Method:
- Loop through the 2024 tables, keep tables that also exist in 2025, have the same number of cells and are not empty
- For each table, count the number of geographies in the table (number of records and the number of suppressed estimates (-666666666 or -999999999) for 2024 and 2025
- Join the 2024 and 2025 counts
- Count geographies in 2024 but not in 2025 and vice versa
- Aggregate 2024 and 2025 missing values for geographies that appear in both years
Findings:
- A lot of variation in the number of the aggregate missing values by table, but no discernable patterns between years.
- There are tables with many more geographies in 2024 than in 2025. For example:
o B05008: Sex by Place of Birth by Year of Entry for the Foreign-Born Population
2,745 geographies in 2024 and 2,041 geographies in 2025
o B27010: Types of Health Insurance Coverage by Age
5,533 geographies in 2024 and 3,395 geographies in 2025 (could be due to changes in the questionnaire)
- There are also tables with many more geographies in 2025 than in 2024, most noticeable in several tables with aggregates. For example:
o B19130: Aggregate Family Income in the Past 12 Months by Family Type by Age of Householder
1,982 geographies in 2024 and 6,793 geographies in 2025
o B25080: Aggregate Value (Dollars) by Units in Structure
2,327 geographies in 2024 and 4,658 geographies in 2025
- There are also tables where the total number of geographies was about the same, but many geographies appeared only in 2024 or only in 2025. For example:
o Table B10010: Median Family Income for Families With Grandparent Householders and/or Spouses Living With Own Grandchildren Under 18 Years by Responsibility for Own Grandchildren and Presence of Parent of Grandchildren
5,877 geographies in 2024 and 5,858 geographies in 2025 but only 4,339 appeared in both years
o Table B25086: Aggregate Price Asked (Dollars)
4,073 geographies in 2024 and 3,786 geographies in 2025 but only 2,929 appeared in both years
I haven't done similar analyses between 2023 and 2024, so I am not sure if these differences are "normal", or can be contributed to the new Disclosure Avoidance System
------------------------------
Jan Vink
Cornell University
------------------------------