Post by ASAPbio Fellow Jackie Carozza
Preprints provide an outlet for sharing scientific results outside of the traditional peer review and journal publication process – not only can researchers post their manuscript early before going through peer review, but they can also post work that may traditionally have found less favor with journals, such as articles-in-progress and negative, confirmatory, or contradictory results. But are researchers taking advantage of opportunities to share nontraditional results, and what are opportunities for the future?
I approached that question by analyzing data the preprint server bioRxiv collects and makes publicly available on its API. Authors uploading their preprint to bioRxiv must categorize their work into one of three article types: new results, confirmatory results, or contradictory results. Of course these categories are oversimplifications, since a preprint may include two or more types of data but an author must choose one category. Still, these categories give us a valuable starting point to assess how authors are using bioRxiv to post nontraditional results – let’s take a look.
Preprints categorized as “confirmatory” or “contradictory” make up less than 2% (and dropping) of preprints on bioRxiv
Perhaps unsurprisingly, new results represent the vast majority (98.2%) of all preprints posted on bioRxiv, followed by confirmatory results (a distant second at 1.32%) and finally contradictory results (0.48%). Interestingly, the percent of confirmatory and contradictory preprints out of all preprints is steadily dropping over time.
Assuming there is considerably more than 2% confirmation and contradiction happening in science, I wonder if researchers (1) do not post their non-traditional results or (2) label confirmatory or contradictory results as new results, either because of potential bias in assigning more value to new results or because the article contains a mixture of findings, also including new results.
Preprinted confirmatory and contradictory results are published less often in traditional journals than new results
What happens to confirmatory or contradictory results after they are posted to bioRxiv? Do they get published in journals at similar rates as new results?
The journal publication status of all preprints is also listed on the bioRxiv API, thanks to bioRxiv’s automatic title-matching search algorithm. This metric will underestimate the number published because if the title changes significantly in the published journal article, the search algorithm won’t find it (by the way, you can email bioRxiv to ask them to update the link from your preprint to the journal publication if it’s not already there!).
I looked at the percent of preprints that had been published in journals over time. Journal publication rates steeply decline after 2019, likely reflecting the time articles spend under review. Let’s look at the data before 2019 to compare confirmatory/contradictory publication rates to new results. Strikingly, while about 70% of new results are published in journals, the rate of journal publication for confirmatory and contradictory results hovers much lower, around 50-60%. This could suggest that authors try but fail at higher rates to publish their confirmatory and contradictory results in journals, or that they don’t submit them to journals in the first place.
What can we learn from these trends for the future?
Researchers are posting (self-categorized) confirmatory and contradictory results – perhaps not very frequently, but they’re posting. Furthermore, publication rates in journals are lower for contradictory and confirmatory results compared to new results. One could interpret this trend pessimistically (“Science doesn’t want to hear about anything but shiny new data”), or optimistically: confirmatory and contradictory results that are not being published in journals can still see the light of day.
Submitting to a journal and going through peer review takes time, money, and energy and can feel especially insurmountable for non-traditional data – but researchers don’t necessarily need to go through that to get the results out. Having a publicly accessible record in the form of a preprint enables others (including collaborators and lab mates) to access it in the future, preserving valuable knowledge and preventing unnecessary duplication of effort.
There is much more opportunity for growth in preprinting non-traditional articles, including confirmatory, contradictory, and “negative” results. If you’re looking for something actionable, consider submitting or encouraging others to submit a preprint reporting negative/null or inconclusive results to the ASAPbio competition “Make your negative result a preprint winner” to contribute to this growing body of literature.
A few details on the methods:
I fetched all articles published from 2013-11-01 to 2022-11-01 from the biorxiv API (N = 239,909), then sorted out duplicate versions by only analyzing version 1 (final N = 174,208).
About the author
Jackie Carozza, one of the 2022 ASAPbio Fellows, is interested in promoting ways to communicate in research in a truthful, digestible, and low-stress way. She is a senior scientist at the Arc Institute in Palo Alto, California, where she studies the innate immune response to immune threats such as cancer and inflammation.