COVID Success Stories by State: More Fun in Reading Graphs

This is the last (for now) of my posts about COVID and reading data/graphs. I'm returning to books.

The first post was about possible optimism in the spread of COVID-19 in the U.S.

The second about the political nature of our response.

Here are two graphs of the COVID spread, comparing two states. Clearly, New York has had a way better response than Wisconsin. Right?

Is there anything wrong with this comparison? If so, what?

I am using the same criteria as I had in the earlier post about possible optimism in the spread of COVID-19 in the U.S.

There are no numbers. So we don't know what's being compared. If anyone hides numbers and properly labeled axes, don't trust them. It looks like Wisconsin could still be climbing while New York is improving. That may be the case. However, take a look at the properly labeled graphs below:

Wisconsin seems to be wavering between fifty to sixty cases per million per day. Meanwhile, New York may be descending toward one hundred cases per million per day. So is Wisconsin doing better than New York?

For about twenty days, Wisconsin was hovering at twenty-five cases per million per day. Should that be its normal? Something else? Some states are hovering between ten to thirty, though: West Virginia at ten, Washington state at thirty.  What should it be?

Here are some states or territories that are whooping all the others' butts:



(Apologies for making the graphs small, but here I'm just looking at trends, not numbers. Feel free to make your own graphs to check.)

These guys are at or near zero. Why are they so much better? You'll probably guess as soon as I list them. Respectively, they are Alaska, Guam, Hawaii, Montana, Northern Mariana Islands, Vermont, and Virgin Islands.

A lot of them are islands (with ocean barriers), and/or have mountain/snow barriers, and/or have low population density. In fact, when numbers plateau, that may be that state's or territory's best rate, under the circumstances. Population density may explain the different-sized plateaus we're seeing (for instance, the difference between Wisconsin and New York). Still, can these be improved?

Puerto Rico, an island, throws a monkey wrench in this discussion. Perhaps they are less prepared to accommodate to this virus than other islands. There are mainland states who still have climbing rates as well, so they need to investigate ways to work on this, brainstorm with allies.

The graph for Northern Mariana Islands is interesting in that the case load hovered at or near zero for ten days, then popped back up. This could mean that we will not eradicate this virus. At least it needs to be for longer than ten days. How much longer? And what does that mean for us?

We can try two things: One, open up the country, business as usual but with masks, etc. (except working with masks is problematic, especially work that requires exertion, and brains need oxygen. Will there be later class-action lawsuits against businesses with strict mask policies? But do we care about the health of such people's brains? They're just low-wage workers, right?) Perhaps we need better masks or a switch to face shields, as silly as they might look. We will probably wear the masks until a vaccine is developed.

Two, we can try a near-complete shut-down of the country, save a bare-bones medical workers and bare-bones crew of those running power and a bare-bones police force. As you can see from the aforementioned post about the political nature to our response to COVID, most people are willing to wait until June, so we can try four more weeks of shut down. People would have one day to gather food for a two-week hibernation. If you had any symptoms after two weeks, you have to stay in hibernation. There would follow the final two weeks of partial lockdown. Then open the country up. Period. No masks unless you or your business wants to, except when an outbreak occurs within a company. If it doesn't work, it doesn't work. We live with the consequences.

Of course, people would complain about either action, and maybe there are other avenues to explore.

This last bit is just some ideas that occurred to me to curb economic repercussions. They have nothing to do with graphs.

To limit business failures, if I were President, I'd make it illegal to collect rent [UPDATE: I'd use a carrot-and-stick approach--either collect rent and face full taxes without deductions (this includes all members of a board. These taxes would be distributed among those who lost their businesses) or allow the rent-free period and receive generous tax breaks]. All electrical power is charged at cost. Anyone who has already paid rent, can get refunds or future credits, depending on the business owner's needs in order to reopen. Those who own buildings or land that's being rented can deduct this off their taxes. Those who had a lot of perishable waste can deduct that as well. That's as close to a reset, that I can think of.

I'd probably do the same for low-income people who lost jobs and defaulted on rent or house payments. Middle-class can put off payments (interest free) if necessary (i.e. unemployed).

Other ideas to mitigate income loss for people and businesses?

The Political Nature of Responding to COVID in the U.S.

This continues discussion begun in "The Rate of Covid-19 Spread in the U.S.: Optimism in Reading and Interpreting Graphs".

A friend pointed out that I had posted this COVID graphing website (Flip the Script) earlier, and at that time, the U.S. was in the middle of most countries battling COVID. Back then, many news outlets compared numbers without adjusting for different population sizes, which led to wide-spread anxiety. The friend felt relief when she'd seen the graphs adjusting for different populations. It's possible the news outlets didn't know any better.

However, countries (mostly smaller in size and population) have done better than the U.S. at limiting the spread. I discussed this earlier, albeit indirectly. How this should be resolved is a present debate in the U.S.

According to one poll (see NPR website on April 23 based on a poll taken between April 15-20), some states are opening about when (or before) a majority of agree when the U.S. should open. I have organized NPR's data which was strangely out of order, probably confusing most readers trying to make sense of the data:

• 3% want it open now, meaning April 15-20 [3% of the population*]

• 14% can wait less than a month [that is, 17% of the population want to be back to normal by April 21 - May 15ish]
• 37% can wait for it to open in a month or three [54% of the population ~ May 15 - July 15]
• 10% can wait for it to open in four to six months [64% of the population ~ August 15 - October 15]
• 34% can wait for more than six months [98% of the population  ~ November +]

Why would anyone organize the data any differently?

This means that most of the population wants us out and about by June, more or less.

One determining factor for how long someone is willing to wait might be how hard hit was the area where each person was polled. So a New Yorker or a Californian may want to stay home.

Another determining factor would be how much money a person has in the bank.

But no one is asking about questions about why we do or don't want to stay at home. Could it have less to do with health than with political repercussions? Could how we feel about staying at home have to do with the political party we root for?

The greater the economic devastation = ousting Trump. The least amount of economic devastation = a plus for getting Trump back into office. You can already see writers laying the economic devastation due to stay-at-home orders at the Trump administration's feet (I've read others but this is from The Atlantic Monthly):

The economic devastation wrought by the pandemic, and the Trump administration’s failure to prepare for it even as it crippled the world’s richest nations, cannot be overstated. Tens of millions of Americans are unemployed. Tens of thousands line up outside food banks and food pantries each week to obtain sustenance they cannot pay for. Businesses across the country are struggling and failing. The economy cannot be held in stasis indefinitely—the longer it is, the more people will suffer.

The association is drawn by juxtaposition. However, if it's the Left that wants to stay home and it causes greater economic devastation, then who should be blamed?

I vote for no blame, but who doesn't love the blame game? I'll go first: Let's blame the ewoks.

Could the Trump administration have better prepared for it? Maybe so. Maybe we should have been stricter. But opinions will fall along party lines. The hindsight of armchair quarterbacks is remarkably keen.

Here's an opinion piece that supposedly addresses how lethal the virus is, saying that more of us are probably infected but we haven't been tested. The reason it appears to be more deadly than a flu virus is that we hadn't had a vaccine ready.

I gleaned this second hand. I haven't signed up to read The Wall Street Journal as I don't want extra spam in my email box.

This intriguing idea is at least partially true since not everyone will go to be tested. But I'm not sure I buy the idea completely. That would mean ten to thirty times the current population listed as having been infected were actually infected. So instead of 1.5 million cases, fifteen to forty-five million Americans were infected, most of whom didn't show serious symptoms. Possibly?

Whatever the case, we must move forward.

#

I have another post to write, extracting more data from the graphs. It will lead us to hard questions for both political parties. (I mean beyond, "Why is it so hard to make a good Star Wars movie?" and "Why do people put nuts in cookies and how can we stop these evil-doers?")

* Isn't it interesting how much attention we give to that 3%? How many worried news spots have you seen about them? Well, it is a sizable chunk of the population. But sometimes things are magnified to make them seem like 30% of the population.

On the other hand, how can we know what size a population we're dealing with until we ask the question? But had anyone asked? Now that we know, why haven't more connected the dots?

The Rate of Covid-19 Spread in the U.S.: Optimism in Reading and Interpreting Graphs

Since death is involved, COVID-19 is a serious matter. Dismissal of what's happening should be dismissed.

Yet there is too much disinformation and distortion about the virus. It may not be intentionally deceptive. It may be ignorance of how to read graphs despite being written or conveyed by educated people. If you want a real graph, go to Flip the Script, where they are gathering actual data.

I have toggled the graphs to optimize visually how the virus is spreading. I will justify my selections below.

Select "Scale to Highlight", so we can view the detail of our selected country.

Select "New Cases" because that shows the rate of infection. Anyone who displays "Total Cases" will be aiming to cause panic. Why? To get viewers/readers. To sell politics or whatever they're peddling. We don't know how, when, or if the cases are considered no longer active. That could easily be politicized. New cases, on the other hand, would be difficult to argue with.

Choose "1 week average" because there are a lot of institutions collecting and entering data, and we can probably surmise they aren't all performed exactly the same. This averages out reporting differences.

Choose "Linear" because most of us can intuit what's happening in the graph. Choose "Log" if you can think in exponential terms, which is a rare individual. Anyone showing graphs for public consumption should do "Linear" to maximize understanding.

What do we see? A slow decline. Why is it slow? Probably due to our large population, freedom, and ease of transportation. [See ETA below for new discussion of this.]

How much of a decline? On Day 34, it maxed at 95.8 average cases / million per day. On Day 67, it fell to 70.2 average cases / million per day, which means the rate of infection has fallen 26.7% [UPDATE: Day 69 is at 28.1%]. The rate of change is somewhat less than one percent a day.* Not impressive, but it's dropping.

Another Reason to Select 1-Week Average

Take a look at this graph of New Cases Per Day. What do you notice?

There's a sawtooth pattern. What is its frequency? About a week. This suggests that there are differences in reporting. This doesn't suggest false data, but "spikes" can be misunderstood as something more than what they are.

It is possible that the spikes could be due to patterns in shopping if the nation is more or less regular in their shopping habits, but that would mean that the virus manifests into symptoms at a very regular rate -- not two to fourteen days.

If you look at the the death toll as it changes day by day, you'll see the same weekly sawtooth (this doesn't exclude the possibility of shopping being the cause, but it does suggest that differences in reporting may reveal the culprit of why these patterns appear:

A third possibility for the sawtooth patterns may be healthcare-worker fatigue although that would be strange to see this appear at such regular intervals across the nation, but it may represent a hospital's ability to admit and treat patients. Perhaps there are times of day, or days of the week when a patient receives the best care.

 Death Toll

Some news outlets have used death tolls to indicate spread, which is dramatic and misleading. That number can only go up, which can only lead to anxiety. Once dead, always dead. Ignore any institution who uses this information. Nonetheless, I also looked at New Deaths to see how that's changing.

On Day 30, it maxed at 8.1 average deaths / million per day. On Day 52, it fell to 4.6 average deaths / million per day, which means the death rate has fallen 43%. The rate of change is two percent a day.*

This has fallen quite precipitously. Why? Changes in testing? Health care system? Population better informed (thanks to news / websites / politicians)? Population better prepared to seek medical attention? Something else? I haven't yet seen this discussed. Whatever the cause, it's encouraging.

Misleading

A lot of the anxiety about the virus is caught in a positive feedback loop--not that the result is positive but our tensions rise and we share our anxieties which causes theirs to rise. They share info that causes our anxiety to rise further, etc. What can we do to exit the loop?

Be open to positivity. Be liberal. Avoid rejecting ideas because it doesn't match your tribe's script. Try to understand where someone's coming from. Maybe they are poor. Maybe they need to work for psychological reasons. If we close ourselves off from different opinions, we can do nothing but enter a positive feedback loop.

We can be more careful about language, especially if language is our job. Any time you hear "spike," be wary. The writers (unwittingly?) are fomenting anxiety among readers by using data gathered in a manner that is unlikely to be uniform, but lumpy or spiked. We need to look at what's going on overall. Moreover, each individual has a different immune system and healthiness. Given this, the data is unlikely to be uniform, even should there be perfect data capturing.

If you hear event X "coincides" with a spike on the same or next day, then they are trying to make connections that may not exist. Avoid this writer and possibly the publication or news outlet. It takes time for symptoms to manifest, and time to consider whether it's worth going to the doctor (if they ever do), and time to see a doctor, get tested and get diagnosed. Moreover, how can you tell which event does what? We'd need people to test before and after--not to mention follow whom they come into contact with--to arrive at reliable conclusions.

One news outlet connected a gathering one day with deaths on the next--a connection which should take a week or three to occur, if this were monitored.

Thinking about the Future

One writer proposed testing every citizen. That would make much clear. However, that may not capture the true numbers since one could be infected yet not manifest, so we'd need to test all citizens within one day, and test them all again a few days later. And so on, for a few weeks. We'd know more about the COVID virus this way. Of course, in testing everyone, we might increase the spread.

Also, this is just one of many viruses. Later mutations may not behave the same, so universal testing may be a large expense with little future use.

You could argue that this would resolve the epidemic quickly. Maybe. Assuming that we keep our borders closed until the virus has left the planet (or until we make a vaccine) and assuming that people make themselves easily found.

COVID-19, many conjecture, may be with us for years. What do we do in the meantime? What if another virus appears next year and the year after that? How will we handle these complications? These are questions we will need to consider.

As the weeks progress, we will have more evidence as states stay closed (or people choose to stay at home), and other states open and some re-enter normal life with caution and others with caution abandoned. This graph will change, for better or for worse.*

* The graphs are from yesterday, which is when I wrote this. The rates for today remain about the same, with an uptick in numbers. At the time of writing this, the graphs hadn't been updated. Making new graphs for one additional data point doesn't make sense although, of course, the new data point may look dramatic on the New-Cases-Per-Day graph.

• A new post on the political nature of our response to COVID-19.

• The third and final (?) post about analyzing states with success stories and discussing possible courses of action.

NOTE: Please credit me. I don't own the graphs although I did manipulate them. And I did write this, ideas that occurred to me from reading the graphs, not following news media or talking heads. Yes. I wrote everything on the blog, so it's copywritten. Please follow standard usage. Feel free to quote.