Why focus only on the lowest 25% of postcodes?

Yesterday my U of M boss, Glyn Davis, gave a speech on the difficulties in reaching the government’s target of 20% of all higher education enrolments being from a low socieonomic status background. The current definition of ‘low SES’ is living in the lowest 25% of postcodes according to the ABS index of education and occupation.

It is well established that socieconomic differences in school results are the major reason why low SES students are ‘under-represented’ at university. However, it was not until I recently analysed 2008 Victorian Year 12 results that I realised that lowest 25% seems like an inappropriately narrow target group. As the figure below shows, while students living in the lowest 10% of postcodes are clearly the weakest performers, those in the 2nd, 3rd, 4th, and 5th deciles have very similar (and not very good) school results). Very few receive ENTER scores in the 90s, and nearly half have ENTER scores below 50. While results trend upwards after the 5th decile, it is only a modest exaggeration to say that we really have the top 20% and the rest.


Source: Own calculations based on Victorian Tertiary Admissions Centre and ABS data. Note: ENTER is a ranking of school results, called UAI or TER in other states. The vertical axis shows what percentage of each socioeconomic decile received scores in the color-coded bands; the total of the bands is the proportion who received 50 or more. The 1st decile contains people with the lowest levels of qualifications and occupations; the tenth the highest.

The effects of these differentials on university enrolments are even greater, because the higher deciles have larger populations. Students from the top 25% of postcodes receive half as many again 90+ ENTERs as the other 75%. Even looking at the whole 70+ ENTER group, the main candidates for university entry, more than half are from the top 25% of postcodes.

As with many of the numbers set by higher education policymakers, the lowest 25% seems – if not simply picked out of the air – to be based on long-forgotten historical considerations. But it would seem perverse to focus simply on the lowest 25% when there are many others whose school results leave them no better off than students in the lowest 25% of postcodes.

We are expecting more detail on this policy in next week’s budget. The government’s performance in higher education policy does not exactly fill me with confidence that this problem will be remedied, but I hope it will.

23 thoughts on “Why focus only on the lowest 25% of postcodes?

  1. Now I’m even more confused. What does “ENTER” mean in this context? And just for clarity is the verticle axis showing percent that go to uni or final HSC grade?


  2. TerjeP – I have added a note to the post; as the title says these are school results, not uni transition rates. I hope to do that task, but this is very time consuming work on large data files (more than 45,000 individuals to be classified into more than 600 postcodes for the school results).


  3. So let me get this right, universities, as a result of government policy, are now focusing on educating people by postcode, is that right?

    I guess we’re now near enough to an earlier suggestion you made, Andrew, which was that perhaps some uni places to be given away by lottery… (I can’t really recall the reason).


  4. JC – At the moment, I don’t think unis particularly do focus on postcodes. The Uni of Melbourne’s access policy focuses on various kinds of proven disadvantage, wherever the potential students live. Other unis I am aware of approach the matter in similar ways. However, if there are big $$$ attached to the postcode target, uni policies will probably change.

    The lowest 25% is not quite a lottery, but certainly it would deliver windfall gains to be people with no disadvantage who happen to live in poor postcodes, while penalising those with disadvantage who don’t live in those postcodes.


  5. I don’t understand the problem Andrew. You seem to be suggesting that everyone who could succeed at university is probably already at university. This is exactly the kind of elitist bourgeois thinking that we’ve come to expect from the running dog lackeys of privilege. The jobs of the future are at stake here, and you’re quibbling about such nonsense as standards and integrity. There are illiterate disadvantaged people out there in need of a degree and the university system needs to get right behind the Education Revolution.


  6. Maybe some people are taking the human capital metaphor too literally.

    Universities, TAFEs and labour market programs are treated as if they were factories that manufacture and install human capital.

    The problem from this perspective is that the system is failing to install these human capital upgrades in an equitable manner.

    But maybe getting an education isn’t like installing better insulation into your roof cavity or having the latest software installed on your hard drive.


  7. The policy of targeting low SES students for university places (however you define low) is that they have greater potential for success at university than is reflected in their ENTER scores. The reasons for this that are typically given: a home environment where education is not valued; a poorly resourced school; large numbers of fellow students who have no interest in school and are disruptive or worse, and so on.

    This might be right or wrong, but you have to engage with the premise if you want to seriously criticise the policy.


  8. Some of these comments seem to have strayed from another thread; the post is not about ‘human capital’ or whether low SES students are worth targeting, but if you do target them how you define disadvantage.


  9. It would be useful if we could see a percentage figure on the proportion of the state population that comprises the SES decile. I was trying to understand why there was no single decile with a proportion of ENTER below 50, then realised the population variance between deciles is effecting the results. Could it be that the higher population high SES deciles make a disproportionate contribution to the ENTER ranges below 50 ENTER as well? That would be an interesting graph to put up along side this one.

    By the way – Lin Martin’s Equity and General Performance Indicators in Higher Education from the mid 90s, which sets the scene for the postcode methodology, acknowledges the problems of postcode but puts forward good reasons why it was originally adopted. Worth a read.


  10. “but if you do target them how you define disadvantage”
    You could certainly target schools. Indeed, there is probably enough data for many schools such that you could correct ENTER scores with reasonable accuracy based on how much the contribution of particular schools were to the ENTER score of a student. In this way, those going to crappy schools would only be punished for what they didn’t learn, rather than what they didn’t learn and the fact that they didn’t have great exam coaching. This would indirectly target disadvantage (I can’t see how it would be worse than postcodes). What you would hope is that richer parents stupid enough to send their children to crappy schools are likely not to be very bright, and hence their children wouldn’t do as well as the bright but poor children going to the same schools. This would thus add some poorer kids to the distribution. At the other end of the distribution, if you go to an uber great school but don’t do well, you are probably pretty stupid, and hence your ENTER score could be somewhat diminished. This would remove some richer kids from the system that didn’t deserve to be there.
    Of course, this is never going to happen, since universities need donors, and punishing kids who go to expensive schools, even if based on reasonable evidence, is not exactly making friends with those likely to have the money to donate.


  11. “The policy of targeting low SES students for university places (however you define low) is that they have greater potential for success at university than is reflected in their ENTER score”
    At least looking at the public vs. private school system, what you find is that while there is some effect of this, it isn’t big as people think (This idea seems to be a left-wing fantasy, that somehow everyone is equal after 12 years of shitty schooling). The reason is of course obvious — if you go to a shitty school, you learn less. Thus, whilst I think you could correct for this (as above), most of the correction is probably taking kids with high scores down, rather than kids with low scores up. As far as I’m concerned, many kids with low scores simply haven’t learnt much, and this often represents an insurmountable barrier to many university courses (think of trying to do engineering with Year 10 mathematics). Some universities have TAFE to correct for this, but the demand for it isn’t great.


  12. Matt – The main reason for the low numbers of below 50s is that ENTER is a ranking of the whole age group, whether or not they completed Year 12. Because those who did not complete Year 12 are assumed to get mostly lower scores, only 28% of all people in this study had ENTERs below 50. However, you are right that the population imbalances mean that the top groups have more people in absolute terms than is apparent from the chart above. The top 20% have 18% of all under 50 scores. However, they have 32% of all persons with an ENTER.

    I’ll send you the data.


  13. Conrad, the point of the policy is presumably not to send innumerate low SES students into engineering courses but to increase the number of low SES students, for a given ENTER score, who go to university. On Andrew’s data, nearly 30% of the lowest SES students have ENTERS above 70. Of those, how many fo to university? My guess is very few (but if Andrew has data to the contrary let’s see it).

    Why target the lowest 25% not the lowest 50%? It’s a good question. Presumably students in the middle-lower SES bands go to schools that are not as bad, have home environments that are not as bad, and so on. That’s almost true by definition of SES. So they have more chance of getting themselves to university by their own efforts than those right at the bottom. (So why do they do barely better at school than those right at the bottom? Perhaps the middle lower bands are disproportionately the children of tradesmen and self-made business people who aren’t badly off economically but who don’t put a lot of weight on school education. But that is just speculation.)


  14. Son of the Ratpack – I have not done that analysis yet; I don’t have the final 2009 enrolment figures for the 2008 year 12s. But as the 13% of 50+ ENTER scores coming from low SES postcodes almost exactly matches the low SES share of commencing students in 2007 (2 years prior) we are on track to replicate past research which found that for a given ENTER score low and high SES students make very similar uni enrolment decisions. This is however consistent with significant numbers in the 70+ group not going to uni, as application, offer and enrolment rates steadily decline with ENTER scores.

    The reason why the results are so similar is that though the postcodes are being ranked according to their result in the index of education and occupation, the deciles are often quite similar to each other in absolute terms. In asbolute terms the range is 746-1238. However, after a big leap in the first decile from the low of 746 to 899 at the bottom of the 2nd decile, the next four deciles between them only increase to 971. Compared to the top deciles, they all have residents with relatively low levels of education and relatively low skill jobs.


  15. Andrew, very interesting. The solution might be not to send smart but underprepared low SES students straight to Conrad’s engineering degrees (which they will probably struggle in) but to give them one or two years in a TAFE for some catch up education with the promise of a uni place and if they do well enough there. But that is a subject for another say.


  16. Giving 20% of places to 25% of the population would presumable mean that the middle 55% get 55% of places and the top 20% SES background get 25% of places.

    However if you are talking UniMelb, then I suspect that the top 20% SES background have a much larger share of the places already and that is unlikely to change. Therefore the middle band get squeezed to ensure that those from low SES get in.

    University enrollment is not just about opportunity it is also about choice. Middle class families aspire and to and often insist on university education (as opposed to any number of other career paths). Low income families often value income now over income in 4+ years time.


  17. “Low income families often value income now over income in 4+ years time.”

    I think this is probably true as an average attitude of low income families. But I don’t think it is true of the kids who have in fact made it to year 12 and done ok; they are already showing significant future orientation and the limited research to date (which I plan to test against recent enrolment data) suggests that they make much the same calculations about future costs and benefits as people from high SES backgrounds.

    There is an assumption in this debate that low SES persons are prone to irrational post-secondary education decisions. I think this is an unproven assumption, relying on an over-extrapolation from short-sighted behaviour observed in other members of the low SES group.


  18. “but to give them one or two years in a TAFE for some catch up education with the promise of a uni place and if they do well enough there. But that is a subject for another say.”
    They do this where I work, but I believe the demand isn’t very great, and that, at least for the Aus students, it is generally from older students coming back, not low SES students (which fits with the idea that low income [white] families may value immediate monetary gains more than others). The other reason I think demand is not big (for what is a good scheme with good and measured outcomes) is that it takes an extra year to get your degree, even for the 3 year degrees. So whilst it seems like a good solution, empirically it doesn’t work well.


  19. Andrew I recall seeing some data on SES and drop out and completion rates for undergraduate enrolments a few years back. Are you aware of recent data on this?


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s