Sunday, February 9, 2020

The Information Protection Algorithm is Crucial Test for 2020 Census


In the Census Law, Individual Census Records are to be kept Confidential for 72 years. Fearing that Data Brokers using New Statistical Techniques could De-Anonymize the published Population Totals, the Bureau is testing an Algorithm that will Scramble the Final Numbers. Imaginary People will be Added to some Locations and Real People will be Removed from Others.

The more the Algorithm muddles the Results, the more difficult it will be, for a Data Scientist to combine a set of Addresses and Credit Scores with Census Results to learn the Age and Race of People living on a certain Block.

While the Algorithm helps Protect Respondents' Confidentiality, a Test Run on the Last Census shows it may produce wildly Inaccurate Numbers for Rural Areas and Minority Populations. Using the Algorithm on the 2010 Census, a County on the Northern Coast of the Hawaiian Island of Molokai gained 626 Residents. But in Toksook Bay, Alaska, they Lost 50 Residents.

Getting a slice of Tax Revenue may be the Primary concern for many Local Governments. The Decennial Census will shape the Allocation of 31.5 Trillion in Government Spending.

Some Weaknesses in the Privacy Algorithm Implementation are just now coming to light. A recent Analysis by Randell Akee, a Professor at the University of California, Los Angeles, found that on Reservations with Fewer than 5,000 People, the Algorithm Decreased the Population of Native Americans by an average of 34%.

John Abowd, Chief Scientist at the Census, thinks the Risk could Grow, but the Census has a Legal Obligation to use the Algorithm in the 2020 Census. But Government Agencies won't have Access to the Unmodified Census Counts when they Distribute their Money.

Census Official have Exempted State Population Totals from the Algorithm, so State's Congressional Apportionment will remain Accurate as possible.

Dr. Abowd said that the Census plans to Increase Accuracy for the Populations of some Small areas, such as Reservations, and that the Undercount of Native Americans in the Test Run is Unacceptable.

There is still time to Modify the Algorithm, the Bureau has more than a year before it Releases Results to the State's for Redistricting.










NYC Wins When Everyone Can Vote! Michael H. Drucker


No comments: