Dan Lewer
@danlewer.bsky.social
Consultant in Public Health at Bradford Institute for Health Research. Interested in quantitative research methods and mental health
It makes a lot more sense but no one understands it
November 6, 2025 at 8:47 PM
It makes a lot more sense but no one understands it
Have I read that right? I guess there is probably some kind of collider bias in people with haemotological disease have a lower threshold for hospital admission and/or maybe people who get serious chikengunya might have other more important risk factors?
November 6, 2025 at 8:18 PM
Have I read that right? I guess there is probably some kind of collider bias in people with haemotological disease have a lower threshold for hospital admission and/or maybe people who get serious chikengunya might have other more important risk factors?
I don't know a lot about chikungunya and dengue and suspect there are lots of valuable insights here! In fig 2 there's a curious association in which inpatients with both chikungunya and haemotological disease appear to have lower risk of dying (this doesn't seem to happen for dengue).
November 6, 2025 at 8:17 PM
I don't know a lot about chikungunya and dengue and suspect there are lots of valuable insights here! In fig 2 there's a curious association in which inpatients with both chikungunya and haemotological disease appear to have lower risk of dying (this doesn't seem to happen for dengue).
Lovely maps in this paper!
November 6, 2025 at 8:12 PM
Lovely maps in this paper!
Reposted by Dan Lewer
Sunk costs and spite are basically the vodka red bull of the examined life
June 16, 2025 at 1:12 PM
Sunk costs and spite are basically the vodka red bull of the examined life
the results
rang true with my beliefs
rang true with my beliefs
November 3, 2025 at 8:53 AM
the results
rang true with my beliefs
rang true with my beliefs
They look good though, right?
October 30, 2025 at 3:19 PM
They look good though, right?
Another statistical significance hack I just came up with - assume there IS an effect (as we knew there was before doing the trial), then test how improbable the data are, thus concluding that while our experiment was implausible, the intervention is beneficial
October 30, 2025 at 11:23 AM
Another statistical significance hack I just came up with - assume there IS an effect (as we knew there was before doing the trial), then test how improbable the data are, thus concluding that while our experiment was implausible, the intervention is beneficial
He suggested that we “write a letter to the editor”, which was odd since he was in fact the editor 😂
October 30, 2025 at 11:21 AM
He suggested that we “write a letter to the editor”, which was odd since he was in fact the editor 😂
Futurebuilders??? It was really nice! I remember Lucy but I'm sure she won't remember me. My friends Jonathan Finighan and Rob Abercrombie worked there too! But maybe Rob was later.
October 29, 2025 at 12:43 PM
Futurebuilders??? It was really nice! I remember Lucy but I'm sure she won't remember me. My friends Jonathan Finighan and Rob Abercrombie worked there too! But maybe Rob was later.
Freelance! I worked with Padraic. In that building just by London Bridge. It was MANY years ago
October 29, 2025 at 10:17 AM
Freelance! I worked with Padraic. In that building just by London Bridge. It was MANY years ago
You might not remember but you were briefly my boss at NPC ... I can't actually remember what we did together, but I do remember you very well! Hope you are well and enjoying your MSc :)
October 29, 2025 at 9:34 AM
You might not remember but you were briefly my boss at NPC ... I can't actually remember what we did together, but I do remember you very well! Hope you are well and enjoying your MSc :)
Check out table 4 and the discussion on p412. It's not a classic modern "factors associated with" using multivariable regression, but I wonder if this list of unadjusted descriptive analyses is a kind of early form of "factors associated with"
October 29, 2025 at 9:30 AM
Check out table 4 and the discussion on p412. It's not a classic modern "factors associated with" using multivariable regression, but I wonder if this list of unadjusted descriptive analyses is a kind of early form of "factors associated with"
Great example! My understand was that this came about after a pathologist in the 1940s observed many of the deaths were among infants sleeping on their front - www.jpeds.com/article/S002....
Accidental mechanical suffocation in infants
Attention is called to the excessive number of infants under 1 yearof age who are
dying from accidental mechanical suffocation. These fatalities have shown a progressive
increase each year to the poin...
www.jpeds.com
October 29, 2025 at 9:29 AM
Great example! My understand was that this came about after a pathologist in the 1940s observed many of the deaths were among infants sleeping on their front - www.jpeds.com/article/S002....
I think the "factors associated with" studies actually showed that smoking protected you from COVID! Then more focused studies such as the MR one you highlight found the opposite (obviously)?
October 29, 2025 at 9:23 AM
I think the "factors associated with" studies actually showed that smoking protected you from COVID! Then more focused studies such as the MR one you highlight found the opposite (obviously)?