Quote:
Originally Posted by Festivus
Using 1% of a population for a scientific, randomly-selected survey would be massive overkill. I mean, you can do pretty good opinion polls for all of Canada with 500 respondents. The issue is you need to make it completely random, and representative of the overall population.
The COVID-19 tests aren't a random survey though, they are self-selective, meaning that only people who volunteer (and fit the criteria) get the test. It's like phoning homes for a poll vs an online survey on a single website. One gets you a representative sample (with modelling), the other gets you bunk.
However, for medical testing like this you want a self-selected survey, because people with symptoms are more likely to have it than people without.
I am 100% certain there are other cases out there, but until we have enough tests to randomly start testing extra hundreds or thousands who have no suspicion of being infected, we should keep using tests on people we think are more likely of being infected.
|
I'm sorry what evidence do you have for this? I have heard on multiple occasions that the probability is exceptionally high that a large plurality, if not an outright majority or maybe even a vast majority of cases are either a) asymptomatic, or b) have very mild symptoms. <https://theconversation.com/coronavirus-bmj-study-suggests-78-dont-show-symptoms-heres-what-that-could-mean-135732> (just one article that I've read recently)
If this is even remotely true, this business about favouring symptomatic cases for testing, means that you're getting a wildly biased sample, which means that your actual infection rate is wildly different than the reported case rate. It follows then, that any sort of measurable death rate is also largely useless without a well constructed random sample test to record the actual infection rate. This should also take into account those who have developed antibodies to the virus, so we can see the true extent off all infection. The higher the denominator, the lower the mortality rate, the closer we are to being able to get over the hysteria and panic.
I don't inherently disagree with reserving limited tests for those who present showing symptoms, but we need actual data to start making informed policy decisions. At the moment, we are using badly inflated numbers to justify quasi-fascist lock-downs, and these aren't sustainable in the short term, nevermind for an indefinite period of time. We need to start coming up with restart plans fast, because right now, we're basing our assumptions on models that don't correlate with reality, which is why the estimates for total deaths in the US has been reduced several times in the last couple of weeks, and which already accounted for social distancing where around 50% of people decided not to comply. The most recent suggests around 60k by the end of July, more in line with a bad flu season than anything else.
Look, I understand that this thing is highly transmissible and that everybody is scared because we've put this thing in the spotlight and are hyping it up, but we gotta start looking at letting people make their own risk assessments. If you're vulnerable, maybe you stay home for a while longer, but for those who aren't part of vulnerable groups, it is going to be to our distinct advantage for a bulk of them to get infected and develop a herd immunity.