Earlier in the year I argued that the governments university equity policy focused on the lowest 25% of people by sociecononomic status was fundamentally flawed.
Using NAPLAN and Victorian Year 12 data I had found that the academic results of the lowest SES 25% (by occupation and postcode respectively) were little different from the second quartile. Consequently, the first quartile was too narrow a focus for policy.
An excellent new paper by University of Melbourne economist Mick Coelli, using higher education participation data from the census and the HILDA survey, puts this conclusion beyond reasonable doubt. Whichever way we look at SES: income, education, occupation or postcode the result is the same – the second quartile is very similar to and perhaps even worse off than the first quartile for their kids getting into university.
The main reason I think is that the first two quartiles have very similar levels of education – early school leaving and/or certificate I and II vocational qualifications – which leads to similar education outcomes for their kids.
Possibly the slightly better results for the lowest quartile in some datasets is due to their greater access to full Youth Allowance.
Coelli’s statistical analysis finds other interesting things. Having English as a second language is an advantage after controlling for other parental factors, presumably due to the ambitious pro-education attitudes of many migrant families. Being the eldest child is also an advantage (this would also include only children). Having a mother aged less than 40 is a disadvantage.
I don’t think we need affirmative action for people who speak English as a first language, are second or later children, or who have young mothers. But if we are going to target low SES people, the target group must be increased to the lowest 50%.
Andrew Norton 
Or maybe decreased? Perhaps the big difference is between the lowest 10% and the remaining 90%?
LikeLike
“But if we are going to target low SES people, the target group must be increased to the lowest 50%”
Of course, you’d have to call it something different, because half the population shouldn’t qualify as low. I also can’t imagine many people in the top half being especially pleased if there are going to be different standards to get in because of it (say, compared to if the bottom 10% had some sort of boost where I doubt too many people would complain).
It seems to me that the main way of targeting the bottom half is with things we already have like income supplements.
LikeLike
Tim – I did find with the Year 12 results by postcodes ranked according to levels of education and occupations that there was a distinct lowest 10%. But I did not have the data to see whether that flowed through into uni attendance rates (whether your results are really really bad or just really bad you still don’t get into uni).
The methodological difficulty is that with the most powerful predictive indicator, parental education, we don’t have a way of isolating a bottom 10% because there are too many people with very low education levels.
IMHO all the ‘easy’ policy levels like removing the caps on places and income support will only make a small difference, since the fundamental issues are educational results and career aspirations. The census data clearly shows that blue collar families making $150K per year have much lower uni attendance rates than the poorest professional families. For the sons of working class families, there is a negative correlation between income and uni attendance.
LikeLike
As I noted on your previous post, this sort of study proves absolutely nothing because there is no attempt to account for selection effects. I’d argue that given the barriers they face long before they ever reach the stage of applying for uni, you would expect that only pretty special kids from disadvantaged backgrounds would reach uni, and that therefore your prior expectations for these studies is that they will show such kids doing exceptionally WELL at uni. That they don’t may mean there is a problem.
LikeLike
DD – I don’t get your point. The post is about whether the same social circumstances/forces apply in the second quartile as in the first re education. The evidence all supports the conclusion that they do.
How well low SES do at uni compared to higher ses is a separate issue, though the research suggests at least no worse and probably a little better, given prior academic achievement. That is a selection effect – only the most academically able and persistent of the low SES make it to uni.
LikeLike
“I’d argue that given the barriers they face long before they ever reach the stage of applying for uni, you would expect that only pretty special kids from disadvantaged backgrounds would reach uni, and that therefore your prior expectations for these studies is that they will show such kids doing exceptionally WELL at uni.”
.
You’re wrong on that — the data from where I work shows that low SES students do worse than high SES ones. That might be for lots of reason, not least of which is that they still have money problems. They may also have learnt less at school (I haven’t seen the data equated for entry scores).
LikeLike
The data I have seen shows they do worse in absolute terms (because their average prior academic achievement is much lower) but for a given ENTER/UAI on average do better than someone from a higher SES group, at least in first year.
LikeLike
Andrew, I haven’t read the Coelli paper so don’t know whether it addresses this, but I don’t think SES status is sufficient to examine or address, for want of a better term, inequality in access to tertiary education.
For example, rural and overseas students suffer the huge disadvantage, compared with urban students, of having to find and manage accommodation. Yet most studies seem to ignore this for some reason, especially in relation to rural students.
Secondly, with regards to rural students, postcode studies typically fail to distinguish between students who boarded at city or regional schools and those who studied locally. This effect can mask disadvantage in rural students.
Also, I would be suspicious of using parents’ declared income in these sorts of studies. Wealthy parents are often able to arrange their affairs so that they appear to have low incomes. Does Coelli show awareness of this and similar factors?
LikeLike
Don’t rural students live in colleges? The life might not be Brideshead Revisited, but it’s not bad, and these colleges have in-house tutors.
LikeLike
Inaccurate reporting of income is always a problem in survey research – though there is no incentive to lie in these surveys (unlike say tax) there are also no penalties for getting it wrong and no rewards for getting it right. Also people often fill in forms for other household members and may just guess. The highest individual category in the census ($2K a week in 2006) means that top incomes are under-reported.
There are known problems with the census in particular with implausible numbers of people reporting incomes below their welfare entitlements.
That said, there is little reason to believe that the data for the bottom two quartiles (recall we are talking about millions of people here) are systematically distorted in comparison to each other by people fraudulently or negligently reporting their income.
The fact that the same pattern keeps repeating regardless of measure suggests that it is robust.
Incidentally, income is never used in official measures of SES for higher ed purposes – here it is just supporting the general conclusion.
LikeLike
Census postcode studies are usually based on where the person lived 5 years ago to try to deal with the issue of moving to study, though I don’t think this study does this. It does however find a significant urban advantage while noting that because the study is based on students who still live with their parents this result may have some bias in it.
The HILDA postcode analysis uses address at age 15. It finds an advantage for high SES areas even after controlling for education and income.
Rural location is a known correlate of low uni attendance. For school leavers, a DEEWR statistical analysis put about half the difference with inner metro down to lower education, occcupation and income of parents. However the rest of the difference was unexplained by their model. Interestingly distance from a campus had low explanatory power – though as it noted this may be misleading, as just because there is a campus it does not mean it has the course the student wants.
LikeLike
“Don’t rural students live in colleges? ”
Most of them don’t.
LikeLike
Aaargh. Its a commedable goal in theory but you can just see how bloody horrible this is going to get, with endless layers of bureaucracy, stupid rules and regulations, and opportunities for gaming the system.
Its funny how a desire to make things better gets totally botched by government into a complete fiasco.
LikeLike