This was wonderful. I wish every scientist wrote like this! Love the phrase "sticky priors." I've always thought of them as axioms, or maybe that's related, but am switching to "sticky priors." Thank you for sharing this with us.
Thank you for this! Especially the section on unexpected or absent results, factoring in the supply and demand for different types of data and findings. That's a tough skill for lay people like me, but the essay handles it elegantly and practically. I try to put on my Bayes hat to weigh priors and new data, but sometimes it narrows my perspective. The essay gives such a great model for improvement.
Dr Robinson wrote "... the media knew there was strong demand for stories of children at risk for COVID-19 in childcare settings and that there weren’t enough verified outbreaks to meet that demand." For me this confirms that, along with everything else, we have to deal with a sensationalist (read, profit-driven) media.
Andrey Mir, in his book "Post-Journalism and the Death of Newspapers," deals at length with your point about "sensationalist ... media." see e.g., https://human-as-media.com/
This is a bit late perhaps, but the following link provides some very interesting evidence that schools are major sources of covid transmission. This should not be surprising given two recently accepted data points. First that young kids get a covid at similar rates to adults thought they tend to be less symptomatic and second that covid transmits primarily via aerosols. Combine these two data and there is no reason to expect schools to be peculiarly safe spaces. The following tweet provides data to back this suspicion up:
Thank you, Dr. Robinson. To me, one of the more subtle and challenging parts of your piece is the tension between the individual choice you were able to make for your family and the population-level decision-making sphere shaped by institutional epidemiology. As a public school teacher recently ordered back to in-person schooling by my state's governor, I have facilitated discussions in our school community about how we can best practice consent as families decide whether their child should attend in-person school or remain fully remote. The families have a choice that was not given to me and my colleagues, making consent especially challenging -- and, I believe, adding to the demands upon meta-epistemological thinking. I believe our governor followed the advice of epidemiologists, especially those of the CDC, when ordering schools to re-open. Consequently, a population-level health decision has led to at least three tiers of differing decision-making power, beyond the tier-of-one occupied by the governor: families of school children were provided a choice; teachers and other school staff were not given a choice; and the wider community was also not allowed a say. Are questions of consent admissible within meta-epistemology? What role should consent play in a meta-epistemological toolbox? What questions are asked or left unasked depending upon one's position within the field of consent of a given system?
All excellent questions. This is going to send like a dodge (which it may be!), but I'll tell you how I think about it. I'm an academic epidemiologist -- we do the studies and come up with the estimates and most of us then usually wash our hands of the harder work of implementation and balancing trade-offs and thinking about ethics. There *are* applied epidemiologists and public health folks -- those that work at health departments or even the CDC. Those that are good at their job do the research part but also explicitly incorporate these trade-offs and have ethical frameworks for the way they think about the exercise of public health power. They realize it's a heavy weight. One of the big limitations in all our discourse is the absence of voices from applied and governmental public health. Many of them simply cannot speak openly in an official capacity. Their bosses are the elected officials (governors, Secretaries of health, etc.). And, these elected officials (especially governors but also sometimes local school boards) have final decision-making power. So most of us haven't really gotten much of a good look into how applied public health folks are parsing data and weighing it against the interests of various stakeholders. I know many of these folks and many of them are excellent and thoughtful people. i think that our national understanding and decision-making would have been enriched by hearing more from them.
Wonderful. Thank you Dr. Robinson, and thank you @zeynep for making a place for such contributions in your forum.
Principle 2 is my new Exhibit A for meta-epistemology. It is not epistemology in any sense, yet there's nothing more critical for accurately building real-world knowledge.
I can't do better than to quote Akex K in saying "Thank you Dr. Robinson for this thought-provoking and well-explained piece. I particularly enjoyed the discussion of how to weigh consistent information lacking natural constituencies".
While you put the pieces together and came to the right conclusion in March 2020, it took the American Association of Pediatrics 3 more months and the CDC 6 months after that. Does that suggest that having a deadline and skin in the game gave you an advantage. Or as Ben Jonson said, "Depend upon it, sir, when a man knows he is to be hanged in a fortnight, it concentrates his mind wonderfully.”
Thank you Dr. Robinson for this thought-provoking and well-explained piece. I particularly enjoyed the discussion of how to weigh consistent information lacking natural constituencies.
I'm wondering how you and Dr. Tufekci think about the risk of Long Covid for a vaccinated individual, and whether you feel there is any reason to suspect that the risk level may differ depending on which vaccine one received. Do you feel that we have enough information on this topic to start applying the principles from Dr. Robinson's essay?
I found Dr. Tufekci's "Long Covid: How Bungle Reporting on a Thorny Topic" article very helpful and reassuring. Since then, however, I have not succeeded in finding much information about vaccines and Long Covid. As I recently received the J&J vaccinate, this is a subject I am quite interested in!
Thank you for reading this comment, and if you choose to respond, thank you for that as well!
Hello! Unfortunately, this question is out of my wheelhouse. My first thought: The best prevention against long COVID is infection prevention. So I'd pay attention to all the new data coming out about how well different vaccines prevent primary infection. I do see lots of federal funding gearing up to study Long COVID, so I am hopeful that we will have more answers by the fall re: interactions with specific vaccines.
Great article - especially the analysis of the media by asking the question "is there a demand for stories about X"? The only puzzling bit was the end. She correctly concluded that, even in the middle of the pandemic with no vaccines, it was not in fact risky to send her kid to childcare before there were no vaccines. Yet now she's vaccinated with what she correctly describes as "so, so effective" vaccines, which she must know have reduced such risk as there is even further, she runs through a whole series of reassurers, including the "silent lunch time" of her pre-schooler?
Daniel -- thank you for pointing out where my writing was confusing! In my attempt to be emphatic, I left you with exactly the opposite impression I meant to convey. I do not think that the vaccines are "so-so" (which Oxford Dictionary defines are "neither very good nor very bad." Instead, I think that are so, SO effective. I added an extra "so" for emphasis -- attempting to convey that i think the vaccines are spectacular. Does that make it all make more sense?
How have the recent school hotspots in MI, PA and Ontario,BC affected your views on kid infectiousness? Looks like schools have become locus of infections and earlier data concerning their low levels was more a function of their being asymptomatic than being without infection.
this is a great question. It has layers so I may not address all the relevant points. I do not know much about the situation in Ontario BC. My impression is that they don't mandate in-school masking there and that vaccination rates are relatively low compared to many US communities. So I'm going to leave them aside. I also don't know about the PA hotspot. So I'll stick to the MI hotspot. Please take what I'm about to say with a grain of salt and all the other caveats that I put into my essay. When I started delving into the MI spike in pediatric cases, the headlines were very alarming. But, when I dug deeper, there often was very little evidence of classroom transmission. As I said in my article, schools are one of the only institutions in most states that are legally bound to report cases. So when there is an outbreak in a community (even if there is no transmission in a school), there are going to be pediatric cases (because children, like the rest of us, frequently get infected by their closest contacts: family members and family friends). Now many of these cases might not be reported by family members or tied to non-school related activities if there's no in-depth contact tracing. But, if the child has to miss school, it will be reported to the school who then reports it. I think a lot of school cases are a instances of shooting the messenger. Because the school is the unit that's legally bound to report, then the media reports the rising cases as if everyone got infected at school. As the MI situation evolved, the public messaging shifted. The state officials made it clearer that the pediatric cases associated with schools had mostly occurred in the realm of extracurricular sports (you can tell that if the cases cluster on the network of the sports team versus any given classroom). It's still a big issue because extracurriculars are really important for many students -- but it's not the same as classroom spread. But a lot of the media still report these cases in a way that implies that even when careful contact tracing later finds that the spread was associated with contacts of teens outside of the school building. In brief, after following the Michigan outbreak these past weeks, I do not believe that schools were locus on infections there. So many schools all across the country recently re-opened to in-person schooling in March and April, including a natural experiment in my state, where a law passed by the legislature forced all elementary schools to return to at least 4 days of a week of instruction and middle and high schools to at least 2 days a week (I believe). If schools were locus of infections, you'd expect to have seen a surge of cases within a few weeks of all those students going back to school in North Carolina. But we haven't seen a clear indication of that. I think that the cherry-picking of (mandated) school reports of pediatric cases in hotspots with community spread (but few reports of "tens of thousands of schools operated today without SARS-CoV-2 spread") gives a skewed perception. And I still find that implicating K-12 schools in community outbreaks (when it's really more about low vaccionation rates and adult behavior) is an easy way to get clicks. I see people routinely just overlaying school opening over some chart to imply that schools are causing whatever outbreak people are seeing -- even if schools have demonstrably nothing to do with the matter at hand. Zeynep highlighted one of these on Twitter from Eric Feigl-Deng: https://twitter.com/zeynep/status/1384304239959105536?s=20.
OK, this got long. LOL. In short, these hotspots haven't changed my opinion. But the variants of interest did make me more cautious about my kids mask-wearing and social contacts. Now that I and many others in my community are vaccinated (I recently looked and my county has one of the highest vaccination rates in the country), I feel more relaxed than I did in January 2021. But I still err on the side of caution: masking, social distancing, etc.
This was wonderful. I wish every scientist wrote like this! Love the phrase "sticky priors." I've always thought of them as axioms, or maybe that's related, but am switching to "sticky priors." Thank you for sharing this with us.
Yes, I loved 'sticky priors' - that term (and it's meaning) will stick with me! :-)
Thank you for this! Especially the section on unexpected or absent results, factoring in the supply and demand for different types of data and findings. That's a tough skill for lay people like me, but the essay handles it elegantly and practically. I try to put on my Bayes hat to weigh priors and new data, but sometimes it narrows my perspective. The essay gives such a great model for improvement.
Thank you for this informative article.
Dr Robinson wrote "... the media knew there was strong demand for stories of children at risk for COVID-19 in childcare settings and that there weren’t enough verified outbreaks to meet that demand." For me this confirms that, along with everything else, we have to deal with a sensationalist (read, profit-driven) media.
Andrey Mir, in his book "Post-Journalism and the Death of Newspapers," deals at length with your point about "sensationalist ... media." see e.g., https://human-as-media.com/
wonderfully written- scientific and logical and human at the same time. thank you!
This is a bit late perhaps, but the following link provides some very interesting evidence that schools are major sources of covid transmission. This should not be surprising given two recently accepted data points. First that young kids get a covid at similar rates to adults thought they tend to be less symptomatic and second that covid transmits primarily via aerosols. Combine these two data and there is no reason to expect schools to be peculiarly safe spaces. The following tweet provides data to back this suspicion up:
https://twitter.com/billius27/status/1388584934927912968
Thank you, Dr. Robinson. To me, one of the more subtle and challenging parts of your piece is the tension between the individual choice you were able to make for your family and the population-level decision-making sphere shaped by institutional epidemiology. As a public school teacher recently ordered back to in-person schooling by my state's governor, I have facilitated discussions in our school community about how we can best practice consent as families decide whether their child should attend in-person school or remain fully remote. The families have a choice that was not given to me and my colleagues, making consent especially challenging -- and, I believe, adding to the demands upon meta-epistemological thinking. I believe our governor followed the advice of epidemiologists, especially those of the CDC, when ordering schools to re-open. Consequently, a population-level health decision has led to at least three tiers of differing decision-making power, beyond the tier-of-one occupied by the governor: families of school children were provided a choice; teachers and other school staff were not given a choice; and the wider community was also not allowed a say. Are questions of consent admissible within meta-epistemology? What role should consent play in a meta-epistemological toolbox? What questions are asked or left unasked depending upon one's position within the field of consent of a given system?
All excellent questions. This is going to send like a dodge (which it may be!), but I'll tell you how I think about it. I'm an academic epidemiologist -- we do the studies and come up with the estimates and most of us then usually wash our hands of the harder work of implementation and balancing trade-offs and thinking about ethics. There *are* applied epidemiologists and public health folks -- those that work at health departments or even the CDC. Those that are good at their job do the research part but also explicitly incorporate these trade-offs and have ethical frameworks for the way they think about the exercise of public health power. They realize it's a heavy weight. One of the big limitations in all our discourse is the absence of voices from applied and governmental public health. Many of them simply cannot speak openly in an official capacity. Their bosses are the elected officials (governors, Secretaries of health, etc.). And, these elected officials (especially governors but also sometimes local school boards) have final decision-making power. So most of us haven't really gotten much of a good look into how applied public health folks are parsing data and weighing it against the interests of various stakeholders. I know many of these folks and many of them are excellent and thoughtful people. i think that our national understanding and decision-making would have been enriched by hearing more from them.
Wonderful. Thank you Dr. Robinson, and thank you @zeynep for making a place for such contributions in your forum.
Principle 2 is my new Exhibit A for meta-epistemology. It is not epistemology in any sense, yet there's nothing more critical for accurately building real-world knowledge.
What a treat. Thank you.
I can't do better than to quote Akex K in saying "Thank you Dr. Robinson for this thought-provoking and well-explained piece. I particularly enjoyed the discussion of how to weigh consistent information lacking natural constituencies".
While you put the pieces together and came to the right conclusion in March 2020, it took the American Association of Pediatrics 3 more months and the CDC 6 months after that. Does that suggest that having a deadline and skin in the game gave you an advantage. Or as Ben Jonson said, "Depend upon it, sir, when a man knows he is to be hanged in a fortnight, it concentrates his mind wonderfully.”
Thank you Dr. Robinson for this thought-provoking and well-explained piece. I particularly enjoyed the discussion of how to weigh consistent information lacking natural constituencies.
I'm wondering how you and Dr. Tufekci think about the risk of Long Covid for a vaccinated individual, and whether you feel there is any reason to suspect that the risk level may differ depending on which vaccine one received. Do you feel that we have enough information on this topic to start applying the principles from Dr. Robinson's essay?
I found Dr. Tufekci's "Long Covid: How Bungle Reporting on a Thorny Topic" article very helpful and reassuring. Since then, however, I have not succeeded in finding much information about vaccines and Long Covid. As I recently received the J&J vaccinate, this is a subject I am quite interested in!
Thank you for reading this comment, and if you choose to respond, thank you for that as well!
For anybody who is interested, here's the link to the prior Insight issue I mention: https://www.theinsight.org/p/long-covid-how-bungle-reporting-on
Hello! Unfortunately, this question is out of my wheelhouse. My first thought: The best prevention against long COVID is infection prevention. So I'd pay attention to all the new data coming out about how well different vaccines prevent primary infection. I do see lots of federal funding gearing up to study Long COVID, so I am hopeful that we will have more answers by the fall re: interactions with specific vaccines.
Great article - especially the analysis of the media by asking the question "is there a demand for stories about X"? The only puzzling bit was the end. She correctly concluded that, even in the middle of the pandemic with no vaccines, it was not in fact risky to send her kid to childcare before there were no vaccines. Yet now she's vaccinated with what she correctly describes as "so, so effective" vaccines, which she must know have reduced such risk as there is even further, she runs through a whole series of reassurers, including the "silent lunch time" of her pre-schooler?
Daniel -- thank you for pointing out where my writing was confusing! In my attempt to be emphatic, I left you with exactly the opposite impression I meant to convey. I do not think that the vaccines are "so-so" (which Oxford Dictionary defines are "neither very good nor very bad." Instead, I think that are so, SO effective. I added an extra "so" for emphasis -- attempting to convey that i think the vaccines are spectacular. Does that make it all make more sense?
How have the recent school hotspots in MI, PA and Ontario,BC affected your views on kid infectiousness? Looks like schools have become locus of infections and earlier data concerning their low levels was more a function of their being asymptomatic than being without infection.
this is a great question. It has layers so I may not address all the relevant points. I do not know much about the situation in Ontario BC. My impression is that they don't mandate in-school masking there and that vaccination rates are relatively low compared to many US communities. So I'm going to leave them aside. I also don't know about the PA hotspot. So I'll stick to the MI hotspot. Please take what I'm about to say with a grain of salt and all the other caveats that I put into my essay. When I started delving into the MI spike in pediatric cases, the headlines were very alarming. But, when I dug deeper, there often was very little evidence of classroom transmission. As I said in my article, schools are one of the only institutions in most states that are legally bound to report cases. So when there is an outbreak in a community (even if there is no transmission in a school), there are going to be pediatric cases (because children, like the rest of us, frequently get infected by their closest contacts: family members and family friends). Now many of these cases might not be reported by family members or tied to non-school related activities if there's no in-depth contact tracing. But, if the child has to miss school, it will be reported to the school who then reports it. I think a lot of school cases are a instances of shooting the messenger. Because the school is the unit that's legally bound to report, then the media reports the rising cases as if everyone got infected at school. As the MI situation evolved, the public messaging shifted. The state officials made it clearer that the pediatric cases associated with schools had mostly occurred in the realm of extracurricular sports (you can tell that if the cases cluster on the network of the sports team versus any given classroom). It's still a big issue because extracurriculars are really important for many students -- but it's not the same as classroom spread. But a lot of the media still report these cases in a way that implies that even when careful contact tracing later finds that the spread was associated with contacts of teens outside of the school building. In brief, after following the Michigan outbreak these past weeks, I do not believe that schools were locus on infections there. So many schools all across the country recently re-opened to in-person schooling in March and April, including a natural experiment in my state, where a law passed by the legislature forced all elementary schools to return to at least 4 days of a week of instruction and middle and high schools to at least 2 days a week (I believe). If schools were locus of infections, you'd expect to have seen a surge of cases within a few weeks of all those students going back to school in North Carolina. But we haven't seen a clear indication of that. I think that the cherry-picking of (mandated) school reports of pediatric cases in hotspots with community spread (but few reports of "tens of thousands of schools operated today without SARS-CoV-2 spread") gives a skewed perception. And I still find that implicating K-12 schools in community outbreaks (when it's really more about low vaccionation rates and adult behavior) is an easy way to get clicks. I see people routinely just overlaying school opening over some chart to imply that schools are causing whatever outbreak people are seeing -- even if schools have demonstrably nothing to do with the matter at hand. Zeynep highlighted one of these on Twitter from Eric Feigl-Deng: https://twitter.com/zeynep/status/1384304239959105536?s=20.
OK, this got long. LOL. In short, these hotspots haven't changed my opinion. But the variants of interest did make me more cautious about my kids mask-wearing and social contacts. Now that I and many others in my community are vaccinated (I recently looked and my county has one of the highest vaccination rates in the country), I feel more relaxed than I did in January 2021. But I still err on the side of caution: masking, social distancing, etc.