Written by: Director Robert Groves
There are two principal ways that we try to learn about the size of the US population. One, the decennial census, is the focus of this blog. The other method, called “demographic analysis,” uses the “vital registration” system of the US — birth registrations, death registrations – as well as estimates of immigration and emigration.
These four sources of demographic information, compiled from historical data and supplemented with results from earlier censuses, provide another look at the size of the population. All four of the data sources allow separate estimates of males and females, different age groups, and, traditionally, different racial groups.
In the past, demographic analysis has been used as one tool to evaluate the census. Past census publications present estimates of “net undercount” for age, gender, and race groups, under the assumption that differences between the demographic analysis and the census reflected census coverage weaknesses.
Through their ongoing research, Census Bureau demographers have found, however, some weaknesses with demographic analysis to evaluate a census. Two seem most important. First, the undocumented immigrant population is not included in the record systems, thus becoming subject to various survey-based, indirect sources of estimates. Second, the measurement of racial groups on the records has deviated from that of the US decennial censuses, making estimation of individual racial groups different from what a perfect census might obtain with its racial measurements.
A great conference at Census Bureau headquarters a few days ago gathered experts to give advice about how the 2010 demographic analysis results should be presented. Based on this workshop the Census Bureau is assembling multiple estimates of population counts by age, gender, and race, to reflect the real uncertainties about the current status. Thus, instead of one population count based on demographic analysis, in December, 2010, we’ll present several – an honest statement of what we know and what we don’t know. We also won’t refer to the differences between the demographic analysis estimates and the census counts as the net undercount of the census. Instead, it’s best to view demographic analysis as another way to estimate population sizes, with its own set of strengths and weaknesses.
Bertrand Russell, in commenting on how science progresses, once noted that the more scientists know, the more they also know what they don’t know. By giving the country multiple estimates, we’ll be reflecting this higher state of understanding the difficulties of demographic analysis.