Thread by @charles_gaba

Q for #StatisticsTwitter:

My "White people make up 70% of U.S. COVID deaths since June" post has come under fire because I didn't weight the racial groups by age. U.S. seniors are massively overrepresented by White people, so if you weight it by age they drop to like 40%. 1/

I agree this is true, and I understand why it's important to weight by age/etc. when formulating public health policies and the like.

HOWEVER, I don't understand is why this changes the bottom line: That White people are dying of COVID at disproportionately high rate since June.

Let's take this to an absurd extreme: Let's say that 100% of White Americans were 80 yrs old, and 100% of every other racial group was 25 years old. If that were the situation, you'd expect something like 95% of COVID deaths to be among White people, right? 3/

Now, in this absurd example, would the reason 95% were White be *because* they were White? Of course not; it'd be primarily because they were elderly.

HOWEVER, that still wouldn't change the fact that 95% of those dying were also White. 4/

Also, in my extreme scenario, you'd expect virtually all of the elderly people to have GOTTEN VACCINATED, whereas you'd expect a far smaller % of the other groups to have, since they're all in their 20's.

But what if a huge chunk of the first group REFUSED TO GET VACCINATED?

To put it in political terms: Let's say that 100 elderly White Republican voters and, say, 50 younger Democratic voters of other races were to die of COVID in a given political precinct.

What impact would this have on the next election? 6/

Answer: In that scenario, there'd still be 100 fewer GOP votes & 50 fewer Dem votes, for a net reduction of 50 GOP votes.

You wouldn't "adjust" the 100 GOP votes down to 60 because "elderly Americans are disproportionately White."

This goes right back to the same criticism @greg_travis and I previously debunked in October, that I "wasn't adjusting for age" when running my county-level COVID death rates by partisan lean of the counties.

acasignups.net/21/10/06/rumor…

Now, if I was looking at the racial breakout data as part of trying to figure out how to ADDRESS the issue via targeted public health policies, then OF COURSE it would make sense to weight by age and other demographics to make sure I'm not wasting resources.

For instance, weighting by age would make it clear we should target protecting *older* Americans, not *White* Americans specifically.

Except...we're already doing that, and we've been doing that for nearly 2 years now.

Which actually makes the 70% stat even *more* stunning.

Again, as @greg_travis put it: Yes, seniors are much more likely to die *if they get infected*, but due to the fact that seniors are also far more likely to get vaccinated, wear masks, avoid crowds etc, it also means they're far LESS likely to *get* infected in the 1st place!

Put another way: The fact that 88% of U.S. seniors have gotten their 2nd dose (and 58% have gotten a booster shot) makes it even MORE stunning that White Americans are dying at such disproportionately high rates, *given* that they make up a huge % of seniors.