I recently came across discussions suggesting that some meta-analyses include unpublished data contributed by the study authors themselves. I’m curious about the scientific and ethical perspectives on this practice.
Specifically:
- Is it considered acceptable to include authors’ own unpublished data in a meta-analysis?
- Are there guidelines on how to handle such data to ensure transparency and minimize bias?
- Should the methods and results of these unpublished datasets be fully disclosed, or can they be summarized in a more limited way?
If anyone has examples of published meta-analyses that have followed this approach, or references discussing the methodology, I would greatly appreciate it.
Thank you for sharing your insights!
Before I would include unpublished data in a systematic review/meta-analysis, I would run the meta-analysis with and without the unpublished data. If the overall conclusions are not affected by inclusion of the unpublished data, I would probably feel OK about including those data. However, if inclusion of the unpublished data changes the meta-analysis conclusions, I would feel less confident about the results of the meta-analysis. In such a situation, I would probably feel compelled to report the results both ways, i.e., with and without the unpublished data.
Including unpublished author data in meta-analyses is generally considered acceptable and can reduce publication bias. Such data should be rigorously evaluated for quality and risk of bias, and permissions must be obtained from authors. Results should transparently present analyses both with and without unpublished data. This approach enhances the validity of pooled evidence, though controversies remain.
Including unpublished data in meta‐analysis is scientifically and ethically acceptable, and in many cases desirable, provided that:
- It is done in a transparent, pre‑specified, and methodologically rigorous way
- Risk of bias is assessed
- Sensitivity analyses are used
- Ethical/legal issues (data ownership, privacy, etc.) are respected
It is not unqualifiedly acceptable — you can’t just grab any unpublished data and put it in without thinking about these issues, because that can introduce other forms of bias or error.
I think that is acceptable practice. It promote transparency in research by acknowledging the existence of all relevant studies, regardless of their publication status.
- Badges
- Science topic
- Similar topics
- Engineering
- Materials Engineering
- Powders