"Disappearing" Smiths

The desis are baffled:

Statistical Modeling, Causal Inference, and Social Science asks where asks where all the Smiths have gone:

Sam Roberts writes,

In 1984, according to the Social Security Administration, nearly 3.4 million Smiths lived in the United States. In 1990, the census counted 2.5 million. By 2000, the Smith population had declined to fewer than 2.4 million.

Where did all the Smiths go from 1984 to 1990? I can believe it flatlined after 1990, but it's hard to believe that the count could have changed so much in 6 years.

Perhaps it's the difference between the SSA and Census methods of counting

Here's another explanation, it's the inverse of the phenomenon of those claiming Native American ancestry in the United States doubling in 10 years. Many Smiths were at one point Schmidts, who knows if some of them didn't revert now that WASP surnames aren't as value-added?

Naturally, Newamul Khan is hilariously off base. The Columbia blogger was on the right track. Nathaniel Weyl discusses the SSA data in The Geography of American Achievement:

. . . a little known publication of the Department of Health and Human Services entitled Report of Distribution of Surnames in the Social Security Number File September 1, 1984. This covers all surnames of persons who have been enrolled with Social Security from the inception of the program in 1936 to the date of the report. Total coverage is 346,417,726, or considerably more than the total 1984 poulation of the United States, which was approximately 237 million. The excess is due primarily to deaths and name-changes caused by marriage. [p. 9]

So, there were approximately 2.33 million Smiths living in the US in 1984 [(3.4/346.4)*237], in line with the 2.5 million Smiths in 1990. No mystery exists. That's not to say we don't have problems. From the original NYT story:

The number of Hispanics living in the United States grew by 58 percent in the 1990s to nearly 13 percent of the total population, and cracking the list of top 10 names suggests just how pervasively the Latino migration has permeated everyday American culture.

Garcia moved to No. 8 in 2000, up from No. 18, and Rodriguez jumped to No. 9 from 22nd place. The number of Hispanic surnames among the top 25 doubled, to 6.

Compiling the rankings is a cumbersome task, in part because of confidentiality and accuracy issues, according to the Census Bureau, and it is only the second time it has prepared such a list. While the historical record is sketchy, several demographers said it was probably the first time that any non-Anglo name was among the 10 most common in the nation. “It’s difficult to say, but it’s probably likely,” said Robert A. Kominski, assistant chief of social characteristics for the census.

Luis Padilla, 48, a banker who has lived in Miami since he arrived from Colombia 14 years ago, greeted the ascendance of Hispanic surnames enthusiastically.

“It shows we’re getting stronger,” Mr. Padilla said. “If there’s that many of us to outnumber the Anglo names, it’s a great thing.”