[ghthor]

I don't understand what's wrong with the data they presented. That article doesn't explain it, it just says it's wrong. There applications of data make perfect sense to me, that's how I'd use the data they presented.

That article makes inaccurate statements like, "basically hyped a bunch of data and weren’t transparent about their methods" which is not true. There were 100% transparent about where there information was from and how they were presenting it.

[brlewis]

> That article makes inaccurate statements like, "basically hyped a bunch of data and weren’t transparent about their methods" which is not true. There were 100% transparent about where there information was from and how they were presenting it.

I can vouch for their transparency. The speed with which I was able to see that their statistical methods were totally wrong was almost entirely due to their transparency. The only fact I had to find outside of what they said was the criteria for testing in the county for which they cited statistics, which I found at https://kernpublichealth.com/coronavirus-disease-2019-covid-...

"Testing is NOT indicated for asymptomatic persons."

The doctors' statistics in the quote below don't work because of sampling bias.

“In Kern County, we’ve tested 5,213 people and we have 340 positive COVID cases. Well, that’s 6.5 percent of the population. Which would indicate a widespread viral infection similar to the flu,”

[kolanos]

Did you watch the video or are you just basing this on out of context quotes? What you're quoting was a brief anecdote about what those doctors were seeing on the ground. Their extrapolations are based on state-level aggregates in New York and California. Sample sizes of 300-650K patients.

[brlewis]

I did not get the chance to see the video before it was taken down, but I got the content of their arguments here: http://www.tathasta.com/2020/04/watch-er-doctors-urge-reopen...

They use the same logic for Kern County, California, and New York. Their logic is, that if you test N people and find M positive cases, that in the population at large you'll have an infection rate of M/N. This would be true if the N tests were performed on random members of the population. But because the tests are only performed when there's reason to suspect infection, M/N is a higher infection rate than in the population at large. Their inflated number for total infected persons makes the death rate appear smaller.

[TechBro8615]

FYI the full video is available at the website of the news station that uploaded it to YouTube (where it was removed):

https://www.turnto23.com/news/coronavirus/video-interview-wi...

[brlewis]

Thanks. 51 minutes is too long for me to watch. Anybody who knows of anything in there that addresses the sampling bias issue, please point me to how far in it is.

[runawaybottle]

> I did not get the chance to see the video before it was taken down

I suppose that’s the real problem, in so many words...

[shkkmo]

I did watch the video and the critique by brlewis is perfectly accurate.

[xeromal]

But why should we believe you? Why can't we watch and make our own judgements?

[vannevar]

I did watch the video. At about the 5-minute mark, they make a huge error in extrapolating the percent positive of tests to the general population (in California, 12% positive test rate). As many others have pointed out, that is an inexcusable error that leads them to an wildly erroneous conclusion about the mortality rate of COVID-19. Recent antibody tests in Santa Clara County and Los Angeles showed a rate around 2-4%, and those studies were criticized because those percentages are close to the levels of false positives one could expect in the antibody tests. So at best, the conclusions in the video are off by a factor of three, and most likely, more.

[kolanos]

I'm not saying these doctors are correct in their statistical extrapolations. What I am saying is the Kern County anecdote being spread around to discredit them by people who haven't even seen the video are practicing in the disinformation that they accuse these doctors of perpetrating.

[brlewis]

As I fully explained in a nearby comment, the logic in their Kern County anecdote is representative of their logic for California and New York statistics as well. I am not misrepresenting anything or practicing any form of disinformation.

[fao_]

>I am not misrepresenting anything or practicing any form of disinformation.

I'm not saying that you are misrepresenting or disinforming people. But, how would you know unless someone told you?

[threatofrain]

> The American Academy of Emergency Medicine (AAEM) and the American College of Emergency Physicians (ACEP) jointly and emphatically condemn the recent opinions released by Dr. Daniel Erickson and Dr. Artin Massihi. These reckless and untested musings do not speak for medical society and are inconsistent with current science and epidemiology regarding COVID-19.

https://www.aaem.org/resources/statements/joint-endorsed/phy...

[soamv]

I did watch it. Their extrapolations were wrong. You can't just multiply positivity rate found in a biased sample by the whole population. That's not a meaningful extrapolation. Then using that extrapolation to say the death rate is low is just building up the error even more.

And then, comparing that to the flu and saying this isn't much worse, we're now well into the "this video is a potentially a public health hazard" territory.

[stOneskull]

No, you just debate it. A follow up video where there's a debate. Simple.

[ceejayoz]

That puts you in https://en.wikipedia.org/wiki/Gish_gallop territory, where watching hours of video and an epidemiology and stats degree is needed to tease out the truth.

[thu2111]

And then, comparing that to the flu and saying this isn't much worse, we're now well into the "this video is a potentially a public health hazard" territory.

No, lockdowns are the public health hazard. The data is quite clear at this point that what's going on here is no worse than the flu. Here's a simple graphic from the UK to make that point clear:

http://inproportion2.talkigy.com/

And if you look at when lockdowns could have possibly started working given incubation and death lags, you'll see that most countries seem to have peaked before the lockdowns could have started working. Implication: they weren't responsible for the peak and decline.

But the more important point is that it's critical such things can be debated. Simply assuming you're right and anyone who disagrees is a "public health hazard" is totalitarianism. You won't have any ability to defend yourself, or even any right to, next time someone erases you because e.g. Trump declared your views to be dangerous.

[orwin]

Well, i don't know about the us you probably have more hospital beds than us, but in france it was absolutely criticla to enforce a lockdown at the time we did. Even with this timely lockdown, we still had to stop intubating >90 yo patients for two weeks (it's getting better). And bear in mind that only a part of one region (1/13th of France basically, not really true but...), not even the whole area, and all the hospital in the region (not the hot area) were so full they had to transfert arounf 20 patients abroad and hundreds to hospital in other regions. My cousin in Colmar, eastern France (SMUR chief) has transfered around 20 people to my other cousin (intensive care anesthetist) in western Brittany. If you have a map: its basically the longest East-West travel you can do in France. In Colmar they also had to close almost all of their non-maternity, non-urgency service, and the number of people dying of overdoses or of too much alcohol (don't have the english word sorry) is at least twice as high (we will have the true numbers in a few months).

Also i've said it here before, but young, healthy people who did not had to have invasive respiration (so no ICU) still got out of their sickness with lingering asthma, possibly caused by a pulmonary fibrosis. I found this [0] in English if you don't know what it is. In five to seven years (assuming we find a vaccine for sras-cov2 in two year) we will really start counting our dead, unless we find a non-chirurgical fibrosis cure.

[0] https://lungdiseasenews.com/2017/06/16/12-facts-pulmonary-fi...

[shkkmo]

I did watch the video and I can safely say that these doctors make massive and very convenient statistic mistakes throughout.

Sample size doesn't help if the sampling method is biased. Even if tests were available to everyone regardless of symptoms, you would still have a selection bias that would skew the numbers. The only way to accurately estimate overall infection rates in the general population is to test a random sampling.

Then there is the "comparison" of Sweden (no lock down) and Norway/California. They mention the death counts and population of these countries in another attempt to show that the "number of cases is high and the number of deaths is small regardless of lockdown" but completely gloss over the fact that Sweden has more that 5x the covid deaths per capita that Norway and California have.

Regardless of if this video should be censored, these guys are awful scientists.

[AnthonyMouse]

> Regardless of if this video should be censored, these guys are awful scientists.

This is the exact reason why it shouldn't be censored. Because if their science is bad then it needs to be available for people who know better to tear it to shreds.

If you take it down then the story can't be "this is why they're wrong" because what they said isn't available anymore, so a rebuttal isn't believable because you can't tell if it's an accurate representation of the original presentation.

So then the story becomes about censorship and you're making it all too easy for motivated people to spin a conspiracy theory about how these scientists are speaking truth to power and getting oppressed, even if they are in actual fact totally wrong.

[shkkmo]

Yet if you look at the comment sections for this video, there isn't anyone tearing it to shreds.

This isn't something I have been able to confidently form an opinion on; it's a hard topic. I think we will be strugging with how to balance bias, deliberate misinformation and freedom of speech for a long time.

But I think at the very least, YouTube needed to put up prominent disclaimers and links to limit the harm that this content can cause.

[AnthonyMouse]

> Yet if you look at the comment sections for this video, there isn't anyone tearing it to shreds.

I count several detailed criticisms of it just in this thread. Do none of them have a YouTube account?

Why don't the people proposing to censor it post their own rebuttal?

> I think at the very least, YouTube needed to put up prominent disclaimers and links to limit the harm that this content can cause.

Which is totally different. Leaving the video up while posting a prominent link to a rebuttal isn't censorship, it's more speech.

[ceejayoz]

> I don't understand what's wrong with the data they presented.

They took stats from a massively self-selecting population - people seeking out COVID-19 testing at their facilities - and extrapolated to the general public.

That's inexcusably dumb.

As the bit I quoted states, that's like "estimating the average height of Americans from the players on an NBA court".

[holler]

> That's inexcusably dumb.

Even if that’s true, what does it have to do with banning the video and preventing the public from making their own decision?

Right this moment I can go on youtube and look up a million rap videos that talk about guns, violence, murder, and more. I can look at videos that would lead me to harm myself in all sorts of ways if I didn’t have a brain. By picking and choosing like they’re doing, it leads one to distrust the company which will ultimately hurt them.

[programmarchy]

Come on. Rap videos are often grotesque but it’s entertainment, like a horror film.

This video was more like yelling fire in a theater. I’m willing believe the doctors were sincere, but they were gravely wrong, and their misinformation was endangering the public.

YouTube was correct to turn off the alarm they erroneously turned on.

Granted, YouTube’s method of throwing up a standard “violation of community guidelines” message is crude, and leads to suspicion. Then again, so was ABC’s editorial judgment. It’s a hard problem.

[holler]

can you or anyone factually prove that they’re wrong? 99% of the internet and “experts” said HCQ is dangerous and ineffective, then it just came out that a leading group of ER docs say it’s effective in ~90% of cases.

The point is that even if they’re wrong, youtube and others are playing a dangerous Orwellian game and it’s going to harm their reputability in the end.

[jdkee]

"Even if that’s true, what does it have to do with banning the video and preventing the public from making their own decision?"

Research done on fake news illustrates why ordinary consumers of news cannot be trusted to make informed judgements.

"Right this moment I can go on youtube and look up a million rap videos that talk about guns, violence, murder, and more. I can look at videos that would lead me to harm myself in all sorts of ways if I didn’t have a brain."

Exactly this. This is why it is irresponsible for a global platform to promote this tripe.

[kolanos]

Did you watch the video? That is not what they did. What you're describing is a brief anecdote they shared about what they were seeing on the ground. Their statistics are based on state level aggregates for New York and California.

[Retric]

State level data is collected using the exact same methods. In many places people with symptoms are told to self isolate until they have difficulty breathing etc. This makes state level data a highly biased sample.

There are a few population studies for very specific areas, but no widespread statewide sampling anywhere.

PS: The US has ~50,000 Coronavirus deaths. Assuming a very optimistic mortality rate of 0.5% suggests at most 10 million infections out of 330 million people or 3.3%. A higher mortality rate of say 5% would mean total infections are possibility as low as 0.3% of the US population.

[btilly]

Given the delay between catching it and dying, a lot of the people who have it and will die from it are not currently counted in your deaths.

This could mean that substantially more Americans have it. (Though on current projections, less than a factor of 2.)

[Retric]

I was including that for my optimistic 0.5%. South Korea with a younger population, good healthcare system far from maximum capacity, and very good but not perfect testing has a CFR of 2.3% and rising. Diamond Princess has 14deaths out of 712cases, while not a representative sample it’s still conformation that a high percentage of even young healthy people need hospitalization.

The biggest question IMO is how much we are undercounting fatalities. NYC lists Confirmed deaths at 11,820 plus Probable deaths at 5,395. They also add: Due to delays in reporting, recent data are incomplete. https://www1.nyc.gov/site/doh/covid/covid-19-data.page

[nostrebored]

Your 'very optimistic mortality rate' makes a number of assumptions about the rate of asymptomatic cases, the population which experiences asymptomatic cases, etc. It ignores data coming out which suggest that interventions like ventilators may be harming more than helping.

Making any assumptions about the mortality rate with the abysmal state of testing in the US and around the world is guesswork.

Making assumptions based on the horrific codebase that's spawned our most cited models should also give anyone in software at least a twinge of anxiety.

[Retric]

Diamond Princess had a slightly older population but healthy population, tested everyone, provided good medical care. They still hit 14 deaths out of 712 infected or ~2% and only had 20% asymptotic cases. South Korea has very good testing, a younger than US population, similar rates of hospitalization to DP and a slightly worse mortality rate likely due to missing some people but having a more at risk population.

The US has an older and sicker population than SK, and in NYC an overworked healthcare system. So, a 1% US mortality rate discounted to 0.5% due to recent infections not having time to die is extremely optimistic. It would also mean testing was only discovering 10% of cases.

PS: Many more people are exposed without having sufficient infections to trigger long term immunity. But, it’s herd immunity or deaths/long term issues we care about making such cases irrelevant as their still at risk.

[kolanos]

I'm not saying these doctors are correct in their statistical extrapolations. What I am saying is the Kern County anecdote being spread around to discredit them by people who haven't even seen the video are practicing in the disinformation that they accuse these doctors of perpetrating.

[Retric]

Using state level data is exactly like sampling MBA heights. A biased sample is a biased sample.

So, no I don’t see how it’s misinformation.

[jerry1979]

I watched the videos. Just like many others have said in this thread, the doctors took the percent of people who tested positive for covid and assumed that percent of people overall in the whole state had covid. Then they took the total number of deaths from covid, and divided that number by the population of the state.

[shkkmo]

It isn't exactly what they did but it is fairly close. They consistently incorrectly extrapolate the covid positive test rate from a biased sample (biased by both the testing guidelines and self-selection) to the general population, not just for their county, but also for California, New York, the United States, Italy, Sweden and Norway.

They then use this to massively misrepresent the lethality of Covid 19.

[wesammikhail]

People will always put their fingers in their ears and not read/watch the source because it is easier to bash someone than to actually do the work that may change your beliefs

[TechBro8615]

Well, it’s also hard to watch the source video when YouTube goes out of their way to delete it. So here in this very thread, is a great example of the harms of taking down the content. It makes it hard to even have a discussion about its merits.

I don’t understand why YouTube couldn’t leave the video up, and put some kind of warning or interstitial advising of problems with the content. I would have problems with that approach too, but certainly less so than I do with the wholesale deletion of the video.

Anyway, you can find it hosted here: https://www.turnto23.com/news/coronavirus/video-interview-wi...

[Jd]

The source is gone.

[mixmastamyk]

They compared them to several countries as well, so your statement is misleading.

[djsumdog]

People should be able to see all the opinions of various experts; doctors in NYC, doctors in Minnesota. Just them on their merits and examine the evidence.

By pulling videos, YouTube/Google is saying they don't trust people to be able to discern anything or do research among several dissenting opinions. And yes, it's their platform and yada yada .. they still allow a lot of content to be uploaded and they have become the most typical/standard centralized source of videos.

I will admit, they might be right to not trust people ... at which point why do we even care about humanity anymore? Are those with more power and influence always going to think the rest of the world are sheep who can't think for themselves and need to be led around?

This time sets all kinds of dangerous prescients. I wish Orwell was still alive.

[jdkee]

"People should be able to see all the opinions of various experts; doctors in NYC, doctors in Minnesota. Just them on their merits and examine the evidence."

Most people are not qualified to "examine the evidence" and make public health policy decisions. This is why we rely upon the experts. Otherwise you have measles outbreaks in Beverly Hills.

[wesammikhail]

You mean the experts that said no human to human transmission? or the experts that called for maska being ineffective? or the experts that didnt know how to store and deploy resources as they needed? or the experts that miscalculated projected infection rates?

[seppin]

> You mean the experts that said no human to human transmission?

Perfect example, no expert said there was no human to human nor is there any chance of it. The only language close to that was "there isn't any proof of it now" which might have been informed by CCP misinformation.

Even though they were misinformed, any true expert should have known not to make any definitive statement either way. If you have such a quote, it was not made by an expert.

[thu2111]

So doctors aren't experts in health now. That's what it's come to?

Do you think epidemiologists are? The people who have apparently never predicted a disease outbreak successfully, ever? Why are they experts but not these doctors?

And what makes YouTube employees believe they're qualified to decide who's right? They're experts in HTML5 video serving, not health.

[notacoward]

> So doctors aren't experts in health now.

Even doctors generally recognize that public health (of which epidemiology is only a part) is a separate domain of expertise. There are doctors who specialize in infectious disease who need some public-health training, but there's no reason to expect that e.g. a oncologist or trauma surgeon would know more than the average scientifically-literate person about a different profession.

[reedlaw]

From the video around 4:40:

> 33,865 COVID cases out of a total of 280,900 total tested--that's 12% of Californians were positive for COVID.

This would only be true if those tested were randomly selected. Even in the most optimistic Santa Clara County study prevalence was estimated at under 5%. 1

1. https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v...