The problem is that it illustrates many different things in one article. I'll try and break them down.
I think the NYT is correct in their broader thesis that Right leaning geographies are having a different experience of Covid-19 than left leaning geographies. Partly it has to do with the prevalence of Covid-19 deaths in those different locations but also, and much understated by the NYT, to do with policy responses to Covid-19. But first an overiew.
There are always two movies playing
Scott Adams makes this argument all the time. For any set of events, there are always two movies because different people register different interpretations, different observations, and different levels of importance to the same evidence. A governor shuts down her state. One audience see a movie about a governor trying to save a percentage of the population from death by Covid-19. Another audience sees a governor killing off 30% of small businesses and ruining the financial stability of 60% of the population. Same event, two movies. That is partly what is happening in this article.
What you see is where you live
This is a variant of the argument I have been making for years; that we underestimate the potential bubble effect arising from the geography of the consolidated media sector. Only a small number of companies produce the majority of "broadcast news". And the great bulk of the journalists of those companies reside primarily in geographies which are radically different from the America in which most people live. I am making up the numbers but perhaps 50-70% or more of big time name journalists and pundits live in New York City, Washington, D.C., and Los Angeles. There are a few in outposts such as Chicago, Philadelphia, Miami, San Francisco, and Atlanta. Virtually none in Phoenix or Tulsa, or Peoria, or Louisville, or Cincinnati, etc.
There are actually two bubbles. One is the Ben Rhodes bubble - the journalists are young with expensive credentials from big name universities, overwhelmingly left leaning, language oriented and innumerate, glib rather than considered, no experience outside of media and academia, and little executive experience/awareness.
That bubble is real and important but probably less than the other bubble. The second bubble is that they live in the same few places and experience a distinctly different manner of life from everyone else.
They live in those few big cities which are all hard left governed, last experienced a Republican governor more than a half a century ago. Washington, D.C., New York, San Francisco, etc. are dramatically more unequal than the rest of the country, far more likely to see homelessness up close, extreme poverty, have very unique housing issues, experience much more intense crime, see much more economic segregation, experience far worse performing schools, live among a much higher percentage of foreign born, there are many more government programs, far fewer cars, much more mass transit, etc.
It is not necessarily bleak - those cities also have much greater wealth, much more cultural variance/enrichment, a greater concentration of elite pastimes, etc.
The issue is that they are substantially different from anywhere else in the nation. If you are a journalist, the world you see every day, the concerns most manifest, are substantially different than anywhere else. It comes across to those outside the cities as a hard left, statist, social justice vision of life when I suspect, to a greater degree than we acknowledge, that the difference in reality being reported is simply a product of difference on geographical reality.
It is not that journalist are seeing a different movie from the same facts as viewed by the rest of the nation. It is that they are actually seeing different facts in the first place.
Clickbait writing is pervasive
Mainstream media has been experiencing wrenching consolidation and retrenchment for nearly two decades now. They rely on cheap 27-year old inexperienced journalists more, on press release journalism more, reporting opinions over facts more, and have abandoned editorial standards. They are in the commodity stage of the media S-curve. Massive cheap competition, unsupportable legacy brand structures, declining revenues, commoditization of product, and all the other ills of that territory at the top of the sector sigmoid curve, just before some transformation that cannot be fully anticipated.
They have to get clicks on their articles which drives journalists to write more baldly, with less subtlety, and consequently less accurately.
From the article.
"The staggering American death toll from the coronavirus" - No, not staggering. Average all causes mortality in the US runs about 235,000 per month. Since the first Covid-19 death in early February, the normal all causes mortality would be very roughly 940,000. And we know that virtually all those reported Covid-19 deaths are a function of comorbidities. People are dying a few months earlier due to Covid-19 combined with an underlying deadly condition. The number of people dying solely from Covid-19 is small or vanishingly small. It is becoming increasingly clear, in the US and across the OECD, that Covid-19 seems likely to look pretty similar in its rates and curve shapes to more than a dozen similar Covid like outbreaks since 1945. Outbreaks which did not involve quarantines or shutdowns.
"deaths, now approaching 100,000" - Not yet. As of May 26th, the CDC reports 75,283 deaths so far and a dramatic decline from the peak weekly death total in mid-March of 15,400 to 2,500 in the most recent week. Remembering that these are deaths with associated Covid-19, not deaths from Covid-19. The decline is rapid and the models have dramatically overforecast deaths so far. 100,000 deaths is more dramatic than 75,000 but 75,000 is by far the more reliable number. Will we hit 100,000? Probably, but not with certainty given the current trend lines. Perhaps by the end of the year which is much further than the imminence implied by "approaching."
"The devastation, in other words, has been disproportionately felt in blue America" - Narrowly true and misleading. "Toll" would be the more accurate word. Devastation conflates death toll with Covid-19 impact. Perhaps 40,000 of the 75,000 deaths are in the small blue areas. That is indeed a concentrated death toll. However, the devastation in those areas is largely driven by the responses to the Covid-19 threat, not the Covid-19 deaths per se.
"there are starkly different realities for red and blue America right now." - Well . . . you are going to have to be pretty narrowly precise in your definitions for that to be true. I live in a deep blue neighborhood, in a deep blue mega-city in a red state. We have been shut down at the city level for two months or so. Economically it has been devastating. In terms of death numbers? Not so much. Are we living with stark differences? There are a handful of rural counties with an excessive outbreak. A couple of poor neighborhoods in the big city. Are there stark differences between Red and Blue? In a few places under a few circumstance, but not broadly it would seem. In my state, outbreaks don't align with the red and blue distinction. In New York it does. More on this later.
"Staggering", "approaching 100,000", "devastation", "starkly" - This is not, per se, a left-right issue. This is a journalism issue. It is exaggerated overwriting in order to drive clicks for desperately needed revenue. It is not balanced or accurate reporting.
Data Visualization Struggles
The lead authority in this area is Edward Tufte. There is both an art and a science to data visualization and mapping. Regrettably, the software package tools have outstripped disciplined data visualization good practice, forcing viewers to work hard to interpret what is being presented (or, as often, what is being misleadingly omitted).
Don't get me wrong - I love good data visualization, it can be a great trigger to insight and/or productive discussion. The challenge is that SW can produce data visualization which has not been well designed.
The NYT leads it article with a visually clean pair of maps:
They have been doing a lot of this type of mapping in the past year.
It is clean, but what does it tell us? How easy is it to extract useful information from it? For one thing, there is no meaningful scale. Big circles are more than small circles but how much more? We don't know.
It is a map of "reported cases" rather than the more certain "Covid-19 deaths" measure. We know that there has been marked varaiance in definition of what counts as a Covid-19 case and even about what counts as a Covid-19 death. We know some locations have operationally struggled with aggregating the numbers. We have seen plenty of instances of very large adjustments. Producing a dramatic visualization of bad data comes close to malpractice.
You have to stare for a few minutes to begin to discern anything from the side-by-side maps in order to understand just what it is that they are mapping. Broadly we can sense that Covid-19 deaths are less frequent in red areas than in blue areas (smaller circles) but we can also see that Covid-19 seems more broadly spread in red areas than in blue (more circles).
But if you are cognitively querying the representation, you are now having to integrate facts not in evidence. We are dealing with death counts rather than death rates. Is that the right measure? We know that density is a factor in here, that should push us to rates rather than counts given that Democratic strongholds tend to be dense.
The above representation is not dissimilar in nature to the misimpression left by a map that was widely touted after the 2016 election. Trump won with a pretty commanding 304 to 227 electoral votes (masking a more fragile reality than that conveys).
The map circulated which circulated and made the win seem even more commanding was this
It shows the county level returns as won by Trump (red) or Clinton (blue). It inadvertently prioritizes geography over population. The map is not factually wrong, but by ignoring population counts and focusing on geographical entities, it tends to exaggerate the dominance of the win. An interesting article here covering the different approaches to visually representing the election results.
The NYT, by using side-by-side maps, with count rather than rate, forces us to attempt to impute some sort of equalizer to take into account density. Pretty messy.
Innumeracy and Empiricism
This is a not unsurprising and indeed well-established pitfall for the news media. They recruit from a pool which is focused on words rather than numbers and when attempting to report something that has a measurable aspect, they frequently commit numeric pratfall after pratfall.
You can see an example in the third and fourth paragraphs of this article. Remember they led with maps that represent death counts.
Democrats are far more likely to live in counties where the virus has ravaged the community, while Republicans are more likely to live in counties that have been relatively unscathed by the illness, though they are paying an economic price. Counties won by President Trump in 2016 have reported just 27 percent of the virus infections and 21 percent of the deaths — even though 45 percent of Americans live in these communities, a New York Times analysis has found.And just like that they shift, without notice or explanation, from death counts to death rates. Both views are important and worthwhile but both measures convey something different about the dimensions of the crisis. You should not be eliding the two as if there were no difference.
The very real difference in death rates has helped fuel deep disagreement over the dangers of the pandemic and how the country should proceed.
It is especially true in the Covid-19 instance. Big countries are likely to have bigger absolute death counts. But death rates allow you to determine how effectively you responded to the crisis. The states with the largest death rates are those, including New York, which chose to move Covid-19 cases from hospitals to assisted living facilities with the highest concentrations of those with comorbidities coresident. Counts gives you a weak indication of size of state. Rates gives you a stronger indication of response effectiveness.
For the rest of the article, they primarily focus on rates but it is a sharp discontinuity to lead with a "counts" graphic and then in the text focus on the more appropriate "rates". You keep trying to reconcile the irreconcilable.
"Over all, African-Americans and Latinos have had higher infection and death rates from the virus." - Another example of motivated reasoning and innumeracy. It appears from all the research I have seen so far that the causal factor in infection and death rates between races is entirely due to prevalence of pre-existing conditions (obesity, diabetes and blood pressure in particular) and cultural norms rather than race per se. The NYT is keen on race as a driver of inequities, but the evidence does not seem to support that conclusion. They slip it in anyway.
Motivated Conclusions
"Right-wing media, which moved swiftly from downplaying the severity of the crisis to calling it a Democratic plot to bring down the president, has exacerbated the rift." - At least three issues here.
I don't know about right-wing media, but there was certainly much debate among nominally qualified experts in January and February as to how much peril Covid-19 represented to Americans and to the world. Some models estimated millions of deaths in the US and other experts estimated in the low hundreds of thousands. There were many, not particularly associated with the right, arguing from epistemic experience rather than models, that, while dangerous, Covid-19 would look more like a bad flu season than the Spanish Flu.
This was not downplaying the crisis. This was debate about the possible implications of something broadly unknown. It was also a proxy for the long-standing debate between experiential evidence and model evidence. With the most recent CDC report in the Washington Post of a possible bad flu-like rate, it is inaccurate to characterize the early debates as "downplaying". That is a motivated conclusion on the part of the NYT reporters. From the WP article.
The question of the true lethality of the virus remains the subject of controversy. When the CDC put out its guidance last week, it estimated that 0.2 to 1% of people who become infected and symptomatic will die. The agency offered a “current best estimate” of 0.4%. The agency also gave a best estimate that 35% of people infected never develop symptoms. Those numbers when put together would produce an “infection fatality rate” of 0.26, which is lower than many of the estimates produced by scientists and modelers to date.Though written with some bias, the majority of that Washington Post article is pretty good at acknowledging that the numbers are not certain, reliable or consistent, that the model forecasts have been widely variant and effectively unreliable, and that there is still much unknown and debated.
Those whom the NYT characterizes as "down-playing" the crisis at the beginning were talking exactly about this sort of range now being suggested by the CDC, 0.1% to 0.4%, in which 0.26% falls squarely. At the low end (0.1%) about a normal flu season, in the middle, about like a pretty bad flu season such as we have every decade or so, and at the top end (0.4%), just outside the worst flu seasons.
If the rate turns out to be at the lower end of the range as those debating in January and February forecast, then it is hard to accurate to claim that they were downplaying. They made a lower forecast than that which apparently the NYT reporters believed was reliable. Reality has, so far, proven the low range forecasters to be more accurate. Characterizing those forecasters, whether left or right, as "downplaying" appears to be a shoddy motivated conclusion.
I sit in a sports bar with a friend before a game. We have a heated debate. I think, based on recent performance, playing styles of the opponents, recent injuries, etc. that this will be a low-scoring game. I forecast that my local team will win 2 to 1. My friend thinks it will be a high-scoring game and that we will win 7 to 5.
By the seventh inning the score is my team 2 and the opponents 1. Of course anything can still happen but my low scoring forecast appears much more likely than my friends high-coring forecast. I was not downplaying, I was forecasting.
"Calling it a Democratic plot to bring down the president" - Now we are into political opinion blogging. I read pretty widely and voraciously. I don't recall anyone claiming that it was "a Democratic plot to bring down the president." I recall lots of right leaning commentators reminding everyone of Democrat Party Elder Rahm Emmanuel's famous dictum,
You never want a serious crisis to go to waste. And what I mean by that is an opportunity to do things that you think you could not do before.Influenced no doubt by Saul Alinsky's Rules for Radicals (page 89), in the section marked communication "in the arena of action, a threat or a crisis becomes almost a precondition to communication."
It doesn't help the NYT reporters that there are plenty of instances among Democratic Party leaders echoing this sentiment specifically about Covid-19, ranging from bringing down Trump, to implementing the Green New Deal, to restoring focus on Global Climate Change, etc. Right leaning commentators have been concerned that Democrats would try and leverage Covid-19 to their benefit and Democratic party leaders and journalists have confirmed that they are explicitly trying to do that.
Making a strong claim such as that Republicans were "calling it a Democratic plot to bring down the president" is so easy to demonstrate. Just find a Republican who made that claim and link to their quote. I am sure there is a county Republican Chair somewhere that made some variant of that claim. I suspect that the problem for the NYT reporters is two-fold. They get a quote that is very close to their claim but the link reveals that it is from a peripheral person. Or, they get a muscular statement from a senior, more central Republican and it reveals that they are talking about the larger issue os exploiting a crisis, which would make Democrats appear bad.
Democrats are explicitly seeking to take advantage of Covid-19 to advance their objectives and are quoted as attempting to do so. Republicans are doing the same. Each establishment party is exploiting a crisis to advance their respective objectives. For the New York Times reporters to focus on the one and not the other is simple advocacy. For the New York Times reporters to make a claim without a supporting link is simple advocacy and malpractice.
"And even as the nation’s top medical experts note the danger of easing restrictions, communities across the country are doing so, creating a patchwork of regulations, often along ideological lines." - It is as if the reporters wrote this May 25th article two or three weeks ago and failed to update it. Several states began an aggressive opening up about four weeks ago. It is true that there were "experts" and plenty of media criticism of those state openings and expert forecasts of killing fields in those states as Covid-19 came crashing back without the lockdowns. And none of it happened. Death rates have been dropping or holding steady.
Bubble Effect/Excessive Deference to Experts
There seems an effort by the reporters to convey that right leaning areas are irrational in their response to Covid-19 even though the reporters are affirming that right leaning critics are correct in their assumptions. We have already discussed that the left claimed an epidemiological disaster if states opened and are confounded that that has not happened.
The right leaning critics also appear to have been correct to distrust the motivated forecasts of the "experts."
"Public opinion polls do show widespread support for stay-at-home orders, but also indicate that Republicans are less likely to see the virus as a significant threat to their health." - Which is not surprising given the thesis that the reporters are validating in their own reporting. They are reporting that Republicans are less likely to see the virus as a significant threat to their health because it is in fact, depending on where they live, not much of a threat to their health, particularly given the absence of prior comorbidities.
Similarly:
"Some skepticism around the impact of the pandemic can be traced to a distrust of the government that has grown among conservatives in the last decade" - And indeed, though the reporters do not allude to it, the Covid-19 has stripped away much of the veneer of competence which we hope and expect to be there when needed. The global experts of WHO failed to declare a pandemic long past the point when it was obvious there was a pandemic. Equally, the experts of WHO, along with many others in the US, recommended for the first few months of the pandemic that personal masking was not effective and indeed counterproductive. Until they changed their minds.
CDC, who has all along supposed to be focused on epidemic diseases, stumbled from disastrous decision to disastrous decision. No executable plan. Badly executed plans. Inadequacy of stockpiles. Focusing on the wrong problems (ventilators). Bad modeling. Unuseful of inaccurate forecasting. Failed testing. And on and on.
All the government experts were concerned about overwhelming the health system which never came close to being overwhelmed by caseloads. A handful of states governments insisted that those with Covid-19 infections should be warehoused in assisted living facilities - a position criticized at the time, done anyway, and now acknowledged to have been among the most grievous of failures.
You would have to be insensate to not acknowledge that both over the longer term and in regard to the near term of the pandemic, that many people, not just on the right, are now distrustful of government and for pretty obvious reasons. It might be more prevalent on the right but it is by no means limited to the right. Since the seventies, it is pretty rare that even 50% of Americans at large trust the government to deal with domestic problems. It has been below 40% since 2014.
The NYT reporters want to make the loss of trust in government a right leaning issue when it is clear that the loss of trust in government is common across the spectrum. Another instance of bubble bias influencing what they are reporting.
I won't belabor it past this point. An interesting article with some useful information but horribly twisted by prior assumptions, partly partisan, partly ideological, partly geographic, partly innumeracy, partly absence of editorial oversight, etc.
With a good editor and another day's worth of research and cleaning up, they could have had a great article rather than mush.
But their core focus - that some areas are living an entirely different pandemic is true. And it is also true, though unacknowledged in the article, that a disproportionate degree of the harm arising from Covid-19 is not in the nature of the disease but in the policies which were adopted to fight the disease.
Posts
No comments:
Post a Comment