ASA Connect

At noon today, I again listened to Gov. Cuomo's daily report on Covid-19. He presented an interesting statistical chart where NYC infection rates went down, in six weeks, from 24% to 2%. As a professional statistician I believe that these NYC statistics are not comparable, as the samples on which they were based are different.

Six weeks ago, at the height of the NYC Covid-19 crisis, and with a low testing capability, the testing took place at the hospitals where sick patients arrived, and at testing sites where doctors sent patients that reported synthomps. Currently, Covid-19 has subsided in NYC and testing capacity has grown. Thence, more random testing is now taking place.

It is natural, even without any real change in Covid-19 infection rates, that the former results (in hospitals) were larger, and the latter (random) are smaller (estimation of NYC Covid-19 prevalence). This is a technical issue, that will naturally escape the non-initiated. The ASA, and statisticians, can contribute by providing guidance and advise to the government in this technical area.

This posting is, by no means, intended to be a criticism of the NY Governor or his team, which we believe have stated the above results in good faith. It is intended to provide yet another argument that (1) authorities need to seek technical stats guidance from professionals, and (2) that the ASA has an important role to play, by way of providing such guidance and help to these.

Thanks/jorge.

------------------------------
Jorge L. Romeu
Emeritus SUNY Faculty
Adjunct Professor, Syracuse U.
https://www.researchgate.net/profile/Jorge_Romeu
------------------------------

My only quibble with your post is that it is not just governments that need appropriate statistical advice.  Private industry and great social organizations need help not only with appropriate data analysis but also with display of data when the data and analysis is good.  Here is an example of good data on display (on a San Francisco billboard - originally found on Twitter - but I also saw the billboard):


Whoever designed the graphic used the diameter proportional to the amounts rather than the area proportional to the amounts.  We have good information but exaggerated (not that it needed exaggeration) which could reduce its credibility.  This is not a case for intentional exaggeration but is a case for statistical help being needed everywhere.



------------------------------
Jim Baldwin
Retired
------------------------------

Original Message:
Sent: 06-04-2020 14:28
From: Jorge Romeu
Subject: An example of governmental need for statistical advise

At noon today, I again listened to Gov. Cuomo's daily report on Covid-19. He presented an interesting statistical chart where NYC infection rates went down, in six weeks, from 24% to 2%. As a professional statistician I believe that these NYC statistics are not comparable, as the samples on which they were based are different.

Six weeks ago, at the height of the NYC Covid-19 crisis, and with a low testing capability, the testing took place at the hospitals where sick patients arrived, and at testing sites where doctors sent patients that reported synthomps. Currently, Covid-19 has subsided in NYC and testing capacity has grown. Thence, more random testing is now taking place.

It is natural, even without any real change in Covid-19 infection rates, that the former results (in hospitals) were larger, and the latter (random) are smaller (estimation of NYC Covid-19 prevalence). This is a technical issue, that will naturally escape the non-initiated. The ASA, and statisticians, can contribute by providing guidance and advise to the government in this technical area.

This posting is, by no means, intended to be a criticism of the NY Governor or his team, which we believe have stated the above results in good faith. It is intended to provide yet another argument that (1) authorities need to seek technical stats guidance from professionals, and (2) that the ASA has an important role to play, by way of providing such guidance and help to these.

Thanks/jorge.

------------------------------
Jorge L. Romeu
Emeritus SUNY Faculty
Adjunct Professor, Syracuse U.
https://www.researchgate.net/profile/Jorge_Romeu
------------------------------

I agree with your comment about the need, from industry and organizations from civil society, for receiving stats advice. I have written a "proposal" regarding the need to provide such help, pro-bono. You can find it in:

https://www.researchgate.net/publication/341282217_A_Proposal_for_Fighting_Covid-19_and_its_Economic_Fallout

The main issues I see with your suggestion are that private industry: (1) usually has its own statisticians, (2) doesn't like to "share" their proprietary information, and (3) working pro-bono for them is anathema (they are in this for profit, thence, such work should be "consulting"). I have no issues with organizations from civil society.

Government, on the other hand, is everybody's property. They may also be more inclined to receive/accept such help.

Thanks for your comment. Best/Jorge.

------------------------------
Jorge L. Romeu
Emeritus SUNY Faculty
Adjunct Professor, Syracuse U.
https://www.researchgate.net/profile/Jorge_Romeu
------------------------------

Original Message:
Sent: 06-08-2020 19:38
From: Jim Baldwin
Subject: An example of governmental need for statistical advise

My only quibble with your post is that it is not just governments that need appropriate statistical advice.  Private industry and great social organizations need help not only with appropriate data analysis but also with display of data when the data and analysis is good.  Here is an example of good data on display (on a San Francisco billboard - originally found on Twitter - but I also saw the billboard):


Whoever designed the graphic used the diameter proportional to the amounts rather than the area proportional to the amounts.  We have good information but exaggerated (not that it needed exaggeration) which could reduce its credibility.  This is not a case for intentional exaggeration but is a case for statistical help being needed everywhere.



------------------------------
Jim Baldwin
Retired

Original Message:
Sent: 06-04-2020 14:28
From: Jorge Romeu
Subject: An example of governmental need for statistical advise

At noon today, I again listened to Gov. Cuomo's daily report on Covid-19. He presented an interesting statistical chart where NYC infection rates went down, in six weeks, from 24% to 2%. As a professional statistician I believe that these NYC statistics are not comparable, as the samples on which they were based are different.

Six weeks ago, at the height of the NYC Covid-19 crisis, and with a low testing capability, the testing took place at the hospitals where sick patients arrived, and at testing sites where doctors sent patients that reported synthomps. Currently, Covid-19 has subsided in NYC and testing capacity has grown. Thence, more random testing is now taking place.

It is natural, even without any real change in Covid-19 infection rates, that the former results (in hospitals) were larger, and the latter (random) are smaller (estimation of NYC Covid-19 prevalence). This is a technical issue, that will naturally escape the non-initiated. The ASA, and statisticians, can contribute by providing guidance and advise to the government in this technical area.

This posting is, by no means, intended to be a criticism of the NY Governor or his team, which we believe have stated the above results in good faith. It is intended to provide yet another argument that (1) authorities need to seek technical stats guidance from professionals, and (2) that the ASA has an important role to play, by way of providing such guidance and help to these.

Thanks/jorge.

------------------------------
Jorge L. Romeu
Emeritus SUNY Faculty
Adjunct Professor, Syracuse U.
https://www.researchgate.net/profile/Jorge_Romeu
------------------------------