Are you a data scientist? I need you to peer review our work ASAP
stevekirsch.substack.com
Is my latest survey GAME OVER for the vaccine narrative? I'm pretty sure it is because ANYONE can replicate the results in 24 hours and no "judgement" or "government data" is used. Is it bulletproof?
The cancers that result from the covid vaxxes are incubating and brewing now. Learn how to protect yourself and loved ones from cancer with fenbendazole, a safe, inexpensive, off patent, otc drug. So far it saved two people (make that THREE) close to me. Read the detailed Case Reports https://fenbendazole.substack.com
Breast, colon, prostate, glioma, lung, it doesn’t seem to matter, fenbendazole seems to eradicate them all.
I put in "disagree on both" -- but just because IMO the situation is more complex and nuanced and there's a missing option like "I agree with the spirit, but not with the letter."
First and foremost, in my opinion, Joel's analysis is plausible and Steve's data are a great attempt at getting insights into the real situation.
HHS and public health agencies worldwide should be doing this.
However, the analysis per se is not "devastating" in the slightest.
Let's look at it from the eyes of an average "highly educated scientist":
1) The dataset is biased, as many pointed out. This alone kills the whole thing.
2) The first step, the normalization is not supported by hard evidence, and thus is a conjecture.
3) The step of removing "Fauci protocol" deaths is not supported by hard evidence, and thus is a conjecture.
4) The rest of the analysis is just an attempt to explain the trends, which is not a proof of anything.
To summarize, in my opinion, Joel's analysis is a good, plausible stab at making sense out of a flawed dataset.
The dataset is too small and biased to make any statistically sound conclusions -- at this point.
Without those, this might as well have been a fart in the wind -- as much as I am grateful to both Steve and Joel for taking the time and spending this effort.
That being said, in a normal situation with sane people in science and in office this would've been a Nature "ahead of print" brief and an immediate cause for concern and a wider worldwide investigation.
But then again, VAERS alone should've been sufficient LONG time ago.
All that being said, this is yet another formidable brick in the foundation of the future gallows.
I am a biophysicists, coming from Physics, Mathematics and Computer Science. That's my background. The real problem with your data is, as you said:
"On December 15, 2022, I launched an all-cause mortality survey which asked a simple objective question of my followers"
Your population cohort is biased and you need an unbiased sample. There is an additional problem with that. Even if you do a random sampling within the general public of any given country or region, with enough representation of its constituents, COVID 19 has become a very powerful psychological phenomena and your sample would still be strongly biased.
Therefore you absolutely need to do what Joel has done, which is to compare the survey data with official data and check if it is consistent. In the case that you would obtain a sample as unbiased as possible, still you would only be able to tell that there are inconsistencies in relation to the official data and that further analysis need to be perform. You will never be able to prove that the official data is incorrect with a survey, you absolutely need further analysis, which is exactly what the government of Florida is trying to do. You need access to the raw data.
A big limitation of the survey is that it is only your readers. A truly random survey is needed. I was surprised at how well the survey lined up with the real-world data, but I think that is because most people here are seeking truth and although may be somewhat biased to point blame at the vaccine for deaths and promote therapeutics, they are much more honest than those on the other side of those arguments. However, the fact that the survey is only your readers will be used against you. I strongly suggest engaging with a professional polling organization to conduct a truly random survey to add credibility. It will likely not show as strong an argument, but it will still show that the assertions about vaccine safety, therapeutics, and vaccine effectiveness you have made have validity. It is clear from publicly available data from the CDC, World of Data, and other sources that excess deaths are correlated with vaccine rollout, that the numbers of deaths and injuries from the vaccines are greatly under-stated and that therapeutics and correct health care significantly reduces the likelihood of getting Covid-19 and reducing the death rate.
Good analysis by Joel esp. the last graph separating out deaths without the murderous fauci protocol. A few other thoughts to share:
1. there should be some observations/data highlights/analyses for each graph, cross-references and overall analysis - this will help inform users who may not be able to correctly interpret the data/graphs themselves or have the time to do so. Joel walked through the charts in the video but explaining in a separate video is not ideal.
2. Charts are not numbered, e.g. Chart 1: Kirsch Survey Results - numbered charts will make it much easier to reference
3. Some highlighted data points (e.g. in Chart 1 & 2) have no context or reason (also see #1 above).
4. It'd be helpful to mark key events with dates in those monthly charts, e.g. vaccine roll-out (MM-DD-2021), Booster 1 roll-out [date]. Those milestone events are typically marked as a dotted vertical line crossing the date on X.
5. Data source is not clearly labelled, e.g., Chart 2 "Monthly Excess Covid" didn't specified it was based on CDC data.
6. Re: Chart 3:
6.1 - I heard personally that there were a wave of deaths (relatives of my coworkers) within a few months after the jab roll-out , which matches/validates the increases of both deaths and jab uptake during Mar-Jul 2021.
6.2 - The slight drop in jabbed deaths vs. higher accumulated vaccination rates between Aug-2021 to Feb-2022 may be explained that after the immediate wave of deaths as a direct result of the jab, jabbed population who survived the immediate death wave were then suffering from long term effects (e.g. critical illnesses, cancer) which were not instantly lethal.
6.3 - I believe there is value in cross referencing data on jabbed population developing critical illness post jab - this will provide a more wholesome pic of its total devastating effects
7. Re: Chart 4:
7.1 - According to the video, the 'Ever Vaccinated Population' line was normalized (vs. the same line in Chart 3) but it was not labelled as such
7.2 - the author seems to attributing the slowing deaths after July 2021 to "Younger, healthier vaccinees" - I personally don't agree unless the time period coincided with the jab roll-out to younger demographics (see also #4 above).
7.3 - I also don't see the point/significance of highlighting the convergence of the black line, i.e. natural percentage, and Normalized red line, i.e. a normalized percentage line. Maybe it's because I don't know normalized lines well.
7.4 - I don't understand why the 3rd label "Expected Proportion of deaths that are vaccinated people" was dubbed as "safe".
8. Re: Chart 5 - a very busy chart (see #1 above)
8.1 - it seems pretty clear that the jab was not safe particularly between Mar - Jul 2021 so no ? needed for 'Not Safe'.
8.2 - It seems that the jab was not effective even after July 2021 with deaths slowing down [a period referred to as "Effective?" in a green box on the chart]- the context/rationale would lie in the dotted blue line "Cumulative Covid Deaths', which trended upwards post July 2021; it was very steady before then. So It was definitely not effective.
9. Lastly, it can be very difficult to determine deaths "of covid" or "with covid" to effectively differentiate the two; such vagueness may apply to both survey and public/cdc data which in turn will have an impact on any analysis. There have been known reports that CDC has been purposefully allocating all/most deaths "with covid" as "of covid" to ramp up the fear porn.
10. It would also be beneficial to have some year-over-year all-cause mortality data/analysis to provide further context.
Different topic. I just found out that Amazon will not honor Robert Malones free Kindle offer for his new book, Lies My Government told Me. Basically it's a scam and Amazon charge the full price.
I have suggested many times to you that your playing the wrong data game. Enough with the defense !
(What's already happened with adverse effects & death.)
Please consider OFFENSE. Play their game but better. Use these floating millions $$$ you have and with others, coordinate a real-time blood assay by getting, say 5,000 random people to submit a blood sample for analysis. Use a free Vit-D test as an incentive.
Contract with an Accredited Lab to perform the following tests.
HEART ATTACKS / MYOCARDITIS
- Troponin
- d-Dimer
- C-Reactive Protein
- Platelet Panel
COMPROMISE IMMUNE SYSTEM / CANCERS
CD8
CD4
Analyze this real-time data by the best statistical minds. The major Cohort comparisons. Compare the real-time test values vs. normal test ranges. Better are those with a blood history baseline over years to compare against.
- Vaxxed (2x, 3x, 4x)
- UnVaxxed
- Gender
- Age Stratified
- Shedding in the UnVaxxed
If you want to FULL-STOP these vaccines, just show the results (blood marker message / damage) of the walking to be dead. Or anyone who wants to walk in to test and validate their risk.
This is how you Red Pill the current Blue Pills who have had the Jabs (2-3X) and feel nothing is wrong with them and other family members after more than 18 months out from being vaxxed. You have to scare them with science & reality.
Covid is a blood disease.
Blood has special specific markers to highlight these organ / system damages. Would only take a month to complete with organization.
Please consider playing OFFENSE Steve.
We are running out of time.
We lose the infants & children, it is all over for us.
This video is taped prior to Today's ruling of the judge that resulted in quashing 8 of 10 claims of Plaintiff Lake's charges against Sec. of State / Defendant Hobbs, which now consists of two issues raised the judge will allow to proceed to trial - Lake only needs one to prevail and Hobbs is out.
Per atty. Renz 59% of the Maricopa county election printers were not working the day of the election - that should be all that's needed; odds of more than 1% out during day of election is suspicious, 10% not working is impossible. >20% is sabotage beyond a shadow of doubt.
What percent of Steve's readers are in favor of mass covid vaccination? I bet is is much lower than 30%, which should tell you something when your own base is calling you out.
Just as it is important for the 30% to describe their criticisms, it is equally important to describe strengths. What have the other 70% of readers said about the specific strengths of this analysis, relative to other analyses? They are equally not so active in the comments. Truth be told, 90% of commenters have no clue what they are talking about, in either side of the fence. Sorry if I am in a frustrated mood today.
I greatly appreciate your work Steve, but I am far from convinced by this dataset, as presented in Joel Smalley's video. Here are several problems which spring to mind, given time I may think of more:
1. The assumption that in the absence of Covid deaths, percentages of vaccinated vs. unvaccinated deaths should align with percentages of each category in the population needs a detailed justification. Smalley just vaguely refers to inaccuracy of census data and to UK survey data indicating that official stats overstate % of population vaccinated. Since this initial "normalisation" of the raw data is absolutely key to his entire analysis, such vague hand-waving is totally unsatisfactory. For example, one could argue that, if vaccines reduce Covid illness then they will also reduce non-Covid mortality over time by leaving those who are vaccinated in a generally more healthy state.
2. Another possible explanation for why the percentage of unvaccinated deaths in the raw data exceeds the percentage of non-vaccinees in the population (after Aug 21) is that your readers, and hence also the people they know well, are more likely to be unvaccinated than the average in the population, hence more likely to know people who died who were unvaccinated.
This is just an example of how your dataset might not be a representative sample of deaths in the general population. There are other indications of such bias, for example Covid deaths are signficantly overrepresented in your raw data. Let's take 2021, where you have a full year of data. From Smalley's video (0:26ff), the total number of deaths per month is (approx., I'm estimating from the figure)
26, 20, 34, 26, 28, 29, 35, 40, 61, 49, 57, 75
Total for 2021: 480.
For Covid deaths (18:00ff) the numbers are
8,5,4,9,9,1,8,12,15,9,12,19
Total for 2021: 111
So 111/480 = 23,1% of reported deaths in the raw data are Covid deaths.
Now, according to Worldometers, the total number of US Covid deaths were as follows:
there were approx. 3,458,697 deaths in the US in 2021.
Thus 476,348/3,458,697 = 13,8% of all deaths were Covid deaths.
So Covid deaths are overrepresented by a factor of nearly 2 in your sample. We don't have a full year of data for 2022, but it seems to be a similar picture there.
So your dataset simply doesn't seem to be close to a representative sample of all deaths. It's perhaps not surprising: you asked respondents to submit the death the details of which they recalled best. And given the high profile of Covid, people are more likely to recall the details of deaths that involved Covid.
3. Smalley explains the spike in unvaccinated deaths in late summer 2021 as due to a kind of "survivor bias". But survivor bias could also explain the slow increase in the % of vaccinated deaths after March 2022, that it over time in general. Imagine for simplicity that there are only 2 kinds of people, healthy and unhealthy. The latter are more likely to die in any given time interval. Suppose now you have a vaccine with some positive efficiency, which is taken by the same percantage of healthy as unhealthy people. As time goes on, the percentage of deaths that are vaccinees will increase, not because the vaccine loses efficacy but simply because, amongst the unhealthy, the unvaccinated are dying off faster.
4. I don't think Smalley explains where he got his curve for the percentage of the population vaccinated over time. But his curve doesn't seem to align with, for example, the curve from Our World in Data.
There, for example, it says that as of March 31, 2022 the % who received at least one dose was 77.04%, not the 84% in Smalley's video. (And the % fully vaccinated on that date was only 66.15%).
Summary: Of the points above, #1 is most fundamental. If that "normalisation" could be properly justified, then maybe the dataset is then worth analysing further. Indeed, the overrepresentation of Covid deaths in the raw data should a priori strengthen a "not safe and effective" conclusion, assuming that the vaccines may prevent some Covid deaths but at the expense of deaths from adverse reactions. On the other hand, one would expect the % of deaths amongst vaccinees to be a priori higher than in the population at large, because vaccination rates are higher amongst the elderly and unhealthy. So this would require an extra normalisation. Your data may also be signficiantly non-representative in many other ways I haven't thought of. In short, there are so many problems I am not sure there is much valuable information to be extracted here.
Where did the data come from? It came from a self selected sample from your substack. Is that a random sample of 'western' folk? Of course it is not.
Now the interesting thing is it may well be 'close enough for government work' as we used to say. But in the world of proofs, beyond debate? It is not. "lipstick on a pig" comes to mind.
Sorry man, how it is.
EDIT I should say. I spent most of a career collecting and analyzing messy biologic data, some collected by industry to support their suppositions about what a 'sustainable' harvest level is. Lots of money involved, and the fate of this or that lumber mill. The first question us government gballs would ask as we sat around the table was "where did this data come from? What were the rules for collection?" because that would color everything else. I have a lot of experience in that realm.
Where did this data come from? How was it collected? What potential or real bias is associated with it? First step.
BRIEFLY, HISTORY OF MEDICINE -
From here you reach further articles to understand how not to get sick, possibly how to heal. I also recommend reading James 5 to ask GOD.
https://andreasalvatorebuffa.substack.com/p/briefly-history-of-medicine
The cancers that result from the covid vaxxes are incubating and brewing now. Learn how to protect yourself and loved ones from cancer with fenbendazole, a safe, inexpensive, off patent, otc drug. So far it saved two people (make that THREE) close to me. Read the detailed Case Reports https://fenbendazole.substack.com
Breast, colon, prostate, glioma, lung, it doesn’t seem to matter, fenbendazole seems to eradicate them all.
Scientific cure without big pharma - Stop covid and cancer
http://stopcovid.canalblog.com/archives/2022/12/21/39753747.html
PUBLIC COMPLAINT AGAINST GOVERNMENTS 01/12/2022 - Action, legal action
http://legalaction.canalblog.com/archives/2022/12/02/39730603.html
F mn Espionage NATO, the revolutionary engine -
http://legalaction.canalblog.com/archives/2022/12/21/39753168.html
Best regards,
Andrea Salvatore Buffa
I put in "disagree on both" -- but just because IMO the situation is more complex and nuanced and there's a missing option like "I agree with the spirit, but not with the letter."
First and foremost, in my opinion, Joel's analysis is plausible and Steve's data are a great attempt at getting insights into the real situation.
HHS and public health agencies worldwide should be doing this.
However, the analysis per se is not "devastating" in the slightest.
Let's look at it from the eyes of an average "highly educated scientist":
1) The dataset is biased, as many pointed out. This alone kills the whole thing.
2) The first step, the normalization is not supported by hard evidence, and thus is a conjecture.
3) The step of removing "Fauci protocol" deaths is not supported by hard evidence, and thus is a conjecture.
4) The rest of the analysis is just an attempt to explain the trends, which is not a proof of anything.
To summarize, in my opinion, Joel's analysis is a good, plausible stab at making sense out of a flawed dataset.
The dataset is too small and biased to make any statistically sound conclusions -- at this point.
Without those, this might as well have been a fart in the wind -- as much as I am grateful to both Steve and Joel for taking the time and spending this effort.
That being said, in a normal situation with sane people in science and in office this would've been a Nature "ahead of print" brief and an immediate cause for concern and a wider worldwide investigation.
But then again, VAERS alone should've been sufficient LONG time ago.
All that being said, this is yet another formidable brick in the foundation of the future gallows.
Hi Steve,
I am a biophysicists, coming from Physics, Mathematics and Computer Science. That's my background. The real problem with your data is, as you said:
"On December 15, 2022, I launched an all-cause mortality survey which asked a simple objective question of my followers"
Your population cohort is biased and you need an unbiased sample. There is an additional problem with that. Even if you do a random sampling within the general public of any given country or region, with enough representation of its constituents, COVID 19 has become a very powerful psychological phenomena and your sample would still be strongly biased.
Therefore you absolutely need to do what Joel has done, which is to compare the survey data with official data and check if it is consistent. In the case that you would obtain a sample as unbiased as possible, still you would only be able to tell that there are inconsistencies in relation to the official data and that further analysis need to be perform. You will never be able to prove that the official data is incorrect with a survey, you absolutely need further analysis, which is exactly what the government of Florida is trying to do. You need access to the raw data.
A big limitation of the survey is that it is only your readers. A truly random survey is needed. I was surprised at how well the survey lined up with the real-world data, but I think that is because most people here are seeking truth and although may be somewhat biased to point blame at the vaccine for deaths and promote therapeutics, they are much more honest than those on the other side of those arguments. However, the fact that the survey is only your readers will be used against you. I strongly suggest engaging with a professional polling organization to conduct a truly random survey to add credibility. It will likely not show as strong an argument, but it will still show that the assertions about vaccine safety, therapeutics, and vaccine effectiveness you have made have validity. It is clear from publicly available data from the CDC, World of Data, and other sources that excess deaths are correlated with vaccine rollout, that the numbers of deaths and injuries from the vaccines are greatly under-stated and that therapeutics and correct health care significantly reduces the likelihood of getting Covid-19 and reducing the death rate.
Good analysis by Joel esp. the last graph separating out deaths without the murderous fauci protocol. A few other thoughts to share:
1. there should be some observations/data highlights/analyses for each graph, cross-references and overall analysis - this will help inform users who may not be able to correctly interpret the data/graphs themselves or have the time to do so. Joel walked through the charts in the video but explaining in a separate video is not ideal.
2. Charts are not numbered, e.g. Chart 1: Kirsch Survey Results - numbered charts will make it much easier to reference
3. Some highlighted data points (e.g. in Chart 1 & 2) have no context or reason (also see #1 above).
4. It'd be helpful to mark key events with dates in those monthly charts, e.g. vaccine roll-out (MM-DD-2021), Booster 1 roll-out [date]. Those milestone events are typically marked as a dotted vertical line crossing the date on X.
5. Data source is not clearly labelled, e.g., Chart 2 "Monthly Excess Covid" didn't specified it was based on CDC data.
6. Re: Chart 3:
6.1 - I heard personally that there were a wave of deaths (relatives of my coworkers) within a few months after the jab roll-out , which matches/validates the increases of both deaths and jab uptake during Mar-Jul 2021.
6.2 - The slight drop in jabbed deaths vs. higher accumulated vaccination rates between Aug-2021 to Feb-2022 may be explained that after the immediate wave of deaths as a direct result of the jab, jabbed population who survived the immediate death wave were then suffering from long term effects (e.g. critical illnesses, cancer) which were not instantly lethal.
6.3 - I believe there is value in cross referencing data on jabbed population developing critical illness post jab - this will provide a more wholesome pic of its total devastating effects
7. Re: Chart 4:
7.1 - According to the video, the 'Ever Vaccinated Population' line was normalized (vs. the same line in Chart 3) but it was not labelled as such
7.2 - the author seems to attributing the slowing deaths after July 2021 to "Younger, healthier vaccinees" - I personally don't agree unless the time period coincided with the jab roll-out to younger demographics (see also #4 above).
7.3 - I also don't see the point/significance of highlighting the convergence of the black line, i.e. natural percentage, and Normalized red line, i.e. a normalized percentage line. Maybe it's because I don't know normalized lines well.
7.4 - I don't understand why the 3rd label "Expected Proportion of deaths that are vaccinated people" was dubbed as "safe".
8. Re: Chart 5 - a very busy chart (see #1 above)
8.1 - it seems pretty clear that the jab was not safe particularly between Mar - Jul 2021 so no ? needed for 'Not Safe'.
8.2 - It seems that the jab was not effective even after July 2021 with deaths slowing down [a period referred to as "Effective?" in a green box on the chart]- the context/rationale would lie in the dotted blue line "Cumulative Covid Deaths', which trended upwards post July 2021; it was very steady before then. So It was definitely not effective.
9. Lastly, it can be very difficult to determine deaths "of covid" or "with covid" to effectively differentiate the two; such vagueness may apply to both survey and public/cdc data which in turn will have an impact on any analysis. There have been known reports that CDC has been purposefully allocating all/most deaths "with covid" as "of covid" to ramp up the fear porn.
10. It would also be beneficial to have some year-over-year all-cause mortality data/analysis to provide further context.
We evidently have quite a few allies who very likely have the needed leverage👍.....
LifeSite News headline: "Time we move on’: Half of all US governors urge Biden to end national COVID emergency"
https://daily.news.lifesitenews.com/link.php?AGENCY=LIFE&M=3326938&N=76325&L=38138&F=H&drurl=aHR0cHM6Ly93d3cubGlmZXNpdGVuZXdzLmNvbS9uZXdzL3RpbWUtd2UtbW92ZS1vbi1oYWxmLW9mLWFsbC11cy1nb3Zlcm5vcnMtdXJnZS1iaWRlbi10by1lbmQtbmF0aW9uYWwtY292aWQtZW1lcmdlbmN5Lz91dG1fc291cmNlPWRhaWx5LXVzYS0yMDIyLTEyLTIxJnV0bV9tZWRpdW09ZW1haWwmdXRtX2NvbnRlbnQ9Q016TXlOamt6T0ElM0QlM0Q=
Different topic. I just found out that Amazon will not honor Robert Malones free Kindle offer for his new book, Lies My Government told Me. Basically it's a scam and Amazon charge the full price.
Mathew Crawford is your guy. @EduEngineer on Twitter
Steve.
I have suggested many times to you that your playing the wrong data game. Enough with the defense !
(What's already happened with adverse effects & death.)
Please consider OFFENSE. Play their game but better. Use these floating millions $$$ you have and with others, coordinate a real-time blood assay by getting, say 5,000 random people to submit a blood sample for analysis. Use a free Vit-D test as an incentive.
Contract with an Accredited Lab to perform the following tests.
HEART ATTACKS / MYOCARDITIS
- Troponin
- d-Dimer
- C-Reactive Protein
- Platelet Panel
COMPROMISE IMMUNE SYSTEM / CANCERS
CD8
CD4
Analyze this real-time data by the best statistical minds. The major Cohort comparisons. Compare the real-time test values vs. normal test ranges. Better are those with a blood history baseline over years to compare against.
- Vaxxed (2x, 3x, 4x)
- UnVaxxed
- Gender
- Age Stratified
- Shedding in the UnVaxxed
If you want to FULL-STOP these vaccines, just show the results (blood marker message / damage) of the walking to be dead. Or anyone who wants to walk in to test and validate their risk.
This is how you Red Pill the current Blue Pills who have had the Jabs (2-3X) and feel nothing is wrong with them and other family members after more than 18 months out from being vaxxed. You have to scare them with science & reality.
Covid is a blood disease.
Blood has special specific markers to highlight these organ / system damages. Would only take a month to complete with organization.
Please consider playing OFFENSE Steve.
We are running out of time.
We lose the infants & children, it is all over for us.
for Tom Renz's view of the Maricopa Judicial predicament, see this vid: Start at Min. 8 Sec 10 to 21 min.
https://www.bitchute.com/video/cjtmT0EHqGMv/
This video is taped prior to Today's ruling of the judge that resulted in quashing 8 of 10 claims of Plaintiff Lake's charges against Sec. of State / Defendant Hobbs, which now consists of two issues raised the judge will allow to proceed to trial - Lake only needs one to prevail and Hobbs is out.
Per atty. Renz 59% of the Maricopa county election printers were not working the day of the election - that should be all that's needed; odds of more than 1% out during day of election is suspicious, 10% not working is impossible. >20% is sabotage beyond a shadow of doubt.
Steve, please offer to debate Peter Hortez! He is calling you a killer!
https://citizenfreepress.com/breaking/the-branch-covidians-are-calling-us-killers/
What percent of Steve's readers are in favor of mass covid vaccination? I bet is is much lower than 30%, which should tell you something when your own base is calling you out.
Just as it is important for the 30% to describe their criticisms, it is equally important to describe strengths. What have the other 70% of readers said about the specific strengths of this analysis, relative to other analyses? They are equally not so active in the comments. Truth be told, 90% of commenters have no clue what they are talking about, in either side of the fence. Sorry if I am in a frustrated mood today.
I greatly appreciate your work Steve, but I am far from convinced by this dataset, as presented in Joel Smalley's video. Here are several problems which spring to mind, given time I may think of more:
1. The assumption that in the absence of Covid deaths, percentages of vaccinated vs. unvaccinated deaths should align with percentages of each category in the population needs a detailed justification. Smalley just vaguely refers to inaccuracy of census data and to UK survey data indicating that official stats overstate % of population vaccinated. Since this initial "normalisation" of the raw data is absolutely key to his entire analysis, such vague hand-waving is totally unsatisfactory. For example, one could argue that, if vaccines reduce Covid illness then they will also reduce non-Covid mortality over time by leaving those who are vaccinated in a generally more healthy state.
2. Another possible explanation for why the percentage of unvaccinated deaths in the raw data exceeds the percentage of non-vaccinees in the population (after Aug 21) is that your readers, and hence also the people they know well, are more likely to be unvaccinated than the average in the population, hence more likely to know people who died who were unvaccinated.
This is just an example of how your dataset might not be a representative sample of deaths in the general population. There are other indications of such bias, for example Covid deaths are signficantly overrepresented in your raw data. Let's take 2021, where you have a full year of data. From Smalley's video (0:26ff), the total number of deaths per month is (approx., I'm estimating from the figure)
26, 20, 34, 26, 28, 29, 35, 40, 61, 49, 57, 75
Total for 2021: 480.
For Covid deaths (18:00ff) the numbers are
8,5,4,9,9,1,8,12,15,9,12,19
Total for 2021: 111
So 111/480 = 23,1% of reported deaths in the raw data are Covid deaths.
Now, according to Worldometers, the total number of US Covid deaths were as follows:
Dec 31, 2020: 376,411
Dec 31, 2021: 852,759
=> 476,348 Covid deaths in 2021.
According to CDC data
https://www.cdc.gov/mmwr/volumes/71/wr/mm7117e1.htm
there were approx. 3,458,697 deaths in the US in 2021.
Thus 476,348/3,458,697 = 13,8% of all deaths were Covid deaths.
So Covid deaths are overrepresented by a factor of nearly 2 in your sample. We don't have a full year of data for 2022, but it seems to be a similar picture there.
So your dataset simply doesn't seem to be close to a representative sample of all deaths. It's perhaps not surprising: you asked respondents to submit the death the details of which they recalled best. And given the high profile of Covid, people are more likely to recall the details of deaths that involved Covid.
3. Smalley explains the spike in unvaccinated deaths in late summer 2021 as due to a kind of "survivor bias". But survivor bias could also explain the slow increase in the % of vaccinated deaths after March 2022, that it over time in general. Imagine for simplicity that there are only 2 kinds of people, healthy and unhealthy. The latter are more likely to die in any given time interval. Suppose now you have a vaccine with some positive efficiency, which is taken by the same percantage of healthy as unhealthy people. As time goes on, the percentage of deaths that are vaccinees will increase, not because the vaccine loses efficacy but simply because, amongst the unhealthy, the unvaccinated are dying off faster.
4. I don't think Smalley explains where he got his curve for the percentage of the population vaccinated over time. But his curve doesn't seem to align with, for example, the curve from Our World in Data.
https://ourworldindata.org/covid-vaccinations
There, for example, it says that as of March 31, 2022 the % who received at least one dose was 77.04%, not the 84% in Smalley's video. (And the % fully vaccinated on that date was only 66.15%).
Summary: Of the points above, #1 is most fundamental. If that "normalisation" could be properly justified, then maybe the dataset is then worth analysing further. Indeed, the overrepresentation of Covid deaths in the raw data should a priori strengthen a "not safe and effective" conclusion, assuming that the vaccines may prevent some Covid deaths but at the expense of deaths from adverse reactions. On the other hand, one would expect the % of deaths amongst vaccinees to be a priori higher than in the population at large, because vaccination rates are higher amongst the elderly and unhealthy. So this would require an extra normalisation. Your data may also be signficiantly non-representative in many other ways I haven't thought of. In short, there are so many problems I am not sure there is much valuable information to be extracted here.
Of course it is not bullet proof.
Where did the data come from? It came from a self selected sample from your substack. Is that a random sample of 'western' folk? Of course it is not.
Now the interesting thing is it may well be 'close enough for government work' as we used to say. But in the world of proofs, beyond debate? It is not. "lipstick on a pig" comes to mind.
Sorry man, how it is.
EDIT I should say. I spent most of a career collecting and analyzing messy biologic data, some collected by industry to support their suppositions about what a 'sustainable' harvest level is. Lots of money involved, and the fate of this or that lumber mill. The first question us government gballs would ask as we sat around the table was "where did this data come from? What were the rules for collection?" because that would color everything else. I have a lot of experience in that realm.
Where did this data come from? How was it collected? What potential or real bias is associated with it? First step.