Using Big Data and Algorithms to Foster Equity in STEM

Executive Summary

The ADVANCE Resource and Coordination (ARC) Network convened scholars from multiple disciplines for a two-day workshop to prioritize under-studied research questions under the general theme of Using Big Data and Algorithms to Foster Equity in STEM. The Research Advisory Board of the ARC Network, a National Science Foundation-funded initiative at the Women in Engineering Proactive Network (WEPAN), identified this theme as a primary area in need of further research exploration as well as policy and practical intervention in academic science, technology, engineering, and mathematics (STEM) workplaces.

This theme was selected because of the ways in which big data and algorithms often perpetuate inequity, discrimination, and violence against people from marginalized communities. For example, facial recognition software used to unlock cell phones, for airport passenger screening, in employment decisions, in ride sharing applications, and for law enforcement surveillance not only raises privacy issues, but also dangerously and consistently has the poorest accuracy when used to identify the faces of Black women (Buolamwini & Gebru, 2018; Watkins, 2021). This is a problem not only because we want the technology to work, but because marginalized individuals are disproportionately targeted. At the workshop, the planning committee sought to discuss the possibility of how big data and algorithms might instead foster equity, particularly in STEM fields.

Members of the workshop planning committee nominated scholars working in these areas who represent a diverse array of disciplines, research specialties, institution types, career stages, and social demographic backgrounds. Twenty scholars and practitioners convened in December 2021 and participated in a series of facilitator-led discussions designed to culminate in a research agenda of under-studied questions that will advance understanding of using big data and algorithms to foster equity in STEM.

By the end of our time together and with additional input from the larger community of researchers and practitioners, the group prioritized three research frontiers:

• Missing data: The problem of missing variables and/or values in big data sets
• Mixed methods: The need for qualitative methods to complement quantitative approaches to big data: getting to the “why” and ‘how’ to supplement the “what”
• Interventions: The desire to design interventions to correct inequities identified from analyses of big datasets

The three priority areas emerged from extensive discussion among workshop participants, and suggestions for expanded research needs are provided. In addition, other questions where research is needed include:

1. What are effective practices for dealing with data exhaust, or uses of data different from the intended ones at the time of data collection, especially when we are trying to promote equity?
2. How will we understand outcomes? Big data approaches can rarely tackle causation, yet these approaches can suggest interventions that need to be tied for anticipated out- comes. The most important desired outcome is an understanding of what characterizes well-being or success in STEM.
3. What measures of bias within the data are needed and how can we incentivize explicit descriptions of those biases in publications?
4. How can researchers integrate participatory methods of data analysis with big data in ways that do not jeopardize privacy?
5. How can we encourage more scholars from marginalized backgrounds to engage with big data approaches?
6. How can researchers influence federal funding directorates to support, fund, and en- gage in critical and ethical work to foster equity in STEM through big data and algorithms?
7. How can big data and algorithms help us understand the effectiveness of interventions for STEM equity from middle-school through late career? Where are the unsolved problems, and how do we ethically collect data to tackle those?

We encourage researchers to consider pursuing these topics and exploring the questions described within this report, especially in collaboration across fields and with practitioners.