Wednesday, December 2, 2009

Are There Errors in the Social Security Death Index?

I was searching through the Social Security Death Index (SSDI) this morning and ran across someone born 16 July 1810. And with a reported death date of July 1911.

A little more checking showed the same item in the SSDI on,, and

A little more searching on the database showed:

* 33 persons born in 1800 (check out Margara Morales - she died in 1987!)
* 21 persons born in 1810
* 22 persons born in 1820
* 26 persons born in 1830

There were no entries with a birth date before 1800.

According to the GenealogyBank database, where you can input a birth year range, there are 998 entries for birth dates before 1830, with the earliest in 1800. And there are 1,816 entries for deaths before 1936. Admittedly, these are small errors - 2,814 obviously mis-keyed entries out of about 85 million entries. About 0.0033% - or one in every 30,000 entries.

The errors are probably inevitable. And it looks like some of them have been corrected - there is another entry for Margara Morales with a birth date of 31 June 1898 instead of 1800 (another error there, of course, June has 30 days) with the same death date and location.

Since Social Security numbers were first applied for in the 1937 time frame, almost all of the listings for a birth date before 1830 must be errors. And most of the deaths reported before 1937 must be errors, since the persons should not have had Social Security numbers. What about the 118,950 deaths reported between 1937 and 1950, and maybe even up to 1962? Are they all correct, or are they keying errors also? Undoubtedly, some of them are correct, but many may be erroneous.

There are some cases where the birth and death date are approximately the same and before 1950 - for instance, Samuel N. McNutt is listed as born on 8 October 1950 and died the same day, and he has an SSN issued in Maryland. These are probably not correct - the death date is probably keyed incorrectly as the birth date. A check of newspaper obituaries would probably correct the discrepancy.

The point here is that we researchers cannot rely on the Social Security Death Index as the sole source for a birth date or a death date - we need to find other sources with primary information (if possible). The SSDI is a wonderful finding aid, and leads us to the SS-5 application.

The SS-5 application is often used as a "gateway document" leading us to information about birth date, birthplace, parents names, location and occupation when they applied, etc. The application itself is an "original source" (created by someone with knowledge of their identity) with both "primary information" (the name, residence, location) and "secondary information" (birth date, birthplace, parents names).

Have you obtained the SS-5 application for your family members? If not, you might consider requesting it and perusing the information on it for more leads to earlier generations.

For the record, seems to have the best search field capabilities and permits wild cards, and permit wild card entries and will write a request letter for you, and permits wild cards and the user can create a Footnote Page for each individual for stories, photographs, and additional data.


Sharon said...

Thanks for the article on SSDI Errors. Here's something that might arouse your investigative interest: The newest update of the SSDI on Ancestry list many, many people who died in Oct 2009 with no first names. Just try it: Enter SMITH as a last name and OCT 2009 as death date. See what you get. I tried to check out Footnote (couldn't find place to search SSDI only), Rootsweb (apparently doesn't have Oct 2009 yet) and GenealogyBank (doesn't allow searching by month only). So I don't know if this is an Ancestry glitch, a SSA problem, a change in policy or ????

I've seen a complaint or two on Ancestry, but no response yet.


Sharon said...

Update: I checked Ancestry SSDI. There were 3 Smiths born in 1908 and died in Oct 2009. None had first names. I checked Footnote. I found these same 3 people. All had first names listed. So it seems this is an Ancestry problem.