We'll see. Deaths per day is showing a definite downtrend. I do not have a reason to think to know this or not, but perhaps some of the new case data is lagging the deaths data, due to waiting on testing results and compiling the data. The death rate dropping is very telling, because the less people you have going to the death bed with the virus, the less it was proliferating ~14 days prior to that. Or in theory at least, that should be the case. I certainly DO NOT want to jinx anything though. Let's keep the trend going!
We both want the same thing. I fully admit I'm applying my own pessimistic tendencies when looking at the numbers. I would very much like to be wrong and see a huge, rapid drop off.
Didn't know there had been a new election for us to "cosign" on. So you agree that "Loathing" republicans because of Newt and Trump is the right way to handle things? I mean "loath"? I don't Loath anyone based on our political differences. Loathing Trump, I get. But loathing republicans? C'mon man. There is an entire thread about Amash and how many republicans like myself would vote for him. Heck, I openly said I wish Trump had been impeached and removed just to get Pence some time before November. But sure..."loath" republicans.
It depends on the purpose, but if the goal is to extrapolate the study on the population of even that county much less the country in general, very flawed is correct. An opt-in sample is a very big problem if the goal is external validity.
The peak in the daily new cases was April 4th and the peak in daily deaths was April 15th (if you take out the 3,778 deaths NY lumped into the 14th from previous days) so the lag is 11 days.
I agree, but with that known going in and with in noted from the start, I don't consider that a flaw. I consider it a noted factor. Boston is seeing similar things in their homeless and in their sewer testing of all things. You have Harvard Scientists saying the death rate is mortality is likely less than 1% of people who contract, and the death rate is much lower than originally feared. I doubt the 85% in the Stanford study holds true as it literally can't in some places mathematically, but it can be a sign that things are much better than we think, and that is good news IF true.
Admitting a flaw, while good, doesnt eliminate it or even limit it. Again, if your goal is to extrapolate to any population beyond the sample, it really is a mess of a sample. It is the type of study that, unfortunately, people sometimes run for media attention. It popped an interesting number and many media members cant evaluate a study. But its external validity is essentially non-existent. Until we get fully random testing, we wont have a study with even external validity into their populations. It isnt a sign of anything, unfortunately, because of the sampling issue.
Wasn't always like that for me, and I'm essentially independent (big 2A supporter) anyway; but the vast majority I interact with these days are a whole other breed that is openly racist, angry, and increasingly confrontational... what's sad to me is so many of them don't even realize, recognize, or acknowledge they're doing it even while it's happening, but it's jarring to someone who doesn't expect it in polite conversation.
I don't know the science of it, I'll admit, but there is a justification the model is using for that rapid drop off it's predicting. Honestly, stemming the tide of new cases to a slight down trend is a major feat in of itself, if you think about it. So it's not out of the realm of possibility to see containment of that magnitude over a period of a few weeks.
You need better acquaintances. I attend large church services with thousands of them and they are every race and creed. This is your anecdotal view that should not be lumped into an entire block of people.
Did Mutz not explain the sampling flaw? Yes - it means the study is flawed. Stanford coronavirus study triggers feud over methodology and motives
As I said, it strikes me as the type of study designed to pop a media friendly number, which is a pretty big ethical issue given the stakes right now. It is likely that the effect that they are studying exists to some degree, but the exaggeration of it through such a severe sampling issue, along with the attempts to apply it to populations beyond their samples is a very big issue. If their goal was just to demonstrate some other part of the methodology beyond sampling, it would be okay, but if the goal was to describe the world, it was a very bad flaw and scientists at Stanford likely know better.
Yes. I was in that conversation. I see a flaw and a KNOWN sample limit as different things. They knew AND stated up front that the study was limited. That does not mean there were flaws in the study. It just means it was a very limited sample. They said that right up front in the very initial release. If I conduct a study that says 95% of people in The Swamp on a Saturday are Gator fans, it is not a flaw....but it is a limited sample as you can not extrapolate it to all stadiums. The difference here is that we are starting to see other areas (Boston for instance) say similar things just with a different sample. I know it may be semantics, but their study is not flawed. It is limited.
You're right . . . not flawed. From the link in my previous post: See? mere imperfections - not flaws.
But it would be flawed to conclude that 95% of the total population are gator fans based on this self-selected sampling of game attendees. Which is exactly what the Stanford study is suggesting.
Stanford isn't peddling news. They were clear it was not data that translated elsewhere, or at least not until similar studies could be conducted elsewhere. Their own admission is how we know the sample was limited. They said it up front. No one at Stanford implied that the data could be extrapolated elsewhere.
Limitation: all of our sample comes from one county and therefore external validity is limited to that one county Flaw: our sample was opt-in and now we can't even utilize the study to make inferences about the county from which the sample was taken. If you say 95% of the people in the swamp are Gator fans, that would not be a flaw unless you tried to say it applied to the population as a whole. But if you posted something on Facebook on a Gator group and the only people to show up as you counted were Gator fans, so you have a 100% Gator reading, that is a flaw.
They extrapolated it to the county population. Their sample is not an appropriate sample with which to do that. From the results summary in the paper: https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v1.full.pdf
