Seed grants from the Transdisciplinary Institute in Applied Data Sciences (TRIADS) will support 10 new projects helmed by Washington University faculty members, harnessing cutting-edge tools to address pressing societal questions.
TRIADS has announced more than $480,000 in seed grant funding to 10 interdisciplinary teams whose members represent 12 different WashU departments and schools. This blending of diverse academic backgrounds is a core component of TRIADS and the Arts & Sciences Strategic Plan. Several projects have received co-funding from other WashU academic units, including the Weidenbaum Center on the Economy, Government, and Public Policy; the McKelvey Department of Computer Science & Engineering; the Olin Business School; and Olin’s Center for Economics & Strategy.
In addition to funding, seed grant teams will receive support from TRIADS for software development, administrative support, grant development, and collaborative space.
“We have projects touching on significant social issues ranging from AI in society to environmental justice to the effects of diversity in the workplace,” said Tammy English, TRIADS co-director. “Notably, many teams also promise to advance how we conduct research through developing new methods for data collection and analysis. It is a fantastic representation of the kind of intellectual community TRIADS is trying to build and support.”
Teams will use their funding to conduct preliminary research. The results will enable teams to seek external funding and, eventually, publish high-impact research.
Accounting for Human Bias to Improve AI-Assisted Decision Making
Faculty: Wouter Kool (Psychological & Brain Sciences), Chien-Ju Ho (Computer Science & Engineering)
Now more than ever, artificial intelligence assists and informs human decision making – from what to watch on YouTube to how to interpret crime data. But when AI systems make recommendations or provide information to humans, they also need to consider known biases in how people interpret and act on information. This project aims to develop an AI algorithm that incorporates knowledge about human biases to lead to better decision making. It will also study how people modify their behavior when they know that their decisions will inform AI systems.
Legacy of Neglect: Linking Flood Hazards, Pathogen Exposure, and Health Inequities in Cahokia Heights, IL
Faculty: Elizabeth Mallott (Biology), Theresa Gildner (Anthropology), Claire Masteller (Earth and Planetary Sciences)
In the coming decades, localized flooding will become increasingly common in many communities due to climate change. This project will focus on Cahokia Heights, Illinois, using the town as a test case for how pervasive flooding in a lower-income community can negatively impact residents’ health. The team will study satellite-based data to compare Cahokia Heights’ flooding levels to those in surrounding towns. This data will be paired with soil and water samples to measure human pathogen exposure in the region. Critically, the team will also work to establish clear links between flooding and health outcomes.
Mobile Assessment of Daily Life Contexts Using Smartphones
Faculty: Tammy English (Psychological & Brain Sciences), Jason Hassenstab (Neurology), Sojung Park (Brown School of Social Work), Ariela Schachter (Sociology)
Smartphone technology has the potential to revolutionize how scientists study people and intervene to change behavior. This project will put smartphones to use by measuring participants’ day-to-day activities. The team will recruit 50 younger and older adults who will install an app on their phones to gather a range of data about their daily environments. Combined with short surveys designed to supplement and verify the app’s measurements, the study will provide a richer picture of how people of different backgrounds navigate their everyday surroundings and interact with one another.
A Gaussian Process Framework for Idiographic Measurement of Psychological Traits
Faculty: Roman Garnett (Computer Science & Engineering), Joshua Jackson (Psychological & Brain Sciences), Jacob Montgomery (Political Science)
Does each human being have a unique mental makeup, or do our psychology and personality traits share a common structure? To tackle this longstanding question, this project will apply machine learning to psychological assessments of 50 research subjects. The team will take more than 100 assessments of each subject across three weeks. The study of the resulting data could have intriguing implications for personalizing psychological diagnosis and treatment.
MOBOGEN: Data-Driven Modeling of Mobility, Borders, and Genetics
Faculty: Michael Frachetti (Anthropology), Nathan Jacobs (Computer Science & Engineering), David Carter (Political Science)
Humans are an incredibly mobile species, having long since covered nearly every habitable region of the globe. This project will examine the “how” and the “why” of humanity’s movement across history, using global GPS data and real-time satellite imagery. Along the way, the MOBOGEN team will seek to answer core questions related to how geographic barriers and political borders impact human migration and ultimately our genetic heritage.
A Field Experimental Analysis of the Process and Implications of a Diverse and Inclusive Workplace: Social Identity, Communication, and Performance
Faculty: Brent Hickman (Olin School of Business), Jessie Sun (Psychological & Brain Sciences)
Efforts to support diversity and inclusion in the workforce have attracted enormous attention from businesses, academics, and policymakers. But research into how diverse workplaces function is still lacking. This project will employ a field experiment approach – creating a real-world business environment that will help researchers understand how the diversity of work teams affects real-time interactions between team members.
The Relationship Between State Violence, Trust in Government, and Vaccine Uptake
Faculty: Caitlin McMurtry (Brown School of Social Work), Michael Esposito (Sociology), Matthew Gabel (Political Science), Darrell Hudson (Brown School of Social Work)
The COVID-19 pandemic hit Black Americans especially hard, with notably higher rates of premature mortality than white Americans. But despite taking many other preventative measures to avoid COVID (such as masking and self-isolating), a disproportionately low number of Black Americans opted to receive the COVID-19 vaccine. This project proposes that this behavior could stem from a lack of trust in government, exacerbated by aggressive police practices in Black communities. To connect these two ideas, the group will compare ZIP code-level police data and vaccine adoption statistics, while also conducting focus group interviews to provide insight into the relationship between attitudes towards government, police practices, and the willingness of citizens to comply with public health recommendations.
Study of Polarization in Social Networks through Empirical Modeling and Simulations
Faculty: William Yeoh (Computer Science & Engineering), Dino Christenson (Political Science)
Partisan polarization in the United States has risen dramatically in the past decade and is often linked to ideological “echo chambers” created by social media algorithms. However, these claims are often anecdotal and lack rigorous research. This team will construct an analytical framework to understand how ideology, emotions, and algorithms can impact political discourse. To achieve that goal, researchers will also analyze existing social media data to form a baseline for their simulations.
Conformal Prediction for Uncertainty Quantification in Emerging Application Areas, from Materials to Social Science
Faculty: Robert Lunde (Mathematics and Statistics), Robert Wexler (Chemistry), Betsy Sinclair (Political Science)
Conformal prediction is a powerful tool for predicting outcomes while making very few assumptions about real-world forces. This method for analyzing complex datasets is incredibly adaptable and especially helpful for characterizing uncertainty about potential outcomes. This team will apply and adapt conformal prediction for two sets of questions: Can we better predict how to combine chemicals to create breakthrough materials in renewable energy production? Can we use this methodology to better estimate public opinion for hard-to-reach populations?
Measuring Financial Innovation via Natural Language Processing
Faculty: Ana Babus (Economics), Chenguang Wang (Computer Science & Engineering)
Over the past several decades, innovative investment methods — beyond traditional stocks and bonds — have changed how businesses raise money and opened new streams of potential revenue. These products are generally less transparent, less studied, and more loosely regulated. This has the potential to create market inefficiencies and, in extreme cases like the 2008 financial crisis, systemic risk resulting from poorly understood products spreading in concentrated sectors. This project will use natural language processing to study several decades of financial contracts, uncovering commonalities in these products and the firms that issue them. The goal is to better understand how market forces impact the evolution of these new investment methods.