Metascience Matters

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:

[email protected]

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

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: [email protected]

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

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: [email protected]

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