Study design

This qualitative study used focus group discussions (FGDs) to gain a deeper understanding of how workplace norms and organisational culture around sexual harassment are perceived and articulated by employees at Lund University. The analytical approach employed was qualitative content analysis (QCA), as described by Graneheim and Lundman [22]. This approach is well-suited to studies seeking to explore variation in perceptions and social meaning-making, allowing for abstraction from manifest content into latent themes.

Study population and sampling

Data were collected as part of the Tellus project at Lund University. Located in southern Sweden, Lund University employs around 8,600 staff and enrols 46,000 students across nine faculties. Faculty size range from 200 in Fine and Performing Arts to 1,900 in Science, and the proportion of women from 37% in Science to 65% in Law. International employees account for 40% of the workforce, and 35% hold fixed-term contracts. Faculties are headed by deans, and departments by heads of department. Both of these positions are elected [23].

Lund University follows the definition of sexual harassment found in the Discrimination Act, and has a strict zero-tolerance policy to harassment, sexual harassment and victimisation [24].

The Tellus project was a university-wide initiative launched in 2018 to strengthen the prevention of and response to SH among students and staff. Tellus included both qualitative and quantitative components. Key findings from the survey outlining the prevalence and characteristics of SH in this context are presented in an overview article [16].

All staff employed at the university were eligible to participate in this study, regardless of employment type, position, or personal experience of SH. Recruitment to FGDs was carried out from November 14^th, 2018, to May 7^th, 2019, and was conducted through posters, digital newsletters, and internal staff communication channels. These posters were displayed on staff notice boards in all faculties and internal communication was sent to all employees via staff email. Those employees who expressed interest in participating in the study contacted the research team via a project-specific email address and received detailed information about the study, including scheduled times and locations for the discussions. All employees who expressed interest in participating were offered a place in an FGD, however, several participants did not attend the FGDs once booked.

Data collection

A thematic discussion guide was developed (Appendix 1), covering perceptions of SH, situational contexts, forms of expression, workplace dynamics, and expectations for the university’s future response. Although the university has a definition of SH that is based on the Discrimination Act [24], no a priori definition was provided, in order to avoid prejudicing participants’ responses and to explore how SH is understood and discussed among employees. Instead, participants were asked if they believed SH occurred at Lund university, with follow up prompts exploring what this SH looks like, and what forms it could take. The guide was pre-tested with academic staff at Malmö University. Located in southern Sweden, close to Lund University, Malmö University employs around 2,200 staff and enrols 28,000 students across five faculties. International employees make up about 20% of staff [25].

Data collection took place between February and May 2019. In recognition of the influence of formal power dynamics in group interactions, FGDs were stratified by managerial status. Participants who reported having managerial responsibilities were placed in separate groups from those without such roles. All FGDs were audio recorded and transcribed verbatim by a research assistant at the university. The analysis followed the principles of qualitative content analysis according to Graneheim and Lundman [22]. Firstly, all transcripts were read through multiple times to gain familiarity with the data. Following this, meaning units relevant to the research aim were identified and coded inductively in English by the first author. These codes were then grouped into sub-categories and categories through a combination of inductive and deductive processes, moving iteratively between the empirical data and emerging interpretations.

Categories were subsequently examined for the development of latent sub-themes through a process of pattern recognition and conceptual reflection. Finally, one overarching theme was formulated. Quotes were selected to illustrate the categories, and these were translated to English by the first author. Data analysis was conducted using OpenCode 4 [26]. The final model was developed collaboratively, with all co-authors reviewing coding decisions and contributing to analytical refinement.

Ethical considerations

Ethical approval for the study was granted by the Swedish Ethical Review Authority (reference number: 2018/350). All participants were provided with detailed written and oral information prior to the FGDs, outlining the purpose of the research, the structure of the FGDs, and the voluntary nature of participation. Individuals were informed of their right to ask questions and seek clarification before providing written informed consent.

Given the group format of the discussions, participants were explicitly informed of the limitations to confidentiality. While the research team could ensure secure handling and storage of data, confidentiality among group members could not be guaranteed. To support open dialogue and minimise potential power imbalances, participants with managerial or supervisory responsibilities were assigned to separate FGDs from those without such roles, thereby avoiding situations in which staff and their line managers would be placed in the same discussion.

During the FGDs, participants were given the option to use their real names or a pseudonym. All were encouraged to maintain the confidentiality of what was shared within the group. At the end of each session, participants received an information sheet listing relevant organisations and services available for psychological or emotional support, in case any distress arose during or after the discussions.

All identifying information was removed during the transcription process to protect participant anonymity. All audio recordings were transferred to encrypted hard drives immediately after the FGD and deleted from the audio recorders. These audio recordings and transcripts were securely stored on encrypted drives with access restricted to members of the research team. Consent forms were stored separately in a locked safe.

Overarching theme: Perceiving sexual harassment through the lens of organisational silence, power relations, and negotiated boundaries

The overarching theme reflects how participants made sense of SH not simply as individual acts of misconduct but as phenomena embedded in the fabric of university life, shaped by institutional cultures, informal norms, and relational hierarchies. Perceptions of what constitutes SH were contingent on dynamic and negotiated boundaries, often shaped by peer interaction, managerial response, and broader societal shifts such as #MeToo. Across the discussions, participants highlighted how unclear definitions and what they experienced as changing expectations created ambiguity, fostering environments where boundary-crossing behaviour was normalised or explained away. Power relations, both formal and informal, were seen as key to understanding who was protected, who was heard, and who bore the risks of speaking up. Silence emerged not only as a symptom of organisational failure but as a protective strategy, adopted by employees navigating uncertain terrain. As a result, participants described environments where misconduct could flourish, not due to ignorance, but rather because of varying interpretations of situations, tacit tolerance, reputational concerns, and the absence of consistent institutional accountability. Taken together, these dynamics reveal how harassment was perceived through a cultural lens, one in which silence, power, and contested norms continuously shaped the boundaries of what could be said, challenged, or changed.

Negotiating collective norms and meanings

This sub-theme focuses on how participants collectively shaped, contested, and reinforced understandings of SH. It shows how, in the absence of clearly defined definitions, understandings, and boundaries are actively negotiated. Throughout the discussions, participants contested boundaries between appropriate and inappropriate behaviour, interpreted incidents through cultural and gendered lenses, and expressed evolving understandings over time. This sub-theme is supported by two categories.

Navigating power relations and organisational management

This sub-theme described how employees at the university are critical of some elements of organisational structures and uncertain about how leadership might respond to SH claims. This uncertainty led to a concern that reported incidents were not treated fairly or consistently. Fears of consequences to speaking up had an impact on decisions to react or report SH, and this was true for both consequences for one’s career and their social situation.

Facing a culture of silence and normalisation

When describing the discussion climate at the university, participants spoke of a culture of silence, as well as a perceived normalisation of SH. Rather than viewing silence as mere passivity, however, participants framed it as a deliberate strategy, one shaped by fear of repercussions, hierarchical dependencies, and the perceived futility of reporting. At the same time, the discussions highlighted how inappropriate conduct becomes routinised through everyday interactions, humour, and institutional complacency. Participants described how behaviours such as sexist jokes, boundary-crossing, and dismissive comments are often downplayed, trivialised, or explained away as cultural, generational, or accidental. These dynamics contribute to an organisational culture where speaking out is discouraged and misconduct is rendered invisible.

Offering peer support for collective action

The final sub-theme captures how participants saw employees pushing back against the structures and silence at the university. Resistance was expressed through solidarity, trust-building, and emotional courage. These acts of care helped some participants transform their work environments, even when formal mechanisms were weak, and led to norms concerning solidarity and collective support.

Negotiating collective norms and meanings

Lund University is a large institution with a complex management structure and a wide array of work environments, each governed by its own set of rules, regulations, and workplace cultures. Despite this diversity, the University applies a single, consistent definition of sexual harassment, drawn directly from the Discrimination Act, to all contexts and units across the organisation [24]. It is in this context that participants in this study struggled to define what constitutes sexual harassment. Throughout the discussions the definitions were constantly changing, and grey zones were identified. Seeing the boundary of what constitutes SH as a negotiated reality, mirrors what classic symbolic interactionist theory (coined by Herbert Blumer in 1937 [27]) terms a negotiated order, an interpersonal process in which actors continuously redefine situations through talk, eye-contact and tacit cues [28]. Understanding that the boundaries of what constitutes SH are negotiated through interpersonal processes has consequences for the design of preventative policies. By moving the locus from individuals to groups, policies must focus on organisational level-responses and support the need for ongoing dialogue and layered training provided to all at the workplace.

This struggle to define SH is not surprising given that research has found that both individual and organisational characteristics, as well as broader societal norms and values, shape how behaviours are interpreted as SH [29]. For instance, research has shown how women identify a wider array of actions as harassing than men do, and perceptions shift depending on the sex of both perpetrator and victim [30,31]. Likewise, Otterbach, Sousa-Poza and Zhang found that even in countries with high gender egalitarianism, men consistently report lower levels of perceived harassment than women, suggesting that cultural endorsement of equality narrows but does not eliminate these gender gaps in recognition and reporting [32].

Perceptions of what constitutes SH are also influenced by the disjuncture that can occur between legal/regulated definitions of SH and everyday experiences. Statutory definitions, often confined to quid pro quo or overt ‘hostile environment’ criteria, routinely overlook other forms of SH, leaving many harmful behaviours outside the scope of what victims, bystanders, and even decision-makers recognise as sanctionable [33]. Our findings mirror this disconnect, highlighting the gap between formal definitions and lived experience.

Cultural differences likewise shape perceptions in ways that can impede a shared understanding. A meta-analysis of U.S. studies conducted by Ilies et al. in 2003 reveals substantial variation in reported harassment rates across industries and demographic groups, suggesting that national norms, workplace cultures and individual backgrounds influence whether similar behaviours are labelled as harassment [34]. In a multicultural institution such as Lund University, these findings underscore the imperative for policies and training programmes that not only reflect legal requirements but also explicitly address cultural and contextual factors.

In the absence of a common understanding, labelling events as SH becomes more complicated, possibly resulting in fewer cases being reported and thus being resolved. This hesitancy to label incidents as SH is echoed in findings of a qualitative interview study conducted across five European countries including Sweden. In this study by Hagerlid et al., three factors were identified that acted as obstacles to identifying SH namely preconceived notions, blurred boundaries, and competing interpretations [35], all findings confirmed in this study.

Many participants explicitly requested a clear definition of sexual harassment during the focus group discussions, underlining the importance of a shared framework for recognition and reporting. This ambiguity, especially around non-physical behaviours, is reflected in a large cross-sectional study among public-sector employees in Denmark that found as many as 75% of ambiguous behaviours were unlabelled as harassment [11]. Moreover, our findings show that participants, particularly women, often ‘tested the water’ with a trusted peer before deciding whether to label an incident as SH, a reliance on bystander validation similarly documented in low- and middle-income countries [36].

The gendered nature interpreting scenarios and deciding whether, or how, to respond has also been measured quantitatively in large US surveys, where male respondents classified fewer acts as harassment than their female counterparts and attributed shared responsibility to the (female) victims [37,38]. These findings are reflected in our results where male respondents tended to question cases of SH more often as well as label more instances as benign. Men’s tendency to underestimate behaviours that amount to sexual harassment, a form of misconduct that disproportionately harms women, is compounded by their persistent overrepresentation in leadership roles. National data from Statistics Sweden show that men occupy a significantly higher share of managerial positions than would be expected from the overall gender distribution in both the public and private sectors, a pattern that has changed little in recent years [39]. At Lund University, the 2019 equality report reveals that women hold just 28% of professorships and 41% of senior lectureships [40]. These entrenched imbalances at the top risk perpetuating the ‘yellow zone’ of ambiguous misconduct, as male-dominated decision-making bodies may be less attuned to recognising, addressing, and formally sanctioning inappropriate behaviour.

The collective nature of establishing boundaries was seen to have been both aided and hampered by the #MeToo campaign. Participants insisted that the movement had in some ways expanded the space for dialogue yet simultaneously ‘muddied’ the scale of severity, producing confusion about what is ‘serious enough’ to report. There is some sociological evidence, primarily from the USA, that the #MeToo era has contributed to more reports of SH by females being taken seriously [41], which would broadly correspond to the findings in this study. At the same time, a large study conducted in Norway and utilising a web-based questionnaire found that negative beliefs about the #MeToo movement, such as fear of false accusations overshadowed the effects of positive beliefs about the outcomes, for both male and female respondents [42].

The fluid, peer-centred boundary-setting we observed, where employees ‘sense-check’ ambiguous incidents with colleagues before deciding whether to label them harassment, echoes a wider pattern in which social groups use shared norms to legitimise or downplay troubling behaviour. This process of collectively negotiated meaning‑making echoes findings from studies from South African Universities, where focus group‑based research among students and staff has shown how everyday verbal and physical violations are frequently framed as ordinary expressions of masculinity or institutional life (“dangerous common sense”), rendering them unremarkable and difficult to contest [43,44]. A parallel can be found in Zietz and Das’s qualitative study of young men in Mumbai, India, who collectively reframed street harassment as harmless ‘fun’ and relied on group validation to neutralise personal misgivings [9]. In both contexts, the interpretive authority to name SH is not located in formal policy but in the micro-politics of the peer group; when that group normalises boundary-crossing, individual discomfort is silenced.

Navigating power relations and organisational management

The image of the university that emerged from our data is one in which formal and informal power has direct and intersecting effects on SH. Formal power structures included the siloed nature of departments and the power invested in individuals. These were further exacerbated by the promotion criteria that looked more at achievement then leadership skills. Informal power was found among those with large research grants or long academic careers, or those who had previously held positions at the university and who had strong networks.

In this context, employees at the university expressed difficulties in navigating behaviours that felt inappropriate but were not clearly sanctionable. These sorts of acts occupy what Taylor [45] calls the ‘yellow zone’, acts that generate discomfort yet remain plausibly deniable and, crucially, rarely trigger formal sanction. The structural logic underpinning this silence is recognised in the harassment literature, with a survey study conducted among university faculty, staff, and students by O’Hare and O’Donohue arguing that organisations often overlook misconduct when perpetrators deliver valued outcomes [46]. The persistence of these ‘yellow zone’ behaviours thus signals not a deficit of rules but a misalignment of incentives, with the same performance metrics that bolster institutional prestige simultaneously deterring colleagues from challenging borderline conduct. Moreover, these seemingly trivial behaviours can nonetheless contribute to significant psychological ill-health and lay the groundwork for more severe harassment. This is illustrated in the meta-analysis by Sojo et al. (2015), that demonstrated how frequent low-intensity gender harassment is associated with increased anxiety, depression, and reduced job satisfaction [47]. Cortina and Areguin further argue that such micro-level slights normalise hostility towards marginalised genders, creating permissive climates that facilitate escalation into overt sexual harassment and broader gender-based violence, thereby underscoring the imperative to counteract these behaviours at their inception [48].

One key consideration concerns manager’s own perception of their remit. Many report feeling powerless to act unless a formal complaint has been filed, yet this belief is not strictly accurate. Under current legislation, organisations are explicitly required to work proactively to prevent harm, rather than merely react once an incident has been reported [20]. In practice, however, managers feel constrained to “react” and insist on a report from the affected individual before taking any steps. This represents a clear mismatch between their mandate to engage in proactive safeguarding, and their experienced responsibility, which they perceive as purely reactive.

Participants’ reluctance to escalate such incidents reflects how they weighed up procedural opacity, career dependency, and reputational stakes. In an interview study conducted among HR managers in Sweden, Cicerali found that the majority of victims they had met preferred quiet mediation over formal investigation [14]. Informal power can also be reflected in gender norms. Gender disparities and imbalances in workforce composition were discussed as a contributing factor for SH in this study, and this aligns with observations from Jordanian campuses that found that distal gender norms that were prevalent in society were able to hollow out ostensibly robust policies through protecting perpetrators, preventing reporting, victim blaming, and discouraging bystander intervention [8].

Facing a culture of silence and normalisation

The findings of this study depict silence not as resignation but as a deliberate risk-management strategy where employees weigh the cost of speaking up against the institutional payoff and often conclude that silence protects careers and collegial ties. In a number of cases participants also expressed how they would consider whether the harassment had been serious enough. In a qualitative study of discursive interpretation of organisational policies around SH, these sorts of binary logics were found to undermine policy implementation and failed to match the lived experiences of employees [13]. Such strategic silence can also be understood through the idea of ‘non-performative’ institutions, i.e., organisations, that claim to care but implement policies that do not address power imbalances [49].

Silence is sustained by a repertoire of rationalisations that circulate through both policy language and everyday exchange. Employees explain boundary-testing behaviour as generational (“he’s old-school”), cultural (“that’s just how it’s done in our field”) or intent-based (“he meant it as a compliment”), echoing Hardt et al.’s [36] cross-national evidence that such justifications blunt the urge to act.

A substantial empirical literature positions organisational climate as the strongest antecedent of SH; Fitzgerald, Drasgow et al.’s [5] integrated model and subsequent meta-analysis [50] demonstrate that perceived tolerance predicts both prevalence and harm. The present study advances that work by illuminating the mechanics through which this climate is reproduced in the university setting. This includes the ways in which performance metrics at the university act as incentives and thus shield high-performing perpetrators from consequences, opaque procedures and hierarchical “gatekeepers” that make reporting a potential risky business, and untrained leaders that default to inaction, together sustaining a ‘yellow zone’ where boundary-pushing conduct is rarely sanctioned.

Organisational research indicates that cultures of silence can be disrupted when harassment prevention is woven into bystander-focused training and linked to clear reward or sanction systems. Experimental and field studies show that giving employees concrete intervention scripts and publicly recognising those who act increases the likelihood that peers challenge boundary-testing behaviour [51,52]. Longitudinal panel evidence collected on corporate affirmative action policies further suggests that norm change is most durable when zero-tolerance expectations are embedded in appraisal and promotion criteria, signalling that leadership will back words with material consequences [53].

Offering peer support for collective action

Participants in the study stressed that informal peer networks often provide the first, and sometimes only, line of defence for individuals who feel harassed. These “back-channel” practices mirror the peer-to-peer safety nets Löfgren et al. document among Swedish students [10], and the mentor shields that Jenner et al. observed in hospital teams [54]. Such micro-acts demonstrate a sense of collective responsibility that operates outside prescribed procedures and can, at least temporarily, reduce victims’ exposure. In the absence of formal systems that are trusted, these informal channels, together with contextual factors in the local environment, can be an important element of prevention and response. These results are echoed by those of Handy’s study of ‘communities of coping’, showing that among women in a small New Zealand town, where informal support, alongside organisational culture and the local environment, was found to be a strategy developed to combat workplace SH [55].

Yet bystander theory suggests that willingness to intervene is heavily contingent on perceived personal risk and situational clarity [51]. Our data confirm those contingencies where employees expressed how they would be more hesitant to respond if the perpetrator outranked them, when norms were ambiguous, or when previous helpers had faced backlash. This aligns with large-scale campus studies showing that fears of social sanction sharply curtail bystander action even when moral motivation is high [52,56].

Strengths and limitations

This study used focus group discussions for data collection, a method well suited to studying norms and attitudes as well as supporting research into sensitive topics [57]. Focus group discussions also allowed the researcher to observe the dialogic process of understanding and navigating workplace norms and organisational culture. This allowed the research team to identify excerpts of conversations to show how the categories and sub-themes are grounded in the data, thus enhancing the credibility of the results. The use of a robust analytical procedure according to the methods described by Graneheim and Lundman [22] and further expanded by Lindgren [58] and keeping a clear documentation of the research process through the use of notebooks and seating plans improved dependability. The emphasis on achieving consensus between all authors of the article contributed towards confirmability. Finally, by providing clear descriptions of the study environment and context the transferability of the results was improved.

In addition, the inclusion of both employees with management/leadership responsibilities and those without (in separate discussion groups) enhanced the authenticity of the findings ensuring that complexity and variation in the points of view were included. This in turn further contributed to the transferability of the results.

The authors of this article are connected to Lund University through employment or studies. This gives them a unique perspective as ‘insiders’. Conducting research as an insider offers both advantages and challenges. On the one hand, insider status can provide contextual understanding and facilitate access [59], but it also raises concerns about role conflict, confidentiality, and potential bias [60]. Reflexivity and inter-author dialogue were essential in navigating power dynamics and ensuring methodological rigour.

Limitations to this study include the sampling procedures that may have led to those most interested in the topic attending, thus missing out on other important perspectives. Bearing in mind the fear of consequences of reporting SH raised in this study, certain employees at the university could have chosen to avoid participating due to the nature of the topic. In addition, some participants (n = 4) did not attend the focus group discussions (FGDs) for which they had registered, primarily due to scheduling difficulties. As a result, several FGDs were conducted with fewer than optimal numbers of participants, and in one case only two participants were present. Conducting FGDs with smaller group sizes may have restricted the range of experiences and perspectives captured within those discussions and may have limited opportunities for variability and contrasting interpretations to emerge through group interaction. Findings from these smaller groups should therefore be interpreted with appropriate caution, particularly with regard to breadth and transferability.

At the same time, methodological accounts in the qualitative literature acknowledge that smaller FGDs can still generate meaningful interaction and data where participants engage actively with one another for example [61]. In our study, the FGD involving two participants generated sustained discussion and mutual interaction rather than parallel individual accounts. Its content was not analysed in isolation but interpreted in relation to patterns observed across multiple FGDs conducted with different group compositions. Excluding these data was therefore considered to risk introducing other analytic and ethical limitations.

The authors also recognise the complex and varied nature of different workplaces at Lund University. To protect the identity of the participants this information was not collected. This could, however, create uncertainty about whose voices are being heard, and how these discussions should be interpreted.

Focus groups can also suffer from social desirability bias and group biases that limit dissenting voices, especially around sensitive topics [62]. Indirect questioning and abstraction of framing, as is commonly found in focus group discussions, however, have been shown to minimise these biases [63].

Implications for practice and further research

This study reveals an articulated need among university employees to develop shared understandings and a common language when it comes to sexual harassment. Employees, both in managerial positions and those without, struggle when talking about a phenomenon where perceptions are influenced by gender, culture, legal definitions and personal experience. To contribute to a shared understanding of what can be tolerated and not, a university wide code of conduct, focusing not specifically on sexual harassment, but more broadly on respectful behaviour and highlighting zero-tolerance of any uncivil or harassing behaviour, might be useful. Sustained dialogue is also needed to translate abstract principles into local practice. Regular discussions and joint reflection sessions could help to make visible hidden norms and help align employees around a common threshold of acceptable behaviour. Without this iterative work, even the best-written code of conduct or policy, risks remaining a document rather than a living standard.

Swedish employers have a proactive legal duty under both the Discrimination Act [17] and the Work Environment Act [19] to prevent harassment and investigate any signs of it. This study adds to a growing body of evidence demonstrating how workplace climate and collective norms can either enable or restrain harassing behaviours [50]. Our findings suggest that universities should move beyond interventions aimed solely at individual victims or perpetrators, but instead prioritise preventative strategies that shape the overall work-environment climate. In practice, this means embedding collective responsibility for fostering a culture in which harassment is neither tolerated nor ignored and equipping all staff with bystander skills so they can recognise and intervene in inappropriate behaviour as it occurs [64].

At the same time, institutions should streamline and demystify formal reporting channels, which participants in our study often perceived as opaque and driven more by reputational concerns than by a genuine commitment to justice. Crucially, organisations cannot wait for formal complaints to drive change, since many employees struggle to recognise and label harassing behaviours. Thus, prevention must focus on improving the social climate itself. Indeed, evidence suggests that enhancing overall workplace climate is among the most effective means of reducing the incidence of sexual harassment [65]. Not only does this align with best practice, but it also fulfils Swedish legal requirements for systematic work-environment management.

Another key finding from the results was that employees in managerial positions felt that they lacked the necessary training, while also being torn between collegial loyalty and legal responsibility, amplifying the so-called ‘yellow zone’ vacuum where ambiguous behaviour persists. This feeling was shared by employees not in managerial positions who saw a vacuum in leadership skills. Targeted training should combine emotion-focused skills such as how to respectfully receive disclosures without defensiveness, with clear guidelines for how any investigation should occur. This could also be achieved through external ‘experts’ who could be brought in to support managers in this task.

The results of this study contribute to a deeper understanding of the perceptions of culture and norms surrounding sexual harassment among employees in an academic workplace in Sweden, contexts in which SH occurs, and in which these behaviours are, or are not, reported. The results also highlight a need for further research into how best to create a workplace culture where these behaviours are not accepted as well as how to challenge the culture of silence. This could be, for example, an evaluation of interventions to improve bystander support, longitudinal studies to examine the impact of providing additional support to managers on workplace norms and reporting, and studies that explore the impact of interventions specifically targeted at workplace climate.