A new large-scale, AI-powered study published last week by Elsevier reveals that the share of global #STEM researchers who are women grew quite a bit over the last two decades, from 26% in 2000 to about 39% in 2022. However, at this current pace, women’s participation in STEM research globally will not reach parity with men’s until 2052.
A lack of gender diversity in scientific research can lead to biased research outcomes, skewed funding, policies that fail to prioritize women’s issues, treatments that are less effective or even harmful for women, stifled economic growth, and more. Fighting back against this kind of under-representation in STEM is not just a matter of fairness; it's vital for scientific progress, societal well-being, and global advancement. https://lnkd.in/euFZ4gED
🔬 A common lab tool, a hidden source of irreproducibility?
Magnetic stirrers are everywhere in the lab - from organic synthesis benches to nanoparticle reactors. But could this trusted device be quietly introducing inconsistencies in our results?
A recent study in JACS Au highlights how magnetic stirrers can affect the reproducibility of chemical reactions, even when all other parameters are held constant.
⚠️ Key findings:
In parallel nanoparticle synthesis, reaction rates and particle sizes varied despite identical conditions.
In catalyst preparation, nanoparticle morphology differed based on where the beaker sat on the stirrer.
In organic cross-coupling reactions, conversion rates fluctuated across adjacent vessels on the same plate.
These surprising results call attention to spatial variation in stirring efficiency - a factor often overlooked in experimental design.
To address this, the authors propose a simple control experiment to help researchers identify and mitigate these effects.
Given how integral magnetic stirrers are across chemistry, materials science, and the life sciences, this study is a valuable reminder: even the most familiar tools deserve a second look when it comes to reproducibility.
📖 Worth a read if you're working at the bench or interpreting lab-scale results.
#chemicalresearch#nanomaterials#organicsynthesis#materialscience
SCORE, a collaboration of 865 researchers, is now released as three papers in Nature, six preprints, and a lot of data (https://cos.io/score/). SCORE examined repeatability of findings from the social-behavioral sciences and tested whether human and automated methods could predict replicability.
Our sample was quantitative claims from 3900 papers published from 2009 to 2018 from 62 journals across criminology, economics, education, finance, health, management, marketing, organizational behavior, psychology, political science, public administration, and sociology.
Five independent teams made predictions about the replicability of those claims, two using human methods, three using AI methods. Their methods were diverse: prediction markets, structured elicitations, bot-based markets, knowledge graphs, semantic parsing.
A representative subset of 600 claims were available for repeatability tests: reproductions (same data, same analysis), robustness tests (same data, different analyses), and replications (same question, different data).
For reproducibility, we could obtain data for only 24% of the 600 papers. Of the 143 papers (551 claims) assessed, we precisely reproduced 54%, and approximately reproduced 74%. We were much more likely to succeed if authors shared data and code, versus just data or if we had to reconstruct data from original sources. Paper: https://lnkd.in/e-JC2yjU
For robustness, 34% of reanalyses showed the same result within a narrow tolerance (+/- .05 Cohen’s d), and 57% with a wider tolerance (+/- .20). Limiting to statistical conclusions (p<.05?), 74% of reanalyses reached the same conclusion, 24% observed no effect, and 2% observed an opposing effect. Paper: https://lnkd.in/eeZ9kyuP
For replicability, we tested findings from 164 papers and successfully replicated 49% of them with the common statistical significance criterion. Original studies had an average effect size of r = 0.25, replication studies r = 0.10. Paper: https://lnkd.in/eeZ9kyuP
Best performing human methods achieved about 75% accuracy predicting replication outcomes. AI methods got pretty good at matching the general patterns of human judgments, but struggled to discriminate claim replicability.
Credibility is built from many contributing elements—transparency, bias control, and methodological rigor—not a single underlying trait.
Thanks and congratulations to the amazing community of researchers who contributed in so many ways to SCORE. This project embodies the collaborative spirit of continuous questioning and improving of research. Special acknowledgement to Adam Russell who conceived and launched SCORE at DARPA.
The durability of research findings can be cast in terms of three Rs. Findings should be reproducible (the same type of analysis using the same data should produce the same result); replicable (redoing an experiment to collect fresh data should produce the same result); and robust (alternative analyses using the same data should draw the same conclusion).
This week's issue of Nature Magazine includes four papers that look at these three Rs in the social and behavioural sciences. Three of them are are an outcome of nearly US$8 million in DARPA funding provided back in 2019 to the Systematizing Confidence in Open Research and Evidence (SCORE) programme. The fourth paper reports the outcome of a series of one-day ‘replication games’ workshops organized around the world since 2022 by the Institute for Replication, a virtual, non-profit network.
The results are sobering: researchers could replicate the results of only half of the studies that they tested. Clearly, current rates of replicability and reproducibility leave much room for improvement, to put it mildly. But rather than despair or throw the proverbial stones, we should focus on the value that such methodical retrospection offers. Any insights that can make the path to reliable findings more reliable will accelerate progress. Looking back at previous work is as necessary as looking ahead; it should be funded and it should be published. Rigorous practices to do so demonstrate the scientific method at work, as the papers just published show so clearly.
You can see the papers, our editorial as well as the news coverage, and a Q&A with Brian Nosek, a lead on the SCORE project here: https://lnkd.in/emgPeiXb
Today, as we mark the #InternationalDayOfWomenInScience, I find myself thinking back to my childhood self: a gaze full of curiosity and a desire to understand the world, holding a pen as if she already sensed that science begins by asking questions. That girl, full of dreams and aspirations, reminds me why it is so important that no scientific vocation should encounter barriers along the way.
Every discovery, breakthrough, and innovation we celebrate is born from talent. Unfortunately, that talent has not always had the same opportunities to flourish. The latest She Figures report, published in 2025 by the European Commission, reminds us once again that achieving equality in science remains an urgent challenge.
Despite the progress made, the data is clear:
• Women represent only 34% of researchers in the EU.
• 98% of scientific publications do not take gender differences into account.
• Only 9% of inventors in Europe are women.
• Women’s presence in authorship drops from 48% at early career stages to 36% in senior positions, reflecting the persistent “leaky pipeline,” meaning the gradual loss of women as they progress in their scientific careers.
These figures show that inequality is not only a matter of access, but also of retention, recognition, and leadership. And when talent is lost along the way, we all lose.
Every step toward equality unlocks talent and strengthens society. And although there is still progress to be made, reports like She Figures remind us why it is so important to keep moving forward.
#SheFigures#WomenInScience
Yesterday was the International Day of Women and Girls in Science, so I decided to take a data-driven look at gender balance in one of the most prestigious scientific journals in the world.
I pulled over 30,000 research articles published in Nature from 2002 to 2026 via PubMed, extracted every author's first name, and used name-based gender classification to estimate the fraction of women in first and last authorship positions, the two slots that matter most in science.
Here's what the data shows:
- Female first authorship rose from ~22% in 2002 to ~35% by 2020
- Female last authorship (typically the senior/PI role) went from ~15% to ~25% in the same period
- The upward trend peaked around 2020–2021, right in the middle of the pandemic. While COVID disrupted science in many ways, it seems women were not only holding their ground in high-impact publishing, they were contributing more than ever to the breakthroughs that made it into Nature.
But let's be honest: we're still far from 50%.
The good news is that the trajectory is clear: every generation of women in science is pushing the door a little wider for the next. And the breakthroughs they're leading aren't waiting for parity to arrive.
Half of laboratory mice are not what scientists think they are.
A new genetic survey published in Science analyzed 341 mouse strains from the Mutant Mouse Research and Resource Centers. 47% of strains were genetically inconsistent with how they were described.
7% belonged to an entirely different strain.
26% belonged to a different substrain.
Nearly 10% carried genetic changes, including reporter genes, that were not in the strain name at all.
The reproducibility debate keeps circling the same target. Animal models. NAMs. Translation. Predictive value.
This study reframes the problem.
If you cannot identify the model you used, every downstream comparison is compromised. Cross-study replication, regulatory submission, mechanistic interpretation. All of it depends on knowing what the animal actually is.
One commentator quoted in the article argues that replication still holds if two labs use the same misnamed strain. True for narrow replication. False for everything else. You cannot generalize to humans, compare across cohorts, or interpret mechanisms when the genetic background is undocumented.
This is not only an animal research problem. It is a metadata problem.
The same logic applies to organoids, organs-on-chip, and computational models. A NAM with undocumented donor variability or unspecified culture parameters fails for the same reason a mislabelled C57BL/6 fails. Different layer of the stack. Same root cause.
Data infrastructure is the precondition for everything else. ARRIVE 2.0. SEND. FAIR. Minimal metadata sets. These are not bureaucratic overhead. They are the only way any model, biological or computational, becomes auditable.
The Pistoia Alliance Minimal Metadata Set Working Group exists for this reason. You cannot fix translation without first fixing identification, provenance, and traceability.
The next decade of preclinical research will not be decided by the animal vs NAM question. It will be decided by which organizations build the data infrastructure to know what they are actually working with.
Half the mice were wrong. The metadata was missing.
Reference: Pardo-Manuel de Villena, F. et al. Science 392, 698–700 (2026).
Improving gender diversity in research and publishing has the potential to diversify research questions, research methodology, improve research impact and outcomes for all, and broaden participation in research and publishing. And that is why we set out to understand gender in the context of publishing in our journals. I am pleased to share our latest report, Closing the Gender Gap: Peer Review at Nature Portfolio, where we set out to understand what gender representation looks like among corresponding authors and peer reviewers through the publication process across Nature, Nature research journals, Nature Reviews journals, Nature Communications, Communications Series journals, and npj Series journals.
Highlights below:
- Women represent 18% of submitting corresponding authors of original research articles across the journals analyzed; the proportion increases to 22-23% in more inclusive journals. The gender gap is consistent with what has reported for other selective journals.
- A higher percentage of women are submitting corresponding authors in psychology (33%), medicine & public health (23%), and life sciences journals (22%) - this trend tracks with the relatively higher proportion of women researchers in these fields.
- There is no evidence of negative gender bias in the editorial or peer review process.
- Proactive editorial efforts contribute to improving representation of women in editorially commissioned works, and as peer reviewers.
Women constitute 34% of corresponding authors globally (according to the SHE Figures 2024 report), with regional variation. Across disciplines and regions, we know from multiple benchmarking reports that the gender gap widens with seniority, but the gap is considerably narrower in cohorts of early career and mid-career stage researchers. So, if we are going to narrow the gender gap in research publishing, engaging with researchers earlier in their research journey is vital. We know, for example, that engaging early career researchers in co-review initiatives at Nature Portfolio journals boosts gender diversity.
While we are encouraged that editorial strategies are having a positive impact, addressing the gender gap will need change across the ecosystem to reinforce publisher efforts to make publishing more representative of the research community.
Read the report here: https://lnkd.in/g4m_Dzuk
Find out more in our blogpost here: https://lnkd.in/g7UDJqA7
If a student reads a typical psychology paper, picks an experiment, and tries to replicate it, they have roughly a coin-flip chance of success.
That's the punchline of a decade of metascience.
This is week 3 of a weekly Experimentology series I'm running through the spring and summer, sharing one chapter at a time. Today is Ch 3, on reproducibility and replicability — and what we know about both.
First, the terminology, because it gets confused. Reproducibility means: same data, same analysis, same numbers. Replicability means: new data, same analysis, similar result. Robustness means: same data, different analysis, similar result. They're related but answer different questions.
Reproducibility is the easier of the two to study, but only when authors share their data and code. A set of studies led by Tom Hardwicke used papers from journals that required data sharing and took ~60 articles and tried to reproduce one analysis from each. About a third were fully reproducible without contacting the authors. After author contact (often extensive), the rate climbed to roughly 62% — and the remaining gap was values that no one, including the original authors, could reproduce. The labor involved was enormous: 5–10 hours per paper.
Replicability is the harder problem. The Reproducibility Project: Psychology (Nosek 2015), the first big systematic effort, tried to replicate 100 studies. About 36% of replications produced a statistically significant result in the same direction as the original. Subsequent efforts in economics and in Science/Nature psychology landed around 60% — better, but still much lower than naive expectations would predict. (And the metric of "significant in the same direction" is itself a coarse one; the chapter discusses better ones.)
Why do replications fail? Several explanations get invoked routinely — context sensitivity, experimenter expertise, hidden moderators. The metascience evidence for these as the main drivers is, in our reading, fairly weak. The bigger explanations are analytic flexibility (p-hacking) and publication bias. Both bias the published literature toward effects that are inflated relative to the truth, which by construction makes replications underpowered.
Where does this leave us? We don't think "crisis" quite captures it. Things haven't gotten worse; if anything they've gotten better. Open science practices — pre-registration, materials sharing, code and data sharing — are concrete responses, and they're spreading.
📖 Read Ch 3: https://lnkd.in/gRCpAQHP#OpenScience#ResearchMethods#Psychology#Metascience#HigherEducation