Not Social desireability bias is to blame for binary thinking re safety. I guess most Humans are just naturally not very good to think at continuums especially when it comes to safety. I guess humans are for some reason binary thinkers.
I have noticed the same thing. I am a little shocked at how incapable or unwilling otherwise smart people seem to be at stochastic thinking. Especially since most issues facing us humans as we go through our lives will be a "numbers game" and not stark, binary choices.
It's weird though, because most the continuum ultimately does lead to a binary choice.
Like, a 0.02% chance of dying from COVID vs. a 0.015% chance of dying from COVID Vaccine is still a binary choice. At the end of the day, we have to chose one package or risks or the other (which includes the uncertainties that go along with them).
It's not even clear to me what the true risks are as a pretty educated person who's totally willing to think statistically.
"Reports of death after COVID-19 vaccination are rare. FDA requires healthcare providers to report any death after COVID-19 vaccination to VAERS, even if it’s unclear whether the vaccine was the cause. Reports of adverse events to VAERS following vaccination, including deaths, do not necessarily mean that a vaccine caused a health problem. More than 608 million doses of COVID-19 vaccines were administered in the United States from December 14, 2020, through August 24, 2022. During this time, VAERS received 16,144 preliminary reports of death (.0027%) among people who received a COVID-19 vaccine. "
0.0027% (probably less due to false correlation) is a much smaller risk than your 0.015%., about 6 times smaller risk Also you should have included the risk of serious but survived Covid illness for a proper comparison. In fact, to date there have been 607,503,134 Covid cases worldwide with 6,492,964 deaths. Math says the risk of dying is 1% not your 0.02%, about 50 times higher than your claim. 1% risk of dying from covid is 370 times greater than the 0.0027% risk of dying form the covid vaccine.
Between two small numbers, one will always be smaller, which still makes it a binary choice.
The bigger problem is that the smaller the numbers, the more likely it is that residual uncertainty washes out the calculation entirely.
Your precise number of angels that can fit on a pinhead underscores that. Humans take care into account that undue precision often masks a lot of assumptions.
The point is the risk of dying from Covid is 370 times (a big number) greater than the risk of dying from the Covid vaccine. All your flailing will not change that.
You are implying that very tiny percentages are so close to zero that you can consider them trivial. So here is a non-covid example. An estimated 26,000,000 people participated in Floyd protests all over America for many months. Of those 7000 were arrested. Most were released without charges and without bail. Of those charged, the most common charge was breaking curfew. The rest were responsible for the burning and looting.
Coincidentally, 7000/26,000,000 equals 0.027%. The percentage looting and burning is maybe a fourth of that, in other words, near zero. So nothing happened, right?
> Compare unfamiliar risks to familiar risks. Start with: How does this risk compare to the risk of driving?
The problem here is that people often have underindexed or overindexed fears relative to actual risks, so they can pick comparisons to support their instincts or, more commonly, to bludgeon their opposition with. If you want to make the risk of driving sound trivial, compare it to the flu. If you want to make it sound severe, compare it to terrorism. Cars are the equivalent of ten 9/11s every year, or half of a rough flu season - dealer's choice.
In principle, it's a good approach, but it requires a starting point of rationality that is hard to reach and would require constant effort just to maintain.
This is further complicated by the massively different risk profiles people have. A large percentage of fatal car accidents are likely caused by driver impairment(drugs, alcohol) or extremely bad weather. There are no doubt many people for whom driving is incredibly safe, because they avoid driving in unsafe conditions and don't do stupid things behind the wheel. If you assume that everyone has the same odds of dying in a car wreck(or dying from terrorism, or the flu), you will draw absurd conclusions and implement ridiculous policies.
Okay, I have had two car accidents in my whole life, and I drove away from both of them. I have had some close calls usually because of the carelessness of other drivers. By your logic, either mandated car insurance is a ridiculous policy or my good driver discount is way too low..
First, I don't think mandating car insurance is based on the presumption that everyone has the same risk, just that everyone has some risk.
Second, I'm not sure if you're arguing that driving is relatively safe or dangerous for you. Maybe your risk is average. You could assume that everyone is at roughly equal risk from careless drivers, but the careless drivers would have to be at a higher risk, meaning that your driving would still be relatively safe.
You are missing the point. A driver's car insurance premium is higher or lower based on a number of criteria.. However, the whole analogy is false. The point of insurance in general is to insure against low probability risks. You might pay more or less due to different risk profiles, but it is foolish to go entirely without. Vaccines are a sort of insurance., not only against death, but also persistent health impacts of the disease.
Anti-vaxxers in general believe that vaccine risks for children, however low, are higher than risks (not only of death) from the disease itself. When it comes to Covid, children accounted for about 19% of all COVID-19 cases, but less than 0.26% of cases resulted in death, according to the American Academy of Pediatrics. People pretend death is the only consideration, but no parent wants to see their child suffer from an illness, even if the child eventually recovers. 19% is actually quite high. The real percentage is probably higher as asymptomatic cases were likely not tested and identified. Long-term effects are still unknown.
I wasn't making an analogy, I was stating a general principle. I didn't say anything about vaccines. I do understand my automotive risks pretty well, and I do forgo car insurance except for the legal minimum, because having insurance for a vehicle that only cost $1500 doesn't really make financial sense. Other people in different situations(with different cars) will no doubt make different decisions(which was my initial point, broadly stated).
I think that doing these comparisons more often, in public, would at least help people start to develop better calibrated intuitions about these things. It's not like people actually have good intuitions about prices in the millions or billions of dollars range, but they do see these numbers discussed often enough in the media that it's possible to get people to understand these things.
So is the risk of dying from a car crash or homicide or terrorism. But people *do* care about trivial risks, when it's a trivial risk of *dying*, and it makes sense for people to have properly informed discussions about *how* trivial each of these risks actually is.
I'm not sure the most ill-formed question concerns the vaccines but was for the basics of human existence: going outside, and being near other people. Both of these have been deemed 'unsafe' officially for long periods of time, and this still informs the behaviour of people today, many for the rest of their lives I imagine.
Your point on the risk/benefit tradeoff of vaccines being very favourable needs to be age adjusted in the manner of your masking maths. What matters is the absolute benefit and because of the extremely low likelihood of serious COVID effects to younger cohorts, the absolute benefit can only ever be minute. Describing this as 'very favourable' seems a stretch in that context.
The extremely low likelihood of serious covid effects seems logical as long as you imagine it won't happen to you., and in fact only makes sense NOW that most people are indeed vaxxed, and covid has mutated to a more transmissible, but less serious illness.
Not so for young people. Even in 2020 the absolute risk of serious health problems due to COVID for anyone otherwise healthy and relatively young was minute. Your first point can also relate to vaccine side effects, nothing to worry about unless it is you that suffers them....
We don't even know yet the long term effects of Covid among survivors. The point is that the risk of illness and/or death is 370 times greater than the risk of side effects from the vaccine. "Minuteness" is not the point. Medical researchers do not consider 6 million cases with 1 million dead minute, any more than the public considered the 3500 rioters among the total estimated 26 million Floyd protesters all over America over many months "minute" or the 1000 rioters among the 10,000 who attend Trump Jan 6 rally "minute."
Amazing that this fairly unremarkable blog post (not that it's bad or wrong, just a fairly basic statement of something I completely agree with so not something I'd otherwise have likely remembered long) ends with a quote I'm going to remember for a long time. I'd perhaps extend it from the family environment and make it an instruction: "Lead with good maths before someone else leads with bad poetry" would make a smashing life motto (I'm British so I can't say "math" without some sort of negative reaction :p)
"For now, all we can honestly say is, “The risk trade-off seems very favorable..."
Where have you been for the last two years? It has been completely obvious for a LONG time that the vaccine's effectiveness goes negative within a few months, and that people are dropping dead by the thousands after taking it. The mRNA vax is a total disaster, and there's no excuse for not realizing that.
Bryan has his math wrong: Covid vaccines are clearly not beneficial for people younger than 65, they're dangerous, especially for athletes who exercise. Even for lean people older than 65, the dangers of the vaccines are higher than the potential benefits.
Part of why people are bad with risk is that our brains don't grasp large/small magnitudes intuitively. (I have a gut feel for the difference between of 25%. I do not have a gut feel for 0.00000025%.) This is exactly why it's important, as you say, to "compare unfamiliar risks to familiar risks".
As someone that does statistics for a living, I long ago realized that if most of my fellows were good at statistics then I would not have the highly compensated career I have. Therefore, I must conclude that they are not good at statistics, and accept this as the inevitable tradeoff for my having a rare talent the market demands.
This makes sense based on evolution. How valuable would statistical talent be in the pre-modern world for increased surviving fertility. Not much. Again, statistics dictates that humans will be bad at statistics.
Not Social desireability bias is to blame for binary thinking re safety. I guess most Humans are just naturally not very good to think at continuums especially when it comes to safety. I guess humans are for some reason binary thinkers.
I have noticed the same thing. I am a little shocked at how incapable or unwilling otherwise smart people seem to be at stochastic thinking. Especially since most issues facing us humans as we go through our lives will be a "numbers game" and not stark, binary choices.
It's weird though, because most the continuum ultimately does lead to a binary choice.
Like, a 0.02% chance of dying from COVID vs. a 0.015% chance of dying from COVID Vaccine is still a binary choice. At the end of the day, we have to chose one package or risks or the other (which includes the uncertainties that go along with them).
It's not even clear to me what the true risks are as a pretty educated person who's totally willing to think statistically.
From the CDC:
"Reports of death after COVID-19 vaccination are rare. FDA requires healthcare providers to report any death after COVID-19 vaccination to VAERS, even if it’s unclear whether the vaccine was the cause. Reports of adverse events to VAERS following vaccination, including deaths, do not necessarily mean that a vaccine caused a health problem. More than 608 million doses of COVID-19 vaccines were administered in the United States from December 14, 2020, through August 24, 2022. During this time, VAERS received 16,144 preliminary reports of death (.0027%) among people who received a COVID-19 vaccine. "
0.0027% (probably less due to false correlation) is a much smaller risk than your 0.015%., about 6 times smaller risk Also you should have included the risk of serious but survived Covid illness for a proper comparison. In fact, to date there have been 607,503,134 Covid cases worldwide with 6,492,964 deaths. Math says the risk of dying is 1% not your 0.02%, about 50 times higher than your claim. 1% risk of dying from covid is 370 times greater than the 0.0027% risk of dying form the covid vaccine.
Sure but that was entirely not the point…
Between two small numbers, one will always be smaller, which still makes it a binary choice.
The bigger problem is that the smaller the numbers, the more likely it is that residual uncertainty washes out the calculation entirely.
Your precise number of angels that can fit on a pinhead underscores that. Humans take care into account that undue precision often masks a lot of assumptions.
The point is the risk of dying from Covid is 370 times (a big number) greater than the risk of dying from the Covid vaccine. All your flailing will not change that.
Err… I wasn’t trying to change it.
Maybe you need a non COVID example to think clearly
You are implying that very tiny percentages are so close to zero that you can consider them trivial. So here is a non-covid example. An estimated 26,000,000 people participated in Floyd protests all over America for many months. Of those 7000 were arrested. Most were released without charges and without bail. Of those charged, the most common charge was breaking curfew. The rest were responsible for the burning and looting.
Coincidentally, 7000/26,000,000 equals 0.027%. The percentage looting and burning is maybe a fourth of that, in other words, near zero. So nothing happened, right?
Makes decision making easier, harder to overcome analysis paralysis with data like “this choice involves 2% chance of death from a lion”.
Is your vaxx-safety reading confined to MR and other such bubblies? My view is very much otherwise.
Seconded. The data out of Europe is very troubling, and indeed the data out of the US is pretty ugly, if very incomplete.
Tabarrok’s vaccine
> Compare unfamiliar risks to familiar risks. Start with: How does this risk compare to the risk of driving?
The problem here is that people often have underindexed or overindexed fears relative to actual risks, so they can pick comparisons to support their instincts or, more commonly, to bludgeon their opposition with. If you want to make the risk of driving sound trivial, compare it to the flu. If you want to make it sound severe, compare it to terrorism. Cars are the equivalent of ten 9/11s every year, or half of a rough flu season - dealer's choice.
In principle, it's a good approach, but it requires a starting point of rationality that is hard to reach and would require constant effort just to maintain.
This is further complicated by the massively different risk profiles people have. A large percentage of fatal car accidents are likely caused by driver impairment(drugs, alcohol) or extremely bad weather. There are no doubt many people for whom driving is incredibly safe, because they avoid driving in unsafe conditions and don't do stupid things behind the wheel. If you assume that everyone has the same odds of dying in a car wreck(or dying from terrorism, or the flu), you will draw absurd conclusions and implement ridiculous policies.
Okay, I have had two car accidents in my whole life, and I drove away from both of them. I have had some close calls usually because of the carelessness of other drivers. By your logic, either mandated car insurance is a ridiculous policy or my good driver discount is way too low..
First, I don't think mandating car insurance is based on the presumption that everyone has the same risk, just that everyone has some risk.
Second, I'm not sure if you're arguing that driving is relatively safe or dangerous for you. Maybe your risk is average. You could assume that everyone is at roughly equal risk from careless drivers, but the careless drivers would have to be at a higher risk, meaning that your driving would still be relatively safe.
You are missing the point. A driver's car insurance premium is higher or lower based on a number of criteria.. However, the whole analogy is false. The point of insurance in general is to insure against low probability risks. You might pay more or less due to different risk profiles, but it is foolish to go entirely without. Vaccines are a sort of insurance., not only against death, but also persistent health impacts of the disease.
Anti-vaxxers in general believe that vaccine risks for children, however low, are higher than risks (not only of death) from the disease itself. When it comes to Covid, children accounted for about 19% of all COVID-19 cases, but less than 0.26% of cases resulted in death, according to the American Academy of Pediatrics. People pretend death is the only consideration, but no parent wants to see their child suffer from an illness, even if the child eventually recovers. 19% is actually quite high. The real percentage is probably higher as asymptomatic cases were likely not tested and identified. Long-term effects are still unknown.
I wasn't making an analogy, I was stating a general principle. I didn't say anything about vaccines. I do understand my automotive risks pretty well, and I do forgo car insurance except for the legal minimum, because having insurance for a vehicle that only cost $1500 doesn't really make financial sense. Other people in different situations(with different cars) will no doubt make different decisions(which was my initial point, broadly stated).
I think that doing these comparisons more often, in public, would at least help people start to develop better calibrated intuitions about these things. It's not like people actually have good intuitions about prices in the millions or billions of dollars range, but they do see these numbers discussed often enough in the media that it's possible to get people to understand these things.
The risk of dying from the flu or Covid are trivial for healthy people.
Risk of death is not the only risk.
So is the risk of dying from a car crash or homicide or terrorism. But people *do* care about trivial risks, when it's a trivial risk of *dying*, and it makes sense for people to have properly informed discussions about *how* trivial each of these risks actually is.
I'm not sure the most ill-formed question concerns the vaccines but was for the basics of human existence: going outside, and being near other people. Both of these have been deemed 'unsafe' officially for long periods of time, and this still informs the behaviour of people today, many for the rest of their lives I imagine.
Your point on the risk/benefit tradeoff of vaccines being very favourable needs to be age adjusted in the manner of your masking maths. What matters is the absolute benefit and because of the extremely low likelihood of serious COVID effects to younger cohorts, the absolute benefit can only ever be minute. Describing this as 'very favourable' seems a stretch in that context.
The extremely low likelihood of serious covid effects seems logical as long as you imagine it won't happen to you., and in fact only makes sense NOW that most people are indeed vaxxed, and covid has mutated to a more transmissible, but less serious illness.
Not so for young people. Even in 2020 the absolute risk of serious health problems due to COVID for anyone otherwise healthy and relatively young was minute. Your first point can also relate to vaccine side effects, nothing to worry about unless it is you that suffers them....
We don't even know yet the long term effects of Covid among survivors. The point is that the risk of illness and/or death is 370 times greater than the risk of side effects from the vaccine. "Minuteness" is not the point. Medical researchers do not consider 6 million cases with 1 million dead minute, any more than the public considered the 3500 rioters among the total estimated 26 million Floyd protesters all over America over many months "minute" or the 1000 rioters among the 10,000 who attend Trump Jan 6 rally "minute."
Amazing that this fairly unremarkable blog post (not that it's bad or wrong, just a fairly basic statement of something I completely agree with so not something I'd otherwise have likely remembered long) ends with a quote I'm going to remember for a long time. I'd perhaps extend it from the family environment and make it an instruction: "Lead with good maths before someone else leads with bad poetry" would make a smashing life motto (I'm British so I can't say "math" without some sort of negative reaction :p)
Unfortunately, no matter how good you are at math, the government is going to "lead" (i.e., impose upon you) with poetry.
There is a lot in that, not only SDB but also pro-authority bias and binary thinking.
I think about this question a lot, especially the probabilistic vs. binary thinking angle.
Deserves extended treatment in your upcoming "Capitalism and Freedom" book!
(False) binaries are also a crude technique of argumentation. Not unlike a lawyer's demanding of a witness testifying, "Answer yes or no."
"For now, all we can honestly say is, “The risk trade-off seems very favorable..."
Where have you been for the last two years? It has been completely obvious for a LONG time that the vaccine's effectiveness goes negative within a few months, and that people are dropping dead by the thousands after taking it. The mRNA vax is a total disaster, and there's no excuse for not realizing that.
Bryan has his math wrong: Covid vaccines are clearly not beneficial for people younger than 65, they're dangerous, especially for athletes who exercise. Even for lean people older than 65, the dangers of the vaccines are higher than the potential benefits.
You will have to back up such wild assertions, especially when phrased as "clearly not beneficial."
How would you respond if asked, "are we at a point on the safety curve where is it prudent to take thus-and-such an action?"
Part of why people are bad with risk is that our brains don't grasp large/small magnitudes intuitively. (I have a gut feel for the difference between of 25%. I do not have a gut feel for 0.00000025%.) This is exactly why it's important, as you say, to "compare unfamiliar risks to familiar risks".
As someone that does statistics for a living, I long ago realized that if most of my fellows were good at statistics then I would not have the highly compensated career I have. Therefore, I must conclude that they are not good at statistics, and accept this as the inevitable tradeoff for my having a rare talent the market demands.
This makes sense based on evolution. How valuable would statistical talent be in the pre-modern world for increased surviving fertility. Not much. Again, statistics dictates that humans will be bad at statistics.