Come See the Bias Inherent in the System!

Like many Americans, I was shocked when I learned in early July of two more incidents in which black men were shot and killed by white police officers. The videos of the shootings of Alton Sterling in Baton Rouge, Louisiana, on July 5, 2016, and Philandro Castile in Falcon Heights, Minnesota, on July 6, 2016, were disturbing, to say the least. As a member of the most common demographic group in the American Academy of Psychiatry and the Law (AAPL) (i.e., white, male and (relatively) high income), I was uncertain how to respond to the tragic shootings. After all, I am a part of the criminal justice system, as I make my living doing criminal forensic evaluations, primarily in the form of court-ordered competence and sanity evaluations. As has become more and more evident in recent years, the data show that our criminal justice system has profound biases, which typically run against black defendants and favor those defendants who have more resources.¹ Federal investigation of policing patterns in the aftermath of the police shooting of Michael Brown in August 2014 in Ferguson, Missouri, and the death of Freddie Gray while in police custody in April 2015 in Baltimore, Maryland, have yielded scathing reports and damning statistics on those cities' police practices.^2,3 Given these biases, I wondered how I could reconcile my ongoing participation in this system and whether I have contributed to the bias that is so clearly part of it.

My short-term response to the shootings in early July 2016 was to use the opportunity of a forensic journal club with the psychiatry residents, scheduled for July 15, to discuss the June 2016 U.S. Supreme Court decision in Utah v. Strieff,⁴ which focused on the arrest of a man for possession of methamphetamine after he was stopped for questioning and was found to have an outstanding warrant. The majority agreed the circumstances of Mr. Strieff's arrest were not unconstitutional, but Justice Sonia Sotomayor's dissent was vigorous and, in Section IV, personal: This case involves a suspicionless stop, one in which the officer initiated this chain of events without justification. As the Justice Department notes, many innocent people are subjected to the humiliations of these unconstitutional searches. The white defendant in this case shows that anyone's dignity can be violated in this manner. But it is no secret that people of color are disproportionate victims of this type of scrutiny. See M. Alexander, The New Jim Crow (2010) [Ref. 5]. For generations, black and brown parents have given their children “the talk”—instructing them never to run down the street; always keep your hands where they can be seen; do not even think of talking back to a stranger—all out of fear of how an officer with a gun will react to them.By legitimizing the conduct that produces this double consciousness, this case tells everyone, white and black, guilty and innocent, that an officer can verify your legal status at any time. It says that your body is subject to invasion while courts excuse the violation of your rights. It implies that you are not a citizen of a democracy but the subject of a carceral state, just waiting to be cataloged.We must not pretend that the countless people who are routinely targeted by police are “isolated.” They are the canaries in the coal mine whose deaths, civil and literal, warn us that no one can breathe in this atmosphere. See L. Guinier & G. Torres, The Miner's Canary (2002) [Ref. 6]. They are the ones who recognize that unlawful police stops corrode all our civil liberties and threaten all our lives. Until their voices matter too, our justice system will continue to be anything but.

The discussion of this decision and Justice Sotomayor's dissent was more personal and immediate than our typical journal club discussions, where we usually talk about landmark decisions from the AAPL list or state court decisions that might affect psychiatrists or people with serious mental illness. Those journal clubs are stimulating, but this one was about something that was happening now, and it had ramifications for all of us. It was not hard for the residents to make the leap from concerns about bias in the response by police in their treatment of black citizens to concern about bias in the response of many groups in their treatment of people with serious mental illness. At the end of the hour, the residents were thankful for the opportunity to talk about the police shootings, which was gratifying, but I was still unsettled about my part in the criminal justice system.

The next step in my reaction to the July shootings was to determine whether my evaluations showed any evidence of bias, an analysis that I had thought about doing for some time. First, a bit of background: when I started to do criminal forensic evaluations, I was trained to identify the defendant by age and gender at the start of the report, but not by race. Like almost all physicians, I learned in medical school to begin every note, every report, with the patient's age, race, and gender, and I remember wondering why the court clinic dropped race from this standard identification. When I thought about it, though, the logic of not making a point of the defendant's race made sense; we were meant to be objective evaluators of defendants' competence and sanity. As an aside, during the journal club discussion of Utah v. Strieff, the residents informed me that the use of age, race, and gender in progress notes and reports remains nearly universal in medical school training, even though this information is easily found in the electronic medical record or in the front sheet of the increasingly rare paper medical record. One unintended consequence of learning not to use race in my forensic reports, though, was that, when I set up an Excel database to keep track of my evaluations, I did not include a column for race when I recorded the demographics, charges, diagnoses, and my opinions for each defendant. It did not occur to me to add a column for race until five years ago, when I used my database to look at evaluations I had done for the local mental health court.⁷ Since 2012, I have duly recorded race in the demographic information for each defendant. At the end of each calendar year, I look at my patterns of opinions, to see what proportion of the defendants I evaluated I thought was competent and what proportion I thought was insane. Over the past few years, I had thought of looking at the patterns of my opinions by race, but never had done so; it takes time, after all.

In the summer of 2016, I had good reason to take the time to look at my patterns, as I wanted to see how I fit in the American criminal justice system. I learned from my database that over the past 4.5 years, I evaluated 734 defendants for court-ordered determinations of competence or sanity or both, completing 682 competence evaluations and 380 sanity evaluations. Since all but 15 of these defendants were either white (56%) or black (42%), I restricted all subsequent analyses to the 719 white and black defendants whom I had evaluated. When I analyzed the patterns of my findings by race and gender, I found I had been more likely to find white defendants competent to stand trial than black defendants, more likely to find black men competent to stand trial than black women, and more likely to find white women competent to stand trial than black women. I also found I had been more likely to find female defendants insane than male defendants and more likely to find white women insane than white men.

But I do not control whom the courts order to undergo competence and sanity evaluations, so perhaps this is where we can ‘Come see the bias inherent in the system!’⁸ The large urban county that is the source of the majority of my court-ordered evaluations is 66 percent white, 28 percent black, and 10 percent Hispanic.⁹ The USA Today website has an interactive feature which will bring up the arrest rates by police department for black and non-black residents, based on their analysis of FBI arrest data from 2011 to 2012¹⁰; their analysis showed that the city that comprises this urban county had arrest rates of 222 per 1,000 black residents and 74 per 1,000 white residents. Combining these arrest rates with the census numbers shows that, despite a population ratio of black residents to white residents of 0.4:1, the ratio of the total number of arrests of black residents compared with white residents was 1.25:1. But the ratio of black defendants to white defendants referred to me by the courts in this county for evaluation of competence was even greater, 2:1. Of course, the hundred or so evaluations I did for this county in an average year paled in comparison with the more than 45,000 new criminal cases filed each year in the same county. Even so, in the rare circumstances where a defendant is ordered to undergo evaluation of competence to stand trial, perhaps the lawyers and judges in this urban county have a bias in whom they select for competence evaluation, as black defendants appear to be disproportionately ordered to undergo evaluations of competence. Once referred to me for evaluation, I was more likely to find black defendants incompetent to stand trial than white defendants, by a ratio of 1.25:1. It appears, then, that I showed evidence of modest bias in my competence evaluations and that, even in my very small corner of the criminal justice system, the bias may have come, at least in part, from the criminal system itself.

When I compared the demographics of the defendants evaluated for the large urban county with the demographics of the defendants from all of the other counties I had worked for, I found that defendants referred for competence evaluation from the urban county were much more likely than the defendants from other counties to be black (67% versus 13%). Hispanic defendants were rare in the urban county (1%) and in all other counties (2%), despite representing 10 percent of the population of the urban county and 7 percent of the state population,¹¹ which opens up the question of why Hispanic defendants were disproportionately unlikely to be referred for competence to stand trial. The patterns of my opinions of competence and sanity in the two geographical groups showed relatively little difference, though I was more likely to find white defendants than black defendants competent in both the urban county and all the other counties.

While it was challenging and stimulating to try to make sense of my patterns in the context of where I practice, I do not want readers to get lost in my numbers and my local circumstances; instead, I want to encourage all forensic mental health professionals to look at their own numbers and their own circumstances. I suspect many forensic colleagues already keep track of their evaluations, but we could all maintain databases of the people we have evaluated and the opinions we have put into our reports. It is not hard to set up an Excel database and, once it becomes a habit, adding each report takes little time. We could all look at our data from time to time, to look for our own trends and patterns. It would not take very long if the database is set up right, and what can be learned is worth the effort. Finally, we all could take this analysis one step further and try to compare our patterns to the larger patterns in our communities, to try to understand how our patterns fit or do not fit, with our communities, which of course carry their own biases. While this approach is most pertinent now for those of us who do criminal forensic evaluations and work in corrections, our colleagues who do primarily civil evaluations could also participate in this exercise and look for patterns in who is referred to them for evaluation of disability or damages and how they answer the questions about disability and damages.

Even as you consider your practice patterns, it is important to remember that “every system is perfectly designed to get the results it gets.”¹² Implicit bias has become a hot topic for all of society and this discussion has filtered down to the medical field.¹³ Human experiences and how we interpret them create the hidden biases we all naturally carry within, but evidence of these biases can be found by self-examination, by looking at data from our own practices, by using standardized tests of social biases, like the Implicit Association Test,¹⁴ as well as by thoughtful peer review. It is time for forensic practitioners to think about how bias might affect the work we do and how that bias can affect the world in which we work and live.

• © 2016 American Academy of Psychiatry and the Law