Read on Twitter
Thanks to some friends, I now have a functioning Rt model for #NJ #COVID19 that matches up pretty well with what the state has been providing. While going through the process, I uncovered some interesting details that I'm sure our Gov. and his team has been keeping a lid on 1/
The first thing I noticed is that prior to the recent increase, the curve supplied by the NJDOH appeared to be raw data. Note how the curve has jagged edges, quick changes in direction, etc. This curve does not behave like an averaged dataset. But, since ~7/25, this changed 2/
Now, the data is smoother, and as of Jul 29th, a new statement appears in the footer which states, paraphrasing: "Based on the 7-day trailing average of daily Rt values, to smooth out artificial fluctuations such as batching of lab reporting." This has caused integrity issues 3/
The addition of this average has stretched out the period of time that Rt stays elevated, when you compare Raw data output (Red) vs. the now averaged Rt (Blue). Raw Rt has ALREADY returned to a value under 1.0 (0.93), while the average is still lingering higher. 4/
Finally, the Gov. claimed that despite backlog, the Rt "would still be above 1". As we now can see, this absolutely FALSE. Green redistributes the cases that were dumped while not changing the total number of cases. Rt would be down to 0.97 if Quest data was released on time 5/
The public now has a tool that matches fairly well with what NJ is reporting, despite multiple calls for the Administration of openly sharing the calculation. This model is based off of the same EpiEstim package provided by the Imperial College London. 6/
Its a shame that the press failed to hold the Gov accountable for transparency during this time frame and a concerned resident had to go through these lengths to figure it out. Its disappointing that Rt has already returned Sub 1.0, yet NJ manipulated the calc on the fly. 7/
