California no longer under lockdown - people freak out

Seanchaidh

Elite Member
Legacy
Mar 21, 2009
5,241
3,065
118
Country
United States of America
Looking at the graph, 20 of the results I would say are outliers (basically half), which are 8 or higher or 3 and lower. You do have a whole other line of 7 results at ~7 as well, which is about 2 higher than the average (so ~40% higher, just doing basic 2/5 math, keeping it simple).
sounds like a pretty normal distribution.
 

Phoenixmgs

The Muse of Fate
Legacy
Apr 3, 2020
8,925
784
118
w/ M'Kraan Crystal
Gender
Male
Why "shouldn't" there? Isn't it possible thats just how the data distribution falls?
sounds like a pretty normal distribution.
Half the data were outliers? Do you understand what an outlier is? A fairly typical criteria for an outlier might be anything more than 2 standard deviations beyond the mean, because in science we're not allowed to take a look at the data and decide "I think I'll just ignore all the data points I feel like".

The average was 5.3. You don't even need to calculate that, the paper tells you. Never mind that the entire point of the paper is based on testing whether 2.5 or 5 is more accurate, and concluding the answer is 5.
So normal plotting of something that should be repeatable should be the opposite of a bell curve?


If you're not competent enough to realise half the material you're citing is trash, you're not competent to usefully assess the material other people are citing either.
Again, how the fuck is one person saying X works for early treatment disproved by another person citing studies for late-stage treatment (or prophylaxis use)? What cancer treatment works for terminally ill patients? I guess we have no cancer treatments whatsoever since none of them work for late-stage cancer. I'm guessing you also believe Ivermectin doesn't do anything for covid either.

You cited studies that don't even meet the requirements that you yourself said make for a good study.

Why didn't you answer the question of if 400 IUs/day is all you need of vitamin d, why does 2,000 IUs/day not even keep someone at a previous 30 ng/ml baseline level?
 

Avnger

Trash Goblin
Legacy
Apr 1, 2016
2,069
1,206
118
Country
United States
So normal plotting of something that should be repeatable should be the opposite of a bell curve?
Why do I get the feeling that you don't understand what a bell curve is, why/when it's applicable, or even how statistical analysis is done in general? Your use of "outliers" as shorthand for "data I don't agree with" alone pretty much proves that last part.
 

Agema

You have no authority here, Jackie Weaver
Legacy
Mar 3, 2009
8,598
5,962
118
So normal plotting of something that should be repeatable should be the opposite of a bell curve?
??

Firstly, reproducibility and data distribution are not the same thing.
Secondly, the data did show a bell curve, as demonstrated by the histogram I supplied.

Again, how the fuck is one person saying X works for early treatment disproved by another person citing studies for late-stage treatment (or prophylaxis use)? What cancer treatment works for terminally ill patients? I guess we have no cancer treatments whatsoever since none of them work for late-stage cancer.
The problem for claiming early use of HCQ to help with covid is the large number of studies looking at early use of HCQ for covid that show it is not effective. The way this debate went, all those months ago, is that you said it worked, were shown otherwise, then you started narrowing the parameters to more specific conditions where it was useful, and then were shown the evidence wasn't there for those narrowed parameters either. Then you went quiet on it for a while, and popped up claiming it was useful again, seemingly to me as if all that previous discussion hadn't happened.

I'm guessing you also believe Ivermectin doesn't do anything for covid either.
You're right, I don't. Or more strictly, I believe that the current evidence does not support the hypothesis that it does.

You cited studies that don't even meet the requirements that you yourself said make for a good study.
And if you understood my explanation of science, you'd understand that wasn't a problem. An individual study tends to be weak; it is a body of studies that needs to be considered. As a "representative example", however, even a relatively modest study serves a useful purpose. The problem comes when people cite individual papers as if they answer a question when they are not representative of the wider body of literature.

Why didn't you answer the question of if 400 IUs/day is all you need of vitamin d, why does 2,000 IUs/day not even keep someone at a previous 30 ng/ml baseline level?
Beacuse:
1) the overall evidence doesn't support people needing 30 ng/ml.
2) the 400 IUs guideline is conditional: it is 400 IU "for whom".
3) you are putting too much reliance on a single study - essentially you are making a specious claim.
 

Seanchaidh

Elite Member
Legacy
Mar 21, 2009
5,241
3,065
118
Country
United States of America
So normal plotting of something that should be repeatable should be the opposite of a bell curve?
Did you bother calculating what the standard deviation was? It's fine if you didn't calculate the standard deviation; I didn't either. Agema supplied a histogram which looked pretty fuckin' normal to me, though. It doesn't look like a bimodal distribution. Which means that what you are calling "outliers" are not outliers. (Actually, if it were bimodal, the two groups of results at the two modes also wouldn't be outliers, but I digress.) A normal distribution (i.e. a bell curve) will ideally have 68% of the data within 1 standard deviation and 95% within 2 standard deviations; the other 5% that are over 2 standard deviations away are what you might properly call outliers sometimes. The proportions don't have to be exactly that. In any case: large portions of the data can be substantially far away from the mean while still being an approximately normal distribution.

edit: changed a (parenthesis) to italics.
 
Last edited:

Agema

You have no authority here, Jackie Weaver
Legacy
Mar 3, 2009
8,598
5,962
118
Did you bother calculating what the standard deviation was? It's fine if you didn't calculate the standard deviation; I didn't either. Agema supplied a histogram which looked pretty fuckin' normal to me, though. It doesn't look like a bimodal distribution. Which means that what you are calling "outliers" are not outliers. (Actually, if it were bimodal, the two groups of results at the two modes also wouldn't be outliers, but I digress.) A normal distribution (i.e. a bell curve) will ideally have 68% of the data within 1 standard deviation and 95% within 2 standard deviations; the other 5% that are over 2 standard deviations away are what you might properly call outliers sometimes. The proportions don't have to be exactly that. In any case: large portions of the data can be substantially far away from the mean while still being an approximately normal distribution.

edit: changed a (parenthesis) to italics.
FYI, the sd of those of those 41 studies is 3.0, with a mean of 5.3. Quite impressively given the 68-95-99 rule, the proportion of values within one sd of the mean is 68.3% and within 2 sds is 95.1%.
 
  • Like
Reactions: Seanchaidh

Seanchaidh

Elite Member
Legacy
Mar 21, 2009
5,241
3,065
118
Country
United States of America
FYI, the sd of those of those 41 studies is 3.0, with a mean of 5.3. Quite impressively given the 68-95-99 rule, the proportion of values within one sd of the mean is 68.3% and within 2 sds is 95.1%.
Heh. Very normal indeed.
 

Seanchaidh

Elite Member
Legacy
Mar 21, 2009
5,241
3,065
118
Country
United States of America
What's the betting they settled on 68 because 69 would have undermined the seriousness with which science students should be taking data analysis?
Hmm.

Well, they'd have to make the 34 a decimal in order to have any hope of multiplying it by two to get an odd number. And while there is a more precise figure used sometimes, it isn't enough to round up to 69.



If they massaged that peak just a bit, though, harden it a little so to speak...
 

Phoenixmgs

The Muse of Fate
Legacy
Apr 3, 2020
8,925
784
118
w/ M'Kraan Crystal
Gender
Male
Why do I get the feeling that you don't understand what a bell curve is, why/when it's applicable, or even how statistical analysis is done in general? Your use of "outliers" as shorthand for "data I don't agree with" alone pretty much proves that last part.
Did you bother calculating what the standard deviation was? It's fine if you didn't calculate the standard deviation; I didn't either. Agema supplied a histogram which looked pretty fuckin' normal to me, though. It doesn't look like a bimodal distribution. Which means that what you are calling "outliers" are not outliers. (Actually, if it were bimodal, the two groups of results at the two modes also wouldn't be outliers, but I digress.) A normal distribution (i.e. a bell curve) will ideally have 68% of the data within 1 standard deviation and 95% within 2 standard deviations; the other 5% that are over 2 standard deviations away are what you might properly call outliers sometimes. The proportions don't have to be exactly that. In any case: large portions of the data can be substantially far away from the mean while still being an approximately normal distribution.

edit: changed a (parenthesis) to italics.
Figure 1 =/= a bell curve. There's 10 just on the 2.5 line alone vs 9 actually between in the average range (4-6). There's so much inconsistency and wrong with how these studies are done.
 

Phoenixmgs

The Muse of Fate
Legacy
Apr 3, 2020
8,925
784
118
w/ M'Kraan Crystal
Gender
Male
??

Firstly, reproducibility and data distribution are not the same thing.
Secondly, the data did show a bell curve, as demonstrated by the histogram I supplied.
If something is reproduce-able, how will there much data distribution?

How are there being more results (10) just on the 2.5 CPG line than there are (9) in the average range (4-6) a bell curve?

The problem for claiming early use of HCQ to help with covid is the large number of studies looking at early use of HCQ for covid that show it is not effective. The way this debate went, all those months ago, is that you said it worked, were shown otherwise, then you started narrowing the parameters to more specific conditions where it was useful, and then were shown the evidence wasn't there for those narrowed parameters either. Then you went quiet on it for a while, and popped up claiming it was useful again, seemingly to me as if all that previous discussion hadn't happened.
It was never not useful if you actually look at the data we have instead of cherry picking the few studies that said otherwise (and almost all of those, if not all, are late stage treatment or have shit P values).

You're right, I don't. Or more strictly, I believe that the current evidence does not support the hypothesis that it does.
So all you do is look at official guidelines. It took CDC until August for them to say it was airborne IIRC, everyone knew it was airborne months before that. CDC just said today that you can see people indoors if you're vaccinated; no shit Sherlock. They even hilariously said you need to "vet" people to make sure they aren't lying about getting vaccinated even though you're already vaccinated and it doesn't matter.

And if you understood my explanation of science, you'd understand that wasn't a problem. An individual study tends to be weak; it is a body of studies that needs to be considered. As a "representative example", however, even a relatively modest study serves a useful purpose. The problem comes when people cite individual papers as if they answer a question when they are not representative of the wider body of literature.
I was never using individual studies from the beginning. Look at all the early stage studies, the vast vast vast vast majority of them say it has a benefit, you'd have to cherry pick to only list the few (if there even are any) that show no benefit or worse, then you might have one or two total left that doesn't have a shit P value.

From Oct 2:
Out of 70 hydroxy studies, 56 of them showed benefits (40 were peer reviewed). Of the 14 studies that were negative or neutral, 10 of them were done on ICU patients (which hydroxy doesn't help with). It's not that smart of an idea to give hydroxy to ICU patients and especially not at those doses.
Beacuse:
1) the overall evidence doesn't support people needing 30 ng/ml.
2) the 400 IUs guideline is conditional: it is 400 IU "for whom".
3) you are putting too much reliance on a single study - essentially you are making a specious claim.
-I was never claiming that, it's just that some places say under that is insufficient, thus there's studies getting people up to that. But again, surely taking 5x what someone needs should at least maintain a 30 ng/ml level, right?
-You keep saying 400 IU is all you need. That's all you need to prevent rickets.
-I never use a single study unless there's like no studies like vitamin d.
 

Avnger

Trash Goblin
Legacy
Apr 1, 2016
2,069
1,206
118
Country
United States
Figure 1 =/= a bell curve. There's 10 just on the 2.5 line alone vs 9 actually between in the average range (4-6). There's so much inconsistency and wrong with how these studies are done.
If something is reproduce-able, how will there much data distribution?

How are there being more results (10) just on the 2.5 CPG line than there are (9) in the average range (4-6) a bell curve?
Please, for the love of God, go take an Intro to Stats class at your local community college then come back to us. An applied statistics course would be better, but even Stats 101 would be enough to show you how wrong your current arguments are.

I'll leave the actual why of the correcting to others with more patience, but your questions and statements here are essentially "If animals evolved from fish, why can they breathe air but not water? Checkmate atheists." level analysis. You have little understanding of even the basics of statistical analysis yet are still trying to make conclusions.
 

Agema

You have no authority here, Jackie Weaver
Legacy
Mar 3, 2009
8,598
5,962
118
Figure 1 =/= a bell curve.
You're right, it is not a bell curve. But that's because it is not an appropriate frequency distribution graph (histogram) to be able to form a bell curve. For much the same reason if you want to measure the distance between two points, a ruler is better than a set of weighing scales.

It was never not useful if you actually look at the data we have instead of cherry picking the few studies that said otherwise (and almost all of those, if not all, are late stage treatment or have shit P values).
I guess you've also forgotten the explanation of what a p value represents that I supplied.

So all you do is look at official guidelines. It took CDC until August for them to say it was airborne IIRC,
I'm trying to get across to you the fact there is a substantial body of literature underpinning the decisions of people who write official guidelines. See below.

And the CDC was implicitly saying covid-19 was very likely to be airborne the minute it advised everyone to wear a mask. It may well have held off on an official confirmation until properly scientifically established beyond reasonable doubt, even if it was the working assumption months previously.

I was never using individual studies from the beginning. Look at all the early stage studies, the vast vast vast vast majority of them say it has a benefit, you'd have to cherry pick to only list the few (if there even are any) that show no benefit or worse, then you might have one or two total left that doesn't have a shit P value.
I wrote a very long series of posts that explained just how wrong you were, because you'd been using a website that maliciously misrepresented studies to suggest HCQ was beneficial when in fact these studies had found it ineffective. Or if you hadn't been using this website, you had almost certainly been following some doofus who had trusted it, or had otherwise mightily fucked up in a similar fashion on their own initiative.

-You keep saying 400 IU is all you need. That's all you need to prevent rickets.
Actually, the evidence suggests the majority of children don't even that much to avoid rickets. The average dietary intake for UK children was found to be 80-160 IUs per day, and yet the incidence of rickets in the UK is just 0.5 per 100,000. 80-160 IUs is the average daily intake range for adults, too, and yet the average blood plasma concentration of Vit D was over 40 nmol/L, above the range studies associate with health deficits.

The UK's scientific advisors looked through the studies and evidence, and concluded that 400 IUs per day on average would ensure 97.5% of the population were into the range where there was no observable health deficits, as compared to the 80% from taking 80-160 IUs.

And lots more. But more to the point, the 2016 review on vitamin D conducted by the UK authorities cites a truly staggering over 500 papers. (I estimated by the number of studies cited on one page - 18, multiplied by the number of full pages with references - 30). It is thus somehwat bemusing to see you pop in with a couple of studies from the Journal of Dodgy Bullshit and the opinion of Dr John Campbell as if that's somehow superior than a report based on >500 papers conducted by an expert panel.
 

Seanchaidh

Elite Member
Legacy
Mar 21, 2009
5,241
3,065
118
Country
United States of America
Looking at the graph, 20 of the results I would say are outliers (basically half), which are 8 or higher or 3 and lower. You do have a whole other line of 7 results at ~7 as well, which is about 2 higher than the average (so ~40% higher, just doing basic 2/5 math, keeping it simple). Dr. Ames pointing out that vitamin RCTs are misinterpreted.
Figure 1 =/= a bell curve. There's 10 just on the 2.5 line alone vs 9 actually between in the average range (4-6). There's so much inconsistency and wrong with how these studies are done.
You're calling things 'outliers' that are within one standard deviation of the mean.

5.3+3=8.3
5.3-3=2.3

Around 68% of the results should be within the range from 2.3 to 8.3

5.3+6=11.3
And zero is within two SD.

Around 95% of the results should be within the range from 0 to 11.3

Those that are higher than 11.3 are a little rarer. But you'd expect weird results like that about 5% of the time if the data is normally distributed. And they make up about 5% of the data (4.878%, specifically; the closest you can get to 5% when dividing by 41). Random variation: it's a thing.

I haven't followed the conversation to know what the hell the significance of any of these numbers is, but that's the stats: you're calling things weird that aren't weird and declaring the whole set of results abnormal even though it is exceedingly normal. Is there a lot of variation? Yes. Is the variation of such a character that we should feel comfortable dismissing the results? No.
 

Agema

You have no authority here, Jackie Weaver
Legacy
Mar 3, 2009
8,598
5,962
118
Also relevant...

Recent bad news for the vitamin D - Covid-19 hypothesis here, here and also with the removal and investigation of probably the most dramatic supportive study thus far from preprint, due to concerns over whether the methodology was reportedly accurately.

In the latter case, turns out their patients might not have been so randomised after all: they assigned patients to control or supplement groups by the ward they were in, but it's believed that the patient allocation to ward was not random. Consequently it's not the sort of randomised trial they claimed, and they therefore used the wrong statistical methods to analyse the data. Whoops!
 

Seanchaidh

Elite Member
Legacy
Mar 21, 2009
5,241
3,065
118
Country
United States of America
Also relevant...

Recent bad news for the vitamin D - Covid-19 hypothesis here, here and also with the removal and investigation of probably the most dramatic supportive study thus far from preprint, due to concerns over whether the methodology was reportedly accurately.

In the latter case, turns out their patients might not have been so randomised after all: they assigned patients to control or supplement groups by the ward they were in, but it's believed that the patient allocation to ward was not random. Consequently it's not the sort of randomised trial they claimed, and they therefore used the wrong statistical methods to analyse the data. Whoops!
Ouch. And there almost certainly isn't a good way to fix that other than literally starting over from the beginning.
 

Phoenixmgs

The Muse of Fate
Legacy
Apr 3, 2020
8,925
784
118
w/ M'Kraan Crystal
Gender
Male
Please, for the love of God, go take an Intro to Stats class at your local community college then come back to us. An applied statistics course would be better, but even Stats 101 would be enough to show you how wrong your current arguments are.

I'll leave the actual why of the correcting to others with more patience, but your questions and statements here are essentially "If animals evolved from fish, why can they breathe air but not water? Checkmate atheists." level analysis. You have little understanding of even the basics of statistical analysis yet are still trying to make conclusions.
You're calling things 'outliers' that are within one standard deviation of the mean.

5.3+3=8.3
5.3-3=2.3

Around 68% of the results should be within the range from 2.3 to 8.3

5.3+6=11.3
And zero is within two SD.

Around 95% of the results should be within the range from 0 to 11.3

Those that are higher than 11.3 are a little rarer. But you'd expect weird results like that about 5% of the time if the data is normally distributed. And they make up about 5% of the data (4.878%, specifically; the closest you can get to 5% when dividing by 41). Random variation: it's a thing.

I haven't followed the conversation to know what the hell the significance of any of these numbers is, but that's the stats: you're calling things weird that aren't weird and declaring the whole set of results abnormal even though it is exceedingly normal. Is there a lot of variation? Yes. Is the variation of such a character that we should feel comfortable dismissing the results? No.
You're right, it is not a bell curve. But that's because it is not an appropriate frequency distribution graph (histogram) to be able to form a bell curve. For much the same reason if you want to measure the distance between two points, a ruler is better than a set of weighing scales.
I'm saying data like this should fit a bell curve naturally. If you got a car that's rated as 30 miles per gallon and you're getting 13 mpg, would you be satisfied by Ford or Toyota saying that's just standard deviation? Because that's within the standard deviation of those 41 studies. The point is that vitamin studies are done rather poorly (which experts in the field have said) and that's why the data points are all over the place. Sure, you aren't going to get the same result every time obviously, but very little is done in these studies to control the many variables and each study may have their own different methods obviously.

I guess you've also forgotten the explanation of what a p value represents that I supplied.
Nope, P value is the chances of a null hypothesis.

I'm trying to get across to you the fact there is a substantial body of literature underpinning the decisions of people who write official guidelines. See below.

And the CDC was implicitly saying covid-19 was very likely to be airborne the minute it advised everyone to wear a mask. It may well have held off on an official confirmation until properly scientifically established beyond reasonable doubt, even if it was the working assumption months previously.
And doctors having to quit their positions to prescribe stuff like steroids for covid because of official guidelines is a good thing? Using your medical knowledge to figure out a solution to a problem is now discouraged?

What's wrong with saying what something is thought to be when there is no established beyond reasonable doubt information?

And from articles about CDC and recommending masks April 3, it wasn't because of the virus being airborne, it was due to asymptomatic transmission. That, again, I knew nearly a month beforehand from a Michael Osterholm interview. The CDC is like the last one to the party. Surely if a normal citizen can figure out this stuff, the CDC can.

The UK's guidelines have also been asinine like the "rule of 6" applying to outdoors that was backed by literally no science whatsoever. One of the hospitals I work at still has letters across so many windows saying "wash your hands" when that doesn't do much of anything. I still only wash my hands before eating lunch and after the bathroom because the virus barely spreads via contact surfaces.

I wrote a very long series of posts that explained just how wrong you were, because you'd been using a website that maliciously misrepresented studies to suggest HCQ was beneficial when in fact these studies had found it ineffective. Or if you hadn't been using this website, you had almost certainly been following some doofus who had trusted it, or had otherwise mightily fucked up in a similar fashion on their own initiative.
I told you I didn't use any of the graphs from that website, I only use it as a database for looking at the studies. Many doctors say HCQ works for early treatment, the studies say that (looking at them in their entirety like you say), but you keep linking studies that are for late-stage treatment that literally everyone knows HCQ doesn't work for (literally nobody is arguing that). Just because something doesn't work for late-stage treatment doesn't mean it can't help as early treatment or else we'd have literally no cancer treatments.

Actually, the evidence suggests the majority of children don't even that much to avoid rickets. The average dietary intake for UK children was found to be 80-160 IUs per day, and yet the incidence of rickets in the UK is just 0.5 per 100,000. 80-160 IUs is the average daily intake range for adults, too, and yet the average blood plasma concentration of Vit D was over 40 nmol/L, above the range studies associate with health deficits.

The UK's scientific advisors looked through the studies and evidence, and concluded that 400 IUs per day on average would ensure 97.5% of the population were into the range where there was no observable health deficits, as compared to the 80% from taking 80-160 IUs.

And lots more. But more to the point, the 2016 review on vitamin D conducted by the UK authorities cites a truly staggering over 500 papers. (I estimated by the number of studies cited on one page - 18, multiplied by the number of full pages with references - 30). It is thus somehwat bemusing to see you pop in with a couple of studies from the Journal of Dodgy Bullshit and the opinion of Dr John Campbell as if that's somehow superior than a report based on >500 papers conducted by an expert panel.
Doesn't change any of the facts that the recommendations are set to stop the very worst side effects of vitamin d deficiency, that's what the flow chart in the official UK recommendations is telling people. It's basically do you have any of these ricket-like symptoms? Yes, take vitamin d supplement. No, you're fine.