Social Policy Research Paper No. 36

Contents | Download as PDF [1.1MB]

Contents

• Executive summary
• 1. Introduction
• 2. Outcomes for children in differing circumstances
• 3. Children's use of non-parental care
• 4. Children's health
• 5. Mothers' health
• 6. Family learning environments and child outcomes
• 7. Discussion
• Appendix A: Summary of variables included in the Outcome Index
• Appendix B: LSAC Research Consortium
• Endnotes
• References
• List of tables and figures

2. Outcomes for children in differing circumstances

• 2.1 Introduction
• 2.2 Findings
• 2.3 Discussion

Section summary

• Outcomes for 4 to 5 year-old children were quite powerfully related to broad characteristics of the child, mother, and family context, consistent with an ecological model of child development. In contrast, there was little evidence that these factors impacted on infants' outcomes, suggesting that these contextual factors impact on children's development through a cumulative processes over time.
• Particularly in the child cohort, there was strong evidence that child characteristics were associated with outcomes. The data suggest that girls consistently had better outcomes than boys (except in the physical domain). Aboriginal and Torres Strait Islander children appeared to have poorer outcomes, despite few differences in infancy. Similarly, children from families which spoke a language other than English tended to have poorer outcomes.
• Indices of parental and family psychological and material capital (maternal education, parental occupational status, income and financial stress) were related to child but not infant outcomes. Infants tended to do better in smaller families, while moderate-sized families were optimal for children. However, family type (either one or two parents) showed no evidence of an independent effect on outcomes.
• There was only weak evidence that neighbourhood effects (liveability, disadvantage, remoteness, metropolitan/non-metropolitan location) impact on child outcomes at these ages. Such effects can be expected to become stronger over time.
• The set of sociodemographic factors identified here are important to consider when assessing the impact of more fine-grained biological and environmental exposures, and are used as covariates in analyses in subsequent sections.

2.1 Introduction

As described in Section 1, the Outcome Index permits comparison on the physical, social–emotional and learning outcomes of children growing up in different circumstances. Identification of groups of children who are developing well and those doing less well provides important guidance to policy makers. It is also important to understand whether poor developmental outcomes for a group of children occurs across the spectrum of domains of development or is limited to one or two domains. LSAC provides an unusual opportunity to examine specific or general differences between groups, since it taps a much broader spectrum of child outcomes than is available in most research.

In this section, we describe the LSAC infants and children according to a broad range of sociodemographic characteristics, and then examine how these variables are distributed at both ends of the overall Outcome Index scores. We next examine these variables in multivariable analyses for overall Outcome Index scores as well as physical, social–emotional and learning domain scores. The final multivariable models include nine sociodemographic variables spanning the child, mother, family and community factors. Analyses in all subsequent sections adjust for these variables.

2.2 Findings

The sociodemographic variables

Table 3 describes important sociodemographic characteristics that broadly characterise the child, the mother, the family and the community. The rationale for examination of these variables in relation to child outcomes was outlined in Section 1, but is briefly reiterated below along with details about the derivation of variables and their distribution in the two cohorts. More specific variables (such as prenatal exposures, child care experiences, and educational activities in the home) are examined in later sections, adjusting for a common set of these broad sociodemographic characteristics. Other important influences on children, such as parenting practices and family functioning, are not examined here, since they are covered in detail in another report (Zubrick et al. 2008). The bivariate and multivariable analyses in this section assess the extent to which these sociodemographic variables predict Outcome Index scores, while noting that causal inferences cannot be drawn from cross-sectional analyses.

The characteristics of those who do markedly well or poorly are often of particular policy interest. Figure 3 shows the proportions within the infant cohort falling into the bottom (negative or 'problem') end of the overall Outcome Index distribution according to (a) child and maternal, (b) family and (c) neighbourhood characteristics. Figure 4 shows the corresponding proportion falling above the positive cut-off (that is, in the top 15 per cent of overall Outcome Index distribution) for these same variables. The bars on each variable indicate the 95 per cent confidence intervals for each proportion. Figures 5 and 6 follow the same approach for the child cohort. Findings are discussed below.

Child characteristics

• Age: the mean age for the infant cohort was 41 weeks (9.5 months), with a range of 14 to 83 weeks. The child cohort was on average slightly over 4 years, 6 months old, with a mean age of 250 weeks (4 years, 8 months) and a range of 223 to 295 weeks. Because of the way the Outcome Index is standardised, age is already taken into account in its calculation, so analysis by age within each cohort is not possible.
• Gender: both cohorts comprised approximately 51 per cent male and 49 per cent female children, reflecting the Australian population for children of this age. There is considerable evidence that boys are more vulnerable to developmental difficulties and develop more slowly than girls in the early years (see, for example, Ruble & Martin 1998); this can have policy implications in areas as diverse as health, child care and education. In the child cohort, almost twice as many boys as girls fell below the negative cut-off, and almost twice as many girls as boys scored above the positive cut-off (see Figures 5 and 6). However, as shown in Figures 3 and 4, this disparity was less marked in the infant cohort.
• Ethnic and cultural background can impact on children's development in multiple ways, as discussed in Section 1. The proportion of children who were described by their parents as being Aboriginal or Torres Strait Islander (ATSI) (5 per cent of infants and 4 per cent of children) was a little above population rates. In the child cohort, only 3 per cent of these children fell above the positive cut-off, whereas a third fell below the negative cut-off. Trends in similar directions in the infant cohort were less marked (10 per cent and 20 per cent respectively). As noted in Section 1, due to the sampling methodology employed in LSAC and the exclusion of some very remote postcodes, the sample of ATSI children in this study is unlikely to be fully representative of all ATSI children.
• Here the measure of non-Australian background was whether a child lived in a family which spoke a language other than English at home (LOTE). Approximately the same proportion of children in both cohorts lived in such families (13 per cent of infants, 14 per cent of children). In the child cohort, a greater proportion of these children were below the negative cut-off (21 per cent versus 16 per cent), and a smaller proportion were above the positive cut-off (10 per cent compared to 15 per cent); these differences were not evident for the infant cohort.

Maternal characteristics

• Education: a mother's educational status is a critical component of her psychological capital. For the purposes of LSAC, 'mother' is defined as any female guardian of the study child, or the one who knows the child best if there are two. In almost all cases the mother was the primary care giver. Mothers of infants tended to be a little better educated than those of the child cohort, with 59 per cent of mothers of the infant cohort having completed high school compared to 51 per cent of mothers of the child cohort. While the distribution of maternal education was not related to high and low Outcome Index scores in the infant cohort, a marked stepwise relationship was observed in the child cohort, with fewer children having low Outcome Index scores and more having high Outcome Index scores as the level of maternal education increased.
• Employment: the labour force participation of mothers of young children has increased markedly over recent decades and has been associated with both positive and negative outcomes (Hoffman & Youngblade 1999; Sanson et al. 2002). In the LSAC infant cohort, 61 per cent of mothers were not currently working and 11 per cent were working full-time (30 or more hours per week), compared to 46 per cent and 19 per cent for mothers of the child cohort. Maternal employment status was not related to positive and negative Outcome Index scores in the infant cohort. In the child cohort, a greater proportion of children whose mothers were not working scored below the negative cut-off, and a smaller proportion scored above the positive cut-off, than those whose mothers were working part or full-time. Outcomes for the latter two groups appeared similar.

Family characteristics

• Family type: research has consistently pointed to differences in outcomes across family types, although the causal relationships between them are unlikely to be direct. For young children, a salient distinction is whether there are one or two parents available in the home. For the purpose of LSAC, the primary parent was defined as the person who knows most about the child, and the secondary parent as anyone else with a parental relationship to the study child or a partner of the primary parent. By this definition, 11 per cent of infants and 15 per cent of children were in single-parent families. In the child cohort, only 9 per cent of children in single-parent families scored above the positive cut-off, and 27 per cent scored below the negative cut-off. In contrast, living in a family with one or two parents was not related to positive or negative outcomes for the infant cohort.
• Siblings: while siblings can provide emotional support and socialisation for children, they can also lead to rivalry and make competing claims on parents' resources. The number of siblings was calculated by counting the number of people living with the study child who had a sibling relationship (including full, step, half, foster and adopted siblings) with the child. Nearly 40 per cent of the infants were only children, compared to just 12 per cent of the children. Two-fifths of both cohorts (41 per cent) were the oldest or an only child. As shown in Figures 3 and 4, a greater percentage of infants who had no siblings fell above the positive cut-off than those with one or two siblings. In addition, a greater percentage of those with three or more siblings fell below the negative cut-off than those with no or one sibling. In contrast, number of siblings had similar distributions across high and low Outcome Index scores in the child cohort, except for children with three or more siblings, where a smaller proportion fell above (11 per cent) and a greater proportion fell below (22 per cent) the positive and negative cut-offs respectively. Similarly, more infants who were the oldest child in the family (which indicates they were only children) were above the positive cut-off (18 per cent) and fewer were below the negative cut-off (12 per cent) than those with older siblings, whereas being the oldest child was not related to positive or negative outcomes in the child cohort.
• Household size: the number of people in a household is a key element of the child's home environment, but evidence for its impact on child outcomes is mixed. Large households can be noisy and stressful, and may entail less parental contact and support for each individual child, but can also provide the child with more social interaction and social support. In LSAC, the number of people in the household is a count of all the people enumerated in the household grid. Most of these people were parents or siblings, but other relatives and non-related adults and children were also included. A greater proportion of infants in three person households scored above the positive cut-off compared to infants in households with four or more persons, and a smaller proportion scored below the negative cut-off than those in households of six or more. For the child cohort, non-stepwise trends indicated that medium sized households (3 to 5 persons) may be optimal for children at this age. Fewer children in households of six or more scored above the positive cut-off than those in four person households, and substantially more of those in both two and more than six person households fell below the negative cut-off than those in four and five person households.
• Overcrowding has been linked to poorer health and other outcomes, and can make the household more chaotic and stressful (Office of the Deputy Prime Minister 2004; Reynolds 2005). Defining overcrowded households as those with at least twice as many people as bedrooms, just over 3 per cent of the LSAC children in each cohort lived in an overcrowded household (see Table 3). A higher proportion of infants in overcrowded households scored below the negative cut-off (26 per cent compared to 15 per cent, see Figure 3). In the child cohort also there were clear differences between those who did and did not experience overcrowding in the proportions falling above the positive and below the negative cut-offs (positive: 7 per cent versus 14 per cent; negative: 25 per cent versus 16 per cent, see Figures 5 and 6).
• Combined parental income: both low income and financial stress have negative and accumulating effects on children's development (Bradbury 2003; eds Keating & Hertzman 1999). Parents were asked what their combined present yearly income was, choosing among 16 categories which were subsequently aggregated for this report into approximate quintiles. In the infant cohort, income was similarly distributed across the high and low outcome cut-offs. However, for the child cohort, 9 per cent of those in the lowest quintile and 20 per cent of those in the highest quintile scored above the positive cut-off, whereas the comparable figures for children below the negative cut-off were 28 per cent and 9 per cent respectively (see Figures 5 and 6).
• Financial stress: families were defined as being financially stressed if they indicated that they had experienced four or more of seven different adverse financial situations in the last 12 months (for example, not being able to pay bills, skipping meals to save money), with 6 per cent of both the infant and child cohorts fitting this criterion. Although financial hardship data paralleled the data on income quite closely, the disparities in child cohort outcomes were even more marked for financial stress. Only 4 per cent of those in financially stressed households scored above the positive cut-off, whereas over one-third (36 per cent) were below the negative cut-off. Differences were not evident for the infant cohort.
• Highest occupational status of parents: parental occupational status is linked not only to financial security but also to parental psychological capital, and hence can impact on child outcomes. The Australian Standard Classification of Occupations (ASCO) (ABS 1997) assigns each occupation a number from 1 to 9, with occupations requiring a greater skill base receiving a higher code; for example, Code 1 refers to managers and administrators, Code 3 includes dental therapists and ambulance officers, Code 5 includes advanced clerical and service workers, and Code 9 includes labourers and related workers. For this report, the highest parental occupation of a household was categorised into one of four groups according to its ASCO Code (1–3, 4–7, 8–9, neither parent employed). Almost half of the LSAC children had one or more parents employed in the highest occupational grouping (ASCO Codes 1–3). Parental occupational status did not appear to differ with outcomes in the infant cohort, but for the child cohort there were marked, stepwise increases in outcome scores as parental occupational status rose. Only 6 per cent of children with neither parent in the workforce were above the positive cut-off, whereas this was true of 19 per cent of children in the highest parent occupational grouping; and almost one-third (31 per cent) of children of non-working parents were below the negative cut-off, compared to 10 per cent of children whose parents were in the highest category (see Figures 5 and 6).

Neighbourhood characteristics

Finally, characteristics of the neighbourhood have been shown to be associated with child outcomes, often indirectly through their impact on family functioning (Brooks-Gunn et al. 1997) and also through differential access to services.

• Neighbourhood liveability: the LSAC neighbourhood liveability scale asks respondents to rate their agreement to five statements about the condition of their neighbourhood (for example, cleanliness, availability of parks, quality of street lighting) on a four point Likert scale—from 'strongly agree' to 'strongly disagree'—and takes the mean of these responses. High liveability neighbourhoods score from 1 to 2.5, while low liveability neighbourhoods score from 2.6 to 4. For both cohorts, more than 85 per cent of children were living in neighbourhoods rated by their parents as highly liveable. Liveability seemed to be weakly related to outcomes in both cohorts. In the infant cohort, 11 per cent of those in low liveability neighbourhoods were above the positive cut-off compared to 16 per cent of those with high liveability. In the child cohort, more children in low liveability neighbourhoods were below the negative cut-off (24 per cent versus 15 per cent) and fewer were above the positive cut-off (10 per cent versus 15 per cent).
• Metropolitan or non-metropolitan residence: household location is potentially important for children's development as it may affect access to and choice among medical, educational and other essential and support services, as well as determining their exposure to noise and air pollution. The LSAC non-metropolitan classification refers to the ABS definition of not living in a capital city statistical division. The proportions shown in Table 3 (with approximately one-third living in non-metropolitan areas) are exactly the national proportion for infants and children, reflecting the LSAC weighting process. There were no obvious differences in area of residence across the high and low outcomes for either cohort.
• Remoteness: this variable uses the ABS Australian Remoteness Indicator for Area (ARIA) (ABS 2001) linked to the LSAC data file at the postcode level. The ARIA is based on the distance of the location from key services such as food stores and doctors. As the LSAC sample design excluded some very remote postcodes, the numbers of children in very remote areas (2 per cent in each of the two cohorts) may underrepresent the population in these areas. Therefore, these findings may not be readily generalisable and comparisons between these groups and the three 'accessible' groupings need to be interpreted cautiously. Living in an accessible or remote area did not appear to be distributed differently across infants' outcomes, although only 9 per cent of those in very remote areas were above the positive cut-off. For the child cohort, the 'remote' category (as distinct from 'very remote') differed from the highly accessible and accessible groups in the proportion above the positive cut-off (only 4 per cent), but the 'very remote' group had a larger proportion above the positive cut-off and a lower proportion below the negative cut-off (12 per cent) than all other groups except 'accessible'. The highly accessible, accessible and moderately accessible groups did not appear to differ substantially from each other.
• SEIFA Index of Disadvantage: the ABS Socio-Economic Indexes for Areas 2001 (ABS 2003b) provides a general indicator of neighbourhood advantage or disadvantage by combining information on the social and economic conditions of an area based on information collected at the 2001 Census. For LSAC this has been linked to the dataset by postcode rounded off to the nearest 10 (for example, 937 becomes 940) to protect the identity of respondents' postcodes. SEIFA scores were categorised into five groups defined by the quintiles of SEIFA scores of all Australian neighbourhoods, with the 'lowest' group representing the most disadvantaged and the 'highest' the most advantaged. As can be seen from Table 3, both cohorts are relatively evenly spread across the five quintiles. A smaller proportion of infants living in the most disadvantaged 20 per cent of neighbourhoods were above the positive cut-off than those in the second lowest quintile, but differences with other quintiles were not substantial, and there were no observed differences for the negative cut-off. In the child cohort, a general linear trend was observed on the negative cut-off, and substantially more children in the most disadvantaged quintile were below the negative cut-off than children in the middle, fourth and fifth quintiles (24 per cent compared to 2 per cent, 14 per cent and 9 per cent respectively). There was no linear trend for proportions above the positive cut-off, but fewer children in the most disadvantaged quintile were above the positive cut-off than children in the fourth and fifth quintiles (9 per cent compared 19 per cent and 19 per cent respectively, see Figure 6).

The analyses to this point examined relationships between the Outcome Index and each characteristic independently. Such analyses can be misleading since they do not take into account the interrelationships among the characteristics being examined. The next section reports multivariable analyses which, when calculating the effect of any one variable, account for the contribution of every other factor.

Table 3: Sociodemographic characteristics of the infant and child cohorts

Characteristic	Infants		Children
	n	%	n	%
Study child
Sex	5,107		4,983
Male		51.3		51.2
Female		48.7		48.8
Age in weeks (95% CI)		40.7 (40.2, 41.2)		250.2 (249.7, 250.7)
Aboriginal/Torres Strait Islander	5,107	4.9	4,981	3.9
Main language not English	5,104	12.8	4,983	14.0
Mother
Education	5,098		4,940
Did not complete high school		41.1		49.1
Completed high school		29.8		26.6
Tertiary		29.1		24.3
Employment status	5,093		4,935
Working full-time		10.6		19.4
Working part-time		28.2		35.0
Not currently working		61.2		45.6
Family
Family type	5,107		4,983
One parent		10.5		15.0
Two parents		89.5		85.0
Number of study child siblings	5,107		4,983
None		39.1		11.5
One		36.4		47.5
Two		16.4		26.8
Three or more		8.1		14.2
Number of people in household	5,107		4,983
Two		1.8		2.7
Three		34.5		12.7
Four		34.8		42.5
Five		17.8		26.5
Six or more		11.1		15.6
Study child the oldest or only child in household	5,107	40.8	4,983	41.1
Overcrowded household	5,101	3.4	4,974	3.8
Midpoint of gross combined parental income category in AUD (median [p25, p75])	4,835	$65,000 [$39,000, $91,000]	4,663	$65,000 [$39,000, $91,000]
Income category approximate quintile	4,835		4,663
Lowest		20.2		19.3
2nd		13.7		12.2
3rd		13.9		12.6
4th		25.9		24.6
Highest		26.2		31.2
Financially stressed household	5,075	6.1	4,948	6.3
Highest occupational class	5,104		4,980
Neither parent working		13.2		13.2
ASCO 8–9 (Unskilled labour)		7.2		6.7
ASCO 4–7 (Skilled labour & clerical)		31.8		32.9
ASCO 1–3 (Professional)		47.8		47.3
Neighbourhood
Low neighbourhood liveability score	5,103	14.8	4,976	14.4
Non-metropolitan	5,107	33.5	4,983	36.3
Remoteness area classification	5,049		4,937
Highly accessible		58.9		55.8
Accessible		22.1		24.2
Moderately accessible		15.2		16.2
Remote		2.1		2.0
Very remote		1.6		1.8
SEIFA Disadvantage Index quintile	5,107		4,983
Highest disadvantage		18.5		18.5
2nd		21.5		23.7
3rd		20.8		20.2
4th		19.9		19.3
Lowest disadvantage		19.3		18.2

Figure 3: Low Outcome Index: percentage of infant cohort by sociodemographic characteristics

Figure 4: High Outcome Index: percentage of infant cohort by sociodemographic characteristics

Figure 5: Low Outcome Index: percentage of child cohort by sociodemographic characteristics

Figure 6: High Outcome Index: percentage of child cohort by sociodemographic characteristics

Impact of sociodemographic variables on outcomes: multivariable analyses

In these analyses, the multivariate associations of sociodemographic characteristics are examined for the overall Outcome Index and each of the three domain scores (physical, social–emotional and learning). In each case these outcome measures are treated as continuous with a mean of 100 and a standard deviation of 10.

Nine of the characteristics examined above were chosen for entry into this analysis. Criteria for selection included the strength of their theoretical contribution to children's outcomes, lack of redundancy with other measures, and representation of all four tiers of influence (child, mother, family and neighbourhood). All three child variables (gender, ATSI status and speaking a language other than English) were included. Maternal education was retained, but maternal employment was excluded because of its overlap with the 'no parent employed' category of the ASCO codes. Family type, family income, financial stress and parental occupational status were retained, but measures of family size (number of siblings, child is oldest sibling, number of people in household) were excluded due to lack of a clear theoretical rationale as well as lack of evidence in the bivariate analyses of strong relationships with child outcomes. Overcrowding was excluded because it was uncommon in the LSAC cohorts and, even when present, showed rather weak relationships to outcomes. SEIFA quintile was chosen as the most comprehensive and accurate measure of neighbourhood advantage or disadvantage; the subjective measure of liveability was excluded, as were metropolitan/non-metropolitan location and the remoteness indicator. The retained measures thus tapped key child, parent, family, and neighbourhood characteristics.

Table 4 shows the results of the multivariable analyses predicting the overall Outcome Index scores for each cohort. Tables 5 to 7 present comparable analyses for the physical, social–emotional and learning domains respectively. The results for the infant cohort are described first, followed by the child cohort.

Infant cohort

The left-hand parts of Tables 4 to 7 show that the nine variables which were entered into the analysis accounted for minimal amounts of variation in all four of the outcome measures examined here (1.4 per cent, 1.3 per cent, 1.8 per cent and 1.6 per cent respectively). Seven of the variables were associated, at least weakly, with at least one of these outcomes, as described below.

Child characteristics

• Gender was weakly associated with the overall Outcome Index (p=0.03) and learning domain (p=0.05), and more strongly with physical domain (p‹0.001), in favour of girls, but differences were quantitatively very modest (1 point or less).
• Being an Aboriginal or Torres Strait Islander was associated only with physical domain scores (p=0.008), with a 2-point disadvantage for these infants in comparison to the rest of the cohort.
• Speaking a language other than English at home was associated only with social–emotional outcomes (p=0.002), with an average difference of 2 points.

Maternal, family and neighbourhood characteristics

• Maternal education was strongly associated with overall Outcome Index (p‹0.001) and social–emotional domain (p=0.003) scores, but there was no linear trend (those whose mothers had not completed Year 12 had slightly lower outcomes than those whose mothers had completed school, but higher than those with tertiary education).
• Highest occupational status of parents showed a weak inverse relationship with infant learning, with highest scores for infants of non-working parents and lowest scores for those in the 'professional' category (p=0.07).
• Financial stress only showed a weak trend (p=0.09) towards an association with the social–emotional domain, where scores averaged 1.5 points lower than for infants from non-stressed families.
• SEIFA Index of Disadvantage: compared to being in the most disadvantaged SEIFA quintile, infants in the second lowest quintile scored an average of 2 points higher on the overall Outcome Index. Surprisingly, infants in the second most disadvantaged SEIFA quintile had overall Outcome (p=0.02) and social–emotional domain (p‹0.001) scores on average two points higher than those in the most disadvantaged quintile. On the overall Outcome Index, those in the fourth quintile scored 1.3 points higher than the most disadvantaged group. Differences with the reference category were small for the other quintiles.

Child cohort

The results for the child cohort are shown in the right hand columns of Tables 4 to 7, and show that the variables in the models generally accounted for substantially more variance than in the infant cohort—14.6 per cent for the overall Index, 11.5 per cent for the social–emotional domain, and 13.5 per cent for the learning domain, but notably only 2.1 per cent for the physical domain. All variables except family type contributed to the prediction of at least one of these scores, as described below:

Child characteristics

• Gender: girls were on average 3 points higher on the Outcome Index score (p‹0.001), 2.5 points higher on social–emotional scores (p‹0.001) and 4.4 points higher on learning scores than boys (p‹0.001), but with a difference of less than one point on the physical domain scale (p=0.001).
• Aboriginal or Torres Strait Islander children were on average 2.8 points lower on the overall Outcome Index (p‹0.001), 3.3 points lower on the social–emotional domain (p‹0.001), and 1.5 points lower on the learning domain (p=0.05).
• Speaking a language other than English at home was associated with an average decrease of 2.5 points on the Outcome Index (p‹0.001), 2 points on both the physical (p=0.001) and social–emotional domains (p‹0.001), and 1.4 points on the learning domain (p=0.004).

Maternal, family and neighbourhood characteristics

• Maternal education: children whose mothers had higher education had higher Outcome Index scores (p‹0.001)—in comparison to those whose mothers did not complete Year 12, those whose mothers did complete Year 12 were 1.7 points higher, and those whose mothers had a tertiary qualification were 2.4 points higher, on average. Similar differences were observed for the social–emotional (p‹0.001) and learning (p‹0.001) domains, while differences were less marked on the physical domain (p=0.04).
• Occupational status of parents contributed to the model for the overall Outcome Index (p‹0.001)—if the highest parental occupation was skilled labour or clerical, children scored on average 2.1 points higher than children with neither parent working; and where the highest occupation was professional, this difference was 3.7 points. It was quite strongly associated with learning scores (p‹0.001), with an average difference of 4.2 points between 'neither parent working' and 'professional' groups, and also with higher social–emotional scores (p‹0.001), with an average difference of 3.3 points, but not with physical outcomes.
• Combined parental income: there was strong evidence to suggest that income was associated with overall Outcome Index and social–emotional domain scores (both p‹0.001), with the contrast between the lowest and highest quintiles being most substantial (about 2 points). It was not associated with physical or learning domains.
• Financial stress made a substantial contribution to the models, with those classified as stressed scoring on average about 4 points lower on the Outcome Index and social–emotional domain (both p‹0.001), and about 2 points lower on the physical (p=0.002) and learning (p‹0.001) domains.
• SEIFA Index of Disadvantage: there were no strong associations with neighbourhood disadvantage, but trends for both social–emotional (p=0.06) and learning (p=0.07) domain scores to be higher as neighbourhood advantage increased.

Table 4: Multivariable relationships between sociodemographic variables and the overall Outcome Index for the infant and child cohorts

Characteristic(a)	Infants n=3,591 R2=1.4%		Children n=4,592 R2=14.6%
	Coefficient (95% CI)	p-value(b)	Coefficient (95% CI)	p-value(b)
Study child
Female	0.7 (0.1, 1.4)	0.03	3.6 (3.1, 4.2)	‹0.001
Aboriginal/Torres Strait Islander	–1.9 (–3.9, 0.1)	0.07	–2.8 (–4.4, –1.3)	‹0.001
Main language not English	–0.9 (–2.3, 0.6)	0.23	–2.5 (–3.5, –1.6)	‹0.001
Mother
Education		‹0.001		‹0.001
Did not complete high school	0 (–,–)		0 (–,–)
Completed high school	0.8 (–0.1, 1.7)	0.09	1.7 (0.9, 2.4)	‹0.001
Tertiary	–1.0 (–2.0, 0.0)	0.05	2.4 (1.7, 3.1)	‹0.001
Family
2 parents in the home	0.9 (–1.1, 2.9)	0.37	–0.2 (–1.4, 1.0)	0.71
Combined parental income quintile		0.57		‹0.001
Lowest	0 (–,–)		0 (–,–)
2nd	–0.5 (–2.1, 1.1)	0.55	–0.4 (–1.5, 0.7)	0.50
3rd	–1.2 (–2.7, 0.2)	0.10	0.2 (–1.2, 1.5)	0.78
4th	–0.8 (–2.2, 0.6)	0.28	0.8 (–0.4, 2.0)	0.20
Highest	–0.6 (–2.0, 0.8)	0.41	2.0 (0.8, 3.2)	0.001
Financially stressed household	–1.0 (–2.9, 0.9)	0.30	–4.1 (–5.5, –2.7)	‹0.001
Highest occupational class		0.79		‹0.001
Neither parent working	0 (–,–)		0 (–,–)
ASCO 8–9 (Unskilled labour)	–0.8 (–3.0, 1.4)	0.48	1.3 (-0.4, 3.0)	0.12
ASCO 4–7 (Skilled labour & clerical)	–0.3 (–2.2, 1.7)	0.77	2.1 (0.7, 3.6)	0.005
ASCO 1–3 (Professional)	–0.6 (–2.6, 1.3)	0.55	3.7 (2.2, 5.2)	‹0.001
Neighbourhood
SEIFA Disadvantage Index quintile		0.02		0.09
Highest disadvantage	0 (–,–)		0 (–,–)
2nd	2.0 (0.8, 3.1)	0.001	0.8 (–0.3, 2.0)	0.15
3rd	1.1 (–0.1, 2.3)	0.06	0.9 (–0.1, 1.9)	0.06
4th	1.3 (0.1, 2.5)	0.03	1.3 (0.1, 2.4)	0.03
Lowest disadvantage	1.2 (–0.2, 2.6)	0.09	1.5 (0.4, 2.6)	0.007

Table 5: Multivariable relationships between sociodemographic variables and the physical domain score for the infant and child cohorts

Characteristic(a)	Infants n=4,800 R2=1.3%		Children n=4,599 R2=2.1%
	Coefficient (95% CI)	p-value(b)	Coefficient (95% CI)	p-value(b)
Study child
Female	1.2 (0.6, 1.7)	‹0.001	0.9 (0.4, 1.5)	0.001
Aboriginal/Torres Strait Islander	–2.2 (–3.8, –0.6)	0.008	–1.2 (–3.0, 0.6)	0.18
Main language not English	–1.1 (–2.2, 0.1)	0.08	–2.0 (–3.1, –0.8)	0.001
Mother
Education		0.61		0.04
Did not complete high school	0 (–,–)		0 (–,–)
Completed high school	0.1 (–0.6, 0.8)	0.70	0.9 (0.1, 1.7)	0.03
Tertiary	–0.2 (–1.1, 0.6)	0.60	0.1 (–0.7, 0.9)	0.82
Family
Two parents in the home	0.3 (–1.2, 1.9)	0.69	0.1 (–1.1, 1.3)	0.88
Combined parental income quintile		0.88		0.07
Lowest	0 (–,–)		0 (–,–)
2nd	–0.3 (–1.7, 1.0)	0.62	–0.4 (–1.8, 0.9)	0.55
3rd	–0.5 (–1.9, 0.9)	0.46	0.3 (–1.1, 1.7)	0.65
4th	–0.4 (–1.6, 0.8)	0.48	0.4 (–0.9, 1.7)	0.51
Highest	–0.1 (–1.5, 1.2)	0.84	1.2 (–0.1, 2.5)	0.07
Financially stressed household	–1.1 (–2.6, 0.4)	0.15	–2.5 (–4.1, –0.9)	0.002
Highest occupational class		0.22		0.97
Neither parent working	0 (–,–)		0 (–,–)
ASCO 8–9 (Unskilled labour)	–0.7 (–2.3, 1.0)	0.43	–0.2 (–1.9, 1.6)	0.86
ASCO 4–7 (Skilled labour & clerical)	0.6 (–0.9, 2.1)	0.43	0.2 (–1.2, 1.6)	0.79
ASCO 1–3 (Professional)	0.7 (–0.8, 2.2)	0.37	0.2 (–1.3, 1.7)	0.82
Neighbourhood
SEIFA Disadvantage Index quintile		0.26		0.51
Highest disadvantage	0 (–,–)		0 (–,–)
2nd	0.5 (–0.5, 1.4)	0.31	0.8 (–0.3, 1.8)	0.17
3rd	1.1 (0.1, 2.1)	0.03	0.0 (–1.0, 1.0)	1.0
4th	0.8 (–0.2, 1.8)	0.14	0.6 (–0.5, 1.7)	0.27
Lowest disadvantage	0.7 (–0.4, 1.8)	0.19	0.3 (–0.9, 1.4)	0.62

Table 6: Multivariable relationships between sociodemographic variables and the social-emotional domain score for the infant and child cohorts

Characteristic(a)	Infants n=4,091 R2=1.8%		Children n=4,592 R2=11.5%
	Coefficient (95% CI)	p-value(b)	Coefficient (95% CI)	p-value(b)
Study child
Female	0.1 (–0.5, 0.7)	0.80	2.5 (1.8, 3.1)	‹0.001
Aboriginal/Torres Strait Islander	–0.8 (–2.6, 1.0)	0.37	–3.3 (–4.9, –1.7)	‹0.001
Main language not English	–2.2 (–3.5, –0.8)	0.002	–2.1 (–3.2, –1.0)	‹0.001
Mother
Education		0.003		‹0.001
Did not complete high school	0 (–,–)		0 (–,–)
Completed high school	0.3 (–0.6, 1.1)	0.51	1.2 (0.4, 1.9)	0.001
Tertiary	–1.1 (–2.0, –0.2)	0.01	1.7 (1.0, 2.5)	‹0.001
Family
2 parents in the home	0.5 (–1.2, 2.1)	0.58	–0.1 (–1.4, 1.1)	0.82
Combined parental income quintile		0.24		‹0.001
Lowest	0 (–,–)		0 (–,–)
2nd	–0.3 (–1.7, 1.1)	0.67	0.1 (–1.2, 1.3)	0.92
3rd	–0.3 (–1.7, 1.1)	0.68	0.4 (–0.9, 1.8)	0.53
4th	0.2 (–1.9, 1.5)	0.71	1.1 (–0.1, 2.4)	0.08
Highest	0.8 (–0.5, 2.1)	0.22	2.3 (1.1, 3.5)	‹0.001
Financially stressed household	–1.5 (–3.2, 0.3)	0.09	–3.9 (–5.4, –2.3)	‹0.001
Highest occupational class		0.91		‹0.001
Neither parent working	0 (–,–)		0 (–,–)
ASCO 8–9 (Unskilled labour)	0.2 (–1.8, 2.2)	0.84	1.2 (–0.5, 2.9)	0.16
ASCO 4–7 (Skilled labour & clerical)	0.1 (–1.7, 1.8)	0.94	1.7 (0.3, 3.2)	0.020
ASCO 1–3 (Professional)	0.3 (–1.4, 2.1)	0.70	3.3 (1.8, 4.9)	‹0.001
Neighbourhood
SEIFA Disadvantage Index quintile		0.001		0.06
Highest disadvantage	0 (–,–)		0 (–,–)
2nd	2.1 (1.0, 3.3)	‹0.001	0.8 (–0.3, 1.9)	0.17
3rd	0.4 (–0.7, 1.5)	0.49	1.2 (0.2, 2.2)	0.02
4th	0.6 (–0.6, 1.8)	0.33	1.6 (0.4, 2.7)	0.006
Lowest disadvantage	0.5 (–0.8, 1.7)	0.44	1.4 (0.2, 2.7)	0.02

Table 7: Multivariable relationships between sociodemographic variables and the learning domain score for the infant and child cohorts

Characteristic(a)	Infants n=4,211 R2=1.6%		Children n=4,599 R2=13.5%
	Coefficient (95% CI)	p-value(b)	Coefficient (95% CI)	p-value(b)
Study child
Female	0.6 (0.0, 1.2)	0.05	4.4 (3.8, 4.9)	‹0.001
Aboriginal/Torres Strait Islander	–0.2 (–1.8, 1.4)	0.83	–1.5 (-3.1, 0.0)	0.05
Main language not English	1.0 (0.0, 2.1)	0.05	–1.4 (–2.3, –0.4)	0.004
Mother
Education		0.18		‹0.001
Did not complete high school	0 (–,–)		0 (–,–)
Completed high school	0.4 (-0.5, 1.2)	0.40	1.5 (0.9, 2.2)	‹0.001
Tertiary	–0.4 (–1.3, 0.5)	0.41	3.2 (2.4, 4.0)	‹0.001
Family
2 parents in the home	0.6 (–0.9, 2.2)	0.43	–0.5 (–1.8, 0.8)	0.46
Combined parental income quintile		0.11		0.21
Lowest	0 (–,–)		0 (–,–)
2nd	0.1 (–1.3, 1.5)	0.90	–0.5 (–1.6, 0.7)	0.41
3rd	–0.8 (–2.2, 0.6)	0.24	–0.3 (–1.6, 0.7)	0.62
4th	–0.9 (–2.2, 0.3)	0.15	0.2 (–1.0, 1.5)	0.70
Highest	–1.3 (–2.5, –0.1)	0.04	0.7 (–0.5, 2.0)	0.26
Financially stressed household	0.8 (–0.6, 2.3)	0.27	–2.3 (-3.6, -1.1)	‹0.001
Highest occupational class		0.07		‹0.001
Neither parent working	0 (–,–)		0 (–,–)
ASCO 8–9 (Unskilled labour)	–1.2 (–3.1, 0.8)	0.23	1.8 (–0.1, 3.6)	0.06
ASCO 4–7 (Skilled labour & clerical)	–1.9 (–3.4, –0.3)	0.02	2.5 (1.1, 3.9)	0.001
ASCO 1–3 (Professional)	–2.3 (–4.1, –0.5)	0.01	4.2 (2.7, 5.8)	‹0.001
Neighbourhood
SEIFA Disadvantage Index quintile		0.85		0.07
Highest disadvantage	0 (–,–)		0 (–,–)
2nd	0.5 (–0.8, 1.8)	0.48	0.2 (–0.9, 1.4)	0.68
3rd	0.3 (–0.9, 1.5)	0.57	0.8 (–0.3, 2.0)	0.16
4th	0.5 (–0.8, 1.7)	0.45	0.5 (–0.6, 1.6)	0.40
Lowest disadvantage	0.7 (–0.6, 2.0)	0.26	1.4 (0.3, 2.4)	0.01

2.3 Discussion

Overall, the picture painted by these data is that broad characteristics of the child, mother, and family context are quite powerfully related to 4 to 5 year-old children's development as reflected in the Outcome Index. This pattern of results for the child cohort provides some support for an ecological model of child development in which the child's own attributes, along with their family and community context, exert influence on developmental trajectories (Bronfenbrenner 1979).

In contrast, these analyses suggest only minor impacts of child, family and neighbourhood characteristics on infants' outcomes. The measures of outcomes were weaker in the infant than the child cohort, so the pattern of findings may partially reflect the lower sensitivity of the Outcome Index in this cohort. It may also reflect the fact that the impact of contextual factors on children's development is a cumulative process which occurs over time. Early measures of 'outcomes' may largely reflect infants' biological predispositions, with the cumulative influences of external factors (such as disadvantage) yet to develop over time; for infants, less time has elapsed for these to impact on development. Future waves of LSAC will enable testing of this hypothesis.

• Gender differences were small among infants, but marked for 4 to 5 year-old children except in the physical domain. The fact that girls develop faster than boys in early childhood, especially in language and the social–emotional domain, is well established in previous research (for example, Ruble & Martin 1998). This is thought to reflect principally biological dispositions, although differential parenting practices and expectations for boys and girls also appear to have some influence (Prior et al. 1993). These latter factors may be in part responsible for the greater gender difference in outcomes in the child, as compared to infant, cohort. We can expect that overall differences will lessen as the LSAC children move towards adolescence, but with continuing male vulnerability to difficulties such as learning problems and 'acting-out' behavioural disorders (Prior et al. 2000). While such gender differences are well established in previous research, they may need more recognition in policy and service provision contexts.
• Aboriginal and Torres Strait Islander children were lower on average on all Outcome Index measures except the physical domain, with marked differences in the proportions above and below the 15 per cent cut-offs. This contrasts with the general lack of evidence of a difference for the infant cohort (who were lower only on the physical domain). The data suggest that factors in the children's family and neighbourhood child-rearing environment may be responsible for a decline in functioning over time (Zubrick et al. 2004).
• Children who spoke a language other than English were lower on average on all Outcome Index measures, while there was little evidence of a difference for infants. The slightly lower learning domain scores might reflect the fact that the PPVT and Who Am I? tests were administered in English, in which case scores would be expected to improve in subsequent waves as mastery of English develops in the school setting. However, this would not explain the lower social–emotional or physical domain scores, which might reflect reduced opportunities for the child to interact with other children, or other aspects of the circumstances of families who have a non–English speaking background.
• Children whose mothers had more education were higher on average on all Outcome Index measures, whereas there were no observed trends for infants. Parental occupational status was a clear predictor of social–emotional and learning outcomes. Both income and financial stress made independent contributions to the prediction of child outcomes, with financial stress having the stronger impact. All these variables reflect the family psychological and material capital available to the child, and suggest that by 4 to 5 years of age these each make independent contributions to the child's overall development, creating socioeconomic gradients whose impact is likely to be carried forward through the child's school life and beyond.
• Family type (one or two parents), while associated with more negative and fewer positive Outcome Index scores in the bivariate analyses, failed to contribute to the multivariable models. This suggests that its influence is mediated through family variables which may be associated with family type, such as income, financial stress and occupational status, as well as other factors such as social support and parent–child relationships (Sanson & Lewis 2001; Wise 2003). Future waves of LSAC will enable examination of the dynamic impact of family structure and family transitions (see also Zubrick et al. 2008).
• Bivariate analyses of family size indicators suggested that infant outcomes were better the smaller the family, and in the absence of overcrowding. Neighbourhood liveability also showed some association with infant outcomes. Moderate sized families appeared optimal for those in the child cohort. Effects were relatively small and often non-linear, which may explain inconsistent findings from previous research. It is likely that these effects are moderated by family factors, a hypothesis which could be examined in further research.
• Neighbourhood disadvantage, as indexed by SEIFA, showed only weak associations with infant outcomes, perhaps reflecting differential access to services and/or support for infants and their families in more disadvantaged areas; this hypothesis could be examined in further analyses of the Wave 1 dataset. While there were some bivariate relationships with SEIFA quintiles, they did not predict child outcomes in the multivariable analyses for the child cohort. This may reflect the fact that at 4 to 5 years, the most salient environmental influence on the child is still the family, and that neighbourhood effects are mediated through family factors. In later waves we may see these neighbourhood factors having a more direct impact on the child.

In summary, these analyses indicate that the set of sociodemographic factors examined here have little impact on infants but explain substantial variability in the child cohort. Child, family and neighbourhood factors are all associated with outcomes, supporting an ecological model of child development and the need for multifaceted approaches to supporting families of young children. While we cannot draw causal implications from the findings, they indicate that the set of variables included in the multivariable analysis are important to include as 'control' variables when examining the impact of more fine-grained aspects of the child cohort's experiences and exposures, such as prenatal and postnatal health exposures, child care experiences and educational stimulation in the home. Hence, analyses in subsequent sections control for these variables. Despite the fact that they account for a very modest amount of variance in the infant cohort, for the sake of consistency they are used as covariates in analyses for this cohort also.

Return to top

Next: 3. Children's use of non-parental care

Previous: 1. Introduction

Copyright & Disclaimer | Privacy policy | Contact us | FaHCSIA home

© Commonwealth of Australia, 2008
Last modified 24 April, 2009