Transcript of National Centre for Longitudinal Data Seminar

Speakers:

  • Dr N Biddle – Centre For Aboriginal Economic Policy Research
  • Ms F Skelton – National Centre For Longitudinal Data
  • Ms L Bennetts-Kneebone – National Centre For Longitudinal Data

Facilitator:

All right, welcome everybody.  What a lovely turnout this is, to what is the inaugural event for the National Centre for Longitudinal Data, which was formed in late 2014.

Welcome to all our DSS colleagues, both here and also in the States and Territories.  We've also got some friends from other departments that are present today, both here and also virtually.  We've got representatives from PM&C and Health and a couple of other agencies, I understand.

If I can just ask the attendees in remote locations to mute your mics—if you haven't already—that would be great.  Thank you.

We're meeting today on many traditional lands across the country, and I'd like to acknowledge the traditional owners of those lands on which we're meeting, and to pay my respects to elders, past and present.

If I can also just ask that attendees—particularly those present here in Canberra—can put their mobile phones on silent as well.

We will be taking photos, as you can see from my colleague Frank, down here.  He will try and mainly be taking photos up in this direction, but if anyone has any issues or concerns with that, please just give Frank a bit of a tap and he'll make sure he doesn't include you in any of those photos.

The agenda for today, a little bit of a welcome and introduction, and it would be remiss of me not to take the opportunity to spruik the DSS Policy Office, and the National Centre for Longitudinal Data (NCLD), at the start of the session while I'm here.  Then we've got Dr Nicholas Biddle, who will be doing a presentation for us.  He's from the Centre for Aboriginal Economic Policy Research at the ANU. Then we have two NCLD colleagues, Fiona Skelton and Laura Bennett-Kneebone, who will each be conducting presentations.  We will be holding questions over until the end, and we will then be having what will hopefully be quite an interactive panel discussion at the end.  So, I would encourage you to be taking notes and jotting down questions as we go through today, and we can run those at the end.  We will also be calling on our friends out in the States and Territories and those who are virtually present, if they have any questions as well when we get to that point.  I will ask you to give us a bit of a visual signal, a bit of a wave, and we will then know that you've got some questions there as well.

So, the DSS Policy Office within DSS is a critical enabling area to support DSS staff and their policy capability.

I would encourage you, if you haven't already, to go and have a look at the intranet.  There's some really good information on the intranet and some really good resources that relate to policy.  It's broadly broken up into four key areas, the first of which relates to policy approaches, and you will find there information and key formats about how DSS provides policy advice.

The second area of the intranet relates to policy capability, which is really about how to improve social policy development skills, and again, there's a lot of resources and some frameworks there that are quite useful.

The third area relates to the policy environment, so looking at within the context that we, as officers, develop policy in DSS, what are the local and what are the global policy constructs and constraints and environmental factors that we are conducting our work within.

And fourth, and quite close to our heart at the NCLD, is the policy evidence base- how to research evidence and how evidence can then feed into your area of policy interests. So I would encourage you all to have a look on the intranet for that.

And there is a fifth area which really outlines the DSS approach to policy advice.  You won't be able to read it there but it's just to show you, give you are bit of a look and feel about what it looks like, and again I would encourage you to have a look at that on the intranet as well.

So, spruiking for the Policy Office done, spruiking for the National Centre for Longitudinal Data commences.

So, I'm the Executive Manager of the NCLD.  The NCLD was established in late 2014 as I mentioned, and it really brought together four longitudinal studies that were being conducted in DSS at that time.  There's the "Building a New Life in Australia" study which is the Longitudinal Study of recently settled Humanitarian Migrants.  There's "Footprints in Time", the Longitudinal Study of Indigenous Children, "Growing up in Australia", the Longitudinal Study of Australian Children, and the HILDA survey, the Household Income and Labour Dynamics in Australia survey. So, we now manage all of those studies within the NCLD.

We also do some other things, so the NCLD's mission is to promote an evidence base that inforMs policies and practices to improve the lifetime well-being of people and families in Australia.  So, part of that is managing those four studies, but in addition to that, government through Cabinet, has commissioned us to develop a framework for government investment in longitudinal data which will be kicking off very, very shortly, and will be delivering by August next year.

Also, it is producing and managing quality longitudinal data sets and encouraging their use.  Thirdly, it is fostering collaboration between longitudinal data developers, researchers and policymakers, and part of these sessions are really to fill that brief. And fourthly, to facilitate broader use of longitudinal data both within government, the research community, and other relevant policy influences as well across the country.  So, that's us for the NCLD.

So, I would like to now introduce Dr Nicholas Biddle.

So, Nick, as I mentioned, works at the Centre for Aboriginal Economic Policy Research at the ANU, and will be speaking to us about the importance of engaging Indigenous children in early childhood education and care, and Nick, for some of his presentation today, has used some of the data from the longitudinal studies that we manage within the NCLD.

In addition to his role at the ANU, Nick's also a Deputy Director of the Australian Centre for Applied Social Research Methods or ACASRM, Graduate Convener and Honours Coordinator for the Centre for Aboriginal Economic Policy Research at the ANU and a member of the College of Arts and Social Sciences Research Committee.

Nick is currently working on the Indigenous Population Project, which is funded by Commonwealth and State and Territory governments, and he recently completed a research fellowship for NCVER, the National Centre for Vocational Educational Research, and he's previously held a senior research officer and assistant director position in the methodology division of the Australian Bureau of Statistics.

So, I would invite you now to join me in welcoming Nick.  (Acclamation)

The importance of engaging indigenous children in early childhood education and care

Dr Biddle:

Thanks for that introduction.

So, I would also like to begin by acknowledging that we here, are meeting on the lands of the Ngunnawal people.  People out there are meeting on the lands of many Indigenous people.  I pay my respects to Indigenous elders, past, present and future and recognise the role of Indigenous Australians in Australia, present and hopefully more strongly into the future.

Thank you everyone for coming along.  There's a lot of familiar faces here, including from the ANU and from Turner Preschool and from many other ones, so, thank you for coming along.

The work which I'm going to present, as mentioned, is partly funded by the Prime Minister and Cabinet, but also has a range of supporters, including our Good Start Early Learning, and as mentioned, the Department of Social Services, through the provision of a fantastic longitudinal survey, the Footprints in Time longitudinal survey of Indigenous children.

I should also make clear that this work comes from a range of papers, including work which was co-authored with Lilia Arcos Holziner, who is here today, Talia Avrahamzon, who is a DSS employee, but working with us at the ANU, as well as Naomi Priest at the ANU.

So, this is what I'm going to talk about: I'm going to start very briefly with the research and policy context.  Then, I'm going to talk about some outcomes of early childhood education and care, to what we can find with the data that we have available to us.

Then, I'm going to talk a little bit about some of the other factors which influence Indigenous child outcomes, to try and take a little bit of a different focus than just an area which government focuses strongly on, but I want to take a little bit of a focus on is an area which government maybe doesn't focus much on, and then give some concluding comments and policy implications.

So, as a researcher and as someone who is very interested in the work which you guys do, both in DSS and more broadly, I think it's quite important to put the analysis, the data findings, in that policy context.

So, I think it goes to the constraints which are there in terMs of how we can use this data to inform policy, but I think it also highlights where we should be focusing in our collection of data, in our use of data and our dissemination of data.  So, I'll start very quickly with the policy context.

Now, I think it's vitally important to think about the policy context in terMs of early childhood education, because it has so many names, so many cut-off dates, and for those from different jurisdictions, it must be ridiculously confusing, even more so than for someone like me who has kind of been working in the area for close to a decade now and still loses track of what's called what, where it's called what and who starts what.

So, I come from New South Wales.  I grew up in western Sydney.  I live in the ACT.  So, I'm going to use what we talked about, how we frame things in New South Wales.  But, others use different names for different types of early childhood education.

So, what I am basically focusing on is the year before full-time schooling, so the participation in either preschool, so that's early childhood education and care in a formal education setting, or a long day care centre, many of which have preschool components, in the year before full-time schooling.  In New South Wales we call that "preschool."  Other places, prep, pre-prep, many, many other names, some of which have dropped off here.

And then, the effects, or at least how that anticipation is associated with outcomes in kindergarten - so that's the first year of full-time schooling - and then years one, two, three and so on and so forth.

Now, the other reason why I wanted to raise this is because preschool, early childhood education and care, is in many ways a State and Territory responsibility, and that means that the policy context or the research doesn't just need to take into account Commonwealth policy, but also the policies of each State and Territory, and the data and evidence suggests that that policy context, those different ways in which preschool is funded, where it's located, has a large effect on both participation and outcomes.

So, the other policy context which I think is quite important is the Productivity Commission's inquiry into childcare, but also more broadly, into early childhood education and care.  I think the Productivity Commission report is, in many ways, framed around this statement from the Productivity Commission, but also makes heavy use of the work of Heckman, the University of Chicago economist, who has done a lot of work on early childhood education investment, as well as the outcomes.

So, according to the Productivity Commission, the benefits from participating in a preschool and preschool for children as development and transition to school are largely undisputed.  There also appears to be benefits from early identification of and intervention for children with developmental vulnerabilities.

Now, as an academic, when I see something which says something is largely undisputed, that gets me going, because, nothing is undisputed, and, in particular, the specifics are very rarely undisputed.

So, yes, a type of investment which has been studied in one particular context has been found to have one particular effect on one particular population subgroup.  Does that mean that for all groups that investment is going to have the same effects?  Does that mean that that investment has greater effects than another investment?  Does that mean that those benefits outweigh the costs? And there might be costs, both financial costs to the government, but financial cost to the individual, financial cost to the family, but, more broadly, some of the social costs.  So, yes, things are undisputed or, yes, they may be undisputed but that doesn't mean they're not disputable.

Why I think Heckman is important, is that he focuses on not just the effects of early childhood education on literacy and numeracy, but on a much broader range of outcomes, not just cognitive skills but also non-cognitive characteristics.  And by that, he means things like persistence, self-control, the ability to regulate oneself, the ability to look forward, plan forward, and make decisions which affect future outcomes.

So, yes, it might be undisputed, but that doesn't mean that there's not a lot of specifics which are worth focusing on.  And one of the specifics is the potential effects on Indigenous child outcomes.

So, what I've got up here—and sorry about the tables, we can send the slides around or make them available for people to have a look at.

So, why is it particularly relevant to look at the benefits of early childhood education for the Indigenous population?  Well, for a couple of reasons.  One, it's because the focus on early childhood education in the Closing The Gap framework, which drives much of the Commonwealth policy agenda around Indigenous Australians, the headline target is Closing The Gap In Life Expectancy, but numerically education targets dominate.

So, the first target around education is ensuring access for Indigenous four-year-olds in remote communities to early childhood education.  Okay, let's think about that target for a little while.

Okay, so there's a focus on access.  Now, what do you mean by access?  Is it just having a preschool there or is it having a preschool there which is suitable for the needs of those children?  Is it about geographic access or financial access?  Sure, there might be a preschool there, but if the costs are high, are those kids going to participate?  Okay, maybe the costs are low, maybe the curriculum is okay, but what about the relationship between or the experience of those kids in preschool?  Is that going to be positive or negative?  So, access is important.  You need one there, but it's not the only thing which drives people's decisions, and it's focused on remote communities.

The implication of that is urban communities, are fine.  That's all taken care of.  There's a preschool down the road.  No need to worry about early childhood education and care in an urban or regional context.

Now, clearly setting up the straw man, it is easy to knock the straw man over, but these targets guide both policy decisions but also policy framing.  So, it's quite important.  And the final thing is, despite the limitations of that target, it still wasn't met.

And the other thing to think about with regards to the targets, why the policy context is important, is because the way they are set up implies that flow-through, so preschool participation, which leads to improved school attendance, which leads to reductions in the gap between Indigenous and non-Indigenous kids in reading, writing and numeracy, which then leads to Year 12 attainment. So, there is an implied progression, or an implied kind of sequence, kind of a life course approach to kids' outcomes, and we need to question the strengths and the specificity around those links.

Okay, so that's why I think it's vitally important to use data like the LSIC and other data sets to think about who is participating in preschool, what are the outcomes for those who do participate, and what are some of the other barriers to education outcomes?

Now, it's fair to say that much of the research in preschool, at least within the quantitative side of things, is what the benefits of early childhood education is, and that makes sense, because we're thinking about where to invest.  And I think the assumption, at least implicitly, is that if you can show the benefits are high, then policymakers and families as rational investors in child outcomes, will ensure 100 per cent attendance.  So, large benefits will mean the government will invest and the parents will send their kids to school and everything will be rosy.

But, we need to think about well, even if there are large benefits, there might be still other reasons why kids aren't attending.  So, understanding the decision to attend, especially taking into account what we know about human decision-making, things like loss aversion, high discount rates, the effects of defaults, the broader insights about why people make decisions, that tells us why despite these large benefits, people might be making a particular decision.  So, as I said, it might be that the distances are too large, or the costs are too high, or there might be other competing uses of that child's time which might have larger benefits.  So, we need to think about that decision to participate.

And I'm not going to go into a lot of detail about that research, but, just to frame a bit of the discussion.

So, according to the 2011 census, 63% of Indigenous kids aged 4 to 5 years, who were not attending infants primary, were attending preschool.  That's compared to about 72% of the non-Indigenous population, and 63% of the Indigenous population in 2006.

Now, this is three years into the Closing the Gap targets, and it's really the only data we have which is internally consistent, where we have information on those who are nationally representative on those who are attending and those who aren't attending.  It's different to using admin data on who is attending and dividing by the population.  It's the only consistent data we have on participation rates.

And what have we found?  Basically, no change through time over the five years, from 2006 to 2011, and in some ways, if you control for the change in the distribution of the population, a reduction in participation rate between Indigenous and non-Indigenous kids.

What does that mean? Well, what does that mean for us?  Well, it means that even if there are large benefits - which I will go on and talk about - clearly Indigenous families or the families of Indigenous children, are making a decision to not send their kids to preschool, and there's a lot we need to do about understanding why they are making that decision - and I'll get to that in a little while.

But, the other thing which that data says is, okay, what's driving that?  Is it Indigenous specific effects or is it because of other characteristics of the Indigenous population?

So, when we control for geography, there's still a large difference between Indigenous and non-Indigenous kids.  So, remember back to that target.  It's about access in remote areas.  Well, even when you look at an Indigenous kid in Sydney, in Tuggeranong or an Indigenous kid in Dubbo or an Indigenous kid in Fremantle, they are still less likely to attend preschool than a non-Indigenous kid, so geography isn't the only driver of that participation.  But, once education and income as a family are controlled for, then Indigenous kids are actually more likely to attend preschool than non-Indigenous kids.

So, we need to focus on other barriers to participation.  It's not just where kids live, it's their family income, it's the parents or the carers' prior experience with education.

I mean, if you dropped out of school in Year 9 or Year 10, then you clearly didn't see the value of education, probably because it actually wasn't very valuable for you at that time because of your negative experiences.  So, that's not putting the blame on those parents, it's the intergenerational effect of prior negative experiences.  You are less likely to want to send your kids to preschool.

Okay, so, the decision's important.  The decision about whether to send Indigenous kids to preschool or not is quite important.  But, what about the relationship between early childhood education and child outcomes?

Now, despite the statement made by the Productivity Commission and the research of Heckman, the evidence on the benefits of early childhood education in Australia isn't conclusive.  So, my four-year-old, he goes to preschool. We sent him to preschool when he was three.  So, in some ways my actions show that I think that the benefits outweigh the costs.  But if I'm totally honest most of the benefits are really for me.  He's at preschool so I can be here.  He's at preschool so my wife can complete her degree.  He's at preschool because it's free, and someone is looking after him, someone who I think is actually a fantastic educator.

Now, I think my son would have been fine if his mother had worked and I had looked after him for this year, and last year.  It probably would have been better if his mother had looked after him. This is nothing to do with gender, it’s just because she is much better than I am.  So, he would have been fine if he didn't go to preschool.

So, we need to think about this for Sajan and other kids. Are there measured differences between someone who did attend, or didn't attend, because of who he is or because of his family background, or because of the decision we made?  Or, more importantly, would the money spent by government on our kids, or kids like them, have been better spent on other prograMs which had a larger benefit?  So, we need to look at the causal influence, not just the difference between someone who did attend and didn't attend, but whether it is education itself.  And as I said, the data is inconclusive.

When we looked at the data from the Longitudinal Study of Australian children, we controlled for a range of risk factors for the general population, and we found that there was a small effect of participation in preschool on maths outcomes. , This is shown in the work I did with Robins and Purdy (?), who are now at Griffith University..  There was no effect on the general outcomes on the strengths and difficulties questionnaire, the reading outcomes, and there was a negative effect when you control for the background of those kids who do attend from participation, for attendance in a long day care centre.

So, when you control for the characteristics of those who do attend, sometimes you find positive benefits; sometimes you find negative benefits.  It depends on the measure and it depends on the type of education.  So, we need to kind of unpack that.  Is it because those who are attending would have had better outcomes anyhow?

So, there is a really great headline which I saw for an online article, which is basically that if you are reading this article and wondering about whether you should send your kids to preschool, then you don't need to send your kids to preschool.

Okay, what about for the Indigenous population?  What we did is we looked at short-term outcomes, so that's the outcomes of those kids in their first or second year of full-time schooling; medium term outcomes; and I'll get to some long-term outcomes in a second.

So, we used data from LSIC, we pulled data from the kid cohort and the baby cohort - and I'm not going to go into a lot of detail about LSIC because there are people here who are much more knowledgeable than I am - but basically what we did was we took participation when those kids were aged 3 to 5, outcomes when they were aged 5 to 7, and outcomes when they were aged 6 to 10.

And what we were very importantly focused on was outcomes which were either reported by the teacher, reported by the child, or came from tests.  We downplayed outcomes reported by parents because it's those parents who were making that decision.  So, when those parents made that decision to send their kids to preschool and then you asked them about those child outcomes, well then is it the outcomes themselves or is it the justification of that decision, so we downplayed those outcomes.

We did developmental outcomes, reading and literacy as well as abstract reasoning and maths ability, and importantly, what we did is we looked at the difference between those who did and didn't participate in preschool, and then we looked at the difference controlling for a range of background characteristics.

So, the beauty of longitudinal data - and I don't have to sell this to this audience - is that we can see what are the characteristics of those kids before that investment in early childhood education or that participation, and then for someone with the same characteristics who did attend preschool and who didn't attend preschool, is there still a difference?  Now, that is not experimental.  We haven't randomised participation or not, but at least it's controlling for observable characteristics.  And I'll present both results.

Okay, so this is what we found: just to summarise, if that bar is above the line, it means those who participated in preschool had better outcomes than those who didn't, or higher values than those who didn't participate in preschool.  If it's below the line, it means they had lower values.

Now, for SDQ what you want is something below the line.  Something below the line indicates better outcomes and something above the line indicates worse outcomes.  But for most of the other ones above the line it indicates better outcomes.

Now, if that bar is hollow that means we can't conclude that that difference is statistically significant.  So, we don't know whether it's just randomness in the data as opposed to, something which is representative of the total population.

So, what did we find?  Well, for short-term outcomes - so that's one or two years after participating, or two years after participating in preschool - almost no association, especially when we controlled for background characteristics.  And, if anything, some of them were going in the opposite direction. So in terMs of SDQ, those who did attend preschool, higher values than those - - or worse values than those who didn't.  Pro-social, it's a large difference, but it's not significant.

Whether the child is happy at school - and that's the child telling us themselves - if you didn't control for the background characteristics, those who attended preschool actually had lower outcomes.  But, there was a very large association with the vocabulary test, and that's basically a quarter of a standard deviation.  These are all yes/no and for most of them it's standardised to the mean of zero and a standard deviation of one.  So, this is a pretty large association.

So, what that basically shows is that in the short-term, two years after participating in schooling, in preschool for most of the things we were interested in, there's not much of a difference between those who attended preschool and those who didn't.

When we go to medium-term outcomes, things change, and that kind of fits with the work of Heckman, most of the difference between those who attend preschool and those who don't in terMs of cognitive ability, is driven by the fact that those who do attend preschool would have had better outcomes, than the others, but it's some of those non- cognitive skills and those long-term skills which seem to be affected by preschool participation.

So, now, when those kids are aged 8 to 10, those who attended preschool have significantly lower SDQ scores, ( and lower values are better).So those who attended preschool, had significantly lower SDQ scores than those who didn't, especially when we control for background characteristics.

What about achievement?  Well, now we're starting to find much larger associations.  Reading achievement, large, not statistically significant, but this data is only from one cohort of those kids, so there's much less certainty around these estimations, but still, a large difference in reading achievement by the time they are 8 and 10. And what this shows to me - and it fits in with Heckman's work - is that the short-term effects are on non-cognitive skills which then are scaffolded and add to the long-term reading and writing.  So, it's not that those kids start school better able to read or not better able to read, it's that they start school with those other skills which then help them learn to read and do maths and so on and so forth.

Okay, but the other point about this data is that we, as policymakers, if we just looked at the short-term, say okay, I've got a three-year program.  My three-year program is just the basic funding cycle.  My three-year program is I'm going to spend a whole bunch of money on getting these kids into preschool and, I'm going to do really well, I'm going to randomise, I'm going to have, a large sample and I'm going to follow these kids and test them in two years' time.  And you can test them for kind of reading and a range of outcomes.

And what you might find is that there's no effect.  And that's what happened in many of the trials in the US. They found no effect in some of those short-term outcomes so preschool was a failure.  Basically, this study, which almost all our Productivity Commission style statements are based on was labelled a failure after the first four or five years.  But it's only when you start to follow those kids five, six, seven, 10, 20 years down the track that you find those large differences, and differences above and beyond those measures of cognitive ability.

Now, I'm going to skip through the effects on longer-term outcomes, mainly because it is using a dataset not from DSS, which is nowhere near as interesting.  But, just to highlight, what I wanted to do with this - and as I said, I'm more happy to talk about it at a later stage - is those longer-term outcomes in PISA in the longitudinal survey of Australian youth, to the extent we can control for background characteristics, support some of these effects.

So, in the five minutes which I have, what I want to talk about is some other factors influencing Indigenous child outcomes.  And the reason why I want to do this is to reflect on those Closing the Gap targets that focus on early childhood education, attendance, literacy, numeracy. Focus on the targets can blind us to some of the other things which affect both the broader well-being of kids, but also the effect on those things we do care about.  So, I'm going to focus on the example of discrimination because it is used in LSIC.  So, the same data but now our main influencer or our main explanatory variable is whether the carer of that child reported some form of discrimination.  So, it is whether that family or that child experienced discrimination because of their Indigenous status, and once again we focus on child reports, teacher reports, assessment scales and a range of other characteristics.

The results are presented in a very similar way.  This is just the distribution of experiences of discrimination at Wave three, Wave five, Wave four, basically any discrimination experienced in the years leading up to that child in Wave six.

Once again the model one is difference in discrimination only, and model two is controlling for other characteristics of those children.  So, what do we find?

We find a very large association between the experience of discrimination and that child's self-perception.  So, self-perception is basically their views of themselves.  It's kind of an index, the views of themselves, and also the ability of those kids to control their own outcomes.

Subjective well-being at school - less of an effect.  Cultural identity -  large association.  Small samples, but still, those who have experienced discrimination have lower cultural identity, much lower test scores in maths.  So, something which is quite removed from instruction at school, still has a very large effect on those things which we're interested in.  Less of an effect on reading, but still basically the measures which we have, we find have either a large effect of discrimination or a small uncertain effect which is going in the direction to suggest that discrimination affects child outcomes.

So, the reason why I wanted to highlight that is that in our focus, early childhood education appears to have those larger returns for Indigenous kids. But let's not forget about those other things which affect whether those parents want to send their kids to school, whether the kids are happy at school.  They also matter, and our policy needs to take those into account.

So, in the one minute, two minutes, that I have available I'll try and wrap things up.

So, what are the main findings?  So participation, which is not going in the direction that we would have liked, and that's slightly different to when we use administrative data. So different data tells different stories.

Geographic distribution explains only a small percentage of the difference.  Socio-economics are more important.

Outcomes: in the shorter, medium-term, there's a large degree of uncertainty, so if we evaluate things in the short or medium term then we might conclude no effects, but medium-term outcomes do have a stronger association and what I didn't talk about is those large differences when that child reaches 15, and they tend to be larger for Indigenous kids than non-Indigenous kids.

So, other findings - which I'm happy to share with you - - and this is the paper which I did with ALIA - there is no evidence for dose-response in preschool hours, so it doesn't matter whether that kid is participating for 20 hours.  At least 10 hours seeMs to have a positive association.

Childcare has a small and insignificant effect on most outcomes, but there is evidence which fits with the evidence of LSAC from the general population that large number of hours, participating in long day care centres, controlling for other characteristics, is associated with worse outcomes for Indigenous kids.  So, the type of institution matters, and that's a very important finding, I think.

Okay, so finally, this just kind of fits within the mission of the Centre.  We need better data, and we need to keep pursuing our quest for causality.

So, according to the closing in our Clearinghouse, there's been no rigorous trials or evaluations of early childhood programs.  So, we are using observational data. It is  important, but not as strong as evaluating particular programs.  And we can control for observable characteristics but we can't control for those unobservable characteristics.

So, we need to supplement our data with well designed, ethical, trials which say well, what type of preschool matters, what type of preschool induces Indigenous kids to participate. Although there are limitations of those types of trials, ethical, inclusive trials, with community support, are needed to supplement that longitudinal data to find to figure out what works to get kids into preschool and what works to close the gap in school readiness and life chances for kids.  Thanks.  (Acclamation)

Facilitator:

Thank you, very much, Nick. That's great.  You generously put your email address there so we can spam you as well.

Some really great insights from Nick, I'm sure you would agree, and particularly for me thinking about the decisions that we, as policymakers and policy influencers make, and how these decisions then impact all the way down the implementation of policies and then how we can potentially evaluate policies, is very interesting to think about, and also interesting to think about how we can use data to understand these causal influences.  And I love that quest for data and that quest for causal linkages. I'm going to borrow that or rather, steal it.

And I think it's important for us to think about how we as policy influencers can keep sight of those longer-term objectives in this, which can sometimes be quite a short- termism environment.  I think there's some really great insights there.  Thank you.

So, I'd like to introduce our next speaker now.

Fiona Skelton who, as I mentioned earlier, works for us in the Centre for Longitudinal data. Fiona will be talking about predictors for a positive start from Footprints In Time, which is the longitudinal study of Indigenous children.

Fiona is the Assistant Section Manager of the data and design team within the NCLD.  She's worked on all aspects of Footprints In Time since it began back in 2003, including a lot of very early community consultations and ongoing community consultations, the design of the study, community trials, including in the Torres Strait, quantitative and qualitative pilots and also interviewer training, so she's really worked on the breadth of that study since its inception.  Fiona currently manages the content development for various waves of Footprints In Time.

So I would ask you to join me in welcoming Fiona.  (Acclamation)

Predictors for a positive start from footprints in time

Ms Skelton:

Thanks.  If anyone else needs to go, that's fine.  I won't be offended. I'm happy if you came to see me.

I would also like to acknowledge the traditional owners of the lands on which we meet, the Ngunnawal people, and the traditional owners of the lands where you are today for our people who are in other offices.  I pay my respects to elders past and present and future, and to the continuing cultures of Aboriginal and Torres Strait Islander peoples in Australia.

This presentation was originally delivered at the ARACY Early Childhood Conference in Tasmania earlier this year with Professor Maggie Walter, who is an Indigenous woman from Tasmania, and on our Steering Committee, and also the Vice-Chancellor for research at UTAS.

So, I also want to pay my respects to the Mouheneener people, who are the traditional custodians of the lands in Hobart.  As Maggie said at the Tasmanian conference that there are no living descendants of the Mouheneener people in Tasmania anymore, but Tasmanian Indigenous people in general are custodians of the lands.

And the other thing I should mention is, I don't think Adam actually introduced his name, so if anyone is wondering who is this new person, this is Adam Rowland.  And we also have a new director I'm going to point out in the audience, Anthony, who can wave over there, who is managing the business part of the National Centre for Longitudinal Data and all our interviewers.

Right, I am going to give you, unlike Nick, a bit of an LSIC thing because it's my passion and I've got to try and get 10 years into five minutes, which - - actually it's 13 years now.  I can't count.  That's why I do statistics.

Who's involved?  So it's funded by the Australian Government as an ongoing study, Footprints In Time, guided by a steering committee that's Chaired by Professor Mick Dodson, and has been since 2003.  It includes Professors Karen Martin and Maggie Walter on the steering committee, and many of you will know that Karen Martin has worked in early childhood for a long time.

We manage the study in the National Centre for Longitudinal Data and we employ the interviewers ourselves.  For the other longitudinal studies that we manage, interviewing is outsourced, but our researchers, interviewers, are employed as full-time Indigenous staff, usually from the areas in which they do interviews, and I'll show you a bit of map about that a bit further on.

I'm not sure if any of our interviewers have joined us today.  Hi, if you are there.  They are largely responsible for the fact that we've got really good retention rates in LSIC, greater than 80% every year, and hats off for the interviewers for that.

Okay, quick design overview: we started with lots of consultation around Australia, and community engagement, and now we have an annual quantitative data collection with more than 1,200 parents and carers, and at the moment our interviewers are hoping to get to 1,250 for this year before Christmas - - always a challenge, and they do a great job.

The children themselves answer questions and do assessments, which you have seen from Nick, who has used some of the data from that.  We do interviews for questionnaires with dads in the study, and we send a questionnaire to teachers, which they then complete online or in questionnaire form.

Where are we?  The yellow circles on this map are where our original sites were, where we recruited the people for the study in the beginning, and you can see the faces of our interviewers across Australia there.

The red dots are where a lot of people have moved to, or the site was a little bit spread out to start off with. We are trying get to the people who are the red dots.

The blue dots are a little bit harder.  So, I understand we have Kalgoorlie, here today.  Hi Kalgoorlie. The participants that we may have in Kalgoorlie, it's a bit too hard and expensive for us to get to as yet.  If we get more people who move to Kalgoorlie maybe we will be able to.  So, they're sort of out of scope for us to keep following at the moment, though I did get to the Tasmanian one this year because I got to go to the conference, and I was the one non-Indigenous interviewer this year.  I was very thrilled.

Sample characteristics: our B cohort, our baby cohort, were born in 2006/2007, so at the time we are talking about Wave 6 I think on this slide, they are six and seven years old.

The K cohort, three years older, pretty much, and we have more babies than kids because we really wanted to get the early years. That 1 to 4 years of what's happening for Indigenous children was the bit that's really missing and unless we get another cohort, will be missing for what's happening for lots of Indigenous kids in Australia.

So, we followed the two groups as they grow up, as Nick said.  All the kids are Aboriginal or Torres Strait Islander or both, but not all their parents are, so then I have one parent who is Indigenous and one who is not Indigenous and by the time we get to Wave six it's about 18% who are non-Indigenous.

This next slide is just to remind you of some of the policy implications.  I'm not going to go into the kind of detail that Nick does, but what I would like to show you is since our kids were born, what are some of the policy changes that may have impacted on our families?

So, in 2004 there were the Palm Island riots after Mulrunji Doomadgee died. ATSIC got abolished after that.  Indigenous TV started in 2007.

In 2008, when we started our main wave of fieldwork, the Northern Territory emergency response came in.  I just want you to think about that, starting fieldwork and saying we're from the government…. be in a study.  Full credit to our interviewers that they persuaded them to.

The Closing the Gap targets were agreed.  Then came the apology.  Professor Mick Dodson, our Chair, become the Australian of the Year in 2009, and up the top I put a few things that may have impacted Indigenous people a bit more than others.

So, for example, I put swine flu up there because if you were Indigenous you were allowed to get a vaccination just by being Indigenous against swine flu.

I've mentioned the changes that happened with welfare reform, so when your youngest child turned eight you have to go back to work.  We have about 40%, depending on which wave you are looking at, of our parents who are sole parents.  Many more Indigenous parents are sole parents than in the Australian population in general.  So, a lot of the policy changes that we make, differentially impact Indigenous people.  And you can see our beautiful kids - in very small photos I'm afraid - growing up down the bottom, as all these changes are happening.

Okay, let's get to outcomes.  So, Nick used the SDQ before, the strengths and difficulties questionnaire, and for people who are not familiar with it, I just thought I'd put up a few of the iteMs on the SDQ.

The post social score is made up of five items, and one of those as an example is, is the child helpful if someone is hurt, upset, or feeling ill and the parent can answer certainly true, somewhat true or not true, which we would have liked to have yes, sometimes no, but the person who owns the copyright - and this scale is used internationally all around the world by lots of different cultures - insists that we use that wording, so we do.

Then we have the difficulties score, which is made up of four different scales with five iteMs for each, and you can see some examples of some of those, but they cover things like hyperactivity, emotional and conduct issues. 

Right, some of the independent variables that are used in the analysis:  so, we asked in a range of questions, this particular question, does your child's teacher understand the needs of children from an Indigenous background, we circled, and you can see in the purple bars the responses in Wave four which are only the K cohort children, so nearly 500 of them, and about 60% or so are saying that the teachers do understand the needs of Indigenous families very well or well, but there's also a large proportion who are saying no, and then by Wave 6 in the green bars you can see that's getting a little bit higher.  That's 1,115 respondents there, and of those, 25%, a quarter, are saying it's not done at all, the teachers don't get it.

Getting back to the SDQ, this chart shows you what happens to children's difficulties scores, children in both LSAC and LSIC, the LSAC children are in the sort of goldy-tan bars and the LSIC children are in the orange bars, and I should point out that this is wave 4 of both studies, so the LSIC children are around about four and around about seven and the LSAC children are around about seven and around about 11 and SDQ scores tend to reduce as they get older anyway. So don't take any notice of the difference between LSAC and LSIC, just look at the trend lines.  The more major life events you get, the more difficulties you get, and that's basically all this slide is trying to show you.  But, if you've got 5 or more major life events then your SDQ score is going to be four or five points higher than those who have not experienced the major life events.

And when I say major life events, I'm talking stressful situations for the families such as moving house, having financial problems, parent losing a job, having a baby - which is a great thing but quite stressful, that kind of thing.  Okay, let's have a look at multiple regression.

So, remembering as Nick said before, more difficulties is not a good thing, so this side is not a good thing, and I've borrowed from Nick's way of doing it. The bars that are hollow have no statistical significance, and the dark bars do.

So, you can see if - - and we're thinking children who are three or four here - - this is Wave three data - - what reduces difficulties.  What's good?  Vocabulary scores that are above average reduce difficulties.

Now remember, vocabulary scores are also what Nick is saying, are the short-term positive effects that you get from preschool, and I do wonder whether a good vocabulary then predicts your latest good scores for literacy.  Well, we actually have found that relationship, and lots of other researchers have, too.

If the parent is worried about the child's speech or understanding, more difficulties.  Sleep problems, more difficulties, though I think that's not significant in this now as it has been in others.  Watching lots of television - not so good.  Being a girl - girls just have lower difficulties scores, and that's the same in LSAC as well, it's not unexpected.  It's hard not to preen at this point but, you know.  I think for participants, the differences disappear as the kids get older.

Parent education, so parents with Year 12 plus education, their kids tend to have lower difficulties scores.

But, this is a really important one that I want to draw your attention to.  If parents' well-being - and by that we're talking mental health to a large extent - is good - and we have a continuous scale that we use for this - then children's difficulties scores are much lower.  So, strong mental health, the parent, much better mental health for the kids or social and emotional well-being.

Then the next one, which the Australian Council for Educational Research published in their paper, which you can find on the net - and if anyone wants details, you can ask me about it afterwards - being Indigenous is important to the parent.

So, we asked them a question about being Indigenous. The most important thing, central to who you are, important but not the only thing, something you want to find out more about or - and I can't even remember what the fourth one was.  But, if you take those two top ones and put them together and put them in the multiple regression then the parents who have got more, stronger identity with their Indigenous identity, the kids have fewer difficulties, which is interesting, even after controlling for other things.

Racism, you can see the direction, but we don't get significance in this multiple regression.  And living in a more isolated area is more likely to have higher difficulty scores. Okay.

In Wave one, we asked all the people who participated in the study, what is it about being Indigenous that will help your child grow up strong?  And then in the subsequent waves, with the help of Laura, did some analysis on all those answers that we got, 1600 answers on what it was about being Indigenous that will help their child, and we shook them down into about 12 items, and then we went back to parents and said what is it about Indigenous culture that you want to pass on to your child at this age?  And we used what they said was important about culture, not what we thought was.

And so I've just put up what were the first and second choices of most people because we asked them to pick five.  Some people said well, it's all important and we said can you pick five that are important for this age, and most of them were happy to do that.

I hope you can all read the bits at the bottom.  But just in case you can't, respect, family history, and pride and identity, hit the top things.  And if you're really interested in geographical variation about this and who picks what and where because yes, language is probably likely to be more important in areas where they speak language, we do have some site feedback sheets that compare all our sites with each other from about Wave three, so just send me an email and I'll send them to you.

So, what about when the kids are a bit older, so Wave six when the children are six years old and nine years old?  So again, say the child is a girl, fewer difficulties.  If the child has a health condition, more difficulties.  Major life events - and remember, that looks quite small on the chart but that's for every additional major life event you get about another half a point on the difficulty score.  If the parent has good mental health, that reduces things.

If the parent wants to pass on the cultural value of pride and identity as opposed to all those other things we mentioned, then the difficulty scores are lower, and if they want to pass on showing respect.  And again that question we asked about does the teacher understand the needs of Indigenous families comes up, that reduces difficulties too, and if you live in a more advantaged area, difficulties are lower.

Okay, what happens when we add other variables from previous waves to the model?  So, just those three ones at the bottom are different from the previous thing before.  Just about everything else stays the same in the progression.  And I see my words dropped off there, but it explains about 20% of the variance, I think. There was bound to be something went wrong with my slides at some point in time.

Those sleep problems, again if the child has had sleep problems, and we asked about that every wave, and about a quarter of the children do have sleep probleMs in every wave.  It doesn't tend to persist, usually.

If the child has been bullied or treated unfairly, so the questions that Nick was asking about, then you can see the difficulty scores are increasing.  And if the parent has good social, personal, cultural strengths from a series of questions that we asked, then difficulties are reduced.  And even though that's a tiny amount there, it is a significant difference.

Now, I want to get on to something really exciting - I can see you're all falling asleep so I'll have to try and get really exciting here - teacher rated outcomes.  This is a literacy measure that's used in the Longitudinal Study of Australian Children.  It comes from an American study called The Academic Rating Scale, I think.  And this is an example of some of the questions that are in the Academic Rating Scale.

So, the teachers get the questionnaire about the particular child and they answer this series of questions and then we tot up all the scores and there is a range of about 10 to 50 scores within them.  We've got both B and K cohorts here, so that contributes to the wide range of scores, but you can see, on the top line, contributes to classroom discussions.  We've got a quarter of the kids who are proficient at that.  By the time we get to compose a story with a clear beginning, middle and end, about 10% are proficient, and it ranges quite widely.

This is the exciting bit.  I can see you're all jumping up and down.  That's good.  I hope in the office out there someone is jumping up and down.

We asked - because the Department of Education asked us to - whether the school had implemented the Aboriginal and Torres Strait Islander cross-curriculum priority of teaching Aboriginal history and culture and Torres Strait Islander history and culture in the classroom.

So, in this case, higher is better.  And in a model that controls the parent education for child attendance, the child's hearing problem and the child's age in months, if the school had implemented the cross-curriculum priority then their literacy scores as rated by the teacher were on average two points higher, significantly different from the other, which I find very exciting, because there's quite a bit of data about the cross-curriculum priority and whether it's useful or not, and this tends to suggest to me that, yes, it is.

In conclusion, so kids have lower social and emotional difficulties scores in LSIC when they haven't experienced racist bullying, when the parents and carers see that the teacher understands the needs of Indigenous families, and when they can pass on the value of pride and identity and showing respect.  And teacher rated literacy scores are stronger for a focus on children where the schools have implemented the cross-curriculum priority of teaching Aboriginal and Torres Strait Islander history and culture.  And that's it.  (Acclamation)

Facilitator:

Thank you, Fiona.  Again, Fiona has got her email address there so we can spam Fiona at will as well.

So, some really important takeouts there again.  For me, that slide that had all the events that have happened since the LSIC study started, it's really a reminder that, the audiences, the policies that we develop and the prograMs that result from those that affect the Australians or some parts of the population that they are designed to affect, it's important to bear some of those contextual factors in mind, as we are implementing and evaluating programs, and what impacts they might be having on those populations.

Some very interesting detailed results in relation to impacts on children's difficulties, but also it's helpful in general I think to understand some of the impacts of some of these things on Indigenous peoples more broadly.

And for those that the subject matter might be a little bit less directly relevant, it's important to think about in your policy world how some of this thinking, you can apply more broadly to your specific policy problem as well.

So, to our third and final speaker today.  Laura Bennetts-Kneebone who will be talking about partner violence and looking at findings from LSAC, the Longitudinal Study of Australian Children and LSIC, the Longitudinal Study of Indigenous Children.

As mentioned earlier, Laura also works in the NCLD with us as a research officer.  She's worked on the design and data management of the Footprints In Time study for about eight years, as well as on the Longitudinal Study for Humanitarian Migrants for the past year or so.  And Laura has previously worked as a research assistant documenting the Victorian Aboriginal languages and tutoring at the La Trobe University.  So, please join me in welcoming Laura (Acclamation).

Partner violence and looking at findings from ISAC, the longitudinal study of Australian children, and lSIC, the longitudinal study of indigenous children

Ms Bennetts-Kneebone:

Good morning.  I would like to acknowledge the traditional owners and custodians of country throughout Australia and especially here in Canberra and their continuing connection to land, waters and community.  I pay my respects to them and their cultures, and to elders past, present and future.

I also want to thank the mothers and fathers who responded to survey questions on this difficult and sensitive topic and acknowledged Helene Shin and Helen Rogers, who worked with me on this analysis and presentation.

Back in April, the National Children's Commissioner invited Helen Rogers, our Director, to attend the Canberra roundtable on domestic violence on children to make a submission in relation to how children are affected by family and domestic violence.

We got to work doing some analysis of the great pool of longitudinal data available in LSAC, the Longitudinal Study of Australian Children, which is also known as Growing Up In Australia and LSIC, our Footprints In Time, which you know a lot about by now.

The Human Rights Commission offered us some direction for the research, suggesting a focus on the demographic characteristics of families affected by domestic violence and on children's outcomes.

Okay, let me tell you a little bit about the two longitudinal studies.  This is LSAC, a map of where our sample is in LSAC, the study of Australian children.

For LSAC we interview up to 10,000 parents and children from two age cohorts every year two years.  LSAC is designed to be nationally representative.  The sample focuses on young children rather than households, and excluded remote areas in Wave one, but some people have moved into them since then.

So, Wave five, the LSAC sample, has 64% of households in major cities, 22% inner regional, 12% out of regional, 1.2% in remote areas and 0.4% in very remote areas.

Like LSAC, LSIC also targets families with young children but only Aboriginal or Torres Strait Islander children and LSIC interviews up to 1700 parents and children from two age cohorts every year.

Although LSIC's is site based and not nationally representative, the LSIC Wave 6 distribution is actually similar to the national distribution of Indigenous people around Australia by remoteness category.  So, Wave five LSIC sample had 20% of families living in major cities; 27% in inner regional areas; 16% in outer regional areas; 9% in remote areas; 19% in very remote areas.  So, the yellow is where the LSIC families live.

The questions that you can see above, LSAC obviously has no direct measures of domestic violence, but has a couple of indicators.  There are questions like how often do you have arguments with your partner that end up with people pushing, hitting, kicking or shoving, and other questions like have you ever been afraid of your current partner, in amongst a third of questions about your relationship with your partner.

LSIC collected item 1 in 2010 and 2013, as well as a range of community safety questions, which included a family violence item.  So, that question you can see there, tell me which of these are a problem in the area you live: family violence; a very big problem, big problem, small problem, not a problem.  So, these are some of the main iteMs I'm using in this analysis.

These questions were asked of mothers and fathers in LSAC, and the first question was asked of primary carers who were predominantly mothers in LSIC.  These questions were only asked if the respondent said they had a partner living in the household.

This explains partly why I have selected the term "partner violence" for use in this presentation, so you can see these questions only relate to violence between partners.  They don't tell us to what extent children or other family members may be directly involved, and they do not tell us to what extent which partner is the victim or the perpetrator or whether both partners are oppressors in this case.

So, a quick look at the overall prevalence in the study,  so, you can see the make-up of the families.  So, in LSIC 42% of the families were single parent families; 52% were in relationships which reported no partner violence; and 6% were in relationships where some partner violence was reported.  That could be rarely, sometimes, often or always I think.

In LSAC, the family compositions look very, very different, so you can see that 13% are single parents, again, not asked the violence question; 83% reported no partner violence; and 4% reported some partner violence.

Now, the graph above excludes refusals, and it excludes reports from males.  It is based on mothers of the B cohort in LSAC and the K cohort in LSIC.  So, this is mothers with a six-year-old in the study in 2010.  So, although the samples are different, I tried to get them as close as possible for this.

A few caveats, as Fiona mentioned: there are some non-Indigenous mothers in LSIC.  There are some Indigenous mothers in LSAC.  And the differences between LSIC and LSAC are unlikely to be due exclusively to the Indigenous status of the families.  The LSIC population in general is much more disadvantaged on measures such as education, employment and community disadvantage.

Now, for example, more than 40% of the LSIC sample lives in communities rated as being the most disadvantaged in Australia using the SEIFA index.  In LSAC only 7% of the families live in the lowest decile if the data is not weighted.

If you are looking at this 4% and 6%, these figures may seem like under-estimates to you if you're hearing results quoted from the personal safety survey saying one in three or 38 and-a-half percent of women have experienced violence since age 15, but this is data about a certain point in time, so even in the personal safety survey, 5.3% of respondents experienced violence in the 12 months prior, so it's about on par.

A strength of longitudinal data is the potential to see whether things change over time.  In both cohorts of LSAC - - which is what this graph shows - - so the first three columns are the baby cohort - - not babies by this point, and the second three columns are the older cohort.  In both cohorts mothers who were afraid of their partner at Wave four were more likely to be single parents two years' later than mothers who were not afraid of their partner.

So, you can see in the first three columns the first column is people who are afraid of their partner at the first point in time.  So, green indicates they are still afraid of their partner; purple indicates that two years' later they are not or no longer afraid of their partner - maybe it's a different partner - and grey indicates the proportion who are no longer in a relationship two years' later.

We can see similar patterns for LSIC, even though the measure used, which is about arguments that end up with violence, and the time elapsed, which is three years, are different to LSAC.

So, mothers in a relationship characterised by some violence in 2010 were more likely to be single or still experiencing partner violence three years' later than mothers reporting no partner violence.

Of the 44% who say they are in a non-violent relationship three years' later, almost all us were still with the same partner.  Twenty-six percent were still experiencing partner violence and 30% were single three years' later.

So, you can see along the bottom is the people who were either - - had no partner, were experiencing no violence or what experiencing some violence in 2010 and the column shows you what was happening to them three years' later - purple meaning no partner violence and green meaning some partner violence.

I am just noting here, we don't know whether those who are not living with a partner experience partner violence because we don't sequence to the question in the survey if they are not living with a partner.

A range of demographic characteristics were significantly related to being afraid of your partner in LSAC or partner violence in LSIC.

The relationships above were seen in bivariate analysis and were not reported unless significant, and relationships not tested are left blank.

You can see the overall patterns here that show that remoteness is a predictor of partner violence in LSIC, lower SES, though using different measures, is a predictor in different surveys.  Fewer years schooling, particularly for the father, seeMs to be important.  Mothers' employment, made no difference in LSAC but was significant in LSIC, and fathers who are not employed in those families, there was a higher proportion reporting partner violence.  Mothers and fathers who had a medical condition were also more likely to report being afraid of their partner, and multiple financial hardship was highly significant in both surveys as a predictor of partner violence.

As I said, this is bivariate analysis, so in a logistic regression, restricted to LSIC couples in 2013, in which occurrence of partner violence is the outcome variable, and with SEIFA index of relative socio-economic disadvantage, remoteness, financial hardship, parent and partner employment status and partner education are as the predictors.  The strongest single predictor of partner violence was remoteness followed closely by SEIFA; so disadvantaged community.  However, multiple financial hardship, partner employment, and partners' education level were all significant.

So, here we are looking at the characteristics that predict a higher risk of partner violence, but remember, the majority of people in any of these groups are not victiMs or perpetrators of partner violence.  For example, there is 11% higher likelihood of partner violence if the father's education level is Year 9 or below.  It doesn't mean that the majority of men who have a low education are associated with partner violence and it shouldn't be interpreted in that way.

Zooming in a little bit more on some of these characteristics, in the purple you can see where the respondents considered family violence a big or very big problem in their community.  This is LSIC data.

In the green, you can see whether they personally reported experiencing partner violence.

On the left are the results for urban and inner regional areas.  On the right are the results for moderate and highly remote areas.

So, perception of family violence in the community has more than doubled, so from 9.5% to 22% in relatively remote areas, and experience of partner violence in the home also nearly doubled, so it went from 5.5 to 9.9%.

In addition, people who experienced it were also more likely to describe it as a problem in their area.  But, this isn't just a remote issue.

As in the previous slide in the purple you can see where the respondents considered family violence a big or very big problem in the community and in the green you can see whether they personally reported experiencing partner violence.  But here we can see the same results cut by community disadvantage, not remoteness.  So, using the socio-economic indices for area, ranks Australian suburbs and communities by level of disadvantage of the people who live there.

When these rankings are split into deciles, some 40% of LSIC sample living communities ranked in the most disadvantaged decile which includes urban, regional and remote areas.

On the left are the results for deciles 2 to 10, the more advantaged communities, and on the right are the results for the single most disadvantaged decile.  And you can see that both perception of family violence in the local community and self-reported partner violence in the home, are much more common in the disadvantaged areas.

LSAC has a somewhat similar item about whether this is a safe neighbourhood, which is also looking at those perceptions of community safety, family violence in the community.

Respondents were somewhat more likely to agree that they lived in a safe neighbourhood if they were not afraid of their partner.  And they were more likely to disagree or strongly disagree the neighbourhood was safe if they had been afraid of their partner.

Psychological distress is measured in LSAC using the Kesslar 6 questions and you can see here the green at the top is the proportion experiencing quite high financial - - mental distress, so you've got two cohorts here.  Maybe if you just focus on the first four columns - that's just looking at one cohort.

Mothers who reported feeling afraid of their partner were significantly more likely to report higher levels of distress and you can see that between 13 and 18% - depending on cohort of mothers - reported high distress if they had been afraid.

I tried to replicate this using LSIC, which has a slightly different distress measure, which we call the Social, Emotional, Well-Being Measure.  Is the Menzies Social, Emotional Well-Being, and so the proportions here classified as experiencing high distress are different to the LSAC results, just because it's a different scale.  So, don't worry about those proportions as much, but look at the overall relationship between the experience of partner violence and higher distress levels.  You can see it's similar if a bit less pronounced.

In the older cohort, mothers' distress levels were particularly high if they were single: which is an anomaly we didn't see in the LSAC data.

We would expect there to be a relationship between relationship quality, mental health and parenting efficacy and we can see that here.

Mothers who report being afraid of their partner are much less likely to rate themselves as a very good parent, and much more likely to consider themselves average or as a person who has trouble being a parent when compared to mothers who are not afraid of their partner, especially in the younger cohort of children.

So, being a good or very good parent you can see here in purple; average or less confident parent is in green.

The impact of poor family cohesion on children can also be seen in these LSAC item, so the children were asked to report here how often their family yell at each other, and children whose mothers reported being afraid of their partner were much more likely to say that their family always yells.

So, here is what we all want to know.  How does partner and violence affect children?  The impact of partner violence on children can be further seen through children's difficulties scores, which you have heard about quite a bit this morning and strength and difficulties score. The scale includes iteMs such as whether the children are restless and overactive, often complain of headaches, lose their temper, often seem worried, often unhappy, often picked on.

Children in both LSIC and LSAC had higher difficulties scores if the mother reported feeling afraid of her partner or reported that arguments with her partner sometimes or often end up with violence.  All reported that the child had been upset by family arguments in the past 12 months, so you can see up here the first two results are for LSAC; the next four are for LSIC; with scores of 1 to 2, or more than two points higher, so children's whose mothers were afraid of their partner or experiencing some violence.

So, the bottom line is even low levels of violence, and most of those reporting partner violence, that it happened rarely, were negatively related to children's outcomes.

Remember, we don't know which, if any, of these factors associated with higher levels of partner violence are causative.  Many of these characteristics cluster together when people are experiencing high levels of stress and other probleMs and who may also have difficult lives in other ways and as such, this evidence may help to identify vulnerable group or regions and that may be in need of some extra support.

We will see over time whether the women move out of these situations or remain in them and how this affects their children.

Thank you very much.  And if you're interested in seeing further presentations or data highlights on different topics, please get in touch with us. We can also arrange access to any of the four longitudinal datasets we hold, if you get in touch with us, including the Humanitarian Migrants survey and the household survey, HILDA.

Thank you very much.  (Acclamation)

Questions and answers session

Facilitator:

Thank you, Laura.  I will invite the presenters to please come and join me on stage, and you will notice that Laura had provided her email address there as well so you can spam the three of them.

Some really interesting and quite topical points from Laura's presentation, particularly with White Ribbon Day opening yesterday in relation to partner violence.  It's really interesting I think to look at the impacts of remoteness and relative disadvantage on some of those scores.  And it is interesting to think - - I was thinking about Sussan Ley, the Health Minister's announcement today, about new approaches to mental health and how we are now looking at some very site-specific and location specific interventions, which I think is quite interesting when you are considering about how those levels of disadvantage and levels of remoteness might impact on some of these findings.  So, thank you.

So, we're now going to have a bit of a panel discussion.  So, I would invite the people in the States and on the screen to give me a bit of a visual wave if you have any questions, and I can note that down and I can come back to you in a second if you do have any questions.  I'm not seeing any visual movement at all.

Ms Skelton:

Got a wave.

Facilitator:

I do have a wave, I've got a wave in Queensland.  Excellent.  Fantastic.  Any others up there?  I can see one in New South Wales, Sydney, I think.  Any others?  That's all I can see.  Excellent.  Thank you, guys.

So, I might - - I'll come to the regions in a second, but I'll throw it open, first of all, to the site here in Canberra, so if you have a question could you raise your hand.  We've got some remote mics.  If you can wait for the mics to come around to you so our friends and colleagues in the States can hear you.

Who would like to kick us off?  I think I might go to the States then.  So, Sydney.  Sydney, if you can ask your question for us, please.

Questioner:

Hi.  Thanks for the presentation today.  I hope you can hear me.  Nicolas - - it's Marianne Scanlon in Sydney, and you read some interesting research conducted by Macquarie University about why Indigenous families may not enrol their children in preschools et cetera.  We touched on some of the evidence today.

One of the things they raised there was about the family - - the preschool understanding Indigenous families but also the family feeling that they themselves are part of that preschool so when they engage with the preschool, the staff see them as family and they have a role in that preschool as well.  That seemed to be one of the critical factors for creating a suitable environment that the Indigenous family was connected themselves.  Not - - if they just dropped their children off and went off to do something, that wouldn't keep children enrolled in that at preschool.  It wasn't a relationship the family saw as supportive.

I really liked the use of the LSIC stuff too, that you are showing some very discreet evidence, but there's some very interesting information that expands on that.

I just wondered if you had heard any of that?

Dr Biddle:

Yes, so, I haven't seen the specific research.  I think it's important to distinguish between what explains the difference between Indigenous and non-Indigenous kids and then what explains kind of variation within the Indigenous population, because I think they are both important questions and I think LSIC is really good at looking at variation within the Indigenous population and other datasets obviously, including the LSAC, including HILDA I guess to the extent it has information on carers, it tells you a little bit about that difference, between the two.

And I think - - because the data that looks at differences between the two populations, Indigenous kids and non-Indigenous kids, kind of implies that much of that difference is explained by non-Indigenous specific characteristics, education levels and income, and to a lesser extent, geography. What that implies to me is that there is a commitment and a desire and that recognition of value for the carers of Indigenous kids for early childhood education.

So, it's nothing to do with indigeneity itself which kind of drives that difference.  But then when you go and look at that variation within the Indigenous population, then there are very specific determinants, and I think the one you mentioned about the comfort of families and that view, and that fits within that data which Fiona put up about teaching the cross-curriculum priorities, even though that's for older kids, you would expect that similar prograMs would have effects for Indigenous kids.

And some research which I've done on the early waves of LSIC showed that the experience of discrimination outside of an education setting for the carers of Indigenous kids, affects or at least is associated with whether those kids are attending or not.

So, I guess I think both of those two are quite important, and tell us quite different things.  But I think it's also important not to, when we think about some of those Indigenous specific determinants, not that blame or that focus on the families of Indigenous kids, but to recognise there are the structural barriers, not specific to Indigenous kids, which are driving some of those decisions.

So, that skirted around your question, but I think that data is really important, those findings are really important, but they need to be contextualised within some of those other broader, structural, values.

Facilitator:

Thanks, Nick.  Laura, I think you have come across this issue looking at both the LSIC and the LSAC data from time to time, and that issue of what is causal and what isn't when looking at the two populations.  Could you expand on what Nick was talking about there?

Ms Bennett-Kneebone:

Yes.  So, the tricky thing, when you're looking at both these populations, it is easy to say that something happens because somebody is Indigenous, but when you try to pick that apart from all the other sociodemographic factors that are associated with it, and with other things that are because they are Indigenous but which are not related to their indigeneity, so for example, discrimination, those things occur because they are Indigenous but it's - - being Indigenous isn't the problem there, it's the way society reacts to those things, and that's why I think it's really important not to stigmatise certain groups when the majority of people in those groups are doing the right thing and are, very passionate about their families and about getting their kids into preschool or about looking after their partners.  So, it's a tricky one to untangle.

Facilitator:

Thank you.

Fiona, as someone who has been involved in managing LSIC for a number of years now, it's interesting to hear your perspective from that Indigenous perspective, and perhaps bringing in some of the qualitative evidence that you might have encountered over the years and some of the communities that you visited.  What's the Indigenous view on this issue?

Ms Skelton:

Actually, one of the important things to say here probably is there's lots of them.  There's lots of different views.  And, in fact, when we started LSIC someone said well, the problem you've got is the variation within your population is greater than the variation that you get in the whole of the Australian population, and that's always something we have to deal with, fairly small sample; lots and lots of variation.

One of the things that might be worth mentioning is that a lot of the families who don't send their families to preschool, we did ask them why aren't you sending your child to preschool, and often the answer is they're too young, but I think Nick has done some analysis that shows they're not actually too young to attend in years but it may be that the parent feels that they are too young for whatever reason.  Maybe they're not toilet trained.  Maybe they don't feel like they are socially up to going, maybe they are not confident in the way the program, the preschool program, is being delivered but it's safer to say they are too young to go.

So, there's some work that Rebecca at Macquarie Uni did picking up on your thing - and I'm trying to remember her last name and I can't - in New South Wales about that too, and found similar things.  They're too young, they're not ready, I don't feel confident they're going to be okay.

Facilitator:

Thank you.

Okay, so I think we had a wave before in Brisbane.  Could we have the Brisbane colleagues of ours unmute their microphone and ask their question, please.

Questioner:

Yes, unmuted.  Thank you for all your presentations.  My question is to Nick again.

We talked about the association between outcomes and geography, and that when you control for geography you didn't see a strong link between poor outcomes and geography, but then when you control for income and education you found a stronger link.

I'm just wondering if - and it's a rookie question - there is an association between geography and income, and maybe then therefore if geography might predict income, therefore there might be a stronger link between geography and the outcomes.  It's an A to B to C question, I guess.

Dr Biddle:

It's a really good question.  I think clearly geography predicts a whole range of things, including some of the other variables, so education, participation of the parents, and all of those things are likely to be mixed up in the analysis.

I guess what I would say, though, is when we look at without controlling for income, when we just control for geography, the differences are still quite large.

So, what that implies to me is that the geographic relationship or the geographic predictors of income aren't the only predictor of, or the driver of, that relationship between income and participation.  And I think the reason for that is that we often, I guess, over-emphasise the effect of geography on kids - - in Indigenous outcomes.

So, what we often forget is that Indigenous kids live in disadvantaged - - Indigenous kids in urban areas are still substantially more disadvantaged than non-Indigenous kids in urban areas, and then Indigenous families in urban areas struggle to access jobs in the same way as Indigenous families in remote areas.

Now, there might be some difference, but often there's more variation between Indigenous and non-Indigenous kids in a particular area than there is between Indigenous kids in remote and non-remote area.

So, the geography - - I guess this is both a researcher, a data, and policy issue, is that we often overemphasise geography in our Indigenous analysis.

As an example, we did some analysis of the data on where jobs are located and where Indigenous Australians live, not the jobs of Indigenous Australians, but all jobs, and where they are located.  And what we found is that for an Indigenous person who lives in a remote area, there's a greater number of jobs, total jobs, relative to people in those areas than there are in Indigenous families who live in urban areas.

So, basically - I think that was a bit of a confusing way to explain that.  To reframe it - there are as many jobs per person in the area, remote areas, as there are in non-remote areas.  It's just Indigenous Australians who have access to those jobs.

So, we often overemphasise what remoteness can tell us about Indigenous outcomes.

Ms Bennett-Kneebone:

I agree with that.  I did some research recently looking at financial hardship in all four of our surveys, but particularly I looked at the relationship between financial hardship and remoteness in LSIC, and so we have sort of six iteMs people could report on, whether they'd gone without meals, whether they'd had trouble paying the bills or their mortgages and looking at what proportion of people had multiple of these things happening at the same time, and the proportions weren't all that different between remoteness categories.  The iteMs picked were different.  So, people in urban areas were more likely to have trouble paying housing payments and bills, and people in remote areas might have gone without meals because they were short of money.  Some of the iteMs reported were different but the overall prevalence of having multiple financial hardship was higher everywhere.

Facilitator:

And particularly important in our policy area that we are dealing with.

Fiona, can you bring in an LSIC perspective here again?

Ms Skelton:

I would agree with Laura.  It's the financial hardship that seeMs to make the difference rather than income, as such.  So, the stress iteMs that we collect about whether you experience financial stress in the last year or the iteMs that Laura was talking about are often the things that make the difference, not level of income.

So, I think Palin  who has done some work on who is attending school and why and whether that makes a difference. It's not income, it's the hardship.  So, how you are dealing with what you've got, which is the tool, which means that the prograMs that we run within the department on money management and help with people's financial stressful situations are really worthwhile, for a whole lot of reasons.

Dr Biddle:

So I think that really fits within an increasing body of research around the effect of poverty on stress and then the effect of that stress on the ability to make long-term decisions.

I mentioned during my presentation one of the headlines - and I'm a sucker for a good headline - and the paper which - - or the summary of that work is that bad decisions don't lead to poverty, poverty leads to bad decisions, and there's a book by Sendhil Mullainathan and Eldar Shafir called “Scarcity”, and where that goes into a lot of detail is the way in which poverty in the sense of income relative to your financial needs and financial stress and the effect that has on your ability to make good decisions, and more importantly the effect that that poverty has on exacerbating other biases.

So, not only do you make worse decisions - - and this is - - there's some work which shows that if you look at the IQ of those in poverty and those who are not in poverty - this is in the US - then there's very little difference.  If you look at maths test scores of those in poverty and those not in poverty in the US there is some difference but it's not very large.

If you then add a - - before you measure IQ or before you ask about, or before you take those maths test scores, if you asked those in poverty to make a very hard financial decision, so that's how would you manage if you had to raise $2000 in a certain period of time - I can't quite remember the question.  It's basically making that person think about managing their money.  If you asked somebody in poverty to make that decision, then that has a very large effect on their measured IQ, on their measured test scores.  If you asked someone who is not in poverty to make that decision it has no effect.

So, basically, it's not that those who are in poverty can't make good decisions, it's not that they have low IQ, it's that their poverty itself affects basically your cognitive load, you're making all these very hard decisions, and then when you have to make a decision about do you send your kids to school, do you get takeaway or do you cook your meals and all these things which have short-term costs but long-term benefits, then you are much less likely to favour the future over the present.

So, I guess it's the effect of poverty, it's the effect of financial stress, not that those who are in poverty are in any ways better or worse than those who aren't.

Ms Skelton:

Which is a really good point, and also relates probably to the stuff that Laura was talking about with partner violence.  I don't think the Children's Commissioner particularly asked us to look at financial hardship, but that's one of the things that keeps coming up for the relationship to partner violence.

Facilitator:

Thank you.  We have a question here, Anna, just down the front, so if we could bring the microphone down here.  If you could put your microphone up here so that Anna can see you.  There you go.

Questioner:

Thanks.  I just have a question for Fiona.  I noted in your strengths and difficulty score, that homelessness had a negative but not statistically different impact, and I just wondered if you had any explanation.  I just see homelessness is generally considered a big stressor, and it would have an impact.

Have you looked at homelessness at all and/or have any explanation why it wouldn't have been significant?

Ms Skelton:

The number of people who have experienced homelessness is relatively low so I think off the top of my head about 10% of our sample.  So, in the overall numbers it's relatively low.

If we had a much larger sample you might see that effect becoming statistically significant.  If you just looked at it bivariantly without controlling for all the other things that are in the model, the homelessness probably would make a difference, but it's in the context of all the other things that you're looking at.

Questioner:

Okay.

Ms Skelton:

So, we are asking homelessness questions again, because we asked about the past five years' experience of homelessness in next wave, Wave nine, and it will be interesting to see, if we do have people who have got repeated experiences of being homeless or not having somewhere to stay. But, numbers are always going to be a bit of a problem for us.

If we were talking about LSAC, where we've got 10,000 people or if we particularly targeted homeless people, which we haven't, in order to find the sample we knocked on doors so they weren't homeless when we first picked them up, then you might find it a bit different.  But, yes, okay.

Questioner:

Okay.  Thank you.

Facilitator:

And Laura, from the other studies that you have been involved in what insights do you have in relation to strengths and difficulties?

Ms Bennett-Kneebone:

Well, the strengths and difficulties questions are just in LSAC and LSIC because they are about the children.  So, yes, it was interesting to see exactly the same impacts on the strengths and difficulties scores in both those studies regardless of, you know, the Indigenous status or the socio-economic status of the children.

Facilitator:

Thank you, and Nick?

Dr Biddle:

I think the - so on the homelessness issue, I'd say a couple of things.

One is that the “Journey’s Home” so that which I think is in part - - has a strong involvement from DSS - -

Ms Skelton:

Yes, I think we counted it.

Dr Biddle:

Thank you

Our Institute runs it and there's three years' of data out, and you can get it.

Dr Biddle:

Yes, and that fortunately has an Indigenous over sample, and that focuses on those who were currently homeless as well as those who are at risk of homelessness.

Now, that isn't focused on children, but from my understanding there are some questions which relate to children both as carers as well as children themselves.

The other thing - - I mean clearly there's a sample selection issue as well.  It is that not only do you miss out on those in baseline who are homeless, who do experience homelessness, are likely to be part of that attrition rate, and part of those - - not only that, but also part of those blue dots which Fiona put up.

But, the final thing I would say is that homelessness is a really complex thing to measure, and I would encourage, with a focus on that issue, to look at a broader measure than just a very narrow kind of did you have a roof over your head, basically.  There's housing stress, there's living in an overcrowded dwelling, there's - - someone may be not classified as homeless at this point in time but if there is a potential risk then that stress itself will potentially impact on outcomes.

So, I would say you use as broad as measure as possible.  And I think in LSIC there is really good data on overcrowding, both self-report as well as some of the more objective measures, and that might pick up not the effect of homelessness, but the effect of an inadequate housing stock, which I think is really what we're really interested in.

Facilitator:

Mm, mm.  Thank you.

Ms Skelton:

And in our major life events questions which we have asked every year, we do have a question about have you experienced moving house, having housing probleMs or what's the third bit?  There's a third part to it anyway, and we started to segregate that after a few years.  So, some of those housing probleMs get picked up not just by asking how many bedrooMs but also by have you been stressed in the housing mix, and because we ask it every year it's a possibly more useful general measure than the homelessness.

Ms Bennett-Kneebone:

You can look at that mobility information by, housing tenure type, too, and look at how far people move to some extent, whether they stay in the same area or move outside of their area.

And we're also starting to see some of that in our humanitarian migrants survey.  People move a lot in their first few years in Australia, and you can kind of pick up on how much they're moving and where to and why.

Dr Biddle:

I mean, the data from the National Aboriginal and Torres Strait Islander Social Survey, the 2008 version, for some of the education related measures which we looked at, that implied that overcrowding is important, but what's more important is the quality of the housing stock within which you live, so in the NATSISS, they ask about iteMs needing repair and a range of other things, and that predicts attendance much better than either the number of people who are housed or the level of overcrowding.  And that makes sense.

I mean, if you're in a dwelling with a large number of people but you have a working fridge, you have a working washing machine, you have a roof over your head, you know, you can manage, and in some ways that broader social network can have benefits.  But if your school uniform can't get washed or you have to go to someone else's house to have a shower or use a toilet, then clearly that's going to have a much larger effect. So it's really unpacking what bad houses actually are, and that's quite important I think.  Different surveys have their strengths and weaknesses that inform that but pooling that information together gives you a really good picture of the effect of housing on kids' outcomes.

Facilitator:

Great.  Thank you.

We do only have another minute or so, so we will have to conclude the panel discussion there.

Just before I wrap up I would ask you to thank me in thanking our panellists.  (Acclamation)

If we could just have the slides back up, Peter.  Thank you.

So, after this session, the seminar will be available on the NCLD website as a video and also as a transcript within 10 days, so if you have missed anything you would like to go back to and refer to or if you would like to direct some colleagues across to the information, please feel free to do so.

We have a number of documents down the front here that you can pick up for those in Canberra.  For those not in Canberra, please feel free to email us or also go to the NCLD website, and there's a number of documents available there for download.

We would appreciate your feedback on this session today.  For those in Canberra, we've handed you out a little feedback form and we've got a box at the front for you to leave that form in.  And for those who are attending via video conference we will send that feedback form out to you later today, and we'd really appreciate that feedback as this is the first of these sessions that we've done.  We'd be really keen to learn about worked, what didn't work, what we might like to do differently in future.

So, all that is left for me to do is to say thank you.  I have provided the NCLD email address up there, but also the policy office address up there as well, and as I mentioned earlier, I would strongly encourage you to have a look on the intranet at what the policy office has to offer you.  And our panellists will be milling around here in Canberra for a little while, so if there are any questions that you didn't get a chance to ask, please pop down and ask them as well.  So, thank you very much.  Sorry?

Dr Biddle:

And please, spam is fine, but other emails are more than welcome, and I would love to hear direct any questions or any feedback about the analysis which we've done.  So, please, those in cyberland as well as those here, please email and please engage and tell us what we got wrong about the analysis but also what's useful and what's not.

Facilitator:

Great.  Excellent.  Thank you very much everyone. (Acclamation)

Last updated: