In the Western countries they said that "cases, hospitalizations and deaths are more among the vaccinated people because the number of the vaccinated is much larger than the number of unvaccinated".
But there are at least 2 countries in the European Union, Romania and Bulgaria, where the vaccinated are still outnumbered by the unvaccinated. These 2 countries can be taken as a control group.
Omicron has just arrived in my country, Romania, and I bet that in few days we'll be able to say loud and powerful: vaccinated people, ergo the vaccines, are spreading the disease. We are waiting for some official data now, I hope the authorities will have enough courage to publish them. But they have already acknowledge that 65% from Omicron cases are among fully vaccinated (while only 40% of the population is vaccinated).
Look forward to any updates you might have from Romania. My officemate in grad school was from Romania.... what was really depressing to me is that not only was she much stronger mathematically than me, but she also scored better on the English portions of the entrance exams 🤣
In short, 45.1% infections were among the vaccinated.
But in Romania only 41,5% from the entire population is vaccinated against Covid.
On a relative base (as cases per 100.000) the vaccine efficacy is now at -1,86%. It can't be hidden anymore (during the Delta wave they could hide this by excepting the vaccinated population from testing).
This should be the end of the game. In conclusion: no, in the Western countries cases, hospitalizations and deaths aren't more among the vaccinated because "the vaccinated are much more than the unvaccinated". They are more among the vaccinated because these so-called "vaccines" seems to play an active role in encouraging infections.
While I have no doubt that the conclusion is correct, I think it's also the true that more densely populated areas would have both more per capita vaccinations and more per capita virus cases without there being any causal link. That said, once we have more time passed and more data collected, I believe the positive association between vaxxing and virus will hold up, without regard to population density.
Thank you for doing this. With no background in statistics at all, I am so grateful to have your work to learn from. Above all, it is such a relief to see someone bring their expertise to dig under the narrative to look for the facts. So I now I'm being greedy here by asking basic questions that will further my poor understanding of statistics, and am just trying my luck in the hope other lay people might also benefit from your answers, should you have the time to answer them.
Unfortunately, I am still unable to understand certain critical points, such as, "R-squared is a measure of how much of the variation in the dependent variable is explained by the variation in the independent variable." Am I correct in understanding, in my plain language, that how much the vax % of an area IMPACTS the case rate of that area?
What does "R" represent? Why is it "Squared"? Has it anything to do with the "R"egression line?
Is the regression line an average of all the case counts and an average of vax rates in a point of time for all data representing listed counties?
I'm going to assume the R is because it's related to "R"egression, but I've never asked that question! See, it takes someone outside of the field sometimes to ask a good & obvious question 😁. Don't worry about the "squared", that is just because the calculation involves a things that are squared... and the reason they are squared is because when we measure "errors" in a regression, we don't want positive valued errors to be offset by negative valued errors (i.e. I estimated this value too high by 100 one time, but too low by 100 another time, so in total my estimate is perfect).... squaring the errors before summing them will eliminate that problem.
R-squared values will range from 0 to 1 (or 0% to 100%). An R-squared of 0 essentially means the two variables look completely unrelated (suppose I tried to predict the average temperature of a city based on the average height of the population). A "best fit" line of data with R-squared =0 would simply be a horizontal line at the average value of all the points. An R-squared of 1 would mean perfect correlation (suppose I "predicted" the number minutes in a day based on the number of hours in a day). A "best fit" line with R-squared =1 would be a line with all the points exactly on the line.
Generally speaking, most things we want to look at are based on many factors. COVID infections might be based on a community's age demographics, population density, previous infections, climate, people per household, etc, etc. We would not expect to be able to explain the variation of infection rates across different areas by only looking at only one of these factors. Therefore, above I contend that for most of the regions, where we can explain >17% and up to 34% of the variation simply based on their vaccination rates, this indicates pretty high correlation. Finally, you will notice that the higher the R-squared value, the more closely the points group around the "best fit" regression line. If the points just like like a random scatter, you will have small value for R-squared and a large p-value, while if the points hover closely to the regression line, that will result in the opposite.
Are you able to force the dashboard to show an arbitrary 2020 date range case rate by the same now-1mo vax rate?
As Maria Romana said I suspect the conclusion is correct but it would be nice to rule out population density or health care access biases. One thing I did try is capping population, using the Southeast view. Even down to 100k and the association is still there. Pretty compelling.
In the Western countries they said that "cases, hospitalizations and deaths are more among the vaccinated people because the number of the vaccinated is much larger than the number of unvaccinated".
But there are at least 2 countries in the European Union, Romania and Bulgaria, where the vaccinated are still outnumbered by the unvaccinated. These 2 countries can be taken as a control group.
Omicron has just arrived in my country, Romania, and I bet that in few days we'll be able to say loud and powerful: vaccinated people, ergo the vaccines, are spreading the disease. We are waiting for some official data now, I hope the authorities will have enough courage to publish them. But they have already acknowledge that 65% from Omicron cases are among fully vaccinated (while only 40% of the population is vaccinated).
Look forward to any updates you might have from Romania. My officemate in grad school was from Romania.... what was really depressing to me is that not only was she much stronger mathematically than me, but she also scored better on the English portions of the entrance exams 🤣
Hi, I have an update. They have published new data on the week 3-9 of January. This is the official document: https://insp.gov.ro/download/CNSCBT/docman-files/Coronavirus%20nCoV/analiza_cazuri_confirmate_covid-19/Raport-saptamanal-EpiSaptamana01_2022.pdf
In short, 45.1% infections were among the vaccinated.
But in Romania only 41,5% from the entire population is vaccinated against Covid.
On a relative base (as cases per 100.000) the vaccine efficacy is now at -1,86%. It can't be hidden anymore (during the Delta wave they could hide this by excepting the vaccinated population from testing).
This should be the end of the game. In conclusion: no, in the Western countries cases, hospitalizations and deaths aren't more among the vaccinated because "the vaccinated are much more than the unvaccinated". They are more among the vaccinated because these so-called "vaccines" seems to play an active role in encouraging infections.
The Omicron Lag and what that means for the UK
https://nakedemperor.substack.com/p/the-omicron-lag
While I have no doubt that the conclusion is correct, I think it's also the true that more densely populated areas would have both more per capita vaccinations and more per capita virus cases without there being any causal link. That said, once we have more time passed and more data collected, I believe the positive association between vaxxing and virus will hold up, without regard to population density.
It occurred to me this morning that when I ran a similar regression several months ago, the regression line for cases vs. vax tended to slope downward. Certainly not dispositive, but it might appear that Omicron has reversed the direction of the correlation. https://inumero.substack.com/p/cases-deaths-cfr-by-vax-rates-and?r=tv61s&utm_campaign=post&utm_medium=web
Thank you for doing this. With no background in statistics at all, I am so grateful to have your work to learn from. Above all, it is such a relief to see someone bring their expertise to dig under the narrative to look for the facts. So I now I'm being greedy here by asking basic questions that will further my poor understanding of statistics, and am just trying my luck in the hope other lay people might also benefit from your answers, should you have the time to answer them.
Unfortunately, I am still unable to understand certain critical points, such as, "R-squared is a measure of how much of the variation in the dependent variable is explained by the variation in the independent variable." Am I correct in understanding, in my plain language, that how much the vax % of an area IMPACTS the case rate of that area?
What does "R" represent? Why is it "Squared"? Has it anything to do with the "R"egression line?
Is the regression line an average of all the case counts and an average of vax rates in a point of time for all data representing listed counties?
I'm going to assume the R is because it's related to "R"egression, but I've never asked that question! See, it takes someone outside of the field sometimes to ask a good & obvious question 😁. Don't worry about the "squared", that is just because the calculation involves a things that are squared... and the reason they are squared is because when we measure "errors" in a regression, we don't want positive valued errors to be offset by negative valued errors (i.e. I estimated this value too high by 100 one time, but too low by 100 another time, so in total my estimate is perfect).... squaring the errors before summing them will eliminate that problem.
R-squared values will range from 0 to 1 (or 0% to 100%). An R-squared of 0 essentially means the two variables look completely unrelated (suppose I tried to predict the average temperature of a city based on the average height of the population). A "best fit" line of data with R-squared =0 would simply be a horizontal line at the average value of all the points. An R-squared of 1 would mean perfect correlation (suppose I "predicted" the number minutes in a day based on the number of hours in a day). A "best fit" line with R-squared =1 would be a line with all the points exactly on the line.
Generally speaking, most things we want to look at are based on many factors. COVID infections might be based on a community's age demographics, population density, previous infections, climate, people per household, etc, etc. We would not expect to be able to explain the variation of infection rates across different areas by only looking at only one of these factors. Therefore, above I contend that for most of the regions, where we can explain >17% and up to 34% of the variation simply based on their vaccination rates, this indicates pretty high correlation. Finally, you will notice that the higher the R-squared value, the more closely the points group around the "best fit" regression line. If the points just like like a random scatter, you will have small value for R-squared and a large p-value, while if the points hover closely to the regression line, that will result in the opposite.
R squared is usually the residual of least squares fitting a curve to random data.
Are you able to force the dashboard to show an arbitrary 2020 date range case rate by the same now-1mo vax rate?
As Maria Romana said I suspect the conclusion is correct but it would be nice to rule out population density or health care access biases. One thing I did try is capping population, using the Southeast view. Even down to 100k and the association is still there. Pretty compelling.
See my response to Maria above... but I can also test your idea if I have time.