Saturday, February 14, 2009

Census politics

Most social scientists depend on data from government agencies. That requires some suspension of disbelief; we want to believe that the reports are not politicized. But Wendell Cox has just posted this chilling discussion of the drift at the Census Bureau and how recent moves by the Obama administration suggest more politicization.

The second (and related) problem that many of us are stuck with is that there are fewer data as one moves down along the geographic aggregation scale. The best data are for the nation, less for the states, less for counties and metros and cities, and much less for sub-city spatial units. Data for census tracts and census blocks are useful, but we only get these reports every ten years.

And cities are the "engines of growth" if and only if they achieve spatial arrangements that allow many potential positive externalities (including information spillovers) to be realized while leaving a bunch of potential negative externalities unrealized. So what happens at the sub-city level is critical, but we have a tough time understanding it. Dearth of data is a big problem.

Adding to the cloud of uncertainty over the precious decennial disaggregated data is among the last things we want. All politics may be local. And much of the important economic questions are also local.

I used to wonder why pollsters are always identified by their political affiliation. Until I thought about it. Do we now have to consider the political affiliation of census takers?

I keep telling students that no one looks for national mean weather report. Why then be satisfied with national economy readings?


Here is an interesting piece by Richard Florida on cities and the economic crisis. Note his reliance on "density". But the average population or employment density of a large metropolitan area tells us nothing. The nature of spatial arrangements is much more complex and requires sub-metro area measures.