I hope you're right, but did you read that Twitter thread? That's exactly what he's talking about - if you look at the national data as a single aggregate, then you'll get the misleading idea that the decline in deaths is meaningful, whereas if you look at the situation at a local level we're about to see NYC-scale crises in dozens of locations around the country with a surge in deaths over the next eight weeks.The US is being fairly reckless, but it may not be as bad as that. The case surge is more than two weeks old by this point, but the overall trend in the number of new deaths is still downward.
Arizona coronavirus patients will 'be given a score based on life expectancy and underlying conditions' to determine whether or not they're put on a ventilator as hospitals near capacity in several states
Hospitals in several states where coronavirus cases are surging such as Arizona, Mississippi and Texas are pushing capacity
In Arizona, more than 3,000 people are hospitalized and ICU capacity reached 91%
If Arizona hospitals push past capacity, patients will be given a score to determine whether or not they're placed on a ventilator
Texas Medical Center hospitals announced that their ICU capacity surpassed 100%, requiring new and converted beds to be opened
In Mississippi, more than 70% of ICU beds are occupied and coronavirus cases topped 1,000 for the first time since the pandemic began
.
It seems that deaths are being grossly under-reported...
I saw this sign at Lowe's this morning, I didn't know about this, anyone else see a similar sign around?...
View attachment 10147
Yes. I saw one at a Home Depot the other day.
There's a nationwide shortage of coins because of the pandemic - CNN
I do recall your posts about the importance of highly localized analyses, so I thought you might find Beckett's analysis interesting. There are many factors at work here so it's easy to draw the wrong conclusions (such as "we've gotten much better at saving the lives of COVID patients" and/or "the virus suddenly and inexplicably isn't as deadly"); that's how I interpreted his reference to Simpson's paradox. I've also heard that more younger people are getting infected right now (not sure if that's true) than previously so that also complicates an effort to understand what's happening, and to project outcomes.I think Miles Beckett on Twitter is misapplying Simpson's paradox. That paradox is to do with proportions as evidenced in contingency tables, where one can infer contradictory causal relationships by grouping data in different ways.
Here we are dealing with simple counts, i.e. is the count of the number of dead recorded each day increasing along with the increase in the number of people testing positive? He mentions that the US epidemic consists of lots of small outbreaks in every locality in the US, and I agree with this. You may remember that I made the same point a few times back in March and April in this thread. The national trend does not necessarily tell you what is going on in a locality, but this is not Simpson's paradox.
The current increase in US cases is being driven primarily by outbreaks in California, Arizona, Texas and Florida. However, even when we look at the death data for these states we see the following (this is the smoothed data with a seven-day moving point average):
Arizona:
View attachment 10150
California:
View attachment 10149
Florida:
View attachment 10151
Texas:
View attachment 10152
This compares with New York state like this:
View attachment 10153
New York had a period of rapid exponential growth, and reached a peak in deaths in just over three weeks from the first recorded death.
In the four states of concern, one is recording no increase in the number of daily dead (California), the others are trending upward but more slowly than New York and the numbers are far smaller. The situation could deteriorate, but there does not seem to be a New York-scale disaster yet unfolding in any of these states.
So, although national statistics can obscure what is going on in a locality, the increase in the positive tests is visible in the bulk data, but there is no concurrent increase in the number of daily dead. I think there is more virus about, as the increase in positive tests is only partially attributable to increased testing. However, it also appears that the virus is not as deadly, or not spreading as it was in New York in March.
Friends in Florida reported the same thing.I wonder if the country has gone mad. I read that there have been Covid parties where younger people are trying to catch the virus. If you told me that the end of the world was in sight, I would have to believe you.