Social Sciences

Crime and Punishment

Examining the intersection between race and prosecutions
jail_cell

Daniel Silberman has long been a proponent of criminal justice reform.

On debate teams in high school and at the UO, he has argued for an end to mandatory minimum sentences and even prison abolition. He has been particularly interested in whether the prosecutorial process can result in different outcomes, depending on a defendant’s race.

From this, Silberman formulated a pressing question: How do changes in a community’s racial demographics correlate to changes in prosecutions?

For his final research paper in sociology, which earned him an honors distinction, he explored that very question.

Silberman—who graduated earlier this year and begins at UCLA School of Law this fall—collected data for all 36 Oregon counties from the US Census, American Community Survey and the Oregon Department of Corrections. He established each county’s felony conviction rate over two five-year periods, 1998–2002 and 2008–12; for racial and economic demographics, he collected county data for 2000 and 2010.

Then, with the help of complicated statistical analysis, he compared each county’s racial composition and economic demographics with felony conviction rates for drug and violent crimes.

Daniel Silberman

Daniel Silberman

In counties across Oregon, Silberman’s statistical model correlated a rise in black or Hispanic populations with an even higher rise in felony convictions for drug and violent crimes—despite the fact that there was no significant surge in those offenses in the FBI reported crime rate for the periods under review.

The model predicted that, over the 10 years, if the number of Hispanics in an Oregon county increased by 1 percent, there would be an average increase of 3.5 percent in felony drug conviction rates.

Just to be safe, Silberman tried to eliminate the argument that the predicted rise in prosecutions could have been due to increased crime that the FBI didn’t capture. He did this by accounting for factors that are sometimes correlated with increased crime and, therefore, more convictions—poverty and unemployment, for example. The predictions held even after adjusting for such factors.

Silberman’s conclusion: For the 10-year period, increases in minority populations were strongly associated with increased prosecutorial harshness.

Though quick to stress that his work doesn’t show that authorities are intentionally targeting blacks and Hispanics, Silberman suggested his findings warrant additional review of the issue.

Associate professor Aaron Gullickson, Silberman’s thesis advisor, introduced Silberman to a powerful statistical tool that made the analysis possible: computing software called, simply, “R,” which is, Gullickson said, “the best software of its kind.”

“But it has the steepest learning curve,” he added. “It’s not a point-and-click kind of thing. It’s almost like learning a real programming language.”

Silberman was a quick study. A few weeks after being introduced to the software, Silberman not only had figured out how to use it, Gullickson said, “but he had been downloading soccer stats and was doing analysis on those as well.”

By the time Silberman was done with the criminal justice project, he was an expert with complicated statistical tools such as “fixed-effects linear regression models.” He created graphs that plotted conviction rates against minority populations. He also produced recommendations for the criminal justice system, including uniform charging standards to ensure consistency among prosecutors.

This from a student who initially was taken aback when he realized that his research question would require him to dig deep into quantitative statistical analysis.

Said Silberman: “I hadn’t taken a statistics class since high school.”

—Matt Cooper