Critical Data Visualisation

Critical Data Visualisation

Critical data visualisation applies the insights of critical data studies to the practice of designing and interpreting visual representations of data. If critical data studies investigates how power operates in data systems, critical data visualisation extends this analysis into the realm of visual communication — asking not just what data is produced, but how it is represented, and with what consequences for different audiences.

Two core questions organise this field:

1. What is the data being visualised? — Where does it come from? What choices shaped its collection? Whose perspectives are included and whose are excluded?
2. How can we create more responsible data visualisations? — What design principles and ethical commitments should guide visualisation practice?

Pinney (2020) argues that data visualisation literacy can be a vehicle for social change — by developing the capacity to read visualisations critically, individuals gain tools to question authority, challenge unequal representations, and advocate for more equitable forms of data communication.

Feminist Data Visualisation

D'Ignazio and Klein (2016) propose a framework of feminist data visualisation that directly challenges the conventions of mainstream data design. Drawing on feminist theory and data feminism, they argue that standard data visualisation practice encodes assumptions about objectivity, neutrality, and universalism that must be examined and challenged.

They identify five core principles:

1. Rethink Binaries — mainstream data often relies on binary categories that oversimplify complex phenomena. The critique extends beyond gender to all binaries: nature vs culture, subject vs object, reason vs emotion, normal vs abnormal. Feminist data visualisation develops representation strategies based on multiplicity and nuance rather than forced binary oppositions. Design process questions include: Is our data the right type? What categories have we taken for granted? How can we register responses that do not fit into our categories, especially edge cases and outliers? Design output questions include: How do we communicate the limits of our categories? How can we allow users to refactor the categories we have presented?
2. Embrace Pluralism — mainstream data visualisation often presents a single authoritative perspective as if it were universal. Feminist data visualisation considers subjectivity — who is creating the visualisation and for whom — and challenges notions of objectivity and neutrality. It moves beyond a simple producer/audience binary to consider the full range of subjectivities involved in production and reception. Design process questions include: Whose voices are not represented on the design team but might be important? Who is being envisioned as the ideal user? Whose perspectives have been excluded from the categorisation schema? Design output questions include: Can the artifact communicate the subject positions of the researcher and designer transparently? Whose view of the world does the visualisation represent?
3. Examine Power and Aspire to Empower — data visualisation can reinforce hierarchical power structures or challenge them. Feminist data visualisation rejects hierarchical approaches and acknowledges the user as a source of knowledge. Projects like Queering the Map give the power of data back to individuals — recording and sharing personal and community histories that mainstream data systems erase. Design process questions include: How is power distributed across the design team? How can end-users' voices be more fully integrated? Can we employ participatory design processes? Design output questions include: Can the visualisation empower the end-user and their community?
4. Consider Context — all knowledge is situated; social, cultural, and material context always matters. D'Ignazio and Klein argue that visualisation practice must move beyond the fantasy of a view from nowhere to acknowledge the specific contexts in which data is produced, represented, and received. The Dear Data project (Lupi and Posavec) exemplifies this: two designers exchanged hand-drawn data postcards for a year, each one deeply personal, contextual, and embodied. Design questions include: How can we leverage human-centred and participatory design to learn about and with our end-users? What kinds of terminology, symbols, and cultural artifacts have meaning to them?
5. Legitimise Embodiment and Affect — mainstream data visualisation often suppresses emotion and embodiment in the name of objectivity. Feminist data visualisation argues that emotional bonds with data stories are not a weakness but a strength; that acknowledging the body as a source of knowledge produces richer, more honest representations. This principle invites designers to ask: How can this visualisation create emotional resonance? How can it acknowledge that data describes real human lives and real bodily experiences?

Make Labour Visible

D'Ignazio and Klein add a sixth principle — Make Labour Visible. This addresses data provenance, attribution, and the often-invisible work involved in data science: crediting data sources, acknowledging the labour of data collection and cleaning, recognising emotional and care work, and making transparent the human effort behind apparently automated systems. Making labour visible challenges the mythology of data as naturally occurring or self-generating.

Bodies and Data

A critical dimension of feminist data visualisation is its attention to the relationship between data and bodies. Hill (2020) identifies five ways bodies are incorporated into or absent from data visualisations:

• Bodies are extracted — data is taken from bodies (e.g. biometric data, health records, behavioural traces) and rendered as abstract, 'objective' information, severed from its material origins and the lived experience of the people it represents
• Bodies are absent — certain bodies are systematically excluded from data collection, reinforcing biases and producing gaps in our knowledge of marginalised populations
• Bodies go unaccounted — some bodies are less likely to be formally documented or recognised by official data systems, leaving incomplete and distorted records
• Bodies are rendered invisible — dominant viewpoints shape who is represented in data and how; bodies that do not conform to normative categories are often suppressed or misrepresented
• The body of the viewer — visualisation conventions can mislead or disorient viewers, particularly those whose embodied knowledge does not match the assumptions encoded in the design; this can produce misinterpretation and misinformation

Recognising the relationship between data and bodies is essential for developing more ethical and representative data visualisation practices.