A Lone Gunman? Using Statistics in Forensics
There were five bullet fragments groups in President Kennedy’s assassination, but if those fragments came from more than two bullets, it would have been very difficult to conclude that Lee Harvey Oswald was the only shooter.
In 1978, however, radiochemist Vincent P. Guinn testified before Congress that the composition of each of the five bullet fragments showed that they came from two—and only two—bullets. “There is no evidence for three bullets, four bullets, or anything more than two, but there is clear evidence for two,” Guinn said (8 September 1978 hearings before the Committee on Assassinations).
The problem with Guinn’s conclusion, however, is that he relied only on chemical analysis.
“Bias is a big problem in forensic science,” says Clifford Spiegelman of Texas A&M University, because forensic scientists are often tasked to look for verification of what police officers or federal investigators already suspect.
Indeed, if the Federal Bureau of Investigations had worked with statisticians from the start of their investigation into President Kennedy’s assassination, they might have concluded early on that the five bullet fragments groups allow for up to five bullets, says Spiegelman. As evidence for that claim, Spiegelman’s team purchased old WCC Mannlicher-Carcano bullets—the type associated with President Kennedy’s assassination—and used the same comparative bullet lead analysis as Guinn. One of the 30 bullets Spiegelman’s team analyzed matched an assassination bullet, chemically.
“So according to Dr. Guinn's testimony that the bullets used in the murder were chemically unique,” says Spiegelman, “we must have been involved in the Kennedy assassination since we have fragments that match.”
Further statistical analyses helped Spiegelman and his colleagues show that a bullet’s composition is not at all unique, making it impossible for Guinn to conclude—as he did—that two of the five bullet fragments came from one bullet and the remaining three came from a second bullet.
Decades after President Kennedy’s assassination, the FBI still didn’t consult with statisticians for forensic cases. And it took until 2005 before the FBI abandoned comparative bullet lead analysis. Even then, the agency announced it “still firmly supports the scientific foundation of bullet lead analysis” and only stopped the practice because of cost and its “relative probative value” compared to other evidence.
Today, though, crime labs across the country and around the world are starting to value working with statisticians and scientists in multiple disciplines in order to improve forensics practices. Working with the United States Army and crime labs in Raleigh and Houston, Spiegelman says, “We’re developing some very successful test cases to share with the forensic science community on how information should flow to forensic scientists and how statistical procedures should be used” in order to prevent bias, avoid convicting innocent people, and, instead, continue the search as quickly as possible for what really happened.
This post is published in From the Staff
Connect With Us:
Aug 29, 2015
The 10th anniversary of Hurricane Katrina brings with it memories of a coincident crisis. But it turns out that recalling an emotionally charged event isn't a straightforward process. How documentary photographs in the wake of disaster can help us understand the workings of memory.
Jun 16, 2016
A discussion with two of the team members who helped create the “predictive policing” computer program about the mathematics behind it.
Feb 26, 2016
What could help more African Americans be successful in science, technology, engineering, and math? In this Q&A, hear from Ashanti Johnson, Melanie Harrison Okoro, and Danielle Lee.
Receive notification when new content is posted from the entire website, or choose from the customized feeds available.