BonsaiNut
iMTB Rockstah
According to this graph our collective behavior moving forward will have a big affect on the projected number of cases.
A statistical model is only as good as the quality of its variables. Unfortunately, though I read their method detail, they did not share their math. I am particularly interested in their "social distancing" variable, since it is not social distancing at all, but rather "cell phone travel patterns". Always be careful when a "statistician" starts renaming their variables to describe them as something other than what they are.
There are huge gaps in their logic. There is no mention of true underlying case rates (supposedly 25%+ in New York City and 20%+ in LA), or the impact of of herd immunity. There is no mention of natural seasonality - and the natural reduction in social interactions in the summer as students are out of school. No mention of the structural changes to the economy, and the 20% of people who are now unemployed and won't be returning to work any time soon. There is no behavioral component included - and the assumption is that anyone returning to any non-essential business is going to behave exactly as they did PRIOR to the pandemic, and there will be no health-related behavioral changes (no masks, no sterilizing shopping carts, no increased washing of hands, no gloves at gas stations or increased distancing at restaurants, etc). Basically they are expecting us to believe that the nanosecond social distancing protocols are repealed, everyone is going to instantly return to "business as usual".
I don't buy it.
(1) Correlation does not equal causality. Just because they can get numbers to fit history, does not automatically mean they can use those same numbers to project history forward.
(2) The only thing worse than no math is bad math. This is bad math.
(3) Tell me the result you want to see, and I can create a statistical model to show that result. It doesn't mean that the model is correct... just that the math works.
Good thing you don't live in Snohomish County, WA. If they loosen up social distancing rules by even 50%, the entire population is going to die by October (sarcasm).
Last edited: