�Ӱ���Ƶ

A Mighty Win for Reed Statistics Students

Three students took first place in a national statistics competition.

A team of three Reedies, Robin Hart '23, Maxwell VanLandSchoot '22, and Sung Bum “Simon” Ahn ’23 recently won 1st place in the intermediate section of the American Statistical Association (ASA) and the Consortium for the Advancement of Undergraduate Statistics Education (CAUSE) sponsored Undergraduate Statistics Project Competition (USPROC). 

Although their win was only announced in the last few months, the origin of the project goes back a bit further.

It all started with an internship. Robin Hart ’23, interned as the data analyst for Washington Defender Association (WDA) under Executive Director Christie Hedman ’80. WDA is an organization dedicated to public defense reform that is instrumental in supporting and training social workers, public defense attorneys, and investigators. During their tenure at WDA, Robin encountered their wealth of data. They were particularly attracted to the funding allocations of The Washington State Office of Public Defense (OPD). 

They remembered this data set months later while taking Statistical Learning with Visiting Assistant Professor Johnathan Wells. He wanted the class to form groups and work on a project together. Robin thought back to the OPD and showed its data to a group that included Maxwell and Simon. 

The group quickly realized that OPD’s funding model had issues. First and foremost, the model only considered misdemeanor caseload and population. Secondly, OPD’s decision to make their baseline funding 6%, split evenly between 39 counties, was completely arbitrary. Even more importantly, the model ignored factors such as poverty and median income, which could affect counties' budgetary needs.

The group decided to create a new funding model using caseload resources data and county statistics from 2018 to 2020. The small sample size was initially a problem, but, with Professor Wells’s guidance, they figured out how to best utilize it. They wanted their model to consider more variables to both address inequity and increase transparency. Using an analytic technique known as a "Lasso," they validated their hypothesis that important variables were being ignored. Accounting for these could enhance the way the OPD allocated funds. As such, the funding model that the group created significantly outperformed the OPD's current model, giving funds to counties with higher poverty levels.

Since the group’s goal was to create a whole new model, they also did a Principal Component Analysis (PCA) because it encompasses more variables and decreases information loss. Professor Wells taught them how to incorporate a type of algorithm that would learn from the raw data in their model, making it even more effective. What they found was that their model better accounted for factors like poverty, median income, median home prices, and other resources available to public defenders in Washington State, allowing it to make more equitable allocations.

Professor Wells was enthusiastic about their findings. And though he urged them to submit it to USPROC, they were all blown away when they won. The team hopes that their win shows that stats can be a tool to fight for equity and justice and that though Reed’s Stats department is small, it’s mighty.

Tags: Academics, Awards & Achievements, Students