David Greaber talks about this in BS Jobs. There’s an expectation that you will be paid less and treated worse if your job is productive and fulfilling. How dare you have a rewarding job and want more money! You love teaching at a preschool and would do it for free? Ok, be happy with $15k/yr.
TAKE: Time-lagging to determine CFR is a bad idea. Better procedure:
- Go back to the mortality trough in the summer (t0).
- From then until now, add up all cases and all deaths.
- Estimate the number of deaths attributable to cases that were confirmed as of t0. You could probably use hospitalizations times some factor as a good proxy.
- Estimate the number of deaths expected from open cases. Again, hospitalizations can be used to make a guess.
- Subtract 3 and add 4. And you’re done.
The problem with lagging is that you are not really looking for an instantaneous estimate of CFR, so you will end up averaging your daily estimates. And if you average your daily estimates, you are just doing the procedure described above, dropping N days of deaths after t0 and N days of cases up to the present. But rather than spend all your time drawing graphs and eyeballing the CFR from them, why not use it to come up with a better estimate of the tail adjustments rather than the gorilla math of just dropping an arbitrary number of days?
Joog,
I use two sources. For national and state level data, I use this file:
https://covidtracking.com/api/v1/states/daily.csv
I have a script in Stata that just automatically imports the data and generates the figures/graphs that I look at.
For county-level information, I use these files (one for cases, one for deaths):
https://github.com/CSSEGISandData/COVID-19/raw/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_US.csv
These files were a pain in the ass because the variable names aren’t intuitive and you have to reshape the dataset to do the things you want.
The NY Times is also maintaining data here:
https://developer.nytimes.com/covid
and my guess is that it might be more useful/easier to handle than the county-level sources I posted above. But since I had already taken the time to write scripts for those sources, I haven’t looked closely at the NYT stuff.
The NY Times also has data for excess deaths, in case you want to try to convince any morons that, actually, COVID is killing a lot of people and not just killing people who would have died at the same point anyway. That’s here:
1 Like
this is actually thinking about it backwards. it’s not like someone set the “correct” rate for all jobs and then after that some evil cabal figured out “hey we can fuck the rubes here doing what they love and cut their pay” and adjusted teacher pay down.
1 Like
I think I understand what you’re saying. It sounds like:
- Assuming that CFR is constant, you shouldn’t try to estimate that CFR with short windows of the deaths/lagged cases relation when that lag is unknown.
- Better to measure the relation using longer windows because those longer windows will necessarily increase the proportion of true overlap between the deaths and the cases that led to them. (Interestingly, this is exactly the path that you see in accounting/finance research that studies the relation between earnings and stock returns. Maybe not interestingly.)
- You’ll still have to make an estimate to account for both sides of that contemporaneous relation, but that estimate won’t dominate the estimated CFR the way it does in the short-window calculation.
I’m somewhat averse to this approach because I don’t believe that CFR is constant - one of the reasons I was interested in calculating an estimated instantaneous CFR is because I believed that the demographics of new cases would shift dramatically once schools started opening. (I’ll admit that this is inconsistent with my view that 21-day lag is the right one to use because it appears to be most consistent over time.)
In any event, here’s what I think you’re suggesting. This was very rough just based on eyeballing data from worldometers:
-
Identify the trough as roughly 7/1. Since that date, we’ve had 138,529 deaths and 10,469,841 cases. That’s a starting CFR of 1.32%.
-
Subtract from that number the deaths (early in the window) that were not due to post 7/1 cases. That requires some estimation, but I took new cases over the 21 days prior to 7/1 and multiplied them by Estimated CFR to get the deaths to subtract (1,414).
-
Add to that number the estimated deaths attributable to post 7/1 cases that haven’t shown up in the data yet. Again, requires estimation. I took last 21 days of cases (as of now) and multiplied them by the same Estimated CFR as in step 2 to get the deaths to add (6,499)
-
138,529 - 1,414 + 6,499 = 143,613 deaths attributable to the 10,469,841 post-7/1 cases, for an estimated CFR of 1.372%. (I iterated the numbers until the Estimated CFR converged to the total CFR.)
1 Like
That’s not in any way implied.
“how dare you have a rewarding job” is absolutely implying it
No.
That sentiment doesn’t imply a secret cabal. You make this secret cabal accusation like once or month or something. It’s absurd.
The sentiment is not just wild speculation either, it’s documented sort of more than anecdotally in the book which is based largely on an obviously not rigorous (still not speculation) survey.
How many times does it need to be explained to people that classes of people often have similar feelings and the members act in similar ways without any kind of secret cabal?
Jesus, dude, the point is that you’re anthropomorphizing market forces. The “cabal” is just embellishment, the central point is that you are still fundamentally not comprehending how things work.
I guess the answer to my question is that it can not be explained enough times because some people are unwilling or unable to understand it.
And note my disagreement isn’t with the work you’re citing. It’s with your interpretation of it.
Sure, in a sensibly run country where bars are closed and restaurants are outdoors and distanced, that’s a difference well worth exploring and hoping to be able to take advantage of to minimize the spread while maximizing educational opportunities.
Here in America…
And, uh, so?
The automobile market wants SUVs.
OMG, that’s anthropomorphizing market forces!!!
why are you talking about this nonsense here?
Because the subject of nurses working so hard and being so dedicated that they not only risk their lives during a pandemic, but are loathe to press for better pay because it means threatening to not help people came up.
oh cool, do you want to go over how the idea nurses don’t press to get paid more is ludicrously false or can we drop it now?
y’all are taking something that seemed to be pretty specific to the dutch and assuming it’s true everywhere when both of ya have no idea what you’re talking about whatsoever when it’s tangential to covid at best.
sorry for being short but tbf this is a really annoying derail.
Well, we did accept it as fact that two people on this board who are married to nurses knew something about the condition of their spouses’ employment.
1 Like
And, I’m trying to be nicer, so yeah, instead of getting short myself, I’m just not going to go out of my way to stop annoying you.
1 Like
Dude there’s an entire industry of nurses that are basically traveling free agents that move contract to contract. Turnover is common. Nurses labor actions are not uncommon whatsoever. Your assumptions are just flat wrong and uninformed. Nurses aren’t so altruistic that they don’t stand up for themselves.
Nurses aren’t all underpaid but a lot of the non-contract ones are. Here you get about 40k/year straight out of nursing school which is roughly what you could make sorting packages at Amazon.