Careful, that research might tell you zip

Posted on January 21, 2008 by andrewleigh

According to The Age, a Victorian parliamentary inquiry chaired by Geoff HowardÂ is currently looking into the question of whether certain postcodes are underrepresented at university. Separating neighbourhood effects from family background effects is a potentially important exercise, but the article makes me think that the inquiry is more likely to fall into the trap of using postcode as a proxy for income. Which would be a pity. The questions “how much does family background matter?” and “holding constant an individual’s family background, does postcode matter?” are interesting ones (my colleague Chris Ryan has done some nice work on the former). The question “does postcode matter?” only begs the previous ones.

This entry was posted in Economics of Education, Universities. Bookmark the permalink.

5 Responses to Careful, that research might tell you zip

Andrew Norton says:

January 22, 2008 at 8:18 am

Bob Birrell has done a lot of this analysis before, but the Victorian government’s On Track survey of school leavers and the 2006 census can add to it. The main proximate cause of differences is going to turn out to be school results (if they do the review properly), but the interesting questions as AL suggests relate to the combination of factors that help explain those differences in school results.
derrida derider says:

January 22, 2008 at 12:25 pm

Or, are poor areas poor because that’s where poor people live, or are poor people poor because they live in poor areas? It makes a lot of difference for policy.
christine says:

January 24, 2008 at 12:30 am

DD: Recent work suggests latter is not the case (see studies of ‘Move to Opportunity’ programs, and work by Phil Oreopoulos).
Bruce Bradbury says:

January 24, 2008 at 2:08 pm

One shouldn’t entirely discount location as a measure of socio-economic advantage. In a recent paper I looked at the association of both family income and postcode SEIFA scores with outcome indicators for pre-school children. Among low-income families, those from more advantaged regions did better (there was little difference among high-income families).

One possible explanation is that some low-income families actually have greater economic resources which allows them to live in better areas. So location can sometimes act as a proxy for wealth – which is missed in income measurements. (There are other possible explanations based on variations within postcodes).

My paper (see section 8.2) can be found at http://www.sprc.unsw.edu.au/reports/LSACOutcomesSPRCReport.pdf
Andrew Leigh says:

January 25, 2008 at 7:33 am

Bruce, I think you’re asking exactly the right questions, as is the MTO work (they have the advantage of not having to worry that regional income is correlated with measurement error in own income). I just hope the Vics don’t stop at postcode.