Thanks, Randall, for your reply. Your concerns about complexity are valid and well taken. They were considered: I suggested that an approach akin to industry's EVOP/Evolutionary Ops could also be applied. In it, data would be collected as changes were implemented in the field). Not all regions appliy containment measures simultaneously; some do so incrementally. This would account for factor variation in experimentatioin.
Also, considering the carry-over effect is important. In cattle feeding DOE experiments, where animals are rotated about different grains, there is a period between data collection to account for the elimination of residual effects (carry-over effect) from one grain feed to the next. It could also be applied here. We have an entire century of Design of Experiments experience, since Dr Fisher started applying them. All this experience could be transferred to the Covid-19 situation.
Other parameters should also be considered. For example, how long has the containment measure been applied (e.g. how many weeks of social distancing) and, if possible, at what level (sometimes the distancing can be measured and quantified through cell phone use). These extra measurements could lead to regression analyses. These not only identify which factors (or containment measures) impact the response (virus infection rates) but also help establish how many x-amounts of a given factor induces y-amounts of infection rate decrease.
One of the large benefits of involving the ASA, as an organization member of the American Civil Society, and of its professional statistician members, into this process, is the identification of which data is needed to perform which types of analyses, in what forms, and how are they to be collected. Much more knowledge could be gained with such multilateral cooperation!
A data bank or web site should be created and maintained by a joint government-professional society group. Statisticians could access and use this data to implement different types of analyses. Resultscwould then be submitted to this site, for consideration and use by researchers and government. In many cases, these organizations do not have the necessary expertise to come up with some of these specialized approaches. But once they become aware of them, and see their benefits, these researchers and government authorities may have their staff implement them -or they may contact the analyses authors and request assistance.
You may want to refer to my previous proposal on the cooperation of retired faculty/researchers, that you can find in:
https://web.cortland.edu/matresearch/Covid-19Proposal2020.pdf
Again, thank-you for you excellent concerns.
Keep safe. Best/Jorge.
------------------------------
Jorge L. Romeu
Emeritus SUNY Faculty
Adjunct Professor, Syracuse U.
https://www.researchgate.net/profile/Jorge_Romeu------------------------------
Original Message:
Sent: 05-22-2020 09:50
From: Randall Tobias
Subject: About Spreading the Use of Statistical Analysis for Covid-19 Data
Certainly an interesting idea, but have you considered some of the implementation issues? How well would counties/regions tolerate and comply with turning on and off containment measures on a weekly basis? How well would they tolerate/comply with having different containment measures than other counties? Given that there is an incubation period for infection, would you design and model the experiment taking carryover effects into account?
Many questions ...
------------------------------
Randall Tobias
SAS Institute, Inc.
Original Message:
Sent: 05-21-2020 10:50
From: Jorge Romeu
Subject: About Spreading the Use of Statistical Analysis for Covid-19 Data
Currently, not all regions/counties that have opened up for business are using statistical methods to assess the effects of the containment measures they have implemented to control the spread of Covid-19 in their areas. The present example/tutorial illustrates how Design of Experiments/DOE can be used in Covid-19 virus data analysis. First, it can be used in factor screening, to identify which of the containment measures are actually working to keep the infection rate at bay. Secondly, to quantify the effects of such significant measures/factors in infection containment, by providing confidence intervals. The data used in our example has been made-up, as (1) our objectives are to illustrate the Power of DOE in tackling Covid-19 problems, and to "educate" the authorities in the use of statistics, (2) and we found no real data set available.
Many political leaders are stating how the struggle against Covid-19 should be based on "science and data". Data, is collected (sampling), then analyzed (hypothesis testing), and its results are then generalized (inference, modeling), And Statistics is a "science".
The ASA, as a national professional organization, has an important role to play here.
Example of Design of Experiments to Assess the Effects of Containment Metods on the Covid-19 Infection Rate:
https://www.researchgate.net/publication/341532612_Example_of_a_DOE_Application_to_Coronavarius_Data_Analysis
------------------------------
Jorge L. Romeu
Emeritus SUNY Faculty
Adjunct Professor, Syracuse U.
https://www.researchgate.net/profile/Jorge_Romeu
------------------------------