Since the 10-day stretch in January when I worked very odd hours because of the coronavirus outbreak, a ton of people screaming about how the data is not trustworthy. OK - there are many who have presented reasonable doubts about the quality of China data over the years. 1/?
But at some point the "CHINA DATA IS BAD AND CANNOT BE TRUSTED" yelling, frankly, will get tuned out. There comes a time when the pushback comes and you have to start giving at least some circumstantial, quantifiable and verifiable proof to show how bad the data is. 2/n
Yes, there are anecdotal reports about people who unfortunately died in the streets, hospitals etc at the height of the outbreak. Can you prove that these people were, in fact, not counted? How many of such deaths occured? Do you haev any quantifiable evidence? 3/n
It is common practice to build models to try to get at the trajectory of COVID-19. But whenever I share data points such as fatality rates many folks scream how many countries vastly undercount # of infections. So, which model has the most accurate underlying assumptions? 4/n
There have been attempts trying to hint at under-reporting, including the number of cremations performed in Wuhan. But we also know that there is a lot of non-COVID-19 deaths happening because people with other diseases can't get proper medical care because of the pandemic. 5/n
China has done plenty publicly to attract suspicion and raise eyebrows. Such suddenly adding 13,332 cases onto a single daily toll and totally distorting all trendlines. Or not counting asymptomatic cases as confirmed cases. Or changing diagnosis standards multiple times. 6/n
But there will come a time (we might be there already) when simply saying that Chinese data is not reliable won't be enough to get anywhere. Numbers. Internal documents leaked. Something to show the extent of how "wrong" the data really is. Otherwise, it's more of the same. 7/n
Will we ever see hard evidence proving that China vastly undercounted death/case tolls? I don't know. All I can say is I have yet to see the statistical flag that will get a critical mass to rally around and make their stand on.

8/n
It is also possible that we never quite grasp the true extent of the COVID-19 epidemic in China because of a) what even China acknowledges as poor response/under-reporting by Wuhan and Hubei when the outbreak emerged b) poor quality of initial test kits c) asymptomatic cases 9/n
I can't take a single model as the holy grail. There is an absolute flood of papers - I don't have the expertise to know what is legit. I am waiting for someone/some organizations throwing its authoritative weight behind one. Not sure if/when that happens. 10/n
It's my job to not plainly accept accusations of lies/wrongdoing/cover-up unless there is incontrovertible proof. Can I have private opinions about what I am seeing and whether I believe it? Yes. Can I use those opinions and inject them into the work I am doing? No. 11/n
So yeah. If somebody just screams "THE DATA IS WRONG" and offers nothing credible that would hold up in, say, a criminal court hearing, then I have to tune that out.

The end.
You can follow @Rover829.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: