He hadn’t taken his medication in weeks—didn’t need it. His clarity and focus over the past couple of months were sharper than they had ever been. As he stood in the hallway preparing to enter the campus library through a side door, his sweating hands firmly clutched the grips of the twin Glock 22 pistols he had ordered online.
If only there had been a way to look into a crystal ball and see that this horrific confrontation was about to occur, it could have been prevented. In the aftermath of nearly every large-scale act of campus violence in the United States, investigation has revealed that early-warning signs had been present but not recognized or acted upon. As a response, nearly all college and university campuses have developed threat-assessment teams, whereby key members of various campus groups come together regularly to share information and discuss troubling student behavior.
Unfortunately, these teams are only able to assess students whose issues have already led to problematic behaviors noticed by people on campus. Existing technology, though, offers universities an opportunity to gaze into their own crystal balls in an effort to prevent large-scale acts of violence on campus. To that end, universities must be prepared to use data mining to identify and mitigate the potential for tragedy.
Many campuses across the country and most in California provide each student with an e-mail address, personal access to the university’s network, free use of campus computers, and wired and wireless Internet access for their Web-connected devices. Students use these campus resources for conducting research, communicating with others, and for other personal activities on the Internet, including social networking. University officials could potentially mine data from their students and analyze them, since the data are already under their control. The analysis could then be screened to predict behavior to identify when a student’s online activities tend to indicate a threat to the campus.
If university officials were to learn that a student had conducted extensive online research about the personal life and daily activities of a particular faculty member, posted angry and threatening comments on his Facebook wall about that professor, shopped online for high-powered firearms and ammunition, and saved a draft version of a suicide note on his personal network drive, would those officials want to have a conversation with that student, even though he hadn’t engaged in any significant outward behavior? Certainly.
This information, which may reside in the university’s IT system, would allow the campus to strategize a swift and effective intervention, and take steps to prevent violent behavior from ever occurring. In such cases, an important distinction would have to be made between violations of the law and violations of campus policy, and established guidelines would have to be followed to ensure the student’s rights to due process.
Interestingly, the technology exists to allow university officials to take such actions. Data mining involves applying specifically designed algorithms to electronic data to identify patterns and transform the data into usable information. It is a form of behavioral surveillance, and it can be used to predict, with amazing accuracy, the propensity for a person’s future behavior. Computer engineers design data-mining algorithms to search for specific patterns that, when analyzed collectively, tend to indicate the likelihood of a particular outcome.
Have you ever had a credit-card transaction declined because the bank noticed an unusual pattern of spending on your account? Through data mining, the bank drew the conclusion that your credit card had been stolen. It would logically follow that mining algorithms could be easily designed to predict the potential for planned or considered campus violence as well.
Although university administrators may resist the idea of passive behavioral surveillance of the campus community because of privacy considerations, the truth is that society has been systematically forfeiting its rights to online privacy over the past several years through the continued and increased use of services on the Internet. Social-networking sites and search engines store and divulge personal information accessible to the world each day, yet people continue to use them in increasing numbers.
Indeed, our online activities are already under constant surveillance—by companies eager to learn about our individual interests. Their findings are used for marketing purposes to single out consumers with goods and services based on our specific interests. How does Amazon.com know what types of books I’m interested in reading? How did my Gmail account find out I’m an Oakland Raiders fan? These examples are instances of data mining.
Some people might think this concept will face harsh challenges on the basis of the Family Educational Rights and Privacy Act (Ferpa), which prohibits the release of any information from a student’s education record without written permission from the student. Following the killings at Virginia Tech in April of 2007, though, universities sought clarification from the Department of Education regarding exceptions to the Ferpa regulations. That shooter had demonstrated mental instability to a number of faculty members on the campus, but information was not shared effectively because of potential Ferpa implications. To resolve this, the department revised and clarified acceptable exceptions to Ferpa requirements.
Under the new regulations, campuses may disclose information from a student’s record, without consent, “to appropriate parties in connection with an emergency if knowledge of the information is necessary to protect the health and safety of the student or other individuals.” Ferpa also allows for “information concerning disciplinary action taken against a student for conduct that posed a significant risk to the safety or well-being of that student, other students, or other members of the school community” to be included in the student’s educational record and to be disclosed to teachers and school officials inside and outside the institution. Enhancing the capacity of threat-assessment teams through data mining is the next natural step using what would ordinarily be private information to prevent on-campus violence.
Because campuses can be prime targets for large-scale acts of violence, have their own full-service computer networks, and operate comprehensive threat-assessment teams, the use of data-mining technology to prevent violence should begin there. Certainly, no single crime-prevention program can be 100-percent effective, 100 percent of the time. But if colleges use the crystal ball that’s available to them, they will surely come much closer to that goal.