While visiting London in 1897, Mark Twain learned that rumors of his death were circulating in the American press. In a telegram home, he termed the reports of his demise as "exaggerated." It is unlikely, though, that Twain would have been as amused by the premature inclusion of his death in the Social Security Administration's Death Master File (DMF).

Since the 1930s, deaths of social security number holders have been reported to the Social Security Administration (SSA). These reports now are compiled in the DMF. This database's principal function is for the termination of benefits. Because the DMF is publicly available, it has numerous secondary uses. Pension funds, financial institutions, and law enforcement agencies rely on it to detect fraud. Genealogists and even epidemiologists use it to advance their research.

Erroneous DMF death reports can have very serious consequences for the "deceased." Not only could they lose benefits, they also might become targets of investigations for potential fraud or other criminal activity.

The SSA urges DMF subscribers "not to take any adverse action against any individual without further investigation to verify any death listed." At the very least, however, erroneously listed individuals are exposed to potential identity theft by the public disclosure of their full name, date of birth, and social security number. The SSA recognizes these problems and has detailed procedures for deleting records from the DMF when it determines persons are incorrectly included. On-going DMF subscribers generally receive regular updates, including listings of records to be deleted.

According to our study of four quarterly DMF updates (from October, 2004, through July, 2005), additions for newly reported deceased persons outnumber the deletions for persons erroneously included by about 250 to 1. Nearly 2,000,000 records were added, compared to slightly fewer than 8,000 deletions.

The deletions are not randomly distributed. This indicates some processes involving these errors could be systematic. For example, children 10-14 years of age had the highest error rates (after discounting persons reportedly more than 100 years old). Figure 1 shows the number of record deletions per 1,000 DMF updates involving additions or deletions. Keep in mind that the deletion rates for children of this age have the smallest denominators, which would tend to make the rates for this group somewhat unstable. The same pattern, though, was roughly repeated in every quarter we studied.

Figure 1. DMF deletions per 1,000 updates by age.

The DMF includes the last zip code where benefits were sent. Geocoding this information allowed us to analyze the deletion rates geographically. Figure 2 shows the distribution of the deletion rates arranged by state from highest to lowest. Alaska, Georgia, Rhode Island, and California are outliers (based on the interquartile range). Alaska accounts for a small number of updates and would have correspondingly unstable rates. However, Alaska ranked first for deletion rates in each of the four quarters studied.

Figure 2. DMF deletions per 1,000 updates by state.

Analyzing DMF updates by specific zip codes also might identify systematic causes for deletions. We created Figure 3 by plotting a point for every zip code reporting an update. This established the base map. We then used a method to rank order zip codes according to the actual number of deletions compared to the number expected based on the average rate for all zip codes to spot the outliers. To stabilize these rankings, we also weighted them by the total number of updates for each zip code. Finally, we color coded the zip codes that ranked highest and superimposed them on the base map to look for possible error clusters.

Figure 3. Outlier zip codes.

This map shows a cluster of zip codes of potential interest in and near Los Angeles, California: 90740 (Seal Beach), 91423 (Sherman Oaks), 90035 (Los Angeles), 90077 (Los Angeles), 91767 (Pomona), 91304 (Canoga Park), 92548 (Homeland), 90293 (Playa del Rey), 92337 (Fontana), and 92647 (Huntington Beach).

It might never be possible to completely eliminate erroneous reports of deaths in the Death Master File. The SSA faces difficult challenges from a large population of people who move, change names, change the spelling of their names, or misreport their dates of birth. Still, the consequences of DMF errors have become increasingly serious. Improving the quality of data in the file should have a greater priority than in the past.

Email inquiries or questions:

Link to:

QCS Corp. Home Page

Detecting Potential Fraud with SecureID™

Simulation Studies of Social Security Number Validation

How to Contact Us

Copyright © Quality Control Systems Corp. All rights reserved.