3 ms·
Later in thread > Trevor Bedford @trvrb · 14h I'd assess the p-value for this coincidence at 2/59=0.03 and so is statistically significant. Additionally, these
by sixhobbits 7y ago
Later in thread
> Trevor Bedford
@trvrb
·
14h
I'd assess the p-value for this coincidence at 2/59=0.03 and so is statistically significant. Additionally, these two cases are geographically proximal, both residing in Snohomish County. 5/9
That's... really not how p values work. Does everyone have to mention p values now just to try sound like they're authoritative?
- mlyle 7y agoI cringed when I read it. At the same time, you can have a meaningful p value from a single observation. If you were assuming novel spread from China and a random selection of virus genotype, you'd expect a coincidence like this to happen about 3% of the time if the null hypothesis held. So it's not completely bullshit. Big caveat is-- inferring this from a single genetic difference with 59 reference sequences taken at different times is dubious. That 3% number could be drastically different in reality. Still-- cryptic spread from the earlier case is probably the most reasonable explanation for what we've seen.