From my first introduction to the HRDAG community at the annual retreat it was clear to me that mentorship is an organizational priority and that the contributions of interns are valued. Much of my first couple weeks as a summer intern at HRDAG were spent familiarizing myself with Patrick’s paradigm for principled data processing. At the same time, I was learning the skills and tricks (bash, make, vim, git) that promote an effortless programming workflow, a pursuit that Patrick calls “sharpening the saw” (just like in programming, you can cut down a tree with a dull blade, but your life will be much easier if you take the time to sharpen your saw beforehand).
In these pursuits, returning intern Gus was invaluable with his unbeatable knack for vim customization, and my co-intern Maria was learning right alongside me. Throughout the summer, the three of us would often be gathered around a whiteboard or screen to discuss some particularly troubling piece of syntax or write an intricate regular expression. In addition to more practical skills, I learned that the word “hack” can be a noun (“hacker”, “a hack”), a verb (“to hack, hacking”), or an adjective (“hacky”), and that “vim golf” might as well be an official sport.
My work with HRDAG this summer focused on bias and accountability in the US criminal justice system. Starting with my first project, which involved data from San Francisco, I became familiar with the idiosyncrasies that plague data from police departments and other institutional sources. I gained practice in dealing with missing values, misspellings, and inconsistent formatting of all shapes and sizes, and I acquired a trivia-like knowledge of certain sections of the California Penal Code. With this project, I examined and found statistical evidence of racial bias in charging for certain offenses that are “wobblers” in the state of California (those that can be charged at either the felony or misdemeanor level). Throughout the summer, Kristian provided expert guidance drawn from her vast experience with criminal justice data, particularly concerning automated risk assessment.
In my statistics courses I’d learned to apply an increasingly nuanced set of statistical models to well-behaved, textbook-quality data, but heard little about how the data came to look so neat. Real data is made by humans, is plagued with bureaucratic constraints, and rarely comes with a key. The results of a statistical analysis mean nothing if the data is meaningless. The skills I developed in cleaning and standardizing messy data (often referred to as “data munging”), were some of the most important abilities I gained from this summer. As the summer progressed, I was able to jump onto new criminal justice projects, as they came to HRDAG from around the country. Through this I gained an appreciation for HRDAG’s project model, one of a collaborative partnership recognizing the unique areas of expertise that each organization brings to the table. It was satisfying to see the many ways in which data science and statistics are being used to uncover and illuminate the most egregious inequalities in the criminal justice system.
Working in the HRDAG office was an absolute delight. Maria, Gus and I spent the summer noshing through the tastiest snacks that Trader Joes has to offer, and I got to learn tidbits about their work on HRDAG’s Guatemala projects. From Suzanne, I learned about Bay area gems such as the “Poetry Nap.” We even had a visit from Kristian’s two adorable poodles, who seemed more interested in lying on our shoes than contributing statistical insights.
I also spent time diving into the literature on quantitative fairness, familiarizing myself with this area of research that has become increasingly important, as algorithms play a larger and larger role in decisions about pre-trial detention. There is no single way to translate the idea of “fairness” into mathematical language; and many pairs of mathematical definitions of fairness are provably at odds in real world applications. In addition, some organizations have critiqued the focus on algorithmic fairness, characterizing automated risk assessment tools, “fair” or not, as inherently dangerous for the criminal justice system. This reading gave me much to ponder about the interface between machine learning and society and the social responsibility of statisticians. I was fortunate to have the opportunity to learn more about quantitative fairness from talking to HRDAG’s Shira Mitchell, when she spent a week in the office.
A highlight of the summer was traveling to the Joint Statistical Meetings (JSM) in Vancouver with Megan, Patrick, Kristian and Maria. Boarding a plane full of statisticians of all stripes and spending the week sampling a buffet of talks from respected statisticians is an invaluable experience for any young statistician, but having the HRDAG team as guides made it even more useful. Upon arriving at JSM, I sat in on a meeting of the ASA Committee on Scientific Freedom and Human Rights chaired by Megan with HRDAG’s Robin Mejia as a member. In this meeting I learned about the work that American statisticians are doing to protect freedom and autonomy of statisticians and statistical agencies worldwide. I was honored to be there to watch Patrick receive the Karl E. Peace Award for Outstanding Statistical Contribution for the Betterment of Society, and to celebrate, in HRDAG fashion, with a meal of oysters. Despite the busy week, I found the time to wander along Vancouver’s scenic coastline, and to sample the local cuisine at Canadian staple Tim Hortons.
I started in statistics out of a desire to use data for social good. Having completed the first year of my Master’s program in Statistics at the University of Chicago, I was eager to start doing just that. I couldn’t have asked for a better introduction to the field of human rights statistics. My wonderful mentors and colleagues at HRDAG were never lacking in wisdom, be it about nonprofit data science, programming tips, or the best places to get burritos and coffee in the Mission (shout-out to La Taqueria and Arizmendi Bakery). I’m thrilled to have the opportunity to continue to be involved in HRDAG projects in the coming year. Perhaps most importantly, I’ve emerged with insight and inspiration that will guide me as an early-career statistician committed to using statistics in service of human rights.