Combating tolerance for sexual harassment with information: Evidence from a field experiment

Sexual harassment remains widespread, and women face higher exposure from colleagues and managers in male-dominated work environments. Can simple information change workplace culture and reduce harassment? This column presents evidence from a randomized field experiment in the Norwegian military. The results show that information that corrects misperceptions can substantially reduce tolerance for sexual harassment. The results also offer guidance for future experiments by identifying an important measurement challenge for detecting behavioral change.

Sexual harassment carries significant costs for individuals and organizations, including poorer health, higher turnover, and higher levels of economic gender inequality (e.g., Folke and Rickne 2022). While prevention is urgent, the most common prevention methods face substantial critique. Employee training can lead to resistance or backlash from potential harassers, and reporting systems fail when fear of retaliation holds victims and witnesses back from making reports. These problems have led experts to call for research on organization-level solutions that prevent sexual harassment by reshaping workplace contexts and cultures (Cortina and Areguin 2021).

Targeting misperceptions

A growing literature shows that misperceptions of others’ attitudes can sustain gender inequality (Bursztyn et al. 2020). In the context of sexual harassment, individuals may underestimate how negatively others view these behaviors or hold inaccurate beliefs about women’s competence. These misperceptions can contribute to environments in which harassment persists. Our newly published research article (Folke et al. 2026) designs an information intervention to correct misperceptions related to common sexual harassment behaviors and, in turn, reduce their prevalence.

A field experiment in the Norwegian military

We implemented a randomized field experiment among recruits in the Norwegian military during an eight-week boot camp. The intervention was embedded in a standard enrolment survey and delivered in a low-salience way to minimize backlash. It was randomized across rooms in which recruits live and collaborate for the duration of the boot camp.

The treatment provided two pieces of information based on survey data from previous cohorts. This prior survey data identified the two most common forms of harassment in our setting, crude sexual jokes and negative comments about women’s competence. We designed the information intervention to directly target misperceptions related to these behaviors. Treated work groups were informed that a majority of their peers consider sexualized jokes to constitute harassment, and that women perform equally well as men on military performance tests. Both pieces of information were based on actual data.

The randomization at the group level allowed us to compare treated and control groups immediately after the randomized treatment, but before they met for the first time (baseline), as well as at the end of the eight-week boot camp (endline). We measure both attitudes towards harassment and self-reported experiences during the training period.

Large and persistent effects on attitudes

The intervention produced substantial changes in attitudes. Treated recruits became significantly less tolerant of sexual harassment immediately after receiving the information, and about half of this effect remained eight weeks later.

Figure 1 shows the impact on an index of tolerant attitudes. The immediate effect at baseline is large, and while it declines over time, it remains statistically and economically meaningful. These effects are sizeable relative to existing evidence on workplace training programmes, which often show limited or short-lived impacts (Roehling et al. 2022). We find no evidence of backlash among individuals with initially high tolerance levels.

Figure 1. Effects of the information intervention on tolerance of sexual harassment (baseline and endline)

Source: Figure 1 from  Folke et al. 2026

Behavioral effects and a measurement challenge

Turning to behavior, the results are less conclusive. The intervention’s impact on sexual harassment prevalence is directionally negative but statistically insignificant (see Figure 2).

The results reveal an important challenge for field experiments seeking to evaluate prevention methods. Any intervention that increases awareness of sexual harassment may struggle to detect a reduction in exposure. Increased awareness makes people more likely to notice and report borderline behaviors in the endline survey, which raises self-reported harassment in the treatment group and offsets a potential negative treatment effect on exposure. In our case, results from participants’ evaluations of a vignette about sexual harassment show clear shifts in their interpretations and evaluations of the same situation. Our intervention increased the likelihood of recognizing problematic behaviors and assigning responsibility to the perpetrator. A likely implication is that this increased awareness also led participants to recall and report more incidents in the endline survey, masking real reductions in harassment.

Figure 2. Effects of the information intervention on self-reported harassment prevalence.

Source: Figure 2 from  Folke et al. 2026

Implications for policy and research

The paper suggests that simple, low-cost interventions can shift workplace norms in meaningful ways. Policies that correct misperceptions about peer attitudes and competence may complement or outperform traditional training programs.

At the same time, the paper provides important takeaways for future evaluation research. Randomized controlled trials that evaluate prevention interventions for sexual harassment are extremely rare (see Sharma 2024 for an exception). Future evaluations should design the experiments to account for the fact that the intervention may affect awareness and, in turn, bias the treatment effect on prevalence downward. Power calculations in pre-analysis plans may account for this downward bias. To correctly capture behavioral changes, designs should ideally supplement data on self-reports of sexual harassment with objective measurements from other sources. Modern methods for analyzing data from text or video may offer new metrics and more reliable results.

Our research shows that changing what people believe others think may be a powerful lever for changing behavior. In settings where misperceptions sustain harmful practices, information can be an effective policy tool to reduce the tolerance of sexual harassment in work groups. Future field experiments should test more interventions while being mindful of the important empirical hurdle revealed by the present research.

References

• Bursztyn, L., González, A. L., & Yanagizawa-Drott, D. (2020). Misperceived social norms: Women working outside the home in Saudi Arabia. American Economic Review, 110(10), 2997-3029.
• Cortina, L. M., & Areguin, M. A. (2021). Putting people down and pushing them out: Sexual harassment in the workplace. Annual Review of Organizational Psychology and Organizational Behavior, 8(1), 285-309.
• Folke, O., & Rickne, J. (2022). Sexual harassment and gender inequality in the labor market. The Quarterly Journal of Economics, 137(4), 2163-2212.
• Folke, Olle, Hanson, Torbjørn, Johnsen, Åshild A., Kosadam, Andreas, and Johanna Rickne (2026). Targeting Attitudes to Combat Sexual Harassment: A Randomized Intervention in the Norwegian Military. Journal of Economic Behavior & Organization,
• Roehling, M. V., Wu, D., Dulebohn, J., & Choi, M. G. (2019, July). The Effect of Sexual Harassment Training on Knowledge, Skill, and Attitudes: A Meta-Analysis. In Academy of Management Proceedings (Vol. 2019, No. 1, p. 19436). Briarcliff Manor, NY 10510: Academy of Management.
• Sharma, K. (2024). Tackling Sexual Harassment: Short and Long-Run Experimental Evidence from India. Available at SSRN 6460764.

Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.