why do i continue to cite these low statistics - 0.5% mortality rate of those infected - when the official numbers are higher than that? won't msnbc be angry with me?

because i'm statistically literate. i know better.

i think it's widely understood now that there's a lot of people evading medical attention, which was initially a point of contention, and is why it's impossible to actually stop. it's also why travel restrictions are more about the government finding a scapegoat than they are about actually reducing the spread of anything. a proper empirical analysis would be that we should stop trying to prevent this thing from spreading and spend more time adjusting to the new reality: this is here to stay, and at-risk people are going to have to change their lifestyles as a result of it. you can blame vacationers or foreigners all you want, but it's not a science-backed policy position. this thing is everywhere, and it's going to be everywhere forever.

so, you want to look at the cases that come in as a representative sample, and then build a model from that representative sample, not just take the data at face value. unfortunately, self-selection produces kind of a shitty bias.

i've suggested doing random sampling to get a real understanding of this...

most of the places where there are a lot of cases are dealing with overburdened health care systems, which means the weak cases aren't getting tested, which is producing these high morality rates. the highest rates are in italy, spain and iran, right now - over 7%. china has reduced this number quite a bit, but the fact that it still has a mortality rate much higher than 4% indicates that there are a lot of unaccounted for cases.

there are some countries that have seen very low mortality rates, as a consequence of extremely aggessive testing.

in germany, the mortality rate is 0.6% - still, even after almost 40,000 cases. austria, which directly borders the affected area in italy, has many fewer cases, but the same 0.6%.mortality rate. norway is at 0.5%, after 3000 cases, as are the australians after 2400. after less cases, but still more than 1000, the czechs are at 0.4%, luxembourg is at 0.6%, ireland is at 0.5%, chile is at 0.3% and pakistan is at 0.8%.

and the jews, who have border security that works, are outsmarting everybody at 0.2%, after 2369 cases.

these are the best health care systems in the world, clearly. but they're also the countries doing something right. but, careful - they might also be doing things that are wrong or ineffective. causality is not the same thing as correlation.

nobody else has 1000 cases yet, which seems arbitrary, but you could work this out using the central limit theorem - there is an n that is enough. i'm guesstimating 1000. other countries are nearing an n>1000, and many of them have mortality rates <0.5%, too.

but, i'm chery-picking, surely! no...

if x (i can't use fancy greek letters here. sorry.) is the true mortality rate, an example of the actual term statistic, then you'd have to estimate it being less than anything observed due to the nature of the study. if you do two hundred real-world hospital studies, x would have to be some kind of limit of the minimal values, because even in the best real-world hospital studies you're going to miss patients. if you did studies in places like dorm rooms or fitness centres, you'd overshoot the other way. so, i'm pointing out a sampling bias, here, i'm not cherry-picking the best results. if you're lucky, you'll get pretty close - but it's essentially impossible to overshoot it.

so, that would mean that the actual mortality rate is less than the 0.2% in israel, unless you think that the jews are the master race, or something. no, really, that's your argument.

so, generating a bound for x, x<0.002, means that these other countries that have mortality rates higher than that, and especially higher than 0.01, are missing a very larger number of cases, and this is likely to get worse as the cases spread. they don't know know where they are....

there's an upside to that, though.

if there are 7500 deaths in italy, and the true mortality rate is 0.1%, that would indicate that there are over 7.5 million cases in italy, right now - not the 75,000 they know about. most of these will be in the lombardy area.

the population of lombardy is 10 million people.

they may be nearing herd immunity, and that's when this slows down and stops - not in spite of their failure at containment, but because of it. 

it would also indicate that there's only 35,000 cases in canada, and we have a very long way to go before this burns itself out.