On August 12, the U.S. Census Bureau released the Redistricting Data Summary Files, also known as PL-94. This release contains information on the following: 

• Housing unit counts  
• Occupancy status for housing units (vacant or occupied) 
• Population totals  
• Population totals by race  
• Population totals by race and Hispanic/Latino origin  
• Voting-age population (age 18 and older) totals by race and Hispanic/Latino origin  
• Population totals in group quarters by major group quarters type  

These data are available at multiple geographies, including counties, municipalities, census tracts, block groups, and census blocks, the smallest census geography. 

MAPC has published PL-94 data for all of Massachusetts in a simple table format on the MAPC DataCommon. Use this link to access it directly, or navigate to www.datacommon.mapc.org and select “Datasets,” then “2020 Census,” and then “Redistricting Data.” 

PL-94 data can also be accessed directly from the Census Bureau via their FTP site: however, the current release is in a format that is not very user friendly. The Census Bureau will re-publish the PL-94 data in a more user-friendly format on or before September 30.

In addition to the PL-94 data, the Census Bureau will publish at least two collections of data: 2020 Census Summary File 1 (SF1) and 2020 Census Summary File 2 (SF2). These releases will include more detailed population and housing information (for example, population by detailed age categories, household tenure (owner or renter), and year-round housing unit counts) and cross-tabulations of the data (for example, population in households by age). MAPC expects SF1 to be published in early 2022 and SF2 to be published after SF1, likely in spring 2022.  

As in 2010, the 2020 Census did not include a “long form” questionnaire that asked questions about income, education, or employment—that data collection has been replaced by the American Community Survey, which is conducted on an ongoing basis. The decennial census is designed to provide a complete enumeration of the population, whereas the ACS samples a small number of households every year to assess detailed population and household characteristics. ACS results are available down to the block group level, and results for most municipalities and smaller geographies are only available as five-year averages. The next ACS release will cover 2016-2020 and will be released in December 2021.    

The PL-94 data cannot be used to definitively update a municipality’s Subsidized Housing Inventory (SHI) percentage. The SHI percentage is calculated as the number of qualifying subsidized units divided by the count of year-round housing units in each municipality. (Year-round units are housing units that are intended for consistent, rather than occasional, use. Examples of housing units that are not considered year-round include vacation rentals, second homes, and properties under renovation.) However, the PL-94 data does not distinguish between year-round units and those for occasional use. This information will be available in the SF1 data due to be released next year.   

In the meantime, the PL-94 data can be used to assess overall housing unit growth and determine if changes in housing stock may have had an impact on the SHI percentage.  An increase in the number of year-round housing units would decrease the percentage of SHI Units in the municipality, which could cause a municipality to fall below the 10% statutory minimum required for safe harbor from Chapter 40B. 

As part of the forthcoming regional plan MetroCommon, MAPC has developed regional population and household scenarios out to the year 2050. These projections will be released in October 2021. Due to the late release of the PL-94, MAPC will not be incorporating the 2020 enumeration into those scenarios. However, MAPC is in the process to update municipal level population and household projections for the next Long Range Transportation Plan, with data to be released in 2022. These projections will incorporate 2020 population and housing units counts.  

We anticipate that decennial census data will provide a high-quality picture of our region, despite myriad challenges present last year. Thanks to a two and half-month extension of the census response deadline and tireless work by Census Bureau, non-profit, and local government staff or volunteers, the 2020 Census performed well on two key measures of operational quality: Massachusetts self-response rates (percent of households that complete a census questionnaire) and proxy response rates (percent of questionnaires completed with third-party information or high-quality administrative records). Both of these measures were at or above 2010 levels. As a result, relatively few records were computed though imputation, a statistical method for populating missing data. 

This said, it is important to note that additional metrics on the accuracy of the census have not yet been released. These metrics, for example, a comparison of population data from the decennial census to other population records and a post-enumeration survey that re-surveys a sample of the population, should help us further understand the quality of the census. It is also important to note that the accuracy of census data is not uniform: accuracy will vary by location (for example, despite an overall higher self-response rate statewide, Berkshire and Hampden counties actually had lower self-response rates in 2020 than in 2010) and by types of respondents, as certain characteristics (for example, respondents of color, of lower incomes, who are foreign-born, and more) are associated with lower census response rates. The forthcoming metrics should also provide more clarity on these shortcomings. 

All college and university students should have been counted at their school-year residence in the 2020 Decennial Census, regardless of whether they were sent home due to the pandemic. These responses should be attributed to the geography where the student lives while attending school, as this is their “usual residence,” the place where they live and sleep most of the time. This attribution is very likely to be correct for students living in on-campus housing, as college and universities respond to the census on the students’ behalf through the Group Quarters Operation. Students living in off-campus housing would have received an invitation to complete their census questionnaire to that address. Students who relocated before completing their questionnaire would have received subsequent invitations to their forwarding address; students without mail forwarding would have been able to complete their census online, though they would have had fewer direct reminders. Again, we will know more about the impact of the pandemic when the census releases additional quality metrics. 

This year, the Census Bureau adopted new protocols designed to prevent unauthorized attempts to combine census tables with consumer listings and other commercial data sources.  This new method, known as Differential Privacy, adds “noise” to the data, so the population tables do not exactly match the responses of individuals in a given area. At this time, it is not clear exactly how much “noise” was inserted into the data; however, we do know that the effect is greater for smaller, less populous aeras (e.g., block groups, rural towns) than it is for municipalities and counties, or aggregations of block groups, etc. Therefore, results for an individual block, block group, or tract should be reviewed to ensure they are reasonable.  It is also important to note that the total population at the state and state-equivalent level (D.C. and Puerto Rico), the total housing units at the census block level, and the number of group quarters facilities by tape at the census block level were held ‘invariant’, so these results can be relied on.   

Despite these caveats, the decennial census remains the most comprehensive and complete enumeration of the population in Massachusetts and remains a more reliable data source than less robust municipal censuses or other sources.