Sophie E. Hill
@sophieehill.bsky.social
PoliSci PhD student @ Harvard / 🇬🇧🏳️🌈 / Creator of MyLittleCrony.com
Signal: @sehill.11
Signal: @sehill.11
Pinned
My thread on the Neurology paper on sweeteners & cognitive health: skywriter.blue/pages/did:pl...
And a follow-up thread on some weird patterns in the Appendix tables: skywriter.blue/pages/did:pl...
Collated by Skywriter onto a web page for easier reading!
And a follow-up thread on some weird patterns in the Appendix tables: skywriter.blue/pages/did:pl...
Collated by Skywriter onto a web page for easier reading!
Reposted by Sophie E. Hill
Double bonus question: can you spot TWO "sibling" tortured phrases (i.e., that derive from the same phrase)?
November 19, 2025 at 11:24 AM
Double bonus question: can you spot TWO "sibling" tortured phrases (i.e., that derive from the same phrase)?
Reposted by Sophie E. Hill
NEW: Lord Dannatt and Lord David Evans of Watford face suspension of 4mo/5mo respectively from the House of Lords after a parliamentary watchdog ruled they had broken lobbying rules. Reports from the conduct committee follows Guardian investigations into the peers.
w/ @robevansgdn.bsky.social
w/ @robevansgdn.bsky.social
Two peers suspended from House of Lords for breaking lobbying rules
Lord Dannatt and Lord Evans of Watford were filmed breaking rules in undercover footage recorded by Guardian
www.theguardian.com
November 24, 2025 at 12:13 PM
NEW: Lord Dannatt and Lord David Evans of Watford face suspension of 4mo/5mo respectively from the House of Lords after a parliamentary watchdog ruled they had broken lobbying rules. Reports from the conduct committee follows Guardian investigations into the peers.
w/ @robevansgdn.bsky.social
w/ @robevansgdn.bsky.social
Reposted by Sophie E. Hill
❌ Error 1 (level: EASY)
Solid waste has mean (0.624) > max (0.525)
❌ Error 2 (level: MEDIUM)
Solid waste and fiscal revenue have SD > range/2
0.53 > (0.525-0.1)/2 = 0.2125
0.926 > (1.2-0.024)/2 = 0.588
Solid waste has mean (0.624) > max (0.525)
❌ Error 2 (level: MEDIUM)
Solid waste and fiscal revenue have SD > range/2
0.53 > (0.525-0.1)/2 = 0.2125
0.926 > (1.2-0.024)/2 = 0.588
November 24, 2025 at 10:18 AM
❌ Error 1 (level: EASY)
Solid waste has mean (0.624) > max (0.525)
❌ Error 2 (level: MEDIUM)
Solid waste and fiscal revenue have SD > range/2
0.53 > (0.525-0.1)/2 = 0.2125
0.926 > (1.2-0.024)/2 = 0.588
Solid waste has mean (0.624) > max (0.525)
❌ Error 2 (level: MEDIUM)
Solid waste and fiscal revenue have SD > range/2
0.53 > (0.525-0.1)/2 = 0.2125
0.926 > (1.2-0.024)/2 = 0.588
Good morning!
Today is a day ending in "y", which means I found another mathematically-impossible Table of Summary Statistics in a peer-reviewed "green economics" paper!
What can you see? 🔎
#FablesofFlummeryStatistics
Today is a day ending in "y", which means I found another mathematically-impossible Table of Summary Statistics in a peer-reviewed "green economics" paper!
What can you see? 🔎
#FablesofFlummeryStatistics
November 24, 2025 at 10:18 AM
Good morning!
Today is a day ending in "y", which means I found another mathematically-impossible Table of Summary Statistics in a peer-reviewed "green economics" paper!
What can you see? 🔎
#FablesofFlummeryStatistics
Today is a day ending in "y", which means I found another mathematically-impossible Table of Summary Statistics in a peer-reviewed "green economics" paper!
What can you see? 🔎
#FablesofFlummeryStatistics
Reposted by Sophie E. Hill
Reverse Image Forensics Challenge: Try to find a unique area in Figure 9 of this recent Scientific Reports paper: 10.1038/s41598-025-17456-6 [Aug 2025] - Annotated by ImageTwin.ai
November 11, 2025 at 1:49 AM
Reverse Image Forensics Challenge: Try to find a unique area in Figure 9 of this recent Scientific Reports paper: 10.1038/s41598-025-17456-6 [Aug 2025] - Annotated by ImageTwin.ai
Reposted by Sophie E. Hill
next up we have one coefficient that looks off by a factor of 10.
god knows how that happened. but the lesson, as ever, is ABP (Always Be Plotting)
plot your raw data. plot your summary statistics. plot your model coefficients. plot your enemy's downfall. plot your residuals.
god knows how that happened. but the lesson, as ever, is ABP (Always Be Plotting)
plot your raw data. plot your summary statistics. plot your model coefficients. plot your enemy's downfall. plot your residuals.
November 22, 2025 at 2:29 PM
next up we have one coefficient that looks off by a factor of 10.
god knows how that happened. but the lesson, as ever, is ABP (Always Be Plotting)
plot your raw data. plot your summary statistics. plot your model coefficients. plot your enemy's downfall. plot your residuals.
god knows how that happened. but the lesson, as ever, is ABP (Always Be Plotting)
plot your raw data. plot your summary statistics. plot your model coefficients. plot your enemy's downfall. plot your residuals.
Reposted by Sophie E. Hill
It's 2025 and you can still claim an effect with results like this at JAMA Psychiatry. cc @urisohn.bsky.social @lakens.bsky.social @heinonmatti.bsky.social
jamanetwork.com/journals/jam...
jamanetwork.com/journals/jam...
November 22, 2025 at 10:28 AM
It's 2025 and you can still claim an effect with results like this at JAMA Psychiatry. cc @urisohn.bsky.social @lakens.bsky.social @heinonmatti.bsky.social
jamanetwork.com/journals/jam...
jamanetwork.com/journals/jam...
Reposted by Sophie E. Hill
There's a fund manager in Australia that says he uses the implausible golf scores of C-suite executives as a 'tell' for companies he should consider shorting (as they're more likely to fudge their financial reports).
Ridiculously inaccurate summary stats feels like the research equivalent.
😂
Ridiculously inaccurate summary stats feels like the research equivalent.
😂
lads i found ANOTHER Table of Impossible Summary Statistics in a "green economics" paper 🚩
Can you see the problem?
Can you see the problem?
November 22, 2025 at 10:03 PM
There's a fund manager in Australia that says he uses the implausible golf scores of C-suite executives as a 'tell' for companies he should consider shorting (as they're more likely to fudge their financial reports).
Ridiculously inaccurate summary stats feels like the research equivalent.
😂
Ridiculously inaccurate summary stats feels like the research equivalent.
😂
Reposted by Sophie E. Hill
other sleuths: *using AI to detect pixel-by-pixel similarities indicating image duplication*
me: ... max minus min
me: ... max minus min
November 22, 2025 at 2:34 PM
other sleuths: *using AI to detect pixel-by-pixel similarities indicating image duplication*
me: ... max minus min
me: ... max minus min
Reposted by Sophie E. Hill
this example is unusual because there are a few breadcrumbs indicating what went wrong...
* Means got duplicated in SD column
* SD values appear in Min column (or at least, these values = sqrt(var))
* Means got duplicated in SD column
* SD values appear in Min column (or at least, these values = sqrt(var))
November 22, 2025 at 2:29 PM
this example is unusual because there are a few breadcrumbs indicating what went wrong...
* Means got duplicated in SD column
* SD values appear in Min column (or at least, these values = sqrt(var))
* Means got duplicated in SD column
* SD values appear in Min column (or at least, these values = sqrt(var))
🤑 SPECIAL OFFER 💰
Dear academic papermills: for the low, low price of $9,999 (bitcoin only) I will check your tables of summary statistics before publication using my proprietary technology (lean back, squint at numbers for 90 seconds). DM for details.
Dear academic papermills: for the low, low price of $9,999 (bitcoin only) I will check your tables of summary statistics before publication using my proprietary technology (lean back, squint at numbers for 90 seconds). DM for details.
November 22, 2025 at 2:46 PM
🤑 SPECIAL OFFER 💰
Dear academic papermills: for the low, low price of $9,999 (bitcoin only) I will check your tables of summary statistics before publication using my proprietary technology (lean back, squint at numbers for 90 seconds). DM for details.
Dear academic papermills: for the low, low price of $9,999 (bitcoin only) I will check your tables of summary statistics before publication using my proprietary technology (lean back, squint at numbers for 90 seconds). DM for details.
Reposted by Sophie E. Hill
The people that will fix science are the ones doing the deep digging required to find errors, statistical shenanigans, and out-and-out fraud.
Hats off.
Hats off.
lads i found ANOTHER Table of Impossible Summary Statistics in a "green economics" paper 🚩
Can you see the problem?
Can you see the problem?
November 22, 2025 at 2:31 PM
The people that will fix science are the ones doing the deep digging required to find errors, statistical shenanigans, and out-and-out fraud.
Hats off.
Hats off.
lads i found ANOTHER Table of Impossible Summary Statistics in a "green economics" paper 🚩
Can you see the problem?
Can you see the problem?
November 22, 2025 at 2:29 PM
lads i found ANOTHER Table of Impossible Summary Statistics in a "green economics" paper 🚩
Can you see the problem?
Can you see the problem?
Reposted by Sophie E. Hill
Origins:
In 2000s, Narayan develops a formula: add "nexus" to a classic econ literature about 2 vars
In 2010s, "nexus" independently becomes a policy buzzword ("Water-Energy-Food Nexus")
Papermills love to combine trendy words + topics
Result: 1000s of green econ "nexus" papers, esp 2021-24
In 2000s, Narayan develops a formula: add "nexus" to a classic econ literature about 2 vars
In 2010s, "nexus" independently becomes a policy buzzword ("Water-Energy-Food Nexus")
Papermills love to combine trendy words + topics
Result: 1000s of green econ "nexus" papers, esp 2021-24
November 21, 2025 at 1:01 AM
Origins:
In 2000s, Narayan develops a formula: add "nexus" to a classic econ literature about 2 vars
In 2010s, "nexus" independently becomes a policy buzzword ("Water-Energy-Food Nexus")
Papermills love to combine trendy words + topics
Result: 1000s of green econ "nexus" papers, esp 2021-24
In 2000s, Narayan develops a formula: add "nexus" to a classic econ literature about 2 vars
In 2010s, "nexus" independently becomes a policy buzzword ("Water-Energy-Food Nexus")
Papermills love to combine trendy words + topics
Result: 1000s of green econ "nexus" papers, esp 2021-24
Reposted by Sophie E. Hill
🙋♀️ @paulnightingale.bsky.social me too!
I call it: "The "Nexus" Nexus" 😅
Recipe
🟢 Pick 1 econ var
🟢 Pick 2+ "green" vars
🟢 Pick 3+ methods (unit root, cointegration, Granger causality, GMM, ECM, FMOLS, PMG, PVAR, wavelet)
🟢 Misinterpret results
🟢 Make absurd policy recommendations
Rinse & repeat!
I call it: "The "Nexus" Nexus" 😅
Recipe
🟢 Pick 1 econ var
🟢 Pick 2+ "green" vars
🟢 Pick 3+ methods (unit root, cointegration, Granger causality, GMM, ECM, FMOLS, PMG, PVAR, wavelet)
🟢 Misinterpret results
🟢 Make absurd policy recommendations
Rinse & repeat!
November 21, 2025 at 12:34 AM
🙋♀️ @paulnightingale.bsky.social me too!
I call it: "The "Nexus" Nexus" 😅
Recipe
🟢 Pick 1 econ var
🟢 Pick 2+ "green" vars
🟢 Pick 3+ methods (unit root, cointegration, Granger causality, GMM, ECM, FMOLS, PMG, PVAR, wavelet)
🟢 Misinterpret results
🟢 Make absurd policy recommendations
Rinse & repeat!
I call it: "The "Nexus" Nexus" 😅
Recipe
🟢 Pick 1 econ var
🟢 Pick 2+ "green" vars
🟢 Pick 3+ methods (unit root, cointegration, Granger causality, GMM, ECM, FMOLS, PMG, PVAR, wavelet)
🟢 Misinterpret results
🟢 Make absurd policy recommendations
Rinse & repeat!
Sometimes I can just tell by the vibes* that a table of summary statistics is going to be mathematically impossible.
(*Probably a combination of: unrounded numbers, centre alignment, and putting the min column at the end)
(*Probably a combination of: unrounded numbers, centre alignment, and putting the min column at the end)
November 20, 2025 at 10:38 AM
Sometimes I can just tell by the vibes* that a table of summary statistics is going to be mathematically impossible.
(*Probably a combination of: unrounded numbers, centre alignment, and putting the min column at the end)
(*Probably a combination of: unrounded numbers, centre alignment, and putting the min column at the end)
Reposted by Sophie E. Hill
In the pooled analysis, voters are:
🔵 more right-wing
🔵 less in favour of redistribution
🔵 more supportive of state intervention and spending
🔵 less supportive of same-sex marriage
🔵 more restrictive on immigration
🔵 more willing to restrict privacy rights to combat crimes
🔵 more pro-Europe
4/10
🔵 more right-wing
🔵 less in favour of redistribution
🔵 more supportive of state intervention and spending
🔵 less supportive of same-sex marriage
🔵 more restrictive on immigration
🔵 more willing to restrict privacy rights to combat crimes
🔵 more pro-Europe
4/10
November 19, 2025 at 1:39 PM
In the pooled analysis, voters are:
🔵 more right-wing
🔵 less in favour of redistribution
🔵 more supportive of state intervention and spending
🔵 less supportive of same-sex marriage
🔵 more restrictive on immigration
🔵 more willing to restrict privacy rights to combat crimes
🔵 more pro-Europe
4/10
🔵 more right-wing
🔵 less in favour of redistribution
🔵 more supportive of state intervention and spending
🔵 less supportive of same-sex marriage
🔵 more restrictive on immigration
🔵 more willing to restrict privacy rights to combat crimes
🔵 more pro-Europe
4/10
Reposted by Sophie E. Hill
Do policy preferences of voters and non-voters differ in European democracies?
My new article in @ejprjournal.bsky.social finds: Yes, sometimes, but this is not the norm.
Let's break it down. 🧵 1/10
My new article in @ejprjournal.bsky.social finds: Yes, sometimes, but this is not the norm.
Let's break it down. 🧵 1/10
November 19, 2025 at 1:39 PM
Do policy preferences of voters and non-voters differ in European democracies?
My new article in @ejprjournal.bsky.social finds: Yes, sometimes, but this is not the norm.
Let's break it down. 🧵 1/10
My new article in @ejprjournal.bsky.social finds: Yes, sometimes, but this is not the norm.
Let's break it down. 🧵 1/10
Reposted by Sophie E. Hill
Now this is how you detect whether an election was stolen. Humans choose rounder numbers.
by @TheEconomist
by @TheEconomist
November 6, 2025 at 9:55 AM
Now this is how you detect whether an election was stolen. Humans choose rounder numbers.
by @TheEconomist
by @TheEconomist
🏆 Tortured Phrases Quiz: Round 2 🏆
Can you guess the original phrase?
(e.g. "cloudiness contamination" ➡️ "air pollution")
1️⃣ carbon discharges
2️⃣ refillable energy
3️⃣ depressing influence
4️⃣ activist liaison
Can you guess the original phrase?
(e.g. "cloudiness contamination" ➡️ "air pollution")
1️⃣ carbon discharges
2️⃣ refillable energy
3️⃣ depressing influence
4️⃣ activist liaison
This is a great intro to tortured phrases.
"Bosom peril" for "breast cancer" is an all-timer.
The best one I've personally found was a statistical method being described as "vigorous" (instead of "robust").
thebulletin.org/2022/01/boso...
"Bosom peril" for "breast cancer" is an all-timer.
The best one I've personally found was a statistical method being described as "vigorous" (instead of "robust").
thebulletin.org/2022/01/boso...
"Bosom peril" is not "breast cancer": How weird computer-generated phrases help researchers find scientific publishing fraud
Many fraudulent scientific papers contain "tortured phrases," standard scientific terms paraphrased into gibberish. Breast cancer is a scary disease, but what is bosom peril? These phrases are likely ...
thebulletin.org
November 20, 2025 at 8:52 AM
🏆 Tortured Phrases Quiz: Round 2 🏆
Can you guess the original phrase?
(e.g. "cloudiness contamination" ➡️ "air pollution")
1️⃣ carbon discharges
2️⃣ refillable energy
3️⃣ depressing influence
4️⃣ activist liaison
Can you guess the original phrase?
(e.g. "cloudiness contamination" ➡️ "air pollution")
1️⃣ carbon discharges
2️⃣ refillable energy
3️⃣ depressing influence
4️⃣ activist liaison
6+ goals scored in BOTH the 5:45 #UWCL fixtures:
Juventus 3 — 3 Lyon
Wolfsburg 5 — 2 Man Utd
Kicking off now: Real Madrid v Arsenal, with the Gunners in their worst run of form for several years...
Juventus 3 — 3 Lyon
Wolfsburg 5 — 2 Man Utd
Kicking off now: Real Madrid v Arsenal, with the Gunners in their worst run of form for several years...
November 19, 2025 at 8:01 PM
6+ goals scored in BOTH the 5:45 #UWCL fixtures:
Juventus 3 — 3 Lyon
Wolfsburg 5 — 2 Man Utd
Kicking off now: Real Madrid v Arsenal, with the Gunners in their worst run of form for several years...
Juventus 3 — 3 Lyon
Wolfsburg 5 — 2 Man Utd
Kicking off now: Real Madrid v Arsenal, with the Gunners in their worst run of form for several years...
Reposted by Sophie E. Hill
To balance my impossible question, I'll offer a trivial one that remains my favorite tortured phrase.
Who's "Glove Romney"? 🤣
pubpeer.com/publications...
Who's "Glove Romney"? 🤣
pubpeer.com/publications...
PubPeer - A Survey Report on Data Analytics as a Tool in PoliticalCamp...
There are comments on PubPeer for publication: A Survey Report on Data Analytics as a Tool in PoliticalCampaign (2019)
pubpeer.com
November 19, 2025 at 12:30 PM
To balance my impossible question, I'll offer a trivial one that remains my favorite tortured phrase.
Who's "Glove Romney"? 🤣
pubpeer.com/publications...
Who's "Glove Romney"? 🤣
pubpeer.com/publications...
Reposted by Sophie E. Hill
Extra impossible bonus question: figure out why some papers torture "by" into, and I swear I am not making this up, "eventually Tom's perusing".
scholar.google.com/scholar?hl=e...
scholar.google.com/scholar?hl=e...
scholar.google.com
November 19, 2025 at 12:28 PM
Extra impossible bonus question: figure out why some papers torture "by" into, and I swear I am not making this up, "eventually Tom's perusing".
scholar.google.com/scholar?hl=e...
scholar.google.com/scholar?hl=e...
Reposted by Sophie E. Hill
Oh my gosh - its worth reading through @sophieehill.bsky.social's thread
CHALLENGE: can you decode these "tortured phrases"?
(i.e., if I say "bosom peril", you say "breast cancer")
1️⃣ "the upset U-molded EKC"
(i.e., if I say "bosom peril", you say "breast cancer")
1️⃣ "the upset U-molded EKC"
This is a great intro to tortured phrases.
"Bosom peril" for "breast cancer" is an all-timer.
The best one I've personally found was a statistical method being described as "vigorous" (instead of "robust").
thebulletin.org/2022/01/boso...
"Bosom peril" for "breast cancer" is an all-timer.
The best one I've personally found was a statistical method being described as "vigorous" (instead of "robust").
thebulletin.org/2022/01/boso...
November 19, 2025 at 11:14 AM
Oh my gosh - its worth reading through @sophieehill.bsky.social's thread