April 26, 2020

"Density alone doesn’t seem to account for the scale of the differential between New York’s fatality rates and those of other cities."

"New York has twice the density of London but three times the deaths, and the differential is even higher [comparing NYC to] cities such as San Francisco and Los Angeles. Deaths have occurred disproportionately in poorer areas, where the incidence of long untreated morbidities such as heart disease and diabetes have contributed significantly. But the same is true in all other cities. The high dependence on mass transit also seems to be a factor. In other major cities, car commutes are much more common. As Joel Kotkin, a scholar of cities at Chapman University in California, says, it may be the lethal convergence of all three factors. 'If you put together density, levels of poverty and reliance on a mass-transit system, you have a hat trick,' he told me.… But even that may not explain the extent of New York’s unique catastrophe. Around the world, the highest death rates have occurred where hospital systems were overwhelmed in the early stages of the crisis. This is especially true in northern Italy. Anecdotally, at least, it seems that the same happened in New York: Large numbers of sick people never got to hospitals, arrived too late or, in the impossible circumstances that medical personnel were confronted with, were given ineffective treatment.… It will be a while before we get a proper understanding of what went so tragically wrong...."

From "The Covid-19 Catastrophe Unfolding in New York Is Unique" (Wall Street Journal), quoted at my son John's Facebook page.

John writes:
I'm not sure this is a logical argument:
"Density alone doesn’t seem to account for the scale of the differential between New York’s fatality rates and those of other cities. New York has twice the density of London but three times the deaths, and the differential is even higher for cities such as San Francisco and Los Angeles."
Doesn't that assume there's a linear relationship between density and infection rates, and isn't that not necessarily the case?
My question is about the comparison of New York to northern Italy, where hospitals were overrun. Were NY hospitals overrun? I thought they weren't.  I think the 3 factors named — density, reliance on mass-transit, and the bad health conditions represented by the term "poverty" — are enough to explain what happened. These things are interactive. Shouldn't we talk about Bayes theorem?

219 comments:

1 – 200 of 219   Newer›   Newest»
Shouting Thomas said...

Italy is, I believe, ranked #5 in median age of population.

Italian woman have been declining to have children for decades.

Italy’s population is very predominantly elderly. That accounts for the high death rates.

MikeR said...

Some NY hospitals were indeed overrun. I recall seeing an article about Maimonides.

Lucid-Ideas said...

NYC is a hellhole. I've been to most of the world's great metros and NYC is propped up by media visibility, it's ratio of single women to single men (look it up), and tbr gullibility of African potentates who believe Baywatch. It is a horrible city infested with intellectual turds, that goes doubly for Wall Street. I would not shed a tear if it was inundated by water or fractured from the continent in some way.

You can shine an apple. You can't polish that turd.

rhhardin said...

I don't see Bayes' theorem in there. Bayes' theorem says that if you're black you're more likely to be a criminal, in the absence of other information.

MikeR said...

Sorry - what about Bayes' theorem?

MayBee said...

NY is dirty. Tokyo, London, Chicago, and LA are all cleaner. But I still think there is something about the infrastructure of NY that is helping spread it.

Balfegor said...

Anecdotally, people presenting with coronavirus-like symptoms (fever, difficulty breathing) who were nevertheless able to move about were not tested and
just sent home to recover. That sounds like the hospital system being stretched past capacity. Also, while Javits and the hospital ship turned out to be mostly unused, there were stories recently about nursing homes pleading to have the coronavirus patients that had been transferred to them by order of the governor retransferred to Javits or the ship, in order to protect staff and other patients given the lack of PPE and appropriate facilities for isolating infectious patients, only to be denied. The statistics that suggest NYC's hospitals weren't overwhelmed seem somewhat suspect to me. No evidence it was intentional, but I wouldn't be surprised if there was pressure to game the metrics in order to minimise the city's failure.

rcocean said...

How absurd. SF is nothing like New York. NYC has 8 million people and large numbers can't get around with taking the subway, Bus, or taxi. Most people is SF have cars. No street in SF is jammed with people 24/7 like Times Square or 5th Ave. The population in SF is largely Asian and rich white liberals, NYC has huge amounts of blacks and Hispanics.

The outer boroughs of London include lots of well-to-do almost self-contained ares. NYC - huge numbers of poor people outside Manhattan and large numbers riding the subway into Manhattan to work. Large number go to central London but its not the same. Above the demographics are different.

Why not compare Tokyo to NYC? More people ride subway. Density is even higher. Or Seoul ?

rcocean said...

Bayer's Corollary: Take an aspirin - it'll cure anything.

rcocean said...

My wife and I had some tough Flu back in early Feb. - we're hoping it was actually CV Virus. Certainly it was the worst flu either of us ever had. Of course, my wife never gets sick, except of me, so that's no saying much.

Cheryl said...

"Were NYC hospitals overrun?" I would love to know the answer to that question. I just did a search of the news and it seemed like there were hospitals that had to send patients elsewhere for treatment, but that isn't no treatment. That's just balancing out the load. And the Javits Center field hospital and Mercy ship seem to have had extremely low usage rates. It does seem that military personnel were needed to fill in for health care workers who were themselves getting sick.

I saw someone on Facebook praise Gov. Cuomo for his work, "in spite of not getting any federal support." Is this true in any way? It seemed to me that Cuomo and Trump worked really well together, and NYC/NY got everything they needed and more. Am I not seeing something?

rehajm said...

Are they counting differently? What about counting covid positive deaths by other factors as covid deaths to pump up the stats?

Why does China have so few deaths for a country it's size? It's a mystery...

rcocean said...

SF Chronicle says first CV patients' heart "exploded". Scaremongering at its finest.

JML said...

Yes, many people in NYC are dense.

rcocean said...

Controversy over padding death count is interesting, but not applicable to NY-NJ-Conn since we have so many deaths. Cases in Hospital for CV-19 are probably accurate, so look at death vs. CV-19 Hospital cases. Excessive ratio of deaths to Hospitalization might be padded death count. OR could be bad medical practice.

GatorNavy said...

As per Cuomo’s order, nursing homes had to admit patients with COVID-19 symptoms. As per Cuomo and DeBlasio the subways were not shut down, neither were the buses and masks or other PPE was required. Lastly, the nonsensical restrictions placed on caregivers by the Cuomo caused further spread of infections during the early phase of the disease as per ED staff.

Fernandinande said...

I don't see Bayes' theorem in there.

Me neither, but how would you weigh in on the idea that
#cases ~ density^2
since density is (mostly) two-dimensional?

Bob Boyd said...

My theory about New York's high rate of infection is that using the word the "fuck" sprays more virus-laden droplets than most words.

Spiros said...

Data is still coming in, but initial reports are that 95% of the total COVID-19 deaths in Europe are men older than 60. In the U.S., fatality rates are much, much more evenly spread for different age groups. Many more young Americans are dying from the disease. Young Americans, especially young African-Americans and Latinos, are more likely to suffer from asthma, diabetes and hypertension than their European and Asian counterparts. Americans are unhealthy. Bad news.

rcocean said...

Correlation between old age/pre-existing conditions and CV-19 deaths/hospitalizations is almost exact. So you'd need to compare those groups in SF/London Vs. NYC to get a real answer. Which no one really cares about. We've just kibitzing - its just coffee talk. No big Whoop.

robother said...

Ann is right, exponential growth is to be expected: the more crowded the subway car is, spread of virus is exponential, not linear. Barbie also was right: "Math is hard," especially for journalism majors.

PJ said...

NY hospitals may not have been actually overwhelmed (for the most part, anyway), but there was certainly a widespread fear that they would be, and the fear may have been as potent as the fact when it came to undertreatment of COVID-19 victims. Also, there seems to be no uniformity from place to place as to how COVID-19 deaths are counted, and perhaps NY is atttributing deaths to COVID-19 more “generously.”

Fernandinande said...

Maybe NY just needs more people like David The Narc.

Cheryl said...

As far as counting other deaths as COVID-19, here's an example from Ventura County CA, where a 37 yo man died of a drug overdose but was counted as COVID because he was positive (for the virus, not for life, apparently) when he died. "Ventura County's coronavirus death toll increased to 16 on Thursday as county officials reported two additional deaths, including a 37-year-old man. The man died as a result of a drug overdose while infected with COVID-19, a significant contributing condition, according to county spokeswoman Ashley Bautista." Here's the link: https://amp.vcstar.com/amp/3015868001?fbclid=IwAR1Owi6iP5ZQZZMg4OY4epWui-vuL8tPP1hzKPw6XuTednNxKDS1VyHlRCA&__twitter_impression=true

It's stories like this that make me question the demographics of the death count.

Ann Althouse said...

Bayes theorem is about the way probabilities of different factors combine. Your intuitive sense of how likely a set of factors is to produce a particular effect can be very wrong.

Ann Althouse said...

But I'm no expert about that. I'm asking for someone who knows how to do that sort of probability analysis to weigh in. If the answer is that it is inapplicable here, please explain in a serious way. Don't just make wisecracks or flat statements that it's inapplicable.

Ann Althouse said...

I asked a question "Shouldn't we talk about Bayes theorem?"

Don't just say "no." Say why.

Paco Wové said...

Maybe NYC just has really bad governance.

Ann Althouse said...

Remember 3 factors were identified: "density, reliance on mass-transit, and the bad health conditions represented by the term 'poverty.'"

I intuitively feel that these are enough to explain the extreme effect in NYC, and I suspect that a scientific approach to statistics might support that. Is there anyone here who can do that kind of analysis?

Feel free to express your own intuitions, but that's not what I was hoping to hear.

dddave said...

In NY and NJ you must add in a massive fail in nursing home management in this crisis. Many of these deaths were clearly available.

I Have Misplaced My Pants said...

Cheryl-my county is reporting as a death a man who was from here but had been living in Nevada for some time, working on a long term contract, and who was infected there, was treated there, and died there. But somehow that’s a death in my county’s statistics. Was he included in Nevada’s numbers too? Does this happen often?

Relatedly, this is our “second covid death.” The “first” was an elderly woman who was in hospice and had been at death’s door for months.

gilbar said...

Remember the olden days?
Back when the "experts" told us that we'd ALL be getting rid of our cars?
and replacing them with mass transit, or a 'shared car', that you'd use WITH other people?

What ever happened to those "experts"?
Also, what ever happened to Vaxers?

Joe Biden, America's Putin said...

All failures by democratics will be gamed and minimized. (See Balfegor's post above for that theory)
All failures by Trump or any R, will be magnified and maximized. Manipulated and

Another example is the various meds that have so far worked in some patients. When first discovered -the media reported "yay!" then when Trump mentioned the very same meds, the media collectively said... "Boo! dangerous! scary! bad!"

Same with very name of the virus. Wuhan. Media were fine with it, until they could weaponize it and call Trump xenophobic.

High Density spread is a no-brainer. If not , why are we all required to social distance? I suppose at some point the hacks will find a way to twist that as well.

dwshelf said...

NYC operated a superspreader event every day, all day long, for a month.

The subway, late Feb to late March. Even after that, they kept it running, it was up to regular people to understand the certainty of exposure.

dddave said...

Avoidable, sorry

Laslo Spatula said...

"Remember 3 factors were identified: "density, reliance on mass-transit, and the bad health conditions represented by the term 'poverty..."

With NYC, you also have to take into consideration the hot dog carts.

I refuse to believe there ISN'T bat meat in some of those hot dogs.

I am Laslo.

Spiros said...

When you apply Bayes Theorem to real life, the results live or die on the quality of the models and the relevance of the evidence used to train them. Garbage in garbage out.

wendybar said...

There are pictures of filled subways in March. Standing room only. Geez.....I wonder how it spread so fast there???

Ray - SoCal said...

Homeless are using subways as hotels...

And turnstile jumpers are ignored

And I question how often the cars and stations are cleaned.

<a href="https://nypost.com/2020/04/25/passed-out-passengers-trash-fill-subway-cars-amid-coronavirus-crisis/ ”> Passed-out passengers, trash, urine fill subway cars amid coronavirus crisis </a>

wendybar said...

But get those Floridians off the beach!!!!

Ray - SoCal said...

Ignored so far is why are Blacks such a high amount of the deaths?

Joe Biden, America's Putin said...

Why are the subways running at all?
oh right - the elites in NYC need their toilets cleaned. I'm look at you DeCommie-Blabio.

Levi Starks said...

“What went wrong”
From what belief are you allowed to make the moral argument that people living in NYC didn’t get exactly what they deserved?
Death is certain. Clearly there are many factors including g mass transit and and an unwillingness to maintain your own health that make it more certain.

Mike (MJB Wolf) said...

Excluding hit spots like NYC we need to immediately:

- Open the schools and free parents to work, if healthy
- Adopt public health protocols such as temperature checks to enter crowded buildings or restaurants
- Keep promoting reasonable distancing, hand washing, taking extra care interacting with vulnerable people

Thirst steps would help mitigate the effects of the huge unemployment spike, assist in creating a more V- than U-shaped recovery. And help position us to better handle any second wave in the Fall, should COVID pull an H1N1-type trick on us.

gilbar said...

MikeR said...
Some NY hospitals were indeed overrun. I recall seeing an article about Maimonides.


You mean THIS ONE? The desperate struggle at one NYC hospital on the front lines of the coronavirus fight

They talk about IV bags in the hall (SOUNDS CROWDED!), then you read that The People are in their rooms, with tubes going out into the hall (so the nurse don't have to go in)

where 26 coronavirus patients were packed into the ICU on one particular day when the normal number is half that much
is it NORMALLY FULL? or is it NORMALLY less than half full?

The hospital normally has about 10% of its 711 beds devoted to critical care, but it’s added about 200 since the outbreak of the coronavirus — and some 250 beds are now occupied by the most worst-off patients.

SO...
250 beds, out of 711 are full; the other 461 beds ARE EMPTY? (the hospital was closed but for Covid)

You CERTAINLY read an article, that IMPLIED that NY hospitals were overrun
But NOT any articles about them ACTUALLY BEING OVER RUN

Danno said...

This is the WSJ article I linked to the other day when Jaltco was worrying about the protest in Madison and I said I would gladly walk through the open-air protest as opposed to being in NYC. And I didn't even delve into the details such as riding the NYC subway. Too funny.

Jeff said...

P(A|B) means the probability of A given B, i.e., the probability of A occurring given that B has occurred.

Bayes theorem: P(A|B) = P(B|A)*P(A)/P(B)

In words: the probability that A will occur given that B has occurred is equal to the probability that B will occur given that A has occurred times the probability that A occurs divided by the probability that B occurs.

I don't see the relevance to this discussion.

Levi Starks said...

Another theory I’m working on is rent controlled apartments which make it nearly impossible for for New Yorkers to live close to where they work, and force them to use mass transit.

NorthOfTheOneOhOne said...

MayBee said...

NY is dirty. Tokyo, London, Chicago, and LA are all cleaner. But I still think there is something about the infrastructure of NY that is helping spread it.

Spot on. Particularly when it comes to things like mass transit. The London Underground is clean and modern, not a graffiti covered relic like the NYC subway.

Oso Negro said...

Chemical Engineer here, trained in process analysis by statistical methods. Tutored in a statistics lab back in the day when students didn't have their own computers. My INTUITION, honed over a life time of watching the world, tells me there is no accounting for the POLITICAL factor in a prospective Bayesian analysis. If you are trying to understand a problem scientifically, you use the best factors you can. If you are trying to exploit a problem politically, you just make shit up.

Danno said...

Laslo said..."With NYC, you also have to take into consideration the hot dog carts. I refuse to believe there ISN'T bat meat in some of those hot dogs.

Does that account for why so many New Yorkers are bat-shit crazy?

tim maguire said...

How about the late arrival at the party by local political leaders? I’m betting the other factors would not have had such a drastic effect without de Blasio as a catalyst.

Matt Sablan said...

Maybe they should have banned more soda.

Tommy Duncan said...

Did the health officials learn something in the subways, trains and buses that moved us from "wash your hands" to "wear a mask"?

Jamie said...

"My theory about New York's high rate of infection is that using the word the "fuck" sprays more virus-laden droplets than most words."

Wouldn't it be fascinating if, in years to come, one factor in the R0 number turned out to be the language spoken? Surely different languages have different rates of fricatives and sibilants and so forth. And obviously different cultures have different "normal" speech volume and personal space habits.

When my husband and I lived in London years back, after college, he worked in a bookshop and I in a university department office. It took us probably the first month of work to get used to how quietly Londoners spoke - and how, when we couldn't hear or understand and said, "I beg your pardon?" they would repeat themselves even more quietly. (Doesn't apply in a crowded pub.) Then, when we returned to the States, we were overwhelmed by how loud everything was.

Milo Minderbinder said...

There's no consistency in measuring incidence and morbidity. If you're taking a statin, sneeze next to a window and fallout to your death, you've arguably died of COVID-19.

There are now economic incentives to report illness and death as caused by COVID-19.

Fernandinande said...

"density, reliance on mass-transit, and the bad health conditions represented by the term 'poverty.'"

Mass-transit is a subset of density, and failing to know any other indicators of general health, NYC, Italy and Los Angeles all have similar predicted life-spans (~82 to 83 years) and many of the poor in NYC and LA have longer than average life-spans because they're Mestizos.

mikee said...

First you needgood data. Deathbed recorded as actually or probably caused by Covid, versus simultaneous with but caused by other illness. And then factor in avoidable deaths such as the ones caused in nursing homes by placement of virus patients among other elderly, per city policy decision. After that, trace subway users, family coinfections, coworker infections, and other reasons for infection like the Health Services director encoueaging Chinese New Year participation.

New York's high count wasn't just one thing. It was everything.

Original Mike said...

"Remember 3 factors were identified: "density, reliance on mass-transit, and the bad health conditions represented by the term 'poverty.'"

I intuitively feel that these are enough to explain the extreme effect in NYC, and I suspect that a scientific approach to statistics might support approach to statistics might support that. Is there anyone here who can do that kind of analysis?"


With respect, you're asking for a hell of a lot. To start, you're going to have to quantity your three factors. Looks like doctoral thesis level stuff to me. And even then, I bet the end result would be mushy.

Michael said...

A genius move was to shrink the number of subway cars because there were fewer riders. Thus replicating full crowded cars. No theorem by any name would have predicted that what was wise in one scenario was catastrophic in another. Theorem would need to be replaced with common sense.

chuck said...

Italy’s population is very predominantly elderly. That accounts for the high death rates.

Yet the worst outbreaks occurred in a couple of cities. I think the relevant factors there were the hospitals getting overrun after the virus had spread widely without early response. Speculation is fun, but I don't think we have enough information to do a real analysis.

GingerBeer said...

NYC's MTA cut subway services in ways that increased exposure to the virus. The MTA moved to holiday schedules, cut express trains, and reduced the number of cars. All served to increase congestion on platforms and in cars, and the lengthen periods of exposure and very likely led to higher viral loads. The only winner in all of this is David Dinkins. There no longer is any doubt who the worst mayor of NYC is.

Jamie said...

"...the various meds that have so far worked in some patients. When first discovered -the media reported "yay!" then when Trump mentioned the very same meds, the media collectively said... "Boo! dangerous! scary! bad!""

I know this subject isn't germane to the thread, but again I say, in my perfect world Trump would stop publicly opining on non-policy matters because of this very phenomenon. It's not his fault that the media won't stop doing this - it's their fault, totally, and I wish all sorts of itchy rashes on them for doing it, since their scare-mongering has real effects - but he is the only one with the power to stop giving them ammo.

West Texas Intermediate Crude said...

The Prof is correct is stating that "a scientific approach to statistics" would be helpful in teasing out causes, treatments, and aggravating factors, but we screwed the pooch big time on this one. We could have plugged dozens or hundreds of attributes on thousands of patients into a database and let it tell us what combination of behaviors and treatments were effective or harmful. We could have learned something.
Now we'll never know, because doctors, hospitals and public health people were forced/encouraged/coerced to conflate patients dying with the virus with those dying because of the virus, and in some places if the clinical picture was compatible with Wuhan virus, it was called that, even though the vast majority of people thought to have the Wuhan virus by clinical criteria turned out to have a different virus.
Truly unfortunate waste of an opportunity.
We will not get another chance- nobody will pay attention next time we are told the sky is falling. When the final tally shows that, outside of NYC area, this virus was less dangerous than the yearly flu, we will not be willing to give up our freedom and our economic well being for what will come to be known as a hoax.
Even worse, the CCP is watching and learning. One day, we will have to deal with another of their "accidents."

Sebastian said...

"density, reliance on mass-transit, and the bad health conditions represented by the term "poverty""

Well, you could do a, you know, statistical analysis. Don't even to go all fancy-like Bayesian.

But even if you'd find any significant relationships, the causal argument would still be a little tricky, if you'd want to prove, say, that mass transit and density directly affected infection and complication rates for old, sick people in nursing homes.

I'd plug number of travelers from China and northern Italy into the equations as well.

Anyway, if you limit that analysis to U.S. cities, the one factor that cannot account for any variance is the only one progs are interested in, i.e., Trump. So, which "expert" is going to be eager to do it?

Jeff said...

Density, mass transit, and social distancing or the lack thereof are all plausible reasons why NYC might have a higher infection rate, but none of them seem like a good reason why the deaths per infection are so much higher in New York (IFR 0.66) than in California (IFR around 0.15 - 0.2 in both LA and Santa Clara) or in Florida (Miami-Dade IFR of 0.15). The Florida numbers are from random antibody tests run by Miami-Dade County reported Friday by the Miami Herald.

Back to Bayes:

If we say event A is dying and event B is having the infection, then the IFR is P(A|B). Then Bayes theorem says that the IFR equals (the probability of dying from anything divided by the probability of having Covid) times the probability of having Covid given that you died.

I still don't see how this helps.

Ann Althouse said...

"With respect, you're asking for a hell of a lot. To start, you're going to have to quantity your three factors. Looks like doctoral thesis level stuff to me."

People making decisions in real time and trying to shape policy have to come up with a good enough way to process the limited information they do have. The public health experts like those on the task force must have worked out something that they are able to do.

Here we are presented with the disaster in NYC like it's a big mystery. I'm asking if it's really all that mysterious. We know a lot already, but of course we don't know everything. You don't give up on science in real time because a slow version of it would be better.

Matt Sablan said...

"When the final tally shows that, outside of NYC area, this virus was less dangerous than the yearly flu, we will not be willing to give up our freedom and our economic well being for what will come to be known as a hoax."

-- I'm going to be more curious how next year's flu is going to be treated. I mean, "if it will save only one life."

Bruce Hayden said...

Anecdotal, but the kids who had been in HS with my kid, who went in to medical school are in their residencies now. The one doing their residency in NYC (ob/gyn) was abruptly reassigned to COVID-19 duty. Still, the NY politicians can’t complain too much about the aid the feds gave them - Trump sent the Comfort up there, and it has gone back home because it wasn’t being used. Hadn’t been, never at more than a fraction of its capacity.

One factor that you can also add is that the NY gov, and esp the NYC mayor were completely unprepared. At one point, De Blasio was inviting people to Chinatown for their lunar new year celebration, instead of stocking up in PPE, ventilators, etc. or more accurately, restocking from previous epidemics. I personally had my stock of PPE long before he bothered to order for NYC. Which means that health care workers, along with NYPD, NYFD, etc, spent critical early weeks unprotected. NYPD was apparently running over 10% out sick at one point, many with COVID-19.

Automatic_Wing said...

A genius move was to shrink the number of subway cars because there were fewer riders.

Didn't they have to reduce the numbers of trains because the drivers kept getting sick (and sometimes dying)? Self driving trains would have been helpful here.

buwaya said...

Manila beats NY and even Manhattan in population density - 119,000/sq mi vs 72,000
Manila proper may be the densest major city on earth at this point.

And its got most of the population using public transportation (a light rail system, buses and jitneys), only a small minority use private vehicles, and I can vouch for the extra intimate conditions in all the varieties of public transportation. Even New Yorkers on their subway have it vastly better.

And its got much worse pollution.

And its got vastly smaller scale of medical facilities - ICU beds, ventilators, etc.

And it is known to have been exposed direct from China at least a month earlier. It has also had very heavy travel from China, millions of Chinese visit annually.

But Manila has @400 dead instead of 20,000.

Matt Sablan said...

"Here we are presented with the disaster in NYC like it's a big mystery. I'm asking if it's really all that mysterious."

-- It is, and it isn't. We know that those three factors you listed are factors effecting New York. We know DC, for example, also has a lot of poverty, but it is no where near as compact. So, could we compare the two cities to figure out what impact each of the different factors has? Probably not, since there's a lot of other compounding factors, like DC was willing to slow down use of the Metro compared to New York, and a lot of the people who make daily commutes in DC are government workers who can work from home, so they have been.

So, in mystery writing terms, it is more of a "How Done It" than "Who Done It." We know what is responsible, we just aren't sure to the extent at which those factors matter, and if there are other, unknown factors further complicating it.

Lurker21 said...


Sure, it's those three things, but there are also contributing factors. Nursing homes, for one thing. Padded statistics, too. New York not only called a lot of deaths covid fatalities without testing, they also tested more people than other places.

Also, patterns of socialization. The British are less likely to go to church. That's one mass gathering that they tend to avoid. Funerals played a big role in spreading the virus in southwest Georgia, and synagogues in the NYC suburbs. Will we find that covid spread faster among religious congregations in the city?

Timing matters a lot. Some places had more time to prepare than New York. San Franciscans tended to stop going out to restaurants earlier than New Yorkers did. They took the virus more seriously in the early days. There's much data out there on the 1918 influenza epidemic and experts have been able to study how the measures different cities took affected mortality rates in those cities.

Josephbleau said...

If you have the dependent variable of total deaths, and predictor variables of density, co morbidity, mass transit as a percent of total transit, average temperature, percent plus 55 us in age, etc, you would start with multiple linear regression. In MLR you can use squared terms, or exponential, or trigonometric transforms of all these predictors to build your model. You can even use categorical variables like country. When this is run you check the significance of your predictors and then you know what contributes to variation in death rate. You don’t start with Bayesian regression, now known as Monte Carlo Markov Chain regression Unless you were trying to let prior beliefs influence your model. This is leading edge and you can’t always get a solution. Bayes law is a simple mathematical statement about conditional probability. You have to do much more to use Bayesian theory in modeling.

You would have to build a data set for this and you would really need more than 8 to 10 cities to do much good. I know that’s not the answer but I don’t have a data set now.

mccullough said...

How do you quantify that NYC is filled with assholes out for themselves?

Who is helping anyone there? Do neighbors even give a shit about each other?

buwaya said...

And as for poverty - New York has nothing to compare to Manila.
To call anyone in New York poor, in such a context, is just a bad joke.
Very few Americans understand what poverty is.

Matt Sablan said...

Also, out of curiosity, has New York provided "excess death" numbers yet? Or are these just replacement causes of death? Especially when we're focusing on the already ill; this is a very morbid topic, so expect very morbid questions like this.

If 1,000 people die of COVID-19, but your population was expecting a death toll of 2,000, and the other causes of death add up to about 1,000, then all that happened is what killed people who were likely to die changed. It's important to know that, because maybe we could save those people still, but knowing that the overall death rate didn't increase will be useful too. Likewise, if COVID kills 3,000 people, and your expected death toll is 2,000, that tells you something different.

LA_Bob said...

Jamie,

I recall the old saw that if the person you're addressing doesn't speak English, speaking louder doesn't help. I wonder if that's an American habit or a universal one.

I also understand the Brits like to say that Americans don't drive cars, they aim them.

A variety of subtle cultural differences might greatly undermine Bayesian analysis.

Carhopper said...

NYC has a lot more highrises. Elevators have been shown to be major virus spreaders.

Jamie said...

I wish I knew more about subway systems. Three (smaller, during a time of reduced demand, but still long enough to provide social distancing space) trains per line would be the way to minimize infections spread: each train is taken out of service temporarily for cleaning each time it reaches the end of its line, and a freshly cleaned train starts back the other way. But I have no idea whether there are enough cars, enough side rails, enough time, enough employees.

And then there's the problem that commuting riders will always have a favorite car, near the stairs/escalator or the opening to the next line they have to catch, and they'll stand on the platform where that car's for will open. So do you have "conductors" to manage the crowd and deny commuters entry to the most convenient cars once they're as full as permissible? Or ignore the platform problem but have doors that automatically latch when capacity is reached? Either way, the trains will be slower and people will be pissed. But maybe not infected.

Will Tokyo trains go back to employing those shovers we've all seen pictures of?

Fernandinande said...

Italy’s population is very predominantly elderly.

Italy's median age is 45.5; Germany's is 47.1; Japan 47.3; (they're all on the high end), NYC 35.8, LA 34.6.

The 2014 median age in New York City was 35.8 years, almost two years lower than the national median of 37.7 years (US = 38.1 in Wiki article)

The median age is < 16 years in Niger, Mali, Uganda, and Angola.

Matt Sablan said...

"But I have no idea whether there are enough cars, enough side rails, enough time, enough employees."

-- There are not, given how poor cleanliness is managed even when people aren't dying because of it.

Howard said...

Before you talk about applying Bayes theorem, you need to develop a conceptual model of viral fate and transport in different media and in different localities and environments.

Dilettantes always want to go for the big guns from the get-go to avoid doing the real work.

Jeff said...

What buwaya said. There has been some speculation that immunization with the BCG vaccine (required in the Philippines, rare in the U.S.) may lessen susceptibility to Covid. Or perhaps it's the higher temperatures and humidity. There's a lot we don't know. But the bottom line is that the incoming data from antibody testing is showing NYC as an outlier. Only there is the IFR substantially worse than the flu.

One possibility is that the NY antibody tests aren't very sensitive and are therefore underestimating the number of infections. Or that the people they sampled (grocery shoppers) are less likely to have been infected than other residents.

Moondawggie said...

Ann, employing a linear or logistic regression multivariate analysis would be more appropriate than using Bayes Theorem in this situation. Basically, regression analysis would help you determine which potential independent variables are actually predictive of outcome.

We use logistic regression a lot in oncology survival analysis to determine which of many treatment and patient factors (radiation dose, number of chemo cycles, stage of disease, age, etc.) significantly affect survival. You could do the same with independent variables such as population density, preexisting conditions, age, use of mass transit, etc., to predict Covid-19 survival.

buwaya said...

If you take into account all experience with this disease across the whole world, it is clear that there is no valid model possible using the limited set of factors mentioned. Among other things climate or weather seems to play a much more important part than crowding or public transportation or contagion models.

And we have no idea what possibly critical factors exist that we don't yet know about.

Fernandinande said...

But Manila has @400 dead instead of 20,000.

Philippines median age = 23.5

Howard said...

Anyways there's nothing to figure out here Asians use masks at a very high percentage because they're not embarrassed or afraid of what other people might think. Therefore ipsofacto ergo the main difference in infection rate is vanity. Put that in your prior and smoking

Balfegor said...

Re: Ann Althouse:

I asked a question "Shouldn't we talk about Bayes theorem?"

Don't just say "no." Say why.


This isn't Bayes Theorem as such, but before you can think about the probabilities you need to think about how the factors we're talking about would drive probabilities. Density, for example -- it doesn't really matter how "dense" the population grouping is if they all keep to themselves (the virus can't infect people through walls). What matters, as I understand it, is people gathering together in range for the virus to spread. And there, I think the density does matter more than one would think.

It's been a while since I've done any probabilities, but let me think out loud:

Suppose you have someone infected, and there's a 1% chance that they infect someone in every interaction. Let's say that 2% of people are infected to start with, and we have either 1 person in a crowd of 50 (scenario A), or 2 people in a crowd of 100 (scenario B).

In scenario A, each of the 49 other people have a 1% chance of getting infected, so the expected value ought to be 49 x 1% = 0.49 new infections (the expected value for each is 0.01, and there's 49 independent runs in this round). In scenario B, the chance of infection per person is higher because there are two potential infectious sources (99% x 99% = 98.01 chance of not getting infected = 1.99% chance of infection). Multiply by the 98 uninfected, and you get an expected value of 1.95 new infections, more than twice the number of new infections expected under scenario A. So A went from 2% infection to 2.98% infection, while B went from 2% infection to 3.95% infection.

But this isn't one and done -- it's an iterative process, and in each round, the newly infected people can infect the remainder:

Round 0: 2.00% (A) 2.00% (B)
Round 1: 2.98% (A) 3.95% (B)
Round 2: 4.42% (A) 7.69% (B)
Round 3: 6.52% (A) 14.55% (B)
Round 4: 9.54% (A) 26.18% (B)

Scenario B, with double density, is exploding a lot faster than Scenario A. If we run two Scenarios A, A1 and A2 both with 50 people (for a total of 100, matching Scenario B), the raw number of infected will increase, but the rate of increase in the % infected will be unchanged -- B will still be spreading a lot faster than A.

That's a super simplistic mental model, but the impact of actual local density -- literally, how many people are in a room, enclosed space, whatever, with an infectious person -- would seem to have a huge impact on how fast the virus spreads. I think it's less the overall density than how that manifests -- in packed bars, in packed subway cars, etc. And there, New York's buffoonish mayor did his best to give the virus one final push before the lockdown:

On Sunday, however, shortly before the governor announced that all bars and restaurants in the city would be restricted to takeout and delivery, de Blasio told New Yorkers in a press conference: “If you love your neighborhood bar, go there now.”

320Busdriver said...

After seeing Osterholm on Chuckturd I am worried. Not that I will catch and die, but that unless we can do epic level testing and isolation on an epic scale we will drown in virus and poverty.

Why are we not moving in this direction with gusto? Is Trump not being counseled that this is the way?

No herd immunity

No vaccine

No joy. For a. Long. Time.

I Have Misplaced My Pants said...

we will not be willing to give up our freedom and our economic well being for what will come to be known as a hoax

From your lips to God's ears, although I am doubtful. The email thread I'm seeing go around my church mailing list regarding when we meet in person again is running strongly "I'm not coming to church again until this thing is under control". Not just elderly church ladies; healthy young families. No one can say what that is, but evidently one death of an elderly woman already circling the drain, five people currently hospitalized and and a 0.000255% confirmed infection rate in our entire county of 360,000 people does not qualify as "under control."

I hope the people you know are not as fearful and as media-brainwashed that the sky is falling as the people I know. I pray that somewhere, adults are talking.

Howard said...

I agree with you A320busdriver. The response and reopening should all be about training the public in PPE and developing habits to prevent cross-contamination. We need to learn how to lengthen the viral load distance while shrinking the social distance.

buwaya said...

In such a situation of complete uncertainty one can only use intuition.

My intuition tells me that the healthiest place right now wrt coronavirus is probably the beach at Santa Monica.

Ralph L said...

If 90+% of the people on ventilators die, what difference does it make if there aren't enough of them?

NorthOfTheOneOhOne said...

Of course it could be that New Yorkers thought they were the shit and that they'd be left unscathed by the virus while Flyover country would end up a zombie apocalypse wasteland. If, and that's a big word for only two letters, that happened; maybe these New Yorkers didn't social distance like they should have, because they're the shit and nothing would happen to them.

Wouldn't be the first time New Yorkers found out they weren't strange visitors from an alien planet who are able to leap tall buildings with a single bound.

Balfegor said...

Re: buwaya:

But Manila has @400 dead instead of 20,000.

Google tells me that life expectancy in Manila is 69 years, while life expectancy in New York City is 81 years. Median age in the Philippines as whole is 23.7 years, while in New York State it is 35.8 years. 4.35% of the population of the Philippines is 65+. 12.9% of the population of New York State is 65+.

Zero evidence for this speculation, but I wonder whether in developing countries, the fraction of people who are most susceptible to this illness are basically all already dead of seasonal flu other other illnesses. We've seen that mortality for coronavirus rises sharply with age, such that something like 15% of people over 75 die (not sure of the current number, but it's high).

I guess one statistic that might shed light on this is what the mortality rate for elderly coronavirus patients in Manila is, vs. the mortality rate for elderly coronavirus patients in New York. Of course, in New York, those numbers may have been juiced a bit by the governor's order that nursing homes have to accept coronavirus patients. So it's never going to be apples to apples.

Howard said...

Balfegor: that's a nice conceptual model of the density factor. If walls mitigate the affects from population density than one could say masks do as well probably not as well that's a wall but nearly as good

Fernandinande said...

And we have no idea what possibly critical factors exist that we don't yet know about.

Also the methods and accuracy of the reporting; I'm tempted to put [sic] after all the numbers.

But just for edutainment purposes, Africa has about 1,300 deaths out of a population of about 1.2 billion.

Africa deaths ~ 1/92,000 people.
US deaths ~ 1/6,000 people.

It seems like a population's good health ~ long life-span ~ more people likely to die.

Mr. Forward said...

Italians and New Yorkers talk with their hands.

I Have Misplaced My Pants said...

-- I'm going to be more curious how next year's flu is going to be treated. I mean, "if it will save only one life."

Agree. I am not sure if I want the people who are irrationally terrified of covid -- and I say irrationally because they are not personally statistically in any kind of danger from it -- to realize how much more scared they should be of the flu. I mean, how can we justify sending children to school when, unlike covid, the flu actually kills children?

In late February my whole family took turns with a mysterious flu-like illness and I had to pick up my son from school when he developed a fever. His teacher said that of 21 kids in the class, seven were out with the flu. No one even considered closing public schools for that. How many kids have confirmed covid, again?

In order to continue on our society has to have some hard conversations about peaceful and resolute and above all rational coexistence with risk, and I'm not at all optimistic that we're capable of doing that.

Robert Cook said...

"Spot on. Particularly when it comes to things like mass transit. The London Underground is clean and modern, not a graffiti covered relic like the NYC subway."

The NYC subway--trains and system surfaces (walls, benches, wall maps, etc.--have not been graffiti-covered in many years. One will see occasional new graffiti, but it is quickly cleaned up. There may be parts of the vast NYC subway system that I don't see that have a greater incidence of new graffiti than the rare examples I see daily, weekly, monthly, but, as compared with 30 or 40 years ago, the problem system-wide is almost nil.

Jeff said...

The response and reopening should all be about training the public in PPE and developing habits to prevent cross-contamination.

No, the respone and reopening should be all about training the public not to believe the scare-mongers.

The lockdowns were supposed to be about flattening the curve so the medical systems would not be overwhelmed. The verdict is in, the medical systems are not overwhelmed, not even in NYC where the pandemic is almost over. And the incoming evidence from random antibody testing tells us that there was really never all that much to be afraid of, because Covid-19 is no more than twice as dangerous as the flu.

Jamie said...

Up-thread someone mentioned next year's flu season and the "if we can save one life" standard. Disaster for all if the lockdown side wins this argument; potentially life-saving if Howard's most recently articulated view,a few comments up from here, wins (getting all of us to employ better anti-disease habits personally). I'm pulling for Howard, and a permanently reduced body count from the flu et al.

SGT Ted said...

Does the article talk about how NYC policy that allowed infected people into nursing homes and killed a bunch of old people?

320Busdriver said...

I would not say the population of the US is of good health. Far from it.

wildswan said...

The only part of the Bayesian theorem explanation which I understood was where three factories turn out defective parts in different amounts. What are the chances that a random part came from factory A? The answer was reached by dividing the number of defective items produced by a given factory by the total number of defective items. Factory A produced 500 defectives out of 2400 so the chances are 5/24 that a random part comes from factory A.

So, you notice they have some hard figures to work with before they get to probability. But do we have any hard figures? I feel we don't know how many people got sick of it or died of it although in New York we do know that many more died of it there than anywhere else. Which is the question - why so many more.

But what about this: if you compared density/ deaths from flu with density / deaths from covid? The density is unchanging but there were 7x as many deaths from covid as from flu. And the other factors were also unchanging and also they multiplied in with each other, overall for the flu the same as for covid. So those three facts don't explain it when we look at NYC as a whole. So what we need is some more granular subset than "NYC density" where flu deaths were 7X higher. For example, were NYC commuters 7x more likely to get the flu? Westchester and New Jersey which have many commuters have excess deaths as well as NYC. Were NYC commuters on some lines or at some times 7X more likely to get the flu. Morning rush hour commuters? Bankers' hour commuters?. Rail commuters? If you could find a flu pattern or several patterns where people were 7x more likely to die or possibly just 7x more likely to get the flu, and compared it with the pattern of corona deaths then you might see how there was an excess of infection in the case of flu in some pattern. Then maybe you'd see that this pattern of excess infection corresponds to the pattern of excess deaths in the case of corona. (Medicine didn't know how to treat corona so excess infection became excess death.) Maybe poor people are 3x more likely to go to work no matter how they feel and 2x more likely to travel by subway and 4x more likely to go to a doctor late and you could see that with flu. And maybe somehow that adds up to 7X in Bayesian theory. Or maybe it's age, obesity and smoking - how did they affect flu deaths?

The comparison of flu and covid deaths is at
https://www.realclearpolitics.com/?state=nwa

rp said...

I find it astounding that no one (eg, a "journalist") seems to note the ubiquity of "street drug" use in NYC. I've worked off and on throughout the city -- and I've been startled by seeing little old ladies buy both heroin and antibiotics on the street corner.
Rent control indeed has created a problem -- as multiple generations cram into "grandma's" rent-controlled apartment -- which allows "grandma's" sons and grandsons to live rent-free and to spend their days on drugs. In addition to about 350,000 subsidized apartments (rents as low as $8/mon in "the projects"), there are about 70,000 shelter beds for the homeless (with many more preferring to live on the streets).
I am waiting for the first article that mentions that drug addicts have been hit harder by the virus than many others.

Howard said...

Graffiti has nothing to do with viral load or transmission. This is just a throwback to the racist theory of disease. Also it is very interesting to note that the deplorables depiction of how New Yorkers live like rats is very similar to the Nazi description of how Jews lived in Germany and Poland.

These are the fans you have cultivated Althouse, how does that make you feel? Is that what you wanted for your art project? getting them to come out of the closet and feel comfortable about expressing their true feelings without thought without emotion without judgement. Personally I think it's been a resounding success

buwaya said...

That median age difference may be another factor, so far also unmentioned.
Perhaps there is a more complex effect involved as well.

It would explain, on the face of it, mortality differences if the difference was proportional to the age distribution. But in the national cases I have compared it doesn't. Mortality is even lower than can be explained by simple demographics.

It may be that youth also reduces the risk of transmission.

But we just dont know.

Balfegor said...

Re: NorthoftheOneOhOne:

Of course it could be that New Yorkers thought they were the shit and that they'd be left unscathed by the virus while Flyover country would end up a zombie apocalypse wasteland. If, and that's a big word for only two letters, that happened; maybe these New Yorkers didn't social distance like they should have, because they're the shit and nothing would happen to them.

More concretely, we already know that there are people with confirmed infections, who ought to have been self-quarantining, but deliberately flouted those restrictions. I'm speaking, of course, of noted assholes Chris Cuomo and George Stephanopoulos. And we only know about them because they're famous. I am certain there are many other New Yorkers who have similarly flouted quarantine. If you're trying to get control of the infection, this kind of bullshit is a real problem. In Seoul, when you're under self-isolation either because you are infected or because you were in close contact with someone who was infected, you "voluntarily" install tracking software in your phone, so it will ping your government minder if you ever leave your quarantine zone (or disable location services). They aren't working on the honour system.

Temujin said...

All of the trio of lifestyle that is indigenous to New York City- density of population, poverty, pubic mass transportation play a part in this. So does the fact that it's just not that clean a city. You can smell it, see it in many places. It's hard to keep a city that large, that busy- clean. Giuliani made a very good dent in it, but DeBlasio completely tore that down. Completely.

He also paraded around in Chinatown telling people to come out and enjoy things, and was late on the draw to realize what had to be done in a city like New York. One could look at his entire time in office and, without reservation note that he was the wrong man for the job from Day 1. He has to be listed as a factor. Not a general factor, like some politicians did not take action but as a direct problem. Also the Governor. And the Governor's staff. The geniuses behind sending Covid patients into nursing homes in New York, and not allowing nursing homes in that state to remove those who were infected. Hence, the highest mortality rate in the nation in senior homes.

New York has some real issues. Its simply how the city is made up. Those that live there will need to choose their leaders with more thought about how they will deal realistcally with the inate problems of that city. DeBlasio brought the city to it's knees before Covid even came onshore. Cuomo was busy with his Albany Corruption Company. There are things to work on there. I would add those two to the splendid mix of a perfect storm that hit New York. Of course, I don't expect the NY Times to agree with this view. They are another problem.

320Busdriver said...

Posted this last night.

He gets it. His credentials are terrific.

https://paulromer.net/covid-sim-part1/

Ken B said...

Ann
Bayes theorem relates conditional probabilities. It is frequently used to incorporate new data into a model: Bayes's Rule uses the theorem in a particular way to do this. The nub of it is this. You have a belief about how probable something is. You see new evidence. You use Bayes's Rule to modify your belief about that original probability.

So your fingering of Bayes' here is too vague to be useful. If however you had a specific theory about for instance density and the rate of disease spread you might be able to incorporate this new data.

Bayes's Theorem is uncontroversial. Bayes's Rule is a lot trickier, and is not universally admired by statisticians, but is used in most machine learning.

Michael said...

Lawyers say that the side that has to explain Bayes' Theorem in a trial - loses.

Robert Cook said...

"Of course it could be that New Yorkers thought they were the shit and that they'd be left unscathed by the virus while Flyover country would end up a zombie apocalypse wasteland. If, and that's a big word for only two letters, that happened; maybe these New Yorkers didn't social distance like they should have, because they're the shit and nothing would happen to them."

Spoken by someone who apparently hasn't been in NYC in the past few months. Social distancing and the diminution of people on the streets and the transit system happened pretty rapidly in NYC.

Fernandinande said...

Zero evidence for this speculation, but I wonder whether in developing countries, the fraction of people who are most susceptible to this illness are basically all already dead of seasonal flu other other illnesses.

A paper by the Brit gov't (from when they changed their official strategy - can't find it again) said exactly that re. Africa, along with graphs of population vs age.

William said...

I'm no statistician. I actively avoid the study of numbers whenever possible, but I occasionally ponder how pc affects outcomes. The writer states "long untreated such as heart disease and diabetes have contributed significantly" to the higher death rate. The writer does not mention that the leading co-morbidity is obesity. Obesity might be a disease of the lower quintiles, but it is not a disease of poverty...I'm not knocking people in the lower quintiles. They might have greater reason for poor impulse control than more affluent people, but overeating is not caused by poverty. Perhaps black people, living as a minority, have more stress than white people, but white people don't seem to have dodged the obesity pandemic either. Some thing are racial without being racist.

Yancey Ward said...

"Does that account for why so many New Yorkers are bat-shit crazy?"

No, that is caused by the bat shit in the hot dogs, not the bat meat.

TRISTRAM said...

If poverty, density and insufficient medical facilities were an issue, why aren’t Brazil’s favelas, and similar communities through the 3rd world having mass deaths, or homeless communities for that matter?

Anonymous said...

Yet, it was the self-appointed and self-anointed intellectual superiors of the NYC-based news media who condemned Jacksonville, FL, for re-opening it’s beaches, while super spreader NYC never even closed down its subways.

buwaya said...

Howard, thats stupid and opposite to the facts.
Across the world it is the untermensch that resist this virus better.

The President of Mexico has been the first and only major leader to note that this thing is on the whole a rich persons disease. So far anyway.

Ken B said...

Ann
If you play Bridge and understand the Principle Of Restricted Choice, that is an application of Bayes.

Mike (MJB Wolf) said...

As others have noted there are more than three variables premised by Althouse, including travel from hot zones, the cold spell in early March combined with an early Spring feel in February, hygiene (especially hand-washing), and a host of demographic factors that can only be properly analyzed once this is over and antibody surveys are complete. Did California enjoy some partial herd immunity that made our cities far less virulent than NYC? We don’t know. We don’t know a lot.

We do know most people can return to normal outside NYC.

Balfegor said...

Re: buwaya:

My speculation (again, zero evidence) is that in every age cohort there are people who are more and less susceptible to infection, perhaps because there are people with weaker and stronger immune systems, and if your immune system is strong enough, you just fight the virus off and rather than having 5 weeks to spread, it has only 5 days.

Younger people in general have stronger immune systems than older people, so in NYC, it might be that for people aged <30 yrs, 15% are susceptible, and for people 30-65, 25% are susceptible, and for people 65+, 35% are susceptible. And that in developing countries, those percentages are all lower, e.g. <30 yrs 12% are susceptible, and 30-65, 15% are susceptible, and 65+ 20% are susceptible. Because the missing people (cumulative from one age cohort to the next) already died of other illnesses.

Again, all just speculation, but it's not something where you could just rebalance the ages to see. That's why I'm curious to know whether age-wise mortality of people confirmed to be infected is lower in the Philippines. If so, that would suggest that even when people are infected, the people who are getting infected are less "vulnerable" in some sense, possibly because the most vulnerable fraction already died in the past.

I Have Misplaced My Pants said...

No, the respone and reopening should be all about training the public not to believe the scare-mongers.

I agree completely and have been saying this for weeks, that our leaders must have an honest conversation with us, and the media must report on it faithfully, about what the real risks are here (not covid - for most of us it is irrelevant while other related risks are certainly not), and to train their minds to listen to what is true and not to what gets them all excited and drama'd-up. But, as the saying goes, a person is smart, but the public is dumb panicky herd animals.

Fernandinande said...

He gets it. His credentials are terrific.
https://paulromer.net/covid-sim-part1/


He says: "Purple squares, which were infectious before but have recovered and now can neither catch nor transmit the virus."

But "WHO says no evidence shows that having coronavirus prevents a second infection"

Everybody is guessing.

buwaya said...

Every last thing people are knocking New York for are much worse elsewhere.
Not least among them is dirt.
And usually the whole list of suggested factors and deficiencies.

jaydub said...

Spiros: "Data is still coming in, but initial reports are that 95% of the total COVID-19 deaths in Europe are men older than 60. In the U.S., fatality rates are much, much more evenly spread for different age groups. Many more young Americans are dying from the disease. Young Americans, especially young African-Americans and Latinos, are more likely to suffer from asthma, diabetes and hypertension than their European and Asian counterparts. Americans are unhealthy. Bad news."

Complete and utter BS. Here's the spread for NC:
- Two percent of the cases and 0% of the deaths are in school age children.
- Seven percent of the cases and 0% of the deaths are in young adults aged 18 – 24.
- Forty percent of the cases and 4% of the deaths are adults 25 – 49.
- Twenty-seven percent of the cases and 11% of the deaths in ages 50 – 64.
- Twenty – five percent of the cases and 85% of the deaths in ages 65 and older.
- Cases (deaths) by ethnicity or race: White 54%(61%), Black 39%(35%), Hispanic 14% (2%)
- The ten Appalachian counties, which are the poorest in the state, have the fewest cases and fewest deaths, four of those counties have no cases.

Why do you always post such garbage and always completely devoid of any supporting documentation? You're nothing but a propagandist.

chuck said...

Unless you were trying to let prior beliefs influence your model.

I believe the IHME models are basically curve fits (erf and logistic) using three parameters with prior estimates and correlations between the errors. Those curves are the epidemiology equivalents of fitting a straight line.

One way of thinking of Bayes is that you start with a big, hopefully exhaustive, collection of possibilities, each with an assigned probability (prior). Given a measurement, one then uses Bayes to update the probabilities. It is important for this to work that the initial collection be complete and that the probability distribution of the measurement can be computed from the prior probabilities of the possibilities. After the measurement is made, Bayes is used to update prior probabilities to new values ready for the next measurement input.

William said...

Can anyone see how a day full of safe social distancing can be negated by a half hour ride on a packed subway train?...In the world of the pc, a packed subway car is less risky than waiting in line to vote....Side thought: the MTA workers seem to have a high infection rate. Wouldn't it be a good idea to furlough those workers who are obese, diabetic, or hypertensive....I think that NYC has significantly mismanaged this crisis in many ways,but, because of pc, the blunders are under-reported....In places where the epidemic is contained, it is because of the wisdom of the Democratic leaders. In places where the epidemic is rampant, it is because of the deficits of the Federal Government and the ineptitude of Trump. There's an Acosta corollary to Bayes Theorem. The chances of a disastrous outcome in any given situation are directly proportional to Trump's involvement in that situation.

Howard said...

A320 bus driver that model is kind of interesting. It reminds me a little bit of a random walk groundwater contaminant plume model we used back in the day it was called plasm

Josephbleau said...

I would also comment that there are mathematical models like the SIR epidemiological model that start with a differential equation based on a theory that at day one there is one infected and on day two she has infected 3 more and so on like rabbits breeding. These are always wrong because they don't include the reality that sometimes a rabbit has a headache and does not want to breed. They are wrong but useful.

What we talk about in statistics are models with random effects and we use actual data to observe things like, of all the cities that we have observed the death rate increases by 3.4 times the density, or 2.8 times the cube of the density. Then we check to see if this relationship is due to random chance or if it is so different from 0 that we can claim it is true. These models are wrong but useful, because data is imperfect.

What you are asking is, do deaths in cities increase in proportion to their measured density? and a linear regression can answer that. A Bayesian modification would be the prior belief that I think that the death rate only goes up with density, never down. So I would use a prior distribution for my density coefficient that is not negative, for example.

wildswan said...

Or the other possibility is that covid is super-infectious so that everyone gets it but only the same percentage (.01 %) or less die from it as die from flu. But, since about five times as many get covid and all at the same time, you have five times as many deaths in absolute numbers as from flu and these deaths are occurring more or less all at once. Then people don't get treated because the system is overwhelmed so it looks as though people are dying because the system is overwhelmed (Italy). But Trump got hospitals and workers and ventilators into New York city so very few died because the system was overwhelmed. In this scenario, they died because 8 or 10 million people were infected and out of that vast pool of disease, 21,000 had severe cases. Naturally modern medicine should find out what makes covid severe for some people.

jaydub said...

NYC/NJ area TSA agents have a very high incidence of Wuflu, presumably because of the exceptionally high volume of arriving international flights. So, they likely got the virus from folks arriving from such hot spots as Madrid and Milan and passed it on to many of the other travelers who passed through the airports. That was probably a major factor until Trump cut off the air travel from China and Europe.

Mike (MJB Wolf) said...

Anyone still relying on WHO data is not paying attention.

Mike (MJB Wolf) said...

Update: Anyone still relying on WHO opinions is not paying attention.

daskol said...

I think Jon nailed it: not Bayesian probability, but simply multiplication vs. addition. Take the three factors and ADD them together, assuming a linear relationship, and it's baffling what happened in NYC. Multiply them against one another, use some some function so that each is a multiplier of the others, and you can easily get there: use Bayesian methods to set the initial weightings for each of the factors, and then update their weightings with new information until you get a model that predicts what we're actually seeing.

In the same way, masks are multiplicative and not liner. If one mask if 50% effective at reducing a person's spread, and two people are wearing masks, you would see a benefit much greater than 50% reduction. Add in another person or two, and the multiplicative benefits of masks in reducing the spread approach 1 or perfection using pretty simple math. That may be particularly true in NYC on the subway. And you can make the math much more complicated and the model more predictive by considering that even if masks are less than totally effective at stopping spread, reducing viral output from a person by 50% may reduce chances of infecting others by 90% since it takes a certain amount of virus to infect someone else (and more virus seems to lead to a worse infection).

We really should have started wearing masks at the beginning, and there's probably not a great deal of the NYC puzzle yet missing. Morbidities of poverty probably explains the rest.

daskol said...
This comment has been removed by the author.
MarkCh said...

Are New Yorkers louder than Londoners? https://twitter.com/asymmetricinfo/status/1253799883598856199

Yancey Ward said...

chuck gives the best fundamental explanation of Bayes- it all comes down to updating your assigned probabilities (priors) when new information becomes available, and there is still no better tool for developing an understanding of this than the Monty Hall problem. If you can wrap your mind around the Monty Hall problem, then you have a grasp of Bayes' Theorem- at least enough to start seeing how it can be used in different situations.

Balfegor said...

Re: Robert Cook:

Spoken by someone who apparently hasn't been in NYC in the past few months. Social distancing and the diminution of people on the streets and the transit system happened pretty rapidly in NYC.

Uh, no. It didn't. It really didn't. Or rather it happened fairly quickly, but came way too late. The stay at home order didn't start under March 22, when they already had 15,000 cases. It's true that people began to pull back on outings earlier regardless of what their doltish mayor said -- looks like they started to pull back between March 7 and 14 or so. Manhattan starts on the earlier side (and I think has not been quite as hard hit). But Bronx and Queens don't really start pulling back until ~March 14, when cases are over 600 and doubling every two days. Assuming a lag of ~10 days from infection => symptoms => detection, at the time the state caught those 600 cases, close to 20,000 cases may already have been baked in the cake, as it were. It was too slow.

buwaya said...

Balfegor,

You can easily do a ballpark on age structure of the Philippines or Mexico vs the US or Europe. The answer is that mortality that we have seen is disproportionately low even adjusting for age distribution.

For instance, 4.8% of the Philippine population is 60 years old or over.
13% of the US population is 65 years or over.

The difference in mortality is disproportionate even adjusted by age. Its not 4.8/13= 0.37 but more like 400/15,000= 0.027

Any other related explanations in this direction are hypothetical. There may be merit in them, but we don't know and wont know until long after this thing is over.

daskol said...

The downside of being the center of the globally connected world in our overoptimzied interconnected economy, like London is for all business too shady to take place in NYC.

Howard said...

Loud obnoxious spittle-flecked talking of a New Yorker is definitely one of the prime Bayesian priors

buwaya said...

I am very well informed on this, thanks to Skype -

Masks started being commonly used in Manila (well, the upscale parts of it anyway) about the last two weeks of February.

By early March of people out in public maybe 2/3 or more were voluntarily using masks. Most were locally made cloth masks that started appearing in clothing stalls and public markets, and quickly became hot sellers. After all anyone with a sewing machine can crank them out rapidly.

In Spain hardly anyone used or even had masks until the lockdown order came.

Jersey Fled said...

I was in the NYC area several times in the early stages of the pandemic and was amazed at how nonchalant people were about it. Few masks and gloves, no social distancing,etc. People where i live in the Philadelphia area were wearing full body condoms.

Maybe it has something to do with the NYC psyche.

That and they have the dumbest mayor in world history.

Balfegor said...

RE: buwaya:

You can easily do a ballpark on age structure of the Philippines or Mexico vs the US or Europe. The answer is that mortality that we have seen is disproportionately low even adjusting for age distribution.

For instance, 4.8% of the Philippine population is 60 years old or over.
13% of the US population is 65 years or over.

The difference in mortality is disproportionate even adjusted by age. Its not 4.8/13= 0.37 but more like 400/15,000= 0.027

Any other related explanations in this direction are hypothetical. There may be merit in them, but we don't know and wont know until long after this thing is over.


I agree -- but the comparison you're making is not the one I'm suggesting. I'm suggesting that, the 65+ age cohort in the Philippines may be smaller proportionately than the 65+ age cohort in New York in part because a bunch of people who would have been part of that age cohort in the Philippines died when they were 20, or 30, or 40, or 50. And those people who died include substantially all the people who would have been susceptible to coronavirus.

As you say, it's hypothetical, and we don't have the data. But if 15% or New York coronavirus patients who are 65+ die, and only 2% of Manila coronavirus patients who are 65+ die, that would suggest that the Manila patients are, on average hardier than the New York patients, which would be suggestive. (Or it could suggest that Cuomo's nursing homes order caused the deaths of hundreds.) Simply adjusting the relative figures based on the demographics of the population tells us absolutely nothing about whether this hypothesis is true or not.

Drago said...

SlickWillie Clinton: "Yet, it was the self-appointed and self-anointed intellectual superiors of the NYC-based news media who condemned Jacksonville, FL, for re-opening it’s beaches, while super spreader NYC never even closed down its subways."

Bears repeating.

To date, not a single alarmist at Althouse blog has offered any criticism of NYC for keeping subways and Central Park open.

Meanwhile, ever single alarmist at Althouse blog had criticized every single republican governor who has opened any public space as well as any protest group going out in public.

Looks like having a "D" after your name confers quite a bit of criticism protection from the alarmists.

Balfegor said...

Re: buwaya:

Or if the Philippines adopted widespread mask usage in mid-February, that might be sufficient, frankly. I think masks have had a huge impact on why East Asian rates of spread, even among elderly populations like in Japan, were so much lower than in the US and Western Europe pre-lockdown.

Drago said...

Buwaya: "Or it could suggest that Cuomo's nursing homes order caused the deaths of hundreds."

Careful. The Althouse alarmists do not appreciate anyone mentioning Cuomo's very specific directive mandating the shoving of elderly infected back into the nursing homes.

They do not like that one bit. You see, the OrangeManBad Quotient of that story is far too small.

Howard said...

Shut the fuck up Drago, the men are talking.

Drago said...

Howard: "Shut the fuck up Drago, the men are talking."

See Buwaya?

These simple facts have, once again, triggered Howard.

Drago said...

Howard could just as easily written Yeah, Cuomo made a mistake on that one.

But after 4+ years of collusion lies, he just cant bring himself to do it.

buwaya said...

Balfegor, indeed we are ignorant.

In the Philippines and Mexico it seems that the case does not seem to be a difference in the mortality rate of those detected and treated, but the rate of serious illness in the first place. Of course we have no idea of the number of vulnerable people exposed and we wont, probably not for years.

Another hypothesis is that younger people are less effective carriers of the virus.

And so forth.

n.n said...

this thing is on the whole a rich persons disease

Very much like polio, where the risk was exacerbated in overly sanitized environments and underdeveloped immune systems in older (i.e. not young) people. Today, a likely transmission path, and cross- contamination (a la medical facilities and nursing homes) in these quarters are the laborers in their service. Spreaders of social contagion forced an acute response and resource mismanagement causing excess deaths. Suddenly, the lessons of past, annual contagion were forgotten.

Howard said...

Drago drives the bus looking through the rearview mirror.

n.n said...

Diversity, including color, sex, age, class? Immigration reform?

Drago said...

Some weeks back you had MIT scientists in analyzing waste in the Boston area which demonstrated ChiCom Flu exposure was far greater than the alarmists were claiming.

Howard made sure to reject that science pretty darn quickly.

Meanwhile we now supposedly have proof the earliest deaths from ChiCom Wuhan Lab Bat Flu occurred in very early Feb, which tells us alot about how far this has already spread.

Drago said...

Howard: "Drago drives the bus looking through the rearview mirror."

What Howard really means: please stop talking about all the things Howard got wrong in the recent past so his newest pronouncements wont look dumb.

Drago said...

Howard: "Drago drives the bus looking through the rearview mirror."

Note: Howard writes that at precisely the moment his dems have already established Sham-peachment III committees to frame Trump for the ChiCom/WHO flu.

buwaya said...

Thanks narciso.
A more complete indicator would be total mortality.

The effect of coronavirus/covid on overall mortality is clear in places that collect timely (almost real time) data, such as Spain.

The US data collection of mortality seems to lag several weeks, I suspect because it has to come from many sources with different reporting periods.

Fernandinande said...

The President of Mexico has been the first and only major leader to note that this thing is on the whole a rich persons disease.

Sailer mentioned that quite a while ago.

mandrewa said...

I'm sure it's already been said above, but I'll toss this off anyway. There is an implicit fourth factor other than density, poverty, and the subway system that the author of those three factors surely must be aware of, ie. climate.

Because if we don't assume climate is the first and most important factor in explaining where this epidemic hits then there are literally thousands of cities around the world that have greater density, far worse poverty, and at least a comparably crowded mass transportation system as that of New York.

In fact, climate is such an obvious factor in all of this that every comparison of one city to another should surely adjust for that somehow. And if we don't adjust for climate in our comparisons then it's all kind of silly.

So how do we adjust for climate?

Well there is more than one way of course.

But what I would advocate doing as of right now is to start off by assuming that climate is the only factor that matters and then collect and compare coronavirus deaths per million for different cities and calculate what numbers (adjustment factor) would be consistent with that assumption. The numbers would improve as we get more data but surely there is enough already to start doing it.

That would be the first order approximation.

Second order would be where you apply this first order adjustment to the data and then look at the new discrepancies and postulate what the factor is the second most important is explaining differences in deaths per million. And for that produce a new set of numbers (adjustment factors) depending on the assumption that both the climate and whatever factor is postulated as the two most important variables in determining what happens.

I can already guess that the second factor will be the percentage of the population that uses mass transportation each week. But of course others will disagree and that means the math gets more complicated for second order calculations as the numbers (adjustment factors) will be different for different narratives.

It goes on from there, from second to third, and from third to fourth. But most likely over 95% of the story will have been explained by the time we hit the fourth factor.

Openidname said...

"Fernandistein said...

"Africa deaths ~ 1/92,000 people. US deaths ~ 1/6,000 people."

You're comparing apples to oranges. It started much later in Africa. The relevant comparison would be, say, deaths at 30 days after the first 100 people infected.

And even then, if the testing criteria are different, the comparison is meaningless.

Openidname said...

"Fernandistein said...

"But 'WHO says no evidence shows that having coronavirus prevents a second infection.'"

Everything the WHO has said has been wrong. Everything.

To the point where, if the WHO says there's no evidence that coronavirus confers immunity, I start working off of the assumption that it does.

Unknown said...

Occam's Razor answer:

Because New Yorkers are weak.

buwaya said...

It may not be just a fourth factor, but a hundred factors.

daskol said...

There is a lot of indirect evidence that previous infection confers immunity, based on what we know about pretty much all other viruses. We haven't yet proven this to be the case for this virus, but it's a strong assumption.

Spiros said...

Jaydub, 60% of Americans are obese and over a tenth of Americans suffer from diabetes:

https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(14)60460-8/fulltext

The "inferior health status" of Americans is well documented:

https://www.ncbi.nlm.nih.gov/books/NBK154469/

Young Asians and Europeans are not like this. Coronavirus is going to impact our youth much harder.

mandrewa said...

Or perhaps percent of the population that is elderly would be the second most factor.

Then to explain deaths per million we have, in order of importance: 1) climate; 2) age; and 3) mass transportation.

Or maybe government policy also should be in there. If lockdowns made any difference surely it is going to show up in the math. Of course by this point I think the evidence is strong that lockdowns, unless draconian, as in China, have little impact.

And probably the time since the first person in a region was discovered to have the virus is another variable that should be on the list. Although as time passes this will become less and less significant.

In any case we should let the math sort out which factors are most important. To test whether a factor is important or not, someone has to imagine that it might be. But after something has been proposed, then surely it should be possible to test that hypothesis against the data (assuming one knows how to do the statistical analysis and that one is ethical and honest.)

buwaya said...

All these in a complex web of relationships and feedbacks.

Throwing a joker in here - Poland, say.

Its climate, in winter- spring, is awful by my standards. But next to no covid.
And its got plenty of exposure to an infected Western Europe.
There's no obvious reason why it isn't as bad as Belgium.

iowan2 said...


AA said
People making decisions in real time and trying to shape policy have to come up with a good enough way to process the limited information they do have. The public health experts like those on the task force must have worked out something that they are able to do.

The medical experts decisions are all ass covering. Plausible deniability. Their assumptions always veered to the "perception" of the safest protocols. Social distancing sounds good, but not supported by any replicated tests. Masks sound good, but CDC studies showed no protection was afforded the wearer. Locking down schools sounds logical, until you realize, death is linked to senior population, not schools.
The list goes on and on.

daskol said...

How much testing they doing in Poland? We're still a bunch of drunks searching for keys under the streetlamps.

buwaya said...

But the American places with the fattest youth are not necessarily the worst off.
It seems like on the whole the opposite is true.

buwaya said...

Death is the best metric.
No need for tests, just mortality.

Unknown said...

NYT - ‘Playing Russian Roulette’: Nursing Homes Told to Take the Infected

At the epicenter of the outbreak, New York issued a strict new rule last month: Nursing homes must readmit residents sent to hospitals with the coronavirus and accept new patients as long as they are deemed “medically stable.”

On Thursday, Mr. Cuomo reiterated that nursing homes had to accept the patients — but only, he clarified, if they could do so safely.

buwaya said...

Best answer is some polity just needs to bite the bullet and end lockdown.
If nothing terrible happens the rest will follow in a rush.
I suspect the rush will follow quickly after the first few, even without waiting for results.

Wince said...
This comment has been removed by the author.
Wince said...

Shouldn't we talk about Bayes theorem?

One of these days, Althouse, one of these days... POW!

Bayes at the Moon!

daskol said...

Governor Kemp of GA is pushing us in that direction.

I don't know if it's because of the order to take in infected patients, but the nursing home a few blocks from my house, the one my kids visit quarterly with their school to read to old folks, is devastated: as of today 60 of 300 patients dead of COVID-19. It's been in the news as one of the hardest hit places fatality rate wise.

chickelit said...

If climate is a fourth factor and stipulating that winter is a higher risk climate, shouldn't we see COVID deaths ramp up in the Southern Hemisphere in the coming months?

chuck said...

UV? Ban ozone, not Freon.

Yancey Ward said...

"shouldn't we see COVID deaths ramp up in the Southern Hemisphere in the coming months?"

The Southern Hemisphere's winter is less severe than that of the Northern Hemisphere- there is little land mass further south than 37 degrees south latitude (the equivalent to Oak Ridge, TN in the north) that isn't part of Antarctica. For example- in Melbourne, Australia- on the southern coast of the continent, is at 37 degrees south latitude, and the average high in June and July, the Winter, is about 59 degrees F- much warner than here in Oak Ridge in December and January. Buenos Aires is similar to Melbourne, and Winter in southern Africa is a kind of misnomer in the way we would understand the word.

Tomcc said...

buwaya: I've been focused on the information coming out for the US and haven't really considered the ROW. Manilla is an interesting example. I'd like to know your thoughts relative to poverty and obesity in the Philippines. I traveled in Eastern Europe (Romania) twenty years ago. A very poor country by US standards. I saw very few overweight folks. In the US, my own observations are that many poor folks are overweight to the point of obesity. Hell, it may even be a predictor of obesity at this point.
It seems very clear to me that the vast majority of the US has "bent the curve" and we need to re-open quickly and as safely as possible. Stop making respirators and start making N95 masks and gloves!

Ingachuck'stoothlessARM said...

Italy's hospitals were overrun in 2017

...they typically also have higher than normal respiratory cases

Jim at said...

I'm your density. - George McFly

Sebastian said...

iowan2: ""AA said: People making decisions in real time and trying to shape policy have to come up with a good enough way to process the limited information they do have. The public health experts like those on the task force must have worked out something that they are able to do." // The medical experts decisions are all ass covering. Plausible deniability. Their assumptions always veered to the "perception" of the safest protocols. Social distancing sounds good, but not supported by any replicated tests. Masks sound good, but CDC studies showed no protection was afforded the wearer. Locking down schools sounds logical, until you realize, death is linked to senior population, not schools."

All correct. But Althouse's "reasoning" also indicates one of the basic problems with public policy. She is smart. She has a skeptical turn of mind. She doesn't normally take things on authority. Yet here we have rationalizations of obvious "astronomical errors"--obviously absurd at the time, on which skeptics called BS at the time--as "good enough." Experts "must have worked out something"--without any consideration of whether those experts made good use of their vaunted "expertise," had any expertise in going relevant cost-benefit analysis, or were swayed by political motives. When even Althouse thinks this way, what's a sane politician to do?

I could have imagined another Althouse treatment of the entire WuFlu fiasco--the fearless Althouse calling BS on every single outrageous "projection," every single false invocation of false "expert authority," and every bad-faith political use of already bad "models." But no. The absence is telling.

mandrewa said...

My last comment was three hours ago. Obviously I'm working on something else and I come here as a break from that something else, which is usually my situation. But while I was doing other things I was thinking about this.

I should preface this by stating I'm not an authority. All of those people that normally reason by authority should ignore me. Now I don't normally say that, thinking that it should be obvious, but once in a while I feel like I should make that clear.

But my preferred approach for most things is to try to engage with the subject, whatever it is, and use my mind and try to make sense of it. I know I will make mistakes, indeed I will make many mistakes. But it is has been my experience that the more I make an effort the more I start to understand a subject. And it's my opinion that it is only by people, in general, individually making such efforts that we actually get anywhere.

Bayesian statistics has long been on my list of things to study. But somehow I have never gotten around to it. And at this point recognizing my current limitations I probably never will. But part of the reason for my interest in the subject is a feeling, an intuition, that there is something wrong with statistics. Or perhaps with the way people typically use it.

But in that same direction, I was thinking more about what I suggested above about building a narrative and then testing it against the data and comparing it with other narratives. And thinking uncommon that approach actually is. Or at least how uncommon it is in the studies typically cited by the media.

In this situation we have what seems to be a complicated situation with many factors potentially affecting coronavirus mortality per million.

Now using statistics, there is no simple mechanical way we can get at the truth because we are always limited by the possibilities we do not imagine. But it is possible to compare the hypotheses we do come up with against each other.

Are we doing that? No.

Here's a simple example. Imagine you hypothesize that poverty is a significant factor in explaining coronavirus mortality per million. Many people think that is highly plausible and in fact there are probably many that think that has to be true.

And here's a way that statistics is used to promote misunderstanding. You do a study looking for a correlation between poverty and coronavirus mortality per million. You will, of course, find a correlation between poverty and coronavirus mortality per million.

Now what's wrong with that? Well part of the problem is that there are probably at least a thousand other single factors that correlate with coronavirus mortality per million. Oh and by the way, people are ignoring almost all of them. Because as it happens most of these other hypotheses don't appeal to human reasoning in the same way that poverty causes x is popular.

What would be a better way to handle this? What would be a sounder foundation?

The better approach would be to always compare your hypothesis against the best other hypotheses that are you are aware of. And not to just compare single factors against single factors but instead compare narratives against narratives.

Of course that would radically change the nature of most papers using statistics. It is not nearly as impressive if you have to say I hypothesized "narrative z" and it was one-quarter as good as "narrative a" when tested against "dataset b." Of course this is what 90% and more of papers using statistics would turn in to if that were the standard.

mandrewa said...

But this is the way it should be done.

And if things were done this way there would be some surprising consequences including possibly real understanding.

For instance many favored hypotheses like poverty causes x would disappear. It's not that you wouldn't have narratives that include poverty causes x but such narratives would be distinctly less impressive, and most people would understand that they are probably not true, if it was made clear that there are other narratives that correlate with the data better.

But of course for some things poverty causes x would be part of the dominant narrative. Those would be the situations where it likely is true that poverty causes x or poverty causes x in combination with other things.

Instead we have a situation where statistics is routinely used to suggest or prove that poverty causes x where in fact there is no such relationship and alternative, better explanations, as demonstrated by the math, are routinely being ignored.

320Busdriver said...

Fernandistein.....” But "WHO says no evidence shows that having coronavirus prevents a second infection"

Osterholm covered that this morning. WHO must have walked that back already. There are monkey studies that show they have immunity when re-challenged.

Out of the box thinking, like Manhattan Project, hell thats a perfect name. Manhattan2

We need to scale up tests biggly. Like never before. Big Beautiful tests. Reagent supplies are currently a choke point worldwide. Saliva test maybe. Otherwise we’re going to drown. Blind me with science.

Big Mike said...

@Althouse, I learned Bayes' Theorem in a single class session taking a probability course as an undergraduate (I was a math major) and have been applying it the rest of my life building -- and validating -- mathematical models. My intuition is that it doesn't really apply in the case of New York. What are the contributing factors to New York City's mortality rate? (Last time I looked New York State accounted for roughly 40% of the fatal COVID-19 cases.) Here are a few:

(1) There is a possibility that Italy is dealing with a more virulent strain of the virus than elsewhere, and that it came to New York City from Italy. We have no way of evaluating either the likelihood of the former (i.e., that Italy has suffered from a more virulent strain) or the second (that it came from there to New York City) without DNA testing. At which point the probabilities become either zero or 100%.

(2) The New York system is a unique mechanism for spreading the disease. I note that there are other cities which rely heavily on subway transportation but which don't appear to have been hit nearly as hard as New York -- Washington with its Metro, San Francisco with BART, Boston with the MTA. I also not that New York's response to the shelter in place unless essential was not to add cars and increase the number of trains so that people could distance themselves better from fellow passengers, but to decrease the number of trains and the number of cars per train so that passengers remained crowded together. The comparison between New York's subway and the other three cities I mentioned is difficult to reduce to a set of probabilites.

(3) Density, as noted, must also correlate. If one lives in an apartment building then one breathes air in the lobby that includes exhalations from other apartment dwellers, some of whom may be positive for the virus.

(4) New York, city and state, are uniquely poorly governed. I judge that Jonathan Althouse Cohen would rate this probability as zero while, looking on from afar, rate it at 100%. Andrew Cuomo's policy of forcing nursing homes to accept patients who have tested positive is virtually guaranteed to increase the mortality rate, and his "if they can do so safely" is an ass covering made of the thinnest tissue paper. I also note that De Blasio was still keeping gyms open (at least the Y that he went to) long after other states had ordered all gyms closed.

(5) New Yorkers may be unusually unserious about taking precautions. Here I am generalizing from the likes of Chris ("Fred") Cuomo, which may be an error. But New Yorkers I have known in the past have treated rules as being for "the little people," even when objectively they were little people themselves. In my experience New Yorkers are uniquely bad in that way. But how does one turn that into a probability suitable for Bayes' Theorem?

Big Mike said...

A good explanation for Bayes' Theorem is here.

Leland said...

Yikes. If you are going to claim something is illogical, then you best under some Algebra. The relationship doesn't have to be linear.

William said...

There are many variable and unknowns about the virus. I suppose in retrospect some strategies will look better than others, but we will only know these self evident truths in retrospect. This much is clear: NYC and Italy did something wrong....I note in passing that for more than one hundred years after the Spanish Flu, Woodrow Wilson's role in spreading and intensifying that disease was never considered or even mentioned. I presume that after this outbreak he will catch some flak, but it's hard to accurately judge who are the heroes and villains in real time.

William said...

In the quandary between lock down and re-open, it's possible that there's no right answer....I'm reading a bio of Grant. In his second term, the economy crashed. There was a demand to print greenbacks and re-inflate the currency. Grant chose to stay on the gold standard and have paper money redeemable in gold. His decision caused the depression to last longer and be more severe. His decision also caused the Europeans to invest heavily in American railroads and industries. This was directly responsible for later economic growth and perhaps it's why the USA became the world's industrial powerhouse....If you lived through Grant's depression and were adversely affected, it was the wrong decision. If you came online a little later, it was the right decision.

Bruce Hayden said...

“(1) There is a possibility that Italy is dealing with a more virulent strain of the virus than elsewhere, and that it came to New York City from Italy. We have no way of evaluating either the likelihood of the former (i.e., that Italy has suffered from a more virulent strain) or the second (that it came from there to New York City) without DNA testing. At which point the probabilities become either zero or 100%.”


Actually, there is a way. Apparently the virus has mutated a bit, and there is a dozen or two sequence of bases that can be used as a fingerprint. The sequence appears to be nonfunctional. I can’t find a link right now (likely on another iPad). It is interesting to see which strains went where. I will repost when I find it.

The Gipper Lives said...

New York had some shitty hospitals to begin with. Now they're Super-Shitty. The word "murder" has been used.

The Gipper Lives said...

ps; No family is allowed in to advocate for patients. They're being left to die in some cases.

RichardJohnson said...

Density doesn't work even within New York City. Here is a recent update of COVID-19 deaths in NYC. Without correcting for deaths per population, it is evident that the lowest death rate is in Manhattan, which is also the densest of the five boroughs.

COVID-19 Deaths, April 25 update
Bronx 2480
Brooklyn 3420
Manhattan 1487
Queens 3511
Staten Island 556

mandrewa said...

So off working on something else again and it occurred to me that there is an opportunity here. This is a situation where a small group of people with a relatively small effort could help the situation.

Let's imagine a website called the "Battle of the Coronavirus Models" or "Coronavirus Statistical Correlations." The goal of this website would be to compare different narratives. Part of it would be maintaining links to different datasets since when we talk about understanding coronavirus different people are going to have different preferences for what data we should use and what adjustments to the data should be made.

Now it isn't so much that the website would be maintaining different datasets, although in some cases maybe that is possible, as it is linking to and using the coronavirus datasets that other people are maintaining.

The second part of the website, and the heart of it, would be the alternative narratives or models. These narratives could be very simple models such as climate explains it all or exceedingly complex with many factors and complex relationships between them. But as long as each can be reduced to a mathematical formula that can be applied to a dataset then that is all that is needed.

Much of the work of doing this website would be reading through the papers looking for the models or hypotheses that researchers are proposing and then naming these hypotheses and encoding them for the website.

The third part of this would be to each day, or each week, run all of the models against all of the datasets and then list in order from the best to the worst.

I'm willing to help out on such a website.

«Oldest ‹Older   1 – 200 of 219   Newer› Newest»