During my service on the ASA Board of Directors, 2004-2007, I participated in several lively discussions about whether and how the Association should engage in contentious issues--political, scientific, economic, cultural--in which statistical science played a key role. I recall a clear majority of us voicing that the ASA should do more than organize meetings, publish journals, and be a catalyst for member-to-member interactions throughout the year. We recognized that in promoting sound statistical practice, the ASA should, at select times and in respectful ways, openly criticize bad practices that could result in misinforming the public and our policymakers.
I doubt no then-member of the Board envisioned how the American political landscape would change over the coming decade. Data collection and analysis, and how statistical information is interpreted and communicated, have always been at risk to be warped by those striving to "prove" their points. Today, the hurricanes of politicized statistics are evermore frequent and consequential. When those storms hit America, the American Statistical Association should respond, not by opposing a given political position, but by exposing a bad statistical practice regardless of political position. Whenever all three branches of the U.S. government are controlled by the same political nexus, those in that nexus will inevitably engage in statistical practices that warrant just criticism. Our pilots of our ship of state will cause far more accidents than us passengers.
You don't have to be the CEO of Gallop to understand that placing a "loaded" question within a survey can ill-affect participation rates and produce response bias for other questions in that instrument. Professional statisticians should be involved in planning all studies of import, including helping to scrutinize the questions with potential biases in mind.
When the "survey" in question is the U.S. Census, this planning is especially important and sensitive. While Census Bureau statisticians are certainly involved in creating that instrument, they can be overruled at any time by those in the ruling nexus. Keep in mind that .gov statisticians have little recourse to voice their concerns publicly. If and when the ASA leadership speaks out instead, so must that leadership expect to draw criticism, some of it from ASA members. That was bound to happen in the American experience of 2018, but it will also hold in 2028, 2038, ..., regardless of which political nexus controls the Census in those years.
------------------------------
Ralph O'Brien
Professor of Biostatistics (officially retired; still keenly active)
Case Western Reserve University
http://rfuncs.weebly.com/about-ralph-obrien.html------------------------------
Original Message:
Sent: 08-22-2018 13:44
From: Jay Beder
Subject: ASA Comments on 2020 Census Federal Register Notice
In support of the ASA's statement, I would like to add that it is not the ASA that has introduced politics into this question, but rather the Administration. The ASA is protecting a vital data-collection policy from political intrusion. That is a core function of the organization, and I think it is unfair to turn the political accusation against the ASA.
The proposed citizenship question did not arise because a group of nonpartisan constitutional scholars uncovered some basic flaw in our reading of Article I, Section 2. Documents released through a lawsuit reveal the political provenance of the proposal. (Scroll down in the article, Why Was a Citizenship Question Put on the Census? 'Bad Faith,' a Judge Suggests.)
I would also point out that while we don't yet know what effect the inclusion of this question would have on response rates, we are in an era of heightened fear and concern among immigrants, both documented and undocumented. It is certainly reasonable to believe that a non-response effect could be quite large, and so we should stick with the principle of primum non nocere. In any case, we should not use the 2020 census as an experiment to test for this effect.
------------------------------
Jay Beder
Professor
University of Wisconsin-Milwaukee
Original Message:
Sent: 08-21-2018 15:17
From: Dennis Sweitzer
Subject: ASA Comments on 2020 Census Federal Register Notice
Politics aside, there is the issue of consistency.
(1) a major statistical issue is that when one changes a survey (like the census), results will differ from the previous version, and it will take some research to determine if the difference is overt or subtle. So far, indications are that the new question will under count residents.
(2) Citizenship is already asked as part of the American Community survey, which is an annual survey of 1 in 38 households. The fact that it is a random survey of a subset of households gives respondents more assurance that they won't be identified.
It seems obvious that the only reason to change the questions from the 2010 census is for political reason, as it will introduce an inconsistency between the 2020 and 2010 censuses.
--
Dennis E. Sweitzer
><>
?? ><;;>
???? <;;>
???? -
;=;>
???? -;=;o
???? >-|o
?????? >-|O
??
Original Message------
- The addition of a potentially sensitive question, such as the question about citizenship, is a change in instrumentation that has the potential to compromise the validity and integrity of results from the census.
I rather doubt that "change in instrumentation" really motivates this objection.
------------------------------
Gavan Tredoux
Lead Data Scientist, Customer Analytics
Bayer
------------------------------