As I mentioned in an earlier post, “Demographic Analysis” And The Census, one method of measuring the size of the US population relies on historical birth registration, death registration, as well as estimates of in-migration to and out-migration from the United States. Census Bureau demographers have completed the assembly of national estimates for the April 1, 2010 population of the United States, and we will release them on December 6, 2010. Since demographic analysis produces only national estimates, it cannot be used for the reapportionment and redistricting purposes required of the 2010 Census. It is, however, a useful comparison to the Census.

Following advice of demographers from around the country and our own internal experts, we have developed multiple estimates – ones that make different assumptions about the components of international migration and different assumptions about completeness of vital statistics records. Estimates of international migration vary depending on the source of the data and the judgments of the demographer constructing them. All five of the demographic analysis estimates are considered plausible and no one estimate is clearly the best.

For each age, sex, and race group, we will present five different estimates, which each result from a different set of plausible assumptions.

It is unusual to show multiple estimates in this manner, but it is the honest way to portray our uncertainty about the demographic analysis estimates.

We will not be able to answer the question, “Which estimate is best?” Each is based on different assumptions, and we cannot eliminate any of those assumptions. We think that reports of the demographic analysis should show a range of possibilities and not one single number.

However, the presentation of five different estimates complicates the comparison of the national demographic analysis estimates to the single official national 2010 Census count for the same group. Even so, we will make those comparisons, attempting to identify patterns of differences. When the patterns of differences can be connected with other information (either from field results of where difficulties arose in data collection or ethnographic studies of how different kinds of households reacted to the census request), we will then have information that we will be able to act on – to bring about improvements for a more effective enumeration for 2020 and improved demographic analysis in the future. Throughout the decades using this process, the decennial census has become more accurate and demographic analysis more sophisticated.