Hacker News Reader

Comment by bitbuilder | Hacker News Reader

>Is there a dashboard which shows increase in testing capacity over that same period

Here's the best resource Ive found for comparing all the relevant data:

http://91-divoc.com/pages/covid-visualization/

Playing around with the data there, it looks like yes NC has had an increase in cases recently, but it tracks an increase in testing. Test positivity rates are about flat. It really is completely pointless to discuss increases in cases without also comparing to increases in testing and positivity rates.

Deaths are up, but as another poster pointed out it's hardly statistically significant. The rolling 7 day average is 2.2/1m now, vs 1.7/1m on May 1st.

View on HN

Be careful not to swing too far the other way and assume that a flat positivity rate implies a flat actual infection rate. As states increase testing capacity, lower risk individuals will start to become part of the testing pool (e.g. those without severe symptoms, those with only fleeting contact with someone infected, those with no connection at all as part of a disease surveillance measure, etc.). Thus a flat testing positivity rate can very well coexist alongside an actual increase in rate of infection, and given the aforementioned factors we might even infer the latter given the former.

That's a fair point, and definitely something to keep in mind. Thank you for highlighting that.

I know you're not debating NC, but looking again at the state, it was mostly the last few weeks that looked flat. If you compare to the peak in positivity 42 days ago, they've gone from 16% positive to to 7% positive, using rolling 7 day averages. So, if I were in NC I wouldn't be overly concerned.

>"Deaths are up, but as another poster pointed out it's hardly statistically significant. The rolling 7 day average is 2.2/1m now, vs 1.7/1m on May 1st."

How is a 34% increase not statistically significant?

That's a statistical/maths term[1], not an opinion of the number.

[1] https://en.wikipedia.org/wiki/Statistical_significance

>How is a 34% increase not statistically significant?

Because going from a daily death rate of 0.00017% to 0.00022% is still such a very small number, what might be called statistical noise.

Because we're talking relatively small numbers. It's easy to create large movements in small numbers.

If one person in my town is murdered this year, and 2 are murdered next year, that's a 100% year over year increase in the homicide rate. But it's obviously a meaningless increase.

2.2 out of a million is a small number. More people in NC are dying every day from cancer or heart disease.

Reply on news.ycombinator.com