perhaps i misheard it, but i literally heard the 95% for over-65s figure this morning on the radio, which seemed to be quoting the cdc (but again, it didn't have my full and undivided attention). the difference could also be first dose vs. 'fully vaccinated'.
At least this link says even for first dose it's only 91% and 87%, again, which could be off, yet seems as if it would be accurate.
Again, I don't know. What I've learned this year is that statistics can be very unintuitive and I know so little about how they work, that in uncertain times like these, I almost feel reliant on choosing someone to trust who I think understands this stuff better than I.
@nuerow cited nytimes[0] on the 95% estimate in another comment[1]. for the purposes of my argument, the precise number isn’t important, just that it equates to “most”.
and incidentally, that’s the shape of most arguments on most things, because we’re embedded in unbounded uncertainty. the most precision we can have about most things is ‘a little’, ‘some’, ‘a bunch’, ‘most’, etc., not something like 54.736%, despite how enticing precision like that can be. it’s mostly engineering where precision like that is practical, available, and useful.
I think precision also matters when dealing with large numbers. If it's 60% plus/minus 30%, not such big deal if dealing with 10 people, it'd be either 3, 6, or 9 people, a range of 6 people. But if it's plus/minus 30% when dealing with a population of 100,000,000, that's a range of 60,000,000 people.
That being said, yes, I agree there are many situations where precision doesn't matter that much and that qualifiers work sufficiently to describe the general sense of the situation. I think what actually drove me to write my precision rant was ironically where you stated that "easily" was hyperbole, which I took to mean as you believing it was not an accurate or precise enough description of the actions that might kill the grandma. Perhaps this thread wouldn't have happened if the person wrote "I might possibly kill grandma." I don't say that facetiously—I think these disagreements about the accuracy and precision of our observations of the world can lead to so much conflict, as it has for me in my life.
So, in a way, I'm grateful for you going back and forth with me, as I've learned even more deeply that while other people may seem to use too much or too little precision to me, I may use too much or too little for them.
yes, it’s miscalibrations of risk, but not necessarily numerical imprecision, that prods my reaction. it’s possible to kill grandma, in many different ways, but it’s not probable, unless intentionally reckless or malevolent. risk is inherent to life, and we need to be continually adjusting our risk assessments so that our individual and collective behaviors approach consistency.
I think the challenge with "easily" "possible" and "probable" is that they are imprecise and often very subjective. One person's "possible" could be another person's "probable."
Looking at stats from Nov 2020 for infection fatality rates[1] (I didn't know the difference among infection fatality rates, case fatality rates, and crude mortality rates until after I had read this[2]), a female over the age of 80 who contracts covid-19 had a 5.759% chance of dying, or approximately a 1 in 17 chance. Is that probable or possible? I suppose it might be relative to one's normal chance of dying. Perhaps her odds of dying from breaking her hip are much higher and therefore much more probable, so covid becomes a possible cause of death. But also, that was for 80 and above, so maybe at 100 the odds are much higher than 1 in 17. However, with new monoclonal antibodies perhaps they're much lower.
I think sometimes I struggle to have conversations on accuracy of risk calculations without getting more precision. Yes, I don't think it needs to go into 4 decimal places, as that precision is very hard to achieve in non-engineering settings as you had mentioned. I also think sometimes words can be way too nebulous and imprecise to bring clarity between two people, as I've done a lot of work with language, emotions, and conflict, and very often (how often is very often?), conflict comes down to different interpretations of the same word.
yes, that's the calibration aspect of it. for instance, we implicitly accept a 1% lifetime chance of dying in a car accident because of the great utility of automobiles. that's a useful calibration point because it has (nearly) universal acceptance and application.
a 6% chance of dying is nothing to sneeze at, but it's still far from probable, which by definition is more likely than not (>50%). and while it's 6 times death-by-car, it's also dependent on contracting covid in the first place (a prior, in bayesian terms). and that prior probability depends largely on whether the grandchild has covid themselves and how contagious they are (how much and where the virus is), and given that as an additional prior, what relevant precautions they take. this chain of dependent probabilities is also why most precautions have marginal effect (because multiplying lots of fractional numbers results in a smaller fraction). none of this chain of reasoning requires great precision by the way; a single digit of significance at each step being plenty for sound decision-making.
finally, note that a 100-year-old has a 30-35% all-cause chance of death: https://www.ssa.gov/oact/STATS/table4c6.html . the addition of covid risk is not immaterial, but relatively not overwhelming.
I appreciate these points, especially about the chain of dependent probabilities. Things can become much less likely depending on many different variables and such.
I guess for me, and maybe for the person with the grandma as well, emotions sway my perception of how risky something is or not, especially with covid. While my mom was going thru chemo, I was already worried that she might die from cancer, and then more so that I would bring a regular flu to her when she was immunocompromised and accelerate that process. Covid-19 stopped me in my tracks, a level of fear I can't recall feeling, that just by being in the same room with my mom and having conversation with her, I could possibly cause her death.
If a grandma of mine were to die from a broken hip, which is probably much more statistically likely than covid-19, I most certainly wouldn't feel as much anger towards myself and regret as if she would die from covid-19 and I had thought I gave it to her.
When my parents got vaccinated, I felt a huge relief, knowing that their probability of dying seemed to drop even more than where it was, almost to zero (I believe), which I probably had originally exaggerated due to fear, and yet, almost all of that fear disappeared for me.
I guess if anything, this thread has helped me pause and reflect on just how much my own emotional state can influence how I see risk, both amplifying or minimizing it away from its more mathematical reality. Thank you for helping me realize this.
awesome, outsized emotional influence on risk assessment is tough to talk about rationally, but it's so important to public policy discussions, not to mention our daily decisions. it takes a lot of emotional sturdiness to self-examine that way.
regarding covid risk, it makes sense to take precautions around the elderly and infirm, because while the risk isn't overwhelming, it's still relatively significant, and feeling relief by taking those precautions is totally understandable. it's just the extrapolation of that kind of situation to blanket public policy that really doesn't make sense (because risk varies so widely for the whole gamut of daily circumstances).
> awesome, outsized emotional influence on risk assessment is tough to talk about rationally, but it's so important to public policy discussions, not to mention our daily decisions.
Yeah, the more I think about, I wonder just how many pandemic decisions have been made by conflicting emotions, especially conflicting fears. Me afraid of my mom dying of cancer + giving her covid-19, others afraid they'll lose the restaurant that they've poured their life into, others afraid the government is using covid-19 as an excuse to gain more control over their lives and steal their property, etc. I talked with a friend the other day who isn't vaccinated and said, "Ya know, if I get covid and die, maybe it's my time." Made me think just how much more I seem to fear death than that person does.
I was listening a little to Sanjay Gupta's conversation with Joe Rogan and one point he made that I liked was talking about how sometimes the conversation about covid-19 risk becomes life or death, overlooking the possible negative effects of just having the virus itself.
> it's just the extrapolation of that kind of situation to blanket public policy that really doesn't make sense (because risk varies so widely for the whole gamut of daily circumstances).
I agree in that risk is widely distributed and it's one of the reasons why I love the idea of decision making at different levels and the concept of federalism in general. I think where it can be a struggle, especially in our current forms of government, is when people cross geographical boundaries. As people move from one place to another (and maybe even as we communicate and therefore virtually feel in many places at the same time or in this one large place together), standardization almost becomes required to make things work. Setting aside the conversation about medical validity of vaccines for a second, one thing they do is help to standardize someone's status for communication to other entities. As someone who (used to) travels a lot internationally, something like a vaccine passport (which they've done for yellow fever for years) makes it easier to verify someone's medical status. Possible to do with covid-19 positive tests as well, just maybe more complex, not sure.
Anyway, I get the feeling you and I could go back and forth for hours on probably any topic lol.
