The shower explanation for exposure and dose is very good, because while we are certain that I did not get dosed with 25 gallons in 10 minutes, we can see very quickly that we don't have any idea how much I was dosed with. I might have drunk none or a mouthful or quite a bit. But someone with much less exposure, say a fellow who spent 30 seconds beside a cup of water, might have been dosed with considerably more.
Briggs, this paper is an absolute disaster. It’s one of the worst I’ve ever read. Rubbish I say. And frankly, I don’t trust Italians. Let me explain why.
During my second semester of advanced calculus, I had an Italian professor. At first, he seemed competent enough, but I started noticing some peculiar patterns. The class was small 6 as I recall, and consisted primarily of international students, mostly from Germany and Jordan, with just two U.S. citizens: myself and Bill, a PhD candidate in physics.
Bill and I were demonstrably stronger in proofs and overall mathematical ability. We knew this because the international students frequently sought our help to understand and complete assignments. Despite this, our test grades consistently lagged behind theirs.
This really upset me and I became suspicious. So I marched to the department chair’s office at the end of the term, demanding an explanation. The chair didn’t mince words: “Oh, that’s Merullo. He hates Americans.”
With recent advances in "AI" it should now be much less difficult and time consuming than it was even last year to comb through all the major medical and medically-related journals over the period since perhaps Jan 1 2004 for articles claiming world changing results from stats/prob models. Each such usage could then be classified according to the errors, if any, made in in either developing or interpreting the model (explicit or not) used. That result could then be used to predict whether or not the work could be replicated.. (or, in some cases, might be fraudulent).
My bet? more than three quarters would fail on the replicable prediction.
There's an article on wattsupwiththat this morning (https://wattsupwiththat.com/2024/12/13/climate-science-settled-until-its-not/ ) that bears on this: it's a somewhat gushy thing about the recent discovery that an apparently major variable has been missing from climate models - but, (surprise!) the authors do not ask the obvious question: if the models are missing important variables, how is that that they retro-cast so well? (umm.. on 2nd thought, that 3/4 guess may be optimistic).
"By administering a questionnaire, we investigated occupational history and leisure-time habits of the 41 ALS patients diagnosed in the 1995-2006 period, and of 82 age- and sex-matched randomly sampled population controls."
As they compare the self-reported exposure numbers from the 41 with ALS to the "matched randomly sampled" 82 without ALS, I am reminded of drug companies touting Relative Risk Reduction, rather than Absolute Risk Reduction. In this case, touting Relative Risk Addition for the self-reported use of pesticides. "I developed ALS; was it the diazinon I used on my roses...?"
And thank you for your ongoing efforts over the past year in exposing what passes for my brain to some truly impenetrable maths.
Your formulas bounce off my eyeballs, but your plain-language explanations of the fallacies inherent in probabilities and models and cause/effect, etc., have been really helpful and I've won some arguments with things I have learned here.
In the 1970's, Dr. Spocks film about the impending ice age and the resultant mass starvations made the rounds and the news and newspapers and magazines were full of hersterical stories about how scientists were absolutely certain that air pollution and acid rain and running out of oil, etc., had us on the express train to a cold and hungry death by the year 2000. I remember at that time thinking this sounded like nonsense and the entirety of the Klimate Kerfuffle since then has done nothing to convince me otherwise. But repeatedly saying "ain't so!" is not really an effective counter-argument.
So what I HAVE learned from your classes is to ignore your formula's and frequent descents into the Faustian regions of mathematical hell and focus instead on what you say when you use words that I can understand. I now have enough of an understanding of how/why models fail, why probabilities fail, and how correlation can drive claims of causation, and am able to produce some really confused looks on the faces of those who claim "carbon" is going to kill us all and that right soon. And I don't have to use a single formula!
Anyway, thanks for teaching me these things, and don't forget that the only use some of your students find for Jaynes's, et al's, books is when employed as doorstops.
Yes in situations such as Briggs described about a chemistry experiment. The problem has always been that scientism spread from the hard sciences into the soft "sciences" where control is tepid at best and ignored for the most part.
I'm having trouble with the examples of 'exposure' vs. 'dose'. Is the simple availability of something 'exposure'? If so, how do we define 'available'?
Getting a cable package that includes CNN, but never watching CNN, means that CNN is available but that I do not come into contact with it. For that matter, CNN is equally available if I do not get the cable package at all, but could have it if I paid the subscription.
My sense of 'exposure' in this case would be if I live in a home with CNN turned on much of the time, whether or not I am paying attention to it. It might be affecting my ideas about the world, either consciously or subliminally. A 'dose' seems more intentional, and might be the case of me sitting on the sofa, turning on the TV, and consciously watching a CNN program to be informed or entertained.
The shower example seems perverse. If I am taking a shower, I am doing it to cleanse myself, not to rehydrate my biosystem. The 'dose' is the water I apply to my surface, just as a dose of sunscreen is what I put on my skin, not something I swallow. By analogy to the original example of CNN exposure being subscription to a cable service that includes CNN, then my 'exposure' to water would be all the water I could get by turning on all my faucets and letting them run forever, even if I do not actually do that.
Maybe this has been covered earlier in the course, but I think this article needs a little more work to define 'exposure' vs. 'dose' in general terms, for the benefit of those of us who have skipped ahead. Thanks for all your efforts in producing this course!
If the pesticide is securely encased, then how am I exposed? In that scenario, my system is no more touched by the pesticide than if it were a million miles away.
It seems that way, but I'm still looking for what 'exposure' actually means, in a functional sense. In this example, it would seem to mean 'proximity.'
Thinking it over, especially in the context of pesticides, I would propose that 'exposure' is any kind of contact between the biological system (me) and the dubious substance in question. 'Dose' is a measured amount of that contact. 'Exposure' is a simple yes or no.
If I use that pesticide I brought home on my lawn, then I have been exposed, and so have my neighbors. Minimally, we can smell it, and are inhaling some of the vapors it gives off. I may get some of it on my hands, and later eat food with those same hands. Someone may walk on my pesticide-covered grass, track it into a house, and the baby of that house crawls on the floor and then sucks its thumb.
That's what I would call 'exposure': someone has likely come into direct contact with the pesticide, whether gastro-intestinally, through respiration, or through eye- or skin contact. But there may be no practical way to measure it, or even prove that it actually happened. 'Dose' is only if we know how the contact took place and the amount involved.
You've hit on the problem, all right. Exposure can mean those things, or it can mean next to nothing. It depends on the paper. It runs all over the place.
The shower explanation for exposure and dose is very good, because while we are certain that I did not get dosed with 25 gallons in 10 minutes, we can see very quickly that we don't have any idea how much I was dosed with. I might have drunk none or a mouthful or quite a bit. But someone with much less exposure, say a fellow who spent 30 seconds beside a cup of water, might have been dosed with considerably more.
Briggs, this paper is an absolute disaster. It’s one of the worst I’ve ever read. Rubbish I say. And frankly, I don’t trust Italians. Let me explain why.
During my second semester of advanced calculus, I had an Italian professor. At first, he seemed competent enough, but I started noticing some peculiar patterns. The class was small 6 as I recall, and consisted primarily of international students, mostly from Germany and Jordan, with just two U.S. citizens: myself and Bill, a PhD candidate in physics.
Bill and I were demonstrably stronger in proofs and overall mathematical ability. We knew this because the international students frequently sought our help to understand and complete assignments. Despite this, our test grades consistently lagged behind theirs.
This really upset me and I became suspicious. So I marched to the department chair’s office at the end of the term, demanding an explanation. The chair didn’t mince words: “Oh, that’s Merullo. He hates Americans.”
This paper is one of a legion.
Yes - well...
With recent advances in "AI" it should now be much less difficult and time consuming than it was even last year to comb through all the major medical and medically-related journals over the period since perhaps Jan 1 2004 for articles claiming world changing results from stats/prob models. Each such usage could then be classified according to the errors, if any, made in in either developing or interpreting the model (explicit or not) used. That result could then be used to predict whether or not the work could be replicated.. (or, in some cases, might be fraudulent).
My bet? more than three quarters would fail on the replicable prediction.
There's an article on wattsupwiththat this morning (https://wattsupwiththat.com/2024/12/13/climate-science-settled-until-its-not/ ) that bears on this: it's a somewhat gushy thing about the recent discovery that an apparently major variable has been missing from climate models - but, (surprise!) the authors do not ask the obvious question: if the models are missing important variables, how is that that they retro-cast so well? (umm.. on 2nd thought, that 3/4 guess may be optimistic).
"By administering a questionnaire, we investigated occupational history and leisure-time habits of the 41 ALS patients diagnosed in the 1995-2006 period, and of 82 age- and sex-matched randomly sampled population controls."
As they compare the self-reported exposure numbers from the 41 with ALS to the "matched randomly sampled" 82 without ALS, I am reminded of drug companies touting Relative Risk Reduction, rather than Absolute Risk Reduction. In this case, touting Relative Risk Addition for the self-reported use of pesticides. "I developed ALS; was it the diazinon I used on my roses...?"
Also, why 82? Why not 41? Why not 410?
Merry Christmas!
Thanks!
A blessed and joyous Advent and Christmastide to you and yours as well!
Elegant explanation.
Sir, Merry Christmas to you as well!
And thank you for your ongoing efforts over the past year in exposing what passes for my brain to some truly impenetrable maths.
Your formulas bounce off my eyeballs, but your plain-language explanations of the fallacies inherent in probabilities and models and cause/effect, etc., have been really helpful and I've won some arguments with things I have learned here.
In the 1970's, Dr. Spocks film about the impending ice age and the resultant mass starvations made the rounds and the news and newspapers and magazines were full of hersterical stories about how scientists were absolutely certain that air pollution and acid rain and running out of oil, etc., had us on the express train to a cold and hungry death by the year 2000. I remember at that time thinking this sounded like nonsense and the entirety of the Klimate Kerfuffle since then has done nothing to convince me otherwise. But repeatedly saying "ain't so!" is not really an effective counter-argument.
So what I HAVE learned from your classes is to ignore your formula's and frequent descents into the Faustian regions of mathematical hell and focus instead on what you say when you use words that I can understand. I now have enough of an understanding of how/why models fail, why probabilities fail, and how correlation can drive claims of causation, and am able to produce some really confused looks on the faces of those who claim "carbon" is going to kill us all and that right soon. And I don't have to use a single formula!
Anyway, thanks for teaching me these things, and don't forget that the only use some of your students find for Jaynes's, et al's, books is when employed as doorstops.
Ah, yes, we did that one:
https://wmbriggs.substack.com/p/the-logic-of-global-warming-aka-climate
Can you ever have a truly predictive model in a world with an incalculable number of variables in any situation?
Yes in situations such as Briggs described about a chemistry experiment. The problem has always been that scientism spread from the hard sciences into the soft "sciences" where control is tepid at best and ignored for the most part.
Sure. Casinos use them all the time. As do you, but you were likely unaware. Any situation in which there is uncertainty.
I'm having trouble with the examples of 'exposure' vs. 'dose'. Is the simple availability of something 'exposure'? If so, how do we define 'available'?
Getting a cable package that includes CNN, but never watching CNN, means that CNN is available but that I do not come into contact with it. For that matter, CNN is equally available if I do not get the cable package at all, but could have it if I paid the subscription.
My sense of 'exposure' in this case would be if I live in a home with CNN turned on much of the time, whether or not I am paying attention to it. It might be affecting my ideas about the world, either consciously or subliminally. A 'dose' seems more intentional, and might be the case of me sitting on the sofa, turning on the TV, and consciously watching a CNN program to be informed or entertained.
The shower example seems perverse. If I am taking a shower, I am doing it to cleanse myself, not to rehydrate my biosystem. The 'dose' is the water I apply to my surface, just as a dose of sunscreen is what I put on my skin, not something I swallow. By analogy to the original example of CNN exposure being subscription to a cable service that includes CNN, then my 'exposure' to water would be all the water I could get by turning on all my faucets and letting them run forever, even if I do not actually do that.
Maybe this has been covered earlier in the course, but I think this article needs a little more work to define 'exposure' vs. 'dose' in general terms, for the benefit of those of us who have skipped ahead. Thanks for all your efforts in producing this course!
For pesticides, suppose you carry a case home. That’s exposure. Later you might accidentally ingest some. That’s dose.
If the pesticide is securely encased, then how am I exposed? In that scenario, my system is no more touched by the pesticide than if it were a million miles away.
That's just what exposed means. It is not dose. Screwy ain't it.
It seems that way, but I'm still looking for what 'exposure' actually means, in a functional sense. In this example, it would seem to mean 'proximity.'
Thinking it over, especially in the context of pesticides, I would propose that 'exposure' is any kind of contact between the biological system (me) and the dubious substance in question. 'Dose' is a measured amount of that contact. 'Exposure' is a simple yes or no.
If I use that pesticide I brought home on my lawn, then I have been exposed, and so have my neighbors. Minimally, we can smell it, and are inhaling some of the vapors it gives off. I may get some of it on my hands, and later eat food with those same hands. Someone may walk on my pesticide-covered grass, track it into a house, and the baby of that house crawls on the floor and then sucks its thumb.
That's what I would call 'exposure': someone has likely come into direct contact with the pesticide, whether gastro-intestinally, through respiration, or through eye- or skin contact. But there may be no practical way to measure it, or even prove that it actually happened. 'Dose' is only if we know how the contact took place and the amount involved.
Thoughts?
You've hit on the problem, all right. Exposure can mean those things, or it can mean next to nothing. It depends on the paper. It runs all over the place.
We'll see something far worse next time.