James Buehler: Data and Disparity

Recently, an article by Ronald G. Fryer found no statistical differences between White, Black,or Hispanic individuals in the use of lethal force by police.

Fryer, an Economics Professor at Harvard, based most findings on 4 sources of data:

1. Information from New York City’s Stop, Question, and Frisk Program
2. The Police-Public Contact Survey
3. Police reporting in Austin, Dallas, Houston, Los Angeles County, and 6 Florida counties.
4. A random sample of police-civilian interactions from the Houston Police Department

Fryer’s research found racial disparities at several levels of police interaction with civilians.  Uses of force, such as “slapping or grabbing,” or “pushing individuals into a wall or onto the ground,” were fifty percent more likely to occur if the individual being detained was identified  as black or Hispanic.

He and his team did not find racial disparities in those  interactions that ended with the police firing their weapon at the civilian.  According to Fryer, “in the raw data, blacks are 23.8 percent less likely to be shot at by police relative to whites.”

James Buehler, a professor of Public Health at Drexel University, and a former Commissioner of Health for Philadelphia, was intrigued both by the paper and by the publicity it garnered.

“It’s a beautiful, elegant, incredible, very rich and deep piece of work,” said Buehler. “I had a concern that it could be misconstrued in a way that might be misleading.”

He wanted to see if he could find different results, when starting with a different kind of data.

Using WONDER, an online data-gathering tool created by the CDC, Buehler looked at death certificates in the US, from 2010 to 2014, filtering for those deaths linked to legal intervention by law enforcement.

“The value of death certificates is that, because everybody gets a death certificate, it’s a way to study broadly what are the patterns of death,” said Buehler.

In all, he examined 2285 death certificates. Among his findings were:

• 96% of deaths resulted from the discharge of a firearm.
• All but 5 deaths were people aged 10 and older.
• 96% of those who died were males.

Examining the cause of death compared to the total population, Buehler found several racial disparities. Over the same span of time, there were an average of 2.5 deaths-per-million among whites, 6.8 deaths-per-million among African-Americans, 1.5 deaths-per-million among Asian and Pacific Islanders, 6.9 deaths-per-million, and among Hispanics, 4.1 deaths-per-million.

Buehler included data from a study by Alex Crosby and Bridget Lyons, whose research into legal intervention deaths in the same years yielded different total numbers of deaths (38% fewer in 2013), but similar ratios of disparity.  That’s part of Buehler’s point.

“It’s almost inevitable that if you start to dig further, you see that there might be some under-recognition of a particular cause.”

Death certificate information, which is widely accepted as acredible source for researchers, has implicit imperfections.  The certificate could have been filled out incorrectly.  Perhaps the particular way a person died has yet to be defined by the CDC.  Maybe the death was caused by an illness, like AIDS or Ebola, which carries cultural stigmas, and so was purposefully whitewashed.

Buehler’s findings, which were not meant to detract from Fryer’s research, were published in part to caution hasty interpretation on the part of interested parties like media outlets, to consider not just what the final numbers say, but where they came from.