Because there are still many people who continue to doubt systematic racial and gender bias in law enforcement, I have been working on some new statistical tests of these claims. I recently released new statistical evidence of racism and sexism in the enforcement of “stand your ground” (SYG) laws, using a sample of cases from Florida in the period from 2006 to 2013. A detailed explanation of the data, methods, and findings (including fully reproducible data and code) can be found in the full working paper linked above. Below, I summarize the two key findings.

The two key findings

1) The probability of conviction for a white defendant against a white victim in a typical case is fairly high at around 90% but with a large margin of error, whereas the probability of conviction for a black defendant against a white victim approaches 100%, even after controlling for more than 10 different objective factors related to the circumstances of the incident.

The probabilities of conviction for white and non-white defendants against white and non-white victims (95% confidence intervals in grey)

2) The probability of conviction for a male defendant in a typical domestic case is found to be about 40%, but for a female defendant in an otherwise objectively equivalent case the probability of conviction increases dramatically to 80%.

The different probabilities of conviction for male and female defendants in domestic and non-domestic cases (95% confidence intervals in grey)

What these models teach us about recent cases

1) The probability George Zimmerman was going to be found guilty of murdering unarmed black teenager Trayvon Martin in 2012 was marginally greater than 50% but would have been about 98% if Trayvon Martin had been white.

2) Black female Marissa Alexander who fired a physically innocuous warning shot to deter her husband in 2010 faced a probability of conviction marginally greater than 50%, but the probability of conviction for a male defendant in an otherwise objectively equivalent situation would have been around only 12%.

What is new about these findings and why they matter

1) The only previous statistical analysis of racial bias in the enforcement of stand your grown laws at the level of individual cases concluded that there has been no racial bias. Conservative gun advocate John Lott analyzed the same data I analyzed and submitted his analysis as testimony to the US Senate Judiciary Committee. As I explain in the paper, his statistical models were very superficially reported and deeply flawed. Thus, my findings overturn the last most influential study which, I believe, incorrectly informed US policymakers.

2) This analysis controls for a very large number of contextual factors which are frequently invoked to dismiss claims of racism and sexism. In other words, if anyone ever says the legal system is not racist or sexist but people of color and/or women are simply more likely to do/be some other thing which rightfully earns them convictions and just happens to be correlated with race or gender, my study provides evidence of racism and sexism in SYG enforcement even after subtracting out any effect which might come from race or gender correlating with a long list of common conservative excuses: whether the victim initiates, whether the defendant can retreat, whether the defendant pursues, whether the defendant has a gun, whether the incident takes place on the defendant’s property, whether the victim dies, age of victim and defendant, whether there is physical evidence, whether there is a witness, and whether the victim is committing a crime.

Murphy, Justin (2015): A Statistical Analysis of Racism and Sexism in “Stand Your Ground” Cases in Florida, 2006–2013. http://dx.doi.org/10.6084/m9.figshare.1301532

You can follow me on Twitter (@jmrphy) or sign up to my weekly newsletter on The Art and Science of Overthrowing All Currently Existing Institutions.


Originally published at jmrphy.net on February 5, 2015.