@MandaJo, that’s what I’ve been trying to tell him!
@Magiver, also note, that when you are using a moving average, you are not measuring how things look “now”, you are measuring the midpoint of your date range. So someone using a 10 day moving average is showing what the average was 5 1/2 days ago, a 7 day average is 4 days ago. This is fine if someone is analyzing old data, but if you’re trying to figure out what’s going on currently, it’s much more important to analyze trends than just the most current data point.
Florida. The issue there is not only that the death rate has climbed. Death rate rise follows diagnosis by about 3 weeks. We are seeing the now the impact of just the beginning of the rise in case numbers; over the past three weeks confirmed case numbers have more than doubled … and they may be moving from the lower risk younger population to a higher risk older population who are ignoring taking any protective measures as a political statement of faith … and spreading out to the rural population as well. This is their first wave and even if they got most of the population behaving themselves today some pretty horrific numbers are now pretty much cast to occur over the next three weeks. Since they are not doing much to alter behaviors, then the rise will likely keep going past that, to who knows where?
Louisiana is at a point that should be very telling. For them this is a second wave clearly triggered by opening up too much too quickly. Their total deaths/million to date are less than half of New Jersey’s and New York’s, but still eighth highest of all states. Total confirmed cases/million not far off the very top of the leaderboard. They are now about three weeks into their new wave of confirmed cases. So far not much of a rise in death rates. Any predictions of what will happen there?
FYI here’s a full quote of my post in the Covidiot thread on what at least our local county is trying to do to alter behaviors. This is only a couple days old but we shall see when/how much it pays off.
What I observe locally even before this latest order is only a few maskholes in stores. But I do see almost universal thinking that outdoors = 100% safe regardless of density or duration. In a region totally dedicated to year round outdoor recreation in groups that’s not really very safe.
Our other big gotcha is restaurants, both indoor and outdoor. Humans of course have to demask to eat; bad design if you ask me. But folks seem to think that means no masks once past the door for the entire duration of their stay. Not really.
My own view is that the issue of masks is just a distraction - sure they may help to some degree, but the real answer has already been demonstrated - its lockdown, its social isolation.
The use or non-use of masks is just a way to offload responsibility of leaders to make effective policy and enforce it to blame ordinary citizens.
The fact that the use of masks itself became politicized probably wasn’t predicted and in fact has actually made everything all the worse.
Quality masks that protect the wearer and reduce the infectivity coefficient (whatever the correct term of art is) by a factor of 5 or 10 slow the whole thing down enough that COVID does turn into the flu; a tragedy for the small percentage who have very bad outcomes, but not a society-stalling crisis.
Masks that are little more than merit badges are just that: badges. My own N95 worn properly and with 100% compliance isn’t COVID-proof whole-body armor. But it’s not zero either. For me or for the people standing near me at the store.
The neat part about masks is we know how to make those. And in quantity. Unlike vaccines. If we already knew how to make vaccines that’d be even better than masks for the same reason an IUD or Depo Provera is better than a condom. People actually wear their damn IUD. Mostly. Condoms? Not so much.
And yes, that US mask-wearing got politicized is a true disaster. For it’s own sake but for larger reasons as well.
We all just know that vaccine-accepting will be likewise politicized. Hell, there’s already a sizeable groundswell here that any future vaccine, regardless of country of development, technology, etc., is just a Leftist social control tool in disguise at best, and a genocidal sneak attack at worst. Thank you Mr Putin.
Ultimately, vaccines, like masks, are just a filter with some statistical level of capability. In any given year did I not get the flu due to the vaccine working in me, or due to the vaccine reducing the number of infectious individuals I encountered? No way to know. We (humanity) need both sides of the equation working for us to win this one. Fewer infectious people times less infectivity per infected person.
And yes, that times less exposure to other people through significant, but not socioeconomically suicidal, levels of physical isolation.
We saw breathless news reports yesterday about Oregon being a “hot spot.” Yes, our cases are increasing and we need to make it stop. However, per capital we’re 39th on the state list. I conclude that the frenzy is based on the change rate, not the absolute or per capita rate.
“All time per capita death count”? We’re talking about NOW. When I said “in the midst of of a growing natural disaster,” is that not clear I’m talking about what’s happening now?
Glad to hear but it happened only after local hospitals are near capacity and positivity rates are in excess of 16%. Already very late in the game. And not going to influence what happens in the more heavily Trumpian portions of the state.
[quote=“susan, post:326, topic:854191, full:true”]… I conclude that the frenzy is based on the change rate, not the absolute or per capita rate.
[/quote]
I am sure it is.
Reasonable question - given the potential for exponential growth, and the lag time between interventional changes and impact, how much should concern be based on per capita numbers and how much on directionality and its slope?
If you look at a 10 day average for New Jersey they’ve had a couple of large spikes that should drop off. “should” being the operative word as they should never have had the spikes in the first place.
Again, I’m not picking on New Jersey. I just pointed out their rank in comparison to Florida. New Jersey is still in play and is geographically linked to NY yet it goes unnoticed. Arizona is the area most heavily hit area by far but I’m watching the news as I type this and it’s a Brady Bunch news day. Florida Florida Florida. New York City got mentioned because of a street party video.
There is a reason why Florida is in the news and New Jersey isn’t.
Here’s the rolling seven day average of daily new case per million residents. Florida ranks #1, it isn’t close, and they are trending up. New Jersey is 45th and trending down.
Rank
State
Cases
WoW
1
Florida
552.45
+28.9%
2
Arizona
418.74
-15.9%
3
Alabama
387.12
+46.3%
4
Louisiana
362.21
-12.8%
5
South Carolina
358.29
+14.1%
6
Nevada
354.3
+45.1%
7
Georgia
342.71
+20.8%
8
Texas
332.6
+2.1%
9
Tennessee
320.68
+41.1%
10
Mississippi
308.5
+35.4%
11
Idaho
303.53
+21.1%
12
California
227.72
+16.6%
13
Arkansas
221.02
+0.3%
14
Utah
199.49
+3.8%
15
North Carolina
192.57
+16.2%
16
Oklahoma
190.51
+27.6%
17
Kansas
163.44
+17.5%
18
Iowa
153.18
-5.9%
19
Wisconsin
142.45
+25.6%
20
New Mexico
133.74
+14.8%
21
Missouri
133.05
+25.1%
22
North Dakota
124.47
+43.1%
23
Washington
119.07
+36.4%
24
Ohio
117.68
+24.3%
25
Maryland
111.98
+51.6%
26
Virginia
110.31
+41%
27
Nebraska
109.52
+26.6%
28
Montana
106.13
+55.7%
29
Delaware
100.64
-8.2%
30
Minnesota
98.77
-1.2%
31
Kentucky
97.94
+11.6%
32
Indiana
97.08
+25.4%
33
Illinois
86.71
+20.9%
34
Alaska
80.07
+49.6%
35
Colorado
79.31
+17.3%
36
Oregon
77.83
+19.6%
37
Michigan
71.06
+31.1%
38
Wyoming
66.4
+15.5%
39
South Dakota
65.88
-4.2%
40
Rhode Island
64.86
+49.8%
41
Pennsylvania
61.94
+4.6%
42
West Virginia
61.86
-17.5%
43
New York
47.78
+25.9%
44
Massachusetts
38.14
+17.9%
45
New Jersey
36.41
-9.7%
46
Connecticut
24.28
+6.3%
47
New Hampshire
17.23
-1.8%
48
Hawaii
15.54
-23.4%
49
Maine
13.39
+2.4%
50
Vermont
12.59
+22.2%
51
District of Columbia
2.02
-9.1%
Also it has been explained why a ten day average is a poor choice, but I’ll give it another go. The data quite clearly shows an underlying seven day periodicity (weekly seasonality) therefore it is best practice to average over a period that is a multiple of this periodicity. Use a seven day average, or if that is too noisy blow it out to 14, 21, … For most of the states we are dealing with seven is fine, but you can get some weird results in the lowest population states. The conversation isn’t really about these tiny states, so seven across the board is suitable for almost all analyses.
The question narrowed down is concern for what? People dying? People getting the virus? ICU capacity? Medical delays for other diseases? Financial lives ruined? Suicides due to financial or emotional distress?
We’re talking about a virus with an extremely low casualty rate along with an aggressive contagion rate, We also have a recognizable demographic group most susceptible to injury or death.
Per capita numbers are the best way to assess the situation as it relates to the population.
As for what to look at it depends on the accuracy of the information. Death counts aren’t perfect but if the same metrics are used to determine them then they produce an accurate but late trend. ICU capacity produces a real-time trend and can be acted on immediately to relocate patients. That leaves covid test results. These are the least accurate numbers but the earliest indication of what to expect.
Different data should be looked at as a group and not singular points of data . It should also be screened for accuracy to look for shoddy data.
Some of the reporting is getting a bit strange. I saw a map where they were reporting solely on how much cases have dropped. I was wryly amused because it showed NH has only dropped 0-10% over the past week, which is less of a drop than a few other states. Well…when you’re reporting 20-odd cases a day, there isn’t much room to drop…
I laughed out loud at a news reader who said a study of 65,000 cases showed that children could get the virus and spread it. Um… OK… we knew this since forever.
Do these people not read this stuff ahead of time or have any input on what goes out to the public.
the idea of a trend is to smooth out the numbers. we already know it has a weekly up and down. running it out in multiples of the cycle doesn’t produce anything useful and it also doesn’t really matter as long as you’re using the same spread for all the states.
To put it another way, ALL the states appear to follow the 7 day cycle so regardless of how many days you use it comes out the same. The difference is how many days you use and I chose 10 because it seemed to smooths out numbers on any given day.
To take away seasonality from a series so we can better see trend, we would use a moving average with a length = seasonal span. Thus in the smoothed series, each smoothed value has been averaged across all seasons.
This is just the first cite I found, but there are many. This is pretty standard data analysis stuff so it won’t be hard to find a bunch of cites. It’s not going to make a huge difference, but a ten day average will include meaningless noise that can easily be canceled out with a seven or fourteen day average for these particular data sets.
For example, the gray daily line is quite noisy and we observe weekly seasonality. For this reason the the red seven day average line is slightly smoother than the green ten day average line. A fourteen day average would be smoother still, but may lag behind current trends.
All that said, no matter which average you use, it is obvious that things are right now as bad as they’ve ever been in Florida, and getting worse rapidly. That is not the case with New Jersey at present and that’s why people are talking about Florida as opposed to New Jersey.
Of course they are all connected. The metrics under consideration inform little about financial lives ruined, about the real morbidity and mortality impacts caused by mitigation choices, so on … but they do, in aggregate (and yes per capita is the meaningful level of analysis) give a sense of where things are and where they are going locally in the next several weeks at least.
The questions of concern include the stability of current rates, for some how much hope they should have that rates can be driven down to very low levels and kept there by way of contact tracing and isolation, and for others how much there risk there is of exceeding regional healthcare systems capacity with much higher death rates from all causes resulting from that.
So to use @susan’s Oregon as an example - compared to other states their current rates are very low (deaths per week < 1/million, new cases/w under 80/m, so on) but the directionality and slope in what for them is really a first wave, still reasonably has many more concerned about the next several weeks for them than those same people would be concerned about, say, New Jersey today, with death rates nearly three times as high but dropping and with lower and dropping new cases rates.
I just have to comment on how interesting it is that the weekly average is not just a balance of noise to recency but actually LESS noisy than the 10 day numbers. Thank you for explaining and demonstrating this!
