(no subject)

[ny_quant]

Добавление ко вчерашнему:



the data in Figure 2 shows a decrease in infection rates after countries eased national lockdowns with >99% statistical significance""

Comments

[ny-quant.livejournal.com]

And you believe yourself to be reasonable while at that?

[misha-b.livejournal.com]

Yes, I look at the data. I can see that most points are below the line, I don't need some arbitrary number like 99% for that.


What I don't know is the protocol for measurements. Without knowing what is being measured and how they are measuring it, these graphs are not useful.

[ny-quant.livejournal.com]

Вы какие-нибудь таблетки когда-нибудь принимали?

// What I don't know is the protocol for measurements. Without knowing what is being measured and how they are measuring it, these graphs are not useful.

On this level, everything we "know" at this time are varieties of garbage. But these charts provoke most decisive rejection for some reason.

[misha-b.livejournal.com]

> Вы какие-нибудь таблетки когда-нибудь принимали?

Most new medicines pass pretty rigorous double blind testing. If it is something important or life threatening, I would definitely review the literature.

> On this level, everything we "know" at this time are varieties of garbage. But these charts provoke most decisive rejection for some reason.

It is not quite true. Most models are indeed garbage, but we know for sure that the virus needs physical contact or physical proximity to propagate. Therefore lockdowns are pretty much guaranteed to decrease the spread.

Thus the narrative being built on these graphs must be incorrect. Without knowing what is being measured, it is not possible to provide any alternative interpretation though.

I feel you are thinking too much in liberal/conservative terms. We just need to look at the data and combine it with what we know to come to reasonable conclusions.

[ny-quant.livejournal.com]

// lockdowns are pretty much guaranteed to decrease the spread

That's a false narrative. Lockdown is not one measure it's a complex of measures. Some are effective some may be not. The point is that relaxing of ineffective measures is a win-win.

// I feel you are thinking too much in liberal/conservative terms.

Maybe but I disagree.

// We just need to look at the data and combine it with what we know to come to reasonable conclusions.

This is exactly the goal of the exercise. And then some people reject the reasonable conclusions because they contradict their priors or political views.

[misha-b.livejournal.com]

> That's a false narrative. Lockdown is not one measure it's a complex of measures. Some are effective some may be not.

You should think about lockdowns as a complex of measures designed to decrease physical contact. Some of them are effective, some are not. Lockdown as a whole slows the viral spread.

> The point is that relaxing of ineffective measures is a win-win.

Sure. Even effective measures may have to be relaxed if the cost of maintaining them is too high.

> And then some people reject the reasonable conclusions because they contradict their priors or political views.

Perhaps. Personally I don't have any political views about the virus. My only prior is what we know from physics, biology and mathematics.

[ny-quant.livejournal.com]

// Lockdown as a whole slows the viral spread.

Correct but it doesn't follow that lockdowns as implemented are the optimal way to do it. In fact, the data suggests that the implementation was excessively strict.

[misha-b.livejournal.com]

Actually, I would like to see some experimentation. It is hard to judge what is the right level of strictness without reference points.

[nameless--one.livejournal.com]

Я из многих источников слышал что главная причина, по которой ОРВИ в основном приходятся на холодный сезон - это то что люди проводят больше времени в помещениях. Соответственно, я не вижу причина априорно полагать что локдаун (когда людям велят не выходить без крайней необходимости на улицу) замедлит распространение. Ну, возможно и замедлит, но это нужно обосновывать.

[misha-b.livejournal.com]

If you don't interact with people, you would not get the virus, no?

[nameless--one.livejournal.com]

You still interact with your family members. Much more than normal. And at least some of these family members go shopping, work not from home i.e. have outside interactions.

Also - why prohibit leaving home which does not involve interacting with other people - parks, forests?

[misha-b.livejournal.com]

Well, you interact with your family members a lot either way, more than enough to spread the virus. The key is interacting with other people. I am not sure shopping is a major risk factor -- you don't do very frequently and it does not take that long.

> Also - why prohibit leaving home which does not involve interacting with other people - parks, forests?

Beats me.

[ny-quant.livejournal.com]

// Beats me

But it is part of the point: excessive ineffective measures that I mentioned above.

[ny-quant.livejournal.com]

No. Our close friends didn't interact with anyone and still got infected somehow. Possibly via food shopping which BTW was allowed under all lockdowns.

[misha-b.livejournal.com]

Very strange. I did not see any info that shopping could spread the virus but I guess it's possible.

[tandem-bike.livejournal.com]

интересно. а они просто позитивны или болеют?

[ny-quant.livejournal.com]

У двоих из троих "заболевших" пропало обоняние и больше ничего. У третьего вообще ничего. Сейчас уже все они негативны. У четвертого члена семьи все тесты негативны, в том числе и на антитела. Т.е. просто полный и ничем не объяснимый иммунитет.

[misha-b.livejournal.com]

Btw, I was thinking about why the 99% thing is so annoying to me. The issue is the following -- we need to separate our hypothesis from the null hypothesis. 99% is the probability that the data does not come from the null under some probabilistic model.

Without describing the model for the null 99% makes no sense. And the definition of the null is non-trivial -- for example it does not make sense to compare this to iid data, since the rates in different states or countries are obviously correlated. Thus, giving a number like 99% seems manipulative, designed to give some scientific veneer to potentially problematic inferences.

[ny-quant.livejournal.com]

OK, I see your point. But what is the practical outcome? To throw the baby with the bathwater?

[misha-b.livejournal.com]

First we have to make sure that there is a baby. For that we need to know something about their methodology, which they seem to be hiding.

I actually find it hard to believe that those JPM guys do not understand statistics or how to properly deal with data. They seem to be solid quantitative types. Therefore it feels like they have some agenda.

[ny-quant.livejournal.com]

I meant it in general. If you are generally annoyed by the hypothesis testing methodology then why stop with this paper? Why not throw out everything that statisticians have done this way?

But yes, the guy is very smart and silly mistakes are not to be expected.

[ny-quant.livejournal.com]

Я согласен, что p-values могут быть misused and abused. Но это ещё не повод видеть желтый флаг в significance level.

[misha-b.livejournal.com]

Without an explicitly stated model it definitely is. If there is a model, we have to consider its validity.

[ny-quant.livejournal.com]

// 99% is the probability that the data does not come from the null under some probabilistic model.

Could you remind me: if we're testing a linear regression, what is the null hypothesis? I don't recall this being discussed in the last class I took almost 30 years ago, nor in anything I've read later. I may be wrong but I believe it is not necessary b/c we're hiding behind the CLT.

[misha-b.livejournal.com]

It might be something like this. Say we are in a fixed design setting, i.e. we have pairs (x_i, y_i), where x_i are assumed to be fixed. We write a linear model y = w*x + e, and put some distribution on the noise e , e.g. e is a Gaussian. Then we can work out various probabilities in terms of that.

Take this with a grain of salt -- I have not looked carefully at such problems in the testing context.