How can teams make credible new claims when analyzing data collected by others for different purposes?

Host: Welcome to A.I.S. Insights — powered by Living Knowledge. I'm Anna Ivy Summers.

Expert: And I'm Alex Ian Sutherland.

Host: Today we are examining a study presented at the International Conference on Information Systems in 2026 by Tim Kilgus, titled "Authorization Work: The Case of Clinical Trial Data Repurposing." The study investigates how researchers take secondary data shared through platforms and turn it into defensible claims.

Expert: That's right. Organizations often assume that once data is standardized and made available on a central platform, teams can readily use it to answer new questions. But in reality, when data leaves its original collection environment, it loses the specific protocols, regulatory approvals, and institutional context that made it trustworthy in the first place.

Host: And simply fitting the data into a new analytical model does not automatically make the resulting conclusions credible to outside evaluators, does it?

Expert: Exactly. Technical alignment—formatting data to fit a task—is only part of the process. In scientific and high-stakes settings, analysts must also prove to an evaluating community that their new claims are legitimate. Kilgus identifies the labor required to achieve this as "authorization work."

Host: How did the author examine this challenge empirically?

Expert: The author conducted an interpretive qualitative case study of Vivli, the largest global platform for clinical trial data sharing, hosting studies that cover millions of patients. The empirical material included thirty-one semi-structured interviews with data users across academia and industry, along with hundreds of formal data requests and resulting publications.

Host: What did the study find regarding how users make secondary data findings defensible?

Expert: The study identified three primary mechanisms of authorization work: anchoring, importing, and warranting.

Host: Let's walk through those. What does anchoring involve?

Expert: Anchoring means tying the secondary analysis back to authorized reference points. For example, before running new models, researchers frequently reproduced the published results of the original trials using the raw datasets to verify data correspondence. They also selected balanced trial sets and maintained individual trial identities rather than simply pooling data, preserving the standing of the original studies.

Host: And the second mechanism, importing?

Expert: Detached datasets do not carry the interpretive judgment of the original collectors. Importing involves bringing in community-specific resources from outside the dataset. In the study, researchers consulted disease expert panels to choose variables, adopted recognized disease-specific endpoints, and applied established methodological conventions to satisfy the expectations of their target evaluators.

Host: That leads to warranting. How does that operate?

Expert: Warranting licenses the inferential leap from the data to the final claim. Because repurposed data lacks prospective experimental design, researchers explicitly framed their findings as hypothesis-generating rather than confirmatory. They disclosed the structural biases of secondary analysis and paired their findings with independent observational studies, external validations, or follow-up trials.

Host: This highlights that while digital platforms provide scalable access to data, they cannot automate credibility.

Expert: Precisely. Access does not equal authorization. When the distance between the data's origin and the evaluating community grows, the effort required to defend derivative claims increases.

Host: What are the key takeaways for practitioners managing secondary data or enterprise data platforms?

Expert: Data governance strategies must account for claim defense, not just data preparation. When organizations repurpose external or historical data for predictive models and strategic decisions, teams should validate baseline integrity, integrate relevant domain experts into analysis design, and calibrate the strength of their conclusions to what retrospective data can legitimately support.

Host: It demonstrates that making data usable is distinct from making the resulting decisions defensible before external stakeholders.

Expert: That is the central point. Without deliberate authorization work, technically aligned data can still fail external scrutiny.

Host: Thank you, Alex, for breaking down this study, and thank you to our listeners. Join us next time on A.I.S. Insights — powered by Living Knowledge.