Someone Check The ACS Please

Image result for american community survey data

The 2011-2015 5 year “wave” estimates for the American Community Survey are out.  These are exciting because we now have enough “waves” of the ACS to look at changes in Census tracts over time without having to utilize those painful crosswalks…

My hope is that someone will take a look at the 2006-2010 ACS estimates for tracts against the 2011-2015 estimates and let us know a few things:

  1.  Are the samples robust enough to give us significant differences where we know there are significant differences?
  2. What can they tell us about urban change, if anything?
  3. How has demographic change itself impacted the stability of the estimates in certain places?

Many people use 5 year wave data like the 2011-2015 dataset and, in their manuscripts, refer to it only as the “2015 ACS.”  This is ok if you have acknowledged towards the beginning of the paper that you are referencing what is a five year rolling sample… but many practitioners, media folks and the public at large sees “2015 ACS” and thinks that is what you are talking about.

Failing to appreciate this difference means ignoring what the data is really telling you…

California’s Legislative Analysts’ Office (LAO) twisted the ACS data used by Karen Chapple and Miriam Zuk at Berkeley to try to say, essentially, the opposite of what Zuk and Chapple argued in their own work… to the point where they had to respond.

The biggest critique I have in all of this is much more basic: what does a median rent in a Census Tract really tell you?  When you tell me that adding affordable housing to a census tract over a period of time reduces its median rent change in that period, all you are telling me is that adding numbers to the bottom of a distribution of numbers shifts the distribution’s median downwards.  How much can we really say with the blunt instrument that is the ACS 5 year tract estimates?

Last point on that Berkeley-LAO debate: they relied on the 2009-2013 5 year wave data, but kept constantly referring to their findings as measuring change from the 2000 Census and 2013 ACS.  The 2013 wave includes 2009, when we were in a recession. It also includes 2013, which was undoubtedly a ‘boom’ year for housing in the Bay Area…  I’m wondering if someone ran the Berkeley analysis with the latest ACS wave, what would they find? It’s the first wave we’ve got tract-scale estimates for that is clear of the recession.  Someone please dig in!

Published by mattdpalm

Doctoral Student UC Davis USDOT Graduate Eisenhower Fellow 2014-2015

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: