Using Crosswalk to transform 2005-09 ACS to 2010 Tract Boundaries

For my research, I have to perform a crosswalk of 2005-09 ACS data for several SES indicators to 2010 Tract boundaries. What I gathered is that the block-level crosswalk is the most precise and reliable method. Since I will be using census tract level data, do I need to go further and aggregate the block-level crosswalk conversion into the census tract level?
Also, I am not sure if the SES indicators I have chosen are counted at the block level. In that case, how to go about it?



There are a couple issues that will make your goal uniquely challenging…

First, no ACS data are published at the block level, so you won’t directly be able to use the NHGIS block-to-block crosswalks. We also recently added crosswalks for “block group parts,” the smallest units used in 1990 and 2000 long-form summary tables, but again, no ACS data are available for block group parts either.

The smallest units identified in 2005-2009 ACS summary tables are for block groups, so I’d recommend you create a crosswalk from those. (Note: you say that you “will be using census tract level data”, but that doesn’t need to determine the level you start from; it often only determines the target level that you want to crosswalk to. As explained on our crosswalks page, it’s generally best to crosswalk from the smallest possible units. If you currently have the SES indicators only for tracts, it’s still likely you could obtain most or all them for block groups, too, which is worthwhile if you’d like the most accurate possible stats for 2010 tracts.) In this previous forum post, I describe how you can use the NHGIS crosswalks for data from block groups. The same principles apply in your setting.

Second, as noted in the NHGIS GIS Files documentation, the tracts and block groups identified in the 2005-2009 5-year summary tables generally correspond to the 2000 census definitions of these areas, but in 19 counties, they don’t correspond exactly to either the 2000 or 2010 definitions of tracts and block groups. This Census web page provides more details, including a list of the affected counties. If you need data in any of these 19 counties, you’ll need to do additional research to determine how the 2005-2009 ACS data correspond to 2010 tract boundaries.

Thank you so much Mr. Schreoder!
This explanation really helped me a lot!