Sean Harrison, PhD
@sean-h.bsky.social
Also, agreed, I have zero interest in cause-specific mortality, it's a crappy outcome that is borderline uninterpretable (euthanasia would do wonders for cause-specific mortality stats).
All-cause mortality all the way.
All-cause mortality all the way.
December 1, 2025 at 11:35 AM
Also, agreed, I have zero interest in cause-specific mortality, it's a crappy outcome that is borderline uninterpretable (euthanasia would do wonders for cause-specific mortality stats).
All-cause mortality all the way.
All-cause mortality all the way.
Or:
"A small number of people could have lived (slighly) longer, but lots of people didn't have unnecessary investigations and treatment that reduced their quality of life!"
vs
"Lives were saved, and some people had unnecessary investigations and treatments"
"A small number of people could have lived (slighly) longer, but lots of people didn't have unnecessary investigations and treatment that reduced their quality of life!"
vs
"Lives were saved, and some people had unnecessary investigations and treatments"
December 1, 2025 at 11:33 AM
Or:
"A small number of people could have lived (slighly) longer, but lots of people didn't have unnecessary investigations and treatment that reduced their quality of life!"
vs
"Lives were saved, and some people had unnecessary investigations and treatments"
"A small number of people could have lived (slighly) longer, but lots of people didn't have unnecessary investigations and treatment that reduced their quality of life!"
vs
"Lives were saved, and some people had unnecessary investigations and treatments"
I suspect this comes up a lot in pro-screening ideas: there's a chance of "saving a life", so why not do it?
I'm very QALY & cost-effectiveness focused, but QALYs do not make an emotionally compelling argument.
"Your quality of life is as important as your quantity of life" doesn't resonate...
I'm very QALY & cost-effectiveness focused, but QALYs do not make an emotionally compelling argument.
"Your quality of life is as important as your quantity of life" doesn't resonate...
December 1, 2025 at 11:33 AM
I suspect this comes up a lot in pro-screening ideas: there's a chance of "saving a life", so why not do it?
I'm very QALY & cost-effectiveness focused, but QALYs do not make an emotionally compelling argument.
"Your quality of life is as important as your quantity of life" doesn't resonate...
I'm very QALY & cost-effectiveness focused, but QALYs do not make an emotionally compelling argument.
"Your quality of life is as important as your quantity of life" doesn't resonate...
Even knowing there was a misprint, I missed it!
December 1, 2025 at 11:27 AM
Even knowing there was a misprint, I missed it!
"By continuing some of our current practices, the only people we will mislead are our colleagues and our students"
Ahahaha, hate that this is still an issue 50 years later, but love that someone did something very similar to me 50 YEARS AGO.
Bet they were equally irritated to have to do it.
Ahahaha, hate that this is still an issue 50 years later, but love that someone did something very similar to me 50 YEARS AGO.
Bet they were equally irritated to have to do it.
December 1, 2025 at 11:26 AM
"By continuing some of our current practices, the only people we will mislead are our colleagues and our students"
Ahahaha, hate that this is still an issue 50 years later, but love that someone did something very similar to me 50 YEARS AGO.
Bet they were equally irritated to have to do it.
Ahahaha, hate that this is still an issue 50 years later, but love that someone did something very similar to me 50 YEARS AGO.
Bet they were equally irritated to have to do it.
I wonder if we should stop saying "lives saved" in contexts like this.
The emotional weighting is so much larger than: "X additional years of life", where X is not massive given age at diagnosis...
Also, are MRIs standard now!?
You stop doing PCa research for 8 years and suddenly...
The emotional weighting is so much larger than: "X additional years of life", where X is not massive given age at diagnosis...
Also, are MRIs standard now!?
You stop doing PCa research for 8 years and suddenly...
December 1, 2025 at 10:11 AM
I wonder if we should stop saying "lives saved" in contexts like this.
The emotional weighting is so much larger than: "X additional years of life", where X is not massive given age at diagnosis...
Also, are MRIs standard now!?
You stop doing PCa research for 8 years and suddenly...
The emotional weighting is so much larger than: "X additional years of life", where X is not massive given age at diagnosis...
Also, are MRIs standard now!?
You stop doing PCa research for 8 years and suddenly...
And, of course, a 12- or 18-bore spring-loaded needle shot into your prostate from your rectum isn't exactly harm-free...
December 1, 2025 at 10:04 AM
And, of course, a 12- or 18-bore spring-loaded needle shot into your prostate from your rectum isn't exactly harm-free...
This stuff is super interesting. I'm hoping that the propagation of reference management software alleviates some of the "incorrect elements of citations", but obviously will do nothing to stem the blight of AI-generated citations.
Hopefully people are learning, but you never know...
Hopefully people are learning, but you never know...
December 1, 2025 at 10:01 AM
This stuff is super interesting. I'm hoping that the propagation of reference management software alleviates some of the "incorrect elements of citations", but obviously will do nothing to stem the blight of AI-generated citations.
Hopefully people are learning, but you never know...
Hopefully people are learning, but you never know...
"The findings indicate that not only do we lack data from which to derive an accurate numerical estimate, but we lack a measurement system within which any numerical estimate would be meaningful."
Ahaha, love burns like this in journal articles!
Ahaha, love burns like this in journal articles!
December 1, 2025 at 9:59 AM
"The findings indicate that not only do we lack data from which to derive an accurate numerical estimate, but we lack a measurement system within which any numerical estimate would be meaningful."
Ahaha, love burns like this in journal articles!
Ahaha, love burns like this in journal articles!
"ChatGPT take the wheel!"
November 29, 2025 at 1:15 PM
"ChatGPT take the wheel!"
There's a lot wrong with using an LLM instead of doing the thinking oneself.
But here it seems like no one has the skills to determine whether the code is correct.
So it's blind faith in something known to, quite often, give incorrect (but superficially correct) results.
But here it seems like no one has the skills to determine whether the code is correct.
So it's blind faith in something known to, quite often, give incorrect (but superficially correct) results.
November 29, 2025 at 8:57 AM
There's a lot wrong with using an LLM instead of doing the thinking oneself.
But here it seems like no one has the skills to determine whether the code is correct.
So it's blind faith in something known to, quite often, give incorrect (but superficially correct) results.
But here it seems like no one has the skills to determine whether the code is correct.
So it's blind faith in something known to, quite often, give incorrect (but superficially correct) results.
People thinking chatbots are real has been a thing for, surprisingly, almost 60 years (1966), and received the name "The ELIZA effect" after the program: en.wikipedia.org/wiki/ELIZA_e...
![Quote from wikipedia: Weizenbaum first implemented ELIZA in his own SLIP list-processing language, where, depending upon the initial entries by the user, the illusion of human intelligence could appear, or be dispelled through several interchanges.[2] Some of ELIZA's responses were so convincing that Weizenbaum and several others have anecdotes of users becoming emotionally attached to the program, occasionally forgetting that they were conversing with a computer.[3][page needed] Weizenbaum's own secretary reportedly asked Weizenbaum to leave the room so that she and ELIZA could have a real conversation. Weizenbaum was surprised by this, later writing: "I had not realized ... that extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people."[23]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:yv43iheuhokouafy55y4xjcb/bafkreigrxco4q5lcyp5b47edyafvkgkbpbep5i57jt5ypnjuijzoo4x5za@jpeg)
November 27, 2025 at 5:02 PM
People thinking chatbots are real has been a thing for, surprisingly, almost 60 years (1966), and received the name "The ELIZA effect" after the program: en.wikipedia.org/wiki/ELIZA_e...
Did you mean:
"... - you *don't* achieve this by ..."
or
"... - you *can* achieve this by ..."
?
"... - you *don't* achieve this by ..."
or
"... - you *can* achieve this by ..."
?
November 27, 2025 at 1:36 PM
Did you mean:
"... - you *don't* achieve this by ..."
or
"... - you *can* achieve this by ..."
?
"... - you *don't* achieve this by ..."
or
"... - you *can* achieve this by ..."
?
Wild guess:
Blocks of 4 had, FOR WHATEVER REASON, different chances of being 2/2 or 3/1, to make the overall randomization 3:2 after each set of 5 blocks.
e.g. 1211 2211 2112 1121 2121
Never seen it done, it's stupid (blocks of 5 would be perfect?), but it would work.
Probably.
Blocks of 4 had, FOR WHATEVER REASON, different chances of being 2/2 or 3/1, to make the overall randomization 3:2 after each set of 5 blocks.
e.g. 1211 2211 2112 1121 2121
Never seen it done, it's stupid (blocks of 5 would be perfect?), but it would work.
Probably.
November 27, 2025 at 9:25 AM
Wild guess:
Blocks of 4 had, FOR WHATEVER REASON, different chances of being 2/2 or 3/1, to make the overall randomization 3:2 after each set of 5 blocks.
e.g. 1211 2211 2112 1121 2121
Never seen it done, it's stupid (blocks of 5 would be perfect?), but it would work.
Probably.
Blocks of 4 had, FOR WHATEVER REASON, different chances of being 2/2 or 3/1, to make the overall randomization 3:2 after each set of 5 blocks.
e.g. 1211 2211 2112 1121 2121
Never seen it done, it's stupid (blocks of 5 would be perfect?), but it would work.
Probably.
Also a fan of the "lifetime gift threshold", where individuals are unable to receive any gifts above a certain amount, including inheritance.
Lot of bereaucracy though.
Lot of bereaucracy though.
November 24, 2025 at 3:21 PM
Also a fan of the "lifetime gift threshold", where individuals are unable to receive any gifts above a certain amount, including inheritance.
Lot of bereaucracy though.
Lot of bereaucracy though.
Declaring something "important" or "not important" on the basis of effect size alone is like declaring something "important" or "not important" on the basis of a p value alone: you don't have enough information to make that call (again, unless effect size is 0 and so p=1).
November 24, 2025 at 9:12 AM
Declaring something "important" or "not important" on the basis of effect size alone is like declaring something "important" or "not important" on the basis of a p value alone: you don't have enough information to make that call (again, unless effect size is 0 and so p=1).
"No one's going to say 'answer this question but don't use the internet', so why are we still doing exams under exam conditions?
Why do we make students write at all? They'll be using computers!
Why do we make students learn?
Or think?"
Why do we make students write at all? They'll be using computers!
Why do we make students learn?
Or think?"
October 20, 2025 at 5:00 PM
"No one's going to say 'answer this question but don't use the internet', so why are we still doing exams under exam conditions?
Why do we make students write at all? They'll be using computers!
Why do we make students learn?
Or think?"
Why do we make students write at all? They'll be using computers!
Why do we make students learn?
Or think?"
There's also the argument, which I haven't seen elsewhere, that the costs of treatment for cancers that will never cause harm mean funding will be reduced elsewhere in the NHS.
So one or 2 deaths prevented *from prostate cancer*.
Unknown numbers caused by reduction in services elsewhere.
So one or 2 deaths prevented *from prostate cancer*.
Unknown numbers caused by reduction in services elsewhere.
October 14, 2025 at 6:23 PM
There's also the argument, which I haven't seen elsewhere, that the costs of treatment for cancers that will never cause harm mean funding will be reduced elsewhere in the NHS.
So one or 2 deaths prevented *from prostate cancer*.
Unknown numbers caused by reduction in services elsewhere.
So one or 2 deaths prevented *from prostate cancer*.
Unknown numbers caused by reduction in services elsewhere.
Anyway, that's my (possibly incorrect) contribution to Lord's paradox: there's no paradox, it's not about interpretation, statistician 2 is just wrong.
Statistician 1 is right, if you can make some (fairly heroic) assumptions.
I didn't see anyone else conclude this, but they could have!
Statistician 1 is right, if you can make some (fairly heroic) assumptions.
I didn't see anyone else conclude this, but they could have!
October 12, 2025 at 3:06 PM
Anyway, that's my (possibly incorrect) contribution to Lord's paradox: there's no paradox, it's not about interpretation, statistician 2 is just wrong.
Statistician 1 is right, if you can make some (fairly heroic) assumptions.
I didn't see anyone else conclude this, but they could have!
Statistician 1 is right, if you can make some (fairly heroic) assumptions.
I didn't see anyone else conclude this, but they could have!
Incidentally, this is why the initial weight regression coefficients aren't 1 in the sex-specific regressions: measurement error pushes them away from 1.
Sex takes up some of the slack through its effect on true weight, and the constant takes up the rest.
Sex takes up some of the slack through its effect on true weight, and the constant takes up the rest.
October 12, 2025 at 3:06 PM
Incidentally, this is why the initial weight regression coefficients aren't 1 in the sex-specific regressions: measurement error pushes them away from 1.
Sex takes up some of the slack through its effect on true weight, and the constant takes up the rest.
Sex takes up some of the slack through its effect on true weight, and the constant takes up the rest.