Craig, do you know if the CDC death data backfills in for recent days, causing recent 7-day avgs to be artificially low? They have a footnote to that effect, but I don't necessarily trust that since it seems like boilerplate language they have on every page associated w/ COVID.
I'm trying to better understand the gap between CDC and NYT daily death averages (373 CDC / 511 NYT).
If you're not sure, maybe a tweet out could get a response from other data nerds I know follow you (eg @CohoKelly).
It used to be the case that deaths would revise upward, but lately not much movement at all. In fact, if you look at my previous weekly updates, the number has dropped as the 7DA settled over time:
3/29 update: 7DA deaths as of 3/25 = 705, today CDC shows 698
4/4 update: 585 to 580 (4/1 data)
4/11 update: 516 to 501 (4/8 data)
I think it may be a combination of (a) some states' probable deaths making it into NYT and not CDC, and (b) the CDC allowing states to allocate older deaths to other dates over time if they have a backlog (or mini) dump in a particular date.
Thanks for the reply - that sure is a big gap appreciate your two hypotheses. I wonder if a 3rd could be some with/from parsing happening either at the State or CDC level that NYT is ignoring, but that's total conjecture. Either way, all of those point to the CDC number being more reliable, which I assumed but good to have higher certainty.
Craig, do you know if the CDC death data backfills in for recent days, causing recent 7-day avgs to be artificially low? They have a footnote to that effect, but I don't necessarily trust that since it seems like boilerplate language they have on every page associated w/ COVID.
I'm trying to better understand the gap between CDC and NYT daily death averages (373 CDC / 511 NYT).
If you're not sure, maybe a tweet out could get a response from other data nerds I know follow you (eg @CohoKelly).
Thanks for the updates,
Jason
It used to be the case that deaths would revise upward, but lately not much movement at all. In fact, if you look at my previous weekly updates, the number has dropped as the 7DA settled over time:
3/29 update: 7DA deaths as of 3/25 = 705, today CDC shows 698
4/4 update: 585 to 580 (4/1 data)
4/11 update: 516 to 501 (4/8 data)
I think it may be a combination of (a) some states' probable deaths making it into NYT and not CDC, and (b) the CDC allowing states to allocate older deaths to other dates over time if they have a backlog (or mini) dump in a particular date.
Cheers!
Craig
Thanks for the reply - that sure is a big gap appreciate your two hypotheses. I wonder if a 3rd could be some with/from parsing happening either at the State or CDC level that NYT is ignoring, but that's total conjecture. Either way, all of those point to the CDC number being more reliable, which I assumed but good to have higher certainty.