[netcan]

Covid has really driven home a point to me. As a society, we are statistically illiterate. Politicians and journalists particularly so. Most of our important, newsworthy information today is statistical.

This is a real problem. In the context of public schooling, I think statistics needs to become the primary discipline taught in high school maths. It's more useful to our work life, and (relevant in the context of public schooling) essential to informed citizenship.

Literacy is a pretty close analogy here. The average person is totally ill equipped to to read politically relevant news and form an opinion about it. Often, the person who wrote it is just as ill equipped.

Statistical statements have a tricky form. They seem like a statement of fact. They are, kind of. It's a fact that this researcher measured what she measured. The implication though, that's conjecture, and it may or may not be a good one.

[tomp]

If anything, COVID drove the opposite point for me.

The whole "masks don't work" spiel that the WHO did was statistically legitimate... We really don't have proof (or whatever the medical community considers is "proof" - like double blind large scale trail with less than 5% chance of being false) that masks work. Statistically, we don't know.

But operationally masks have negligible risk and practical burden, while having a huge potential benefit (stopping the pandemic in its tracks), so even if the overall probability of this benefit is low (or at least not necessarily 95+%), it's the correct decision from an executive perspective.

Basically: scientific / statistical opinion: masks aren't proven to work; executive decision: recommending masks has minimal downside and massive potential upside;

[jkingsbery]

The latest EconTalk episode talks about this: https://www.econtalk.org/nassim-nicholas-taleb-on-the-pandem... ... Once the Pandemic hit, we should have worn masks. Statistical studies aren't the only form of human knowledge - we understand that when we speak, air and saliva leave our mouth, carrying viruses along. Masks catch air and saliva (and as you say, have low cost). Therefore, wearing masks makes sense as a matter of logic. We don't need to drag p-values into everything.

[imtringued]

Masks reduce spread of the virus by already infected people. They work because even people who don't even know they are sick are wearing them from the beginning to the end of their infection. When it comes to exponential growth it doesn't matter if the masks reduce infection rate by 10%, 1% or even 0.1%. Alternatively, even if masks don't work physically there is a psychological component to wearing masks that can reduce the infection rate by reminding people that there is a virus out there and as long as you are wearing the mask you should be extra careful.

[DanBC]

There's not much evidence to support anything in your comment.

If masks were so effective it would be easy to find that benefit in any of the good quality trials that have been run, and we don't see that benefit. To see any benefit of mask wearing we have to drop the quality of evidence right down.

One thing often missing from these discussions is the concept of "number needed to treat". How many people need to wear a mask to prevent one additional infection?

https://www.fhi.no/globalassets/dokumenterfiler/rapporter/20...

> Given the low prevalence of COVID-19 currently, even if facemasks are assumed to be effective, the difference in infection rates between using facemasks and not using facemasks would be small. Assuming that 20% of people infectious with SARS-CoV-2 do not have symptoms, and assuming a risk reduction of 40% for wearing facemask, 200000 people would need to wear facemasks to prevent one new infection per week in the current epidemiological situation.

[tacocataco]

I was under the impression that masks only work to protect you from transmitting to others. Has that aspect not been studied yet?

Or are you talking about the efficacy of masks protecting the person that wears it?

Appreciate your time.

[ryanmcbride]

The largest reduction in rate of transmission is when the infected person is wearing a mask. However if both the infected, and uninfected person are wearing masts, it drops just a little bit lower. But you're right that if an unmasked infected person comes in contact with a masked healthy person, the masked person is still very likely to get infected.

[xapata]

It's much like flossing. The mechanism is so obvious, we didn't need a statistical study to be confident. We now have statistical evidence from Covid-19 spread.

[tomp]

No idea. I’m not an expert.

My physics intuition says that air filter is almost equally effective whereever it is (your mouth or another person’s mouth). It depends on the details though - how big the virus is? How far does it travel? How much of it comes out “dry” vs in water droplets? How big are these water droplets? What’s the distance it can travel in a droplet vs “dry”? And so on and so on. My prior says that eyes aren’t a big transmission factor (as the virus has particular affinity towards lung cell receptors) and that the only way that mask efficiency is radically different depending on who’s wearing it is, if the virus is mostly exhaled in water droplets and most of them evaporate in the next 1-2 meters (so when you inhale through a mask, it doesn’t filter the tiny virus particles, whereas exhaling into a mask does filter the less tiny water droplets).

But the bigger point is, it doesn’t matter. Masks can help, and that should be enough to make a decision, like a general rule or recommendation to wear masks (or face coverings).

[marcosdumay]

The last few years have made it very clear that worldwide the medicine regulation sector sucks at cost-benefit analysis.

[Joeri]

Statistical literacy doesn’t necessarily help. Economists have proven to be “excellent” armchair virologists. They understand the statistics well enough that they think they can give an informed opinion, but because they don’t actually understand viruses and infection their opinions are dangerously uninformed.

Knowing statistics isn’t enough, you also need to know the field those statistics are applied to, because otherwise you can reach statistically sound but inherently meaningless conclusions.

[r00fus]

If I could upvote this twice I would. The basis for our childrens' education is missing statistics and finance/economics.

My kids' school does teaches basic consumer protection and critical thinking "how to spot ads/scams"

Teaching those should be the foundation on which we teach other things.

[netcan]

Thanks.

I agree on extending the point past statistics. When we talk a lot about schooling tha should prepare kids for the world, that talk is usually directed at work-ready type skills. Coding, resume writing, etc.

We really need to update our fundamentals, near the epistemological level.

[dfxm12]

Just like COVID-19, education is also a politicized platform in America. It's not a coincidence that the same politicians who are trying to downplay the global pandemic and effects of the virus are colleagues with, if not the same, politicians that are trying to reduce access to education.

If you agree with the premise of the parent poster that improper education is "a real problem", change towards a more educated citizenship begins in November.

[netcan]

Not everything is conspiracy. The US isn't unique here. Statistics is more important than it used to be. Schooling just hasn't changed.

[cagenut]

I call it "functional innumeracy". Our society is stuck at a local-maxima for communicating via headlines and short-conversations where we can only really compress and encode signed-keywords to each other. Meaning like, musk+ or musk-, bitcoin+ or bitcoin-.

If we want to take things to the next level of numerical understanding, where we "graph" all sorts of rates, distributions, curves, crossover and inflection points, etc, then what do we do? Fuck around in a spreadsheet for an hour and screen shot that and upload it to imgur and put the link in here? Comb through google images for something close enough and maybe photoshop some arrows on it?

We need a better communication toolkit than a few hundred bytes of ascii to make it possible for people to introduce a heightened degree of numeracy in everyday conversations and decision making. Image macros and emojis and infographics are pigin attempts to go down this road, but we're not there yet (that i've seen, links if you got 'em).

Imagine you're a journalist right now, you're logged into wordpress, you need to explain the insane disaster of today's GDP report, and you have a deadline of finishing your post before lunch. What do you do?

[taneq]

Statistically and scientifically illiterate, yes. There is this overall belief that one person's opinion, however gained, is equal to another person's opinion, however gained. If one opinion is the result of multiple well designed scientific studies and the other is the result of some late night YouTube indulgement then no, they're not equal.

[throwaway_pdp09]

Upvoted because it's a good point but I think the problem goes deeper. Simply expressed most people will deny or plain block out what they don't want to hear, and education can't cure that. There are plenty of people still denying that covid is of significant risk - and we're right in the middle of it!

[ryathal]

As a society we are increasingly illiterate. Mathematical or statistical knowledge isn't achievable without reading comprehension that is above the average person.

The real problem is we aren't teaching kids to read well enough to have a chance of them establishing a foundation in anything else. Statistics needs a base understanding of arithmetic and algebra (calculus would help a ton but lets be realistic in expectations), it's just even more math the average student won't understand. We need to get the basics actually taught to the point of mastery for most students before adding stretch goals.

[samizdis]

Possibly tangential, so apologies for that, but I am reminded of the "Linda problem"[1], as posed by Amos Tversky and Daniel Kahneman.

Michael Lewis's book The Undoing Project[2], which in my opinion provides an accessible and informative look into the work of Tversky and Kahneman, is worth a read.

[1] https://en.wikipedia.org/wiki/Conjunction_fallacy

[2} https://en.wikipedia.org/wiki/The_Undoing_Project