Years ago, a (statistical) colleague advised me that, if it is necessary to conduct a test to tell whether or not a relationship is significant, then it isn't significant. These scatter diagrams are a good example of that: when R-squared is only around 1-2%, it is clear that the vaccinations (including boosters) are ineffective.
It certainly is. Much higher. Too high, really. I wonder if that's because states with high vax rates are more likely to have lots of people doing lots of (self-)testing? Are the cases PCR-certified or RATs? [Here in Australia, where cases are still pretty high, most ( >70%) of the cases are a result of self-administered and self-reported RATs. I suspect that there is a lot of testing as people are obliged to confirm a negative test for some purposes, not only because of anxiety or caution. Idle speculation, not data ... but no data available on how many tests are actually conducted, just how many are positive, so case rates are unknown and unknowable.]
A good point, although arguments by analogy can sometimes be problematic. A statistical significance test is a different kind of test from a medical test. Some medical conditions (eg some cancers) may be detectable via a test even when people do not (yet) feel ill; that’s not to say those tests are entirely reliable - in fact, they’re not - but they do suggest some caution with the analogy.
Thanks! It's great to see this official data, even though it's very sad too.
If only people had started eating real, unprocessed, unsweetened food early in 2020, we might have had half the Covid deaths. The side-effects of eating real food are overwhelmingly positive. We might even see all-cause deaths drop, even with Covid and monkeypox...
I looked at Missouri. The urban counties are the high death rate counties. There are higher percentages of people of color in urban counties than rural in most states. Color strongly correlates with vitamin D deficiency.
Vitamin D deficiency seems to correlate strongly with covid mortality.
It is al Danish check for graph in the middle "intesivlaegelser". Green line are boostered people.
At firs glance this seems to contradict your findings here.
But it appears that the boostered people do get an infection easier. Also visible in the Danish data. Common explation ist that thes test more often than unvaccinated or double vaxed, also in the Danish data to see.
Could it be that boostered do get an infection easier than double vaxed an unvaxed but do have better outcome? If the increase in infection rate is higher than the protetcetion for hospitalisation both findings would match.
I guess this is right, please check at ourworldindata the current surge of Covid Cases in Portugal. They are heavily vaxxed and boostered.
If correct that would be the thing that Geert VanBosche predicted earlier. The vaxed become a risk for the unvexed.
Denmark and Portugal do have a bit more then 20% share of positive PCR Tests. A bit to much for an artifact. It seems that C19 is not as deadly anymore. But we used to have much fewer cases at this time in the last year.
It’s is often interesting with scattered data like this to look at the upper bounds .. rather than the line of best fit.. for the boosted the upper bounds of hospitalisation rate ie potential risk.. is doubled when you go over 40%...
Hi Joel, isnt there a time from injection factor as well to how effective the vaxx is? From what I have read it seems the efficacy wears off over time, and seems to wear of quicker with each additional dose. No idea how you would take this into account in what you have done
I don't get why we are receiving opposing reports when analysing the data for vaccination status and hospitalisations/deaths. I get that this is covid hosp specifically, but I don't know why we get two opposing results from the officials and a lot of top data analysts. Confusing!
The UK ONS is saying it's mainly the opposite for ASMRs: "Monthly age-standardised mortality rates (ASMRs) for deaths involving coronavirus (COVID-19) have been consistently lower for all months since September 2021 for people who had received a third dose or booster at least 21 days ago, compared with unvaccinated people and those with just a first or second dose."
how do you calculate the t-stats? wouldn't it just be the coefficient/standard error of coefficient? r-squared looks very low. not coming here to troll you. just curious. I'm studying quantitative analysis right now for my cfa. I don't support the jabs.
Yes, visually comparing the maps confirms some correlation between the disease and the jabbing levels, but there are significant inconsistency regions, e.g. in Florida and Texas. It would be interesting to try and figure out why...
Am I reading these maps correctly? I see a lot of areas in NY, NJ and FL that have very high uptake of dose two, but zero of the booster?? How does that make sense? Colleges alone (in NY and NJ) would be impacting those areas wouldn’t they? Are most students getting exemptions? I’m blown away to see this.
The VT data looks wrong in Dose 2 rate. They are more vaccinated than NH (I’m from NH). They boast 100% rate in over 65s. I’m curious how states do hospitalization rates as well. For instance, Dartmouth Hitchcock is a large teaching hospital on the VT border and gets lots of VT patients. It also pulls from several rural counties, not just Grafton County where it’s located. Does the hospitalization go to Grafton County or the county of the patient? NH is also monkeying with the data - only calling a covid patient a “Covid hospitalization “ if patient is on RemDeathIsNear or dexamethasone.
I think I ran into this problem in doing some analysis for VT. They seem to report vaccinations at state level for some period of time, then allocate to counties after some delay. I suspect recent vaccines for VT are in the “unassigned” county.
Huh. The first map of 2-dose rate stuck out to me as NH was darker (more vaccinated) than VT, but that’s not the case. The booster map coloration look better. I hate socialized , nationalized medicine and vax data collection mandates, but it is nice to see the UK data in all its naked glory when using it for the desired purpose!
Thanks for the updates.
Years ago, a (statistical) colleague advised me that, if it is necessary to conduct a test to tell whether or not a relationship is significant, then it isn't significant. These scatter diagrams are a good example of that: when R-squared is only around 1-2%, it is clear that the vaccinations (including boosters) are ineffective.
Ineffective *at best.
The R-squared for cases is much higher, https://inumero.substack.com/p/vaccines-and-cases-a-look-at-us-states?r=tv61s&s=w&utm_campaign=post&utm_medium=web
It certainly is. Much higher. Too high, really. I wonder if that's because states with high vax rates are more likely to have lots of people doing lots of (self-)testing? Are the cases PCR-certified or RATs? [Here in Australia, where cases are still pretty high, most ( >70%) of the cases are a result of self-administered and self-reported RATs. I suspect that there is a lot of testing as people are obliged to confirm a negative test for some purposes, not only because of anxiety or caution. Idle speculation, not data ... but no data available on how many tests are actually conducted, just how many are positive, so case rates are unknown and unknowable.]
For lots of illnesses "if it is necessary to conduct a test to tell whether or not" you are ill then it isn't very significant either.
A good point, although arguments by analogy can sometimes be problematic. A statistical significance test is a different kind of test from a medical test. Some medical conditions (eg some cancers) may be detectable via a test even when people do not (yet) feel ill; that’s not to say those tests are entirely reliable - in fact, they’re not - but they do suggest some caution with the analogy.
Please do an analysis of rural v. urban counties.
Thanks for the data.
Obesity and vitamin D deficiency correlate, as does metabolic syndrome with obesity.
https://onlinelibrary.wiley.com/doi/abs/10.1111/obr.12239
Thanks! It's great to see this official data, even though it's very sad too.
If only people had started eating real, unprocessed, unsweetened food early in 2020, we might have had half the Covid deaths. The side-effects of eating real food are overwhelmingly positive. We might even see all-cause deaths drop, even with Covid and monkeypox...
The big gray area is Jesus saying no to vaccines.
Not sure if we're using the exact same data sets, but you can see the same thing on the "Key Metrics Vax by County" tab on my dashboards: https://public.tableau.com/views/CovidDashboardsbyTCoddington/KeyMetricsVaxbyCounty?:language=en-US&publish=yes&:display_count=n&:origin=viz_share_link
I looked at Missouri. The urban counties are the high death rate counties. There are higher percentages of people of color in urban counties than rural in most states. Color strongly correlates with vitamin D deficiency.
Vitamin D deficiency seems to correlate strongly with covid mortality.
Posting as a resource for anyone... updated weekly, lots of ability to filter, choose different metrics, etc.
Thanks for the Public Tableau. Nice job.
I’ve touched on quite a few of these in some substack posts. Take a look and leave comments if there’s additional interesting questions unanswered.
I have a question. There is good data available from Denmark that still shows that boostered people do have a lower chance to get into hospital.
https://covid19danmark.dk/
It is al Danish check for graph in the middle "intesivlaegelser". Green line are boostered people.
At firs glance this seems to contradict your findings here.
But it appears that the boostered people do get an infection easier. Also visible in the Danish data. Common explation ist that thes test more often than unvaccinated or double vaxed, also in the Danish data to see.
Could it be that boostered do get an infection easier than double vaxed an unvaxed but do have better outcome? If the increase in infection rate is higher than the protetcetion for hospitalisation both findings would match.
I guess this is right, please check at ourworldindata the current surge of Covid Cases in Portugal. They are heavily vaxxed and boostered.
If correct that would be the thing that Geert VanBosche predicted earlier. The vaxed become a risk for the unvexed.
Denmark and Portugal do have a bit more then 20% share of positive PCR Tests. A bit to much for an artifact. It seems that C19 is not as deadly anymore. But we used to have much fewer cases at this time in the last year.
Also very strange in the moment is South Korea.
3 snots for kids. Wow will denial continue as we lose our kids to odd health issues?
It’s is often interesting with scattered data like this to look at the upper bounds .. rather than the line of best fit.. for the boosted the upper bounds of hospitalisation rate ie potential risk.. is doubled when you go over 40%...
that’s significant
Hi Joel, isnt there a time from injection factor as well to how effective the vaxx is? From what I have read it seems the efficacy wears off over time, and seems to wear of quicker with each additional dose. No idea how you would take this into account in what you have done
Joel, is it possible from the data to tell if hospitalizations lead or lag the vaccinations?
I don't get why we are receiving opposing reports when analysing the data for vaccination status and hospitalisations/deaths. I get that this is covid hosp specifically, but I don't know why we get two opposing results from the officials and a lot of top data analysts. Confusing!
The UK ONS is saying it's mainly the opposite for ASMRs: "Monthly age-standardised mortality rates (ASMRs) for deaths involving coronavirus (COVID-19) have been consistently lower for all months since September 2021 for people who had received a third dose or booster at least 21 days ago, compared with unvaccinated people and those with just a first or second dose."
https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsinvolvingcovid19byvaccinationstatusengland/deathsoccurringbetween1january2021and31march2022
Where is this data from? I neither believe nor disbelieve it until I can confirm it.
how do you calculate the t-stats? wouldn't it just be the coefficient/standard error of coefficient? r-squared looks very low. not coming here to troll you. just curious. I'm studying quantitative analysis right now for my cfa. I don't support the jabs.
Yes, visually comparing the maps confirms some correlation between the disease and the jabbing levels, but there are significant inconsistency regions, e.g. in Florida and Texas. It would be interesting to try and figure out why...
Thanks!
The first question "they" will ask is: why 24 Feb 2022?
Am I reading these maps correctly? I see a lot of areas in NY, NJ and FL that have very high uptake of dose two, but zero of the booster?? How does that make sense? Colleges alone (in NY and NJ) would be impacting those areas wouldn’t they? Are most students getting exemptions? I’m blown away to see this.
The VT data looks wrong in Dose 2 rate. They are more vaccinated than NH (I’m from NH). They boast 100% rate in over 65s. I’m curious how states do hospitalization rates as well. For instance, Dartmouth Hitchcock is a large teaching hospital on the VT border and gets lots of VT patients. It also pulls from several rural counties, not just Grafton County where it’s located. Does the hospitalization go to Grafton County or the county of the patient? NH is also monkeying with the data - only calling a covid patient a “Covid hospitalization “ if patient is on RemDeathIsNear or dexamethasone.
I think I ran into this problem in doing some analysis for VT. They seem to report vaccinations at state level for some period of time, then allocate to counties after some delay. I suspect recent vaccines for VT are in the “unassigned” county.
Huh. The first map of 2-dose rate stuck out to me as NH was darker (more vaccinated) than VT, but that’s not the case. The booster map coloration look better. I hate socialized , nationalized medicine and vax data collection mandates, but it is nice to see the UK data in all its naked glory when using it for the desired purpose!