Data Activism by Design

Open data is not the same as legible power. Over a legislative year, hundreds of votes by nearly 180 members produce well over one hundred thousand entries—too dense for static charts, too important to stay opaque. The question with public value is simple and charged: Do party members vote as one—or not—and where does the line break?

Our team collected official roll-call records, cleaned and normalized the dataset, and preserved edge cases such as formally recorded misvotes (logged publicly but not retroactively corrected). Two tidy CSVs—votes and descriptions—support a reproducible pipeline: same method, next session. Scraping notes and data decisions are documented for auditability.

A compact civic dashboard with linked views:

• • a party-by-vote table for quick pattern scanning;
• • two synchronized summaries—party-level and member-level—updating on selection;
• • Encodings stay consistent; a distinct hue elevates Mixed to make intra-party disagreement visible at a glance.
• • A unified legend and hover annotations reduce cognitive load without dumbing down the story.

Design decisions are treated as inclusion tactics: a color-vision-safe palette, generous spacing, readable type, rounded percentages, and highlight-on-demand instead of constant noise. Accessibility here is not cosmetic compliance—it’s the line between who gets to read power and who doesn’t.

• • Data feminism: make power visible, design for inclusion, and document the labor behind data.
• • Data & agency: build tools that expand people’s capacity to act—not just to look.
• • Visualization craft: clear encodings, linked views, and cognitive load reduction as pathways to civic comprehension.

• • Journalists & NGOs: embed filtered views in reporting, point to where Mixed spikes and ask why.
• • Citizens & classrooms: explore a real session, trace party unity vs. dissent without statistical jargon.
• • Lawmakers & staffers: verify narratives against the public record, use disagreement patterns as prompts for explanation.

“Party unity” is debated constantly yet rarely seen across a whole session. Linking per-vote context to party and member summaries invites a more nuanced reading of representation—agreement, abstention, absence, and especially disagreement inside the line. The dashboard does not argue a position; it exposes structure so others can act.

Public notes on misvotes remain annotations, not corrections; interpretation must keep that nuance. The model is Denmark-specific; adoption elsewhere needs localized party systems and vote taxonomies. Any civic tool can be misread—explanations and context are part of the product, not a blog footnote.