Ethics

Lesson 23: Ethics in Data Visualization

Objective

Critically examine the power and responsibility designers hold when creating data visualizations, exploring how design choices influence representation, narrative, and agency while considering the ethical implications of these decisions.

Outline

This lecture addresses complex ethical questions in data visualization, exploring how our design choices shape understanding and impact society. We examine ethical dimensions through the lens of power, representation, and agency, challenging students to think critically about their work.

Content Warning

This lecture includes challenging material on ethical issues in data visualization including representation, bias, and power dynamics in design.

Situating ourselves

A starting look at https://www.youtube.com/watch?v=AaU6tI2pb3M&t=690s.

Lenses on the World

Data visualization is not neutral and serves as a way of seeing that shapes understanding.

• John Berger's "Ways of Seeing": how we see things is affected by what we know and what we believe.
• Jones: design choices give visualization rhetorical power that blurs the line between persuasion and deception.
• Designers have power and responsibility for how they shape understanding and narratives.
• These questions extend beyond visual representation into the heart of technology, including AI.

Digging in deeper

Philosophical and theoretical frameworks help motivate practical considerations.

• Michel Foucault on Discourse: power prescribes particular rules and categories which define criteria for legitimating knowledge and truth, and discourse masks its construction and capacity to produce knowledge and meaning.
• Bret Victor's "Media for Thinking the Unthinkable": the medium changes what is possible to think altogether, shaping not just representation but what ideas are possible to conceive.
• Marian Wright Edelman: "You can't be what you can't see" - representation matters and shapes what people believe is possible.

Encoding Choices and Ethics

Let's reivist Cleveland and McGill's hierarchy with these ethical considerations.

• What are the "most important" variables and who decides what matters?
• Which variables receive the highest-accuracy encodings and what narrative does that tell?
• Baseline choices: why is distance from zero standard, when is a broken axis appropriate or problematic?
• Example: Income Gaps Visualization demonstrates how encoding choices emphasize certain comparisons.

Possibility Space (Action)

The possibility space defines what questions we can and cannot ask of data.

• What actions are allowed in our visualization and why did we choose these specific interactions?
• What questions are made easy versus difficult or impossible?
• Example: Global Plastics AI Policy Tool defines which policy interventions are considered and what futures can be explored.
• How are we warping the possibility space through our accessibility choices and who can access the full range of actions?

Possibility Space (Narrative)

Beyond what actions are possible, we must consider what narratives are possible.

• Propaganda Games (Extra Credits): games and interactive visualizations can promote particular viewpoints through mechanics. https://www.youtube.com/watch?v=UP4_bMhZ4gA
• Parable of the Polygons: what is gained through abstraction (accessibility, clarity, universal applicability) versus what is lost (specificity, individual experiences, historical context).
• Three Treatments of Gun Deaths: comparing Periscopic Gun Deaths (emotional, individual focus), Rand Longitudinal Study (state-by-state trends, policy impacts), and MDSR Book (statistical presentation, data accuracy).
• No single answer exists but each makes different choices about emotional versus analytical framing and individual versus aggregate perspective.

Agency and Representation

Consider how design choices change agency and influence representation.

• How are we choosing what is good and what is bad?
• How are we choosing what is visible and what is invisible?
• How are we choosing what is possible to think?
• How would those we are representing feel about that representation?
• Unlike some other forms of technology, these choices in data visualization are often more explicit through encoding devices, user actions, ludonarrative, and possibility space.

Design Justice Network Principles

A framework for ethical design practice that centers the voices of those directly impacted by design decisions.

• Provides practical guidance for ethical considerations.
• Available at https://designjustice.org/read-the-principles

Moving Forward: A Positive Framework

This lecture offers guidance rather than paralysis.

• Broaden the perspective of who is an author through co-creation and participatory design.
• These choices are essential, make them wisely with acknowledgment of responsibility.
• Think about which narratives are elevated or diminished and be intentional about perspective.
• There is no formula to get to a right or wrong answer - we have to think critically and do the work.

Take Aways

Data visualization involves unavoidable ethical choices about representation and power that require critical examination and thoughtful decision-making.

• Data visualization is not neutral - we are creating "ways of seeing" that shape understanding.
• Design choices involve power and responsibility including encoding devices, baselines, possibility space, and accessibility.
• Theoretical foundations from Foucault, Victor, and Edelman help us understand how visualization shapes what is possible to think and see.
• Cleveland and McGill's hierarchy requires asking who decides what variables are "most important" to determine encoding.
• Possibility space defines what questions and narratives are possible through allowed actions and interactions.
• Agency and representation require considering how those represented would feel about their representation.
• Design Justice Network Principles provide a framework for centering impacted voices.
• This work requires critical thinking without a formula: we must do the work of thoughtful ethical design.

Citations

[1] A. Shatov, "White Digital Device at 12 00," Unsplash, 2021. Available: https://unsplash.com/photos/white-digital-device-at-12-00-DHl49oyrn7Y

[2] J. Berger, Ways of Seeing, Penguin Books, 1972.

[3] R. Jones, "Deceptive by Design: Data Visualization and The Ethics of Representation," MIT. Available: https://technologist.mit.edu/deceptive-by-design-data-visualization-and-the-ethics-of-representation/

[4] Wikipedia Contributors, "Ciclo de conferências do filósofo francês Michel Foucault, no Hospital das Clínicas da Universidade do Estado da Guanabara (UEG)," Wikimedia Foundation. Available: https://en.wikipedia.org/wiki/Michel_Foucault#/media/File:Michel_Foucault_1974_Brasil.jpg

[5] W. Pragher, "Photograph of Martin Heidegger," Wikimedia Foundation. Available: https://en.wikipedia.org/wiki/Martin_Heidegger#/media/File:Heidegger_2_(1960).jpg

[6] R. Adams, "Michel Foucault: Discourse," Critical Legal Thinking, 2017. Available: https://criticallegalthinking.com/2017/11/17/michel-foucault-discourse/

[7] Wikipedia Contributors, "Bret Victor." Wikimedia Foundation, Inc., Jun. 22, 2023. Available: https://en.wikipedia.org/wiki/Bret_Victor

[8] B. Victor, "Media for Thinking The Unthinkable," MIT Media Lab, 2013. Available: https://vimeo.com/67076984

[9] CDC, "Marian Wright Edelman," Wikimedia Foundation. Available: https://en.wikipedia.org/wiki/Marian_Wright_Edelman#/media/File:Marian_Wright_Edelman_01.jpg

[10] M. Edelman, "You can't be what you can't see," Goodreads. Available: https://www.goodreads.com/quotes/536048-you-can-t-be-what-you-can-t-see

[11] A. Thorn, "AI is an Ethical Nightmare," Philosophy Tube, 2023. Available: https://www.youtube.com/watch?v=AaU6tI2pb3M

[12] W. S. Cleveland and R. McGill, "Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods," Journal of the American Statistical Association, vol. 79, no. 387, pp. 531–554, Sep. 1984, doi: 10.1080/01621459.1984.10478080.

[13] A. Pottinger, "Income Gaps." 2023. Available: https://incomegaps.com/

[14] A. Tierney, "Median annual earnings of full-time workers in the United States in 2023, by gender," Statista, 2024. Available: https://www.statista.com/statistics/1186135/us-median-annual-worker-earnings-by-gender/

[15] B. Baumer, D. Kaplan, and N. Horton, Modern Data Science with R, 2nd Ed, CRC Press, 2021.

[16] K. Rees and Periscopic, "U.S. Gun Deaths." Periscopic, 2018. Available: https://guns.periscopic.com/

[17] K. Sumah, L. Floyd, and H. McCracken, "Changes in State Firearm Mortality," Rand Corporation, 2024. Available: https://www.rand.org/research/gun-policy/longitudinal-firearm-mortality.html

[18] A. Pottinger, R. Geyer, N. Biyani, C. Martinez, N. Nathan, M. Morse, M. de Bruyn, C. Boettiger, E. Baker, K. Koy, and D. McCauley, "Global Plastics AI Policy Tool," University of California, 2024. Available: https://global-plastics-tool.org/

[19] A. Pottinger, R. Geyer, N. Biyani, C. Martinez, N. Nathan, M, Morse, C. Liu, S. Hu, M. de Bruyn, C. Boettiger, E. Baker, and D. McCauley, "Pathways to reduce global plastic waste mismanagement and greenhouse gas emissions by 2050," Science, 2024. doi: 10.1126/science.adr3837

[20] J. Portnow, "Propaganda Games - Ethical Game Design," Extra Credits, 2012. Available: https://www.youtube.com/watch?v=UP4_bMhZ4gA

[21] V. Hart and N. Case, "Parable of the Polygons," Nicky Case, 2022. Available: https://ncase.me/polygons/

[22] Design Justice Network, "Design Justice Network Principles," Design Justice Network, 2018. Available: https://designjustice.org/read-the-principles

[23] S. Costanza-Chock, "Design Justice" (Eyeo 2019). Available: https://vimeo.com/354276956

License

This lesson is part of Interactive Data Science and Visualization and is released under a CC-BY-NC 4.0 license.