What does this economist think of epidemiologists?

I have had fringe contact with more epidemiology than usual as of late, for obvious reasons, and I do understand this is only one corner of the discipline.  I don’t mean this as a complaint dump, because most of economics suffers from similar problems, but here are a few limitations I see in the mainline epidemiological models put before us:

1. They do not sufficiently grasp that long-run elasticities of adjustment are more powerful than short-run elasticites.  In the short run you socially distance, but in the long run you learn which methods of social distance protect you the most.  Or you move from doing “half home delivery of food” to “full home delivery of food” once you get that extra credit card or learn the best sites.  In this regard the epidemiological models end up being too pessimistic, and it seems that “the natural disaster economist complaints about the epidemiologists” (yes there is such a thing) are largely correct on this count.  On this question economic models really do better, though not the models of everybody.

2. They do not sufficiently incorporate public choice considerations.  An epidemic path, for instance, may be politically infeasible, which leads to adjustments along the way, and very often those adjustments are stupid policy moves from impatient politicians.  This is not built into the models I am seeing, nor are such factors built into most economic macro models, even though there is a large independent branch of public choice research.  It is hard to integrate.  Still, it means that epidemiological models will be too optimistic, rather than too pessimistic as in #1.  Epidemiologists might protest that it is not the purpose of their science or models to incorporate politics, but these factors are relevant for prediction, and if you try to wash your hands of them (no pun intended) you will be wrong a lot.

3. The Lucas critique, namely that agents within a model, knowing the model, will change how the model itself operates.  Epidemiologists seem super-aware of this, much more than Keynesian macroeconomists are these days, though it seems to be more of a “I told you that you should listen to us” embodiment than trying to find an actual closed-loop solution for the model as a whole.  That is really hard, either in macroeconomics or epidemiology.  Still, on the predictive front without a good instantiation of the Lucas critique again a lot will go askew, as indeed it does in economics.

The epidemiological models also do not seem to incorporate Sam Peltzman-like risk offset effects.  If you tell everyone to wear a mask, great!  But people will feel safer as a result, and end up going out more.  Some of the initial safety gains are given back through the subsequent behavioral adjustment.  Epidemiologists might claim these factors already are incorporated in the variables they are measuring, but they are not constant across all possible methods of safety improvement.  Ideally you may wish to make people safer in a not entirely transparent manner, so that they do not respond with greater recklessness.  I have not yet seen a Straussian dimension in the models, though you might argue many epidemiologists are “naive Straussian” in their public rhetoric, saying what is good for us rather than telling the whole truth.  The Straussian economists are slightly subtler.

4. Selection bias from the failures coming first.  The early models were calibrated from Wuhan data, because what else could they do?  Then came northern Italy, which was also a mess.  It is the messes which are visible first, at least on average.  So some of the models may have been too pessimistic at first.  These days we have Germany, Australia, and a bunch of southern states that haven’t quite “blown up” as quickly as they should have.  If the early models had access to all of that data, presumably they would be more predictive of the entire situation today.  But it is no accident that the failures will be more visible early on.

And note that right now some of the very worst countries (Mexico, Brazil, possibly India?) are not far enough along on the data side to yield useful inputs into the models.  So currently those models might be picking up too many semi-positive data points and not enough from the “train wrecks,” and thus they are too optimistic.

On this list, I think my #1 comes closest to being an actual criticism, the other points are more like observations about doing science in a messy, imperfect world.  In any case, when epidemiological models are brandished, keep these limitations in mind.  But the more important point may be for when critics of epidemiological models raise the limitations of those models.  Very often the cited criticisms are chosen selectively, to support some particular agenda, when in fact the biases in the epidemiological models could run in either an optimistic or pessimistic direction.

Which is how it should be.

Now, to close, I have a few rude questions that nobody else seems willing to ask, and I genuinely do not know the answers to these:

a. As a class of scientists, how much are epidemiologists paid?  Is good or bad news better for their salaries?

b. How smart are they?  What are their average GRE scores?

c. Are they hired into thick, liquid academic and institutional markets?  And how meritocratic are those markets?

d. What is their overall track record on predictions, whether before or during this crisis?

e. On average, what is the political orientation of epidemiologists?  And compared to other academics?  Which social welfare function do they use when they make non-trivial recommendations?

f. We know, from economics, that if you are a French economist, being a Frenchman predicts your political views better than does being an economist (there is an old MR post on this somewhere).  Is there a comparable phenomenon in epidemiology?

g. How well do they understand how to model uncertainty of forecasts, relative to say what a top econometrician would know?

h. Are there “zombie epidemiologists” in the manner that Paul Krugman charges there are “zombie economists”?  If so, what do you have to do to earn that designation?  And are the zombies sometimes right, or right on some issues?  How meta-rational are those who allege zombie-ism?

i. How many of them have studied Philip Tetlock’s work on forecasting?

Just to be clear, as MR readers will know, I have not been criticizing the mainstream epidemiological recommendations of lockdowns.  But still those seem to be questions worth asking.

Comments

You don't seem to understand that "real science" such as medicine often suffers from weak statistical analysis and modeling capability. There is not just p-hacking in journals but a gross misuse of basic stats that would not get through a normal economics seminar. Epidemiological models are hardly in the same class as Newtonian mechanics and scientists know that. They don't have the same experimental base and they don't have the level of statistical sophistication of top statisticians or econometricians.

Does not seem to realize that Newtonian mechanics is anything but a system based on a model that is incorrect, covering only a portion of such modern endeavors as maintaining satellite fleets in precise orbits. Unless you believe that Newtonian mechanics is sufficient for a global GPS system.

Or alternatively, should be aware that engineering is not science - Newtonian mechanics is useful, but is more than a century out of date providing a correct foundation.

Remember, wiki, if you give the troll an opportunity to change the subject, he will. He's only here to hijack the discussion. He doesn't actually care about the topic, and his bits of knowledge are just gleaned quickly from the Wikipedia articles he often quotes so heavily.

This website truly has all the info I needed about this subject and didn’t know who to ask.

Every scientific model is incorrect, but some are useful.

General relativity is an incorrect model. We know that because it's incompatible with quantum mechanics, and a grand unified theory still eludes us. But it lets us do accurate GPS calculations.

Newtonian mechanics is an incorrect model. But we use it all the time because it gives extremely accurate answers under nearly all "normal" conditions.

The flat earth theory is incorrect. But we use it all the time in practice. If you calculate the surface area of a football field by multiplying 100 yards of length by 160 feet of width—instead of using spherical trigonometry—then you've used the flat earth model.

"Epidemiological models are hardly in the same class as Newtonian mechanics" because Newtonian mechanics tells us precisely where Neptune will be in a million years, and meanwhile we have no clue why New York City has more Covid deaths every single day than the cumulative all-time total for all of Canada.

Not to just overly lost in the weeds, but this is absolutely not true - §Newtonian mechanics tells us precisely where Neptune will be in a million years.' Mainly because Newtonian mechanics covers only a few of the relevant variables. Which is fine, because much like epidemiology, we use whatever relevant and reliable data is available for the model, then continue to refine it over time. The main point of epidemiology is not the model, it is ensuring reliable data on which to deal with something like a pandemic. It is reliable data which allows for better models, something that would have seemed straightforwardly obvious before this post appeared. Particularly when talking about a disease that has not even existed long enough to know what a 12 month cycle looks like.

People are often surprised to learn the precise location of Neptune cannot be calculated to arbitrary precision over an arbitrary time, it is an n-body problem.

Effectively, one cannot calculate the orbit of any planet accurately beyond the point at which gravitational influences other than the Sun and the planet come into play... fortunately these effects don't matter much on scales relevant to humans.

https://en.wikipedia.org/wiki/N-body_problem

That's right. All of the planetary orbits are mathematically chaotic. We can do fairly long term predictions, but as they get farther out (or backward) in time, they get less accurate.

Computers can simulate the n-body gravitational perturbations. Chaos theory eventually prevails, but you can accurately predict for maybe a few tens of millions of years. You need general relativity for Mercury's precise orbit, but Newtonian ought to be good enough at the distance of Neptune.

Within certain limitations... even for Newtonian physics, some orbits cannot be calculated to arbitrary precision. One might account for the influence and interdependent movements of the planets, but not the vast numbers of small objects in (say) the Oort cloud or asteroid belt, many of which are as yet undiscovered. Again, very small effects, of course.

Although, to be fair, I don't know at what point the effect of those influences are smaller than the Planck length, at which point you could presumably ignore them and call the result completely accurate. Might not matter unless the orbit you're interested in is actually in the Oort belt or asteroid belt.

“...meanwhile we have no clue why New York City has more Covid deaths every single day than the cumulative all-time total for all of Canada”

Really?

Here's a thought.

You have superspreaders whose R is very high and some of these superspreaders interact heavily with other superspreaders.

Celebrities, politicians, the famous and quasi-famous are superspreaders. They meet and greet many, often strangers, on average on a daily basis. Many of these are also famous or quasi-famous superspreaders themselves, also often relative strangers. The rest are ordinary folk — “the little people”.

Included amongst these famous and quasi-famous superspreaders are the jetsetters. They frequently flit around the world including to relatively exotic places like Asia, including China.

New York is the jetsetting superspreader capital of the world. A place where the jetsetting famous and quasi-famous or simply wealthy and connected superspreader socialites socialise heavily.

It is no wonder that New York City got hit more quickly and hit hard.

Northern Italy? That's where the fashion jetsetting superspreaders congregate. No mystery there either.

Northern Italy also has a large Chinese population employed sewing clothes. Where are they from? Why, coincidentally, Wuhan, with direct flights from Milan to Wuhan.

Actually the incidence of the virus in Prado (where much of the Chinese manufacturing is done) was lower than in other areas of northern Italy. This theory is appealing intuitively but the data doesn't back it up. It is more likely that there was travel by Italians to China for business that wasn't being tracked and people unknowingly brought the virus back to Italy

"they don't have the level of statistical sophistication of top statisticians or econometricians."

Have you noticed that nothing important has been discovered or unearthed by statisticians or econometricians? This turf war brewing between economists and epidemiologists is really silly since both are bit players at best.

Prefect comment, Gangs. From a scientific standpoint, econometrics is pure garbage.

Au contraire GoNY. They have discovered AGW. Prior to that they discovered the forthcoming mini ice age. Prior to that they discovered peak oil. I think there was even a wager on peak copper. Their contributions have been significant for political careers and in getting government grants.

One of these things isn't like the others. Peak oil is a certainty due to the non-sustainable nature of oil - the fringe Russian abiogenesis theory isn't true - the only question is when will it occur, assuming it hasn't occurred already. AGW is true as well, but it's not as blindingly obvious as peak oil.

I get the impression that epidemiologists are extremely sophisticated in their use of statistics. That's why their predictions tend to be so accurate in the face of incomplete and poorly understood data. As data accumulates and experiments are performed, they adjust their models. We see this in the current epidemic.

Economists, in contrast, seem to be very naive, and while they use statistics, they rarely adjust their priors or update their models. For example, after failing to predict the 2007 meltdown, they continued using their failed models and repeatedly expressed surprise at the lack of inflation and slow rate of recovery, the kind of stuff anyone with a primitive acquaintance with accounting would have expected.

YOU. CANNOT. BE. SERIOUS.

Economists are naive? And yet Neil Ferguson is bathed in attention for a BS model that assumes NO BEHAVIORAL RESPONSE.

Behavioral sciences are kinda sort of important as Ferguson and Murray have brought nothing but disgrace upon their discipline.

"Epidemiological models are hardly in the same class as Newtonian mechanics and scientists know that. "

You could say the same thing about econometrics.

Where is the evidence that economics research is somehow more statistically robust than say medicine, epidemiology or psychology? My impression is that economists are completely uninterested in issues like publication bias, p-hacking and (lack of) replications. If you look at a typical econometrics textbook like Wooldridge's, it's all about the NHST ritual, with no attention paid to how the incentive structures of scientific publishing will virtually guarantee that the NHST ritual will produce a deeply biased literature. This is exacerbated by the fact that economists prefer an alpha level of 10%, making it childishly easy to publish anything as significant.

Has there even been analyses of statistical power in economics research, the kind that have been common in psychology since Cohen (1962)? In psychology it's at least long been acknowledged by leading researchers that the system is fundamentally broken, whereas in economics no one seems to care about the GIGO nature of the published literature.

In general, the grasp of basic statistical concepts is weak across the natural and social sciences. Economics is no exception.

Found the high schooler.

GRE scores are useless, btw, especially for those gifted, divergent thinkers.

Extremely useless. I almost spit out my water when he actually said that.

He IS an academic, these guys need to defend the castle they've created otherwise what is the point

As a guy with a nearly perfect GRE score, I agree that GRE scores are extremely noisy, basically useless.

As a person who has never taken a GRE nor even knows what it is, sounds a bit like standard ivory towerism, I.e. "I have a PhD therefore I will always make better decision than you".

It sounds that way, but it's not. The GRE is a set of standardized exams used in the evaluation of applications to master's and PH.D program. It is precisely as relevant here as an SAT score is to a software developer with decades of experience.

Humble brag.

GRE score is just an euphemism for "what is their IQ"

Science and Policy is iterative and models for rare events are going the imperfect. More complicated model is no fix for crappy data.

Gifted, divergent thinkers are useless if they don't have high cognitive ability (as measured by GREs, for example).

Yes Matt, thank you for stating the obvious. But it’s been found over and over again that GRE lacks concurrent and predictive validity. Princeton Review will explicitly state thot in their workbooks.

Validity for what? It validity seems reasonable enough for the typical academic outcomes, especially given that we're dealing with situations characterized by reduced variance and collider biases.

Ah yes, if you're good at test taking, you're going to perform well in academia, the mothership of test taking. Thank you for this profound insight.

IQ tests only tell you how good you are at being intelligent.

"Intelligence" is not well-defined, no matter what IQ fundamentalists keep asserting. Insofar as IQ measures something, it's your ability to take tests and guess what the test designers want you to answer. A useful ability that certainly comes in handy in many career paths but you'd be hard-pressed to convince everyone, from neuroscientists to AI researchers to developmental biologists to animal behavior scientists that's all there is to the all-encompassing and far-reaching concept of "intelligence".

"Cognitive ability" is not well defined and it's certainly not "answering a bunch of high school level questions very quickly".

This is just Tyler tiptoeing around asking about their IQ scores.

John Ioannidis only Truth seeking Epidemiologists. His US death est was 40k with CFR < 0.1

I think epidemiologists come with a variety of backgrounds, which makes some generalizations difficulty. Some are PhDs, some are MDs. Some might even be MScs.

Those guys always seem to forget their Kahnemann and Tetlock. What are their biases? Are they aware enough of them? Are they aware that, as experts, they too often tend to confuse their own subject matter expertise with expertise in issuing forecasts? Do they know that in Tetlock's experiments often the subject matter experts came in dead last, after the regular people and the dart throwing monkey? That Rosling made it a subject of his Ted talk that the health policy types of the Karolinska are, in fact, worse than the monkey? There again I remind people of this dashboard.
https://goodjudgment.io/covid/dashboard/

@yo- Good rant but the Good Judgement dashboard is saying likely over 23M C-19 cases by a year from now. You saying that's too high? Your phrase "then again" seems to undercut that. The year is not yet over and despite some curve flattening I don't see new cases radically going down yet; new cases being a function of testing, the more tests, the more new cases. BTW in most Tetlock type experiments the hierarchy found is: Truth significantly beats expert beats layperson significantly beats random guess. That holds in both life and chess (life is like chess). If Average Joe plays chess with expert me, statistically they will lose, even if they think they're good.

"If Average Joe plays chess with expert me, statistically they will lose". Nonsense. I'm going to ask for a handicap, otherwise why should I waste my time playing against you?

Have you read his book? "Foxes" beat hedgehog experts on average (and by quite a margin).

Just wanted to note here that at least one of Tetlock's original superforecasters is an epidemiologist.

This particular "good judgment" dashboard seems like garbage to me. Why in the world are the bins so large?

For U.S. deaths by March 2021, the category with 79 percent of the "super predictors" is "35,000 to 350,000". And the second highest category, with 24 percent, is "350,000 to 3.5 million".

How is that useful, from a policy making standpoint? There's a huge difference between 35,000 deaths and 350,000 deaths, in terms of impacts on the health care system and even the economy. And that's ignoring completely the 24 percent that say deaths will be 350,000 to 3.5 million. (The 3.5 million is greater than all U.S. military deaths from all the wars the U.S. has fought, *combined*. And we're talking about 3.5 million deaths in less than a year!)

I don't know why people who supposedly specialize in "good judgment" would set up such ridiculous large bins.

The bins are large because the uncertainty is large. I think they don't want to have many more than 5 or 6 bins, and with only that many bins, they need large ones to include the full range of plausible uncertainty.

We're talking about forecasting something that grows exponentially in its early phases, so order of magnitude bins is reasonable.

It illustrates that Tetlock's game is giving intentionally useless forecasts to encourage despair and discourage policy action on climate change, etc.

"The bins are large because the uncertainty is large. I think they don't want to have many more than 5 or 6 bins,..."

And what is their reason for not wanting to "have many more than 5 or 6 bins? It better be a pretty damn good reason, because with the bins they have, their "superforecasting" isn't worth squat, from a policy-making standpoint. Their two "most likely" bins cover a factor of *one hundred* in deaths in the U.S. in the next year...from 35,000 to 3.5 million. That's essentially worthless, because 35,000 deaths are not dramatic (a "normal" flu season's worth), and 3.5 million deaths in the U.S. is far worse than the 1918 Spanish Flu.

And even if they had limited themselves to six bins, they could have had far more useful bins, from a policy standpoint. For instance if the bins had been 1) <5000; 2) 5000 - 25,000; 3) 25,001 - 125,000; 4) 125,000 - 625,000; 5) 625,000 - 3.1 million; and 6) 3.1+ million....

...that would have been only six bins, but far, far more useful bins from a policy making standpoint. So the "Good Judgment" website showed shockingly poor judgment in setting up the bins, if their goal was to actually help policymakers with accurate forecasts of the likely U.S. death toll.

If elections were held every 100 years, how would you decide how to run a campaign, whom to vote for, or whether to run for office yourself? The facts are few and the events too infrequent to build a solid body of knowledge. And if that's true, how legitimate would an interdisciplinary effort be?

Taking this analogy as read (high lethality pandemics = elections), you might be able to build an estimation from more generalized social dynamics and popularity contests, which go on all the time, and even in other species, and then you'd build up... if you had a good means of gathering good data. (Politics is kind of a special case of course where it is particularly challenging to get a true look at what is going on because there are specific interests in you not having a true look at what's going on if it's bad for them.)

Conjecture: The limitations may be more around marshaling enough data on the viral spreads we do have, because we don't gather much data on them, because the costs were until recently considered too high relative to benefits and computer / diagnostic penetration too low. Less than that the actual viral spreads (usually involved diseases of much lower lethality and/or lower novelty) do not present in theory enough data to use for inference.

My view - as a trained "hard scientist" - is that epidemiologists (like most people) are awful at statistical / probabilistic reasoning. They seem to have enormous faith in models built on incomplete and almost certainly skewed evidence. They default to accepting data as it comes versus taking into consideration the conditions under which it has been gathered. They have the natural human tendency to exaggerate the importance of what they do. They have insufficient epistemic humility, and fail to think beyond the very next step to the endgame of how we will exit.

Overall, I am extremely unimpressed with the "experts" on this one. There needs to be a reckoning for those who spread panic among policymakers and the public, while at the same time parroting official guidance that somehow masks were extremely important for health care professionals but completely ineffective when used by the general public. I am actually taken aback by how much respect I have lost for epidemiology and public health professions more generally as a result of seeing them react to this crisis.

I've never understood the line of thinking in your last paragraph. They are government bureaucrats for the most part, lost how anybody had any respect or faith in them in the first place. Sure there are some good ones, they don't get promoted, much less make it on TV.

I work in a "hard science agency" and the amount of science that goes into decision making processes or output it is next to nill. I would go as far as to say it's actively surpressed because it's the sacred cows of those that do get promoted. It's all about incentives and the incentives are mediocrity rises to the top.

That would be the administration leadership, rather than epidemiologists, you should be focused on. Also not clear your "spreading panic" comment makes any sense. We already lost tens of thousands of lives to this thing and that is WITH the state orders to stay home in the most populous places. I am taken aback by how totally incapable some are of trying to grasp the counterfactuals in this whole scenario.

12/31: Taiwan begins screening of all flights from Wuhan

1/14: WHO issues statement that there is no human to human transmission of Corona

1/20: Taiwan bans mask exports, activates CECC, issues mask guidance for entire population

2/1: CDC says Corona low risk to America, bans use of influenza study to test for community transmission, threatens Helen Chu for researching transmission, FDA says testing for Corona is illegal unless with CDC test that does not work

2/10: CDC says low risk, tells Americans to not wear masks

3/5: CDC once again issues guidance to not wear masks, Cuomo and de Blasio tweet about which movies New Yorkers should attend in theaters

4/1: CDC reviews mask guidance

4/3: CDC issues mask guidance for entire population

Your timeline suggests we should fire a bunch of people in the federal government, replace them with Taiwanese (and Korean) infectious disease experts, and question why public health decisions of huge import are entrusted to the paper-pushing commercial lawyers who typically occupy state and local elected office. Sounds like a plan to me!

Even for fools who have faith in technocratic rule, it's a stupid plan based on the false assumption that the U.S and its institutions, its customs, etc. are or could be made similar to those of S. Korea or Singapore.

I don't understand your argument. It is the politicians that ultimately decide whether there will be stay-at-home orders, not the CDC "experts" or the lower-level officials in the executive branch. The plan is not to make the institutions and customs similar to Koreas, but to replace clueless "experts" recommending do nothing with smart, accurate experts, and hope that the politicians do a better job through being better advised. I think the people who make these kinds of comparisons are suffering from Dunning-Kruger syndrome in that they understand how government works much less than they think they do.

New Zealand doesn't have similar customs to Taiwan, but it has had only deaths so far.

*only 5 deaths

"They default to accepting data as it comes versus taking into consideration the conditions under which it has been gathered."

Epidemiologic training focuses heavily on measurement error (artifacts in data), study design (i.e., approaches to collecting and analyzing data), proper analysis under uncertainty about the data-generating mechanism, avoidance of naive use/interpretation of results from hypothesis testing, etc, etc.

Then why has the credulousness of the epidemiologists and public health experts in the public eye been so on display? My guess is that in an academic setting they are trained without internalizing the lessons, and then find their training difficult to put into practice, defaulting to taking numbers as reported, making idiotic simplifying assumptions to project forward, and presenting the results as fact.

Seems plausible. Compare the teaching in science education that one must report what one planned to do and what happened in an experiment, with the historical reality of publication bias and p-hacking, only recently beginning to be corrected in fields like psychology (where p-hacking was endemic) and medicine (where publication bias is still an issue).

Well, as they say, you can guess in one hand...

You must hang out with different epidemiologists than I do, because all I hear about are the difficulties inherent in modeling or making sound inference based on incomplete information and discussion about priorities for data collection to fill in those gaps. Almost everyone I know working in ID modeling has been talking about these issues incessantly. And some have raised another point: mechanistic models are not necessarily predictive models in the sense of giving a well-behaved forecast. A lot of modeling focuses on trying to figure out how various structural assumptions affect inference about the potential course of epidemics. If we do things right, of course, the worst-case scenarios in the models should never come to be.

Besides, the reason to take this seriously *is* the uncertainty.

Easily the best comment on the thread.

The lack of proportion in the thinking of the “scientific community” during this “crisis“ was shocking.

If we want to be charitable, we could observe that the scientists did not predict the behavior of the media which took those while raised the most alarm and amplified their signal tremendously.

Sadly it is the scientists who will bear the brunt of the world ire over this debacle, when it is the media which is the issue.

Epidemiologists are utilitarians; they deal in lives. They do that better than economists.

They care not about the cost of a life saved.

Epidemiologists are fully aware they also deal in deaths. Whether they do that better than economists is open to debate, but they do seem to be fairly skilled in preventing death better than the centuries long practice of letting a disease run its course.

And whatever epidemiologists think about the cost of life, they are not in charge of the political system. A system where essentially everyone thinks their life is beyond price.

Like doctors, epidemiologists (I hate to generalize) tend to try to solve the problem that is right in front of them -- a population of sick people or the threat of one -- without any awareness of opportunity costs. A life saved, at the cost of how many other lives not saved (in the future, so invisible at the moment)? What is seen and what is not seen. Bastiat.

Which contradicts that perspective. That is triage, which most definitely involves the life and death weighing of opportunity cost. In novel pandemic terms, the question being who gets the ventilator, the 30 year old or the 86 year old?

That an individual doctor working on an individual patient may become focused is not the same as saying that those involved in practicing medicine are completely unaware of opportunity costs.

No it's not mere triage if the lost life is "in the future, so invisible at the moment".

For example, a cancer patient who will die a year from now because biopsies aren't being performed anymore. Or someone who commits suicide from despair in the fifth year of Great Depression II.

Of course one can define triage as being extremely short term, but biopsies are also not being performed because people are cancelling appointments to avoid the chance of catching an extremely contagious virus which may lead to fatal consequences. These cancellations are not being done by doctors or epidemiologists, but by individuals in the U.S. who just may have been aware of this disease back in mid-Mach, as European cases began their steep climb.

From a broad enough perspective, opportunity costs represent nothing more or less than making choices.

Triage means assessing known victims with evident acute medical conditions and sorting them by priority. You can't "triage" someone you've never seen, who currently seems fine and is perhaps far away, and neither of you even know yet of their future misfortune.

You're not going to win arguments by inventing your own definitions of words.

The notion that elective procedures are being canceled mostly at the initiative of patients themselves is absurd. Just read the news. Doctors who normally perform those procedures are either being laid off or reassigned to coronavirus emergency duty.

Do you think triage guidelines are made up moment by moment? You are of course correct that in a given situation, different situation dependent decisions will be made. However, the guidelines for triage, adapted as they might be to any given situation, are nothing but guidelines for making decisions involving opportunity costs. So yes, starting on March 21 in Strasbourg, patients who had yet to be infected or arrive were already being subject to being triaged, in a statistical sense during a novel pandemic- 30 year old gets a free ventilator (or a trip to Germany where there were free ventilators), 80 year old gets palliative care before dying without ever going on a ventilator. One can hope that the situation is no longer so desperate, but someone infected on March 23 and who arrived at a hospital in Strasbourg on March 30 had already been fully triaged, without anyone seeing them.

In this thread prior_approval somehow misunderstands the difference between triage and opportunity costs

prior_approval prepares to write a children’s book: Bastiat and the candlemaker’s guild, a children’s guide to medical triage

My dentist cancelled all non-emergency appointments. My dental hygienist is very capable, but was not reassigned you COVID duty at a hospital. My understanding is that is was due to both patient cancellations and reducing risk to staff.

"to avoid the chance of catching an extremely contagious virus "

This has been considered a medium level contagious virus. You might be thinking of the measles.

An epidemiologists job is not to weigh opportunity costs of things outside their scope. They can say "we can probably reduce X infections by doing A, B and C." It is the policymakers that need to weigh the cost of those measures. The problem is, this country is pretty bad at putting a cost on lives, at least in the health care space. For example, drug companies can regularly get Medicare and other insurers to pay hundreds of thousands of dollars for a drug that might extend the life of any already quite sick person by a couple months. Is that worth it? Who knows--we refuse to grapple with this question as a society.

Exactly.

Thanks to the politicization of everything, the pursuit of information in an imperfect data environment has been hopelessly contaminated with the pushing of political agendas and turf wars.

Instead of fighting about opinions and options, the war is being waged over the underlying facts. And in that environment, uncertainty, estimates, assumptions, and statistical limitations tend to get lost.

As some Bush admin cretin said, we create reality, we don't do nuance.

And yet those experts and bureaucrats cannot help themselves from making those policy judgments themselves in their work and becoming advocates. E.g., behavioral scientists in the UK questioned the behavioral fatigue argument used as one of the justification the UK's initial policy before its U-turn. Despite being unable to provide any significant contrary empirical studies themselves, they nonetheless advocated a lockdown policy: "Experience in China and South Korea is sufficiently encouraging to suggest that this possibility [radical behavioral change] should at least be attempted."

So until these experts and bureaucrats learn to STFU (which will be never) and only give their policy advice when asked, they should be treated like any other political actor.

Superb post. Thank you. I will use this in my work.

I'm not sure what you mean by Mexico and Brazil being the worst countries. Worst for reporting of COVID-19? Worst for preventive measures? Worst for treatment?

I think he's bought into the "Bolasanario man bad" trope. The Brazilian president has done the math and even if 1% of the country would die from this bug, it would hurt less than a prolonged shutdown of their economy. He's said something to the effect of "we are economically on the verge of becoming Venezuela."

And he's right. And that would hurt more than that having a few hundred thousand dead 80 year olds.

I think people overstate the downside of an economic shutdown and understate the costs of killing millions of people. Brazil, during the shutdown, might be as bad as Venezuela but odds are it will recover, unless, of course, Brazil is in as bad structural and political shape as Venezuela. Killing two million people to avoid a bad Q2 economic report strikes me as rather callous and bloodthirsty. That seems to be the tradeoff.

Much to the chagrin of many, epidemiology, especially behavioral, has a lot of qualitative measurements due to the *random* variability in human behavior. But do check out the work of Dr Nicholas Jewell at Berkeley's School of Public Health for more info on what *academic* epidemiologist do: https://statistics.berkeley.edu/people/nick-jewell

"What is their overall track record on predictions, whether before or during this crisis?"

Probably better than most economists. Then again my 8 year old is smarter than most economists.

Thank you for the deep insight.

Fantastic. The first chink of light suggesting Tyler is starting to cotton on to the fact that we have been scammed, and seasonal flu (with sci-fi name) has been used as a cover for economic shenanigans and removal of civil liberties. Let’s hope he puts that big brain to use working out who was behind this and what their intention is. Maybe he could start by looking at Robert Kennedy Jr’s recent attack on Bill Gates’ sinister vaccine activities. https://childrenshealthdefense.org/news/government-corruption/gates-globalist-vaccine-agenda-a-win-win-for-pharma-and-mandatory-vaccination/

" Bill Gates’ sinister vaccine activities." LOL

Bill Gates status has done nothing but gone up during this crisis.

Ridiculous conspiracy theory. This is a novel human virus, with different symptoms and outcomes to seasonal flu, and its RNA has been sequenced. In fact, there's a new test which involves sequencing the RNA.

Prior to covid-19 times I was conscious of epidemiology only as the science that develops and perpetuates suspicions about the effects of various foods, such as the idea that salt causes high blood pressure or that butter and eggs cause heart disease. Since then I have been assuming that there must be a higher sort of epidemiology that is applied to real public health problems, such as pandemics.

You may want to read this - The Ghost Map: The Story of London's Most Terrifying Epidemic – and How it Changed Science, Cities and the Modern World is a book by Steven Berlin Johnson in which he describes the most intense outbreak of cholera in Victorian London and centers on John Snow and Henry Whitehead.

Marketing and Advertising. Various industries employ pay "experts" to persuade the public that some Product A is bad for them and that they should be using Product B. Later, it is shown that the dangers of Product A was well overstated. Economic interests, both personal and other, drive the data.

#1 = Le Chatelier's principle

Somebody doesn't understand what the inputs to the science of epidemiology are. It really exposes prior as a troll when he races to be the first comment.

When you look back at Pandemic 2020 are you going to be proud that you spent much of it replying to online trolls?

Since your scorn of actual citations is well known (in part because you believe all citations are from wikipedia), will an OED definition suffice?

The branch of medicine which deals with the incidence, distribution, and possible control of diseases and other factors relating to health.

So, the inputs are basically sick/unhealthy people and their location, with modelling being the least important part of what epidemiology concerns itself with. Determining what is occurring with an epidemic involves data collected and reported with as much reliability as possible.

I think you have a very uncharitable reading of Tyler, basically putting words in his mouth that I can’t imagine him saying.

He almost certainly thinks epidemiology is a social science. You do too, right? What made you think he doesn’t think it is a science?

'You do too, right?'

Only if you consider medical science a social science, which I don't. We could debate endlessly, but medical science actually works best when applying the rigor of the hard sciences to collecting and evaluating data.

And they’re off!

Happy Easter Sunday.

You should be asking the same about economists. It's funny to see economists getting outmaneuvered by another class of "experts" that have the ear of the ruling class. I hope more economists get kicked out of the room when adults are making decisions.

Literally everything in this post comes from critiques of economics or economists, especially macro. It's very clever in a way. You should read the rude questions as a measure of the vulnerability of epidemiology to a "Revolt of the Public" once the dust has cleared a bit.

The ongoing contempt from this blog for epidemiology, a field that Tyler evidently lacks even a passing knowledge of, is very disappointing. And thinking that GRE scores mean anything is just juvenile...

If you've spent much time on this blog, you prolly would have noticed that despite its profound (& documented) inadequacies TC worships at the altar of IQ.

IQ/GRE denial is very juvenile, and often comes from people with above-average IQ who don't care if they're kicking away the ladder to higher levels of education for certain smart members of historically-discriminated-against groups. Why would tests which don't do anything be used so widely?! This kind of thinking shades very quickly into conspiracy-theory nonsense and identity politics conspiracy-theory nonsense in particular - why, because black people on average score, it must be a conspiracy to keep black people out of academia!

Just check out the books on the research on this stuff. IQ has high predictive ability and THIS REPLICATES.

*score lower

The 'researchers' behind IQ are known for quackery or outright fraud. Lynn misreported the 'national IQ of Guinea' using the results from a school for intellectually disabled children in Spain. He'd also infer missing data from a country by averaging the IQs of neighboring countries. Rushton tried to apply the pleiotropy behind the melanocortin system of some animals to humans i- order to show blacks were more agressive and dumber, despite multiple biologists repeating that this model specifically did not apply to humans. Kanazawa wrote a paper that attempted to prove the cold winter theory by assuming the Earth was flat and populations just crossed the atlantic. Both Jensen and Lynn would routinely remove IQ scores where blacks scored higher or as well as whites. Burt literally made up data.

It's not that there's a conspiracy. It's just that IQ fundamentalists have a very poor track record and aren't known for their credibility.

Ooh, I'd missed the GRE dick measuring competition on first read. Mine is 2380/2400 and 8". Its nice knowing that very few people are beyond me on that particular "efficient frontier"

2380 isn't a GRE score. Seems all your measurements are 7x larger than reality.

and we all should know that the important metric is shaft length + 3* circumference.

It was when I took it. Logic / Math / Verbal, each out of 800 points.

Libertarians are showing their Trumpiness. This is the social science equivalent of asking the black guy for his birth certificate. They can't win the debate so they ask for GRE scores(?!).

"equivalent of asking the black guy for his birth certificate"
Er, that was Hillary actually...

One thing both economists and epidemiologists seem to be lacking is an awareness for the problems of aggregation. Most models in both fields see the population as one homogenous mass of individuals. But sometimes, individual variation makes a difference in the aggregate, even if the average is the same.

In the case of pandemics, it makes a big difference how that infection rate varies in the population. Most models assume that it is the same for everyone. But in reality, human interactions are not evenly distributed. Some people shake hands all day, while others spend their days mostly alone in front of a screen. This uneven distribution has an interesting effect: those who spread virus the most are also the most likely to get it. This means that the infection rate looks very higher in the beginning of a pandemic, but sinks once the super spreaders has the disease and got immunity. Also, it means herd immunity is reached much earlier: not after 70% of the population is immune, but after people who are involved in 70% of all human interactions are immune. At average, this is the same. But in practice, it can make a big difference.

I did a small simulation on this and came to the conclusion that with recursively applied Pareto-distribution where 1/3 of all people are responsible for 2/3 of all human interaction, herd immunity is already reached when 10% of the population had the virus. So individual variation in the infection rate can make an enormous difference that are be captured in aggregate models.

My quick and dirty simulation can be found here:
https://github.com/meisserecon/corona

+1000. This is such a critical point. Tyler, if you get this deep in the comment thread, I sincerely implore you — read Kronrod's comment, and if you agree with it, signal boost it.

As noted below, using actual preliminary epidemiological collected data using antibody tests, it is absurd to talk about herd immunity at 10%. Unless something beyond the death rate is truly unusual about Germany, which seems to be the latest ploy among those who have no interest in data to support whatever it is they want to believe. Which at an economics blog is an a priori given.

Not sure it's reasonable to make deductions about herd immunity requirements using data from the hardest-hit location in Germany. There's every reason to think Heinsberg has elevated R0 compared to the national average.

Also, as I discuss in a comment below, the 10% number can be wrong without refuting Kronrod's underlying point.

Responding to myself to add: I don't actually know if Heinsberg is the hardest-hit region in Germany, I'm just going off the comment below by 'I am very interested, can I sign up for your newsletter?', who is arguing the opposite side of this from me.

It had that honor, notably so, and is a major reason why Germany was also a bit ahead of the curve compared to Italy or Spain, as cases spiked rapidly from a single married pair as the initial source. And its initial lockdown was two weeks ahead of the Germany wide lockdowns, though not as stringent at first. Heinsberg provided a fact based view into what covid19 was like, and why Germany never entertained the idea of letting it spread uncontrollably.

It is also why the first large scale antibody tests of an entire regional population are being done there. The Germans are very interested in what comes after the lockdowns, but are going to base their policy making on as much data as possible. Much like how the initial lockdown decisions were based on what was seen in Heinsberg. Along with Italy and Spain, of course, but having reliable local data made it much easier for German policy makers to make decisions without pretending for a second that 'it is different here.'

There is a common misconception about herd immunity: reaching it does not make the outbreak stop immediately. It just means that a level of immunitiy is reached where the virus will eventually disappear. For example, if the infection rate drops to 0.8 at a point in time where there are still 10'000 active infections, you have technically reached herd immunity. But those 10'000 will infect 8'000 more and those 8000 will infect 6400 and so on. In the end, this will result in 30'000 additional infections despite herd immunity. So even if you go for herd immunity, it might still be wise to go into lockdown once it is reached to prevent the outbreak from doing further damage. But after that, you can safely end the lockdown.

There is more than one common misconception about herd immunity, including the fact that the term is close to meaningless unless we are talking about vaccination.

I have to agree with Sadly that the term doesn't seem to add much value and is (almost?) tautological. SARS1 disappeared globally because of containment. Likewise, diseases such as yellow fever, malaria, and the plague used to be common in what is now the first world but basic containment and public health measures have kept them in check for decades.

Kronrod, while your premise sounds reasonable, in your version of herd immunity doesn’t it only work in practice if the critical level of group immunity is met within those who are responsible for the majority of human interactions PLUS some smaller subset of those whose interactions occur much less frequently?

Interesting.

I wonder how often Tom Hanks shakes hands compared to the average American. Presumably, there is a never ending supply of people who want to tell their friends they shook Tom Hanks' hand, so the limit is whatever Hanks thinks is enough.

What if 20% of the population do 80% of the social interactions? What is herd immunity when that is true?

Cashiers, waitstaff, people at airline/rental car counters, or school teachers? Along with Uber/Lyft drivers being part of the larger public transport group.

The hardest hit neighborhoods in New York City are the ones near La Guardia. I wouldn't be surprised if airport workers were especially hard hit.

I do agree that Ro is not homogeneous and varies with the population but also with events. There are the superspreaders, celebrities and politicians come to mind ( Bolsonaro).
Clusters also occurs because of events for example, the Gangelt Carnival near Heinsberg was the biggest cluster in Germany, the Church gathering in South Korea, in Washington State a choir meeting that lasted 2 hours where no one actually shook hands or touched anyone else resulted in 1/3 if the people infected. The singing alone was forcefully releasing droplets into the air.
These can't be attached to people per se, but only to events. They may have to be modeled differently

Google "social networks epidemic model"

And check the results.

Most of the epidemiologists seem to use a variant of the one size fits all SIR deterministic model. The population is compartmentalized into 3 buckets Susceptible-> Infectious ->Removed. There is an estimate of the fraction of infected individuals per unit time and the fraction of recovered individuals per unit time and the differential equations are solved and we’re off and running.
These deterministic models do not account for the messiness of the actual world, people quarantine themselves or are forced to, or take precautions, Ro is not a constant in time or in place, the population is not homogeneous, super spreaders and clusters exist. at what size and time delay clusters are detected is important. What is the minimum number of infections at one time and place for the epidemic to grow, How much testing is required to make informed decisions about the course of the epidemic.
Nothing in the epidemic is static, travel can be free or banned, different countries use different strategies and change strategies from one week to the next, a shortage of beds is temporary, treatments improve, people start to wash their hands more, contact tracing may come into play etc..
From the simplistic models and the data being incomplete or confusing at the beginning, we end up with widely divergent predictions; 2 M deaths or is it 100K ?
A good model should be able to answer questions like this:
How many people are infected in the US today ? surprising hard to know without actually measuring it
We close all gatherings of > 100 people, what’s the new infection rate ?. Is it effective/ineffective
60% of the population wear masks , what’s the new course of the epidemic?
In the end they give a general idea on how the curve moves but no precise predictions. People just look at the actual data in real time more than the model and say Ah, yes , the new cases are peaking, we passed the worst.
More complicated stochastic models, where every individual has some probability to act/travel and meet people in a certain way, coupled with real time observation of demographic/health/movement proxies ( thermometer data, internet searches, GPS localization, medicine purchases etc.. and the incorporation of public health scenarios is probably the future. The current models are too simple.

AI has a role to play in spotting patterns that escape humans, perhaps by learning from training sets of previous epidemics

Yeah, right. No. When data quality is low simple models are more accurate than more complex ones.

I distinctively remember reading reports from models that took pretty much what you talk about into account. It estimated how much people are going to be compliant and was also pretty open about it being an estimate.

What I find odd are people who don't read papers and don't read reports, but make wrong assumptions about mistakes or omissions in those. Not that actual papers and reports have no faults, but you can't find these when you dont bother to read them.

The Neil Ferguson model that was so influential was still the basic SEIR model.

Great post!

>g. How well do they understand how to model uncertainty of forecasts, relative to say what a top econometrician would know?

Given that we were just presented a model that said "best case scenario" and then 2 days later we are supposedly on track for a better case today . . .

But while I am skeptical of the apocopolyptic severity of CV19, I don't see how we can stay under 100k USA deaths unless the Hydroxicloriquine / zpack treatment really is working that good in the field.

I think the 50k number is purely political and was constructed as trap so no matter which decision Trump makes (but especially if we hit 100k+), he can be blamed for the extra deaths.

Even with a great treatment, since every person who dies WITH CV19 is recorded as a CV19 death in America, we will surpass 100k in official count.

Every person who dies WITH CV19 is recorded as a CV19 death in Germany and Italy also.

Further, both in France and the UK, there were (and to a certain extent, remain) systematic problems recording any covid19 deaths outside of a hospital (in France) or an institution (UK). The undercount is unknown in the NYC region, but that there is an undercount is acknowledged by numerous figures, in part because someone who dies at home is not tested, and they are buried/cremated as quickly as possible.

I've seen no evidence of undercounting in the USA. I read those same "could be undercounting" posts, but they seem to be written by those who would like the official number to be higher for political purposes.

Today in New York, an 88 year old obese man with stage 4 lung cancer is going to pass away and test positive for cv19.

And the counter will tick.

Would this be empirical enough for you, in the light of the fact that covid19 not only involves the lungs, but the heart as well?

"he FDNY reported a nearly 400 percent increase in "cardiac arrest" home deaths in late March and early April, a spike that officials say is almost certainly driven by COVID-19, whether they were formally diagnosed or not.

Between March 20 and April 5, the department recorded nearly 2,200 such deaths, versus 450 in the same period last year, according to data it provided on Friday.

The numbers are dramatically higher across the board -- the numbers of calls, the number of deaths and particularly the percentage of such calls that end in death.

In just the first five days of April, more than 70 percent of cardiac arrest calls ended in a pronunciation of death every day. Some days, the numbers were up tenfold versus a year earlier -- and everyone acknowledges there's only one likely cause." nbcnewyork.com/news/local/massive-spike-in-nyc-cardiac-arrest-deaths-seen-as-sign-of-covid-19-undercounting/2368678/

The Big Apple future looks grim, as does the rest of the country if the path of the virus continues.

Everybody dies, so one assumes that the FDNY can take a few months off dealing with any heart attacks at all after this continuing spike.

Maybe we could cut their budget during that slack period too.

>"spike that officials say is almost certainly driven the insane levels of fear-mongering and hyper-overreaction to COVID-19"

I fixed it for you. It's Easter, so no charge. You're welcome!

That's not how to read that data. You see, heart attack patients in the hospitals are down 66%. Why? You get symptoms, you're like "maybe it's not - why risk getting covid?"

By the time they call, they're toast. Hence the FDNY numbers vs hospital cases.

Also - you think this onslaught of "the end is neigh!" trash on TV that boomers watch nonstop might be stressing them into into a heart attack?

'herd immunity is already reached when 10% of the population had the virus.'

Factually, that statement is demonstrably incorrect, based on the preliminary result of antibody testing in Heinsberg, the most heavily infected region in Germany. 15% seem to have been infected, but there is zero evidence that the disease has stopped spreading on its own. In the last week, there were almost 140 new infections, or a bit under 10% of the region's total cases. Two graphics in this list just might be helpful, along with this little piece of information - Heinsberg went into its first lockdown on Feb. 28, which was then overtaken by the federal state shutdown on March 16. de.wikipedia.org/wiki/COVID-19-Pandemie_im_Kreis_Heinsberg

I don't see how the math works in this model"

"Also, it means herd immunity is reached much earlier: not after 70% of the population is immune, but after people who are involved in 70% of all human interactions are immune. ... I did a small simulation on this and came to the conclusion that with recursively applied Pareto-distribution where 1/3 of all people are responsible for 2/3 of all human interaction, herd immunity is already reached when 10% of the population had the virus."

If 1/3rd of the population had 2/3rds of the interactions, wouldn't that mean that after the upper 1/3rd is infected, we are only at 66% of interactions, so still just below the hypothesized 70% needed for herd immunity?

I could see doing this calculation with, say, 20% of the population has 80% of the interactions (so herd immunity would be at around 17%), but I don't understand the math with 33% and 67%.

We make up a model, assume it is true, then use it for decisionmaking while saying other models must be incorrect? Why yes, epidemiology and economics really aren't that different, are they?

The point of my comment is that I don't understand the math in this made-up model.

I think you guys are counting those antibody test results too early.

It can take 8 weeks after an infection for those antibodies to show up on a test.

In one sense, you are of course right - the data is extremely preliminary and extrapolated from a small number, compared to the much better results expected in the next few weeks involving 100,000 people. Including ongoing household survey of how people live in an attempt to better understand how the disease is actually spread, without making models based on pure speculation such as Kronrod or Sailer.

Don't make the mistake thinking that epidemiology is primarily concerned with making models. It is actually primarily concerned with collecting facts in an attempt to understand what is happening in a population.

I don't think this undermines the logic of Kronrod's post. Kronrod's 10%-infected-for-herd-immunity figure is (I think) derived from an R0 of about 4, the same R0 that standard epidemiological models use to get to 70%-infected-for-herd-immunity.

Heinsberg's serological data implies a significantly higher R0, right? At least within the population of Heinsberg itself. So Kronrod's number would need to be adjusted accordingly. But the underlying idea, that standard epidemiological models are seriously overestimating time to herd immunity, still seems plausible.

And note, of course, that Heinsberg's data implies a lower mortality and morbidity rate, to go along with the elevated R0.

This was supposed to be a response to 'I am very interested, can I sign up for your newsletter?'.

'So Kronrod's number would need to be adjusted accordingly.'

Or simply ignored as the method used to derive it seems contradicted by actual data based on actual testing in an actual region. That is one of the differences between epidemiology and economics - economists are used to making assumptions in their models that have no requirement to be supported by facts.

The idea that a large majority of social interaction is done by a much smaller minority would seem to completely ignore an entire significant population - those represented in a setting that ranges from daycare to high school, where they mix with many others, then return home. (University students are considerably more complex, as spring break will undoubtedly demonstrate in similar fashion to Mardi Gras).

Or do you honestly think that 10% of everyone from 2 to 18 would provide herd immunity? Or that only 20% percent of 6 year olds have 70% or 80% of all social interactions, such as walking in the halls, sitting next to each other at their desks, in a cafeteria, or on a school bus?

I am interested in epidemiological models that make sense. Three weeks ago I was hearing "naive" (no immunity, early spread) R0 = 4 and herd immunity requires 40%-70% of the population infected, as the best guesses of epidemiologists. Those numbers look completely wrong to me, in part for the reason Kronrod laid out.

I don't care that much about the 10% number because I never expected a random MR commenter's model to accurately predict something inherently unpredictable.

The fact that you're spending so much energy arguing that Kronrod is *wrong*, versus standard epidemiological models being *right*, is a sign that we may be on the same side here. I think the standard models are probably borked. Do you agree or not?

Or maybe you think modeling details just don't matter: herd immunity clearly requires a very wide spread, end of story. But I'm interested in herd immunity under unusual circumstances, especially social-distancing-without-complete-lockdown. There's no good data on that, so I'd like to have some functional models to throw at it, if you don't mind.

'I am interested in epidemiological models that make sense.'

I, on the other hand, am much more interested in epidemiological data, and care much more about its reliability than whether it makes sense compared to some model. Data is how to judge a model.

'The fact that you're spending so much energy arguing that Kronrod is *wrong*'

Well, the data does not care about what he, you or I think. Which is what makes epidemiology a science, as it first attempts to collect data as the basis for proving anything.

As for what I believe about herd immunity, it can be reduced to saying the term is essentially valueless unless we are talking about vaccination, and anybody using herd immunity without reference to (currently non-existent) covid19 vaccines is probably more interested in what they wish than in whatever data is available.

Good luck finding data of sufficient quantity and quality to make sense of this pandemic before it’s too late.

+1 to Donskoy.

In about two weeks, the full results of a German study involving antibody testing of 100,000 people in a region are expected, and this group will continue to be tracked into the future.

(Very) preliminary results already exist, and as an expanded quantity and quality of those results are available, they will be used to guide German policy making. Germans seem to be big fans of actual epidemiological data being the basis of decision making, without seemingly wasting too much time on models.

Especially if those models are not based on fundamental epidemiological principles involving public health measures during the spread of a new contagious disease. It is not a matter of luck to have such data, it is a matter of actual effort using available resources.

Chicago has started antibody testing. However, it is estimated to have a 30% false-positive rate for antibodies

I'd emphasize that the newness of this phenomenon has made it particularly hard to forecast. Tyler calls it the heterogeneousness of the data: the patterns don't play out the way one would expect.

For example, I made one good pattern recognition call about a month ago: skiers are a big vector. But otherwise I've time and time again come up with a preliminary theory about what a big pattern will be that doesn't play out as consistently as I'd expected.

This. You can't trust anyone's model of what will happen until we can find some models that explain what has happened.

I'm not aware of anyone who can explain why Europe has been hammered while Southeast Asia has largely been spared —Ireland has more cases diagnosed than India— or why Latin America is doing so much better than the US or even why numbers differ in various European countries. Russia and Sweden are following the very course that was disastrous in the UK and doing much better. Why?

People need to be able to answer all of that before you can have any confidence in their forecasts about how various policy choices and personal behaviors will affect future case loads.

Some data here: https://blog.plan99.net/is-epidemiology-useful-a4ec54e59569

Most people understand that there is a difference between social science and hard sciences like physics and chemistry. It's easy, but mistaken, to attribute the difference to the topic studied: social sciences study human behavior while hard sciences study nature. The most salient difference, however, actually relates to methodologies. Hard sciences gather data through highly controlled laboratory experiments, which are impossible for social sciences.

That distinction seems to be important when considering epidemiology and climate science. Both study natural phenomena, so many people seem to group these sciences with hard sciences as examples of natural science. In reality, however, the methodologies of epidemiology and climate science seem to share much more in common with economics: reliance on empirical (non-laboratory) data mixed with some theoretical foundations. In turn, economics seems to differ from other social sciences, which seem to have much shakier and less coherent theoretical foundations.

Instead of the dichotomy between natural and social sciences, we seem to need a third category of science for climate science, epidemiology, and economics (and perhaps others) that distinguish them from both hard sciences like physics and chemistry and social sciences like sociology.

They are statistical formalisms. The models they build, the data they collect, the domains they study are all statistical in nature. What these fields lack is causal explanations so they use much weaker approximations like correlation or regression instead. This means they cannot make strong statements and the conclusions they reach must always follow a list of explicit caveats.

"Hard sciences gather data through highly controlled laboratory experiments, which are impossible for social sciences."

So are geology, astronomy, astrophysics, meteorology, ecology, paleontology, and evolutionary biology not hard sciences? They can't do experiments either.

The scientific method necessarily includes rationality, empiricism, and falsifiability. Those fields you mention do have experimentation in varying amounts but where they do not they have only untested hypotheses.

Ask those same questions about economists; based on relative performance over the last 20 years, I favor the doctors.

This reads like someone showing off their knowledge of music theory to explain why, in their opinion, a famous and well respected painting is rubbish, when all they actually mean to say is 'I do not like this painting'. Could go on to become comedy classic.

What does this economist think of Bill Gates? He is the most visible person pushing for the complete and total shutdown of the global economy. His opinion matters more than nerds debating on Twitter over R0 and other jargon. Or do we give him a free pass because of his philanthropy on fighting the pandemic?

He is also the person most insulated from the consequences of the complete and total shutdown of the global economy.

That might be a misconception in some ways. While familia Gates might have a well-protected home with a large and competent staff, an extensive garden, its own electrical generation, water and sewer system, livestock, etc., basically a medieval manor, most of his wealth is of necessity an abstraction, records of shares in various businesses and enterprises and enpixelated funds stored away on servers somewhere. His true liquidity and access to trade is more limited than one might think. A rancher in southeastern Colorado or a gulf coast Cajun fisherman might be better insulated from economic catastrophe than any industrial titan.

You're joking, right? Even if 99.9% of his wealth was virtual, Bill Gates could still buy a hundred ranches and fishing boats tomorrow with spare change he found behind his sofa cushion, so even by your reasoning he'd come out ahead.

Not sure how ranchers and fishermen are doing these days anyway with all the restaurants closed.

the consequences of the complete and total shutdown of the global economy. wouldn't be good for Gates, John Legere, David Zalaz, Warren Buffet or any other of the Yankee oligarchs. I've probably got more filthy lucre in my pants pocket than any of them do and those engravings of Andrew Jackson won't buy a cup of coffee when the poop hits the fan, especially after a $2 trillion money drop from the sky. Most of the money in the US won't be in circulation, it will be hidden inside a plastic card and no one knows if they'll actually be able to access it at some point or if anybody will accept the card for payment. It'll be hard on the lowliest in the hierarchy but their chances are better than the oligarchs when the going gets really tough.

This crisis is probably going to make us all think what we really mean by “wealth” and “value” and cost”. It might even reveal that these are purely political constructs.

Do not worry about economists, financial constraints will soon put them back in charge.

It's cute that you believe in financial constraint.

"In the short run you socially distance, but in the long run you learn which methods of social distance protect you the most."

If you're talking about individuals, this is really overestimating people's ability to differentiate risks. For me, in the short run, I socially distance, but in the long run I just stop caring and do what the hell I like.

On "Zombies", I'm pretty sure you are not giving (Australian) economist John Quiggen credit here. Krugman gets the term from him (as closer reading, and less concern about status, would have made obvious.). Krugman, at least, managed to credit Quiggen.

I don't think epidemiology is unique.

It is a field that could fascinate smart people and draw them in. But such fields tend to also draw people .. beyond their means.

There are fields that reward you for being good at certain sorts of math, and then after you pass that hurdle, want you to be sensible.

lol, did anybody filter for that?

I doubt the epistemological problems here that the field is too imaginative, too intellectually ambitious and too intellectually independent.

(Though if you are the typical anonymous, and given human tendencies towards self serving bias, I can see why you would construct a view of these qualities as a problem.)

lol, note that I don't fashion my opinions into tawdry personal vendettas.

I'm about the big picture.

Your only opinion that you’ve been sharing for 5 years is literally a tawdry vendetta

Actually this is kind of an interesting case in point. Why do you follow me around, and I don't follow you around?

It is because I don't think you're important.

And I concentrate on the governance of the United States of America.

One would wonder (if one phrased things in the manner of prior_approval) if someone is really being "followed around" by someone when they write hundreds of blog comments a week at the same blog as a second person, and that second person responds to a perhaps few percent of them.

I would add, the thing with you that annoys people is that your thinking seems to run that if Trump were simply turfed out on his ear, and people were simply more boosterish and trusting in pre-Trump neo-con and mainstream con elites, all would be well again.

You have nothing substantive to say on the US's entanglement in neocon wars, on the China threat, on the "coming apart" and deaths of despair of post-industrial American regions, on the chilling SJWisation and the socialist shift of academia and the concomitant restrictions on speech (anthropologists who have to sign commitments to diversity or who are blacklisted, etc), on reducing illegal immigration, on the US being subject to free-riding in international institutions (most recently the China-centric WHO response mostly bankrolled by American money). Etcetc.

This is why people mock you as "Orange man bad". Because you don't seem to ever have anything substantive to say on most of the topics and policy issues that brought Trump to power. You are most often seeming to be simply indulging in digging your heels and wanting things to go back to what they were, and dressing it up as concern with the "quality of US government", without acknowledging all the above massive failures and proposing any alternative policy programme to set them right. It's the thing of people who simply think that if they "resist", they don't have to modify or compromise in their vision of politics at all, or actually resolve any failures of the decades leading up to 2016, and life will simply go back to how it was, if they "resist" hard enough.

GRE scores? Might as well ask them for their GPA and letters of rec. Personal essays if you have the time.

What about their Mensa score?

One might describe Cowen's musings as the secular explanation of (the suffering from) the contagion, just as one might describe Douthat's musings as the theological explanation of (the suffering from) the contagion. https://www.nytimes.com/2020/04/11/opinion/sunday/coronavirus-religion.html Which explanation works for you?

Douthat:

"This obligation to discernment applies in a pandemic as much as in any lesser circumstance. As the Dominican theologian Father Thomas Joseph White wrote this week for First Things — in a message directed specifically to Christians inclined to rebel against the quarantines that have closed so many churches — there is a religious duty to interpret the present moment, not just seek to endure it or escape: “What does it mean that God has permitted (or willed) temporary conditions in which our elite lifestyle of international travel is grounded, our consumption is cut to a minimum, our days are occupied with basic responsibilities toward our families and immediate communities, our resources and economic hopes are reduced, and we are made more dependent upon one another? What does it mean that our nation-states suddenly seem less potent and our armies are infected by an invisible contagion they cannot eradicate, and that the most technologically advanced countries face the humility of their limits? … We might think none of this tells us anything about ourselves, or about God’s compassion and justice. But if we simply seek to pass through all this in hasty expectation of a return to normal, perhaps we are missing the fundamental point of the exercise.”'

The contagion has provided all of us a clear view of the selfish and the selfless, the hoarders who would deprive others of masks and toilet paper, the emergency workers, the nurses, and the physicians who would risk their own health for the health of the suffering; the voracious appetite for more, more wealth, more assets, by the economically privileged, the giving spirit of philanthropists who choose to live with less in order for the poor and destitute to have more. Christians do not interpret suffering as punishment for their sins or the sins of others but as a window to see what's important in this life and the life that is promised for those who are willing to accept it.

If the percentage of those under 40 years of age that die of all causes, is higher than the percentage rate of that same age group that die from coronavirus/COVID19, how can it make any sense to subject them to a quarantine and its economic consequences?

Maybe because they can spread infection even if the disease is less likely to kill them. That makes them each more likely to get sick driving up their chance of death, and it makes them more likely to infect someone with a higher likelihood of dying. It's about not doing jumping jacks in a life raft.

Secondary effects of the virus. Such as heart damage, damage to male reproductive organs, etc. Add on costs to society for the life of those affected. The cost of doing nothing is far more than the cost of doing something - usually. Unless incompetent individuals are making the decisions, in which case the response can be worse. Economists have no way to model the decision making of individuals to account for that. Gross estimates through statistics is what substitutes for the ability to model each individual's decision making on a minute basis.

Economists lack the facility to actually predict anything and when they do get something right, it is by accident or through the application of personal experience (hint: most people do this). Mathematically, economists require perfect knowledge of all future events for them to be a viable science. Since such is impossible, economists are more like Miss Cleo - some are better con artists than others. There is not a "law" in "the science of economics" that can not be routinely violated or falsified in the real world. Thus, economics is not a science.

And note that right now some of the very worst countries (Mexico, Brazil, possibly India?) are not far enough along on the data side to yield useful inputs into the models.

Define 'worst'. None of these countries are suffering nearly as severely as western Europe and it's quite possible all three will reach the apex of their death tolls within the next week. Your previous shtick was pretending Japan was a shambles when we all know it is not. Why do you do this?

There are many epidemiological studies in Australia but they have primarily, if not solely, been about agricultural pest and disease incursions. They have involved economists as well as the appropriate scientists and in most cases the industries likely to be affected. Most of the studies have involved known problems elsewhere in the world but the lessons have been learned for unknown new pests and also for the deliberate introduction of diseases that impact on pest animals. I think that in the current coronavirus incursion, the human health agencies should have spoken to the agriculture agencies. I am attaching a link to the summary of an exercise on a potential foot-and-mouth disease incursion in Australia.
https://www.agriculture.gov.au/sites/default/files/sitecollectiondocuments/biosecurity/emergency/exercise-odysseus-report.pdf

Excellent post. How many models have you examined? Who pays for model development? What, if any, outside parties help fund these models (Gates Foundation, Big Pharma,...?}. How different are the models structured, as opposed
to just adopting different assumptions in an Excel spreadsheet format? How proprietary are the models? How much peer review are they subject to?

The virus is a mutation we had no knowledge of prior to its mutation. The learning curve concerning it is still rising. The response can it afford to wait for perfect knowledge or almost perfect knowledge. The key question is what is the plan for reopening the economy and how will it be implemented. To date there is no plan and no benchmarks. The state of economics was exposed in 2008. A lot of inductive reasoning in the discipline giving results at variance with real data. Last thing I want is some economist making decisions of how to respond to the pandemic. Advice and input yes. Decision triggering authority no. Between the two disciplines there is far more science in the medicine than in economics.

Stewart Brand wrote a good book about buildings, 'How Buildings Learn'. The worst thing one could say about a building was that it won an architectural award. Architects are about impressing other architects. He found that the people who got the right answers and built the buildings that suited their organizations were the space planners. Those are the folks who have to figure out square footage, circulation and facilities.

It's like this with money too. Economists try to impress other economists. The people we should be talking to are the accountants. Unlike economists, they understand family and business budgets and financial constraints. Their problem, aside from not winning awards, is that they aren't politically connected.

I am sure that
The question could be posed
The other way:
What
Do epidemioligists
Think of
Economists.

Would you rather have
Dr. Faucci or
Dr. Larry Kudlow

I think the difference is that epidemiologists understand networks and network analytics better than economists who were trained as far as calculus.

Economists who focus on networks clearly understand and support epidemiologists and use the same terminology. See, econ Prof. Michael Jackson's at Stanford and Nicolas Economides at NYU.

Some good questions here, but biggest problem with the post is that it often confused epidemiologists with policy makers. They inform policies, but don't specialize in things like public risk perception around masking or liquidity needs of small business.

Competent ppl need to be in place who can synthesize information from various specialties. Unfortunately, competent people are not in charge.

I'm doing an interdisciplinary PhD in a school of public health (in a program that allows me to take courses in different schools of a large university). About 1/3 of my training has been from economists, 1/3 from statisticians, and 1/3 from epidemiologists. My observations would be: (1) The pay is mostly "soft money" (academic appointments where you need to cover most or all of your salary through grants); (2) I think there is a lot more variation in the type of nature of the training that PhD-level epidemiologists have than PhD-level economists. This is true in their general training (e.g. some institutions are much stronger in causal inference than others) and in their sub-field training (with some very specialized sub-fields, like infectious disease epi); (3) I think there is more variation in how quantitatively strong people in PhD epi programs are than people in PhD econ programs. I think the math requirements for admissions to epi programs are less demanding. That being said, certain subfields (infectious disease epi being one of them) are very quantitative and bring in people who are probably stronger quantitatively than most PhD econ students.

Economics? Epidemiology? How about Sleep Research? Real expertise seems to be in short supply but slick presentation of data seems to be in large supply. Guzey's takedown of Walker's sleep studies rages on.

https://statmodeling.stat.columbia.edu/2020/03/24/why-we-sleep-a-tale-of-institutional-failure/

Carl Bergstrom would make an excellent guest on CWT.

I second this recommendation.

My two cents
as a former scientist
and current predictive modeler
The predictions that are generated by any model are only as good as the data used to build and update the model.
If we want to predict deaths from infections we need good data on the number of people who die. We have that.
And, the number of people who are or have been infected. We have never had that and still don't.
In part because we devoted our limited testing capacity to people who were already sick. Which is understandable in terms of how people usually work. But, may have been a mistake.
If you really want to understand how deadly the virus is.
We could have assumed that anyone who had the symptoms had the virus. After all, all we have been doing is treating the symptoms.
Some sort of random testing would have yielded better data.
And, we would have learned, I think, that this virus is far less lethal than we thought it was.
I'm guessing that many epidemiologists would have recommended this approach.

Experience has shown that medical facilities can become a spreading mechanism for the virus. When all hell breaks loose, you protect your most valuable asset, which is the provisioning of medical care. The testing regime is for that purpose.

You are right of course, but it assumes access to a resource that is unavailable. And ultimately you are saying that if we had more data we could make better decisions. That is obvious.

Everything about this pandemic will be obvious in retrospect.

Of what use are models at all in this situation? They will be useful in about 18 months to look back and try to understand things.

By definition you have a situation with multiple parameters and exponential growth. It is a profoundly unstable model; a slight change in any of the parameters has huge changes in the end result. By definition it is impossible to predict; it is impossible to know the inputs in a timely way. Anyone using a model to predict anything in this situation is a fool.

The only rational response is
1. not put yourself in this situation,
2. to look at your resources, do some simple math, ie. we can handle this many ICU cases, get more of everything quickly,
3. implement infection protocols in hospitals and nursing homes,
4. start praying.

If you are economist the answer in such a situation is
5. have Federal Reserve conjure $15 trillion dollars. I don't know what the equivalent is here.

Knowing what we now know would have meant on January 20th all travel into the country and between states would have been stopped. Infection protocols implemented in every health facility. Emergency declaration to override normal regulatory constraints so that testing and supply distribution could be done in a timely way. Social distancing and limiting of crowds implemented. Maybe in an alternate universe.

So really the difference between economists and epidemiologists is economists have been almost 100% behind opening trade with the world. Epidemiologists would wonder if it was wise to expose ourselves to the pathogens originating in third world food markets.

That is #1 on my list by the way. Economists were and are profoundly wrong and incapable of understanding the short term effects of their crusades. But their math is better.

Imagine on January 20th doing a lockdown like we are experiencing now and it turned out to be like SARS.

There is no way of knowing beforehand. It is impossible. The only rational thing is to not be exposed.

Something very remarkable. 1957 early in the year there was an avian flu virus (another chinese food market public health catastrophe) that spread through asia. It hit North America in the fall of 1957 and killed 68000 or so over that winter.

If we were dealing with that same time constraints, by the time it would hit here there would be vaccines and treatments and the story would be very different. But that luxury of time no longer exists because of the globalized nature of our world, so something that started late 2019 has circled the world in 5 months.

Economists would say, and have said, and have no hesitation to insult the intelligence of anyone who disagrees, that this is a wonderful thing. Epidemiologists would say what the f**k are you thinking?

The score so far this year:
Economists 0
Epidemiologists 1.

By the way, in any of the models economists have used to describe globalized trade, did any predict 10 million unemployed over a two week period?

Is there a body of work who would predict the flatfooted response of the bureaucracy? Some tidy model of number of bureaucrats * pages of regulations = length of the delayed response?

Or something simple. Calculate the odds of a single path development failing. Applies to any field.

You have only stated the cost of a global food market as 68,000 Americans dying in 1958 but nothing of the gains of having more variety and cheaper food for the past 70 years.

This feels like lame excuse-making. Mainland China really is like a disease petri dish and sophisticated neighbors like Taiwan, South Korea, Singapore and Japan have clears plans in place to mitigate the risk of pandemics. Their strategies were largely successful in buying time in the early months and, even though some of them have started to see a rise in cases, they are still doing much better than the U.S. despite their close proximity and openness to China.

As for the 1957 flu, the Wikipedia article cites sources saying the time from patient zero in China to the first cases in the U.S. may have been as short as February to June, only four months. For covid-19, it was three months. It's true there was no outbreak in June but seasonality and luck almost certainly played a role there.

Their strategy comprises diligent testing/tracing/isolation/quarantine, airport surveillance, MASKS. No need to do lockdown. Same methods could have been done in the West with limited economic damage. Complacency got us here. The U.S. government will make masks mandatory so as to be able to open the economy. What an about-face, from masks being harmful to perhaps useful to mandatory. Shameful lies.

The Lucas critique-

I'm not sure this is the purpose of epidemiology. I mean while we are at it why not consider that after this thing is over R&D on viral research will probably increase and let's map out what impact that will have on infections in 2065! Predict the entire universe while you're at it!

I think it's important to 'stay in our lanes' here. There is some room for sociology to cross over, but let's keep a limit on it. For example, it now looks like this virus has a R* of maybe 5.5-5.7 instead of only 2-3 as originally estimated. R* is a moving target in that it naturally goes down as the infection runs out of people to infect. It's also sensible to consider that different cultures will register different R*s for the same infection. A more socially distant culture may experience a virus as having R*=4 rather than R*=7 for a less distant one.

But is it really necessary for epidemiology to try to figure out how a culture will react and adjust to an infection. Maybe after this is over the US will behave more like Japan, that might change the 'default R*' for diseases.

Budding ID epidemiologist here. Where should I start if I want to give myself a good education on public choice, for the purpose of working toward resolving no. 2? Just read everything by Buchanan?

The most arrogant article/blog post regarding “economics” I have ever read. Only confirms that economics is an ideology that unfortunately have much political power in today’s world. Someday you will be combined with theology departments where you can all study anachronistic belief systems that once held great power over people.

Would you care to elaborate?

Literally in the second sentence: "because most of economics suffers from similar problems"

I agree. This is really an argument that economists are superior to epidemiologists.

Like comparing hammers and screwdrivers.

Use the tool appropriate for the task.

Agree. But the fact the Black-Scholes option formula won a Nobel economic prize shows the low level of sophistication of economic discipline. A level of 2 out of 10 in mathematics complexity.

Not to worry. The Fed has a plan as it does for all economic models.

Are there any serious epidemiologists whose work does not come with standard blanket disclaimer that they are operating with very little/contradictory base data, no variables controls, little time to crunch numbers thoughfully, in a rapidly evolving environment both terms of data and in terms of the impact of changes in behavior, and that therefore their projections - hello, they called them projections, that's your first hint - are blunt and heavily dependent on the assumptions?

I am not aware of any epidemiology forecasts that do not openly admit that their shelf life can be measured in hours. And that they are merely trying to cast some sort of light into the cave to at least try and assess the shadows.

If you want to pin blame, look to the media and politicians, who tend to regurgitate the forecasts after first stripping out the disclaimer.

"Thick, liquid academic markets"? Only a tenured twit would imagine that to be the case. They're barely liquid in the first few years after a PhD is granted and then after that, they're close to completely locked up. Senior hires are rare and so fraught with issues that they rarely occur.

As 2020 turns out to be little more than just another flu season in the USA, I'm glad we've reached the "Blame the Modelers" phase of the hyper-overreaction.

Happy Easter everybody!!

"The Straussian economists are slightly subtler"

Fauci is playing a better Straussian game than any economist I'm aware of.

I believe the authoritative take on this is Mr. Allen: https://www.youtube.com/watch?v=1yCeFmn_e2c

In epidemiology, the devil is in the "Re" and R(o) terms of the equations (the effective reproduction rate of the virus). These are not properties of the virus particles even if words like "basic" are used. They are purely social constructs where human interactions combine with the virus properties to create the apparent reproductive rate.

If we put a bubble around people (the whole purposes of PPE -- personal protective equipment), the physical distance in "social distance" is irrelevant.

Ironically, social scientists have been making radical claims for decades that things like glaciology (feminist glaciology) and even physical viewpoints like particle physics and thermodynamics are just "social constructs of dead white males. Now they can't see how, when they have a real "social construct" that will determine the future, to make changes that can save the economy and us geezers at the same time.

R0 has been taken as a type of constant since mostly we have not undertaken society wide changes in response to common infections unless they are highly problematic (example HIV).

That doesn't mean, though, that a society can't 'hack R0' by changing behaviors if it is willing.

I just read that in 10 states, 30 to 60% of coronavirus deaths are in nursing homes.

I remember how once-frequent it was for the media to write stories about "the cost of raising a child," suggesting one should give it careful consideration, features which seem less frequent now given the recent history of immigration-driven population growth.

Will the media turn, when this is over, to writing stories about how expensive it is to raise an elderly person?

My conscience pricked me so far as to propose that my parents and I drive out and meet somewhere midway between our homes. Something I read made me think if they got this virus and went into the hospital, I might never see them again. Plus my father enjoys a drive. So now I am searching Google street view to find a picnic table somewhere, our usual meeting places being parks, all closed.

Now Google the median length of stay in a nursing home til death.

Its especially relevant if is anywhere near as widespread as I

I am going to come in here and defend epidemiologists.

It is not their job to forecast the spread and # of deaths of a novel disease. It's more their job than anyone else's, but that is an impossible job to do with much precision and so it is not why we keep them around. We keep them around because while most of us know the super basic stuff like hand washing and trusting doctors to tell us what our symptoms mean, there are second level characteristics of every disease that make a big difference. And honestly I think the epidemiologists have done OK here.

The stealth: many infected show no symptoms, and one infected person can potentially spread it to lots of people. So masks are more important than they would usually be, but it is especially important that medical personnel wear masks. It is better for a nurse to have 10 masks and a software engineer to have 0 than for the nurse to have 9 and the engineer to have 1. Tests are important than they would usually be, but the same logic. Telling sick people to stay home from work would usually make a difference, but wouldn't help much here.

The tenure: people require hospital care for weeks, not in on Sunday and out (one way or another) by Saturday. We never hear about "flattening the curve" for flu, or I think for any other disease, because even though covid-19 will kill a similar number of Americans as the seasonal flu there is a danger that cohorts will keep entering the hospital until you run out of resources.

So: this is the time you lie to people about mask effectiveness, this is the time you call for working from home, and this is the time you call for social distancing. I'm sure epidemiologists have other tricks up their sleeve that they judged weren't useful here, but might come in handy another time.

The error we have made is that everyone in public health and especially MDs are trained to be "protect human life at all costs" hedgehogs. And just like we don't turn to the prosecutor at the end of a criminal trial and say "well, what do you think?" we shouldn't just say "well, the epidemiologists say to lock down the entire economy."

+1

I think this piece should be reversed. Epidemiologists rate other professions for how they did, economists esp.

What's their GRE or how smart epidemiologists are, is not relevant. Economists tend to be smart but they, typically, can't think. You can tell by how eager they are for power, instead of proving themselves by exploiting their insights through, say, trading.

In other words, the relevant question is: do epidemiologists have skin in the game? Do they suffer consequences for being wrong.

So GRE is getting beat up here. I have no side on this, but a question. Would a whole discipline's GRE be telling while any individual's not be so much?

At least epidemiologists are able to make a reasonably modell predicting the degree of awfulness of whatever epidemic is in front of them. Economists are unable to spot any recession before it jumps up and bites them in the arse.

Almost all of this chatter misses Tyler's point. This is a problem of specialization:

Epidemiologists know a lot about disease behavior but very little about human behavior.

Economists know a lot about human behavior and almost nothing about diseases.

Our current epidemic models all appeared skewed toward epidemiology (not surprisingly) instead of economics. As such they may be very good at predicting how the virus behaves in a lab dish, but by failing to account for human behavior changes, they make poor predictions about real human transmission patterns. Based on the last month, it appears those errors tend to average negative, resulting in models that tend to be too pessimistic.

Find me an MD who also has a PhD in economics and I will pay closer attention.

https://en.wikipedia.org/wiki/Economic_epidemiology Just weird to think like this man.

What a credible post. A person criticizing epidemiologists for their prediction models asks about the GRE score of epidemiologists and doesn’t realize GRE is not a good predictor of success (https://www.sciencemag.org/careers/2017/06/gres-dont-predict-grad-school-success-what-does). The court jester would be proud of this tomfoolery.
Also, he asks: What is their overall track record on predictions, whether before or during this crisis? Answer: As good as economists track record in predicting the 2008 worldwide recession.

Perhaps more interesting than the average GRE scores of epidemiologists is the percentage of whiz kids the field of epidemiology captures. For the purposes of this exercise, we can define a verbal reasoning whiz kid (quantitative reasoning whiz kid, analytical writing whiz kid) as someone who obtains a perfect score in the verbal reasoning (quantitative reasoning, analytical writing) section of the GRE. We can then grade intended graduate school majors by the share of these three types each captures. (Note that in the available data, epidemiology is rolled up into a major called "health & medical sciences".)

For verbal reasoning, the top ten fields (% of whiz kids captured) are:
1. Biological & Biomedical Sciences (10.1%)
2. Political Science (8.4%)
3. Computer & Information Sciences (8.1%)
4. Mathematical Sciences (6.6%)
5. English Language & Literature (6.4%)
6. Economics (5.8%)
7. Physics & Astronomy (4.9%)
8. Psychology (4.0%)
9. Health & Medical Sciences (3.3%)
10. History (3.3%)

For quantitative reasoning, the top ten fields (% of whiz kids captured) are:
1. Computer & Information Sciences (18.5%)
2. Engineering, Electrical & Electronics (14.9%)
3. Mathematical Sciences (13.6%)
4. Engineering, Mechanical (7.0%)
5. Physics & Astronomy (5.6%)
6. Economics (5.5%)
7. Banking & Finance (4.9%)
8. Engineering - Other (3.7%)
9. Biological & Biomedical Sciences (3.7%)
10. Engineering, Chemical (2.7%)

In this arena, health & medical sciences comes in 17th out of 52 with 1.1% capture.

For analytical writing, the top ten fields (% of whiz kids captured) are:
1. Health & Medical Sciences (11.0%)
2. Psychology (9.8%)
3. Biological & Biomedical Sciences (9.6%)
4. Political Science (8.3%)
5. English Language & Literature (5.6%)
6. Computer & Information Sciences (4.4%)
7. Economics (4.0%)
8. Business Admin & Management (3.8%)
9. History (3.4%)
10. Engineering - Other (2.9%)

It would seem that economics has a big though not enormous advantage in quantitative reasoning, while epidemiology has a big advantage in analytical writing. Economics also has a small advantage in verbal reasoning. I think this is a more useful way to look at things, because what we care about is not the average epidemiologist, but the top epidemiologists, since these are the ones who have the most influence. Share of whiz kids captured gives you an idea of how much top talent the field has, and how much competition for the top spots there is in the field. Of course, all the usual qualifiers apply in terms of the informativeness of GRE scores, and using perfect scores as our definition of whiz kid, while very specific, is probably not terribly sensitive.

Sources:
(1) https://www.ets.org/s/gre/pdf/dept_major_field_codes.pdf
(2) https://www.ets.org/s/gre/pdf/gre_guide_table4.pdf

I think it's ironic that those criticizing prediction models of epidemiologists don't realize that GRE is not a good predictor of success. Also, again remind me how economists fared in predicting the 2008 recession?

As I mentioned, GRE is only somewhat informative about intellectual abilities, but some information is usually (though not always!) better than none. And I’m not an economist, but I agree that evaluable track record of the field is the more pertinent question.

Thanks.

Have they done anything about the low ceiling on GRE quant scores yet?

Someone should also inform the court jester that many prediction models are in fact conducted by biostatisticians who have had 10 years of statistics training and know far more about statistics than economists. Ignorance is bliss for the author of this post.

When you ask

a. As a class of scientists, how much are epidemiologists paid? Is good or bad news better for their salaries?

Maybe those epidemologists have not figured out how to run a remunerative grift for the tobacco lobby as the economists at George Mason did?

https://www.scribd.com/document/257705494/Tobacco-Economists-Network

I think this piece should be written from the reverse point of view, have epidemiologists rate economists and other professions for how they did.

For example, how many ideas from economists were innovative or simply trying to reinvent the wheel or whip old hobby horses? Robin Hanson would not come out well here, I think. How about that lawyer/think tank guy who pronounced the virus would max out at 500 dead in the US without taking any special precautions?

you might be interested in david a freemans work on misapplication of statistics, ie, http://psychology.okstate.edu/faculty/jgrice/psyc5314/Freedman_1991A.pdf

Tyler, I have to say that I am very disappointed at the shallowness of this piece. First, many of the questions you ask could have been answered with a few clicks of your mouse using a decent search engine. Second, you seem to have a deep if not intentional misunderstanding of the purposes of epidemiological models--or even of the fact that there are several different types of epidemiological models. The epidemiologists do, in fact, have a pretty good grasp of this fact, and they understand the limitations of the models. This said, a very simple epidemiological model is very useful for illustrating how an epidemic progresses and how certain countermeasures can be effective. Maybe you should look into one of these simple models.

And you final questions are simply ignorant and insulting.
1) Most epidemiologists do not go into the field to make money. They go into the field to help people in need. That you do not understand this or even acknowledge it as a motivation does not speak well of you personally.
2) GRE score has a very low correlation to academic and professional success. (Actually, most studies show that one of the highest correlations is the time a student spends at the department.)
3) The rest of your questions merely show that you cannot conceive of academic inquiry being independent of nationality and politics. In my own field of physics, we had a group try to push a "German Physics". It did not go well.

In fairness I think the financial dynamics are important. It's not about whether people go into a field to 'just make money' or 'help people', reality is both are true. People go into a field often with positive motivations but they also have to make a living. What are the financial dynamics of a field given those constraints? Who hires? How important are 'gatekeepers' who can deny tenure or how about alternative paths such as private companies?

Agreed. To me it reflects that Tyler is not very familiar with network and epidemiological models.

An inch deep.

The posts would be much better if he did a little research first.

This is an arrogant post. I did not expect anything like this on this blog. Especially from someone who advocates a slow and steady approach to learning new subjects. So after two months of paying attention, this? If this is meant as a call to arms, that’s not how you built collaborative teams. Poor timing and poor approach to getting information.

Asking about the IQ of epidemiologists is certainly legitimate in a setting where their policy recommendations will have consequences that will in shape the course of most of the the next decade.

"Epidemiology" is a weird word, most epidemiologists do not study epidemics. Fortunately, infectious disease epidemiologists - those who actually focus on epidemics, like Marc Lipsitch and Neil Ferguson - tend to be very smart, certainly comparable to (or exceeding) top economists.

If infectious disease epidemiologists instead had IQs that were comparable to social epidemiologists or nutritional epidemiologists (e.g. Walter Willett), I would be legitimately scared. It is not surprising that some people are uneasy if they have interacted with this type of epidemiologist in the past.

Most of the twitter reaction is coming from social epidemiologists who are working to get rid of GRE entry requirements in order to increase representation of historically marginalized groups in their "research".

IQ is bullshit, it's Gaussian by construction for no apparent reason, g is a statistical artifact that arises from factor analysis because all tests results correlate, intelligence is not well-defined (except by IQ fundamentalists), most of the "outcomes" that correlate well with IQ come from circularity of "being good at test taking will make you advance quickly in a career path that requires a lot of test taking", most of the prominent IQ researchers have been found guilty of data manipulation or outright fraud (Lynn 'inferring' national IQ of countries by taking the average of neighboring countries when data is missing, etc.), none of the theories of intelligence behind g or IQ line up well with actual neurological data.

More reading:
https://medium.com/incerto/iq-is-largely-a-pseudoscientific-swindle-f131c101ba39
http://bactra.org/weblog/523.html

I would expect the genetic component of intelligence to be close to Gaussian. Any polygenic trait that depends on many alleles where each allele has a small positive or negative effect, the sum of all them should be approximately normal by the central limit theorem (provided each term is independent). But I don’t think the scale is all that important anyway. If you prefer, go ahead and convert raw test scores into percentiles. It will still be predictive.

>I would expect the genetic component of intelligence to be close to Gaussian

The "genetic component" doesn't make sense, this stuff is not additive. There is no known mechanism for genetic determinism behind intelligence beyond very low hanging fruit like obvious developmental defects. And no, polygenic traits don't always work that way. "Provided each term is independent" is a hell of an assumption in a world of epistaxis, differential expression, epigenetic and other such various messes.

But even that has nothing to do with the actual distribution of "intelligence" which has no reason to be Gaussian or even symmetric. Why should there be a genius for each person with hydrocephaly?

> But I don’t think the scale is all that important anyway.

It is, because if we suspect that, as is the case for many performance tasks, their ability distribution is fat tailed, it means the true covariance between IQ and performance is undefined statistically, so the sample covariance you're measuring is completely uninformative.

> It will still be predictive.

Yes, a test will be predictive of people's ability to do well in career paths that require tests (the fabled "educational attainment"). Is it any surprise? It doesn't even correlate that well for a circular metric.

Taleb is an okay guy in my book, but he completely embarrassed himself with his IQ commentary. I didn't find his article convincing when it came out, nor am I compelled by your regurgitation of it.

NOBODY says that IQ explains a lot of variance in the highest levels of performance. But in most domains, higher IQ is better, all else equal. Some domains are cognitively demanding and high intelligence is a *necessary though not sufficient* condition to perform in such areas. You say performance on tests only predicts performance on tests and has zero practical value (hence "circular") but this is simply wrong. Say you're Britain during WWII and you're trying to crack Germany's Enigma code. This is a highly practical problem and you need smart people working on it. A guy that scores average on an IQ-style test is not going to be of any use and you damn well know it.

Regarding genetics, we already know that intelligence is in large part genetic. Intellectually, adopted kids are more like their biological parents than their adopted parents. And we know for certain that intelligence is a polygenic trait. And we know that other polygenic traits like height ARE INDEED miraculously Gaussian despite "epistaxis, differential expression, epigenetic and other such various messes."

I do agree with a lot of Taleb's comments about the "Intellectual Yet Idiot" class and his critiques of the so-called elite. But this is more an issue with incentives ("skin in the game" and so forth).

> I didn't find his article convincing when it came out, nor am I compelled by your regurgitation of it.

Yet you didn't refute any of the points he adresses, most of which have nothing to do with the rest of your post.

>But in most domains, higher IQ is better, all else equal.

But 'all else' is *never* equal. This is such a contrived and artificial hypothetical, typical of the IYI trends he denounces. Real life is messy, real life aptitudes are hard to quantify and can't be easily reduced to a couple numbers no matter how much psychometricians want it to be true.

>You say performance on tests only predicts performance on tests and has zero practical value (hence "circular") but this is simply wrong.

I did not say IQ isn't predictive. What I said is its predictive value is circular insofar as the outcomes it so-called 'predicts' are heavily biased toward good test takers. In order to enter and graduate from a good university or apply to various programs you do need to pass various tests whose content is strikingly similar to IQ tests (e.g. Cowen's cherished GRE). It's no surprise that doing well in one of those means you're going to do well on similar tests, but only IQ fundamentalists would argue that answering a bunch of pattern recognition and grammar questions correctly and very quickly is a satisfying way of measuring one's intelligence.

> Say you're Britain during WWII and you're trying to crack Germany's Enigma code. This is a highly practical problem and you need smart people working on it. A guy that scores average on an IQ-style test is not going to be of any use and you damn well know it.

There you are again, making up contrived hypotheticals, without even proving them.

>Regarding genetics, we already know that intelligence is in large part genetic.

This doesn't mean anything. Seriously, 'X is partly genetic' is an utterly meaningless statement, let alone putting a percentage behind it. That's not how modern genetics work. If you want to show genetic determination driving a trait, you have to demonstrate an explicit genetic mechanism complete with pathways and prove it indeed influences your trait, usually with animal models. Otherwise you're just staring at a bunch of correlations. You don't get to hide behind 'but I did a bunch of controls', that's nowhere near the standard of rigor required by actual geneticists to prove genetic determination. As far as cognitive ability is concerned little has be found beyond very trivial examples such as developmental defects.

>And we know that other polygenic traits like height ARE INDEED miraculously Gaussian despite "epistaxis, differential expression, epigenetic and other such various messes."

Height, that famous trait that follows a gaussian distribution regardless of sex, ethnicity, amount of nutrition when growing up, and a thousand other factors. 'But I meant Gaussian within each population controlled for factors that'd make it non-Gaussian!' Oh ok then.

I vaguely remember reading something about this topic, I think by an economist. The guy was saying that the epidemiologists whiffed badly on AIDS. They predicted it to penetrate (ahem) much deeper than it actually did. The reality is that the vast vast majority of infections were gay men along with junkies and some from tainted blood supply. It never spread much among the general population. As I recall the economist’s take was that this was because people mitigated the risk much more effectively than expected (“safe sex” etc) and this threw off the models. But I don’t think that’s quite right. More like they simply concocted a myth of “heterosexual AIDS” for political reasons.

Coronavirus is not as hot button as AIDS, but I have noticed they won’t give much data on ethnicity (presumably so as not to “stigmatize”).

Heterosexual AIDS is very much a real thing in Africa. 'Concocted' means you think they:

1. Had good models that predicted how things would pan out.
2. Hid the results in order to make a political case for more AIDS funding.

Since the models are published all you have to do is take them, plug the numbers in and show that they never showed heterosexual AIDS would become a thing.

I suspect the answer is no the models did not account well for how both gays and straights would change their behavior in light of AIDS. Such models require making more and more second and third order predictions about how society will change.

I think they were lying about it because they did not want people to think of it as GRIDS or a “gay” disease. “We’re all at risk.” And it would be more likely to get funding. Of course, the reality was that the earliest CDC reports indicated that among the gay men in NYC, SF and other hot spots, that those testing positive had average partner counts of around 1,000. (I mean, think about it, it’s all men.). The disease was never going to spread the same way among the general population because heterosexuals are not anywhere near that promiscuous. Additionally, “unprotected” vaginal sex has quite low transmission risk relative to activities favored by gay men.

Now to be fair, maybe serious epidemiologists were more sober than the media and government officials. But if so they were not very effective in communicating their reservations or correcting the misinformation.

Africa's problem seems to be general promiscuity. Possibly Western promiscuity was overestimated.

https://en.wikipedia.org/wiki/HIV/AIDS_in_Africa#Behavioral_factors

HIV transmission is most likely in the first few weeks after infection, and is therefore increased when people have more than one sexual partner in the same time period. In most of the developed world outside Africa, this means HIV transmission is high among prostitutes and other people who may have more than one sexual partner concurrently. Within the cultures of sub-Saharan Africa, it is relatively common for both men and women to be carrying on sexual relations with more than one person, which promotes HIV transmission.[22] This practice is known as concurrency, which Helen Epstein describes in her book, The Invisible Cure: Africa, the West, and the Fight against AIDS, in which her research into the sexual mores of Uganda revealed the high frequency with which men and women engage in concurrent sexual relationships.[39]

The CDC actually publishes estimates of HIV risk for different sexual behaviors. Based on their numbers, it would be very hard to spread HIV through heterosexual sex and in general it is quite hard to spread from woman to man. Even if you had sex with thousands of prostitutes without a condom, you would probably not get HIV (although surely you’d get lots of other types of VD). The risk is surprisingly low, only 4 per 10,000 exposures.

https://www.cdc.gov/hiv/risk/estimates/riskbehaviors.html

Some of the numbers coming out of Africa are so fantastic (20% HIV rate in some countries) that I find myself wondering if the stats are accurate.

For those numbers to be right, it seems like they would have to having a lot of anal sex and there would have to be a lot of men “on the down low.” Or they would have to be way more susceptible to it biologically for some reason.

gregor, you assert they lied yet you decline to support that. All you have to do is take the models they were using, plug in their assumptions and show the numbers that come out never indicated any real risk to heterosexuals.

"The CDC actually publishes estimates of HIV risk for different sexual behaviors...."

You mean you're publishing data from today. What was known about relative risks of different types of sexual behavior in 1984? In supporting your case, I should not be seeing CDC links unless you're linking to old archival documents.

This raises a question, how well did or even does the CDC know the different sexual behavior the population engages in? Like right now how many 19 yr olds engage in unprotected sex at least once a month? I think any answer we have will be a guess but suppose a newly sexually transmitted disease was discovered to be spreading among a few Americans tomorrow. It seems pretty heroic to assert the average American "shouldn't worry", unless you had really excellent evidence and models you'd bet people's lives by.

"that those testing positive had average partner counts of around 1,000"

True but make an assumption a new STD will penetrate eventually deep into the population of all sexually active Americans. You would still expect it to show up first among people who have 1000 partners versus people who have 1. The fact that in the early stages you notice it first in the people with 1000 does not justify you assuming it isn't a problem for everyone eventually. It just justifies the assumption that you may have a limited window of time to stop the infection now or else you'll never be able to control it.

No. It was obvious at least as early as 1982 that you generally needed A LOT of "exposures" to get HIV. And just in terms of basic physiology you'd expect high volume anal sex to be much higher risk.

"Epidemiologists from the Centers for Disease Control have done studies among homosexual men with and without the immune disorder but matched in age, background and other characteristics. After testing for more than 130 potential risk factors, they found that the median number of lifetime male sexual partners for affected homosexual men was 1,160, compared to 524 for male homosexual men who did not have the syndrome. The study also found more use of sexual stimulants and illicit drugs among the GRID patients."

524 for the guys that were negative!

My guess is that it was so hard to get coronavirus data by race until the last week because the public health establishment figured that the next epidemic would be concentrated among Marginalized Minorities, so they set up a policy of occluding race/ethnicity data. But then it turned out that it was, early on, concentrated among skiers, movie stars, political leaders, and so forth.

Yawn, your hobby horses are so boring. Epidemiological data moves very slowly. It takes money and effort to update data quickly. We have rapid, accurate updates on the weather and stock prices. That is because people pay good money to have this information generated quickly and accurately.

Demographic and epidemiological data like deaths move very slowly because it's rare that anyone finds up to the minute data valuable.

We know about health shocks on celebrities, political leaders and sports stars for the same reason we know when the price of Tesla stock moves dramatically but not so much the price of vintage 1984 Star Wars toys. Because we pay to generate this information and reward those who can produce it quickly and accurately.

I think we all know what is going to happen once the epidemic dies down. The much maligned government will bail out the big banks, the big corporations and the billionaires and then, for the next few decades, the economists will marvel at the slow rate of recovery and blame everyone laid off or forced into bankruptcy for their own financial incompetence and inability to pull themselves up by their bootstraps.

No PhD. But Surgeon with MBA who worked for McKinsey for 3 yrs before Med school. “Easier to tell someone they are going to die than fire a 50yr immigrant female from her union job knowing her family can never replace that income”

2 questions for the court jester
A) every fired a bunch of mid life union employees who never finished high school?
B) ever held someone’s hand while they die?
I’ve done both. Multiple times. Econ “market” models don’t have a great reputation spotting common outlier events. Try telling those women they can “move” or “retrain”.

Echoing others in here
1) the biostatisticians advise epi people. Why is that important? Almost every statistical Econ trick was lifted from Biostats. The very concept of data aggregation and distribution is from Biostats.
2) Ask Gates who he has more time for - oh it’s public health and epi people and educating females (I forgot. - you don’t believe in formal education).

Biggest gains in life years added and quality of life on the earth cane from - epi studies in public health leading to clean water, clean sewage and tidying up some vaccines.

As they say. “Those that can do, those that can’t teach - and throw out ridiculous questions”

Several of your claims are not supported by data.

For instance, it's possible that wearing a mask might encourage people to go out more. But it's also possible that it might people to go out less.

I wear a mask when I have to go out now and I feel silly. It's uncomfortable, hot, and smells funny. I want to go out out even less.

We need data to understand these behaviors, not your conjectures.

I know you are in a weird niche in econ, in a very unconventional department. But there are lots of economists working with epidemiologists, and whole field of health economists that often doesn't look that different than epidemiology. There are journals that cross both areas.

I guess I'm saying maybe you should do a few minutes of research before posting stuff like this on a well-read blog.

https://twitter.com/sanjum/status/1249374580831064072?s=21

Hah. pretending economics is a science. it is an art. Meanwhile, Americans die.

In Minnesota, our U of Minn epidemiological models are saying the peak will be sometime between late May and July and that we will need 3-5,000 ICU beds (With current swing capacity, we are close to 3,000 ICU beds, but would still face equipment shortages.) The U of Wash model in contrast says Minnesota COVID cases will peak on April 26 with a need for 126 ICU beds, a staggeringly different conclusion. The Governor is relying on the Minnesota model for now; will he be laughed out of office for being ridiculously over-prepared, compared to other states?
For example, in Florida, they appear to be relying on the U of Wash. model which suggests a need in Florida for a peak of 1,241 ICU beds on April 26 (with 1,695 available).
Much as we Minnesotans are prone to the natural tendency to want to be right, let’s hope our model is wildly wrong and that the U of Wash model is the better predictor, as Florida officials are presently assuring their citizens that their State have sufficient ICU capacity for the coming surge,

In Minnesota, our U of Minn epidemiological models are saying the peak will be sometime between late May and July and that we will need 3-5,000 ICU beds (With current swing capacity, we are close to 3,000 ICU beds, but would still face equipment shortages.) The U of Wash model in contrast says Minnesota COVID cases will peak on April 26 with a need for 126 ICU beds, a staggeringly different conclusion.

Wow, those *are* staggeringly different predictions!

I don't see how the U of Minn model could be correct...it appears that daily additional deaths in the U.S. are already a couple days past their peak, and dropping rapidly. (But I'm not an epidemiologist.) (Or an economist.)

"if you are a French economist, being a Frenchman predicts your political views better than does being an economist (there is an old MR post on this somewhere)" - anyone can link to this post?

Thanks for this. As someone who entertains the fantasy of getting a PhD in Economics once I retire, this would undoubtedly be my focus. The fields are ripe for epistemological enquiry.

I would look at it in the context of the history of economic thought, starting with past pandemics and looking at any collaboration between economists and epidemiologists.

Brazil is actually doing far better than the US, at equivalent stage

There have been many diverbent paths taken by economists on these topics. I suggest a quick reading of a 1898 essay by Veblen entitled "Why is Economics not an Evolutionary Science?" An evolutionary science being a close knit body of theory...a theory of a process, of an unfolding sequence" and what he writes about "realism in data" https://www.jstor.org/stable/1882952?seq=2#metadata_info_tab_contents

This article makes excellent points, which most commenters seem to have missed. perhaps due to their own apparent biases or desire to show off that they know a few, albeit irrelevant, things.

My take-away is that economics and epidemiology are both suffering from the same malady - a refusal to accept the psychology of human nature. We do not all behave the same, and we most certainly are NOT rational. As Sister Mary Holywater taught us in 6th grade Catechism class; "That humans are rational beings simply means that we are "capable" of rational thinking; not that we are rational all the time." Both econometrics and epidemiology model what they see, cherry-pick easily accessible data, pop it into their model, and rejoice at whatever their equation spits out as knowledge until subsequently refuted by equally flawed models and studies. Yet both fields persist.

Of course, there are economists and epidemiologists who do factor human nature into their thinking, and both predict rather well. Unfortunately, they are drowned out by the din of colleagues who still do not get it.

He’s back! We missed you, Tyler. I appreciate that you take your role as a public figure seriously. That said, it’s time to muddle through this mess. I look forward to your lessons learned in 2022 - what do you think would have actually saved the most lives?

The funniest thing is that you, and economist, seem to truly believe you are more qualified than an epidemiologist to handle public health matters. "How smart are they," really? "How meritocratic" are their "institutional markets"? You've got to be kidding.

This very open and public attempted usurpation on public health matters by members of the economics field is not only absurd on its face (no, you do not know how to deal with public health crises because you know how to read a spreadsheet), the political consequences could only be dire (who benefits when 'orthodox' economists make the rules is crystal clear, and quite unacceptable).

So maybe there is a better object of your study here. Perhaps instead of trying to reassert the primacy of economic matters over those of public health, perhaps you should determine how many economists should be deported to Antarctica for the safety of our citizenry? Because that, you are clearly showing, would be a far more productive use of your time.

Hi Tyler,

Get off your high horse.

Mandy

The field of epidemiology I am most familiar with is nutrition epi. It's the data that is the problem there, as may also be true with the COVID-19 data. Around 80 percent of diet/disease relationships studies rely on 24 hour diet recall data. About 60% of the people reporting do not report eating enough food to stay alive.

I have no horse in this race other than being human, and pass no judgment other than the time spent (wasted) by those (who allegedly believe in the application of data and math) bantering without just answering the questions posed. We don't have to agree on the social value or righteousness of any individual or that individual's ideas, and such bantering adds little value to the wholistic intent of the questions: what data is available that adds credibility to the forecasts to the extent such forecasts have predictive value such that policy makers can rely upon such forecasts in their decision making processes? There are clearly a lot of people far more intelligent than me on this post, but intelligence has little value if it cannot effectuate action. I challenge you all to try to answer the initial questions collectively, and then we can vigorously debate action. Thank you to all of us trying to come to great solutions via positive collaboration.

Comments for this post are closed