BA.2 became the majority strain in the United States this week, as expected. Mostly things are continuing as expected, with unsurprising news on a variety of fronts. The biggest surprise is that the second booster shot was approved. Given I expect to be asked about that every so often, I split that off into
> I am still confused by there being far more people who declined the second shot than the first shot
Seems likely to be a data issue - iirc one major issue is that people who took two shots in different times/places might have been registered as taking the first shot twice, which would artificially raise the one shot numbers (and lower the two shot numbers).
Maybe I don't get involved in enough stats arguments, but the "thinks Bayesianism is subjective and meaningless..." seems like the straw man part to me. I'm a data science/statistician who works almost exclusively in frequentist stats/p-values world, but I don't believe any of those things _nor do I know anyone who does_. I've taken several Bayesian courses and would love to use more Bayesian stuff in my work but am constrained for a variety of reasons.
While I'm sure there exists _someone_ who holds the opinion that EY is taking about there (the world is a big place!), I'm not sure I believe that there are enough/important enough people who do that writing about it is worthwhile.
I also think that "p-values need to be burned down" is an _equally dumb opinion_ to be quite honest. Yes, there are problems with them, and absolutely it is is _incredibly_ easy to misuse them. These are not minor problems, and in certain fields, this misuse (either intentional or not) is rampant. That does not remove the very real value and use that frequentist tools can have when properly used.
While I'm less familiar, I'm not sure I believe that it's impossible to misuse Bayesian stats in similar ways (even if it may be more difficult). If someone snapped their fingers and suddenly all statistics were in a Bayesian framework, I'm not convinced that fraud/misuse would be significantly decreased, just that it would probably be different kinds of misuse that would manifest in different kinds of ways.
When you make an idiot-proof thing, the world just goes and makes a better idiot. Right now, the majority of people who use Bayesian stats are the people who _care enough about the problems to be using Bayesian stats_. The fact that they are using it means that, if they were doing frequentest stats they would likely not be _mis_ using them. If you forced all the incautious, misinformed, or fraudulent people to use Bayesian statistics, then you would get incautious, misinformed, and fraudulent Bayesian statistics.
On the UK numbers - worth consulting the ONS infection survey, if you’re not aware of it. This is a weekly prevalence survey released on Fridays. Since they just test a fixed number of people and extrapolate, it shouldn't be influenced by testing levels etc but obviously with the trade-off that it's a lagging indicator compared to daily case numbers.
Their last results suggested 1 in 16 people in the UK had Covid during that week, with higher rates in some areas (Scotland had 1 in 11 people infected).
> I am still confused by there being far more people who declined the second shot than the first shot
Seems likely to be a data issue - iirc one major issue is that people who took two shots in different times/places might have been registered as taking the first shot twice, which would artificially raise the one shot numbers (and lower the two shot numbers).
Maybe I don't get involved in enough stats arguments, but the "thinks Bayesianism is subjective and meaningless..." seems like the straw man part to me. I'm a data science/statistician who works almost exclusively in frequentist stats/p-values world, but I don't believe any of those things _nor do I know anyone who does_. I've taken several Bayesian courses and would love to use more Bayesian stuff in my work but am constrained for a variety of reasons.
While I'm sure there exists _someone_ who holds the opinion that EY is taking about there (the world is a big place!), I'm not sure I believe that there are enough/important enough people who do that writing about it is worthwhile.
I also think that "p-values need to be burned down" is an _equally dumb opinion_ to be quite honest. Yes, there are problems with them, and absolutely it is is _incredibly_ easy to misuse them. These are not minor problems, and in certain fields, this misuse (either intentional or not) is rampant. That does not remove the very real value and use that frequentist tools can have when properly used.
While I'm less familiar, I'm not sure I believe that it's impossible to misuse Bayesian stats in similar ways (even if it may be more difficult). If someone snapped their fingers and suddenly all statistics were in a Bayesian framework, I'm not convinced that fraud/misuse would be significantly decreased, just that it would probably be different kinds of misuse that would manifest in different kinds of ways.
When you make an idiot-proof thing, the world just goes and makes a better idiot. Right now, the majority of people who use Bayesian stats are the people who _care enough about the problems to be using Bayesian stats_. The fact that they are using it means that, if they were doing frequentest stats they would likely not be _mis_ using them. If you forced all the incautious, misinformed, or fraudulent people to use Bayesian statistics, then you would get incautious, misinformed, and fraudulent Bayesian statistics.
"They decline to note particular commercial devices, but yes they are available."
Got a link?
On the UK numbers - worth consulting the ONS infection survey, if you’re not aware of it. This is a weekly prevalence survey released on Fridays. Since they just test a fixed number of people and extrapolate, it shouldn't be influenced by testing levels etc but obviously with the trade-off that it's a lagging indicator compared to daily case numbers.
Their last results suggested 1 in 16 people in the UK had Covid during that week, with higher rates in some areas (Scotland had 1 in 11 people infected).