BRAZILIAN SCIENTISTS TRY TO PREDICT MURDERS USING MATH AND POSSIBLY HOW TO PREVENT THEM

RIO DE JANEIRO, BRAZIL

Scientists in Brazil are using mathematical models and "urban metrics" in 15 cities to forecast murders and possibly prevent them.
A man reads a newspaper as Brazilian police take up positions during a raid of a slum in Rio de Janeiro in 2011. Scientists are using mathematical models to help forecast murders in Brazil.

 

In Brazil, where the homicide rate in 2011 was more than four times higher than in the U.S., predicting who dies from violent crime could be a numbers game.

So says a team of scientists using mathematical models and "urban metrics" in 15 Brazilian cities to forecast murders, and possibly prevent them.

The study, which appears in the latest edition of the peer-reviewed journal PLoS One, is based on data from public health records from 2000.

But can an algorithm that factors in variables such as GDP, illiteracy and male population size also identify crime trends?

"This is a very hard question to answer," acknowledged Haroldo Ribeiro, a co-author of the study by researchers at the State University of Maringa in southern Brazil.

Ribeiro, who has a Ph.D. in physics and specializes in data analysis and statistics, told MSN News in an email that a "clear relationship" exists between murders and population.

For example, he said, "One would expect that by doubling the population of the city, the number of homicides will also double. However, it happens that the number of homicides will be larger than the double. It will be multiplied by 2.2."

Ribeiro explained that this non-linear relationship is known as a "scaling law," which says that when combined, two elements can be more potent than the actual sum of their parts. In the case of a population that doubles, urban indicators such as income and unemployment can combine to create a more volatile scenario than expected.

However, Ribeiro said that using the "well-defined mathematical law" has its limits, because it can only point to possible factors that are connected with homicides on a broad level.

"What I can say is that if we know the 'macro' factors related to crime, we can find better political strategies aiming to reduce the number of homicides," he said.

Ribeiro said one macro factor in Brazil is impunity, which mathematical models show driving up criminal acts.

"When the penalties are promptly applied, the results show a reduction in crime rates," he said. "This point is another problem for us Brazilians. The justice here is very slow, [which] ... helps to spread the 'culture of impunity.'"

Criminologists have pointed out that the study cuts out the emotional component of human behavior, which is too unpredictable to be measured by a math formula.

Paul Appelbaum, director of Columbia University's division of law, ethics and psychiatry, cautioned that the Brazil study only offers a "snapshot" of specific circumstances and relationships in Brazil in one year.

"It's based on data from the year 2000 and in a single country. For these models to be predictive models, they would have to be generalized for other countries, at other points in time," he said. 

Appelbaum noted that meaningful indicators for Brazil such as illiteracy and child labor may not apply to New York or London, where rates of both are low and likely not correlative with crime. He also noted that although U.S. unemployment rose in the five years since 2008, crime rates have remained steady.

The Brazil study is not the first to cite math as a crime-fighting tool. University of California-Los Angeles researchers have applied earthquake prediction models to show how initial crimes can spur even more crimes in at-risk neighborhoods.