Class effects on class performance?

The latest results of the Australasian Survey of Student Engagement were released today, and like previous research finds very little evidence that socioeconomic status matters much once students arrive at university.

On various learning outcomes, low SES students – defined in this case as being from a low SES area and being the first in their family to attend university – rate themselves slightly more favourably than middle or high SES students.

(click on image for clearer view)
Source: figure 40, ACER, Doing more for learning: Enhancing engagement and outcomes

Perhaps this is because higher SES students have a less favourable comparision between uni and school, but low SES students do not perceive themselves as being particularly disadvantaged at university.

Unfortunately the report doesn’t go into detail about grades, though it does say that the average reported grade for middle SES students is only 0.7 on a 0 to 100 scale above low SES students. However this is only a modestly useful finding, since as I have argued the definition of low SES based on living in the lowest 25% of postcodes has no basis in social science. Many so-called middle SES students are sociologically pretty much the same as ‘low’ SES students.

The limited past research would suggest, however, that once university entry scores are controlled for low SES has no effect or even a positive effect on grades.

In the government’s proposed equity funding policies additional money is to be spent on services for enrolled low SES students, but the evidence suggests that SES is not a reliable indicator of academic need. There have been long delays in releasing the final version of this policy, and I can only hope it is because they realise that the original policy was misconceived.

16 Responses to “Class effects on class performance?

  • 1
    May 17th, 2010 22:15

    I wonder if there is much difference across SES, in starting salary and salary 5 years from graduation, after controlling for academic performance in the penultimate year of university.

  • 2
    May 18th, 2010 05:02

    I was under the possibly incorrect impression that the problem with the new rules was that the government was going to give money to low SES students no matter what — this will be a mess because universities will use different cut-offs for low SES students and others (make sure you rent a place in a crappy neighborhood before you enroll if you need to get in), which will have an effect depending on how low universities are willing to go for the money and since most have few services to help low performing students catch up. Given that I think it is safe to assume that some will go as low as need be to get it, this will mean there will be classes where there is a reasonable difference in performance between the low SES students and the rest. That might not matter much in some courses where both normal and low SES groups still have TERs in a range where the TERs don’t predict course outcomes well, but it will make a difference to the other courses which will need to be diluted to accommodate them.

  • 3
    Andrew Norton
    May 18th, 2010 05:41

    Brendan – Not controlled for entry scores, the Graduate Pathways Survey found that low SES students 5 years out were slightly less likely to be in professional or managerial employment, 59% to 64%. No information was given on salary.

    Conrad – It’s a while since I read the draft guidelines, but from memory there is a funding formula that pays unis according to what proportion of all low SES students they have. But oddly this is not an incentive scheme – if unis have to spend all the money on low SES students there’s actually no financial reason to take more of them, even if they will win ideological points from the government.

  • 4
    May 18th, 2010 08:39

    Andrew how were the survey questions phrased?

    One simple bias could be that students are comparing themselves to their friends/family rather than other students in their course. Therefore low SES students at uni are more likely to be the smartest kid from their school. Compared to high SES kids where being successful at uni is the expected outcome for all students.

    Lots of students at uni (particularly in the early years) actually have relatively little academic interaction with other students.

  • 5
    Andrew Norton
    May 18th, 2010 09:41

    M – If you download the report, the questions start at page 62. I think the students are being encouraged to think about themselves in the past compared to now. What the data show is that the SES groups are far more alike than different, but I would think the idea that low SES students do feel that they have developed more is more likely than bias in explaining the small differences that we observe. Because they are more likely to start from a lower skills base, they gain more from the uni experience.

  • 6
    derrida derider
    May 18th, 2010 11:33

    The biggest bias here is due to what is NOT in the survey sample – low SES students who ought to be at uni but aren’t.

    If you’re worried about inequality of opportunity (and you damned well should be), the real damage is done by schools policy, not higher ed policy. The egregious support for private schools by the Howard government, continued by the Labor goverment, will have long term consequences for social mobility in this country.

    For two recent discussions from differing viewpoints on these questions, have a look at LP and scepticlawyer.

  • 7
    Andrew Norton
    May 18th, 2010 12:14

    DD – Indeed, I have persistently argued that by year 12 class explains very little of the observed differences in SES participation, and higher ed policy will only make a marginal difference, mainly by making places available to people historically excluded by the quota system (class doesn’t seem to have anything to do with application rates, but because entry is usually merit based the weaker school results of low SES students has seen them disproportionately affected by quotas).

    There’s nothing wrong with this survey in that regard – it provides valuable information showing that spending additional money on low SES students who are already at uni will produce very low returns. If we had evidence-based policy in higher ed, which of course we don’t, the money could be saved or spent elsewhere more effectively.

    Unsurprisingly I don’t agree with you on private schools. Indeed, less funding for private schools would make them even more elite and exclude more people from the institutions that most effectively educate young people – pretty much what happens in England by the way, where a low % of students attend private schools.

    Hilariously, the LP software identified a post on England and produced an ad for Savile Row suits amidst grumblings about the English class system.

  • 8
    Peter Patton
    May 19th, 2010 09:29


    You’re probably the best person to confirm this, but my theory is that the most important factor in education success is a child’s IQ. In the absence of the sort of compulsory IQ testing that was a feature of the education systems a generation ago, parental income/SES would be the best proxy of that child’s IQ.

    If this correct, we might ask why government bothers micromanaging as much as it come to.
    Is there any data?

  • 9
    Peter Whiteford
    May 19th, 2010 12:01

    I don’t recall compulsory (or even non-compulsory) IQ testing in Australian schools a generation ago (when I was at school).

  • 10
    May 19th, 2010 12:16

    Peter Patton: Presumably your implication is that parental income is correlated with parental IQ, and parental IQ is correlated with the child’s IQ; thus parental income should be correlated with child’s IQ (if I’ve misunderstood you, apologies).

    I believe that this is unlikely to be the case, because there just isn’t evidence that either of your assumed correlations really exist; certainly not as strong correlations, anyway.

    Additionally it has been shown that IQ tests are strongly culture-specific; their accuracy as a general measure of innate intelligence is believed to be poor.

  • 11
    Andrew Norton
    May 19th, 2010 13:21

    Parental IQ and child IQ are correlated, though I cannot off the top of my head remember the level of correlation. And IQ is strongly correlated with educational success. In that sense, arguments that it is ‘culture-specific’ don’t really matter except to side-debates about international or inter-ethnic comparisons; it is predictive of things that are regarded as important in Western cultures.

    However I am not sure that it is a more useful test than specific tests for uni admission etc. Probably not, given its relatively limited use. Like Peter W, I am not aware that IQ tests have ever been widely used in Australia.

  • 12
    May 19th, 2010 14:54

    “and IQ is strongly correlated with educational success”
    I think you’ll find it’s only moderate actually, and that doesn’t even take into consideration “under the drawer effects”, i.e., those not reported because they don’t find an easily publishable result. Even this doesn’t tell you what you want to know, because part of your IQ score is no doubt based on your education achievement, so there is causation both ways (i.e., education affects IQ and vice-versa).

  • 13
    Andrew Norton
    May 19th, 2010 15:23

    It’s a while since I have read up on this, but I thought it was high, but this is more your area than mine. But there are two questions here: do IQ tests reflect innate intelligence, and do they predict performance? Whatever the answer to the first question, IQ tests may be useful for some purposes.

  • 14
    Andrew Norton » Blog Archive » A hopelessly flawed university ‘equity’ policy
    May 19th, 2010 15:48

    […] As we have been reminded this week, the core assumption of the policy, that low SES students are particularly in need of additional […]

  • 15
    May 20th, 2010 04:56

    “But there are two questions here: do IQ tests reflect innate intelligence, and do they predict performance?”
    I think that the only people that don’t think that there is a genetic component which influences scores on IQ tests are left-wing academics that are not in touch with reality (you’ll find some), and the same would be true of IQ and test performance. One reason the correlation with grades isn’t high is because, even after controlling for environmental influences which have reasonable effect sizes (e.g., teachers, parents, peers, etc.), there are other variables, like conscietiousness, that are important, and no doubt stuff that people don’t measure or skills IQ tests don’t measure well (e.g., sustained attention). None of this is an argument to say that IQ tests are the be all and end all of everything as some people like to believe, but just that if you create a measure which measures innumerable different types of mental processes, it isn’t surprising that it correlates with other tasks that use them, and also that people have different genetic dispositions for doing different types of things (many of which are probably quite weak).

  • 16
    Peter Patton
    May 22nd, 2010 09:45

    These studies that allegedly show public school kids get better marks than private school kids with the same TER/UAI at uni are surely badly structured.

    1. While they ‘blame’ the UAI/TER differential on the differences on the private/public divide, they do not ‘blame’ the university for any differences that occur there. It is not the purpose of high schools – public or private – to train kids to get high marks in 1st year Commerce! Their job is purely to help kids get high HSC marks. Then their job is done.

    2. The HSC is designed explicitly as a very competitive qualification to rank students, in order to allocate a limited supply of university spots in different degrees. University is nothing like this. Once you get in, you can cruise, and just get your degree. Sure, you can choose to get all competitive, and try to get Distinctions and above, but unlike the HSC that is not the point of getting a degree. So it is meaningless to compare the first year Commerce grades of private school kids vs public school kids.

    3. Quite possibly, the private school kids already have a large social network and/or confidence, so quite deliberately party and network. So, maybe the difference these studies are really picking up are not the quality of teaching/academics/student abilities, but the different social norms inculcated at public v private schools.

    4. It would be interesting to see a split on those who end up with either 1st Class Honors or Distinction + GPA/WAM.