Worried that the introduction of a citizenship question to the 2020 Census "without appropriate testing introduces unknown accuracy risks due to the potential that it will deter legal or illegal immigrants from responding," the Insights Association, the leading nonprofit association representing the marketing research and data analytics industry, urged that the question be removed.

"To ensure accuracy, the census requires the highest possible representation of our population. Every subsequent survey and study that intends to be statistically representative of the U.S. population will be built on decennial data, including the American Community Survey (ACS), the Economic Census and every other federal government survey, and any inaccuracies will be felt for at least a decade."

The Insights Association filed comments with the Census Bureau today, flagging the threat to accuracy posed by the citizenship question (which the Department of Commerce announced at the end of March), as well as the increased costs that would result.

It is estimated that more than 100,000 comments have been submitted so far, and the tally should be significantly higher by the deadline this evening. The association recently urged the industry to submit their own concerns.

Read the Insights Association's full comments in PDF or below:

Thank you for the opportunity to comment on the finalization of 2020 Census operations and questionnaire, on behalf of the Insights Association[1] and the marketing research and data analytics industry.

Accurate Census data underpin every statistically representative survey and study in the U.S. and informs decision-making in both the public and private sectors, so we appreciate moves on the part of the Census Bureau and the Department of Commerce to modernize the decennial headcount for the 21st Century.

However, we are severely concerned by the addition of a citizenship question to the 2020 Census questionnaire, announced by the Department of Commerce on March 26, 2018. We urge that the citizenship question be removed from the decennial.

The problem with a citizenship question

Adding a citizenship question to the decennial census without appropriate testing[2] introduces unknown accuracy risks due to the potential that it will deter legal or illegal immigrants from responding. Without testing and with fewer respondents, the decennial headcount likely will be less accurate, less valuable and unnecessarily expensive.

To ensure accuracy, the census requires the highest possible representation of our population. Every subsequent survey and study that intends to be statistically representative of the U.S. population will be built on decennial data, including the American Community Survey (ACS), the Economic Census and every other federal government survey, and any inaccuracies will be felt for at least a decade.

Marketing researchers know the expense of respondent cooperation better than most, as well as the potential downside of making any survey longer. The most expensive part of the 2020 Census, according to the Bureau, will be door-to-door non-response follow-up. For the 2010 census, it cost $90 million more for every 1 percent of households that didn't respond up front. The budgetary impact of adding a citizenship question could be severe and require significant emergency appropriations.

Warnings from the Census Bureau

While none of the lawsuits proceeding against the question's addition have been decided, the discovery process has revealed interesting background to how the decision was made. Some of those documents from within the Census Bureau indicated that the citizenship question addition could depress response rates, increase costs and hurt accuracy,[3] just like we explained at the time of the announcement.[4]

Even before those legally-required disclosures, Census Bureau researchers in 2017 had publicly warned about a recent “increase in respondents” who “spontaneously” express their “concerns to researchers and field staff about confidentiality and data access relating to immigration.”[5] The researchers also observed “increased rates of unusual respondent behaviors during pretesting and production surveys (data falsification, item non-response, break-offs).”

Conclusion

Six former Census Bureau Directors (Republicans and Democrats), in a letter to Commerce Secretary Wilbur Ross on January 26, 2018,[6] warned that “adding an untested question on citizenship status at this late point in the decennial planning process would put the accuracy of the enumeration and success of the census in all communities at grave risk.” Further, the former directors cite “a great deal of evidence that even small changes in survey question order, wording, and instructions can have significant, and often unexpected, consequences for the rate, quality, and truthfulness of response. The effect of adding a citizenship question to the 2020 Census on data quality and census accuracy, therefore, is completely unknown. Also of import, overcoming unexpected obstacles that arise as 2020 Census operations unfold would add to the cost, without assurances that such efforts would yield a more accurate outcome.”

As explained above, we share those worries, and urge the Department of Commerce and Census Bureau to rescind this decision and remove the citizenship question from the 2020 Census questionnaire. That would be in the best interests of an accurate, complete and efficient decennial census.

 

[1] The Insights Association was formed through the merger of two trade organizations with long, respected histories of servicing the market research and data analytics industry: the Marketing Research Association (MRA), founded in 1957, and the Council of American Survey Research Organizations (CASRO), founded in 1975. Representing more than 4,000 members across the United States, the Insights Association is the leading nonprofit trade association for the market research and analytics industry, and the leader in establishing industry best practices and enforcing professional standards. The Insights Association’s membership includes both research and analytics companies and organizations, as well as the researchers and research departments inside of non-research companies and organizations.

[2] According to the Congressional Research Service, “The 1950 census was the last to date that collected citizenship data from the whole population, rather than a population sample.” After that, the “citizenship and/or related questions” were on the old census “long form” and now on the rolling annual American Community Survey (ACS) that replaced the long form.

[3] "Census Bureau’s chief scientist warned secretary in memo against adding citizenship question." By Jacqueline Thomsen. The Hill. June 10, 2018. http://thehill.com/homenews/administration/391527-census-bureau-official-warned-in-memo-against-adding-citizenship

[4] "Citizenship Question on 2020 Census Could Harm Accuracy and Costs." March 27, 2018. https://www.insightsassociation.org/article/citizenship-question-2020-census-could-harm-accuracy-and-costs