Metascience Matters

Randy Ellis

The research integrity channel for people who want science to work. Metascience Matters covers scientific fraud, replication failures, peer review, research incentives, AI/data reliability, and the people building better science. Hosted by Randy Ellis, a computational biologist and neuroscientist. New interviews, clips, and explainers on how science breaks and how to fix it. Guests, tips, and collaborations: metasciencematters@gmail.com.

Épisodes

  1. -3 j

    Replication research, diamond open access publishing | Lukas Röseler

    Lukas Röseler is the Managing Director at the Center for Open Science at the University of Münster. He is the Editor-in-Chief at the journal Replication Research, part of the Replication Journal Federation, is an active member of the Framework for Open and Reproducible Research Training (FORRT) community, and developed a video game called Guess the Replication, where players guess whether studies successfully replicated.  Replication Research: http://replicationresearch.org/ Replication Journal Federation: https://forrt.org/rjf/ Framework for Open and Reproducible Research Training (FORRT) community: https://forrt.org/ Guess The Replication: https://lukasroeseler.github.io/GuessTheReplication/ Dominance and Prestige: Meta-Analytic Review of Experimentally Induced Body Position Effects on Behavioral, Self-Report, and Physiological Dependent Variables: https://www.researchgate.net/profile/Robert-Koerner/publication/360577810_Dominance_and_prestige_Meta-analytic_review_of_experimentally_induced_body_position_effects_on_behavioral_self-report_and_physiological_dependent_variables/links/62823b133a23744a728598c2/Dominance-and-Prestige-Meta-Analytic-Review-of-Experimentally-Induced-Body-Position-Effects-on-Behavioral-Self-Report-and-Physiological-Dependent-Variables.pdf OUTLINE: 0:00 - Intro 2:10 - Leading the Münster Center for Open Science 6:17 - How Lukas became interested in metascience 10:53 - Cultural change in science 13:43 - Being Editor-in-Chief of Replication Research 20:00 - Funding the journal, diamond open access 24:55 - Prestigious journals' money extraction scheme 28:17 - Distributed institutional support for diamond OA journals 31:50 - The Replication Journal Federation 39:24 - Journals for replication, or replications for journals? 44:34 - Guess The Replication 50:57 - Prediction markets for replication 54:32 - Meta-analysis of body positioning effects/power posing 1:02:46 - If science were aligned to truth, how would researchers be evaluated? 1:14:36 - Advice for young people

    1 h 19 min

  2. 27 mai

    Data Provenance in Medical Machine Learning | Alexander Gibson

    Alexander Gibson is a PhD student at the Queensland University of Technology and the Australian Centre for Health Services Innovation studying the intersection of metascience and clinical machine learning. One of his focus areas is data provenance, the Who, What, Where, When, Why, and How of datasets, and how neglecting this can lead to bad outcomes in medical machine learning not only in research, but also for clinical practice and medical device approval.  CONTACT RANDY: Feedback: metasciencematters@gmail.com EPISODE LINKS: Alex's preprint on unreliable diabetes and stroke datasets: https://www.medrxiv.org/content/10.64898/2026.02.24.26347028v2 OUTLINE: 0:00 - Introduction 3:14 - The beginning of Alex's interest in clinical predictive modeling 5:05 - Alex's interest in metascience 6:42 - Choosing a dissertation topic/metrics hacking in machine learning 9:49 - Preprint on data provenance in medical datasets 12:33 - The diabetes and stroke datasets Alex investigated 16:46 - Major irregularities in the data 23:29 - TRIPOD+AI guidelines for auditing machine learning studies 25:26 - How unreliable studies can impact clinical practice and medical device patents 26:42 - Citation networks 27:37 - AI-generated formulaic medical machine learning studies 31:50 - Strategies for high-quality data provenance 33:53 - Patents citing unreliable studies, and how to integrate data provenance into peer review 35:23 - The biggest problems for clinical predictive modeling studies 37:02 - Resources and tools for improving rigor in machine learning 38:45 - Metrics reporting 40:45 - Choosing decision thresholds in predictive models 42:59 - The importance of clinical context in metrics reporting 45:21 - The unreasonable effectiveness of age and sex as predictors 47:53 - The roles of academia and industry in improving clinical machine learning studies 50:07 - Explanation versus prediction 52:51 - Advice and resources for students 54:27 - Outro

    55 min

  3. 22 avr.

    Fraud in Alzheimer's research | Charles Piller

    Charles Piller, an Investigative Correspondent for Science Magazine, is the author of the 2025 book Doctored: Fraud, Arrogance, and Tragedy in the Quest to Cure Alzheimer's. He has received more than 40 honors or awards from American Association for the Advancement of Science/Kavli Foundation, National Institute of Health Care Management, Society of Professional Journalists, National Academies/Eric and Wendy Schmidt, Los Angeles Press Club, American Society of Tropical Medicine and Hygiene, Gerald Loeb Foundation and UCLA, Western Publications Association, First Amendment Coalition, California State University, Computer Press Association, American Society of Business Press Editors, California Newspaper Publishers Association, National Council on Crime and Delinquency, and other organizations. His work has been nominated for the Pulitzer Prize nine times by STAT, the Los Angeles Times and The Sacramento Bee. CONTACT RANDY: Feedback: metasciencematters@gmail.com EPISODE LINK: Blots on a field?  https://www.science.org/content/article/potential-fabrication-research-images-threatens-key-theory-alzheimers-disease Picture imperfect https://www.science.org/content/article/research-misconduct-finding-neuroscientist-eliezer-masliah-papers-under-suspicion Doctored: Fraud, Arrogance, and Tragedy in the Quest to Cure Alzheimer's https://www.amazon.com/Doctored-Fraud-Arrogance-Tragedy-Alzheimers/dp/1668031248 OUTLINE: 0:00 - Introduction 6:05 - How Charles became a science journalist 7:58 - Charles' 2022 piece, "Blots on a field?" 14:39 - Responses from scientists to this piece 22:22 - Charles' 2024 piece, "Picture imperfect" 30:29 - How Charles began working on his book "Doctored: Fraud, Arrogance, and Tragedy in the Quest to Cure Alzheimer's" 33:15 - Charles' experience working with scientific sleuths 36:45 - The incentive for positive data in science 46:09 - Charles' NYT op-ed about how to combat fraud in Alzheimer's research; attacks from the Trump administration on science 53:22 - Charles' thoughts on registered reports and other ways to improve scientific incentives 58:34 - Upcoming work 59:22 - Advice for listeners 1:02:20 - Outro

    1 h 2 min

  4. 1 avr.

    Why So Many Cancer Biology Findings Fail to Replicate | Tim Errington

    Tim Errington is the Senior Director of Research at the Center for Open Science. He led the Reproducibility Project: Cancer Biology, Systematizing Confidence in Open Research and Evidence (SCORE), as well as the implementation and evaluation of initiatives such as Registered Reports, Registered Revisions, responsible conduct of research trainings, and open science badges. CONTACT RANDY: Feedback: metasciencematters@gmail.com EPISODE LINKS: Investigating the replicability of preclinical cancer biology (Reproducibility Project: Cancer Biology led by Tim Errington): https://elifesciences.org/articles/71601 Bayer replication study: https://www.nature.com/articles/nrd3439-c1 Amgen replication study: https://www.nature.com/articles/483531a Reproducibility in Cancer Biology: Challenges for assessing replicability in preclinical cancer biology (Companion paper to Reproducibility Project: Cancer Biology): https://elifesciences.org/articles/67995 What is replication?: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3000691 Study comparing standard reports and registered reports in psychology: https://journals.sagepub.com/doi/10.1177/25152459211007467 Blog post on the seemingly magical success of revision experiments: https://rajlaboratory.blogspot.com/2013/06/the-magical-results-of-reviewer.html Google's AI co-scientist paper: https://arxiv.org/abs/2502.18864 Machine-readable documents: https://journals.sagepub.com/doi/10.1177/2515245920970949 How open science helps researchers succeed: https://elifesciences.org/articles/16800 ZBW's Expedition to Open Science Land: https://expedition-open-science.org/ OUTLINE: 0:00 - Introduction 4:47 - Tim's origin story as a cancer biologist 6:38 - Initial interest in metascience 9:24 - Starting the Reproducibility Project: Cancer Biology 12:07 - How were the studies that were replicated chosen? 14:41 - Publishing the Reproducibility Project: Cancer Biology as a registered report 17:26 - Results from the Reproducibility Project: Cancer Biology 20:28 - Tim's experience throughout the years running the Reproducibility Project: Cancer Biology 25:21 - The difficulty of running cancer biology studies 27:54 - Judging whether a replication is successful 31:23 - What has the response to the Reproducibility Project: Cancer Biology been? 37:52 - Why aren't replication rates higher? 40:26 - Challenges of running cancer biology replication studies 45:43 - Caveats of preclinical disease models 49:13 - The incentive for positive data in science 57:05 - Systemic intervention vs. Individual policing 1:01:04 - The value of registered reports 1:07:38 - Registered revisions 1:10:48 - Falsifying theories early at the preclinical stage 1:15:21 - Different institutions (e.g., academic, industry) conducting different studies (e.g., preclinical, clinical) 1:17:34 - New initiatives at the Center for Open Science (Replication project of social/behavioral sciences, automated tools for predicting replication success, LifeCycle journal) 1:23:02 - AI scientists are trained on biased literature; distrust of academic literature in drug discovery 1:28:46 - Peer review 1:32:51 - Narrative in science 1:35:06 - 100-200 years into the future 1:40:29 - Advice for high school/undergraduate listeners 1:42:51 - Metascience manifests in every field 1:44:46 - Philosophy of science 1:47:49 - Outro

    1 h 48 min

  5. 1 avr.

    The Replication Crisis Changed Psychology. Did It Fix It? | Brian Nosek

    Brian Nosek is the cofounder and Executive Director of the Center for Open Science. He co-developed the Implicit Association Test, a method that advanced the study of implicit bias. He then co-founded three non-profit organizations: Project Implicit to advance research about implicit bias, the Society for the Improvement of Psychological Science to improve the research culture in his home discipline, and the Center for Open Science (COS) to improve rigor, transparency, integrity, and reproducibility across research disciplines. He led the Reproducibility Project in Psychology, a replication of 100 studies from psychology, as well as multiple other replication projects, along with policy reforms such as open science badges, and the investigation of prediction markets for study replication. CONTACT RANDY: Feedback: metasciencematters@gmail.com EPISODE LINKS: Estimating the reproducibility of psychological science (Reproducibility Project in Psychology led by Brian Nosek): https://www.science.org/doi/10.1126/science.aac4716 False-Positive Psychology: Undisclosed Flexibility in Data Collection and Analysis Allows Presenting Anything as Significant: https://journals.sagepub.com/doi/10.1177/0956797611417632 What is replication?: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3000691 Metascience conference: https://metascience.info/ International Conference on the Science of Science and Innovation: https://icssi.org/ International Congress on Peer Review and Scientific Publication: https://peerreviewcongress.org/ MetaROR: https://metaror.org/ OUTLINE: 0:00 - Introduction 2:56 - Brian's origin story as a psychologist 6:35 - Initial interest in metascience  10:57 - "Scientific Utopia," starting the Reproducibility Project in Psychology, and major fraud cases 17:57 - How did the Reproducibility Project in Psychology come together? 23:40 - Results of the Reproducibility Project in Psychology 27:03 - Response from the field, institutional changes 30:39 - Advice for high school/undergrad listeners 34:03 - Outro

    34 min

  6. 18 mars

    Litigating multimillion-dollar scientific fraud cases | Eugenie Reich

    Eugenie Reich is an attorney committed to taking on scientific fraud, understanding the incentives that drive it, and recovering misdirected research funding. She is also a former investigative science journalist committed to correcting the scientific record. Her 2009 book Plastic Fantastic, details a major fraud cause in physics at Bell Labs. Two of the cases she has litigated cases were against Biogen ($900 million settlement), and the Dana-Farber Cancer Institute ($15 million settlement). CONTACT RANDY: metasciencematters@gmail.com EPISODE LINKS: Cargo Cult Science: https://calteches.library.caltech.edu... The PubPeer Conundrum: https://www.tandfonline.com/doi/epdf/... New Scientist: https://www.newscientist.com/ Skeptical Inquirer: https://skepticalinquirer.org/ Special thanks to Dylan Bouscher OUTLINE: 0:00 - Introduction 4:20 - How Eugenie became interested in science 6:25 - Interest in scientific fraud 7:38 - Deciding to become a science journalist 8:58 - Fraud in physics at Bell Labs (the subject of her 2009 book Plastic Fantastic) 20:15 - Bell Labs coming under private ownership 22:53 - Interviewing scientists for her book 26:50 - The response from the physics community 31:10 - Deciding to become a lawyer 33:28 - The False Claims Act 36:29 - The qui tam provision of the False Claims Act 37:46 - The $900M case against Biogen 44:02 - The day-to-day of working on this case 46:30 - Impact of the case on biotech and pharma 48:40 - The $15M case against Dana-Farber 50:42 - Do universities have an incentive to protect accused researchers? 54:15 - Is the scale of fraud too large to be fixed? 57:56 - Does fraud damage public trust in science? 1:00:44 - Tools/solutions to combat these problems 1:04:32 - Advice to practicing scientists 1:05:55 - Advice and resources for listeners 1:08:28 - Outro

    1 h 9 min

  7. 4 mars

    Forensic Metascience, the GRIM test, and technology for checking papers | James Heathers

    James Heathers is the Founder and Director of the Medical Evidence Project, a venture of The Center for Scientific Integrity. He aims to reduce medical harm and improve patient outcomes by identifying and publicizing errors and miscondcut in the medical literature. He uses forensic meta-analytical techniques to detect and deconstruct errors arising from low-quality science and fraudulent work in areas that involve large numbers of patients. CONTACT RANDY: metasciencematters@gmail.com EPISODE LINKS: The original GRIM test paper: http://www1.psych.purdue.edu/~gfranci... Machine-readable documents: https://journals.sagepub.com/doi/10.1... Chaos in the Brickyard: https://www.science.org/doi/10.1126/s... Books: The Man Who Only Loved Numbers: https://en.wikipedia.org/wiki/The_Man... The Emperor of All Maladies: https://en.wikipedia.org/wiki/The_Emp... OUTLINE: 0:00 - Introduction 5:27 - How James has been the last couple days, and his radio voice 6:40 - Writing the first book on Forensic Metascience 10:47 - James' PhD work on heart rate variability 14:12 - Starting to work on metascience 18:40 - The GRIM test 26:40 - Programmatically scaling statistical checks 33:25 - Centering quantitative results and peripheralizing narrative in scientific papers 39:36 - AI for metascience, unearthing data underneath narrative 48:11 - Pursuing cases of misconduct 52:45 - Founding and Directing the Medical Evidence Project 56:10 - Incentives for positive data, post-publication review, challenging the binary of positive and negative data 1:05:41 - Advice and resources for listeners 1:12:40 - Optimism over pessimism 1:16:58 - Outro

    1 h 17 min

  8. 8 févr.

    300+ retractions, image manipulation, and why science should be boring | Mu Yang

    Mu Yang is a behavioral neuroscientist at Columbia University, and a scientific sleuth responsible for more than 300 retractions. She led an effort that discovered more than 130 fraudulent papers in the publication record of Eliezer Masliah, former head of the Division of Neuroscience at the National Institute on Aging at the National Institutes of Health. Her sleuthing work has been documented in the book "Doctored" by Charles Piller, Science Magazine, and other outlets, and is unpaid. CONTACT RANDY: metasciencematters@gmail.com EPISODE LINKS: Books:  Doctored by Charles Piller: https://www.simonandschuster.com/books/Doctored/Charles-Piller/9781668031254 Unreliable by Csaba Szabo: https://cup.columbia.edu/book/unreliable/9780231216241/ Scientific integrity blogs:  Dorothy Bishop: https://deevybee.blogspot.com/ Leonid Schneider: https://forbetterscience.com/ Podcasts for critical thinking Plain English by Derek Thompson: https://www.theringer.com/podcasts/plain-english-with-derek-thompson The Gray Area by Sean Illing: https://www.vox.com/the-gray-area The Ezra Klein show (NYT): https://www.nytimes.com/column/ezra-klein-podcast OUTLINE: 0:00 - Introduction 2:58 - Mu's origin story 4:35 - Moving to Columbia 6:15 - How Mu became a sleuth 8:13 - Reporting her first case 13:09 - Red flags Mu looks for in papers 17:30 - Reductionism in behavioral neuroscience 18:04 - Standardization vs. Generalizability 19:58 - Data sharing standards across fields 21:09 - Difficulties of reporting irregularities in papers, university incentives 23:54 - Allocating time between images, numerical, other kinds of data 24:37 - How she searches through papers 25:45 - Examining the chemistry literature 31:10 - Types of misconduct vary by field, risks of reporting 35:43 - The case of Eliezer Masliah 40:31 - Why demonizing individual scientists isn't productive; the system isn't working 56:59 - Academic incentives for positive data 1:07:31 - Hard to publish null data; "unhealthy codependence" between academia and publishing 1:13:08 - Changing incentives 1:21:42 - Are we even making a dent in the scale of scientific misconduct?  1:27:35 - Mu's toolkit 1:29:38 - Mu does this work because it's fun! 1:34:38 - Protecting students; telling them that null data is ok 1:37:52 - Evaluating researchers  1:43:15 - Is peer review still relevant? 1:51:38 - How much better could science be? 1:55:14 - What will science look like in a century? 1:58:13 - Advice and resources for listeners 2:00:54 - Online presence 2:01:35 - Outro

    2 h 2 min

  9. 23 janv.

    Esketamine for depression, registered reports, and alcohol use disorder | Florian Naudet

    Florian Naudet is a Professor of Therapeutics at Rennes University. As a metascientist and psychiatrist, his research interests lie in developing and evaluating methodological solutions to treatment assessment, primarily but not exclusively for mental health conditions. His work has also made inroads to quantifying and understanding research waste and the prevalence of substandard data-sharing practices. CONTACT RANDY: Feedback: metasciencematters@gmail.com EPISODE LINKS: Efficacy and safety of esketamine for “treatment resistant depression”: registered report for a systematic review with an individual patient data meta-analysis of randomized, double-blind, placebo-controlled trials https://link.springer.com/article/10.... Vibration of effects from diverse inclusion/exclusion criteria and analytical choices: 9216 different ways to perform an indirect comparison meta-analysis https://link.springer.com/article/10.... Pharmageddon (book) https://davidhealy.org/pharmageddon-i... A manifesto for reproducible science https://www.nature.com/articles/s4156... Using reporting guidelines to improve the reproducibility of cooking Christmas tree meringues: the “People tasting trees” cluster-randomised controlled trial https://pmc.ncbi.nlm.nih.gov/articles... OUTLINE: 0:00 - Introduction 0:36 - Florian's origin story as a psychiatrist and metascientist 4:46 - How Florian became interested in metascience 11:55 - Systematic review of clinical trials of Esketamine for depression 28:45 - Publishing this study as a registered report 40:49 - Vibration-of-effects analysis of double-blind randomized controlled trials assessing nalmefene and/or naltrexone for treating alcohol use disorders 59:58 - Advice for listeners interested in pursuing research like Florian's 1:03:38 - Outro

    1 h 4 min

  10. 15/12/2025

    Environmental impacts on health, and the future of AI in medicine | Chirag Patel

    Chirag Patel is an Associate Professor of Biomedical Informatics at Harvard Medical School, renowned for his expertise in using computational methods to understand human health and disease from high-throughput data streams. He specializes in understanding the role in the intersection of genetics and environmental exposures (the exposome) in human health, as well as various disease areas such as diabetes, cardiovascular disease, and Alzheimer’s disease. CONTACT RANDY: Feedback: metasciencematters@gmail.com EPISODE LINKS: Chirag’s Lab: https://www.chiragjpgroup.org/ TEDx talk on the exposome:    • Exposome: decoding human health and diseas...   OUTLINE: 0:00 - Introduction 0:37 - How do you decide what to work on? 1:57 - Where does metascience fit into your work? 3:57 - Vibration-of-effects analysis 6:41 - How does VoE change how we see existing scientific work? 9:22 - The biggest challenges in the replication crisis within biomedical informatics 11:48 - Environment-wide association study of Type 2 diabetes 15:04 - The exposome 17:30 - Policy vs. precision medicine applications of the exposome 20:09 - The patient exposome 22:38 - HBA1C for diabetes as a bridge to the exposome 23:54 - Broader metascientific issues of the exposome 25:01 - The effects of extreme weather events on human health 29:35 - AI for biomedical informatics, the exposome, metascience 31:19 - Advice for listeners interested in pursuing research like Chirag’s 32:53 - Outro

    33 min
  • 10 épisodes

À propos

The research integrity channel for people who want science to work. Metascience Matters covers scientific fraud, replication failures, peer review, research incentives, AI/data reliability, and the people building better science. Hosted by Randy Ellis, a computational biologist and neuroscientist. New interviews, clips, and explainers on how science breaks and how to fix it. Guests, tips, and collaborations: metasciencematters@gmail.com.

Informations

  • Création
    Randy Ellis
  • Années d’activité
    2025 - 2026
  • Épisodes
    10
  • Classification
    Contenu explicite
  • Copyright
    © Randy Ellis
  • Site web de l’émission
    Metascience Matters