Every year people cross over the U.S. border illegally and are not detected by American surveillance systems. As a result, some are able to successfully bring illegal drugs, people, or counterfeit goods into the U.S. As Professor Emeritus Paul Kantor also explains, “A particularly big worry is that some of the people might be terrorists who cannot come in on commercial planes or ships.”

Kantor is part of a Rutgers team of three researchers who are working to develop novel methods the U.S. Department of Homeland Security (DHS) can use to estimate how many people and goods are “missing” at the border.

The project is titled “Missed Detections:  From Data to Actionable Estimates” and is funded by the Borders, Trade and Immigration Institute, at the University of Houston, established as a DHS Center of Excellence by the Department of Homeland Security Office of Science and Technology (DHS-S&T).

Kantor, who had “retired” from SC&I, and moved to Wisconsin, decided to join the team, because, he explained, “The truth is that we wrote the proposal when I was on my way into retirement. So you might call it a ‘planned accident.’  I have always been interested in problems where some rigorous mathematics has to be combined with methods of social science, to ensure that the solution matches to the real problem.”

For this new project for the DHS-S&T, Kantor will be collaborating with Research Professor Dennis Egan (who is heading the project) and Distinguished Professor of Mathematics Fred Roberts. In addition, Kantor anticipates that they will work closely with analysts at the DHS, and he also hopes to find a SC&I doctoral student who can work with them. Kantor has already begun working with SC&I colleagues to begin the search. Kantor said the project will span approximately three years and will take place in New Jersey, Washington, and Wisconsin, to start.

Explaining the nature and goals of the project, Kantor said, “Rutgers University will extend formal tools of data science, adapting techniques from ecology and operations research. The resulting insights into missed detections will help Customs and Border Protection (CBP) and other agencies assess and document performance; get early warning of change; assess trends in a timely fashion, and understand the effect of specific resource allocations on deterrence and detection.

“Simple capture-recapture models have already been applied in this context, and underlie some Border Patrol metrics. The currently employed ‘naïve’ Capture-Recapture models assume that all apprehended persons try again, and are as likely to be caught as any other person. This project will develop more sophisticated Extended Capture-Recapture (ECR) models of this complex process, define the data needed to apply it, and validate it with both simulated and actual data.

“The DEA method arose in governmental and non-profit settings where multiple ‘Decision-Making Units’ (DMUs) deal with similar problems. DEA recognizes that every border station or sector differs from the others. Using mathematical techniques including Linear Programming, DEA provides a principled way to assess resources and their relations to impacts. The management benefit of DEA is that each unit can be considered in its own context and with its own specific mix of resources and impacts.  The engineering and scientific challenges lie in adapting models to the peculiarities of the border security problem, and in dealing with practical operational limits on the data that are, or can be, available to decision-makers.”

“The Rutgers research will have to use synthetic data on interdictions, recidivism, and on effort, by station,” as Kantor explained, because “DHS may not be able to release real data because of classification or personal data, so we expect that they will be making up data that reflects the situations at the border, but does not reveal any sensitive information. We will be working with them on that effort, during the first year of the project.”

Explaining examples of a few of the techniques they will use, adapted from the fields of Ecology and Operations Research, Kantor said, “In Ecology, there has always been the problem of estimating how many bears there are in a forest.  And there’s a long-time solution, which is to band or mark captured bears and let them go again. After a while, they are supposed to be mixed in with all the other bears. Then, if, say one out of every six captures is a banded bear, we estimate that the total number of bears is six times the number that have bands.  Of course, the number is more like one out of a hundred, so the statisticians have lots of tricks for setting ‘confidence limits.’  As you might guess, this does not work so well for people. The ones who have been caught crossing the border don’t randomly mix in with the newcomers. Some of them go home. And they learn from their experiences, so they are not likely to try crossing in the same way, or the same place, next time.

“From Operations Research we can look at this question: if there are several missions to be accomplished, and you are scoring some particular percentage on each one of them, what is your overall score?  Of course that depends on how important each mission is.  But it would not make sense to simply pick the mission where you have the best score, and announce that it is the only one that matters. Operations Research lets us figure out ‘the best score you might have, while being reasonable about the relative importance of the various missions.’ This research will help the DHS make that kind of analysis to understand the impact of various resources, on the complex realities at the border.”

Kantor explained in more detail how ECR will work. “This is research, so we don’t exactly know the answers already. But, going back to our bears, we’ll be looking for more realistic models of how real people behave, depending on who they are, and what happened to them when they were caught. We can call that ‘passive.’  We might also find some data on how that behavior of ‘trying to cross the border illegally’ is affected by changes in policy. We can call that ‘active.’ Combining observations with mathematical models we may be able to estimate what the total flow is, from observations with several levels of effort.” 

In terms of how the team will employ optimization techniques of Data Envelopment Analysis (DEA) Kantor said, “This is the stuff I mentioned about assigning relative importance to the several missions. In DEA, they do the same thing also for the resources that are used. Then the information can be combined to come up with suggestions about ‘which policies are more efficient.’  This has been applied to libraries (that was work with one of our star SC&I students, Wonsik ‘Jeff’ Shim, who is now a professor in South Korea); and it has been applied to public schools. But we will have to find out how useful it can be for securing the border.”

In terms of managing the operational limits on the available data, Kantor said, “In real problems, almost none of the data is at hand. And the people protecting the border are too busy to spend time being interviewed by us, or filling out a lot of forms. So we will work with DHS and people at headquarters, who already have a great deal of information. And if our methods are making sense, and providing valuable insights, they will work with us to get additional data from the border.”

The three members of the research team will apply their particular areas of expertise to the project.  As Kantor explained, Dennis Egan “is a psychologist who worked for many years at Bell Laboratories. In addition, he will head up those aspects dealing with perception of importance and impact, as judged by various stakeholders. He will be concentrating on what kinds of data we need, and how to get information on subtle questions like the relative importance of missions.

“Co-PI Fred Roberts, who was a long-time head of the center for Communication Control and Advanced Data Analysis (CCICADA), established as a Center of Excellence by DHS-S&T, will help look for mathematical ways to model the missed detections. He has done a lot of work on mathematical models for the social sciences, and he will be contributing to the development of the ECR models and making sure that they are mathematically sound.

“I have worked in both areas, and will be working on developing models that can work with the data that are available. This mixing of qualitative and quantitative research is exactly the kind of thing that I have been doing for decades.  I like to think of it as ‘qualantative’ research.”

This is not the first time the three have collaborated, Kantor said. “We have done numerous projects together, most notably a series of three major projects on finding Best Practices to protect stadiums and other large venues against the threat of terrorist attacks.  Prior to that Fred and I worked together on other data-oriented projects related to countering terrorism, for the Intelligence Community, and later the Department of Homeland Security. That earlier work resulted in a joint publication that we are not allowed to read because it is classified.”

This project, however, will not be classified, according to Kantor, because, “I think that our contract requires that we give the BTI and DHS a chance to review things before publication. But under the DHS rules, we cannot do any classified work at or for a University Center of Excellence. So we don’t expect that anything we do will become classified.”

Asked about the greatest challenges involved in this research, Kantor said, “I suspect that integrating the whole thing, and making the social science and the mathematical science hang together to produce some useful results will be the biggest challenge. As you know, any issue related to the border generates a lot of emotion these days. So we don’t know.  I will say that if any students want to protest against our trying to help the government in this way, they are totally welcome to visit me in the land of Lake Mendota.”

“There are two horses to this project,” Kantor said, “and I guess I am standing with one foot on each of them. My years at SC&I, where qualitative and quantitative methods are often yoked together, provide an excellent platform, and I hope that these two horses will keep running side by side, for a successful conclusion.”