in the past year told someone about the incident (63%); however, only about 5% of faculty and staff who experienced an incident of SH in the past year formally reported the incident to the university. Because the number of participants who formally reported a past-year incident of SH was insufficient to support separate analyses by employment group, faculty and staff participants were combined into a single category for all disclosure analyses.
There was a positive association between witness presence and disclosure (F(1, 388) = 4.38, p < .05). Among participants who experienced past-year SH (n = 395), those who indicated that a witness was present for the most impactful incident were more likely to disclose the experience to someone (68.8%, n = 152) compared to those who indicated that no one witnessed the incident (56.5%, n = 99). There were also significant differences in formal reporting by bystander action versus inaction (F(1, 389) = 5.09, p < .05). Participants who experienced past-year SH and indicated that a witness was present for the most impactful incident and the witness took action were more likely to formally report the incident to the university (15.9%, n = 8), as compared to those who indicated a witness was present and did nothing (5.3%, n = 9).
Discussion
Academic workplace settings were identified as “breeding grounds” for SH in a recent systematic review (Zara et al., 2024, p. 20) due to their power dynamics, lack of clear SH policies, and competitive nature. As BI models become increasingly utilized in academic workplace settings to address SH (Bond & Haynes-Baratz, 2022; Collins et al., 2021; Elias-Lambert et al., 2022; Lee et al., 2019), it is critical to understand the occurrence of witness presence during these incidents and their impact perceived by those who experience SH. Though most research on SH in academia focuses on faculty, it is important to include staff to better understand their experiences (Moylan et al., 2022). We found that over half of our sample of faculty and staff reported the presence of witnesses during the incidence of “most impactful” SH, suggesting that strategies to engage individuals as prosocial bystanders are an important, yet untapped resource, in academic settings. Although no participants reported that bystander action had a negative impact on the situation (i.e., “made things worse”), only slightly more than half of faculty or staff reported that the witness action had a positive impact on the situation (i.e., “made things better”), similarly suggesting that bystanders can, and do, play a critical role in mitigating harms associated with SH, but that training to support more effective intervention is warranted.
While many of the findings were consistent across faculty and staff samples, some noteworthy differences emerged. Regarding types of harassment, for crude gender harassment, faculty reported higher witness presence (63.2%) than staff (51.5%)—a meaningful difference even if not statistically significant. For unwanted sexual attention, the pattern was reversed: staff reported higher witness presence (52.6%) than faculty (41.2%), though cell sizes were small. This seems to suggest that faculty might experience more public or group-based verbal forms of harassment (e.g., crude comments in meetings or academic spaces) and staff might face more relational or proximity-based incidents (e.g., inappropriate attention from colleagues or supervisors), where witnesses are limited but not absent.
In addition, for staff, witnesses were more likely to be present in virtual SH incidents (70.2%) than in-person incidents (51.3%), a statistically significant difference. For staff, virtual workspaces (e.g., email, Teams/Zoom, chat) seem to increase the visibility of SH in shared communication channels or group settings. Results also indicated, however, that bystanders were less likely to take action when the transgressor was a supervisor. If incidents involving supervisors occur in shared or observable settings, the SH might be more visible but power hierarchies might inhibit BI due to perceived risks or organizational power dynamics. However, further work is needed to investigate these exploratory findings to better understand the context.
Implications for BI Training
Our study indicates that BI training and research are needed for both faculty and staff in academic settings. While much of the emerging literature on BI training in academia focuses on faculty, there is a dearth of information on programs for academic staff (McMahon, Manley, Amponsah et al., 2025). It might be beneficial to tailor BI training to staff separately, though research is needed to evaluate the best approaches. A recent study with library staff at 10 University of California campuses found that over half of participants reported experiencing or observing SH at work and desired small-group, scenario-based training to address the particular needs of their department culture (Barr-Walker et al., 2021).
Our study also highlights the importance of including the perspectives of targets in BI research, as their perspectives, along with those of bystanders, offer a more realistic perspective on what bystander actions are most helpful. Both targets and bystanders should be consulted when designing BI training, so they can describe real-life scenarios, offer various interventions and outcomes, and capture the nuances of academic culture (McMahon, Power, Snyder, et al., 2025). By using real scenarios and case studies, participants can discuss the opportunities, options, and risks associated with intervention in an engaging and applicable way (Banyard, 2024). This might allow programs to effectively present the range of bystander opportunities available to intervene and discuss which intervention opportunities would be most successful in a given context. More evaluation of training approaches is needed to better understand their impact.
Our findings emphasize the role of power dynamics in BI situations within the context of academia. There is increasing recognition of the way that bystanders’ intervention decisions in academic workplaces are impacted by power-based factors such as fear of retaliation, social location, and anticipated consequences for their well-being and career trajectories (Haynes-Baratz et al., 2022; McDonald et al., 2016; Scarduzio & Adams, 2022). For example, in our study, those whose SH was perpetrated by someone with supervisory power, compared to someone without such power, were more likely to report the presence of others. However, the decision to intervene when the transgressor has supervisory power is especially complex in academia due to its hierarchical nature and inherent risk (Haynes-Baratz et al., 2022). Both faculty and staff with less power depend on supervisors for work performance evaluations, mentoring, accessing networks, letters of recommendation for tenure, and opportunities for promotion or career development. The expectation for those with less power to intervene in these situations may create unfair “bystander burden” and place them in vulnerable positions (McMahon, Manley, Johnson, et al., 2025, p. 4). Scholars and practitioners have also highlighted that employees with marginalized identities face additional risks when intervening as bystanders, and thus training efforts must explicitly address and account for these vulnerabilities, understanding that safe options for BI look different based on identity, institutional factors such as the larger campus climate, and whether individuals feel supported or not (Bang et al., 2016; Harris & Linder, 2017; Hoxmeier et al., 2022; Linder, 2024; McMahon et al., 2020). As such, it is incumbent upon those with power—particularly those who are supervisors or managers—to intervene. Bystanders in these situations who also have supervisory power can help by holding their peers in supervisory roles accountable and working to create a culture of nonviolence and support.
Though further development of nuanced BI training efforts to address SH in academia are needed, these programs must be considered as a part of a larger package of prevention and response at the community and institutional levels, including the incorporation of policies, procedures, and structures to address SH and other forms of harm, as well as to build trust and accountability (Buchanan, 2014; Klein et al., 2021). An over-reliance on BI programs to resolve SH can place burden on bystanders and reduce institutional accountability (McMahon, Manley, Amponsah et al., 2025). Addressing the culture within academia that poises it as a “breeding ground” for SH must also occur (Zara et al., 2024, p. 29). The National Academies report suggests that mechanisms such as anonymous reporting, hiring protocols that review applicants’ history of SH, and protections from retaliation for whistleblowers must also be in place alongside BI training (NASEM, 2018).
Limitations and Future Directions
Several limitations must be noted. First, the study was conducted at one university, and results might not be generalizable. Replication at other institutions of higher education would be useful. Furthermore, while all faculty and staff at this university received the survey, there is potential for selection bias, and the response rate was low; therefore, there were likely perspectives on BI that were not captured in this study. Relatedly, the survey item inquiring into bystander action only captured doing “something,” or not, and without knowing the specific intervention tactic, we cannot make claims about what intervention looked like, and importantly, whether, and how, the impact on the target of SH differed depending on the action. While the finding that participants were more likely to disclose their SH when bystanders were present is interesting, it requires further investigation, as the directionality cannot be assumed based on the cross-sectional nature of the survey.
As researchers continue to investigate the nuances of SH in academic settings and the role of bystanders in mitigating its harm, measures should similarly evolve to capture the nuances in bystander action, power dynamics of those involved, among other factors, to better understand how these characteristics and contexts impact experiences and associated outcomes. Understanding the relationship between the various parties (transgressors, targets, and bystanders) are all important aspects of the context, yet we did not capture all of these relationships in our survey. Qualitative work with both targets and bystanders (and potentially bystander-target dyads) could help understand the contextual factors related to BI, its perceived helpfulness, and its relationship to outcomes such as the target’s decisions about help-seeking, and their well-being. Employee roles are fluid, and they may be bystanders as well as targets and/or transgressors within the same or different contexts and over time. Longitudinal research to better understand the overlap of these roles would offer important insight.
In addition, given the exploratory nature of this study and the relatively small analytic subsample of participants who reported past-year SH (n = 395), our analyses were limited to bivariate tests. Although multivariate analyses could provide a more nuanced understanding of how multiple factors interact to predict bystander presence, action, and outcomes, our sample size—particularly when stratified by incident characteristics and sociodemographic variables—limited statistical power to detect reliable effects in multivariable models. Certain cell sizes became too small to separate out faculty and staff, such as the interaction of those positions with various sociodemographic variables. Future research should build on these findings with larger samples and multivariable approaches to better understand the complex interplay of power, identity, and context in shaping bystander dynamics and the experiences of those targeted by SH.
Conclusion
This exploratory study contributes to the emerging literature on the potential role of prosocial bystanders for the mitigation of harm associated with SH in academic settings. The study focused on the perspectives of SH targets, who indicated that BI action was sometimes helpful, and none reported that it made things worse. These findings suggest that further work is needed to incorporate the viewpoint of both bystanders and targets into BI training efforts in order to understand barriers to taking action as well as what types of intervention are most helpful based on the particular context. The findings underscore the need for further refinement, implementation, and evaluation of BI models for prevention for SH among faculty and staff.
Footnotes
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.