The Current

Data-based investigation could have led to Bruce McArthur's arrest much sooner, says expert

Criminologist Michael Arntfield has been using algorithms to catch serial killers in the U.S. — a strategy that is proving far more effective than traditional detective work but difficult to implement in Canada.
Bruce McArthur, 66, is accused of killing 6 men, and police believe there may be more victims. (Bruce McArthur/Facebook)

Read Story Transcript

WARNING: This story contains a graphic image of an unidentified man believed to be dead

Toronto police on March 5 released the photograph of an unidentified man who detectives believe to be another victim of alleged serial killer Bruce McArthur.

The move shocked many who have been following the case, and reignited criticisms from Toronto's LGBT community, again raising the question: Could police have done more to stop this from happening?

The data-driven work of Western University criminology professor Michael Arntfield suggests they could have done, and still need to do, much more.

Arntfield advocates using big data in addition to traditional detective work to identify and find serial killers.

His team's algorithm has allowed police in cities like Cleveland and Chicago to identify patterns of serial criminality more effectively than traditional detective work, which has led to earlier arrests. But Arntfield says institutional barriers stand in the way of implementing such a program in Canada. 

This combination of photos shows the six men McArthur is charged with killing. Top row, from left to right: Skandaraj Navaratnam, 40, Andrew Kinsman, 49, Selim Esen, 44. Bottom row, from left to right: Dean Lisowick, Soroush Mahmudi, 50, and Majeed Kayhan, 58. (CBC/Toronto Police Service)

Cluster analysis

Arntfield describes the patterns he and his team have discovered at the non-profit homicide databank Murder Accountability Project as "remarkably revelatory."

Their software uses specific data points to parse through 800,000 American murders since 1965 to find patterns that may link offences in an approach they call "cluster analysis."

The approach puts together the information of otherwise disconnected sources — like different jurisdictions, or long periods of time between reported murders.

"Data between 2000 and 2016 reveals there are 10 times the number of active serial killers than previously estimated by the FBI," Arntfield told The Current's Anna Maria Tremonti.

"In larger cities, you've got two and three [serial killers] active at once."

Michael Arntfield is a criminologist and professor at Western University. (Michaelarntfield.com)

Their approach has uncovered the largest cluster of unsolved strangulations in American history, with over 130 in Atlanta.

Another set of data shows that 15 per cent of all unsolved strangulations in the U.S. occurred in an 11-square-mile radius in Chicago since 2002 —- a pattern that Arntfield believes suggests the work of a serial killer.

"We know that, unfortunately, a lot of these investigators just don't share their notes. They don't talk," Arntfield said. "So the algorithm will sort of shore up that difference and we'll find patterns that thus far have not been detected."

And Arntfield says their "map of murder" has revealed there are far more serial killers in the U.S. than previously thought.

"Data between 2000 and 2016 reveals there are 10 times the number of active serial killers than previously estimated by the FBI."

Data between 2000 and 2016 reveals there are 10 times the number of active serial killers than previously estimated by the FBI.- Michael Arntfield

Arntfield said the Murder Accountability Project's cluster approach is helping police identify and catch suspected killers.

He said that recently in Cleveland, the algorithm detected serial activity, the team offered their findings to the police and an arrest was made in one of the cases.

But in Canada, there is no data-driven approach akin to the Murder Accountability Project. Arntfield argued this will hamper future investigations of suspected serial killers, including the case against Bruce McArthur.

Investigation an 'uphill battle'

​Arntfield argues that if police had a wider gamut of data and an easier way to coordinate and compare that data, it's likely they would have arrested McArthur sooner.

He uses the term 'linkage blindness' to describe the difficulty of connecting the dots in a wide-scale investigation like this one.

"I've been stressing this in the MacArthur investigation," he told Tremonti.

Toronto police announced March 5 they discovered a seventh set of human remains in garden planters seized from the grounds of a midtown home where McArthur worked as a landscaper. (David Donnelly/CBC)

"Right now, Toronto police is shouldering the entirety of this investigation because, again, all the [allegedly] confirmed victims [were found] within their jurisdiction," he said. "We also know that he was active for many, many years in other parts of Ontario prior to arriving in Toronto. So this case, as it expands, may take on dimensions that require countless other police departments, including police services no longer in existence."

But Arntfield said police and researchers face an "uphill battle" in implementing a murder database in Canada.

The primary data vehicle used in Murder Accountability Project, the Supplementary Homicide Report, has no Canadian equivalent, and Arntfield said federal authorities are reluctant to cooperate.

"Access to data, in spite of platitudes offered by the government to the contrary, is not that easy ... In Canada, I have found [a] good data vehicle, but StatsCan suggested it will not comply with FOIA requests, or access to information requests."


You can share this article across email, Facebook, Twitter and other platforms.

This segment was produced by The Current's Danielle Carr, Samira Mohyeddin and Julie Crysler.