Readme April, 2014 SOCIAL SECURITY DEATH MASTER FILE NICAR Record count: 91,465,830 Data covers: 1937 - Nov 30, 2013 The Social Security Death Master File (DMF), obtained from the National Technical Information Service (NTIS), is the most complete listing of all deaths in the United States since 1937 -- although it's not entirely complete and still has errors. To be listed, the person must have a Social Security number. It does *not* include any details about the death of a person. It does list the name, Social Security number, date of birth, date of death (in some cases, only month and year), the state (for deaths between 1937-February 1988), and ZIP codes of last known residence and location where lump-sum payment made to widow(er) are sent (for deaths up to Feb 2010). Important information: The data NICAR purchased from NTIS had 88,254,202 records, with no state or zip information (see the next paragraph for more information on that). In order to give our users the most information possible, we added back in 1) state and zip information where we had it, and b) records that the SSA had excluded from the latest version. These records have been flagged (see the Record Layout.txt). In 2011 The Social Security Administration changed what information it releases. It no longer releases information on where the person was when they died, such as state or zip, or where the last payment was sent (the last three columns). The SSA had wiped those columns entirely. For more information, see the "Change to DMF.pdf" file included with the data. In 2013 Congress passed a law as part of the Budget Act of 2013 that limited the most recent years of the data to licensed entities only. NICAR purchased a copy of the full database before the law went into effect. The Social Security Administration goes out of its way to emphasize that it does not guarantee the accuracy of the Death Master File. For good reason -- journalists have identified numerous people who, according to the database, are dead but are in fact alive and well. Often, it's a matter of the SSA mistakenly identifying someone as dead when in fact the person's spouse died. So it's important to verify that someone is in fact deceased before stating so in an article. What you can do with the data: Journalists from across the United States have done stories about election system weaknesses by matching this database with the voter registration data in their area. Some have identified actual cases of people who are recorded as casting a ballot after their death. Other journalists have documented problems with local voter registration lists that contain people who are long dead, allowing possible voter fraud and making counts of registered voters inaccurate. If you've ordered the entire U.S., you should receive ten CSV files that are COMMA-DELIMITED with " as a text qualifier: DMF1.CSV DMF2.CSV DMF3.CSV DMF4.CSV DMF5.CSV DMF6.CSV DMF7.CSV DMF8.CSV DMF9.CSV DMF10.CSV Each contains 10,000,000 records, with the exception of DMF10, which contains 1,465,830. Other files included: Change to DMF.pdf -- A document from NTIS explaining the changes to what information is released in the Death Master File starting November 1, 2011 RecordLayout.txt -- The NICAR-created record layout LEGAL.txt -- Terms and Conditions Governing NICAR's Data Services and Agreement Concerning Sale of Data/Databases to Journalism Organizations and/or Individual Journalists Mandatory Requirements and Record Layout as of Nov. 2011.pdf -- The NTIS-provided record layout; note that the first field was always empty in the data sold to NICAR, so we eliminated it. The last three fields are no longer released Record Layout before Nov 2011 change.pdf -- The NTIS-provided record layout, which included the last three fields GAO_d05478.pdf -- A GAO report about: "Additional Data Could Help State and Local Elections Officials Maintain Accurate Voter Registration Lists." It mentions the Death Master File specifically. If you are working in Microsoft Access, here's how to open the CSV files: 1. Copy the tables to your hard drive. 2. Open a blank database in Access, name it and save it. 3. Inside the new database, go to the "external data" section and choose "From text". 4. A wizard will ask you to locate the file(s). Choose "link" rather than "import". The files are comma-delimited with a " text qualifier. 5. Each file will need to be linked separately. ****************************************************** STORY EXAMPLES, AVAILABLE FROM THE IRE RESOURCE CENTER ****************************************************** Story No. 24546: In 2009 WFOR-TV (Miami) reported that some farmers who have died during the last 20 years are still being paid. By comparing the "Farm Bill database" and the "Social Security Death Index," WFOR-TV found 234 deceased farmers in South Florida continue to receive taxpayer money through the U.S. Farm Bill. The amount still being paid is estimated to be $9.5 million. Story No. 22767: In 2006, a Poughkeepsie Journal investigation found that New York's new statewide database of registered voters contains as many as 77,000 dead people on its rolls. As many as 2,600 of them have cast votes from the grave, according to a Poughkeepsie Journal computer-assisted analysis. The analysis was examined the potential for errors and fraud in New York's three-month-old database. No fraud was found however errors had been made. Story No. 18192: The Atlanta Journal Constitution looked at the existing and the potential scope of voting errors and fraud in Georgia. The investigation finds that "the actual number of ballots cast by the dead is fairly small - 5,412 in the past 20 years," but "the ranks of potential dead voters have grown dramatically in recent years." The newspaper analyzed the most common errors in voting records and the loopholes in some election laws. The report also describes the voter identification requirements. Story No. 21805: Times of Northwest Indiana (Munster, Ind.) used voting records and Social Security death records to find that deceased residents voted in the primary election in 2003. It was also found that thousands of dead people remain as registered voters on the voter rolls. State and local officials say this can create an "opportunity for ghostly corruption." The Resource Center also offers a NICAR newsletter Uplink October 1994 article by Paul D'Ambrosio of the Asbury Park (N.J.) Press about how he pursued a story about matching the data with voter registration in New Jersey. You can search for more stories and tipsheets on the Web at www.ire.org/resourcecenter. You can order stories from the Resource Center at 573-882-3364 or by e-mailing the IRE Resource Center at rescntr@ire.org. If you have any questions about how we processed the data, you can contact the IRE and NICAR Database Library at 573-884-7711. ***************************************************************** ADDITIONAL IMPORTANT NOTES FROM THE SOCIAL SECURITY ADMINISTRATION ***************************************************************** While NICAR is not a "subscriber" of the Death Master File (we purchase what's called a "single issue" copy and update it annually at most) and therefore not covered within the rules about "subscribers," here are some excerpts of notes to subscribers from the NTIS Web site with some additional cautionary notes: As a result of a court case under the Freedom of Information Act, SSA is required to release its death information to the public. ... DMF does have inaccuracies and SSA does not guarantee the accuracy of the DMF. SSA does not have a death record for all deceased persons. Therefore, the absence of a particular person on this file is not proof that the individual is alive. In addition, there is the possibility that incorrect records of death may have been entered on the DMF. If an individual seeing your copy of the DMF has a complaint that they find erroneous data/death information on that DMF, you should advise them to follow the procedures listed below. In fact, you should be providing the information below in your publication, if any, of the DMF: Errors: If an individual claims that SSA has incorrectly listed someone as deceased (or has incorrect dates/data) on the Death Master File (DMF), the individual should contact his/her local Social Security office (with proof) to have the error corrected. That local Social Security office will: (1) make the correction to the main NUMIDENT file at SSA and give the individual a verification document of SSA's current records to use to show to any company recipient/purchaser of the DMF that had the error, or (2) find that SSA already has the correct information on the main NUMIDENT file and DMF (probably corrected sometime prior) and give the individual a verification document of SSA's records to use to show to any company subscriber/purchaser of the DMF that had the error. ... ***************************** ABOUT SOCIAL SECURITY NUMBERS ***************************** As of June 25, 2011 Social Security Numbers are assigned randomly, and do not contain geographical significance. Below are explanations of the change, as well as the way the numbers used to be assigned SSN RANDOMIZATION (From the SSA website): The Social Security Administration (SSA) is changing the way Social Security Numbers (SSNs) are issued. This change is referred to as "randomization." The SSA is developing this new method to help protect the integrity of the SSN. SSN Randomization will also extend the longevity of the nine-digit SSN nationwide. The SSA began assigning the nine-digit SSN in 1936 for the purpose of tracking workers' earnings over the course of their lifetimes to pay benefits. Since its inception, the SSN has always been comprised of the three-digit area number, followed by the two-digit group number, and ending with the four-digit serial number. Since 1972, the SSA has issued Social Security cards centrally and the area number reflects the state, as determined by the ZIP code in the mailing address of the application. There are approximately 420 million numbers available for assignment. However, the current SSN assignment process limits the number of SSNs that are available for issuance to individuals by each state. Changing the assignment methodology will extend the longevity of the nine digit SSN in all states. On July 3, 2007, the SSA published its intent to randomize the nine-digit SSN in the Federal Register Notice, Protecting the Integrity of Social Security Numbers [Docket No. SSA 2007-0046]. SSN randomization will affect the SSN assignment process in the following ways: -- It will eliminate the geographical significance of the first three digits of the SSN, currently referred to as the area number, by no longer allocating the area numbers for assignment to individuals in specific states. -- It will eliminate the significance of the highest group number and, as a result, the High Group List will be frozen in time and can be used for validation of SSNs issued prior to the randomization implementation date. -- Previously unassigned area numbers will be introduced for assignment excluding area numbers 000, 666 and 900-999. -- These changes to the SSN may require systems and/or business process updates to accommodate SSN randomization. If you have any questions regarding SSN randomization or its possible effects to your organization, please see the related Frequently Asked Questions or email your question(s) to ssn.randomization@ssa.gov. THE PREVIOUS ASSIGNMENT METHOD (also from the SSA website): The nine-digit SSN is composed of three parts: * The first set of three digits is called the Area Number * The second set of two digits is called the Group Number * The final set of four digits is the Serial Number Area Number The Area Number is assigned by the geographical region. Prior to 1972, cards were issued in local Social Security offices around the country and the Area Number represented the State in which the card was issued. This did not necessarily have to be the State where the applicant lived, since a person could apply for their card in any Social Security office. Since 1972, when SSA began assigning SSNs and issuing cards centrally from Baltimore, the area number assigned has been based on the ZIP code in the mailing address provided on the application for the original Social Security card. The applicant's mailing address does not have to be the same as their place of residence. Thus, the Area Number does not necessarily represent the State of residence of the applicant, either prior to 1972 or since. Generally, numbers were assigned beginning in the northeast and moving westward. So people on the east coast have the lowest numbers and those on the west coast have the highest numbers. Note: One should not make too much of the "geographical code." It is not meant to be any kind of useable geographical information. The numbering scheme was designed in 1936 (before computers) to make it easier for SSA to store the applications in our files in Baltimore since the files were organized by regions as well as alphabetically. It was really just a bookkeeping device for our own internal use and was never intended to be anything more than that. Group Number Within each area, the group number (middle two (2) digits) range from 01 to 99 but are not assigned in consecutive order. For administrative reasons, group numbers issued first consist of the ODD numbers from 01 through 09 and then EVEN numbers from 10 through 98, within each area number allocated to a State. After all numbers in group 98 of a particular area have been issued, the EVEN Groups 02 through 08 are used, followed by ODD Groups 11 through 99. Group numbers are assigned as follows: ODD - 01, 03, 05, 07, 09------EVEN - 10 to 98 EVEN - 02, 04, 06, 08------ODD - 11 to 99 See the latest Social Security Number Monthly Issuance Table [NICAR note: http://www.ssa.gov/employer/highgroup.txt] for the latest SSN area ranges issued to date. Alleged Social Security numbers containing area numbers other than those found on that table are impossible. Serial Number Within each group, the serial numbers (last four (4) digits) run consecutively from 0001 through 9999.