Spatial dimensions of child social exclusion risk: widening the scope Paper presented at the 11 th Australian Institute of Family Studies Conference, Melbourne, July 7-9 th 2010 Annie Abello, Cathy Gong, Justine McNamara and Anne Daly
2 Acknowledgements ●This paper was funded by ARC Discovery Grant DP : Towards an enhanced understanding of child and youth social exclusion risk at a small area level in Australia The authors would like to thank the other Chief Investigators and Partner Investigators on the grant – Prof Laurie Brown, Dr Asher Ben-Arieh, Professor Michael Noble and Ms Leanne Johnson, as well as Ann Harding and Robert Tanton from NATSEM and staff of the Bureau of Infrastructure, Transport and Regional Economics.
3 Background ●Earlier ARC-funded research into child social exclusion ●Development of NATSEM’s original Child Social Exclusion (CSE) Index ●Work under new grant (2010 – 2012): -Further development and refinement of CSE Index -Creation of an index of youth social exclusion risk -More analysis
4 Refining the index ●Re-examination of conceptual and measurement frameworks ●Investigation of new sources of data/variables ●Re-visiting methodology (first version used Principal Components Analysis to create index – similar to SEIFA indexes; this version we are creating domains, using PCA within domains and then equal weighting to combine domains) ●Comparing results ●Work still ongoing
5 Conceptualising social inclusion/exclusion Very large literature on conceptualising and measuring social exclusion, and much debate. Issues include: -Differences between social exclusion and poverty -Individual/structural -Relational aspects -Normative judgements -Overlap of risk/causal factors with outcomes -How important is persistence -Wide and deep exclusion
6 Social exclusion and children ●Levitas et al. (2007)UK work on matrix of social exclusion measures which can be applied to different age groups ●UK social exclusion and poverty audit indicators for children (Opportunity for All) ●SPRC Australian work on social exclusion measures related to children ●Small but increasing number of international small area indicators of child deprivation/disadvantage (eg UK, South Africa)
7 Some additional conceptual and measurement issues ●Data availability, especially for some concepts/dimensions ●The role (and availability) of data on children’s subjective well-being ●Importance of policy relevance ●Composite index vs individual variables ●Use of domains
8 Domains and variables used for original and revised NATSEM CSE index DomainsVariablesOriginal CSE indexRevised CSE index Socio-economicSingle parent family√√ In bottom income quintile√√ No family member completing year 12√√ Highest occupation of family members√× No parent working√√ EngagementNo internet at home√√ No parent volunteering√√ No motor vehicle√√ HousingPublic housing√× High renting cost×√ Health services & disability Ratio of GPs×√ Ratio of dentists×√ Children with disability×√ Data source: ABS Census We also intend to include some administrative data, such as crime, education outcome, environment and transport data if they are available for small area.
9 Refinements to methodology ●Principal Components Analysis (PCA) (1) To transform a set of correlated data into a smaller set of uncorrelated components. (2) PCA is used for all variables to estimate original NATSEM CSE index, but used for variables within each domain to estimate the revised CSE index. ●Equal weighting: for the revised CSE index only, we take the mean of each of 4 domains using equal weights, after exponential transformation of the index for each domain.
10 Statistics of main variables, Australia, 2006 VariableUnitMeanSD Single parent family% of children In bottom income quintile% of children No family member completing year 12% of children No parent working% of children No internet at home% of children No parent volunteering% of children No motor vehicle% of children High renting cost% of children Children with disability% of children Ratio of GPsPer 1000 persons1.71 Ratio of dentistsPer 1000 persons0.44
11 Correlation matrix of main variables Variables Single parent Low income No year 12No parent working No internet No volunteer No motor vehicle High renting cost Ratio of GPs Ratio of dentists With disability Single parent Low income No year No parent working No internet No volunteer No motor vehicle High renting cost Ratio of GPs Ratio of dentists With disability 1
12 Scree plot of domains (To test PCA)
13 Loadings for domains Original variablesSocio-economicEngagementHealth services & disability Single parent family 0.80 In bottom income quintile 0.91 No family member completing year No parent working 0.91 No internet at home 0.92 No parent volunteering 0.58 No motor vehicle 0.95 Ratio of GPs 0.89 Ratio of dentists 0.89 Children with disability Note: Loading is the correlation between the first component and original variables
14 Proportion of children by CSE quintile by capital cities/balance of Australia Original version of indexRevised version of index
15 Areas with most and least social exclusion risk, old and new version 50 areas with greatest risk: -In both old and new versions, 98% in non-capital city areas -70% of greatest risk small areas in new version were also in this group in old version 50 areas with least risk: -In both old and new versions, 94% in capital city areas -72% of least risk small areas in new version were also in this group in old version
16 Correlations between CSE index (new version) for children aged 0 to 15, 0-4 and 5-15, 2006 CorrelationCSE quintile for children 0-15 CSE quintile for children 0-4 CSE quintile for children 5-15 CSE quintile for children CSE quintile for children CSE quintile for children
17 Social exclusion characteristics by capital city/balance of Australia VariablesUnitCapital cities Balance of Australia Single parent family% of children No family member completing year 12% of children No parent working% of children In bottom income quintile% of children No internet at home% of children No motor vehicle% of children No parent volunteering% of children High renting cost% of children Children with disability% of children Ratio of GPsPer 1000 persons Ratio of dentistsPer 1000 persons
18 Characteristics for areas with greatest and least risk (n=50) MeanUnit 50 small areas with highest risk 50 small areas with least risk Single parent family% of children No family member completed Yr 12% of children No parent working% of children No internet at home% of children No motor vehicle% of children No parent volunteering% of children Bottom income quintile% of children High renting cost% of children Children with disability% of children GP to 1000 populationPer 1000 persons Dentist to 1000 populationPer 1000 persons0.20.7
19 Future work ●Additional variables, especially for domains currently not covered/poorly covered (e.g. physical environment; crime and safety; education outcomes) ●Continue to trial index creation techniques ●Map and further analyse results ●Youth index