The Census Bureau plans to release the state population counts used to apportion Congress by April 30.  In August, it will release the first wave of redistricting data, which includes population counts and racial/ethnic distributions for all Census geographies — states, cities, counties, towns, villages, voting districts, tracts, block groups, and blocks.  Data will be distorted to ensure privacy, but the process is likely to make the data much less useful for many purposes, including redistricting. 

Test versions of the new methods applied to the 2010 data showed that serious inaccuracies were introduced, making the data much less useful and suspect for even the simple task of reporting the population and racial and ethnic distributions for small geographies.  In early March, Alabama sued to block the distortion. It was joined last week by 16 states.  A court hearing is scheduled in two weeks, and the outcome may be immediately appealed to the U.S. Supreme Court.

Since the plans to distort the data were announced in spring of 2019 (and especially since a test version based upon the 2010 Census was released in October 2019), many users have raised serious questions about the data and its implementation, made without consulting states.

According to the Census Bureau, a new “accuracy metric” is being implemented for the next test version to be released around April 30.  The Census Bureau says the metric will “ensure that the largest racial or ethnic group in any geography entity with a total population of at least 500 people is  accurate to within 5 percentage points of their enumerated value at least 95% of the time.” 

This approach immediately raises several questions:  

1. Within the entities defined (presumably by assemblies of Census blocks), how will the second-largest racial or ethnic groups be affected?.  If the groups are roughly the same size, it may be important to define the majority group for voting rights enforcement.  And how will this affect outliers that are not within 5 percentage points? 

2. What entities used for voting (e.g., precinct-like entities, as well as other Census geographies) would likely be ignored because they have fewer than 500 people? We have presented a set of tables that shows the number and percentages of counties, places, minor civil divisions, tracts, block groups, voting tabulation districts, and blocks with fewer than 500 people.  (We used 2010 Census data for voting tabulation districts and blocks; for all other geographies, we used the 2015-19 American Community Survey).

These tables show most blocks have fewer than 500 people. Roughly one-fifth of voting tabulation districts have fewer than 500 people. A substantial number of other geographies – other than counties, tracts, and block groups – also have fewer than 500 people.

Once the new demonstration product and metrics are released, we will continue to analyze their impact using tools provided by the Census Bureau.  We will also continue to monitor the outcome of the court case attempting to block the use of the new distorting privacy method.

Author: Andy Beveridge