What is the best way to reconcile the results for these three polls?
How good is the following heuristic?
The resources you spend mitigating a problem should be directly proportional to its overall severity.
— Bryan Caplan (@bryan_caplan) August 18, 2020
Medically speaking, how bad is coronavirus compared to flu?
— Bryan Caplan (@bryan_caplan) August 17, 2020
How much time, inconvenience, and resources should we spend fighting coronavirus compared to flu?
— Bryan Caplan (@bryan_caplan) August 17, 2020
I’m tempted to just say “cognitive dissonance.” The effort heuristic makes great sense, and the medical estimate seems about right. But that in turn implies that past and current coronavirus efforts (public and private) are grossly excessive. Indeed, do we even spend five hours per year fighting flu? If so, why should we spend more than twenty five hours per year fighting coronavirus? But almost no one feels comfortable with that relaxed attitude, hence the dissonance.
READER COMMENTS
Philo
Sep 16 2020 at 10:27am
The first poll question is garbage, since it ignores the issue: How much good do you get for your marginal mitigation buck? Many people do not see this when answering your poll, but in practice this is how they think—cost-benefit style. A few trivial measures eliminate most of the flu risk, but COVID, being much more contagious, calls for more expensive mitigation.
suddyan
Sep 20 2020 at 7:51am
[A few trivial measures eliminate most of the flu risk, but COVID, being much more contagious, calls for more expensive mitigation.]
The only way the purported conclusion (“more expensive mitigation”) follows from the proposition (“much more contagious”) is if an extremely simplistic view is taken.
How “contagious” COVID supposedly is, is merely one aspect of a much more involved situation.
Holistically, COVID is no more dangerous than the flu over many past years.
KevinDC
Sep 16 2020 at 10:43am
Here’s my attempt at an explanation – although in the spirit of full disclosure I should say I’m only on my second cup of coffee and have only given this 90 seconds of though so I might not have thought this through very well.
The reason one might think we should spend such a greater proportion of resources dealing with Covid 19 compared to the flu is that the flu has far more options readily available at lower costs. Flu vaccines are readily available, and can mostly keep up with even the changes in seasonal flu. And we have a variety of antiviral and other treatments available known to be effective in treating people infected with flu. By contrast, Covid 19 has no vaccine yet, and no treatments yet proven to be effective. Without these highly convenient, low cost options, you could believe Covid therefore requires the use of higher cost and less convenient methods to contain, especially amid tens or hundreds of thousands of deaths occurring nationwide in the span of a few months. That’s why we can spend comparatively and even disproportionately little time worrying about the flu – because we already have lots of low cost tools to deal with it, which isn’t the case for Covid.
That’s not an airtight argument, but it’s not obviously absurd either, and I can see how someone might reasonably think that.
dede
Sep 20 2020 at 6:54am
“we have a variety of antiviral and other treatments available known to be effective in treating people infected with flu”
Are you sure? Quick check on the internet tells me that lying in bed, drinking a lot of water and maybe some aspirin or paracetamol is the standard treatment. Apparently, some antiviral drugs are available but, given the number of people dying every year from the flu, their efficacy could be debatable…
Daniel Carroll
Sep 16 2020 at 10:55am
Covid-19 impacts are non-linear and it mostly presents a coordination problem. The poll suggests your readers are more concerned about the societal impacts – thus the non-linear impacts – than they are about their individual risk. The second poll isolated medical impacts, and Covid-19 impacts are more than medical. Also, the third poll had 38% of the respondents of the first two, and may not represent the same cross-section of opinion. Twitter polls are not exactly known for random selection.
Steve
Sep 16 2020 at 6:01pm
I think this is the correct answer. Medically speaking, COVID may be 2-5x worse than the flu. But as far as impact to GDP (whether due to an appropriate response or a completely disproportionate one) it is probably 100-1,000x greater. So higher spending is justified to put an end to it.
I see no contradictions in those two assessments.
Anonymous
Sep 16 2020 at 11:15pm
Bingo. And the lockdowns aren’t the reason for the economic damage. People don’t go out when there’s a high risk to their personal safety, as there is when a pandemic is raging.
robc
Sep 17 2020 at 10:16pm
Woodstock suggests otherwise.
nomadscientist
Sep 18 2020 at 4:59pm
Circular reasoning. Corona is expensive because people choose to overvalue mitigation of corona. If we had not tried to mitigate corona, pure medical damage would not have reduced GDP by more than <1%.
Aleksander
Sep 20 2020 at 7:38pm
If Corona is only 5x worse than the flu, why would an appropriate response make a 100x impact on GDP? Such an impact would suggest that people really think the medical risk is much higher than 5x the flu.
DK13
Sep 16 2020 at 11:49am
It’s largely sunk cost fallacy. We want to believe our efforts at staying home, wearing masks, etc. have been effective, and that we’d be so much worse off but for those efforts. Which in turn leads us to a certain blindness and obliviousness about the costs of those efforts going forward.
Kailer
Sep 16 2020 at 12:32pm
You can’t compare COVID to the Flu! How many times do we have to say this. In April, more people were dying each day from COVID in the US than died on September 11th! Almost twice as many Americans have died of COVID than died fighting in World War 1! And that’s only if you count measured COVID deaths. If you count the excess deaths over the number of deaths in the last few years it’s even higher!
To summarize: you can’t compare COVID and the flu, you can only compare COVID and terrorist attacks, war deaths and deaths from all causes including flu over the last five years. It’s not that hard people! Science is REAL.
Mark Z
Sep 16 2020 at 3:31pm
Of course you can. Covid is n times as bad as the flu, for some finite value of n. If what you’re saying is n >> 1, who here is disagreeing with that?
Anonymous
Sep 16 2020 at 11:16pm
I honestly can’t tell if this is satire or not. That last sentence just about sealed it for yes, but… not sure.
MarkW
Sep 17 2020 at 8:56am
In a normal year (I found the numbers for 2017), the U.S. averages about 7700 deaths per day. That’s more than 2X the deaths of 9/11! Every day! Without a pandemic! For a full year, that works out 2.8M deaths — more than 2 1/2 times more than the number of Americans killed in all U.S. wars in history!
But therefore what? What is the value of comparing all of these incommensurate numbers?
Dan
Sep 16 2020 at 12:56pm
There are a host of issues here.
Different samples. Your third question has just over a third of the respondents as the other two questions. That’s kind of a red flag. Twitter polls themselves are a red flag. Ilya Somin had a good post about this around a month ago – I can share it with you if you’d like – but I think you’re empirical enough to already know why these are dubious at best. Yes, they’re better than Robin’s posts citing polls or comments from return of kings or some other garbage MRA site. But surely you know that twitter polls are not that much better than a “clap-ometer” someone’s making with their arms at a tween talent show, right? You have access to some great exerimentalist colleagues at GMU who would surely be happy to talk through some of the issues presented here.
The first question is poorly defined as Philo notes above. Maybe global warming or possible asteroid strikes are 100x more severe than covid or the flu, but I could rationally decide my actions have no impact on global warming or asteroid strikes, so I shouldn’t bother spending much time at all on either problem, let alone 100x the effort I spend on disease prevention/mitigation.
Even if we zoom out to the societal level, the problem is still there. Maybe we (“society”) can do something about global warming, but can’t really do anything to mitigate asteroid strikes. Even if both have the same severity (or asteroid strikes are a bigger deal), we probably should allocate our efforts toward global warming – the problem we can solve in this scenario.
It’s not a terrible heuristic, but neither is the labor theory of value. If you don’t think about it too much, both heuristics make “great sense.” But not when you think like an economist (which you usually do)! They’re both bad/incomplete enough that when you try to apply them too literally/broadly it leads to some really goofy thinking. I believe you’ve done some work cataloging the harms that came from efforts to apply the latter heuristic. (http://econfaculty.gmu.edu/bcaplan/museum/faqframe.htm)
You already know this. You taught me a class with this stuff!
The second question is poorly defined and has multiple plausible interpretations. Is Covid 2-5x worse than the flu for an individual conditional on having one or the other? Will it kill 2-5x more people than the flu? Are you 2-5x more likely to get it than the flu? There are presumably other reasonable interpretations of the question, and each of them will have different implications about the proper response.
The issues with this are further compounded by the fact that we still don’t know how bad covid will end up having been. Will deaths plateau around 200k? Double in the next six months? At this point it looks like 5-10x more people have died from covid than would have died in a typical flu season. There’s presumably some crowding out, so we could argue about the precise multiplier. But at this point it still seems radically premature to be overly certain about how much worse covid is.
None of this implies that you’re definitely wrong about the reaction to Covid. But this series of polls really doesn’t show what you seem to think it shows.
Denver
Sep 16 2020 at 1:06pm
Question: suppose we actually spent fewer resources on coronavirus. Would the third question still come out the same?
In other words, are people just rationalizing what has happened. Or do they truly think we should be spending so much time fighting coronavirus?
caldera
Sep 16 2020 at 1:12pm
I think any “cognitive dissonance” is instead explained by the ambiguity in the framing of this question: “Medically speaking, how bad is coronavirus compared to flu?”
Suppose coronavirus symptoms are 2x worse than the flu, but you are 100x more likely to get coronavirus. Is that, “medically speaking,” 2x worse, 100x worse, 200x worse, or some other amount worse? I don’t know.
Similarly, “medically speaking” may obscure the primary motivator of precautionary efforts: individual risk. Suppose coronavirus is undoubtedly 2x worse than the flu based on severity of symptoms, that may not mean it poses 2x the risk to a particular person. That is, the symptoms of coronavirus may be below a certain threshold where an individual thinks that both the flu and coronavirus pose no individual risk to him or herself, such that he or she should take no precautions. Or perhaps 2x (or 3x or 4x, etc.) the severity bumps the risk of coronavirus over a certain threshold where an individual wants to begin taking precautions.
I would be interest in seeing the answers to the following question: “How risky is coronavirus compared to flu?”
nobody.really
Sep 16 2020 at 1:20pm
1: Perhaps the poll results reflect Prospect Theory–the idea that people systemically prefer to hedge against certain kinds of uncertainty, even beyond a game-theoretical optimal level.
2: Perhaps the poll results reflect a lack of precision in the initial proposition, “The resources you spend mitigating a problem should be directly proportional to its overall severity.” How do you measure “overall severity”? In particular, what role does uncertainty play in that calculation? The concept of “Black Swans” reminds us that low probability events can have large consequences.
3: More generally (and as Daniel Carroll noted), perhaps the poll results reflect a non-linear phenomena. I’d characterize it as a problem of scenario analysis (or perhaps network economics?) Imagine you plan to fill a bucket today, and use the water tomorrow–but first you need to decide how much effort to put into patching the holes in the bottom of the bucket. You plot the cost of invested in filling various numbers of holes vs. benefit derived from each scenario. The data reveals that all the incremental benefit seems to come from filling the LAST hole. But if you were to calculate the AVERAGE benefit from each hole patched, you might adopt a sub-optimal strategy. In short, proportionality may make sense in the context of continuous (or discrete and non-correlated) phenomena–but that may not be what we’re talking about here.
Michael Koehler
Sep 16 2020 at 2:38pm
The following is my gut reaction and the gut reaction of people I know. I extrapolate that others think the same way.
I don’t know anyone that has died of the flu in my lifetime.
I get a flu vaccination every year.
I know several people who have died from Covid-19.
We should put a lot of effort into not dying from Covid-19.
Is there a more reasoned view? We certainly know more than we did in February.
robc
Sep 21 2020 at 7:39am
I am the reverse of you. I know several people who have died from the flu(a college professor of mine being the one that leaps to mind, it was 2-3 years after I had him, but he was still teaching at the time). I know of no one (personally) who has died of covid.
A Country Farmer
Sep 16 2020 at 10:15pm
Another fascinating aspect to this is that no one is talking about yearly flu deaths. Plenty of grandmas are also dying. Should we be quarantining every winter? As much as I disagree, I’d at least like to see the pro-quarantine crowd tackle this question.
Peter McCluskey
Sep 17 2020 at 12:40pm
It sure looks like we should be doing more about the flu in normal times – more handwashing, more social pressure to self-isolate when we have flu symptoms.
The people who described the heuristic as “solid gold” seem to be engaged in thoughtless virtue signaling. The right amount to spend on COVID is heavily dependent on whether it can be made harmless via vitamin D pills, etc.
Ankur Aggarwal
Sep 17 2020 at 7:30pm
Bryan,
With respect, “directly proportional” was always the wrong way to look at this. In a world of nonlinearities, ordinal comparisons make a lot more sense. And even then might not explain the full picture.
Suppose, for instance, that our current public spending on the flu, “only” saves 10,000 lives per year compared to a world absent that spending program, because that’s the best that the relevant medical technologies can do right now, but that we knew *for certain* that spending 11x as much on the novel coronavirus could save at least 20,000 and potentially up to 50,000 lives per year.
Given just that information, would it inherently be irrational to pursue the COVID treatment from a cost-benefits analysis (we’re going to assume away philosophical problems like the NAP and taxation, for the time being)?
Of course not! Our ability to spend more money on the flu and save more flu victims is tapped out, so to speak, resulting in low marginal return, whereas our ability to address corona is simply us deciding to pick up extant low-hanging fruit brought about by a new problem in our awareness of the world, so in this constrained scenario, the opportunity cost of this expenditure is quite low.
Now, this hypothetical may not be a realistic description of the world we live in–I don’t know, it’s not my field. But it certainly is possible and conceivable, and it’s up to domain experts to tell us whether such a comparison of solutions is (close to) the case (again, we’re suspending the question of whether or not it’s ethical to force people at gunpoint to pay into such a scheme).
PM
Sep 17 2020 at 11:37pm
One assumtion here is that the spend for Covid response had to be so high at all. With swifter amd bokder action, earlier investment in tracing, and social compliance, there might have been a normalizing of the economy by now. Leadership has made the response expensive.
Second, timescale is overlooked. If we are constantly investing that 5x with no return in decreased imfections/add economic benefit, does that make the investment unwise? It would be like commissioning an addition to your home, but the contractor burns your whole house down. It isnt the addition that is the causal issue.
Last, if you are a vulerable party, your individual weighting of the risk will be mich different. We accept the annoying regularly to accommodate vulnerable or disabled people. Where is that in the reasoning?
Jens
Sep 18 2020 at 10:49am
Why should there be proportionality between the medical severity of COVID-19 and the effort that has to be made to cope with it?
Michael Byrnes
Sep 18 2020 at 7:28pm
“Medically speaking” led me to think (and would have led me to answer, had I seen the question) from the perspective of a person infected with one or the other. That’s not the right way to think about the societal impact of Covid-19.
Ebola is an illustrative example. Nasty, nasty virus far, far more deadly than the flu. “Medically speaking” I’d guess more than 11x worse than the flu. But Ebola is far less impactful at a societal level. (See, 2014).
chris
Sep 21 2020 at 10:55pm
Very easy. Directly proportional does not mean the slope equals 1. All three statements can be true at the same time if people believe a twice as bad problem warrants four or five times as strong a reaction.
Comments are closed.