Price, proficiency, or permission: assessing platform data access for election research

Summary

This article maps the landscape of social media data access tools available to election researchers, cataloguing 72 programs and analyzing them along two axes: independent versus collaborative collection, and content (production) versus consumption data. Lukito and Mimizuka argue that access is gated by a triad of barriers — price, proficiency, and permission — that systematically disadvantage less-resourced and non-Western scholars. They document the chronic instability of platform-permission regimes (epitomized by Meta’s serial cycling through Social Science One, CrowdTangle, FORT, and the Meta Content Library), the near-absence of consumption-data infrastructure, and the WEIRD-centric geography of available tools. Rather than endorsing any single solution, they advocate a plurality of access approaches sustained through multi-institutional archives and regulatory mandates modeled on the EU’s Digital Services Act.

Key Contributions

• A two-dimensional framework (independent↔collaborative × production↔consumption) for classifying data access regimes.
• A curated, publicly released inventory of 72 election-relevant data access tools, coded along 10 dimensions.
• The “price, proficiency, permission” heuristic as a diagnostic for structural inequities in computational election research.
• Policy recommendations including legal protections for independent collection, plural access regimes, and a proposed global public archive of politicians’ social media activity.
• An argument linking methodological access constraints to epistemological limits on what election research can know.

Methods

The authors merged the IDDP platform transparency tracker with Brandon Silverman’s list of 253 research tools, supplemented by SCOPUS and Google Scholar searches, filtering to tools used in at least one peer-reviewed study or preprint. This produced 72 unique programs, each manually coded on dimensions including active status, platforms covered, collection method, data type, pricing, geographic availability, and lifespan. Analysis proceeds via the 2×2 framework, with qualitative case studies of pivotal programs (CrowdTangle, Meta Content Library, Pushshift, Social Science One, the Meta 2020 Election Project, the National Internet Observatory) illustrating each quadrant’s dynamics.

Findings

• Only 6 of 72 tools focus on consumption data versus 66 on production — a stark epistemological asymmetry for understanding how social media affects voters.
• Platform-provided production tools (24, or 33.3%) dominate the permission quadrant; many DSA-mandated programs remain in beta and EU-restricted.
• Independent collection is the largest category (n=42) but fragments into paid third-party services, scraping/unofficial APIs, and donation tools — each privileging a different barrier (price, proficiency, or participant access).
• Meta has cycled through four major access initiatives; CrowdTangle’s August 2024 shutdown disrupted research immediately before the U.S. election, exemplifying “independence by permission.”
• Independent consumption efforts (NIO, PORT, donation studies) are promising but constrained by sample skew, infrastructure cost, and ethical complexity.
• Most programs serve U.S. and European researchers, reinforcing WEIRD inequities.

Connections

This paper sits squarely in the “post-API age” conversation about platform power over research, complementing Freelon2024-sc and Gonzalez-Bailon2024-rq on the precarity of platform-mediated access, and Ohme2026-nv on user-centric data donation as an alternative paradigm. Its critique of permission regimes resonates with Rieder2025-ju and Rieder2026-pp on API politics, while its DSA-oriented policy turn connects to Lewandowsky2026-ob and broader European regulatory scholarship; the documentation of CrowdTangle’s collapse echoes concerns developed in Bak-Coleman2025-pm and Murtfeldt2025-wu about reproducibility under unstable platform infrastructures.

Podcast

A research-radio episode discusses this paper: Listen