Monday, October 05, 2009

Use and abuse of socio-economic rankings in public policy

Last December in Saturday Morning Musings - Byzantium, ARIA and Australian public policy I used Byzantium as an entry point to a discussion about geography, perception and the Accessibility/Remoteness Index of Australia (ARIA). I extended this discussion somewhat in March in Problems with language and definition in public policy

Without re-running the whole argument, part of my point was that the use of statistical constructs such as ARIA in policy development could have quite perverse results. ARIA was based on geography, but actually ignored geography.

I make this point now because of the growing use of another concept, statistical rankings of socio-economic disadvantage. This may sound dry, but bear with me for a moment.

In the Australian of 30 September, Andrew Trounson had an interesting article discussing the Australian Government's plans to increase the number of university students from socially and economically disadvantaged backgrounds. To do this, they appear to be considering increasing the funding loading given to universities for such students to about $A400.  Trounson suggests that the government has yet to finalise a method of identifying low-SES students, but the expectation in the sector is that it will opt initially for a measure linked to the Socio-Economic Indexes for Areas.

This story caught my eye for several reasons.

The first was the provision of funding to universities instead of to students.

On the surface, the best way of helping poorer students to get to university would be means tested scholarships. Giving the money directly to universities certainly provides an incentive for universities to recruit such students. However, of itself it does nothing to help students to actually get there. They still have to support themselves financially, and poor students in particular struggle under current funding arrangements. So I am not sure that the proposed approach is addressing the right problem.

The second thing that caught my eye was the reference to the Socio-Economic Indexes for Areas. I have been vaguely conscious of this one for a while, but had yet to follow up.

The Australian Bureau of Statistics describes these indexes in this way:

SEIFA stands for Socio-economic Indexes for Areas. This suite of indexes ranks geographic areas across Australia in terms of their socio-economic characteristics. The SEIFA indexes are created by combining information collected in the five-yearly Census of Population and Housing (called the Census throughout this paper). There are four different indexes, each representing a slightly different concept. These concepts are abstract and difficult to measure, so the indexes aim to capture these abstract concepts by combining information that is related to the concept. For example, the Index of Relative Socio-economic Disadvantage uses information such as low income and low education as markers of relative socio-economic disadvantage.

The SEIFA indexes are rankings. Each index ranks different geographic areas of Australia according to a 'score' that is created for the area based on characteristics of people, families and dwellings within that area.

The key point to note about this description is that we have another statistical construct like ARIA. From my viewpoint as an analyst, the various rankings based on the 2006 census are very interesting because they allow me to look at the pattern of disadvantage, especially in New England.  As I noted back in March 2007 in New England's Poor Towns - a failure in public policy, nearly all the poorest towns and villages in NSW are to be found in New England.

But while the rankings are interesting, I am not sure that they provide a solid base for funding decisions.

If the objective of the new proposals is to help disadvantaged areas, then the use of SEIFA might make some sense, although I would argue that there are better ways to go. However, if the objective is to help disadvantaged students, the use of SEIFA would appear fraught with the same type of problems that make ARIA such a poor policy guideline.

At the simplest level, it means that a well off student from a disadvantaged area would give a university extra funding, a disadvantaged student from a well-off area would not.

I am not sure that this makes much sense.  

2 comments:

Anonymous said...

I think your analysis is spot on.

Jim Belshaw said...

Thank you, anon. I must say, and this is something I have written on a fair bit, that the continued blind use of various metrics continues to worry me. It gives some very bad results.