Okay. I already did that in recent memory, which is why I felt confident stating what I did.

Since you gave a statistic for car accidents, I'll do the same for the covid side: https://www.cdc.gov/nchs/nvss/vsrr/covid_weekly/index.htm

Since the start of 2020, there have been 576 "All Deaths involving COVID-19" among people under 18 in the US. Compared to 60,811 deaths from all causes, that's slightly less than 1% of all child deaths.

The 576 covers almost two years, so let's call it 300 deaths/year. Not accounting for the slight difference in age ranges, that's 1/8 your number of 2400 deaths/year from car accidents. I'd call that roughly one order of magnitude lower. If you clump in children under 13 in your car accident statistic, I suspect it would fall below 1/10.

Is my math wrong or is your math wrong? If my math is wrong I would love to understand why.

I think your math doesn't distinguish between rates and amounts: indeed, if every kid in the US got COVID and only 600 died, then sure, you could make the claim that it's less risky than driving. (Also, the pandemic would be over, rendering this whole conversation moot.)

The truth is we don't know how many kids in the US have had COVID, but I'm pretty sure it's not 100% of them -- which is why the number I cite is the estimated case fatality rate and not just the total number of cases.

If you throw in an unknown scaler and fudge it, then yes you can make the statistic do whatever you want. I'd argue it's not a relevant or useful statistic, though. It's analogous to you telling me my risk of dying from a bullet to the head is near 100%. It's technically true, but I'm not going to super glue a kevlar helmet to my head.

What’s the unknown scalar? We know the case fatality rate.

Oh, my apologies, I originally misinterpreted one of your earlier messages when you brought up underreporting.

I understand where the 0.05% case rate number comes from (reported deaths divided by reported cases). I do personally think the reported deaths number is probably slightly over-reported and the reported cases is significantly underreported.

Even if the case fatality rate is accurate, I just don't think it's useful when assessing personal risk unless you routinely go out of your way to catch covid. If you start getting into that level of detail, you at least need to correct for comorbidities as well.

Specifically, when I say that it's an order of magnitude less likely for a teenager to die from covid than a car accident, I don't mean a teenager that caught covid or a teenager that was in a car accident. I mean a randomly sampled teenager out of the 60 million or so in the US.

I was thinking about it a bit more. Apparently the fatality rate of car accidents is 0.7%, so comparing that to a covid case rate of 0.06%, it seems like it's still an order of magnitude less fatal for a teenager to get into a car accident than to catch covid. It's been an interesting discussion, and I've enjoyed diving into the numbers more. Thanks!