> for now we assume that total COVID-19 deaths equal excess mortality.

This seems reasonable, although I'd suggest it's an underestimate of COVID-19 deaths.

Excess deaths counts have been consistently showing negative in countries during periods of time where they have coronavirus under control.

Other significant causes of death (including suicide, influenza, road traffic accidents) appear to be down.

As such, I'd assume that COVID-19 deaths are likely somewhat higher than excess mortality figures. I'm prepared to accept the assumption that they're approximately the same.

[Edited to note: the linked article discusses this with further detail and evidence]

I just don't understand where they got such high excess death numbers in the first place. Fitting a sinusoid with a 52 week perios and adding a slope for population growth you can get a very good (R=99.xx) fit up until the pandemic. Using that ramped sinusoid the excess death number comes out about 50k over the reported covid totals.

Edit: data source is cdc total mortality page. Going back to 2014

> Other significant causes of death (including suicide, influenza, road traffic accidents) appear to be down.

Can you source your other claims? At least one seems to be wrong. Traffic deaths seem to be up:

https://www.usatoday.com/story/money/cars/2021/03/05/pandemi...

Other countries are reporting otherwise, typically in the tens of percents down. Singapore, Greece, India, Germany, UK, Zambia, Malta, Philippines:

https://www.todayonline.com/singapore/fall-traffic-accidents...

https://www.marketscreener.com/news/latest/Road-Traffic-Acci...

https://www.hindustantimes.com/cities/delhi-news/road-accide...

https://www.destatis.de/EN/Themes/Society-Environment/Traffi...

https://www.gov.uk/government/statistics/reported-road-casua...

https://zambiareports.com/2020/07/22/332-die-road-accidents-...

https://newsbook.com.mt/en/2021-sees-decline-in-road-traffic...

https://www.topgear.com.ph/news/motoring-news/mmda-edsa-road...

These are mostly unchecked secondary sources, of course. But, your secondary source, the "National Safety Council", appears to over-report deaths compared to the Department of Transportation's later figures. In 2019, they over-reported by... 8%.

> Excess deaths counts have been consistently showing negative in countries during periods of time where they have coronavirus under control.

> Other significant causes of death (including suicide, influenza, road traffic accidents) appear to be down.

That's most probably a direct factor of the containment measures. Influenza has been on sometimes historic lows (https://www.who.int/influenza/surveillance_monitoring/update...), and as many people have stayed home streets have been free and people don't drive home from bars totally drunk, both drastically reducing road accidents.

Suicides however? These are going up by some indications (https://www.nature.com/articles/s41562-020-01042-z), as is domestic violence (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7195322/).

Suicides rates are the same or down in most of the world[0], and were apparently reduced in the US during the pandemic[1].

[0] https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0...

[1] https://www.medrxiv.org/content/10.1101/2021.02.13.21251682v...

> [Edited to note: the linked article discusses this with further detail and evidence]

Here's the sum total of evidence provided:

> An analysis by the Netherlands statistical agency suggested that all excess deaths in the Netherlands were directly due to COVID-19

So they use a single study out of Europe to extrapolate here.

> The second driver of excess mortality is reduced health care utilization for many causes;3 however, the impact of reduced health care use on health outcomes is harder to prove. Many mechanisms have been proposed, including reduced vaccination rates and reduced births in hospital.4 Demonstrated increases in cause-specific mortality related to these causes, however, have not yet been verified. The impact of changes in health care utilization on excess mortality may be observed in later years, rather than in 2020 or the first quarter of 2021.

So they dismiss this.

> Third, convincing evidence has been found that rates of anxiety and depression have increased, which might in turn lead to increases in deaths from suicide.5 To date, the evidence on increased suicide is very limited.6 Opioid deaths, on the other hand, have clearly increased7 in the United States. Compared to past trends, opioid deaths increased by around 15,000 since March 2020. Evidence on whether this has also occurred in other countries awaits further study.

So again, they dismiss this.

> Fourth, we reviewed the evidence on decreases in injuries as a result of reductions in mobility. We analyzed data from 12 countries that provide cause of death data by week or month, which allows us to test whether some causes decreased significantly during 2020 and whether that decrease was related to the decreases in mobility that have been reported. This analysis suggests that globally, injury mortality decreased by 5% in 2020 due to reductions in mobility. At the global level, this translates into a reduction of approximately 215,000 deaths.

Conveniently, when then assumption works out in their favor, they include it in the model.

> Fifth, some infectious causes of death may have declined during the pandemic due to the behavioral changes associated with control of the pandemic, including mask use and reduced contact with others. Causes that have clearly declined are influenza,8,9 respiratory syncytial virus,10 measles,11 and possibly other respiratory viruses and viral diarrheas. For example, influenza cases in the United States declined 99.3% from the winter season of 2019–2020 to the winter season of 2020–2021. Combining the reductions reported in different countries in influenza, respiratory syncytial virus, and measles, the global reduction in mortality from these causes may be larger than 400,000 deaths.

Again, when the assumption works in their favor, they include it.

> Sixth, deaths from some chronic conditions such as ischemic heart disease or chronic respiratory disease declined in some months of 2020, most notably in May and June in Europe. These declines were most likely due to the fact that frail individuals who died from COVID-19 earlier in the year would otherwise likely have died from these chronic conditions. The strongest evidence for this effect is that excess mortality was negative in some countries in Europe in June when the reported COVID-19 death rate was very low. In aggregate, this effect likely reduced mortality by only 2% based on our analysis.

This is so egregious as to be laughable...they find evidence for something, and then "based on their analysis", limit the effects.

> The main potential increases in excess mortality due to deferred care and increases in drug overdose and depression are hard to quantify at this point or are of a much smaller magnitude. Given that there is insufficient evidence to estimate these contributions to excess mortality, for now we assume that total COVID-19 deaths equal excess mortality.

So they didn't even try.

Again...I don't know what the right answer is here, but this is, at best, a biased reading of the available data. All you can do is throw up your hands and go back to first principles: without prospective validation, statistical models are fiction.

On the other hand: "in many European countries there was a spike of excess deaths in weeks 31–35 during a period when COVID-19 reported deaths were extremely low. This period coincided with a heat wave and national reports of deaths due to the heat wave. We excluded these weeks of data from subsequent analyses."

I personally wouldn't have excluded these weeks - unless you're going to exclude historical (near-annual) heatwaves from the figures, this should balance out.

> All you can do is throw up your hands and go back to first principles: without prospective validation, statistical models are fiction.

No estimation is perfect, nor expected to be, and so your criticism is unreasonable. You're doing exactly the things that you are criticising the study for, and more so. You provide zero evidence to counter their sourced evidence, and ignore and dismiss points that don't match what you are trying to say.

> No estimation is perfect, nor expected to be, and so your criticism is unreasonable. You're doing exactly the things that you are criticising the study for, and more so. You provide zero evidence to counter their sourced evidence, and ignore and dismiss points that don't match what you are trying to say.

They have provided no "evidence" -- they made a model. They have provided no validation of that model, so it could be completely incorrect. Nobody knows.

Furthermore, I just showed you that of the sourced evidence they did consider, they dismissed nearly everything that reduced the death rate attributable to Covid, and said it was "too dificult" to quantify.

Mathematical models without prospective validation are not evidence. They are not science. They are bedtime stories.

> without prospective validation, statistical models are fiction.

I wouldn't be completely dismissive. The first attempts at using statistical data to drive medical practice were during the Napoleonic wars, and dramatically improved medicine. I would expect the statistical methods used were pretty crude, as it was a very young discipline.

Crude methods are fine. Fantastic models are made with linear regression.

No matter how crude your methods, if you don't validate your model prospectively, it is a mathematical fairy tale.

Maybe I'm just ignorant of the relevant statistical methods, but how do you suggest models be validated prospectively? All the model validation techniques I'm aware of come after the model is fit on the data (or a subset of it), like testing against a holdout set, or k-fold cross-validation, or checking R^2.

> testing against a holdout set

Not just a holdout set -- a holdout set of prospective data, where you actually predict the future, blindly, and see how you do.

Obviously, you have to wait a little while to gather such data before you make big claims with your model. Doesn't mean you get to skip it.

It's not even clear to me that the authors did a cross-validation here, or even bothered to fit the free parameters of their model based on any kind of empirical data at all. At least with regard to percentage of excess deaths due to Covid, I'm not sure how they could...the parameters seem to be unrelated to any empirical data.

What is the mathematical difference between testing against a holdout set of data collected at the same time as the data used to fit the model, and a holdout set collected in the future? I'm in total agreement that avoiding, e.g., overfitting is a good thing – but what work is prospectivity doing here?

Also, if I understand correctly, none of the modeling done in standard texts such as Gelman et al.'s Bayesian Data Analysis involves prospective validation under your definition. Should we then classify the examples in that book as "mathematical fairy tales"? This seems like a fairly strict standard!

A prospective test is blind. You don't know what the data looks like, so you can't cheat.

> Also, if I understand correctly, none of the modeling done in standard texts such as Gelman et al.'s Bayesian Data Analysis involves prospective validation under your definition. Should we then classify the examples in that book as "mathematical fairy tales"? This seems like a fairly strict standard!

It's surprisingly common for models to pass all of the cross-validation you want to throw at them, and fail in the real world. I don't care what statistical techniques you've applied, if you don't conduct blind tests, you don't know how your model performs.

Setting this aside: TFA did nothing you're talking about. Let's be clear about that.

> A prospective test is blind. You don't know what the data looks like, so you can't cheat.

Suppose I blind myself to the holdout set. What's the difference?

> It's surprisingly common for models to pass all of the cross-validation you want to throw at them, and fail in the real world. I don't care what statistical techniques you've applied, if you don't conduct blind tests, you don't know how your model performs.

What exact failure mode is such that a) prospective testing guards against it, and b) traditional validation methods do not?

Take for example distribution shift. Suppose I'm worried the underlying data generating process is going to change between training and deployment. You propose, I guess, that this is fixed by collecting more data prospectively. OK, suppose and I do that and everything checks out. Now what guarantees there is no distribution shift between the time I do the prospective testing and subsequently deploy the model?

To be more direct: one eventually has to make assumptions of statistical regularity and distributional constancy somewhere, at some point, in order to do any statistical inference at all. If you have good reasons to make such an assumption, I don't see why prospective data collection is any different from a regular holdout set. And if you don't, then you're screwed no matter what you do.

> Setting this aside: TFA did nothing you're talking about. Let's be clear about that.

Sure, but strongly held, unorthodox opinions about statistical practice are more interesting to me than tearing down some mediocre article.

Not the poster, but happy to answer since I have a decent amount of experience in similar areas.

Their point is that while in theory you could blind yourself to the holdout set - that rarely REALLY happens. At least in a truly good way. Almost everyone is going to either peek at the data, massage the data, or try the validation and when it fails tweak the model and try again, or any number of shenanigans that no one admits happen - but happen all the time.

The failure mode that a prospective test would guard against in this case is - you can't peek or massage the future (and if you can, you can retire a billionaire pretty easy and stop worrying about this) - so you can't accidentally fool yourself or others by peeking, massaging, etc. And if you tweak your model, then do another prospective test - then you still can't peek at the future, and you'll have a different set of data, so you can't be overfitting to the old data/p-hacking.

You can do it over and over again, and as long as you don't just say 'all good' on a failed test without doing another different prospective test, you either have a predictive model or you clearly do not.

That is something the traditional validation models cannot do, since they aren't looking for truly never before seen data.

It still doesn't tell you if once in production you will stop fitting and fail over, but it does at least tell you that it fit a never been seen, new set of data at least once. Which is better than the other tests can do by their very nature.

Yes, exactly. But I'd extend this one part:

> Almost everyone is going to either peek at the data, massage the data, or try the validation and when it fails tweak the model and try again, or any number of shenanigans that no one admits happen - but happen all the time.

While "peeking" at the data is certainly something that happens all the time, usually what happens for even the most rigorous of model-makers is that your model fails "out of domain" -- the past doesn't accurately reflect the future.

Even if you get everything else right, it's incredibly common to fool yourself into believing you're capturing "reality", when in fact you're just making a good fit to past random fluctuations.

The point about peeking at the holdout set is fine, I guess. But then the slogan should be "models without a truly blinded holdout set are statistical fiction." Prospectivity doesn't do any work above and beyond blinding there.

I already addressed the point about distribution shift above. You need to assume that the past reflects the future at some point, otherwise you can't justify deploying the model even after the prospective test.

> a holdout set of prospective data, where you actually predict the future, blindly, and see how you do.

According to your theory, publishing would act as a 'nothing up my sleeve' validation of their model.

Since, you can check this model in the future - you can do it for yourself.

And yet, you seem to be complaining about this publication?

(I still don't agree with this as a principle - since I wouldn't now expect the future to reflect the past, and the authors make no claim to predictive powers when they are identifying that the present is unlike the past)

> The second driver of excess mortality is reduced health care utilization for many causes;

Wouldn't that still be technically tangentially an after effect of covid? I mean no covid, no reason not to get to the doctor, so if covid didn't exist you might survive those, no?

I noticed (late summer last year) that especially age group 25-44 is/was affected in the US with a +25% (excess) mortality YoY. It is not entirely unusual that diseases affect different countries differently; However I found it peculiar.

I suspect this is the same in other countries.

25-44 year olds don't usually die (typical mortality is approx 0.1%), so +25% represents a relatively small number in absolute terms.

USA is an outlier on this. You can compare with Italy for example:

https://www.statista.com/statistics/1191568/reported-deaths-...

https://www.statista.com/statistics/1105061/coronavirus-deat...

USA has 30 times the deaths among 30-39 year olds but only 5.5 times the population. It isn't just Italy, almost every other country has a distribution similar to Italy.

Italy has a huge older population and a lot more lax on cigarettes... older smokers = raised chances of death...

USA on the other hand has excess obesity across all metrics (I'm one), so if it hit younger ages harder here it's probably due to obesity epidemic raising death chances.

At least in Germany[0] and Sweden I found nothing like that.

[0]: https://www.destatis.de/DE/Themen/Gesellschaft-Umwelt/Bevoel...

edit: I found no English version unfortunately, but at least the Excel file which is easier to work with.

Americans are incredibly overweight on average

Excess deaths could easily be attributed to non-covid factors like Alzheimer’s.

Suggesting the US had 900k covid deaths is ridiculous. Only 1 of the 6 considered factors lists covid as a direct cause.

> Excess deaths could easily be attributed to non-covid factors like Alzheimer’s.

Why would there suddenly be an 100% YoY increase in the number of Alzheimers deaths?

Hospitals and doctors were basically not handling any non-covid patients.

Alzeimer's currently doesn't have any effective treatments, so I'd repeat the question: how would the fatality rate from Alzheimer's double year over year?

Alzheimers patients are among the worst off in a restricted healthcare scenario.

https://www.wsj.com/articles/coronavirus-pandemic-led-to-sur...

Because we like to ignore facts that counter our worldviews.

> for now we assume that total COVID-19 deaths equal excess mortality.

This part is especially doubtful for me, as there are factors moving it in either direction. For one, the pandemic measures also helped a lot for things like the flu, which had less cases and therefore caused less deaths. On the other hand, problems caused by lack of exercise and social interaction have definitely increased. Lastly, I'm quite sure we have a lot of missclassified deaths in either direction.

These are all quite different and I see no reason to assume these factors even out.

I would generally expect non-covid deaths to be down as is indicated by correlations between mortality and economic decline, the fact that suicides haven't trended substantially upwards, and the fact that cars are off the road.

There are around 30,000 to 35,000 car deaths per year in the US over the last ten years, and around 35,000 to 45,000 deaths from suicide. Those are terrible number sin absolute terms, of course, but even if those went to 0 for a year during the pandemic, that’s still less than 15% of the deaths currently attributed to COVID-19 in the US.

I saw an article to the effect that car-related deaths are up because, since there are fewer cars on the road, the drivers who are left drive faster and/or more recklessly and get in more accidents. So, as with everything in this pandemic, it's complicated...

Yeah I was struck by how fucking insane people got on the roads shortly after the pandemic struck.

You'd think people would chill out because the roads were practically empty, but I guess pandemic stress combined with the immediate rage of "the roads are empty and this asshole camper is only doing 75 mph in left lane" leading to absolute lunacy at very high speeds.

Normal levels of rush hour traffic around here mean nobody can do more than 50 mph at best with an average speed closer to 35, which probably keeps the deaths down.

Interesting. So scratch the bit about cars.

Doesn't honestly surprise me that much , I was driving up i95 in late March and people were pulling some really crazy shit in the early pandemic. I also saw a few (looked deadly to me) wrecks on the highway.

Traffic deaths were up very slightly in 2021. Not enough to move the needle on all-cause mortality.

For my country and some others I monitored, the first wave happened outside of flu season, and excess mortality was higher than COVID deaths at that point and continued to track.

Second wave was more difficult since it overlapped the normal flu season.

You can see graphs here showing normal flu seasons and where COVID fits in: https://www.euromomo.eu/graphs-and-maps/

Some more factors might be hospital crowding early in the pandemic, people avoiding medical treatment, and increased homicide rates.

> people avoiding medical treatment

People avoiding medical treatment is awkward in the figures, because - medical treatment sometimes kills people. First link I found suggests that medical treatments cause 800,000 deaths per year in the USA.

If medical treatments are really down, I'd guess we might actually expect mortality figures to be temporarily down (but, higher later).

> irst link I found suggests that medical treatments cause 800,000 deaths per year in the USA.

That number pops up all the time in clickbait headlines, but the study behind it doesn't really support the claim.

There is a pretty substantial difference between "Deaths caused by medical mistakes" and "Deaths that occur in a certain timeframe after a medical mistake occurs" (which is what the study actually looked at) are two very different things.

The sicker someone is, the more likely they are to receive healthcare (so there are more opportunities for mistakes). Also, the sicker someone is, the more likely they are to die. Those two combine to result in a lot of folks that die shortly after a medical mistake, even if the mistake was very minor, and didn't contribute in any way to their death.

I really ought to reread my comments during the edit window... That second paragraph...

If medical treatment causes more deaths than it prevents on average, we’re probably doing it wrong, though maybe it looks different in the short term as you imply.

Agreed. All medical treatments are about balancing risks.

Whenever you take any surgery (for example), the meaningful risk of immediate death is balanced against the benefit of better quality of life later.

If you measure the impact of medical treatments only on the day they are taken, you'll only find evidence that a bunch of people have died who wouldn't have otherwise died on that day.

You can't then observe the benefit until some time has passed.

I wouldn't call it at all doubtful. The fact there's less cars on the road can only push excess mortality down, regardless of lack of exercise, so it seems sensible to attribute it all to the pandemic.

Also, suicides aren't significantly up,the physical and psychological effects of lockdown don't happen immediately, while a reduction of car deaths do.

e.g. UK suicides have not significantly changed in 2020:

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsde...

So overall I think they're pretty safely assume that any excess deaths are coronavirus, especially as flu has all but disappeared too.

The IHME has been wildly wrong yet highly cited for over a year now. I've still got screencaps of their models that predict around 90,000 total deaths in the USA by the end of 2020. They were lifting a sigmoid curve off of China's numbers and applying it to the United States entirely blindly. It didn't even predict the shape of the decline off the first peak correctly since it was implicitly assuming the US would take similar steps to China (martial law) and drive the R0 down to 0.4 or so.

Yes, I don't understand why the IHME is still taken seriously after their awful models a year ago where they were repeatedly, confidently wrong.

Link predicting deaths drop to 0 by June 21, 2020 (i.e. last year). Spoiler: they did not. https://ktla.com/news/nationworld/influential-ihme-model-pro... The original paper https://www.medrxiv.org/content/10.1101/2020.03.27.20043752v...

There has been a ton of terrible modeling this year, and the IHME is just one example. This week the CDC published an ensemble of models trained on Covid case data up to March 27.

The models completely mispredict the actual number of cases since that time:

https://pbs.twimg.com/media/E0qZTvVVkAIs484?format=jpg&name=...

This is a total failure of validation that would make any other practitioner question the whole exercise, but it didn't stop the CDC from publishing them. Nor did it stop the news media from trumpeting the results. The Washington Post and CNBC and others wrote headline articles fawning over them.

But this wasn’t written to confirm that COVID was underreported, nor predict future results, it was written to estimate the mortality rate of the past year and a half. Your criticism doesn’t seem to hold here.

Also: The author cited the sources for their assumptions, specifically on COVID mortality classification.

Surely you'd agree that "2020 excess deaths are going to be predominantly covid" is a very reasonable hypothesis though, right? And you agree that "2020 has an extraordinarily high excess death rate", because that part is easily measured data, right?

Basically: your response seems a bit nitpicky to me, which tells me that perhaps you're starting from a perspective where you're assuming the article is wrong and looking for holes to poke.

Look, nothing about this paper's conclusions is surprising at all. At all. We all know true covid deaths are higher than reported, because all such metrics underreport, for all diseases (and frankly almost all causes of death). The only question is "By how much?". So you don't like 400k. What's your counter?

I think the critique is more that the 'study' is really just a complicated statistical restatement of the assumption, so it doesn't prove or even highlight anything new or interesting.

That's some kind of relativist argument I don't understand. Modelling and theory is an important part of science. We have something that is difficult to measure independently, so we build models to derive an explanation from things we can measure. And they did that. And the results line up with the hypothesis.

I mean, duh. Usually hypotheses are right! You're right that it would be more interesting if it showed some kind of confounding effect or whatever.

But it didn't. It turns out that the boring explanation seems like the right one: the anomalously high number of people who died of unexplained reasons in 2020 was just the giant global pandemic. Oh well.

> Surely you'd agree that "2020 excess deaths are going to be predominantly covid" is a very reasonable hypothesis though, right? And you agree that "2020 has an extraordinarily high excess death rate", because that part is easily measured data, right?

I don't know the answer, and neither do you. But given that the top-line is asserting that there are 900k excess deaths due to Covid, the choice of assumption here is not some minor detail. There are lots of other plausible alternative hypotheses.

Aassuming X, then building a model that shows X is true, is not validating a hypothesis.

> There are lots of other plausible alternative hypotheses.

Please list them.

We have data about deaths going back many decades. It's very safe to say that if in 2010 you made a prediction about the number of expected deaths in 2011, you would have been very close to the real number. Same if you had predicted the deaths for 2019 in 2018.

Suddenly, after decades and decades of being almost perfect in your predictions, you would have been wildly wrong for the predicted number of deaths in 2020.

After decades of predictability, there was a whopping 20% increase in 2020 [1]. How can you possibly say it wasn't caused by Covid?

[1] https://jamanetwork.com/journals/jama/fullarticle/2778361

And that's the very beginning.

We also have plenty of time series data from effectively instituted lockdowns worldwide. Excess deaths rise prior to the lockdown, in line with but greater than reported COVID deaths. They continue rising until median COVID- time-to-death days after the start of lockdown. Then they fall, sharply, dipping below regular excess deaths (and tracking the fall in COVID deaths). That isn't a pattern that fits the "maybe some other factor like the lockdown itself caused the spike" hypothesis, and it is also the pattern that was predicted by the scientists advocating the lockdowns.

Sure, some deaths are caused by lockdown, but the evidence suggests a pattern of the deaths which have been[1] caused by lockdown being [more than] netted out by normal deaths prevented by lockdown.

Some days it's cold, and anthropogenic climate change models are just fancy math, but this doesn't make all climate hypotheses equally plausible

[1]there's a plausible argument lack of medical attention during lockdowns and/or long term economic consequences may have a significant death toll in future, but that doesn't affect the "excess deaths reported during COVID have matched COVID spikes evidence

Not sure why I can’t reply to the person who replied to you - “timr” who is asking for data, but there is data here: https://www.euromomo.eu/graphs-and-maps/

Not one of those graphs contains annotations on lockdowns, so not one of them provides evidence for the claims you're making.

Nobody is disputing that excess deaths are happening, the question is why they're happening.

Cut/paste from the other reply

Again, not sure why I can’t reply directly to timr, but the data in the graphs on EuroMOMO is not spelled out, it’s just data. That’s no reason to dismiss it?

You can start by asking questions like- why do some neighbouring countries have different profiles? Why do some not have any excess mortality? Why do some have multiple spikes?

Cut/paste

Your claims of "vast time series data" would be more believable if you actually linked to some of it.

It's interesting how you can't suspend your credulity to encompass the idea that there are alternative hypotheses for excess deaths, but you are more than happy to assert the existence of "vast datasets" to defend your own opinions.

Here are a few plausible alternative hypotheses that actually have substantial backing evidence, which can be obtained with trivial googling:

* Excess deaths due to missed treatments (we know this is happening, e.g. for cancer)

* Excess deaths due to untreated acute medical emergencies (e.g. physicians in NYC noticed the curious phenomenon of heart attacks disappearing from ERs during the height of lockdown

* Excess deaths due to drug and alcohol abuse (again, we know it's happening)

I'm sure there are many others worth consideration. My point is not to assert that any of these are automatically true, but that they merit consideration, and anyone who doesn't fairly consider them is probably not engaged in good-faith debate.

> There are lots of other plausible alternative hypotheses.

I would be curious which ones that would be.

Sure, here's a much better treatment of the same subject:

https://ourworldindata.org/covid-excess-mortality

Which is an excellent article. I just fail to see how you think it's inconsistent with the numbers in the IHME story above or their methodology. Lots of these confounding factors are expressly enumerated. I mean, do experts like the authors of that OWiD article find the same fault with IHME that you do? Can you cite some making similar criticisms?

Again, I get the distinct feeling that you're arguing with methodology here not because there's anything particularly suspicious with IHME at all, but because the conclusion (that the US is approaching 1M covid deaths, something that should surprise no one) is politically inconvenient.

> I mean, do experts like the authors of that OWiD article find the same fault with IHME that you do? Can you cite some making similar criticisms?

I don't know. I don't cross-check my thoughts with every "expert" in the world before I express them. I have expertise in this field; I can think for myself.

> Again, I get the distinct feeling that you're arguing with methodology here not because there's anything particularly suspicious with IHME at all, but because the conclusion (that the US is approaching 1M covid deaths, something that should surprise no one) is politically inconvenient.

You have no idea what my politics are. I am saying that this model reflects the assumptions used to create it. Nothing more, nothing less. It is not a validation of the assumptions -- it is a regurgitation of the assumptions.

You have deaths due to Covid directly, additonal deaths that come as a result of good policies, and additional deaths that come as a result of bad policies.

It's disengenuous to say Covid is the cause if politicians used hamfisted responses that led to more deaths than necessary.

Sorry, what exactly is the mechanism for those policy deaths you're invoking? Can you cite the relevant study you're thinking of and explain why it's more convincing than people simply dying of a serious disease?

It seems like you have an issue with this paper not because of the methodology but rather because the conclusion disagrees with your worldview.

At the very least this has no way to distinguish directly between direct and indirect effects of the epidemic. The average person has put on weight, gets less exercise and is less likely to go visit a doctor. I think the distinction is important if this type of statistics are used to consider e.g. the easing of outdoor restrictions.

Yep, and it's a suspect one at best. Personally I wouldn't trust a stastical model from someone who advertises that they own the book "how to lie with statistics" and obviously used it

I trust someone who acknowledges that statistics can be used deceptively vastly more than I trust someone who doesn't.

I mean that book is basically required reading for Statistics 101. Doesn't everyone have a copy?

A comment from a high school math teacher concerning that book (which yes, was commonplace to read even decades ago, even among high school AP students) has stuck with me: "It's definitely possible to lie with statistics, but wait until you see how much you can get away with without them!"

The book "how to lie with statistics" (1954) was written before tags like /s were invented. Referencing "how to lie with statistics" is an indicator that the author is trying to avoid common pitfalls in statistical reasoning.

I own an inherited blue Pelican paperbook copy from my pharmacist grandfather.

https://en.wikipedia.org/wiki/How_to_Lie_with_Statistics

>Personally I wouldn't trust a stastical model from someone who advertises that they own the book "how to lie with statistics" and obviously used it

why?