The Census Bureau continues to roll out the latest data from the American Community Survey, last month announcing the availability of the first real nationwide data for 2010 in the form of the 1-year ACS Estimates. I’ve been conducting some trainings on how to get and use ACS data for local-level community planning (self promotion: check out the Mel King Institute’s training in Boston, or wait for us to offer it again), which has prompted me to pay some more attention to the new “American FactFinder” platform, which has prompted me to write this post.
In short, if you want to use the new numbers (including the latest 3-Year and 5-Year Estimates when they are released), you’ll probably end up needing the new FactFinder. And while I am impressed with some of the features (the ability to quickly make thematic maps and print-ready tables on the fly is fun to show off in the trainings), my general sense is that the “legacy” FactFinder was just fine, thank-you-very-much, and I find myself missing the ordered approach it imposed. In the old version, you methodically navigated through a set of choices, refining your search as you went: first choose a data set and a year, then select geographies, then tables, then get the data. If you wanted to change anything, just follow the breadcrumbs to back up to and edit your selections. Easy-peasy.
The new version is certainly prettier–it’s nicely rounded and pleasant on the eyes; selection boxes emerge from the background like good English butlers when needed, and fade away when their work is done; and those added bells and whistles are fun. But gone from all of this is the structure of the old FactFinder, which seems to be essential to the practice of data analysis. Rather than viewing the process as akin to following a recipe or writing a program, the experience feels more like a shopping spree, with no clear difference between selecting geographies, tables, datasets, or even other options (“industry codes”? “document types”?). One unfortunate byproduct of this approach is a general confusion about whether one is even “selecting” or “filtering” as one goes—are these logical “ors” or “ands”? An even larger problem is the temptation to keep clicking willy-nilly, rather than following out a clear mission to get the data. In short, I fear that where the old FactFinder provided a user-friendly introduction to the way Census data is organized, the new version moves us one (albeit small) step in the wrong direction.