Skip Navigation
Resource

Digitizing the Census

The 2020 Census will be the first to be completed largely online — a process that comes with both opportunities and risks.

  • Alexis Farmer
Published: September 10, 2019

The 2020 Census will be the first to be completed largely online – if the Census Bureau’s plan goes off without complic­a­tions. An online census is one of several tech­no­lo­gical innov­a­tions that the Census Bureau has designed to respond to the chal­lenges of count­ing an increas­ingly large and diverse soci­ety, while also comply­ing with strict cost constraints that Congress has imposed. 

The Bureau developed these tech­no­lo­gical innov­a­tions – which also include redesign­ing its address canvassing process and incor­por­at­ing admin­is­trat­ive records as sources of inform­a­tion on house­holds – to help save it an estim­ated $5.2 billion on the upcom­ing census. Nonre­sponse follow up – the process through which the Bureau captures responses from house­holds that haven’t submit­ted their census forms them­selves – is the Bureau’s largest and most costly field oper­a­tion. It is also a signi­fic­ant contrib­utor to the census’ escal­at­ing costs. Each decade since 1970, the Bureau has had to invest more resources to improve initial responses to census forms. After running the cost­li­est census ever in 2010, the Bureau has decided to scale back door-to-door canvassing and follow-ups and, instead, rely more heav­ily on new tech­no­logy to count every­one.

Under the Bureau’s plans for 2020’s online census, 80 percent of house­holds will receive an invit­a­tion to submit their responses over the inter­net. The Bureau estim­ates that 45 percent of those house­holds will respond to the census online. The Bureau will mail paper ques­tion­naires to the remain­ing 20 percent of house­holds, target­ing those with low inter­net access or large older-adult popu­la­tions. Ques­tion­naires will also be mailed to those house­holds that do not respond online in the first instance. House­holds that receive paper ques­tion­naires will still be invited to respond to the survey online, but will have the option to submit their answers by mail. Each house­hold will also have the abil­ity to report its answers by phone. If house­holds still do not respond, the Bureau will send census field work­ers, known as enumer­at­ors, door-to-door to collect their data using mobile devices and tablets.

Although moving to a digital plat­form has its advant­ages, it also has its risks. The Census Bureau must address key design elements to ensure that it and others are not suscept­ible to cyber-attacks. Carol Cha Harris, the Director for Inform­a­tion Tech­no­logy Acquis­i­tion Manage­ment Issues at the U. S. Govern­ment Account­ab­il­ity Office, stated in testi­mony before the House Commit­tee on Over­sight and Govern­ment Reform that, given the volume of personal inform­a­tion the Census receives, “it will be import­ant for the Bureau to ensure that only respond­ents and Bureau offi­cials are able to gain access to this inform­a­tion.” In Janu­ary, the Bureau repor­ted that 24 out of 44 systems needed for the 2018 End-to-End Test – the only process approach­ing a dry-run for the census before 2020 – are ready for use. Several crit­ical IT systems, includ­ing cyber­se­cur­ity meth­ods, will ideally be tested in the lead up to the 2020 Census to ensure that they are func­tional.

In addi­tion to these cyber­se­cur­ity chal­lenges, the move to an online system faces another hurdle that the Census Bureau must over­come: the diffi­culties many tradi­tion­ally under­coun­ted communit­ies face access­ing the inter­net. Racial and ethnic minor­it­ies, urban and rural low-income house­holds, immig­rants, and young chil­dren have been histor­ic­ally under­coun­ted at dispro­por­tion­ately high rates. Trans­ition­ing to an online plat­form could lead them to being under­coun­ted even more severely. Some rural areas lack broad­band or any inter­net service. People with lower incomes are less likely to have a smart phone or inter­net at home.

In addi­tion to encour­aging people to respond online, the Bureau is also consid­er­ing using admin­is­trat­ive records – data that people have already given to the federal govern­ment for purposes like tax returns, govern­ment assist­ance programs, and the like – to reduce the need for door-to-door visits to house­holds and help improve the qual­ity of the data that it collects. The Bureau estim­ated that using admin­is­trat­ive records in these ways could save it $900 million

But previ­ous results show dispar­it­ies in those whom admin­is­trat­ive records bene­fit and hurt. A 2016 report invest­ig­at­ing the impact of admin­is­trat­ive records and the inter­net on hard-to-count popu­la­tions revealed the diffi­culty of match­ing admin­is­trat­ive records to people who belong to racial and ethnic minor­ity groups, increas­ing the risk of these groups being under­coun­ted.

Over­all, new tech­no­lo­gical devel­op­ments present both oppor­tun­it­ies and risks to the Census Bureau. How it navig­ates them over the next two years will have a signi­fic­ant effect on the ulti­mate fair­ness and accur­acy of the 2020 Census.