A Safer Monitoring Tool to Help Workers Report Harassment in Organizations
- Firms
- Workers
- Empowerment
- Violence
- Mental health
- Worker satisfaction
- Behavioral economics
- Information
- Monitoring
Measuring the extent of harassment in the workplace is challenging, as reporting is often associated with fear of retaliation and stigma. Researchers conducted a randomized evaluation to test how different ways of asking questions affect workers’ likelihood of reporting harassment at a large apparel manufacturer in Bangladesh. A survey technique that gave people an option to deny they had reported issues so that they could give answers more safely led employees to report more abuses. Using the new survey data, researchers were able to estimate the extent of harassment at the firm more accurately.
Policy issue
Harassment at work is a concern in many industries and workplaces. Firms and employers struggle to tackle this problem in part because victims may choose not to report abuses due to fear of retaliation or damage to their reputation. Traditional surveys that directly ask people to report their experiences of harassment often capture fewer reports than what actually occur, given how sensitive the topic is. Increasing the security of organizations’ approaches to asking workers about their experiences with harassment is one potential strategy to get more accurate information on harassment in a workplace. However, there is limited direct evidence testing this approach or showing how it changes what policymakers learn from reporting data.
Context of the evaluation
It is estimated that high proportions of women in both low- and middle-income countries and high-income countries experience sexual harassment at workplaces. In Bangladesh’s garment industry, harassment is also a persistent issue. Earlier studies suggest that low wages, limited access to legal support, and traditional gender norms give managers considerable power to harass workers. Although people believe harassment to be widespread in this industry, workers rarely report it.
Researchers partnered with an anonymous large garment manufacturer in Bangladesh that employed more than 25,000 workers to conduct a phone survey aimed at obtaining more reliable data on workplace harassment. The firm’s senior leaders sought to better understand workers’ experience in the workplace and improve their well-being. The senior leaders also wanted to use this data to inform the firm’s human resources policies.
In apparel production, the firm often organizes workers according to line teams. Line teams are likely to be either predominantly women or men, depending on their function – for example, only men typically work in cutting sections, while mostly women work in sewing and finishing. By contrast, more than 90 percent of managers are men.
Details of the intervention
Researchers collaborated with the firm’s human resources department to conduct a randomized evaluation on the impact of different survey methods on how workers answered questions about experiencing harassment during work. Using phone surveys, they randomly selected 2,245 workers into 9 groups that each received different combinations of the following techniques aimed at encouraging people to give accurate answers on sensitive topics:
- Hard garbling (HG): HG is a technique that provides people with plausible deniability in surveys by automatically marking some responses as complaints. The interviewer told workers assigned to this version of the survey that one in five answers asking about harassment would be automatically recorded as "yes". This gave people an option to deny reporting issues and made it safer for people to report without worrying about being blamed later.
- Rapport building (RB): In this version of the survey, the interviewer spent time chatting with the worker about life outside of work to create a sense of trust and emotional connection. The aim was to help workers feel at ease to answer sensitive questions.
- Low personally identifiable information (Low PII): In this version, interviewers did not ask workers about information that someone who saw the survey could use to identify the worker. In particular, interviewers did not ask workers the name of their direct supervisor or which production team they worked on.
- Comparison group: In this version, workers received a business-as-usual survey—interviewers asked questions about harassment directly, did not apply extra effort to build trust, and gathered personally identifying information.
Interviewers asked workers about workers’ experience with three types of harassment in the past year—physical harassment, sexual harassment, and threatening behavior—by their direct supervisor. They also posed questions on worker well-being like mental health and job satisfaction.
The researchers were also interested in how improved reporting data could be used to answer policy-relevant questions about the nature of harassment in the firm, such as: How prevalent is harassment? What share of employees are responsible for the damage? How isolated are victims within teams?
All respondents were assured of a strong commitment to confidentiality as part of the process for consenting to participate in the study. To design the survey questions for the context, the research team conducted focus group discussions, interviews, and pilot surveys with garment workers before the main surveys.
Results and policy lessons
Making it safer for workers to report if they experienced harassment by using the hard garbling (HG) method, which automatically records a subset of cases as harassment, gave workers an option to deny having reported. This increased the estimated rate of harassment reports from both women and men, and across different types of workplace harassment.
Harassment reporting: In the comparison group using the business-as-usual survey method, 9.9 percent of workers reported experiencing threatening behavior, 1.53 percent reported physical harassment, and 1.78 percent reported being sexually harassed by their supervisor. When interviewers used the HG survey method, workers were 4.5 percentage points more likely to report threatening behavior, 4.4 percentage points more likely to report physical harassment, and 4.8 percentage points more likely to report sexual harassment. Both women and men reported more experiences of harassment, although the effect was larger among men.
Researchers found that the other survey tactics of building rapport with workers or removing identifying information did not have clear effects on helping workers report more cases across all different types of harassment.
Harassment statistics for informing policy decisions: To help the anonymous firm use data to answer policy-relevant questions, researchers used the results of the HG surveys to more accurately estimate the nature of harassment in these factories. They measured how common harassment was, how many managers were involved, and whether a single manager was more likely to harass one or several team members.
Using the new survey data, researchers estimated that 13.5 percent of workers experienced threatening behavior, 5.7 percent physical harassment, and 7.7 percent sexual harassment by their managers.
At the team level (there were seven workers per production team on average), harassment is widespread: 72 percent of teams had at least one worker experiencing threatening behavior, 40 percent had at least one who had suffered sexual harassment, and 25 percent had at least one who had been subject to physical harassment. However, victims were often the only person on their team to experience (or report) the harassment.
Overall, the research showed that workers were more likely to report sensitive harassment experiences when the interview method allowed them to deny that they had reported harassment. The findings also showed that harassment in this context was widespread, that many managers were responsible, and that victims were often isolated.
The researchers used the results to inform the senior management of the firm about the extent and nature of harassment in the firm, suggesting viable policy options to address the problem. Researchers suggested that for this firm, a policy of firing all offending managers would be very costly, and the firm may find it less costly to focus on changing the behavior of existing managers and changes in company norms.