One advantage of mySidewalk’s data library is that it includes historical Census estimates alongside current data, allowing users to see how certain indicators have changed over time. But, comparing data from different Census years is not as straightforward as it may seem. The main reason is that the shapes and boundaries for most Census geographies change over time for a variety of reasons.
Why Geographies Change Shape
Every Decennial Census, a new set of blocks, block groups, and tracts are created. The boundaries of block groups and tracts are based on groupings of populations, and as our communities grow (or shrink), these shapes change too. Blocks are the smallest geography in which the US Census provides data and change to match changes in the physical environments. Blocks are defined by physical environment features such as roads, railways, and bodies of water. Block groups are the next most likely to change, as they are each a grouping of blocks. The US Census tries to maintain some consistency in the shape of tracts over the decades, but Census tracts are often split, merged, or new ones created to accommodate the changes in where people live.
Why does this matter? At mySidewalk we work hard to provide data for the same set of shapes, so that you can use data from many different data sources and see how an area has changed over time.
Here's where things get even more complicated. Not only do geographic shapes change over time, but so does the way that the Census identifies geographic shapes: GEOIDs. Our work involves accounting for these changes too. How might these GEOIDs change?
The GEOID refers to a new shape
The original shape got bigger, or it got smaller. A different number and/or ratio of blocks will now fall within the geography than before.
The GEOID was reused to replace ID for a shape in a different location
This is rarer, but does happen. In this case, a shape from the 2010 Census had a GEOID, but that shape changed to such an extent that the GEOID is no longer used. That GEOID may be recycled for a shape in 2020 that has 0% overlap and may even have replaced an exact 2010 shape that had a different name.
How mySidewalk handles changing geographies over time
Geographic harmonization is a statistical technique used to facilitate comparisons of Census concepts over time. For example, geographic harmonization allows us to “transpose” Census 1990 estimates from Census 1990 geographies to Census 2020 geographies.
mySidewalk provides data from the following Census products:
Decennial Census 1990
Decennial Census 2000
Decennial Census 2010
Decennial Census 2020
Current 5-year ACS Estimates
Major updates to Census boundaries generally coincide with the release of decennial Census data. Since the most recent major update to Census geography boundaries occurred with the release of the 2020 Census, it was necessary to use geographic harmonization to support Census 1990, Census 2000, and Census 2010 data alongside other Census products in our data library.
We used Census block-level geographic crosswalks provided by the National Historical Geographic Information System to perform reapportionment of Census 1990, Census 2000, and Census 2010 block group level estimates to Census 2020 block group geographies. We then aggregated the block group level data to higher levels of geography (states, cities, etc.) via weighted block-point apportionment (WBPA). For more information on how the crosswalk files were developed, see the methodology documentation provided by the NHGIS.
Steven Manson, Jonathan Schroeder, David Van Riper, and Steven Ruggles. IPUMS National Historical Geographic Information System: Version 12.0 Geographic Crosswalks. Minneapolis: University of Minnesota. 2017. https://www.nhgis.org/user-resources/geographic-crosswalks