OT: Corona virus

[obvert]

ORIGINAL: Canoerebel

Wait a minute, Erik. This graph does the same thing that the other one does - shows trends using numbers.

And this one suffers from a similar flaw, if read out of context or isolated. It uses raw numbers instead of cases per capita.

And it's flawed because it doesn't take into account increases in testing, as in the USA.

???

P.S. The graph is fine for what it's used for.

I just posted two graphs. Not sure which one you mean. I think you might mean the testing graph?

I originally posted the wrong pic of the testing graph, and now have the one up I intended. It is interactive on the site, but I couldn't embed that so just have the pictures.

Again, what I like is that it makes it very easy to see how countries relate to each other and see who might be in trouble more, as they fall well to the lower right, where a lot of cases without a lot of testing means probably a lot more out there.

[Canoerebel]

I was referring to your post 1635. That graph is no more useful than the one you criticized and possibly less so, as it seems to have several serious flaws.

[RangerJoe]

I understand that the US VA hospital system is asking everyone questions, taking temperatures, and no allowing unnecessary people into their hospitals. No children under the age of 16 either. This could also be a partial source of the increase in cases and tests performed.

[obvert]

London update.

It's Mother's Day in the UK. That means a flood of people into the sunny parks for both leisure with the family and exercise. We went out I the morning to hopefully find fewer people, but still cycled a bit farther than intended until we managed to get into a quiet woods. We distanced well. A lot though were out breathing heavily next to their jogging teams, and sitting around having picnics. I can't really fault them, but since the UK has ramped up testing case numbers have shot up, with a doubling in three days.

If that trend continues, we will be having a lot of trouble in a week. At 5k+ now, that means 15k+ in three days and about 45k+ by next Sunday. That is not good. I know it's hard for people to look ahead, but I'm worried now when I hear that there was some backlash to closing pubs and then seeing the numbers out in close proximity.

We are hoping to still get our daughter out, like everyone else, but to be very careful, get deeper into nature, and avoid any shops or groups and keep our 2-3 meter distance from everyone. We've had to wait numerous times for people to pass before moving in order to do this, but that's okay.

It's been lovely to spend so much time together, and I've been able to help my wife a lot with her new marketing strategy for the kids books she both writes and publishes. It all has to go online obviously, but some have been getting great reviews and we hope the public need some new books to share with their kids during the time inside.

There are warnings, as from the Mayor, Sadiq Kahn this morning, that things will be restricted more soon.

[obvert]

ORIGINAL: Canoerebel

I was referring to your post 1635. That graph is no more useful than the one you criticized and possibly less so, as it seems to have several serious flaws.

Ahh. Well, the good thing, as I mentioned, is that the FT graph correlates percentage increases while also showing actual number of case by country and tracking where they are in relation to a baseline 33% number. It tracks everyone from the same point, from 100 cases, which makes trajectories easy to correlate, and also shows more countries, which is very useful.

I find the upward curve to be more meaningful next to the 33% line of case increases, since this problem is still growing across the world, than the downward trend in the graph from Bloomberg, which is misleading visually. Without seeing number of total cases the percentage increase is less useful. Cases cold still be growing, slowly, or they could be holding steady with some recoveries (no longer listed as cases) and some new cases, keeping the total number high.

Each can be used for something, but to show downward trends at this point is misleading anyone who isn't aware of these factors and may not be aware of total numbers.

[geofflambert]

http://www.cidrap.umn.edu/news-perspect ... udies-find

http://www.cidrap.umn.edu/news-perspect ... d-severity

I'm curious about death-by-pneumonia for this. In the case of the flu, pneumonia is frequently caused by secondary infections, but nobody seems to have good numbers on how often that's the case. I'm wondering if Covid 19 is more likely to cause pneumonia directly and thus be the direct cause of death. Sure would like to know average length of time between infection and death, I'm suspecting it's shorter with this.

Copper is proving (as with other pathogens) to be a lethal surface to the virus. Wish all hospitals could switch from plastic, aluminum and stainless steel surfaces to copper, brass and bronze. In terms of doorknobs and push plates this would be a good thing everywhere and I hope building codes start reflecting that.

[Canoerebel]

Showing a downward trend in one set of data is misleading even though its accurate, but showing an upward trend that is clearly misleading** isn't misleading?

[&:]

These are Orwellian times.

** If read in isolation.

[Canoerebel]

Shame on Bloomberg for creating such a misleading graph!

[obvert]

ORIGINAL: Canoerebel

Showing a downward trend in one set of data is misleading even though its accurate, but showing an upward trend that is clearly misleading** isn't misleading?

[&:]

These are Orwellian times.

** If read in isolation.

That upward trend is not misleading. If it is please show me how?[&:]

Cases are increasing, right?

So isolating percentage decreases is trying to find something positive in something that is still quite negative.

BBfanboy also commented on this one originally ...

ORIGINAL: BBfanboy

I don't accept the premise of the graph because the early stages were the "discovery" phase when testing started and numbers were spiking because we finally knew what (who) to count. Once you know someone has it you don't count them in the subsequent weeks.

So the volume of testing must be factored in to the number of cases found i.e. the positive per thousand tests statistic we saw in an earlier chart. A plot of that stat over time would be instructive, with annotation of measures put in place on the time scale part of the chart.

[Canoerebel]

It suffers from two serious flaws: (1) using total cases instead of per capita (so that larger countries are inherently overrepresented), and (2) it fails to give the context that some countries (USA) have accelerated testing tremendously, leading to a big rise in cases (meaning the big rise was in test results, not in cases). So the graph is highly distorted.

Both graphs - the one you relied on and the one you criticized - are misleading when taken out of the context they were posted for. Both are fine for the purposes given, but the former suffers from less distortion/reporting bias than the latter.

[obvert]

ORIGINAL: Canoerebel

It suffers from two serious flaws: (1) using total cases instead of per capita (so that larger countries are inherently overrepresented), and

Huh? The one you posted has only five countries. The FT graph has 25 countries. and it's using total cases in relation to percentage growth and number of days since 100 cases. So yes, it is limited, but has quite a clear and precise rendering of what it is intended to show from the data.

(2) it fails to give the context that some countries (USA) have accelerated testing tremendously, leading to a big rise in cases (meaning the big rise was in test results, not in cases). So the graph is highly distorted.

That is true, but the actual text additions do give some extra context to the points, and I would guess that in future, if that showed up clearly as a massive continuing spike, they might post a note about that. The big rise in(positive) test results IS a big rise in cases, and to say it's not a rise in cases is very misleading. That's exactly what it is!

You can't show everything, but from this it's clear that Korea, Singapore, and Hong Kong have (so far) really limited both total number of cases and percentage growth.

[Canoerebel]

I found the presentation yesterday interesting. When it came under criticism for a variety of things, it caught me by surprise. My immediate thought was, "Oh, no, I posted something that was flawed. Darn it."

But as I've looked at the criticisms carefully, and read back through my original post, I've discovered that the criticisms are flawed and that those who make them have various levels of bias or misunderstanding.

That helps me to better understand perspectives going forward and knowing how to present data.

[RangerJoe]

Each graph shows the information in a certain way, in a certain context. If you want to find fault with it and state that it is no good, then I suggest that you make your own graph and post it. In the meantime, use the information presented with the caveat that it is not perfect.

No, I don't mean to insult anybody, either. If I did, I would do it directly.

[Canoerebel]

Number of recoveries in Korea has jumped.

About 7-10 days ago, Korea reported 51 deaths total, 118 recoveries total. Since then, 53 deaths and about 2800 recoveries.

Some countries are reporting strikingly low numbers of cases, including India and Russia. If true, why?

[USSAmerica]

I'm also very curious about the relatively low numbers from India. I'd love to believe they were accurate.

[witpqs]

A study regarding travel bans, social distancing, etc in China.

https://www.medrxiv.org/content/10.1101/2020.03.03.20029843v3

[Kull]

ORIGINAL: obvert

Question: where is the daily info graph Kull and others have posted that shows daily testing trends? Thx

For some reason they didn't publish a report yesterday (for 3/20), but we are back in business today. Since only one day was skipped, I obtained the totals from 3/20 and the per-province percentages from 3/19 and used that to "fill-in the blanks". As a result, the day-over-day numbers at the province level (and even north-vs-south) are a bit suspect, but still we can see that two trends seem to be taking hold:

1) The rate of overall case load increase is holding steady (between 14.6% and 14.9% over 3 days)

2) The rate of increase appears to be diminishing in the north, but rising rapidly elsewhere

Pertinent links:

- Daily Italian Stats by province here

- Original post with regional map of Italy here

- The last post on this topic here

[Kull]

ORIGINAL: obvert

Question: where is the daily info graph Kull and others have posted that shows daily testing trends? Thx

Actually you were probably looking for this chart. As noted in an earlier post, I find the Spanish totals to be highly suspect:

[obvert]

ORIGINAL: Kull

ORIGINAL: obvert

Question: where is the daily info graph Kull and others have posted that shows daily testing trends? Thx

Actually you were probably looking for this chart. As noted in an earlier post, I find the Spanish totals to be highly suspect:

Chart, yes, sorry! That's it. Thanks

[Uncivil Engineer]

I saw the article at the link a few days ago - an analysis of the passengers on the Diamond Princess that was quarantined in Japan (?) back in January (?). I don't know anything about the author or his qualifications (probably not medical), but this raises a lot of questions about the response to COVID-19.

https://wattsupwiththat.com/2020/03/16/ ... mysteries/

Some highlights - "some 83% (82.7% – 83.9%) of the passengers never got the disease at all"

"slightly less than half the passengers (48.6% ± 2.0%) who got the disease showed NO symptoms"

"There were a total of 7 deaths among those on board. All of them were in people over seventy"

7/~3635 = .0019257 or 1.9% death rate (3635 is my estimate from the first graph of the total number of pax. IIRC ALL of them were tested before they were allowed to leave the ship.

A question - the 83% who didn't get the disease; are they still carriers? Or did they test negative? The article doesn't say. I would assume that those who showed no symptoms WERE infected AND were carriers, and could infect others when they got home.