It's Census Day!
Today is Census Day, revered among U.S. statisticians* as the day for which the census reports a snapshot of the U.S. population. The census questionnaire asks: “How many people were living in this house, apartment, or mobile home on April 1, 2020?” April 1 has been the official Census Day for every census since 1930.
Data from the 2020 census will be essential for dealing with the aftermath of the coronavirus epidemic. I wrote earlier about why census data are crucial for the nation and your community. Many states have a website telling the value of an accurate census count for their state. In Arizona, for example, it’s “estimated that up to $3,000 per person, per year is at stake for every Arizonan…. Just a 1% undercount would represent a loss to the state of $62 million per year for a decade, for a total loss of $620 million.”
But the coronavirus epidemic has disrupted the Census Bureau plans for data collection.** Consequently, getting a high percentage of households to self-respond — that is, fill out the form online, by telephone, or by mail — is even more important in 2020 than it was in 2010.
Figure 1. 2020 Census Self-response Rate as of March 31, 2020 vs. 2010 Census Final Self-response Rate. Data source: 2020census.gov/en/response-rates.html, accessed March 31, 2020.
Figure 1 shows the self-response rate (the percentage of households in the Census Bureau’s address file who have self-responded) for each state as of March 31, 2020, plotted against the final self-response rate for that state in the 2010 census, along with the linear regression line. Note that the 2020 self-response rates, reported as of the end of March, are much lower than the final self-report rates for 2010 because self-response for 2020 is still continuing.
As of March 31, the national self-response rate for the 2020 census was 36.2%,*** but with a lot of variation from state to state. As one would expect, states with high final self-response rates in 2010 also have high self-response rates (so far) in 2020; the correlation between the two rates is 0.81.
FIgure 2. Self-response rates (March 31, 2020) for counties in Arizona.
But there is also a lot of variability among counties and cities within the same state. Figure 2 shows the self-response rates for counties in Arizona, again as of March 31. You can find the data for your county and city at 2020census.gov/en/response-rates.html. Maricopa County (Phoenix), Yavapai County (Prescott), and Pima County (Tucson) have relatively high self-response rates, but several counties (including the counties with large Native American populations) have very low self-response, and the planned community events to increase response in those areas have been postponed or canceled. Counties that may have great needs for resources during the 2020s are at risk of being undercounted.
But you can help improve the self-response rate. Here’s how:
Fill out your own census form, if you haven’t already done so. You can respond online, by telephone, or by mail.
Persuade two households who haven’t yet filled out the form to take part in the census. (Note: the earlier you start, the easier it will be to find two households that have not yet participated.)
Ask each of those households to recruit two more households to fill out their census forms.
Voila! Exponential growth in census self-response! If no one breaks the chain, the process will end only when everyone has participated.
When encouraging someone to participate, I was asked, “If I respond, won’t that hurt people in other communities, because my area will have more people responding and will get more resources?”
That’s an excellent question. But, no, your self-response actually helps all communities. The Census Bureau tries to get as many households to respond via internet, telephone, and mail as possible. This reduces the costs of obtaining responses from those households, and results in more accurate data about them.
People who don’t self-respond are still counted in the census, but the Census Bureau has to work much harder to count them. After the initial self-response period, the Census Bureau sends enumerators out to households that haven’t responded. In past censuses (censi?), the enumerators would visit the residence in person and ask the census questions. That’s the method listed in the 2020 census plan as well, although at present field operations for the census have been suspended until at least April 15.
Your self-response means that no one needs to follow up with you to get your data. When field operations resume, census workers (who for the most part are people in your community hired specifically for the 2020 census) do not have to spend time on you and can devote their efforts to households that find it more difficult to self-respond. It’s similar to putting on your own oxygen mask during a flight emergency — yes, you’re helping your own breathing, but you’re also allowing the flight attendants to spend their time assisting other passengers instead of needing to help you.
Of course, some households don’t provide data when approached by a census worker, either, and, according to Salvo, Jacoby, and Lobo (2020), census data for these households may be obtained from neighbors, landlords, administrative records, or other sources. But data obtained from these other sources are likely to be less accurate than the information that would have been obtained from the household itself. After all, your neighbor or landlord does not know as much about your household as you do. In addition, areas with low self-response rates are more likely to be undercounted.
So while you’re self-isolating to flatten the curve for COVID-19, or your state is on pause, why not fill out the census form and encourage those in (and out of) your network to do the same? It will take ten minutes or less, but will help your community, and the country, for the decade to come.
Footnotes and References
Anderson, M. J. (2015). The American Census: A Social History, 2nd edition. New Haven, CT: Yale University Press.
Salvo, J. J., Jacoby, A., and Lobo, A. P. (2020, February). Census 2020: Why increasing self-response is key to a good count. Significance, 17(1), pp. 30-33.
*Statistical data collection is the subject of the fifth paragraph of the U.S. Constitution. The Constitution does not give any other scientific discipline the same prominence.
**Other crucial government surveys, such as the Current Population Survey (used to measure unemployment) and the National Crime Victimization Survey, are also being affected. As of March 20, the Census Bureau has suspended all in-person interviews to collect data for surveys, and is attempting to conduct all interviews by telephone. But these surveys use face-to-face interviewing in normal operations because some households are difficult or impossible to reach by telephone; the in-person data collection gives a higher response rate and better representation of the United States. This change is likely to have serious consequences for the statistics from these surveys.
In the National Crime Victimization Survey, persons interviewed in person report approximately twice as many violent crime victimizations as persons interviewed by telephone. Thus, if it turns out that the estimated number of violent victimizations has decreased in 2020, is that because they really decreased or because the survey missed the crime victims who would only have participated in an in-person interview? It may be possible to weight the data to partially compensate for the lack of in-person interviews, but relationships among population subsets that hold during the rest of the year may well differ during the time period where in-person interviews are suspended, so that weighting adjustments may not entirely fix the problem.
***The national self-response rate is 38.4% as of 4 pm EDT on April 1.