COVID mortality

The 8/8 meant 8th tweet of an 8 10 tweet thread in which he builds up the Simpsons paradox math. Sorry for that part not being clear. I thought the whole thread was neat but only copied the last image here.

Good question on the time period. It looks from his article on this that it was current hospitalizations as of 8/15/21 (?).

He was trying to address the (dumb) question, “if the vaccines work, why are most people hospitalized in Israel vaccinated?” That question is really an issue of the base rate fallacy that @meep has posted about, but splitting the data by age reveals a nice example of Simpson’s paradox.

I happen to be doing my next blog post about this. I looked into the prof’s numbers, and they are consistent (I can’t read Hebrew, and the autotranslate is inconsistent in operation, so I’m trusting the prof read the numbers off correctly… but it all hangs together, and I can replicate some of his ratios)

I did some graphs of his tables:

He preferred to use 30-39 unvaccinated as his reference point, but I decided to use 50-59

4 Likes

Anyway, looks like the risk reduction ranges from 81%-100%, though there should be error bars on all these rates (especially the 0 rates)

1 Like

How different are the COVID rates for the vaccinated from anything else that might send someone in poor health to the hospital or worse?

Splitting hairs a bit, I wonder if there is a small-ish bit of base rate issue within the deciles? If you could split into age 40, 41, 42, and so on, the numbers could nudge up a bit.

Here’s my post:

Seriously, I’m not going smaller than 10-year groupings for Israel. They had only 500-ish people hospitalized w/ COVID in this data set.

2 Likes

[nitpick]You’ve split the ages into 12-50 and 50+, so either age 50 is in both groups or something is mislabeled.[/nitpick]

Good graphs / visualization.

I am seriously not using < > signs nor ( or [

(If I really want to get persnickety, I will say nobody is exactly 50 years old, due to quantum limits on time measurement…)

Anyway, thanks.

4 Likes

Oh, a lot of folks would just say 12-49 and 50+ or 12-50 and 51+ depending on which was applicable. No special characters needed. I think ALB is implied when working with non-actuaries.

Had to look that one up, learned a new TLA today!

1 Like

What’s “ALB”?

1 Like

Age at last birthday. Just the number you’d give if someone asked you today how old you are.

3 Likes

Funny ALB/ANB story with me the actuary in this story and my fiance at the time (now my wife) who worked with actuaries. We had our pre-marriage counseling with our minister. One of the first questions we got asked was how old the other one was. I used ALB like just about everyone in the world would and my fiance used ANB so we got different answers on our papers. The minister was very surprised that we didn’t match and we had to explain the two calculations to her.

2 Likes

But, but, but, Pi to the 62.8T digit says?

Quotes that make you say WTF?!?!

I would never abbreviate that - is ANB Age Next Birthday or Age Nearest Birthday?

Yup. A lot of mortality tables are ANB (age nearest birthday), and that’s often how life insurance is sold. So if you’re buying a new policy, but it before your half birthday as that is the date where the insurer will consider you a year older.

I think this is becoming slightly less true, though. But still mostly true.

Oh, interesting. I sort of assumed they’d use ALB but that makes sense. I’ve only purchased life insurance through my employer where they ask almost no questions (and my DOB is probably sent over by my employer), and I’m phasing that out, so I never knew much about the life stuff.

I am intending ti attend this Webinar

Israel’s 4th Wave The Delta Variant in Arab Society - with Dr. Salman Zarka and Aiman Saif

3 Likes

Group life insurance (including Additional Life and Spouse Life) is often based on your age (ALB) at the end of the year, which is what the IRS requires for imputed income purposes for the employer-sponsored portion of Basic Life. (Now a group and their insurer can calculate age any old way they want but it’s often easiest to just use the same calculation the IRS requires for imputed income.)

I guess I didn’t make it clear that I meant individual policies were ANB.

1 Like