Evaluating a dialogue-based bystander intervention for work-related gender-based violence and harassment: a mixed method pilot study

Work-related gender-based violence and harassment have become a topic of increasing interest since the 1970s (Fitzgerald and Cortina, 2018; McDonald, 2012), with renewed interest from the general public and researchers since the #MeToo movement went viral in 2017 (Johnson et al., 2019; Kessler et al., 2021). Gender-based violence and harassment (GBVH) is defined as “harassment and violence directed at persons because of their sex or gender or affecting persons of a particular sex or gender disproportionately and includes sexual harassment” (ILO, 2019, article 1. b). In contrast to sexual harassment (SH), the term GBVH acknowledges that not all acts (e.g. gender harassment) need to be sexual in nature to constitute harassment (O’Connor, 2020; Sojo et al., 2016).

A 2017 report including 28 European countries, estimated the prevalence of GBVH to be 13%–21% for women in the past 12 months, 45%–55% of women having experienced GBVH since the age of 15 (Latcheva, 2017), with women in employment indicating greater exposure. Indeed, the estimated prevalence of work-related GBVH (WR–GBVH) is alarmingly high worldwide, disproportionately affecting women and sexual minorities (UN Women and International Labour Organization, 2019). A recent Danish study (Larsson et al., 2023) suggested that the estimated prevalence rates of WR–GBVH vary across different sectors of industry between 9.2%–33.7% for men (n = 4,614) and 14.1%–51.8% for women (n = 4,389).

These rates not only reflect wider societal issues but also underline the need for the prevention of WR–GBVH as it is associated with severe individual consequences such as an increased risk of depression (Nielsen et al., 2024; Rugulies et al., 2020) and Posttraumatic Stress Disorder (Hansen et al., 2022), long-term sickness absence (Blindow et al., 2021), dissatisfaction with and disengagement from work (Lipinsky et al., 2022), reduced productivity (Lipinsky et al., 2022), self-harm (Jin et al., 2018) and hazardous substance abuse (Blindow et al., 2024). The disproportionate prevalence for women and minorities can be explained using the intersectionality framework (Crenshaw, 1991), which highlights how multiple marginalised identities (e.g. race, disability and sexuality) interact in complex ways, amplifying risk and discrimination. This framework explains why, for example, minority-race women face a greater risk of WR–GBVH than would be accounted for by gender or race alone. Consequently, some groups are significantly more vulnerable to WR–GBVH (Coulter et al., 2017; O’Connor et al., 2021) and may also experience more severe consequences (Bondestam, 2024).

With WR–GBVH being so prevalent, several approaches to WR–GBVH prevention have been developed, such as zero-tolerance policies, code of conduct implementations and different types of employee training (e.g. altering gender beliefs and increasing self-protection skills). However, there is limited evidence for the efficacy of these approaches so far (Diez-Canseco et al., 2022; Madsen and Nielsen, 2023), and many of these fail to include an intersectional approach (Ryan, 2023). Bystander interventions have been proposed as an alternative approach to WR–GBVH prevention. This strategy focuses on educating and “activating” passive bystanders, framing prevention of GBVH as a collective responsibility. Bystanders are encouraged to engage in direct intervention when witnessing GBVH, participate in primary intervention by speaking out against sexist jokes or assist in tertiary prevention by supporting victims in reporting the offence (McMahon et al., 2020; McMahon and Banyard, 2012). Notably, this approach has been implemented with special attention to intersectionality with promising results (Brush and Miller, 2023; Ryan, 2023).

However, a recent scoping review (Nielsen et al., 2025) points to a lack of research-based and empirically tested bystander intervention programmes aimed at reducing WR–GBVH, as well as important limitations in existing research. For example, most bystander intervention research on GBVH has focused on higher education settings (cf. Mainwaring et al., 2022; Mujal et al., 2021) and emanates from the USA (Madsen and Nielsen, 2023). Moreover, and reflecting an overall trend in studies on primary prevention (DeGue et al., 2014), most intervention studies do not include long follow-up periods (Nielsen et al., 2025). As such, the long-term effectiveness of most WR–GBVH interventions remains unclear. One example of an effective intervention without proper long-term assessment is the study by Potter and Moynihan (2011). The authors report on a WR–GBVH intervention within a military context. Due to the constraint of working with soldiers on active duty, the study only compares data from the 4.5-month follow-up between participants and the control group. Although the study showed positive results, the lack of a baseline prevents evaluating long-term effects. Furthermore, despite researchers recommending a mixed methods design when evaluating organisational interventions (Abildgaard et al., 2016; Nielsen and Abildgaard, 2013; Roodbari et al., 2023), most studies on primary level bystander interventions targeting GBVH–WH have only used quantitative measures (Nielsen et al., 2025). Researchers (Nielsen and Abildgaard, 2013) have also recommended measures of contextual factors to be included in intervention evaluations, yet such measures are often missing in studies of WR–GBVH prevention (Medeiros and Griffith, 2019; Nielsen et al., 2025). In other areas, longitudinal mixed-method designs have been used successfully. One example is the study by Arapovic–Johansson and colleagues (2020), which evaluated the implementation of a new productivity measurement and enhancement system. The study, which included both process evaluation and an analysis of the contextual factors, revealed several positive effects as well as barriers to implementation, pointing to the importance of not only focusing on direct outcome measures.

In Denmark, researchers and practitioners have recently co-developed a promising dialogue-based bystander intervention targeting WR–GBVH called “Intervene: Sexual Harassment! (Intervene-SH)” (Mikkelsen et al., 2021). Intervene-SH is a free-of-charge, user-friendly, standardised intervention primarily targeting gender harassment and unwanted sexual attention that organisations can implement themselves. The main part of the intervention is a two-hour workshop led by a facilitator, such as an occupational health and safety (OHS) representative. In the workshop, participants are presented with six bystander roles, representing both constructive and destructive bystander behaviours and short vignettes that portray real-life situations and harassment dynamics. The intention is for participants to become aware of the negative consequences of displaying passive or colluding bystander behaviours and how to act as constructive bystanders (Mikkelsen et al., 2021). The intervention builds upon core principles for organisational interventions by using active engagement of employees, explicating a distinct programme logic, working with existing processes and developing organisational learning capabilities. These elements have previously been highlighted as essential for maximising the impact of organisational interventions (Nielsen and Abildgaard, 2013; Von Thiele Schwarz et al., 2021).

The present pilot study is the first to assess the impact of Intervene-SH. It was designed as a longitudinal mixed method evaluation examining the intervention implementation across six care facilities within a Danish public organisation. The study’s primary aim was to assess how employees, facilitators and managers experienced the intervention. Secondly, outcomes and effects of the intervention were investigated. The third aim was to assess how contextual factors would affect intervention implementation and its effects. Focus group interviews were designed to address all three aims: investigating participants’ experiences of the 2-h workshop, noting any perceived positive or negative effects, as well as contextual factors that facilitated or obstructed its implementation. For aim two, a longitudinal questionnaire survey aimed to measure changes in bystander behaviours, attitudes, perceived barriers to intervention and the prevalence of GBVH before and after the intervention.

The study was quasi-experimental, with a within-subjects design. A concurrent mixed method approach was used, using a questionnaire survey and focus group interviews. Six care facilities (hereafter worksites) within the same organisation participated in the present study (n = 170 employees). The organisation was recruited using LinkedIn. Participation in the intervention workshop was intended to be mandatory for the employees. Participants completed questionnaires and participated in focus group interviews approximately two months before the intervention (baseline – T1) and again three months after the intervention (follow-up – T2) (see Figure 1).

Due to the organisation’s internal structure, it was not feasible to include a control group in this study. The departments themselves were too small to meaningfully divide into a “treatment” group and a control group. Besides, employees frequently interacted across departments, which would increase the risk of contamination in a control group (Nielsen et al., 2006). Additionally, each department had its own manager and target population, making them quite distinct from one another. This heterogeneity would have made it challenging to ensure that the control and intervention groups were comparable.

As a final step in the research planning, all necessary ethical and legal approvals according to Danish legislation were obtained for the present study, including ethics approval from the University of Southern Denmark Research Ethics Committee (approval ID 22/66992).

“Intervene: Sexual Harassment!” is a structured dialogue-based intervention which includes vignette-based discussions, practical intervention strategies, and strategic planning to create a safer work environment. The intervention is an adaptation of the bystander intervention “Intervene - Good working relationships without bullying”, which focuses on fostering collegial support and preventing bullying (Mikkelsen and Høgh, 2019). The four-step intervention consists of a series of meetings and workshops led by either internal (e.g. union or OHS representatives) or external facilitators. A plug-and-play manual for the intervention and all intervention materials are available free of charge at several Danish websites.

As described in the intervention pamphlet (Mikkelsen et al., 2021), the purpose of Step 0 is to thoroughly plan the intervention process. Among other things, this includes forming a steering group, getting familiar with the intervention itself, deciding who should facilitate and participate in the intervention, ways of communicating about the intervention process and its purpose, upskilling facilitators, information and communication, preparing materials to be used in the intervention, and deciding on how to evaluate the intervention process.

During the current project, the first and last authors facilitated an introductory Intervene-SH workshop for managers and employee representatives (ER). This was done to increase intervention fidelity (the degree to which a specific intervention is implemented as intended) and ensure that facilitators felt more ready to effectively lead the Intervene workshops. Following the introductory workshop, the facilitators met and prepped together for their respective workshops.

Steps 1 and 2 describe the facilitated dialogue meeting and constitute the main part of Intervene-SH. In Step 1, participants first learn about various forms of harassment, such as unwanted sexual attention and lewd jokes. They are then introduced to six “Bystander types” (see Supplemental Materials 1), three of which represent negative or destructive bystander behaviour and three positive or constructive behaviour. The participants, who are divided into small groups, then read two or more written vignettes (four representing colleague-on-colleague GBVH and two representing client-to-employee harassment) and discuss the displayed bystander behaviour and its consequences. The group dialogues are followed by a plenary discussion led by the facilitator.

The purpose of Step 2 is to initiate a dialogue on constructive bystander intervention and to equip participants with skills to address inappropriate behaviours. Specifically, participants engage in group dialogues on how bystanders could intervene in the presented vignettes. The dialogues are guided by vignette-specific questions and “suggestion cards” describing different bystander actions and intervention phrases. For example, the three questions for vignette 1 are “What would a ‘Defender’ say if she were present in the playground?” followed by “What effect will it have for Jeanette [the victim]?” and “What could a ‘Sympathiser’ do?”. The ready-made suggestion cards contain phrases such as, “Oops – we don’t use this tone here! Please refer to my colleagues in a respectful manner.”

This final step takes place after the dialogue meeting. The steering group (e.g. the work environment organisation) meet to develop and subsequently implement action plans to prevent future GBVH instances and to promote a supportive work culture. It includes summarising insights from previous steps and integrating them into the organisation’s policies and practices.

Nine focus group interviews with a total of 50 employees (50% female) were conducted at T1. Participation in the focus group interviews was voluntary. The number of interviews conducted at each worksite was adjusted in proportion to the number of employees to ensure that larger worksites had more interviews for equitable representation. Additionally, one focus group interview was conducted with elected ERs and another with worksite managers. At T2, eight employee group interviews were planned, but only five were completed (n = 19; 42.1% female) due to unforeseen circumstances such as employee illness and patient emergencies. The managers and ERs participated in separate group interviews at T2.

A separate semi-structured interview guide was constructed for each of the three interview types. The guides focused on the form, content and delivery of the intervention and whether participants had noticed any changes in their own or their colleagues’ behaviours. The last author facilitated the first two out of nine group interviews at T1, while the first author conducted the rest.

For the qualitative analysis, the first author transcribed the group interviews verbatim and anonymised them. During this process, managers, ER, employees and worksites were all assigned random numbers. To ensure anonymity, the manager’s and ER’s numbers do not necessarily correspond to that worksite number in the transcriptions. We used thematic analysis (TA), which is recommended for analysing qualitative data in organisational intervention evaluations (Roodbari et al., 2023). The first author conducted the initial analysis, which followed the iterative five-step process proposed by Braun and Clarke (2006, 2022). This approach to TA is inspired by realist methods that focus on participants’ experiences, meanings and reality (Braun and Clarke, 2021, 2022). Firstly, transcripts were read to gain familiarity with the data. All interviews were then inductively coded in NVIVO 14 (Lumivero, 2023), with continuous refinement of the codes. After coding, the established codes were grouped into initial themes, with obstructing and facilitating factors being the only a priori themes. Themes were refined and finalised through discussions between the first and last authors, ensuring robust final themes that accurately represented the data.

The questionnaires were distributed electronically via email. In total, 92 of the 167 participants invited at T1 completed the questionnaire (response rate: 55%). At T2, 70 of 160 questionnaires were completed (response rate: 44%). Seven email addresses were no longer in use at T2 and were excluded from the response rate calculation. Fifty-two participants completed the questionnaires at both T1 and T2 (dropout 43.5%). As shown in Table 1, most participants were female, above 35 years of age and worked full-time.

Assessing bystander behaviours, attitudes, and opportunities to intervene presents unique challenges because of their complex and multifaceted nature (Hoxmeier et al., 2023; McMahon et al., 2017). To address these challenges, this study used multiple outcome measures adapted from various sources.

The questionnaires administered at T1 and T2 were identical, except for additional questions and statements at T2 regarding participation in the intervention and how this was experienced. All scales were translated from English to Danish for the present study using translation and backtranslation, except for the inventory of workplace sexual- and gender-based harassment (IWS), as a Danish version already exists. Only measures relevant to the current study are reported here.

The estimated rate of GBVH was measured using 12 items from the IWS (Larsson et al., 2022; M. B. D. Nielsen et al., 2024). The original IWS consists of 21 items covering three main categories of GBVH: unwanted sexual attention (14), gender harassment (3) and coercion and bribery (4). Three additional questions relate to who the perpetrator was, whether the respondent received help afterwards and whether they had witnessed GBVH. Nine of the original 21 items were deemed irrelevant for the current study based on dialogue meetings with representatives from the target organisation and low item prevalence rates in the Danish reference material (Larsson et al., 2023). The final scale included eight items measuring unwanted sexual attention, three items measuring gender harassment, and one item measuring sexual coercion. Cronbach’s alpha for the total score on the 12 IWS items was acceptable at T1 (α = 0.67) and T2 (α = 0.74). Respondents were asked how many times they had experienced each behaviour within the past six months, rated on a five-point scale: 0 (never), 1 (once), 2 (2–5 times), 3 (6–9 times) and 4 (10 times or more).

This 12-item scale, developed by Mikkelsen et al. (in preparation), assesses self-reported bystander behaviours in GBVH situations across multiple dimensions, including passive/active and constructive/destructive behaviours (see Bowes-Sperry and O’Leary-Kelly, 2005; Paull et al., 2012). The scale consists of a positive and negative subscale with nine and three items, respectively. Respondents rated their agreement on a five-point scale from 1 (strongly disagree) to 5 (strongly agree), with a “don’t know” option. Cronbach’s alpha for the positive subscale was poor at T1 (α = 0.52) and T2 (α = 0.63) and very poor for the negative subscale at both T1 (α = 0.37) and T2 (α = 0.30). The low alpha level may be due to the low number of items on the scale. It could also be an issue with unidimensionality, as bystander behaviours might conceptually differ from each other. This would breach the assumption of essential tau equivalence, meaning that all items in the scale might not measure the same latent trait with the same precision, producing a significantly lower alpha value, especially for a scale with such a low amount of items (Graham, 2006; Hayes and Coutts, 2020). Using McDonald’s Omega, a measurement recommended when tau equivalence is not met (Hayes and Coutts, 2020), the negative subscale shows slight improvement at T1 (ω = 0.45) and T2 (ω = 0.48). The positive subscale also improved at both T1 (ω = 0.58) and T2 (ω = 0.68).

The seven-item barriers to bystander action scale was adapted from Edwards et al.’s (2021) five-item subscale on School Personnel Barriers to Bystander Action by modifying the wording to fit a workplace context and adding two items. The new barriers added to the scale were “I don’t intervene because the victim brought it on themselves” and “If others don’t do anything, I don’t need to do anything either”. This scale assesses the extent to which potential bystanders perceive various impediments (e.g. lack of skills) as barriers to engaging in prosocial bystander behaviour. Respondents answered on a four-point scale from 1 (strongly disagree) to 4 (strongly agree), with the addition of a “don’t know” option. Cronbach’s alpha for this scale was good at T1 (α = 0.83) and T2 (α = 0.79).

Grounded in Ajzen’s (1991; 2019) theory of planned behaviour, this scale was developed to measure perceived bystander behavioural control in situations with bullying (Evensen et al., 2024) and later revised for a Danish context (Mikkelsen et al., in preparation). This nine-item scale assesses respondents’ perceptions of their ability to deal with negative acts (NA) as a potential bystander or target and their beliefs regarding normative behavioural responses to such acts in the workplace. Respondents answered on a five-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). Cronbach’s alpha for the scale was good at T1 (α = 0.83) and T2 (α = 0.86).

To evaluate the impact of Intervene-SH on other NA, we used the short-negative acts questionnaire (SNAQ) (Notelaers et al., 2019) along with four items from the original Negative Acts Questionnaire-Revised (Einarsen et al., 2009), deemed relevant for this study. The SNAQ is comprised of three subscales, each with three items: person-oriented NA, work-related NA and social exclusion. We added one question to each subscale and one on physical intimidation to create the Adapted Short-Negative Acts Questionnaire (A-SNAQ). We used the original response scale of 1 (never), 2 (occasionally), 3 (monthly), 4 (weekly) and 5 (daily). Both scales have been shown to have good psychometric properties (Einarsen et al., 2009; Notelaers et al., 2019). Cronbach’s alpha for each subscale in the current study was acceptable at T1 (α = 0.70–0.91) and T2 (α = 0.67–0.84).

For T2, we developed five statements to measure the participants’ evaluations of the intervention effects. The statements covered topics such as perceived relevance of the intervention, perceived improvements in intervention abilities and perceived changes in one’s behaviour (see Figure 3 for all questions). Respondents answered on a five-point Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree). The questions were analysed using descriptive statistics.

Quantitative data were analysed using SPSS version 29 and Excel version 16. Paired sample t-tests were conducted to assess the differences between the three T1 and T2 outcome measures (e.g. the estimated effect of the intervention). Three outcome measures had missing data (9.6%–28.8%) due to the inclusion of a “Don’t know” answering option. We conducted sensitivity analyses for these measures using various methods of mean imputation, including case mean, population mean and the expectation-maximisation feature in SPSS 29. No significant differences in the test outcomes were found between the imputation methods. To evaluate the prevalence of WR–GBVH and NA, we created a dichotomous scale using the IWS and A-SNAQ scores, where a score of 0 indicated no exposure and a score of 1 or higher indicated exposure, using the full sample at T1 and T2. These groups were used to assess changes in prevalence using two-proportion Z-score tests. The five statements created to assess the participants’ perceived effects of the intervention were analysed using descriptive statistics, i.e. distribution of responses.

In the following section, our results are reported in order following the three study aims. As such, the qualitative results cover (1) participants’ and facilitators’ experiences with the intervention, (2) the perceived effects of the intervention, and (3) facilitating and obstructing contextual factors. As the quantitative data only measure outcomes and impacts, these are reported together with the related qualitative results in the section “study aim 2” (see Figure 2).

The following analysis for aim 1 is divided into four sections, each covering a major component of the intervention. Sections 1, 2 and 3 explore participant experiences with bystander typologies, vignettes and suggestion cards, which are the core contents of the intervention. As the intervention was designed to allow companies to implement it themselves, the fourth section of the analysis investigated how participants and ER experienced this aspect.

It should be noted that many participants had difficulties remembering details of the intervention, which reportedly was associated with the conflation of other projects. However, when prompted by the interviewer, most participants were better able to recall the intervention material.

One of the few elements of the intervention that most participants could recall unprompted was the six bystander types, which many participants could relate to. A representative comment from an employee was, “I think you can always recognise yourself a bit in all these characters”. Generally, the participants thought that the bystander typologies were the most useful part of the intervention: “I think that the bystander types were something that was particularly exciting about this whole process. So, it’s actually one of the things I remember really well and carry with me when dealing with everyday conflicts” (Manager 4). After the intervention, some worksites hung posters with bystander types in communal spaces as reminders. Bystander posters were also used to spark conversations about the intervention among employees and as a tool to introduce the concept of “Intervene-SH” to new employees.

The interviewees had differing opinions on the four GWBH vignettes. Whereas some found them to be “plausible scenarios where I thought they [the vignettes] are not, like, that far-fetched” (employee, worksite 6), other employees, especially from one worksite, found the vignettes unrealistic and unrelatable: “The problem was that many of the issues [from the vignettes] were completely off the mark” (employee, worksite 2). A recurrent theme reflecting aspects of the critique vented by these interviewees was named “Beating around the bush”. According to these employees, the use of hypothetical vignettes steered them away from discussing “real-life” examples of problematic behaviour in the workplace: “We don’t really need to invent that much. Because there are things [issues with GBVH] we could talk about. But we’re avoiding that conversation” (employee, worksite 4). Some facilitators expressed similar views and discussed how it could be difficult to talk about vignettes that reminded them of behaviours displayed by specific colleagues. Facilitators experienced how this could create resistance amongst certain employees who avoided participating in the discussion as they felt called out:

Most participants found the vignettes, questions and suggestion cards relatable and realistic, though not every question proved to be helpful for all participants: “I’ve […] noticed, after conducting several sessions myself, that some of us, including me, feel that not all the questions are equally useful” (ER 4). Although none of the participants recalled any exact phrases on the suggestion cards, they expressed that they generally felt that they had been useful. An employee reported, “I remember sitting with those cards and saying, one could also do it this way. Or one might also say it like this […] We sat discussing it” (employee, worksite 6). However, one facilitator expressed concern about the actions on some of the suggestion cards: “there are some of them [the questions], or the statements rather [from the suggestion cards], which may be a bit more conflict escalating than anything else” (ER 4). This shows that not all the suggestion cards are universally applicable, as they might be interpreted differently than intended. This makes sense, given that some phrases might be seen as negative or hostile depending on the intonation of the speaker. Some participants also expressed a wish for an increased emphasis on practical, replicable bystander actions in the workshops and felt that the suggestion cards were insufficient in that respect.

The intervention was designed as a “self-implementation” tool, and accordingly, we explored how the participants experienced this aspect. When analysing the participants’ answers to the question “What did you think about it being your ER that facilitated the intervention?”, the results were mixed. About half of the participants found it beneficial that the intervention was delivered by an ER that they considered “one of their own”, whereas the other half found it problematic. Participants who found it beneficial thought that having ERs as facilitators made the intervention more relatable, present and relaxed. One employee (worksite 6) reported, “There can also be some comfort in it being someone you know, right? In terms of being able to speak more openly and such”.

Conversely, other employees questioned the use of ERs as facilitators, highlighting that the intervention could appear disingenuous if the ERs were perpetrators of NA at the same workplaces as they facilitated. One participant reported experiencing this issue. She disclosed that the ER facilitating their intervention “Doesn’t always behave well” (employee, worksite 2), which made her question the integrity and trustworthiness of the intervention. Others believed that using external consultants would create a feeling of authority or seriousness regarding the facilitator’s role and the intervention that using ERs did not.

Even though many participants initially had difficulties remembering the intervention, they stated that they felt “something” had changed. The TA revealed two major themes.

Employees and ERs alike found that the intervention made them more aware of GBVH issues. One employee (worksite 6) stated, “I think perhaps it has made me better at being mindful about different situations that might otherwise slip past my radar”. Across interviews, employees highlighted how learning about bystander typologies provided new insights into the different dynamics of NA in the workplace. A common phrase was that the bystander types was a real “eye-opener”. Interviewees also reported that the intervention supplied them with new knowledge on why and how they should intervene:

Another interesting finding was that participants articulated a desire to apply these insights to various situations beyond GBVH, including conflicts, bullying and challenging interactions with clients.

Many interviewees also reported being more conscious of a personal responsibility to be aware of personal boundaries (own and others) and to intervene when needed. Indeed, one employee (worksite 6) noted “an increased awareness and a greater willingness and motivation to act” as the main takeaway from the intervention. When touching upon the issue of personal responsibility in the interviews, this was often followed by comments about how the intervention also created a “shared” or “communal” responsibility to intervene. According to one ER (3): “It’s like we have given each other carte blanche to act, to interfere, to intervene and so on, right? And I think that’s quite noticeable”. Managers also pointed to a shared responsibility to act as an important takeaway: “It’s about making everyone accountable. That is, there’s no one who can say 'but it had nothing to do with me'. Yes, it does because you witnessed it” (Manager 4). The intervention also appeared to have paved the way for tangible actions, albeit with the acknowledgement of the challenges involved. One employee noted: “I would say that I feel others are more aware of it [GBVH] too. That is, they also speak up and such” (employee, worksite 5). However, participants also described the need for continued focus and practice, as intervening was still considered difficult:

According to the interviewees, the intervention led to new conversations about boundaries, humour and the general tone in the workplace. One manager remarked that during and after the workshops, discussions of boundaries created a “shared language and method” (Manager 5), through which they could understand bystander responsibilities and the social dynamics involved in NA. Employees from worksite four discussed how being part of the project had given them new terminology and created a space for new discussions about acceptable and unacceptable workplace behaviours. Other employees reported that the Danish name of the intervention “Grib ind”, which is also a term for intervening, had become a common phrase used in daily conversations. An employee declared: “Just the fact that we have a word for it means a lot” (employee, worksite 5).

At one worksite, this new shared language has been used to address GBVH issues. A manager explained:

While the qualitative analysis revealed significant perceived effects of the intervention, the quantitative data presents a more nuanced picture. A series of two-proportion z-tests were conducted to compare the proportions of respondents who either experienced GBVH and NA or were bystanders at T1 and T2. As shown in Table 2, the prevalence of GBVH was approximately one-third at both times, with most participants experiencing only one incident. By contrast, a high level of NA was found both before and after the intervention. As the questionnaire asks about experiences from the past six months, the two reference periods do not overlap.

Table 3 shows that respondents did not differ significantly in any of the outcome variables between T1 and T2.

Despite the null findings of the t-tests and z-tests, the intervention-specific statements showed a different picture. Figure 3 illustrates the distribution of responses to the five statements, highlighting several interesting insights. Notably, more than 40% felt better equipped to intervene, and over 20% of the participants reported modifying their behaviour or communication style post-intervention. These quantitative insights resonate with the qualitative findings, suggesting that the intervention fostered a heightened awareness and willingness to act, even if broader behavioural changes were not statistically significant.

The TA also identified several factors that either facilitated or obstructed the implementation and effects of the intervention.

Three major factors were identified across the interviews as having positively influenced the implementation process: leadership support, employee engagement and co-facilitation.

While some employees felt that managers failed to allocate sufficient time or resources to the project, most employees and ERs noted that managers across the six worksites had supported the project. For example, they allowed ERs to prepare for workshops and rehearse the materials and allowed employees to participate in interviews during work hours.

Co-facilitation was implemented by ERs on their own initiative. During the follow-up interviews, all ERs acknowledged that cooperation during preparation and facilitation had proven to be an asset, enhancing both their overall experience with the intervention and the quality of the facilitation process:

The importance of employee engagement was discussed in most of the follow-up interviews. One ER described how colleagues would introduce newcomers to the principles of “Intervene-SH”:

Five major obstructing factors were identified: competing projects, lack of identification with stressors (GBVH), lack of intervention fidelity, management and communication barriers and lack of follow-up on project activities.

The simultaneous implementation of other projects, such as a project focusing on psychological safety and a research project on interpersonal conflicts and psychosocial work environment, meant that interviewees had difficulties recalling aspects of “Intervene”. In one interview, interviewees spent approximately one-third of the time discussing activities from other projects and found it difficult to distinguish between them. According to the interviewees, the implementation of “Intervene-SH” suffered because of these other projects, which demanded time and resources. Moreover, our analysis suggests that some employees may have suffered from “project fatigue”. According to the ERs, a segment of the participants expressed considerable resistance towards implementing Intervene-SH due to the concurrent initiation of multiple projects:

Another ER explained that the resistance was not directed at the content but at the time it consumed:

Follow-up interviews revealed that many participants did not find the intervention relevant, as GBVH was not considered an issue. Across interviews, most managers, employees and ERs framed GBVH as a non-issue, whereas bullying and conflicts were viewed as “real issues” requiring attention. Interestingly, while stating that GBVH did not occur at their worksites, many interviewees mentioned multiple examples of both overt SH and gender harassment, as well as more subtle examples of GBVH, such as “mansplaining” and having to hear about colleagues’ sexual exploits. However, each was conceptualised as a standalone incident perpetrated by one “bad apple”. Conversely, NA such as gossip or bullying were seen as reflecting a cultural issue.

Interviewees from worksites two and three found the intervention to be less relevant. According to the two facilitators, this resistance was also apparent during the workshops, wherefore they changed the intervention’s content. In contrast, some interviewees reported that one facilitator had told them that he was not prepared to facilitate and did not think that they would find the intervention relevant or useful. Unsurprisingly, employee engagement in this workshop was reported to be particularly low. Interestingly, interviewees from this worksite also joked about SH during the follow-up interview. One interviewee commented, “We don’t have sexual harassment, we call it a perk”, while another jokingly noted that he “missed the eighties […] when there wasn’t any fuss about anything”. Both remarks were laughed at by other interviewees. This could be interpreted as a way to minimise or downplay the seriousness of the issue.

In the baseline and follow-up interview, participants from worksite two reported that they coped with the high emotional demands and stress by venting frustrations regarding, for example, clients and colleagues. Questioning the relevance of the intervention could thus be interpreted as a reaction to the risk of losing a valued coping strategy should the general rules of conduct in the workplace change. Following a discussion of how their culture was “special” and that newcomers who did not fit in would quickly leave, some participants from worksite two expressed a fear of being misunderstood and thus labelled a perpetrator, should behavioural norms and expectations change.

During the group interview at worksite two, the interviewees collectively constructed a narrative suggesting that the intervention was necessary only because “some employees” were overly sensitive. The term “some employees” referred to individuals not present at the interview and, more broadly, those outside the core workforce of worksite two. Through various discursive practices, these “other” employees, who might have supported the intervention or recognised GBVH as an issue, were seen as overly sensitive individuals who needed to develop a thicker skin. Again, the intervention and its focus were framed as an attack on the specific subculture and communication style of these workers.

At worksite 3, the facilitators made significant changes to the intervention materials, making them about conflicts and bullying. At worksite 2, the intervention was altered to focus on NA in general, following widespread discontent with materials and vignettes, which were deemed irrelevant and uninteresting by the employees.

The TA uncovered multiple issues pertaining to the organisation’s internal management of the project. As such, there appeared to be significant confusion about the project’s scope and focus, with some participants thinking that it was about bullying, conflict or psychological safety, while others thought that it targeted harassment from clients and their relatives. One ER noted: “There are doubts about whether the project is about sexual harassment or bullying. They’ve [employees] said that it seems more and more like it’s only about bullying”, perhaps reflecting inadequate communication about the project and its purpose on the part of managers and ERs or it could be another example of employees confusing it with other concurrent projects with similar topics.

Some ERs and employees felt that the project was “a management project” which had been forced on them. In particular, the ERs talked about the lack of a bottom-up democratic decision process and that managers should have asked them if they and the employees found the project relevant before comitting to it. In addition, across the follow-up interviews, some employees and ERs expressed that management had failed to prioritise the project, illustrated, for example, by management’s decision to only allow one “Intervene-SH” workshop for each worksite. Consequently, many employees were unable to attend, which again caused frustration, especially in situations where colleagues thought to display problematic behaviours did not participate.

None of the worksites had planned any follow-up activities, although the intervention materials highlight the importance of follow-up and include guidelines and methods to do so. The participants expressed pronounced dissatisfaction with this. One ER (7) elaborated on it: “Some thought that, now there will be a focus on this […] and they haven’t really felt that way afterwards”.

The TA and intervention-specific statements revealed that, although interviewees from especially one worksite expressed predominantly negative views about the intervention, most employees, ERs and managers found it relevant and relatable. They reported experiencing significant changes in themselves and their work environment, such as feelings of personal responsibility and communal accountability. Additionally, the interviews showed a burgeoning change in workplace discourse about GBVH, enabling discussions on previously unspoken topics. On the other hand, the quantitative analyses did not reveal significant differences in the outcome variables. Taken in isolation, these findings might suggest that participants did not benefit from the intervention. However, these analyses are underpowered and should be interpreted carefully. The limitations and possible explanations for the quantitative results are discussed below.

Nevertheless, given that positive effects were demonstrated through qualitative analysis and self-evaluation, our study highlights the importance of using mixed methods in intervention evaluation. The positive effects are similar to those seen in other intervention studies targeting workplace climate (Shea et al., 2021).

While facilitators and participants alike were generally positive about the intervention’s format and design, the use of hypothetical vignettes became a point of discussion during interviews. While generalised vignettes provide total control over the narratives, they may fail to resonate with participants who do not recognise the described issues in their daily work lives, which may again lead to disengagement and scepticism (Hershcovis et al., 2017; Mikkelsen, 2020). On the other hand, while using real-life examples can make the intervention more relatable, it may entail certain risks. An example could be the escalation of existing conflicts or the stigmatisation of individual employees seen as offenders. This may further a psychologically unsafe environment instead of putting focus on constructive bystander behaviour. Whether workplaces should use real-life examples or hypothetical vignettes should be determined on a case-by-case basis, accounting for facilitator skills, workplace psychosocial climate, the material’s relatability to the specific context, and the level of psychological safety within the intervention group.

The use of ERs as facilitators was met with mixed reactions from the employees. While some appreciated the relatability of ERs, it made others question the intervention’s effectiveness and integrity, especially when ERs were part of ongoing workplace conflicts. This raises the question of whether the intervention is suitable for self-implementation in all organisations. Deciding whether to use internal or external facilitators may be a question of economic resources. Yet, the pros and cons must be discussed in the planning phase. Regardless, inexperienced facilitators should receive support and training before facilitating dialogue-based bystander interventions, which could bring about complex group dynamics (Mikkelsen and Holmegaard, 2025).

Lack of identification with GBVH as a relevant issue was found to be a major obstacle to intervention implementation, with some worksites altering or abandoning intervention materials. This points to the potential pitfalls of not performing a comprehensive needs analysis, for example, via organisational diagnostics (Harrison and Shirom, 1999), before intervention implementation (Von Thiele Schwarz et al., 2021). Although we did conduct a needs analysis, it was not possible to adjust the intervention focus as our primary research aim was to test Intervene-SH.

Managerial support and alignment between intervention objectives and organisational priorities are key factors in organisational readiness (Weiner, 2009), which facilitates successful implementation. Our findings align with previous research showing how organisational readiness and the fit between the intervention and organisational context moderate success (Roodbari et al., 2023). Factors like time constraints and competing priorities, though seemingly trivial, are often underestimated by management, yet they may cause significant disruptions. In the present study, a lack of follow-up activities indicated a shift in management’s focus, signalling to employees that the intervention is no longer a priority. This may explain why many employees had trouble remembering it. A constant shift from project to project may also fuel “project fatigue”. Thus, our study highlights the challenges of engaging worksites with multiple concurrent initiatives, echoing findings in other studies (Abildgaard et al., 2016).

Our study design addresses several limitations in previous bystander intervention research within organisations. Using mixed methods allowed for a comprehensive evaluation, identifying key elements affecting the intervention’s implementation. Additionally, a longitudinal design addressed the limitation of short follow-up periods (Coker et al., 2022; Perry, 2023) and provided insights into the intervention’s long-term effects. Including a third follow-up measure after an even longer period (i.e. 12 months) might have further established long-term behavioural and attitudinal changes not present at the 3-month follow-up (Hill et al., 2016; Medeiros and Griffith, 2019). However, as commonly seen in organisational intervention research (Abildgaard et al., 2016), participants had difficulties remembering the intervention at the three-month follow-up. This begs the question of how to balance the need for longitudinal research with the hectic everyday lives of organisational members who may forget nonessential information.

Building on these methodological insights, our study provides valuable insights into the design and implementation of workplace interventions aimed at addressing GBVH. The findings highlight the limitations of a one-size-fits-all approach, which often fails to consider the diverse cultural and organisational contexts in which these interventions are applied (Cortina and Areguin, 2021; McDonald and Flood, 2012). Furthermore, our study showed that using vignettes instead of real-life cases presents both advantages and disadvantages. Thus, effective adaptation of intervention materials, for example, when developing real-life cases or new vignettes, is essential to ensure appropriateness and relevance. This should be done based on a thorough needs analysis and in collaboration with employees, ERs and other stakeholders. Such an approach could improve employee engagement and minimise resistance, reducing the likelihood of unplanned modifications to intervention materials, as observed at two worksites in this study. A larger catalogue of vignettes could be developed in collaboration with industry stakeholders to make an “item bank” from which vignettes could be chosen and further tailored. This speaks to the growing interest in and need for participant research in organisational psychology, especially when addressing sensitive issues (Bal et al., 2019; Mallory, 2021).

Several limitations should be acknowledged. Firstly, without a control group, we cannot determine whether the observed changes (or lack thereof) are due to the intervention or if other factors are responsible. Furthermore, the low prevalence of GBVH, combined with the three-month follow-up period, may represent a limitation, as it reduces participants’ opportunities to apply bystander skills. Similar barriers to training transfer and a decline in bystander skills following interventions have been reported in previous studies (Kania and Cale, 2021; Kuntz and Searle, 2023). Secondly, the sample size was small and limited to one organisation in Denmark, which limits the generalisability of the results beyond this specific context. Furthermore, the dropout rate was quite high which may have introduced bias in our data. This also resulted in underpowered quantitative analyses. Future research should include larger samples, preferably with multiple organisations within various industries and across different cultural contexts. A third limitation is the low reliability of certain scales, especially the two subscales of the bystander behaviour scale (BBS). As such, further investigation and refinement of this scale is warranted. Larger samples would allow the use of factor analysis to examine the scale’s structure and determine whether the two subscales align with the assumed constructs.

Limitations also include challenges related to fidelity and dose (the number of employees exposed to the intervention) of the intervention. DeGue et al. (2014) systematic review of primary prevention strategies for sexual violence perpetration found that longer sessions and exposure to multiple sessions increased the chances of positive outcomes. As such, the one-session design of the intervention might be an issue. Also, given that many employees did not participate in the intervention, the content of which was also changed at two worksites, it is difficult to assess the intervention’s organisational impact. However, high fidelity and employee involvement can sometimes be at odds, and achieving both might not always be possible (Schelvis et al., 2016). Therefore, whether lower fidelity is an issue should be evaluated on a project-by-project basis.

A limitation concerning the intervention material was the helping cards only containing proactive and direct interventions. These might not be suitable for all situations, and could potentially be conflict-escalating (Vranjes et al., 2023).

This study aimed to assess how employees, facilitators and managers experienced the bystander intervention “Intervene: Sexual Harassment”, its effects and how contextual factors affect implementation. Although no significant changes in quantitative outcome measures were found, the qualitative data offered valuable and detailed insights into the participants’ experiences of the intervention and its perceived effects, such as an improved understanding of bystander roles, a new language with which to discuss GBVH, and an increased sense of responsibility to intervene. However, results also indicate that tailoring vignettes to organisational context could be a way to improve employee engagement.

Further research addressing the study’s limitations is necessary to explore the potential long-term effects and scalability of Intervene-SH. Such studies should use longitudinal designs and include larger and more diverse samples, and mechanisms to monitor and ensure both dose and fidelity. Continuous support from researchers offers the possibility of adjusting study parameters to account for potential disruptions in the intervention process. There should also be a focus on in-depth investigations of contextual factors, which could give insight into how the intervention’s effectiveness and applicability across different organisational settings could be improved.

Erratum: It has come to the attention of the publisher that article Nielsen, K.S., Hansen, M., Bjørkelo, B. and Mikkelsen, E.G. (2025), “Evaluating a dialogue–based bystander intervention for work–related gender–based violence and harassment: a mixed method pilot study”, International Journal of Conflict Management, Vol. Ahead of Print No. Ahead of Print. https://doi.org/10.1108/IJCMA-09-2024-0229, contained incorrect sub-heading levels under the sections titled “Facilitating factors” and “Obstructing factors”.

The heading levels has now been amended correctly to appear appropriately as Heading Level 4 under the respective section headings.

This error was introduced during the article publication process, for which the publisher apologises.

The authors would like to thank the participating organisation for their cooperation.

Funding: University of Southern Denmark, Faculty of Health Sciences, NO. PhD Scholarship; Nordic Research Initiative Sexual Harassment, No. 21-OC2-00104.

The supplementary data for this article can be found online.