NISS hosted another Ingram Olkin Forum on Wednesday, March 15, 2023. This session was on "Statistical Methods for Combatting Human Trafficking: Estimating Prevalence Methods." The IOF Committee liaison leading the organization of this forum was David Banks, Duke University, who also moderated the event.
At the start of the forum, Jim Rosenberger, Director of NISS, provided the overview of the IOF S3 series. Since 2018, NISS formed the IOF committee and named this series “Statistics Serving Society,” attributed to Ingram Olkin, who had a passion for both the field of statistics and for the impact of statistics on society. Olkin believed that statistics are beautiful, powerful and that statistics can make a unique contribution in service to society. Ingram was one of the inspirations for the founding of NISS more than 30 years ago. The focus of the forum is to address a societal issue that might benefit from attention from the statistics community, and that's certainly what we aimed to do during this forum.
The forum’s goals are to bring innovations and statistics to new research, accelerate development of innovative approaches, develop statistical action items to inform public policy and generate reliable evidence, and facilitate new collaborations between statisticians and stakeholders. During this forum we hope we achieved these goals in our breakout session, and plan to continue developing working groups after the webinar. You can see past forums and their outcomes for other examples on how we did this here: https://www.niss.org/ingram-olkin-forums. Shortly after COVID struck, the pandemic changed the world and we’ve had several virtual events. Our last IOF on Advancing Demographic Equity with Privacy Preserving Methodologies was just held in person this past January, so we plan to get back to hosting in-person events. Keep an eye on our events page for the upcoming event on Police Use of Force which will be another in-person workshop this fall. We are hoping to plan future Ingram Olkin Forums on topics including the opioid crisis, and systematic reviews.
This was the opening forum on Statistical Methods for Combatting Human Trafficking, which we hope to have an in-person follow up workshop after what we learned during this session. The organizing committee for this forum included David Banks, (Duke University), Daniel Manrique-Vallier, (Indiana University), and Megan Price (Human Rights Data Analysis Group), and they each attended the forum. The motivation of the forum came from a National Academies panel that was held in early 2019. The purpose of this panel was to examine statistical methodology for estimating the prevalence of human trafficking. Our general hope is that this is yet another steppingstone along the way to better interaction between statisticians and the domain scientists who study human trafficking issues, and to frame some of those issues.
We were very fortunate that Margaret Henderson, a faculty member from the University of North Carolina Chapel Hill, and Nancy Hagan who is on the North Carolina Human Trafficking Commission to provide a thorough overview on the overall definition of human trafficking. Margaret explains that there is a broad range of understanding about what human trafficking is and identifies some of the myths and misinformation during her talk. Nancy adds that she and Margaret try to reach across the sometimes ‘great divide’ between academia and stakeholders, advocates and populations studied and served, and they do this across the state of North Carolina and to some degree at the national level. Nancy and Margaret raised a number of issues and concerns regarding current and ongoing research, including the applications, with the intent of again bringing us together to have a shared vocabulary set of contracts, constructs and understanding so that we can move forward.
They share with us one of the important goals in this area is to always center the survivors as the folks who have the lived experience of trafficking. Margaret and Nancy also outline the forum’s learning goals: to define what sex and labor trafficking are, recognizing that different countries have different definitions in place, the typology (how it shows up), what the business models are, the indicators of trafficking and the response. They continued to identify challenges to research designs for data collection and analysis, where participants of this forum went further into that in breakout group sessions later on. One of the main problems is that traffickers are highly adaptive. They unfortunately are able to move from multiple communities quickly, as well as change their business model to be able to take advantage of new markets or new opportunities to advertise.
“The question any community should be asking is not ‘does trafficking happen here,’ but ‘how does it happen here?’ It happens anywhere there are vulnerable people, and it really doesn't matter what the nature of the vulnerability is. Traffickers are predators and they look for people with vulnerabilities that can be exploited. They move quickly, and they can evolve faster than community response or interventions can evolve or that laws can evolve.”
- Margaret Henderson
Nancy explains the legal definition of trafficking and that the international perspective differentiates from the United states, and is quite distinct. This really impacts of course what we see and define as human trafficking how we study it how we determine policy and our response as well.
After the overview from Margaret and Nancy, Tyler McCormick (University of Washington) continues this session with his presentation on “Respondent-Driven Sampling.” Tyler describes that it’s a relatively newer technique in this particular area, but it’s something that’s been used in other contexts with populations that are difficult to access with traditional surveys for some time now. There’s a lot to do in terms of understanding the potential utility here, which he goes into during his talk. Tyler explains the basic plan for respondent driven sampling at a high level, to give a sense of familiarity with the sampling technique, how it works, and then how it might work in the context of trafficking. During this talk, Tyler gives the audience a sense of when this might be a reasonable technique to use. There are still a lot of a lot of directions to go and there is still the potential for a lot more thinking to be done in this area. Respond driven sampling starts with a convenience sample of people in the target group, then encourages group members to recruit others.
“This is a big advantage if you have individuals that interacted with a previous intervention or if they have an ongoing relationship with a public service agency; then those people won’t be representative of the population overall, but are relatively easier to access.”
- Tyler McCormick
It also requires participants to be socially connected. A key feature of response driven sampling is that individuals in the target population are recruiting other individuals in the target population target group and it also means that people need to have some autonomy over their mobility and have the ability to move and interact with other individuals in this group of interests. It provides access to members of the target group directly and is frequently used in studies of sex and labor trafficking along with related groups (i.e., sex workers and documented migrant workers).
In the many challenges that Tyler talked about, he poses the forum with three questions to think about as we address them:
1. How do we actually think about what we do with the estimators once we have them?
2. How do we communicate uncertainty effectively given that this is a highly complicated process (and highly complicated estimators)?
3. How do we foster engagement with the idea that more sophisticated methods are useful in some contexts, but can also exacerbate communication talents?
Rowland Seymour, (University of Birmingham) and Bernard Silverman, (Oxford University) present their talk next on "Statistical issues in the study of human trafficking and modern slavery." Rowland and Seymour provided a range of different methods and ideas that they’ve been thinking about in the fight against modern slavery. Rowland spoke about ‘Network Scale Up Method,’ comparative judgment, and also some work they’ve done together on using looking at IP addresses for online exploitation. Bernard discusses how statistics could contribute to public policy, even if they’re known to be very inaccurate.
Bernard and Rowland discussed several case studies that demonstrate a range of statistical thinking relevant to the fight against modern slavery and human trafficking and answered four main areas in their presentation:
1. How can the network scale up project be used to assess the extent of child sexual exploitation?
2. How can a comparative judgment framework be used to map the risk of modern slavery?
3. How can building a model help analyze the use of different IP addresses for online exploitation?
4. How can statistics, even if known to be very inaccurate, contribute to public policy?
Bernard welcomes anyone who wants to see the technical details which were not covered during this forum to reach out and will give you many references.
After all of the presentations, the forum had a speaker panel discussion where they answered some questions from the audience. Our participants then joined Zoom Breakout Group Discussions where each individual breakout room had two volunteers: one facilitator and one notetaker. All groups were tasked to answer three main questions. Each group discussed which methods they thought would work best for forced labor, child labor, sex trafficking, and so forth; they discussed whether they could think of other strategies for estimating prevalence; and then they identified their group’s list of statistical action items that are needed to better inform public policy, generate reliable evidence, and stimulate research that may mitigate the problem.
After the breakout group discussions were over, the volunteers sent their filled-out document to the IOF organizers. The forum then closed with providing synthesis and future steps from this forum.
NISS would like to thank all our speakers and participants for coming to this webinar and participating in the breakout sessions providing helpful feedback on this session. We have the full recording of this session for you to view on our NISS Communications YouTube Channel below. This video is saved within the Ingram Olkin Forums playlist which you can find here: https://www.youtube.com/watch?v=C3bcgYoxj9E