Consider me as a remedial kindergartener when it comes to Statistics. That disclaimer out of the way, I have learned so much from your articles and conversations. Thank you again Mathew. I am blessed to call you my friend as well as my remedial kindergarten statistics teacher.
Thanks, Mathew. It appears that HUB is a substantial effect which is almost never taken into account in interpreting data, but crucial to making any conclusions about the meaning of any interventional medical data. What is odd, to me, in the U.S. Mortality data, is that it shows a substantial increase throughout the entire period, including all of 2019.
Thanks Mathew for another great post and video. Really interesting to see your exchange with Seheult and good faith attempt to point him towards better statistical reasoning. He is a difficult case since he does make nice content with clear communication and has a lot of hubris. In his Texas tweet, nice to see a few folks pointing out the problems and inconsistencies. He also fails to mention Texas all-cause excess mortality remained positive all through 2022 as seen in the Texas graph at USMortality https://www.usmortality.com/excess-mortality/percentage, and as you have repeatedly explained, we have both HUB and lack of mean reversion.
Side note: I'm happy to relay that I've finally been able to reproduce your correlation time series graph. You briefly showed your Excel table in the video and this reminded me that you are working with daily data. I made scatter plots behind these correlations and see a lot of zeros for daily mortality because a large number of counties only report deaths on a weekly basis. All of these zeros significantly shrink the correlations into the range [-0.1, 0.1]. For what it's worth, I think if the data are aggregated to weekly or monthly, the correlations become much stronger (e.g. -0.5) and the resulting graph becomes more compelling because then there is no doubt we are dealing with significant associations. I'm glad to pitch in on any data analysis tasks you find are stretching your time.
I watched the video above (Episode I: the overconfidence of Dr. Roger Seheult). Maybe you could be more explicit about what you mean by conditional statistics and healthy user bias with simpler examples. The Monty Hall example is too confusing for most people.
Using examples directly related to vaccines, Covid, excess death etc. is a good idea.
Wikipedia does a good job on Healthy User Bias:
"The kind of subjects that take up an intervention, including by enrolling in a clinical trial, are not representative of the general population. People who volunteer for a study can be expected, on average, to be healthier than people who don't volunteer, as they are concerned for their health and are predisposed to follow medical advice,[1] both factors that would aid one's health. In a sense, being healthy or active about one's health is a precondition for becoming a subject of the study, an effect that can appear under other conditions such as studying particular groups of workers. For example, someone in ill health is unlikely to have a job as manual laborer. As a result, studies of manual laborers are studies of people who are currently healthy enough to engage in manual labor, rather than studies of people who would do manual labor if they were healthy enough."
My main point is multiple simple examples are a good idea.
The Simpson's paradox example you brought up might have gone over many peoples heads.
Here's the Wikipedia explanation for those that don't know what it is:
Simpson's paradox is a phenomenon in probability and statistics in which a trend appears in several groups of data but disappears or reverses when the groups are combined.
You're right. I could take what is 5% of my time, make it 100%, and do a much better job. Or I could not sleep. But I'd need help with other priorities. It's all about time right now, and I get help from nobody much of the time. If somebody else could go ahead and plan out and record a better video, I'd be grateful.
This is such a predictable response from you. I've seen you respond to reasonable and well-intentioned criticism like this over and over again here, whinging about how you don't have enough time and nobody helps you. What's wrong with a simple "thanks, I'll take that into consideration next time I do one of these"? That would be the standard.
Is it possible that you're just not as good a teacher as you think you are, since you prefer to only work with self-selected individuals who are highly motivated and talented to begin with?
"What's wrong with a simple "thanks, I'll take that into consideration next time I do one of these"? That would be the standard."
Why would a person running a marathon stop and say, "Thank you," when somebody suggests that they could sprint and overtake the next runner? Either they're being harassed, or the person has no idea what's going on. I assume Jeffrey is in good faith, but I need him to know what's going on related to how little energy and time there is left to produce content that I'm not only the only one producing, but gets iced out in the MFM in favor of often strictly false content.
I'm not being rude or unfriendly. I'm explaining reality, and hopefully spurring people to understand that there is no help for this work (and to ask and interpret why that is).
"Is it possible that you're just not as good a teacher as you think you are, since you prefer to only work with self-selected individuals who are highly motivated and talented to begin with?"
Like the kids in downtown Chattanooga (literally the poorest of the poor) that I gave the same lessons as the self-selected families? Or are you inventing my work in your head so that you can take shots? Why would you do that?
I believe that had I not moved back to Dallas, I would have made the ghetto of Chattanooga a math team powerhouse.
How about working with people who don't have a high school degree?
Wasn't all that different from when I worked with a graduate student at Harvard getting an MBA.
Happy to take criticism, but it has to relate to reality. If you want to show me how it's done, please take over the work and show me. I'm begging you, honestly. I'm running three businesses, and I'd love to simplify back to one.
I totally understand the not having time issue. Your sfuff is valuable because I don't see your criticsim of vaccine statistics anywhere else (or at least I don't dig enough into other people's articles like Fenton's). I'm just pointing out suggestions to get your ideas out more clearly so it is understood by a larger audience.
Just suggesting you consider how your own style, approach, and attitudes feed into that predicament. You often come across as the aggrieved and agonistic quant, rather than the gentle and patient teacher that perhaps you imagine yourself as. And your prose is often unnecessarily convoluted, when concise and lucid explanations aren't that far away. Slow down, marathon runner. Is this really a race you need to personally win? Your choice of metaphor is telling. My 2 cents: do less if you need to, but do it better. Then maybe you won't be so frustrated.
As for those poor kids in Chattanooga, etc... you know full well that "talented and motivated" doesn't have much to do with initial SES or credentials. I think you dodged my point yet again. I trust it wasn't intentional.
Your textbooks had an editor, right? I hope so. I understand your frustration with being such a solo operation. It's funny, I actually started this account looking for serious collaboration, but by now, having read your stack for a year and a half, I'm not interested in collaborating with you. I have enough to do elsewhere. I do hope you find a capable and acceptable taker somewhere, though. You deserve it.
If, on the one hand, you have 30K readers, but on the other hand you think your stuff keeps getting "iced out" of the "MFM"... well... I don't know what to tell you, beyond what I just said. Look inward.
I wish you the best in your business endeavors. At this point, I think you have a better temperament for running a business than for single-handedly enlightening the unappreciative twitter doctors of the world.
Consider me as a remedial kindergartener when it comes to Statistics. That disclaimer out of the way, I have learned so much from your articles and conversations. Thank you again Mathew. I am blessed to call you my friend as well as my remedial kindergarten statistics teacher.
Doctors do not understand probability very well as Sebastian Rushworth explains
http://sebastianrushworth.com/2021/06/23/how-well-do-doctors-understand-probability/
This resulted in every positive Covid test result being counted as a ‘case’.
Here, John Hopkins publishes a list of Covid test kits
https://www.centerforhealthsecurity.org/covid-19TestingToolkit/molecular-based-tests/current-molecular-and-antigen-tests.html
Note that some claim 100% sensitivity and 100% specificity. Truly a gold standard test!
Overconfidence in test results ends up in over treatment of patients.
Great result for big Pharma eh!
Thanks, Mathew. It appears that HUB is a substantial effect which is almost never taken into account in interpreting data, but crucial to making any conclusions about the meaning of any interventional medical data. What is odd, to me, in the U.S. Mortality data, is that it shows a substantial increase throughout the entire period, including all of 2019.
Thanks Mathew for another great post and video. Really interesting to see your exchange with Seheult and good faith attempt to point him towards better statistical reasoning. He is a difficult case since he does make nice content with clear communication and has a lot of hubris. In his Texas tweet, nice to see a few folks pointing out the problems and inconsistencies. He also fails to mention Texas all-cause excess mortality remained positive all through 2022 as seen in the Texas graph at USMortality https://www.usmortality.com/excess-mortality/percentage, and as you have repeatedly explained, we have both HUB and lack of mean reversion.
Side note: I'm happy to relay that I've finally been able to reproduce your correlation time series graph. You briefly showed your Excel table in the video and this reminded me that you are working with daily data. I made scatter plots behind these correlations and see a lot of zeros for daily mortality because a large number of counties only report deaths on a weekly basis. All of these zeros significantly shrink the correlations into the range [-0.1, 0.1]. For what it's worth, I think if the data are aggregated to weekly or monthly, the correlations become much stronger (e.g. -0.5) and the resulting graph becomes more compelling because then there is no doubt we are dealing with significant associations. I'm glad to pitch in on any data analysis tasks you find are stretching your time.
I watched the video above (Episode I: the overconfidence of Dr. Roger Seheult). Maybe you could be more explicit about what you mean by conditional statistics and healthy user bias with simpler examples. The Monty Hall example is too confusing for most people.
Using examples directly related to vaccines, Covid, excess death etc. is a good idea.
Wikipedia does a good job on Healthy User Bias:
"The kind of subjects that take up an intervention, including by enrolling in a clinical trial, are not representative of the general population. People who volunteer for a study can be expected, on average, to be healthier than people who don't volunteer, as they are concerned for their health and are predisposed to follow medical advice,[1] both factors that would aid one's health. In a sense, being healthy or active about one's health is a precondition for becoming a subject of the study, an effect that can appear under other conditions such as studying particular groups of workers. For example, someone in ill health is unlikely to have a job as manual laborer. As a result, studies of manual laborers are studies of people who are currently healthy enough to engage in manual labor, rather than studies of people who would do manual labor if they were healthy enough."
My main point is multiple simple examples are a good idea.
The Simpson's paradox example you brought up might have gone over many peoples heads.
Here's the Wikipedia explanation for those that don't know what it is:
Simpson's paradox is a phenomenon in probability and statistics in which a trend appears in several groups of data but disappears or reverses when the groups are combined.
You're right. I could take what is 5% of my time, make it 100%, and do a much better job. Or I could not sleep. But I'd need help with other priorities. It's all about time right now, and I get help from nobody much of the time. If somebody else could go ahead and plan out and record a better video, I'd be grateful.
This is such a predictable response from you. I've seen you respond to reasonable and well-intentioned criticism like this over and over again here, whinging about how you don't have enough time and nobody helps you. What's wrong with a simple "thanks, I'll take that into consideration next time I do one of these"? That would be the standard.
Is it possible that you're just not as good a teacher as you think you are, since you prefer to only work with self-selected individuals who are highly motivated and talented to begin with?
Statistician, teach thyself. Cheers!
"What's wrong with a simple "thanks, I'll take that into consideration next time I do one of these"? That would be the standard."
Why would a person running a marathon stop and say, "Thank you," when somebody suggests that they could sprint and overtake the next runner? Either they're being harassed, or the person has no idea what's going on. I assume Jeffrey is in good faith, but I need him to know what's going on related to how little energy and time there is left to produce content that I'm not only the only one producing, but gets iced out in the MFM in favor of often strictly false content.
I'm not being rude or unfriendly. I'm explaining reality, and hopefully spurring people to understand that there is no help for this work (and to ask and interpret why that is).
"Is it possible that you're just not as good a teacher as you think you are, since you prefer to only work with self-selected individuals who are highly motivated and talented to begin with?"
Like the kids in downtown Chattanooga (literally the poorest of the poor) that I gave the same lessons as the self-selected families? Or are you inventing my work in your head so that you can take shots? Why would you do that?
I believe that had I not moved back to Dallas, I would have made the ghetto of Chattanooga a math team powerhouse.
How about working with people who don't have a high school degree?
https://www.campfire.wiki/doku.php?id=rounding_the_earth:a_path_to_statistics
Wasn't all that different from when I worked with a graduate student at Harvard getting an MBA.
Happy to take criticism, but it has to relate to reality. If you want to show me how it's done, please take over the work and show me. I'm begging you, honestly. I'm running three businesses, and I'd love to simplify back to one.
I totally understand the not having time issue. Your sfuff is valuable because I don't see your criticsim of vaccine statistics anywhere else (or at least I don't dig enough into other people's articles like Fenton's). I'm just pointing out suggestions to get your ideas out more clearly so it is understood by a larger audience.
I hear you. I know.
I'm actually building a business (two including the education company) now with the purpose of being able to hire help in all my projects.
Just suggesting you consider how your own style, approach, and attitudes feed into that predicament. You often come across as the aggrieved and agonistic quant, rather than the gentle and patient teacher that perhaps you imagine yourself as. And your prose is often unnecessarily convoluted, when concise and lucid explanations aren't that far away. Slow down, marathon runner. Is this really a race you need to personally win? Your choice of metaphor is telling. My 2 cents: do less if you need to, but do it better. Then maybe you won't be so frustrated.
As for those poor kids in Chattanooga, etc... you know full well that "talented and motivated" doesn't have much to do with initial SES or credentials. I think you dodged my point yet again. I trust it wasn't intentional.
Your textbooks had an editor, right? I hope so. I understand your frustration with being such a solo operation. It's funny, I actually started this account looking for serious collaboration, but by now, having read your stack for a year and a half, I'm not interested in collaborating with you. I have enough to do elsewhere. I do hope you find a capable and acceptable taker somewhere, though. You deserve it.
If, on the one hand, you have 30K readers, but on the other hand you think your stuff keeps getting "iced out" of the "MFM"... well... I don't know what to tell you, beyond what I just said. Look inward.
I wish you the best in your business endeavors. At this point, I think you have a better temperament for running a business than for single-handedly enlightening the unappreciative twitter doctors of the world.