They low-key are in jurisdictions that implemented citations for being mask-less in public...
2019-2020 coronavirus pandemic (Vaccination 2021 Edition)
- Thread starter McElroy
- Start date
Tracking the coronavirus around the U.S.: See how your state is doing
View NPR's maps and graphics to see where COVID-19 is hitting hardest in the U.S., which state outbreaks are under control and where cases are still spreading.
www.npr.org
Why would anyone in their right mind shade that map in that way? Percent change relative to itself is worthless information when everywhere experiences the pandemic so differently. You could make a map where blue meant more people have died per capita and red means fewer have died per capita and create a nearly identical map.View attachment 332
Tracking the coronavirus around the U.S.: See how your state is doing
View NPR's maps and graphics to see where COVID-19 is hitting hardest in the U.S., which state outbreaks are under control and where cases are still spreading.
I'm doubly confused by your remark...
1) It is not worthless if the purpose of the map is to display where the coronavirus is on the rise and where it is going down. Which is obviously its goal based on the map's title.
2) Why would you color a map in blue and red and more specifically color "less deaths per capita" red?! Common color convention would suggest red = worse, green (or blue I guess?) = better. Sounds like you're trying to politicize the hypothetical map...
Maybe they shaded it that way because they want to convey whether cases are going up or down by state, as the title clearly indicates they want to. It performs a perfectly good job of doing so.
2 first cause it's a simpler answer: I just meant color it that way to match their map. If you shaded the intuitive way, red bad and blue good, to map out which states have suffered the most, it would be mostly inverted from that map.
For 1: if it was just a binary "where is rising and where is falling" it would honestly be more informative. Shading darker or lighter in the way they did is genuinely misleading. What states look worst? That corner of Montana, Idaho, Oregon, and Wyoming looks pretty bad. But it's actually got less of an increase per capita than most of the less shaded states. Oregon is in the list of examples, the increase is 4 per 100,000. That's a third of the increase per capita that California has, but they look the same, and they both look less hit than those other 3 states despite being hit harder. Those 4 specific states have the fewest covid deaths per capita in the contiguous US, but that map makes them look like they're the same as Arizona right now.
Like, look at Hawaii. It's the deepest red possible. And yet, in that article, NPR says "Some states, such as Hawaii, have declined to just a dozen new cases per day, while others are growing." How does that make sense? It's because Hawaii was down to 1-3 cases per day a few weeks ago, and now they're at 10-15 per day. That's like a 500% increase over 2 weeks, that they tactically decided to exclude from their list of states with rising numbers.
Long story short: they really should have shaded based on the "per 100,000" column.
It does not state that the per capita value listed in the chart is an increase, though. The per capita value is how many people have been infected daily in the last two weeks.
i.e. Oregon has had a weekly average of 175 people a day infected out of a population of 4.2 million. 175 / 4,200,000 * 100,000 = 4 per 100,000 infected per day, just like the data correctly states.
I'm not really sure how removing the different shadings of orange and blue/green would make the map more informative and less misleading. In order to be misleading the map would need to pretend doing/showing something it is not which I do not feel it does. Sure going up, down or leveling off can be described in many ways but percentages of growth is an entirely valid method. Using different shades of colors to show different degrees of growth based on the method used seems to be entirely valid as well and actually makes it more informative as it directly shows where percentages are higher/lower. Off course percentages are percentages and there are pro's and con's to using percentages versus amounts of cases, one could also argue about using growth of cases versus growth of cases per capita, etc. But none of these discussions should lead to the conclusion the map is deceaving, it should lead to the conlusion the map alone doesn't give sufficient information.
Because there's more to effective communication than a technically accurate label. It's like the old "zoom in the bar graph so that a 1% difference between two values is half the graph". It doesn't matter if the axis label is accurate, it matters what the image conveys. This isn't a technical resource where the data is what matters, this is a news graphic from a news organization where communication is everything. "It's not a lie" is not a high enough standard.
Let's say the graph shows the change in infection rate by per 100,000 people.
In state A, the daily infection rate goes from 10/100,000 to 15/100k, an increase of 5. State B goes from 1/100k to 6/100k, also an increase of 5. So it looks like things in states A and B are eqally bad by your measurement.
However, because of the way trends work, if these trends in increase in infection rate continue, 2 weeks later state B will have 36/100k infected per day where state A will be 23/100k. A month later, it will be 208/100k in state B and 34/100k in state A. So would we actually be right to have said that things were eqally bad in those two states?
So what makes your way superior, then?
You don't believe it works that way. You just don't. You don't believe for a moment that Hawaii having 5x as many daily cases as two weeks ago means that in 2 weeks from now they'll have 5 times over again.
My way is superior because it would actually illustrate where it is spreading or receding the most, which is what people want to know. Percentage change is a number that combines two different variables, the increase/decrease and the magnitude of the current spread. By combining those two together, you look at the map and cannot discern either. It's like if you combined calories per package and price per package into price per calorie; there might be some value in a very specific analysis if the raw data is there with it, but to a normal consumer looking, you've managed to conceal the information they actually want.
Why not?
To use a real-life example, using number of deaths as a representative proxy for infections, in the USA around the 10th March there were ~10 people dying per day of covid. Around the 24th March there were ~250 recorded covid deaths per day in the USA. On the 7th April, there were over 2000. It levels off around then in large part because of lockdowns in late March to restrict spread, peaking around mid-late April. Without a lockdown, who knows were it might have got to? Around or even over 10k is far from impossible.
In other words, a fivefold increase in infections over a fortnight is a) plausible and b) sustainable for a protracted period of time (from a relatively low base, at least), and an increase of that size should be a sign of considerable concern.
It's not superior, though, it's just different. Horses for courses: think about how to present data in a way that best emphasises what you most want to explain. You want to explain something different from NPR. That doesn't make it "better", it just means you want to convey a different message.
When I feel people have lost themselves in the argument for arguments sake and are no longer taking stock of what they actually believe, I'm inclined to do that.
Because a bell curve is only exponential for a short period, and especially in these midwestern rural states, outbreaks are localized. Hell, that's why they've avoided outbreaks so far to begin with, it's not an interconnected homogeneously mixed population in Idaho.
What message do you think NPR is conveying with that map? Do you think it coincidental that people are ragging on red states at the moment, as if they handle the pandemic worse?
Oh, is that why your underwear was knotting up over the color choice? As someone who's worked with professional cartographic software, I can say that a spectrum from blue to red is an extremely common choice for choropleth maps like the one being debated, especially when one end of the variable displayed is "bad" and the other is "good". It's very intuitive, very striking, and generally has nothing to do with political affiliation.
The map shows rate of change of new cases. It does so effectively. Tough tiddies if you don't like it or if it makes "red" states look bad. You can argue up and down what the "most useful" statistical variable to inform the public is, but the rate of change of viral spreading is an important variable to inform the public of.
I deduct points from the map for the choice to make states uniform shapes. This only serves to cut out geographic relationships that could be observed, such as whether larger or smaller states tend to trend towards a higher case gain rate. It also makes it more difficult to evaluate regional patterns, when you obscure which states bounds which other states. At the very least, what it gains in simplicity it loses in confusing the location of certain states (North and South Carolina are east and west of each other, Washington and Oregon are not sharing a boundary, etc). I don't know whether to blame deliberate obfuscation of data or new media's fetish for simple geometric patterns for an aesthetic sense.
No, that's not what I'm getting at. What I'm getting at is that Republican states overall have done objectively better through the pandemic (admittedly largely by circumstance rather than policy), and NPR chose to display a gradient metric that effectively punishes states for having a low case count, when there are comparable statistics that could more accurately warn people where the danger is highest without that flaw. People are already chomping at the bit for excuses to say New York is sooper dooper and the midwest sucks, and NPR is feeding it to them.
The map doesn't really show rate of change. It shows percent increase or decrease. To figure out the rate positive cases are increasing or decreasing from that, you have to know the current statistic. One household in Hawaii getting infected would cause the same percent increase as an entire small city getting infected in California. Would you say those places would be changing at the same rate?
It makes more sense on one of the sites I've been linking, because in addition to the color shades being more precise and based on an absolute scale rather than percent, it also has every state represented by the graph of new cases over time, all with the same y-axis scale per capita.
America Is Reopening. But have we flattened the curve?
See New Case Trends in all 50 States
coronavirus.jhu.edu
I think it's common sense public service to let people know that an infectious disease is spreading rapidly in their state, because it might encourage them to take extra care.
I think arguably some of the red states locked down too early. After that, I suspect what earns them some sneers is the attitude some of the governors and activist (mostly right-wing?) groups have had. For instance, I think the governor of Texas on Monday made a big noise about opening up his state, and on Tuesday looked at the figures and promptly reversed course (e.g. Houston hospitals virtually at maximum critical care capacity; the paediatric hospital's facilities are now full of adult covid cases).
I was going to post the numbers in Texas too. It's not like most states that are looking bad are looking bad because they went from like 10 cases to 20 cases.
Compare to the John Hopkins version linked a couple posts up, look at how much more useful that is.
I agree with what you're saying here, but that's not the understanding that most people express. Like, to an extent, they couldn't have known early on that they were locking down too much or too early because the virus was such an unknown, but they should have loosened restrictions partway a long, long time ago and accepted a full open was going to be a more distant future than places that got hit harder. Holding down to nothing and then just releasing the floodgates makes the holding down to nothing utterly worthless.
So like, people in rural states protesting heavy handed lockdowns, the "open up" attitude, were mostly right. The chorus was "we shut down to keep the hospitals from filling, the hospitals are empty, let us go back to work". And the mature, informed answer should have been "ok, but be prepared for more lockdown when it comes later." And like, it's not like bluer states aren't opening now. It's not just Texas. PA is basically all green zone tomorrow. But that's a luxury you get only when your active case count is low AND your potential super spreaders have developed immunity.
Most of the states that are looking bad in that hexagon map are going from like 1 case per 100,000 to like 2 cases per 100,000. There aren't 27 states going to hell, it's like Texas, Arizona, Mississippi, South Carolina, and Florida. The majority of states increasing in count are low, low curves with modest upticks.
Several states are doing worst than entire countries though. The US is just not doing a good job handling the virus from the government down to the public as well. It simply takes common sense measures to keep the virus in check and it's not being done in many areas.