How many times have you read a paper with the reassuring phrase: “Data is available upon request”?
It sounds promising, especially if you want to reuse the dataset, reproduce the findings, or explore it out of curiosity (even just for fun). Just like any motivated researcher, the first thing you would do is email the corresponding author asking whether the data can be shared. If you are too ambitious, you might politely ask for the full replication package: data and code.
Then you wait. You refresh your inbox. Days pass. Weeks pass. Months pass. And your inbox stays empty.
At some point, you start questioning yourself: Did I read that statement correctly?
It turns out that the more we encountered this pattern, the clearer it became that “available upon request” may sometimes translate to something else in practice: “not available upon request.” This pattern is consistent with prior evidence from psychology and economics, showing that authors often do not share data or supplementary materials even when papers claim they are available upon request (Hussey, 2023; Krawczyk & Reuben, 2012).
To what extent is research on tropical diseases transparent and credible?
This has also been our experience in a project focusing on research on tropical diseases. In our previous blog post, “The Reproducibility Gap in Tropical Disease Research,” we introduced our ongoing project aiming to examine the transparency and credibility of published research in this area.
Our central question is simple: To what extent is research on tropical diseases transparent and credible?
Our project is not just an intellectual exercise. Concerns about restricted access and weak data-sharing norms have been raised across other health-related research areas as well, where “available upon request” is increasingly seen as insufficient for reproducibility and scientific progress (Langille et al., 2018). In particular, research on tropical diseases often relies on clinical trials, randomized controlled trials, and field experiments, all of which are costly and time intensive. This work is sometimes funded through research grants or from donors who care about saving lives. More importantly, public health policies and interventions, including vaccines and treatments, are often designed and implemented based on this evidence. The impact of this research is not only theoretical, but also real. That is why it is critical to know whether these decisions are based on accurate and trustworthy foundations.
In our sample, we identified 553 causal and interventional studies published in leading public health and general-interest journals. Out of these, only 23 studies provided full replication packages. The remaining studies were missing data, code, or both.
As a reminder, replication materials were unavailable under one of several common scenarios:
- Only data (or, in less common cases, code) was provided
- The paper explicitly stated the data were “available upon request”
- A link to replication materials was provided, but did not work, or did not include what was promised
- The paper referenced supplementary materials, but no replication package was attached
- The availability of replication materials was unclear (unknown)
- Data were restricted and could not be shared
For all studies where access might have been possible (i.e., categories 1–5), we decided to follow up directly with corresponding authors. The highest hopes we had were for categories 1, 2, and 3, where replication materials seemed most likely to be shared. However, we put our hopes aside and emailed authors across all (the first) five categories. That left us with 448 studies requiring outreach.
The reality slap
We sent 448 emails but received 9 replies. That is a response rate of just 2%. Here, by “reply”, we simply mean a reply. We are abstracting from whether it was positive or negative in terms of sharing the replication package.
What struck us most was not only the low response rate, but, in fact, who responded. None of the replies came from papers that explicitly stated the data were “available upon request,” or from papers that provided data without the code, or from papers that provided broken links to replication materials. In brief, none of the responses came from exactly the cases where we would most expect replication package access to be feasible.
What does this mean?
At the very least, this is a red flag.
When research has serious implications for public health, researchers, in principle, should not have to rely on unanswered emails to reproduce or replicate published evidence. More broadly, this should not be the norm in any scientific field. Many explanations may exist for the low response rate: limited time, lost files, unclear responsibilities for sharing, data governance concerns, or even concerns about what reproduction and replication attempts might reveal, or …
Regardless of the reason, the outcome is the same: access fails in practice. And that brings us back to the question at the heart of this project:
To what extent is research on tropical diseases transparent and credible?
Stay tuned!
References
Hussey, I. (2025). Data is not available upon request. Meta-Psychology, 9.
Krawczyk, M., & Reuben, E. (2012). (Un) available upon request: field experiment on researchers’ willingness to share supplementary materials. Accountability in Research, 19(3), 175-186.
Langille, M. G., Ravel, J., & Fricke, W. F. (2018). “Available upon request”: not good enough for microbiome data!. Microbiome, 6(1), 8.