Beyond the margin of error
A systematic and replicable audit of the TikTok research API
Bibliographic Data
| ID | 20509403 |
|---|---|
| Authors | George D Pearson (0000-0002-9530-7851, American Legacy Foundation, corresponding author), George D H Pearson, Nathan A Silver (American Legacy Foundation), Nathaniel A Silver (0000-0002-0325-2559, American Legacy Foundation), Jessica Y Robinson (University of Oslo), Mona Azadi (0000-0001-8944-7471, American Legacy Foundation), Barbara A Schillo, Barbara Schillo (0000-0003-0847-2996, American Legacy Foundation), Jennifer M Kreslake (0000-0001-6605-1327, American Legacy Foundation) |
| Year | 2025 |
| Volume | 28 |
| Issue | 3 |
| Pages | 452-470 |
| Publication date | 2025-02-17 |
| Open Access | No |
| Type | ARTICLE |
| Venue | Information Communication & Society (JOURNAL) |
| Journal identifiers | ISSN: 1369-118X • E-ISSN: 1468-4462 |
| Publisher | Informa UK Limited (PUBLISHER • GB) |
| DOI | 10.1080/1369118x.2024.2420032 |
| OpenAlex | W4404148247 |
| Language | EN |
| Citations received | 13 |
| References cited | 27 |
The availability of an accurate API is an essential tool for research examining the content and dissemination of messages on large media platforms in the twenty-first century. TikTok launched their research API as part of a commitment to data transparency and a goal of protecting youth and vulnerable populations from harmful content. A systematic audit of TikTok’s API for researchers was performed comparing the API versus information displayed on the TikTok website (i.e., front-end), specifically examining metadata engagement (views, likes, shares, and comments) and assessing the availability of videos. Two research teams with unique access accounts separately audited the API in two distinct contexts: tobacco and elections. Results showed significant discrepancies between the front-end and API metadata for data collected before July 2024. Discrepancies were most severe for view counts, where the API regularly returned values multiple orders of magnitude lower than those displayed on the front-end. Significant discrepancies were found for all four engagement metrics in both datasets. Furthermore, publicly available videos were found in the front-end that were not returned by the API, and errors were found in how the API filters results by account usernames. The implications of these issues for researchers and TikTok’s efforts to support the research community are discussed
Audit · Business · Statistics · Systematic error · Computer Science · Digital Mental Health Interventions · FinTech, Crowdfunding, Digital Finance · Mathematics · Mobile Health and mHealth Applications · Psychology · Accounting
Where there’s smoke (less), there’s fire
Populism over partisanship on TikTok
Harnessing Big Data, Hindered by Bias
‘A multitude of sounds and voices’
TubeStats and TokStats
Mapping Platform-Mediated Proximities With the TikTok Global Observatory
Experimentation on TikTok, Standardisation on Reels? Party Short-Form Video Use in the 2024 UK General Election
Forgetful by design? A critical audit of YouTube’s search API for academic research
From information exposure to legal consciousness
European elections on TikTok
Exploring temporal dynamics in digital trace data
Price, proficiency, or permission
Finally, Access
Validation of Database Search Terms for Content Analysis
On the Psychology of TikTok Use
Stitching Politics and Identity on Tiktok
TikTok and public health
TikTok and Political Communication
After the ‘APIcalypse’
Discreet shipping’ on TikTok enables selling of e-cigarettes to youth
Analyzing radical visuals at scale
Research Note
Beyond challenges and viral dance moves
Computational Research in the Post-API Age
The Use of TikTok for Political Campaigning in Canada
Where Have All the Data Gone? A Critical Reflection on Academic Digital Research in the Post-API Age
Studying Reddit
| Citations per year | 13 |
|---|---|
| Citation span | 2025 - 2026 (2) |
| Highly cited | No |