1/ The following is a recreation of a #COVIDー19 tracking chart that was presented to me this week by a major bank. It is common to what is shown repeatedly in the press. The interpretation in FLAWED. #VIRUS

2/ The bars show a running tally of infections in the US, updated daily. The line chart shows the growth in the total number cases, on a daily basis. This line suggests the growth rate is slowing. I submit that it is not.

3/ I've been trying to come up with an analogy to explain the flaw. Here's my first attempt using Fishing.

4/ A guy goes fishing at lunch. On day 1, he takes one pole, fishes for an hour, and catches 2 fish. On day 2, he takes two poles and catches 4 fish. On day 3, he takes a friend and a total of 4 poles. Together they catch 8 fish.
T1 = 2, T2 = 4, T3=8.

5/ The daily total can be summarized with a formula:
QTY = 2^T where T is the number of days

2^1 = 2
2^2 = 4
2^3 = 8

But what does that tell us about the number of fish in the pond? Is the number growing at an exponential rate?

6/ Tracking the number of raw cases provides us with shocking headlines. But it tells us nothing about the growth rate of the virus.

Ideally we would be conducting randomized testing and mapping the proportion of positive tests over time. But we are not doing that yet.

7/ We are predominately testing people with symptoms. Of those that we are testing, we are still seeing acceleration in the # of infections.

Bars = test results by day
Gold line = % of positive cases each day
Blue line = % of positive cases over time
http://covidtracking.com/data/us-daily 

8/ Caution: the prior chart is not based on randomized testing. Two plausible inferences: 1) infections are increasing; or 2) doctors are getting better at identifying potential cases for testing.

9/ Once we begin randomized testing, we can get a more accurate understanding of what is truly happening. Until then, I urge you to use extreme caution when making inferences from biased data sampling. And do not rely on RAW values.

END.