New "differential privacy" policies from @uscensusbureau could mean the end of public-use microdata. Important report from @ipums.
assets.ipums.org/_files/mpc/MPC…
assets.ipums.org/_files/mpc/MPC…
More on what differential privacy could mean for Census data here: nytimes.com/2018/12/05/ups…
"Public-use microdata" is a wonky, snooze-inducing phrase. But the idea is simple: The government makes available anonymous, individual-level responses to many of its major surveys. This is important for a couple reasons.
Most importantly, researchers, policymakers, nonprofits, advocates and, yes, journalists use microdata to understand their communities and the policies and trends that affect them.
A simple example: Official statistics from @BLS_gov generally only provide educational breakdowns for people 25+. Curious how job prospects differ for recent college grads vs their parents? You'll need microdata for that.
I could point to countless studies, news stories, etc., that drew on microdata from the American Community Survey, Current Population Survey and other major government datasets. If I have time, I'll tack a few on to the end of this thread -- or send me yours!
There's also another less obvious but I think important benefits of publicly available microdata: confidence in government statistics, which feels particularly relevant right now.
When Jack Welch implied that the Obama administration was mucking with the jobs data, economists and econ journalists scoffed. One reason: We literally look at the data every month. Not just the aggregate data, but the individual-level responses.
There are, understandably, important and valid concerns about the privacy implications of government statistics. Those of us who have worked on this know how seriously @uscensusbureau and other agencies take those concerns.
I'm certainly not an expert on differential privacy, and it's not my place to argue for or against any specific policies. And it's certainly possible that policies will need to evolve along with technology.
But I *am* here to say that if we lose public-use microdata, we'll be losing something very valuable, not just to the researchers and reporters who use it but to the communities they -- and we -- serve. <fin>
One addendum to this, because it's come up a bunch in replies: I have no reason to think this discussion is anything other than a good-faith attempt to grapple with difficult privacy issues.
I get why people are on high alert for any evidence of political interference with government data. (And if you have any such evidence, my DMs are open!) But this discussion, and the privacy concerns underlying it, predate this administration.
