Source: CENTER FOR IMMIGRATION STUDIES, 9/21/18

A new article published in the academic journal PLOS One argues that the size of the illegal immigrant population in the United States has been grossly underestimated by prior research. The authors make assumptions about cumulative inflows and outflows since 1990 and arrive at a "conservative" estimate of 16.7 million illegal immigrants in 2016 and an average estimate of 22.1 million — 50 to 100 percent larger than other estimates.

The findings are unsupportable. Accepting that there are 22 million illegal immigrants in the U.S. also requires accepting that every Census Bureau survey missed huge numbers of people and that most administrative data from the federal government is woefully incomplete. There is no body of research that corroborates such a claim.

Most prior research subtracts the legal immigrant population, based on administrative data, from the total foreign-born identified in Census surveys. The difference represents the number of illegal immigrants, which is then adjusted upward to reflect undercount. Using this method, most researchers find that 10 to 11 million illegal immigrants are included in Census Bureau surveys, with perhaps one million missed based on other research. But if the new study is correct, then the Census Bureau is missing another 11 million illegal immigrants, something that would have obvious impacts in other datasets and in the real world.

"No matter how carefully a theoretical model may seem to be calibrated, it is not useful unless it is consistent with the real world," said Steven Camarota, the Center's Director of Research. "It is incumbent upon the authors to explain how their estimate can be reconciled with other data. It seems extremely unlikely that the Census Bureau, the Department of Education, and other records of vital statistics miss so many people year after year."

Specific examples:
• When the Census Bureau recanvassed a subset of the population after the 2000 and 2010 censuses, they found a miss rate of roughly 1 percent. If the new study's illegal estimates are correct, then we would expect to see an undercount two or three times larger than what the Bureau estimated.
• About 4 million women in the American Community Survey (ACS) each year reported they had a child in the prior year. This nearly matches the number of births reported by the National Center for Health Statistics (NSHC) based on birth certificates, which suggests the ACS is not missing large numbers of mothers. If there were 11 million more illegal immigrants in the country, then NSHC data would need to show roughly 300,000 more births, which it does not. 
• Looking at births specifically to immigrants in a forthcoming report, the Center compared ACS and NSHC data in detail and found that the ACS may understate births to immigrant by just 5 percent. We would expect to see a far greater undercount if the ACS missed 11 million illegal immigrants. 
• In 2014 the count of public school students in the ACS differed from Department of Education figures by only 3 percent, which is partly explained by the differences in data collection rather than undercount. But, if there were 11 million illegal immigrants missed by the ACS, roughly 1 one million would have to be enrolled in school, implying a large difference between the ACS and DOE figures that does not exist. 

On the issue of counting illegal immigrants, the Center for Immigration Studies agrees with the Migration Policy Institute's critique of the new paper: "People leave footprints" and millions of illegal immigrants cannot simply be "hidden" from demographic profiles.