Steve Haroz
@steveharoz.com
Visual perception and cognition scientist
(he/him)
My site: http://steveharoz.com
R guide: https://r-guide.steveharoz.com
StatCheck Simple: http://statcheck.steveharoz.com
(he/him)
My site: http://steveharoz.com
R guide: https://r-guide.steveharoz.com
StatCheck Simple: http://statcheck.steveharoz.com
Pinned
Steve Haroz
@steveharoz.com
· Mar 14
Invalid Conclusions Built on Statistical Errors
When you see the statistical result of an experiment like p = 0.003 or a 95% confidence interval of , you might assume a certain clarity and definitiveness in those results. A null effect is very unli...
steveharoz.com
BLOG POST: Invalid Conclusions Built on Statistical errors
If you see a small p-value or a large separation in confidence intervals, you may assume that an effect is reliable. But overly simple reporting may be hiding serious conclusion-altering errors.
Here are some example scenarios:
1/🧵 #stats
If you see a small p-value or a large separation in confidence intervals, you may assume that an effect is reliable. But overly simple reporting may be hiding serious conclusion-altering errors.
Here are some example scenarios:
1/🧵 #stats
Apologies that the #StatsSky feed is acting finicky. I did some spam cleaning yesterday, but now it's dropping too many posts. I need to debug it, but the UI for that is rough (damn regex). Also there's no source control, so I'm always scared I'll irreparably break it.
January 13, 2026 at 11:35 PM
Apologies that the #StatsSky feed is acting finicky. I did some spam cleaning yesterday, but now it's dropping too many posts. I need to debug it, but the UI for that is rough (damn regex). Also there's no source control, so I'm always scared I'll irreparably break it.
Can someone explain to me what problem this solves? I thought Bayesian analyses were already in plenty of clinical trial papers. Was it previously discouraged? Or often misused? What behavior is expected to change because of this doc?
#StatsSky
#StatsSky
FDA drugs and biologics guidance for Bayesian clinical trials is something I've been dreaming of for decades. I am thrilled that this draft has been released!! #Statistics #StatsSky #fda #clinicaltrial #bayes
This is a big deal for speeding up clinical trials, and for pediatric use after adult trials have proved successful. The @USFDA goes Bayesian.
A win for #IBD patients, and especially for kids with #IBD. @crohnsandcolitisau.bsky.social @f2harrell.bsky.social
www.fda.gov/news-events/...
A win for #IBD patients, and especially for kids with #IBD. @crohnsandcolitisau.bsky.social @f2harrell.bsky.social
www.fda.gov/news-events/...
January 13, 2026 at 10:59 PM
Can someone explain to me what problem this solves? I thought Bayesian analyses were already in plenty of clinical trial papers. Was it previously discouraged? Or often misused? What behavior is expected to change because of this doc?
#StatsSky
#StatsSky
Reposted by Steve Haroz
We've got ISSUES. Literally.
We scraped >100k special issues & over 1 million articles to bring you a PISS-poor paper. We quantify just how many excess papers are published by guest editors abusing special issues to boost their CVs. How bad is it & what can we do?
arxiv.org/abs/2601.07563
A 🧵 1/n
We scraped >100k special issues & over 1 million articles to bring you a PISS-poor paper. We quantify just how many excess papers are published by guest editors abusing special issues to boost their CVs. How bad is it & what can we do?
arxiv.org/abs/2601.07563
A 🧵 1/n
January 13, 2026 at 8:27 AM
We've got ISSUES. Literally.
We scraped >100k special issues & over 1 million articles to bring you a PISS-poor paper. We quantify just how many excess papers are published by guest editors abusing special issues to boost their CVs. How bad is it & what can we do?
arxiv.org/abs/2601.07563
A 🧵 1/n
We scraped >100k special issues & over 1 million articles to bring you a PISS-poor paper. We quantify just how many excess papers are published by guest editors abusing special issues to boost their CVs. How bad is it & what can we do?
arxiv.org/abs/2601.07563
A 🧵 1/n
I am happy to share that our preprint “𝗪𝗼𝗿𝗸𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗖𝗶𝗿𝗰𝘂𝗹𝗮𝗿 𝗗𝗮𝘁𝗮: 𝗔 𝗧𝘂𝘁𝗼𝗿𝗶𝗮𝗹 𝗳𝗼𝗿 𝗖𝗼𝗴𝗻𝗶𝘁𝗶𝘃𝗲 𝗮𝗻𝗱 𝗕𝗲𝗵𝗮𝘃𝗶𝗼𝗿𝗮𝗹 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵” is now out.
Huge thanks to @bayslab.org, Julie de Falco, Zahara, @cjungerius.bsky.social, @ivntmc.bsky.social, Adam, and Xiaolu for the lovely collaboration.
doi.org/10.31234/osf...
Huge thanks to @bayslab.org, Julie de Falco, Zahara, @cjungerius.bsky.social, @ivntmc.bsky.social, Adam, and Xiaolu for the lovely collaboration.
doi.org/10.31234/osf...
OSF
doi.org
January 13, 2026 at 5:18 AM
Reposted by Steve Haroz
🔍 Sleuthing challenge 🔍
How many suspicious patterns can you find in this table?
(You don't need to know what the variables are.)
Hints in next post! 👇
How many suspicious patterns can you find in this table?
(You don't need to know what the variables are.)
Hints in next post! 👇
January 11, 2026 at 6:14 AM
🔍 Sleuthing challenge 🔍
How many suspicious patterns can you find in this table?
(You don't need to know what the variables are.)
Hints in next post! 👇
How many suspicious patterns can you find in this table?
(You don't need to know what the variables are.)
Hints in next post! 👇
#statstab #460 Invalid Conclusions Built on Statistical Errors
Thoughts: It's easy to make analysis mistakes, providing code should be the norm.
#analysis #pseudoreplication #error #design #code #guide #educational
steveharoz.com/blog/2024/wr...
Thoughts: It's easy to make analysis mistakes, providing code should be the norm.
#analysis #pseudoreplication #error #design #code #guide #educational
steveharoz.com/blog/2024/wr...
Invalid Conclusions Built on Statistical Errors
When you see p = 0.003 or a 95% confidence interval of , you might assume a certain clarity and definitiveness. A null effect is very unlikely to yield those results, right? But be careful! Such overl...
steveharoz.com
January 11, 2026 at 6:24 AM
Tiptoeing back onto social media after a year away.
January 6, 2026 at 10:08 PM
Tiptoeing back onto social media after a year away.
Reposted by Steve Haroz
RFK Jr is a consequence of decades of anti-science media, promoted by both the right and left. Subtle forms of anti-sciencism like organic food or raw milk or "removing toxins" snowball over time. Calling them out in their early phases may be critical.
The media outlets rightly admonishing RFK Jr. need to reckon with their own roll in his rise to prominence.
In 2005 Rolling Stone and Salon would publish his op-ed ‘Deadly Immunity’ linking vaccines to autism in children, based on tiny - later retracted - British study.
In 2005 Rolling Stone and Salon would publish his op-ed ‘Deadly Immunity’ linking vaccines to autism in children, based on tiny - later retracted - British study.
December 9, 2024 at 5:40 AM
RFK Jr is a consequence of decades of anti-science media, promoted by both the right and left. Subtle forms of anti-sciencism like organic food or raw milk or "removing toxins" snowball over time. Calling them out in their early phases may be critical.
Reposted by Steve Haroz
After spending more than an hour on the phone with Microsoft Support, I have learned:
1. It is impossible to disable Copilot in OneNote, Excel, PowerPoint, or Windows itself.
2. It will not become possible to do so for another month AT THE EARLIEST.
(1/?)
1. It is impossible to disable Copilot in OneNote, Excel, PowerPoint, or Windows itself.
2. It will not become possible to do so for another month AT THE EARLIEST.
(1/?)
Anyone know how to disable Copilot in OneNote? There's no tickybox for it in the "Options" page, I already turned off "online experiences" or whatever, and when I tried uninstalling the Copilot app, the only change seems to be that I no longer have an "uninstall" option on the Copilot app.
January 24, 2025 at 9:32 PM
After spending more than an hour on the phone with Microsoft Support, I have learned:
1. It is impossible to disable Copilot in OneNote, Excel, PowerPoint, or Windows itself.
2. It will not become possible to do so for another month AT THE EARLIEST.
(1/?)
1. It is impossible to disable Copilot in OneNote, Excel, PowerPoint, or Windows itself.
2. It will not become possible to do so for another month AT THE EARLIEST.
(1/?)
Reposted by Steve Haroz
Whenever I see these articles on 'publish or perish' culture being to blame, I always think of Tal Yarkoni's blog post: talyarkoni.org/blog/2018/10.... Yes, the pressures are there and they can force our hand, but that means we should work to improve the structures where we can (not abide by them).
No, it’s not The Incentives—it’s you
There’s a narrative I find kind of troubling, but that unfortunately seems to be growing more common in science. The core idea is that the mere existence of perverse incentives is a valid and…
talyarkoni.org
January 23, 2025 at 9:23 AM
Whenever I see these articles on 'publish or perish' culture being to blame, I always think of Tal Yarkoni's blog post: talyarkoni.org/blog/2018/10.... Yes, the pressures are there and they can force our hand, but that means we should work to improve the structures where we can (not abide by them).
Reposted by Steve Haroz
Are you not a fan of journals' "enhanced online PDF viewers"? Me neither. I put together a little Firefox add-on that helps you skip the "enhanced" reader and download the PDF directly.
January 23, 2025 at 10:08 AM
Are you not a fan of journals' "enhanced online PDF viewers"? Me neither. I put together a little Firefox add-on that helps you skip the "enhanced" reader and download the PDF directly.
Reposted by Steve Haroz
Check out this incredible video from 1941 that details the physics of visual perception using the mechanics of the eyeball from Karl Kurt Bosse
pennstateoffice365-my.sharepoint.com/:v:/g/person...
pennstateoffice365-my.sharepoint.com/:v:/g/person...
Sign in to your account
pennstateoffice365-my.sharepoint.com
January 6, 2025 at 9:20 PM
Check out this incredible video from 1941 that details the physics of visual perception using the mechanics of the eyeball from Karl Kurt Bosse
pennstateoffice365-my.sharepoint.com/:v:/g/person...
pennstateoffice365-my.sharepoint.com/:v:/g/person...
Reposted by Steve Haroz
Hey #StatsSky, what are you favorite papers to cite when you need to justify something that is obvious (I once had a reviewer ask we justify the use of logistic regression on a binary outcome)
or when you need to push-back on silly reviewer requests (e.g., asking for p-values in table 1)?
or when you need to push-back on silly reviewer requests (e.g., asking for p-values in table 1)?
January 6, 2025 at 8:15 AM
Hey #StatsSky, what are you favorite papers to cite when you need to justify something that is obvious (I once had a reviewer ask we justify the use of logistic regression on a binary outcome)
or when you need to push-back on silly reviewer requests (e.g., asking for p-values in table 1)?
or when you need to push-back on silly reviewer requests (e.g., asking for p-values in table 1)?
Reposted by Steve Haroz
Flu cases have quadrupled over the past month in hospitals in England, with NHS leaders warning that “skyrocketing” cases could make this winter “one of the worst we have ever seen”. Just published @financialtimes.com: on.ft.com/4j4g0h5
January 3, 2025 at 12:41 PM
Flu cases have quadrupled over the past month in hospitals in England, with NHS leaders warning that “skyrocketing” cases could make this winter “one of the worst we have ever seen”. Just published @financialtimes.com: on.ft.com/4j4g0h5
Reposted by Steve Haroz
Have tried to make the dish many times, with varying success. This is an important pepe:
"Finally, we present a scientifically optimized recipe based on our findings, enabling a consistently flawless execution of this classic dish."
"Finally, we present a scientifically optimized recipe based on our findings, enabling a consistently flawless execution of this classic dish."
1/n Some time ago my colleague, excellent cook, and friend Ivan told me: "Cacio e pepe is the recipe that I screw up more often. Let's make a project studying systematically the physics of that sauce".
Prepare to get cheesy, I'm glad to share the Cacio e paper preprint:
arxiv.org/abs/2501.00536
Prepare to get cheesy, I'm glad to share the Cacio e paper preprint:
arxiv.org/abs/2501.00536
January 5, 2025 at 10:19 AM
Have tried to make the dish many times, with varying success. This is an important pepe:
"Finally, we present a scientifically optimized recipe based on our findings, enabling a consistently flawless execution of this classic dish."
"Finally, we present a scientifically optimized recipe based on our findings, enabling a consistently flawless execution of this classic dish."
Reposted by Steve Haroz
Everyone is running away with this one, but I took the time to trace the source, and these are 80% confidence intervals, the true meta-science story (and there is one!), was written up as one about selective reporting and calibrating standard errors..
January 4, 2025 at 9:47 PM
Everyone is running away with this one, but I took the time to trace the source, and these are 80% confidence intervals, the true meta-science story (and there is one!), was written up as one about selective reporting and calibrating standard errors..
I've settled on an approach to giving up on TV shows:
If a show has a serialized plot and goes more than 1 year between seasons, it's dead. It doesn't matter how much I like it, I cannot keep track of characters or follow a plot with a gap of more than one year.
If a show has a serialized plot and goes more than 1 year between seasons, it's dead. It doesn't matter how much I like it, I cannot keep track of characters or follow a plot with a gap of more than one year.
January 4, 2025 at 8:32 AM
I've settled on an approach to giving up on TV shows:
If a show has a serialized plot and goes more than 1 year between seasons, it's dead. It doesn't matter how much I like it, I cannot keep track of characters or follow a plot with a gap of more than one year.
If a show has a serialized plot and goes more than 1 year between seasons, it's dead. It doesn't matter how much I like it, I cannot keep track of characters or follow a plot with a gap of more than one year.
Reposted by Steve Haroz
Researcher: "We let the data speak for itself."
Earlier that day:
Earlier that day:
January 2, 2025 at 3:31 PM
Researcher: "We let the data speak for itself."
Earlier that day:
Earlier that day:
Reposted by Steve Haroz
How can sharing our uncertainty with others alter our confidence when we're alone?
Delighted to share the lab's new paper, with Einar Andreassen and @cdfrith.bsky.social. Particularly pleased as it's Einar's first!
Link: osf.io/preprints/ps...
🧵👇
#PsychSciSky
#neuroskyence
Delighted to share the lab's new paper, with Einar Andreassen and @cdfrith.bsky.social. Particularly pleased as it's Einar's first!
Link: osf.io/preprints/ps...
🧵👇
#PsychSciSky
#neuroskyence
January 2, 2025 at 11:54 AM
How can sharing our uncertainty with others alter our confidence when we're alone?
Delighted to share the lab's new paper, with Einar Andreassen and @cdfrith.bsky.social. Particularly pleased as it's Einar's first!
Link: osf.io/preprints/ps...
🧵👇
#PsychSciSky
#neuroskyence
Delighted to share the lab's new paper, with Einar Andreassen and @cdfrith.bsky.social. Particularly pleased as it's Einar's first!
Link: osf.io/preprints/ps...
🧵👇
#PsychSciSky
#neuroskyence
Everything is more advanced and yet shittier.
December 31, 2024 at 4:47 AM
Everything is more advanced and yet shittier.
Reposted by Steve Haroz
Timeline cleanse! Allow me to share with you my favorite images from the 10 billion dollar telescope JWST this year 🧵🧪
Starting off heavenly with this star forming region in the Large Magellanic Cloud…
📸: NASA, ESA, CSA, O Nayak, M Meixner
Starting off heavenly with this star forming region in the Large Magellanic Cloud…
📸: NASA, ESA, CSA, O Nayak, M Meixner
December 29, 2024 at 9:21 PM
Timeline cleanse! Allow me to share with you my favorite images from the 10 billion dollar telescope JWST this year 🧵🧪
Starting off heavenly with this star forming region in the Large Magellanic Cloud…
📸: NASA, ESA, CSA, O Nayak, M Meixner
Starting off heavenly with this star forming region in the Large Magellanic Cloud…
📸: NASA, ESA, CSA, O Nayak, M Meixner
Reposted by Steve Haroz
How did the tech industry get so powerful? It's simple. There are major media outlets that will simply copy paste anything they say
Netflix says more than 200 countries tuned into the "Beyoncé Bowl" and its two NFL games. We may just be starting to learn what that ultimately means for the future of television and sports media.
The 'Beyoncé Bowl' halftime show was a massive hit for Netflix. The football was too
Netflix says more than 200 countries tuned into the "Beyoncé Bowl" and its two NFL games. We may just be starting to learn what that ultimately means for the future of television and sports media.
www.npr.org
December 27, 2024 at 1:02 AM
How did the tech industry get so powerful? It's simple. There are major media outlets that will simply copy paste anything they say
Reposted by Steve Haroz
Stop Forcing A.I. into Fucking EVERYTHING!
December 24, 2024 at 3:11 AM
Stop Forcing A.I. into Fucking EVERYTHING!