The discrepancy with the actual mortality data is staggering: for people aged 18–24, the share of those worried about serious health consequences is 400 times higher than the share of total COVID deaths; for those age 25–34 it is 90 times higher.
I am not so happy with the metric here.
Let A = someone is worried about health consequences of the virus.
Let B = someone dies of the virus
Let C = someone is 18-24.
The claim that young people have a distorted view of their risk would be that p(A/C) is way too high relative to P(B/C). But the data that they are presenting seems to me to compare P(A/C) relative to P(B and C)
But let us stipulate that in fact young people now tend to greatly over-estimate the risk. Why would this be?
1. Tyler Cowen predicted this with one of his very first posts on the virus–that we would under-estimate the risk early and then over-estimate the risk. More recently, he speaks of phantom risk, or the “stigma” of cases.
2. Desai cites partisanship and media hype. I agree that this is a hypothesis, and perhaps when President Biden takes office the mainstream media will decide that it’s time to say that it’s safe for people to go back to work and school. But I think that the virus fears are too deep in our collective psyche for the media to undo them.
3. Respect for authority is still a thing, and the authorities are saying that schools cannot open, etc. So people infer that the risk must be pretty great.
4. I don’t think we can say that fear of the virus is completely irrational. As of a week ago, the average daily death figures continued to hover at around 1000, and if you extrapolate that to an annual rate of 365,000 it is frighteningly high. And young people know that even if they do not suffer adverse consequences, they may come into contact with friends or relatives who are older and more vulnerable.
I believe the comparison that Desai made is actually P(A|C) vs P(C|B), which is no sounder a comparison than P(A/C) vs P(B and C).
Adding to your point 4, another relevant aspect of case A is that the world has seen a lot of serious and potentially long-term effects after surviving COVID-19: lung damage, cardiovascular damage, MIS-C, muscle weakness, and so forth. We have only months of follow-up data, not decades, so we do not know how many young patients could end up living with the effects of this disease for decades. In that respect, the young could be concerned about long-term impacts even if they know their short-term risk of death is relatively low.
I tend to think that the young overestimate these risks as well, but unless we hypothesize mass hysteria in medical science, those concerns do represent true risk.
Michael P. your last paragraph reflects the fundamental contradiction of the few that still think that Covid-19 is a severe disease. To some degree, we are all concerned about not being infected because we don’t know how our bodies will respond. That degree depends both on our perceptions of how little true “experts” know and how hard it’s to identify the true “experts”. Our perceptions are formed by information supplied by “intermediaries” with their agenda (and this includes Tyler Cowen as an agent of Mike Bloomberg –you should read his columns in Bloomberg News). Misinformation is the main cause of misperceptions*. There is no need to hypothesize mass hysteria in medical science or any science or politics or GMU Department of Economics to understand that many “intermediaries” have made grave errors, intentionally or not, in communicating relevant science and all other relevant knowledge. Those “intermediaries” include mass and social media, politicians, bureaucrats, academics, and charlatans. Those errors are responsible for the misperceptions of risks discussed by Sonal Desai in her short paper.
*Read “Facts and Myth about Misperceptions” by Brendan Nyhan in the last issue of the Journal of Economic Perspectives. The last paragraph says “Any evidence-based response to the problem of misperceptions must thus begin with an effort to counter misinformation about the problem itself. Only then can we design interventions that are proportional to the severity of the problem and consistent with the values of a democratic society.
Indeed, Arnold, I think my comment contradicts your points 1-4. In my comment, I don’t consider the possibility that young and old people are concerned about infecting other people (your point at the end of #4) but it adds little to our concern about how our bodies will respond (except perhaps for kids under 12 but they are protected by adults).
There are plausible reports of damage to the heart and lungs, even in asymptomatic cases, that may be permanent. Since immunity does not persist for long, a young person might contract this disease dozens of times over a lifetime. If each case does permanent damage, that could add up to years of disability and early death.
Or maybe not. We don’t know, at this point. But reasonable caution seems prudent, and it seems reasonable to take modest precautions (masks, avoiding crowds, etc.) even for young people.
Also, as a general rule it pays to remember that those who speak with great certainty on complex and rapidly-evolving matters are generally con men. Anyone who’s certain about the long-term effects of this virus is either a genius or an idiot, and one of those is far more common than the other.
Yes, Jay. That is why from the very beginning I assumed that “To some degree, we are all concerned about not being infected because we don’t know how our bodies will respond.” The implications of this basic assumption are related to the demand for information to form our perceptions, but unfortunately, we have to rely on suppliers (the intermediaries) that are not good and often wrong. Given the poor and wrong information, you should not be surprised that misperceptions have become a problem.
If we could know the distribution of our perceptions over any dimension relevant to the virus, the disease, and the pandemic, I’d bet that is uniform over a min greater than zero and a max lower than 100. Under the assumption that the “true” value must be within a much narrower range, misperceptions are a problem. Thus, I’m not minimizing our concerns. To those concerns, I add my concern about the challenge to access and understand the information supplied by “intermediaries”.
Arnold appears to have chosen the potatoes over the meat in Ms. Desai’s analysis.
Here is the meat:
“Six months into this pandemic, Americans still dramatically misunderstand the risk of dying from COVID-19:
On average, Americans believe that people aged 55 and older account for just over half of total COVID-19 deaths; the actual figure is 92%.
Americans believe that people aged 44 and younger account for about 30% of total deaths; the actual figure is 2.7%”
These data points on relative risk have been known for quite some time and were well publicized, yet we collectively reacted as if they didn’t exist. This strikes me as an irrational response.
RE: point #4, I can see “fear is rational, given 1k deaths per day.” But I think that figure is based on shoddy underlying data. In particular, I suspect we’re fundamentally counting “people dying with a positive covid test or suspicion of one” as “covid deaths”. (And I have issues with the PCR tests on top of it)
So that could be (part of) why calculating risk based on those numbers doesn’t match up with the risk we may now *feel* from observing the world?
There’s an equally compelling argument looking at excess death statistics that we’re undercounting COVID deaths.
The NY Times journalists looking at excess deaths is not compelling.
There have not been 1,000 Covid-19 deaths a day in August but 1,000 Covid-19 reported deaths, which include previous weeks much more than in April through June. Also, one would never simply multiply 1,000 deaths a day by 365 to get a scary number like 365,000. We all know deaths will continue to come down as the have since a summer peak of 1,150 deaths a day (reported).
Jul 25 to Jul 31 1,150 deaths a day
Aug 8 to Aug 14 1,090
Aug 15 to Aug 21 1,020
Aug 22 to Aug 28 960
Cases a day are down 35% since mid July at the peak. Covid-19 related ER visits are right at where they were at the beginning of the pandemic. There was a week in May that was 0.1 percentage point higher.
I am intrigued by your use of the phrase, “. . .when President Biden takes office. . ”
Did I miss a post on this? May I ask your reasoning for this conclusion?
Or is my brain that has been too rattled by the onslaught of life-is-stranger-than-fiction absurdist news stories rendered me incapable of any longer recognizing sarcasm and satire?
Your brain has not been unduly rattled. Arnold has jumped the gun on this point. Based on early polling data, the carnage and the violence changes everything.
Kling is stating the condition for what follows in that sentence. It isn’t an opinion on who will win, at least not as written.
If vs. when. not that complicated. Kling said “when” which is substantially different from “if”.
Lighten up. This is one of my humor tropes. On an economics panel discussion early in the Biden Administration, I repeatedly discussed how unemployment might not come down to a particular point before the end of President Palin’s second term.
🙂
I’m pretty sure Biden will get elected. Trump is currently at ~43% in election betting odds, but I think that exaggerates his odds. I would guess Trump’s odds are more like 25-30%. It looks like Nate Silver’s odds are about the same as mine. I doubt riots will benefit Trump much. Generally these things are seen as reflecting badly on incumbents, and Biden is not popularly associated with BLM-type ideology.
And, if you’re wrong, it will be the biggest goal line fumble in U.S. political history. Looking forward to the possibility to lmao for another four years as the left tries to understand how they lost yet again.
When I see polling data that doesn’t make sense, my first instinct is to blame the poll, not the people responding. I note that Desai doesn’t provide the actual questions polled, merely his interpretation of them. Nor do we get any summaries that don’t match the story he wants to tell, but somehow I suspect there were more than three questions asked.
“Tyler Cowen predicted this with one of his very first posts on the virus–that we would under-estimate the risk early and then over-estimate the risk.”
My own experience has been the opposite. Early on, there were a lot of unknowns and I was more cautious. There are still some unknowns, but my assessment of the risk is modestly lower today than it was in March.
More evidence that commenters here are not representative.
One of the more interesting aspects of survey was the partisan differences in risk perception. Democrats have an enormously higher risk perception than do Republicans. According to the study, partisan difference is driven by social media.
As far as some of the comments above about likely election outcomes based on reports of polling, one would think a lesson would have been learned in 2016, but evidently not.
Polls need to be separated: there are media polls, on the one hand, which are heavily partisan loaded to produce striking results so as to generate “news.” (Fake news?) One should never even talk about a poll unless the partisan loading is published, and if inconsistent with the actual turnout in the previous election, adjusted accordingly. Registered voter polls lean about 4 points more Democratic than likely voter polls, but you don’t see that many likely voter polls because they are more expensive to conduct.
Media polls are further questionable if one remembers the Wikileaks of the Podesta emails. In some of them, there is specific discussion of deliberate poll rigging as part of the campaign effort. One technique is to sample independents in areas known to lean heavily Democratic.
There are on the other hand non-media polls like Gallup, and perhaps Rasmussen, which rely on subscriber revenues, and therefore make more effort to get things right, as that is what their subscribers are looking for.
for people aged 18–24, the share of those worried about serious health consequences is 400 times higher than the share of total COVID deaths
The explanation is fairly simple. Children scare easily.