I spent the day at a conference put on by the Triangle Census Research Data Center at RTI, where I saw several interesting talks. Two speakers that I found particularly interesting were Robert Groves, the U.S. Census Bureau Director, and John Haltiwanger of the University of Maryland. The conference featured the ribbon cutting for the new Census Data Research Center facility at RTI, where researchers will be able to access microdata, the raw census data that is not made public in order to protect respondents’ confidentiality.

Robert Groves’ talk touched on the difficulties of addressing falling response rates for the Census and surveys in general at a time when they lack access to new funding. His solution involved addressing inefficiency in the Census Bureau by asking Census staff to send him their proposals for cutting costs – he received several hundred. He also talked about the conflict between the need to protect the confidentiality of respondents and the need to collect and provide rich data; breaching respondents’ trust could hurt the ability to gather data in the future (there was a lot of anger about the potential for this during the 2010 Census), and he encouraged researchers to see the data as something that belongs to the respondents, not the researchers. He also gave some amusing advice to researchers – don’t assume your research begins and ends with the data, because the data probably isn’t that good.

Haltiwanger gave a talk on the Longitudinal Employer Household Dynamics, which links census data on households to data on employers. This makes it possible for researchers to study things such as the career histories of entrepreneurs from their beginnings as sole proprietors (who make up the vast majority of U.S. businesses), to understand outcomes for employees who change jobs or lose their jobs, and to link census with patent data sets to compare the roles that entrepreneurs and innovators in creating new businesses.

I also learned that the Census hasn’t done a great job of preserving its data, to the point that the TCRDC recently had to get an old UNIVAC computer working to recover data from the ’40s and ’50s.  This is somewhat understandable since maintaining data is not their main mission. Nonetheless, the wonk in me cringes at the thought of all that lost data.