Access to early childhood education in Australia

Research Report No. 24 – April 2013

5. Differential access to early childhood education

In Section 4 we discussed the fact that access to ECE is not currently universal, as measured in terms of participation in ECE in the year before full-time school. In the literature and stakeholder consultations, several groups of children were identified as being at greater risk of missing out on access to ECE. This section now explores how rates of participation in ECE vary according to the range of factors discussed in Section 4, through statistical analyses of the three Australian datasets described in Table 5.

This section focuses on access in terms of participation in ECE. Clearly, this is a quite limited definition of access to ECE, considering the various dimensions of access that have been discussed in this report (especially in Section 4). Nevertheless, this measure of participation has the advantage of being easily understood and easily compared over jurisdictions and time. Compared to more sophisticated measures, it is also relatively easy to derive from existing datasets (although not without problems, as discussed in Section 4 and  Appendix B ).

The analyses in this section consider children to be in ECE in the year before full-time school if they are in either preschool or long day care. Preschool refers to ECE programs delivered through preschools or kindergartens, or other equivalent programs offered across Australia. Any participation in LDC is counted as ECE, regardless of whether parents reported that their children had a preschool program as part of LDC. It was felt that any LDC for children of this age is likely to involve a structured program, and would be expected to have some component of early learning built in. Also, the decision to include any LDC as ECE was partly due to data quality concerns about the distinction between LDC with and without preschool programs.

The type of ECE program (that is, LDC compared to preschool) will be the focus of Section 6 and so is not examined here.

The analyses presented here examine how characteristics of children, families and regions are related to different rates of access to ECE, to identify those factors that are related to lower levels of access. The focus is on those groups of children who are frequently acknowledged (by stakeholders and in the literature) to be likely to have higher rates of non-participation in ECE (sometimes referred to as being part of hard-to-reach families) (e.g., Walker, 2004).

This section examines how ECE participation varies with:

  • remoteness of regions;
  • socio-economic status of regions;
  • socio-economic characteristics of families (parental income, employment, single- versus couple-parent families, parental education);
  • Indigenous background of families;
  • non-English speaking background of families; and
  • children with special health needs.

Before beginning these analyses, we first present some analyses of parental decision-making, and barriers that may affect children's participation in ECE. As discussed in Section 4, stakeholders often see parents' decisions about children's non-participation in ECE as being the hardest barriers to both understand and address. These analyses, therefore, help to understand the potential role of parental beliefs on differential access to ECE. For example, these analyses can provide insights into the extent to which issues of accessibility and cost may present barriers to children's participation in early education (Moss, Appendix A; Press & Hayes, 2001).

5.1 Parental decision-making about participation in early childhood education and identified barriers

To explore parental decision-making regarding children's participation in ECE, three datasets were examined, making use of questions asked of parents about why their child did or did not attend preschool (or child care) in the year before full-time school. A full description of these analyses is presented in Appendix D.

In many cases, non-participation is reported across the three datasets to be because a parent is already available at home to provide care and therefore ECE is not necessarily needed; that is, reasons were often framed in relation to the idea of child care, rather than education. For example:

  • In the NSPCCC, the most common reason for children's non-participation in formal care or ECE identified from parents' responses was "belief in importance of home care" (22% of parents). In addition, a significant number of "other" responses were coded to a range of options, all of which suggested that a parent was at home to care for the child, and care was not needed (another 36% of parents, plus another 3% who also referred to the importance of home care).
  • Using LSAC (B cohort, Wave 3), parents of 4-5 year old children who were expected to start full-time school the next year but were not enrolled in ECE were asked why they did not use ECE. The largest response groups were "parent is available - not needed" (20%) and "child does not need it" (19%).
  • Using CEaCS, among children aged 4-8 years who were in school and had not attended preschool or LDC prior to school, the main reason given for non-attendance was "prefer to care for child at home" (73%).

While these responses give the perception that for many parents, non-participation in ECE is a conscious choice, this may be too simplistic an interpretation. It would be useful to gain an understanding of how parents come to this arrangement, and to examine whether parents understand about the availability and benefits of ECE. As discussed in Section 7, qualitative research would be the best source of this information. We have used the survey data here to provide some initial insights.

Within each survey, there were a number of respondents who cited reasons for a child's non-participation in preschool (or child care) that were more indicative of barriers.

  • In NSPCCC, of children who were in the year before full-time school but not in formal care or ECE, 41% of parents gave reasons other than parental availability (although they may have also provided a response around parental availability, as multiple responses were permitted), including that the ECE arrangement (care or preschool) was too expensive (16%), the parent lacked trust in formal child care (9%), the parent already had friends or family looking after the child (6%), the ECE was too far away (6%), or the ECE was too difficult to get into (1%). Quite a large number of responses were recorded as "other" (14%).
  • In NSPCCC, parents were asked why their school-aged children had not attended preschool (5% of children, N = 59). Of these, 6% gave answers indicating that there were no places available and 14% that they could not afford it. However 82% of responses were coded to "other reasons" and could not be further identified. (Percentages add to more than 100% as multiple reasons could be given.)
  • In the LSAC sample, in addition to those reasons mentioned in relation to LSAC above, other reasons given were "can't afford it - cost too high" (16%), "other - quality/program issues" (12%), "child is too young or old" (10%), "problems with getting places" (9%), and "other - accessibility or affordability" (7%).
  • In CEaCS, after "prefer to care for child at home", the next most common reason for children not attending preschool or LDC was "other reasons" (15%), followed by "moved from interstate or overseas" (6%).

More explanatory information regarding the questions and response options upon which these analyses are based is given in Appendix D.

Additional qualitative research about parents' decisions regarding the use of non-parental care in the year before full-time school supports the complexity of findings detailed here. For example, when asked about their decision not to use child care, mothers in the Family and Work Decisions' Study would often state more than one concern about this form of care (Hand, 2005). Like the parents in the above studies, not needing child care because a parent was home was usually the primary reason; however, issues of trust, affordability and a lack of places at their preferred service were also frequently cited. Furthermore, many of those mothers who chose not to work in order to care for their children did so due to concerns about trust and affordability.

It is important to understand which families report the different reasons cited above. Sample size limitations mean these data cannot be analysed comprehensively by the socio-demographic characteristics examined in the rest of this section; however, where possible, we do highlight some key findings that emerge.

5.2 Factors influencing participation in early childhood education

The rest of Section 5 presents analyses of how ECE participation varies by child, family and regional characteristics. These analyses were conducted using each of the NSPCCC, LSAC and AEDI datasets. The analyses compare children in ECE (preschool and/or LDC) to those not in ECE. The latter includes those only in parental care, informal care or non-ECE types of formal care.

The analyses include some straightforward tabulation of participation rates by those factors listed previously. In addition, as multiple factors are likely to be important in explaining how participation in ECE varies, it is appropriate to use multivariate analyses. This allows us to determine whether particular factors have independent associations with ECE, once other characteristics are taken into account. For example, we begin by examining how ECE participation varies with the remoteness of the region in which children live. By using multivariate analyses, we can see whether remoteness is a significant factor in explaining differences in rates of ECE participation when other characteristics, such as the Indigenous status of children and socio-economic status of their families, are taken into account.

All analyses focus on children who were, or were predicted to be, in the year before full-time schooling. For NSPCCC, child age (in months) was compared to state/territory eligibility regarding school starting age, and used alongside information on current ECE or school participation to determine whether children were likely to be in the year before full-time school. For LSAC, children were identified by their responding parent as being in the year before full-time schooling. For AEDI, data on ECE participation before commencing school was collected retrospectively for all children, as the study children were already attending school at the time of the collection.

The following subsection describes the methods used in the multivariate analyses. Refer in particular to Box 1, which describes how to interpret the findings from the multivariate analyses.

Description of multivariate methods and summary of findings

Multivariate analyses were used to identify those characteristics associated with children being more likely to participate in ECE. As discussed above, LDC and preschool were counted as ECE. Classifying children as being in ECE or not in ECE in this way results in a binary classification, which can be modelled using logistic regression.

Where possible, the analyses take account of the child's Indigenous status, having English as a second language, having special health care needs, child age, parental employment and education, household income, remoteness and socio-economic status of the region. Exact details vary depending upon which data source is used (as different information is available for each dataset). See Appendix E for further information about the details of these models.

In all models, an indicator of whether children lived in the larger eastern states of NSW, Victoria or Queensland (as opposed to South Australia, Western Australia, Tasmania, NT and ACT) was used. This distinction was used because ECE tends to be delivered in different ways within these broad grouping of states/territories; in particular, there is a greater reliance on LDC in the eastern states (see earlier discussion in Section 2).

First, an overall model including all children was estimated from each data source. Second, using the AEDI (given the very large size of the dataset), similar analyses were undertaken for each state/territory individually. These models omit the various family characteristics, as these were not available in the AEDI data. Then, in further analyses, the Australian models were replicated for families with specific characteristics (by remoteness, Indigenous versus non-Indigenous, whether English is main language spoken), using the AEDI. Only the AEDI contained a sufficiently large sample to satisfactorily undertake this more disaggregated analysis.

The results from each of the models are presented as odds ratios, which can be interpreted as shown in Box 1.

Box 1: Interpretation of multivariate results

Results from logistic regresssions are presented in this report as odds ratios (OR). The "odds" of having a particular outcome is the probability of having it, expressed as a ratio of the probability of not having it. Odds ratios are an estimate of how the "odds" vary for those with and without a particular characteristic.

In these analyses, the odds ratios provide an indication of whether being in ECE is more likely (when the odds ratio is greater than 1) or less likely (when the odds ratio is less than 1) for those with a particular characteristic, compared to those not having this characteristic. The size of the odds ratio indicates how much participating in ECE varies according to this characteristic. Thus, if the odds ratio is greater than 1, the larger the number is, the more likely it is that the child is participating in ECE. If the odds ratio is less than 1, the closer the number is to 0, the less likely it is that the child is participating in ECE. When the odds ratio is equal (or close to) to 1, there is no difference between those with and without that characteristic in their likelihood of participating in ECE. This applies, for example, when examining indicator variables such as Indigenous status of the child, non-English language status and special health care needs status.

In the case of variables with more than two categories (such as remoteness and primary carer's employment), a reference (ref.) category is established. This category is identified in the results tables and is always shown with an odds ratio of one. The odds ratios for other categories then compare the odds of being in ECE for each of those categories with that of the reference category. For example, in Table 7, for remoteness, the reference category is "major cities". The odds ratios are lower for all of the other categories, indicating that ECE participation is lower in all other remoteness areas than it is in major cities.

Note that a limitation is that these odds ratios only allow comparison back to the reference group in a strict sense, although the relative size of the other odds ratios can be used as a guide to how participation in ECE compares across other groups. For example, using the AEDI results in Table 7, the odds ratios for outer regional areas and remote or very remote areas are 0.76 and 0.55 respectively. These odds ratios are based on comparisons for each group to the reference group of major cities. The relative size of these odds ratios suggests that ECE participation is less likely in the more remote areas, compared to outer regional areas. However, further statistical tests would be required to assert this with certainty.

The stars in the table indicate the statistical significance of each odds ratio. If there are no stars on a figure, this indicates that, according to conventional levels of significance, this odds ratio does not differ significantly from 1; that is, this characteristic is not significantly associated with ECE participation. A greater number of stars indicate that we have greater confidence that this variable has a significant association with ECE participation. Looking at the remoteness example in Table 7, the difference between inner regional areas and major cities in children's participation in ECE is not statistically significant for NSPCCC, but is for AEDI.

The results of the overall multivariate analyses are presented in Table 7, and the state/territory analyses in Table 8. These results are discussed in the subsections that follow, taking one characteristic at a time. We first consider local area characteristics (remoteness and socio-economic status of regions). The socio-economic status of regions is measured using the Socio-Economic Index for Areas (SEIFA) score of relative disadvantage, which captures information about local area disadvantage, such as low income, low educational attainment, high unemployment and relatively unskilled occupations (Trewin, 2004). The distribution of scores in the dataset was used to classify children as living in areas with SEIFA scores in the bottom 20%, the middle 60% or the top 20%. Then socio-economic characteristics of families are examined, including family income, parental employment, single- versus couple-parent families, and parental education. Finally, the results are presented for Indigenous background, non-English speaking background and children with special health needs. State/territory differences are discussed throughout these analyses.

Table 7: Multivariate analyses of which children are in early childhood education in year prior to full-time schooling, AEDI, NSPCCC and LSAC
  AEDI (OR) NSPCCC (OR) LSAC (OR)
Eastern states (NSW, Vic., Qld)
(ref. = All others)
0.78*** 0.95 0.04***
Locational factors
Remoteness (ref. = Major cities) 1.00 1.00 1.00
Inner regional areas 0.91*** 0.86 0.88
Outer regional areas 0.76*** 0.62* 0.53**
Remote or very remote areas 0.55*** 0.50*** 0.38
Socio-economic status of region
(ref. = Most disadvantaged, bottom 20%)
1.00 n.a. 1.00
Middle advantage, middle 60% 0.88***   1.24
Most advantaged, top 20% 1.24***   1.48
SEIFA not available (NT in AEDI) 1.84***   n.a.
Socio-economic status of families
Family income (ref. = Higher incomes, top 20%) n.a. 1.00 1.00
Lower incomes, bottom 20%   0.75 0.49
Middle incomes, middle 60%   0.85 0.72
Parental employment (ref. = Not employed) n.a. 1.00 1.00
Employed part-time   1.96*** 1.90***
Employed full-time   2.71*** 1.39
Single parent (ref. = Couple parent) n.a. 1.58 1.35
Parental education (ref. = Incomplete secondary) n.a. 1.00 1.00
Secondary, diploma or certificate   1.62** 1.59*
Bachelor degree or higher   1.53* 3.21***
Indigenous child 0.53*** 1.09 0.26***
Non-English speaking background 0.46*** 0.79 0.63
Special health care needs 0.70*** n.a. 1.29
Age at survey (months) n.a. 0.93*** 1.05
Constant 13.06*** 203.88*** 11.38
Sample size 236,206 1,610 2,936

Note: In these analyses, enrolment in any preschool or LDC was counted as being in ECE. See Appendix E for further information about the variables listed. * p < .05; ** p < .01; *** p < .001.

Source: AEDI (2009); NSPCCC (2009); LSAC (2008)

We can see that the results do vary somewhat according to which source is used, which may reflect the differences in how ECE participation was captured in each source and in the timing of collections (NSPCCC referred to 2009, while the others referred to 2008). (See Table 4 for descriptions of the different data sources.) Further, the models are not exactly the same; in particular, with family-level details not available for AEDI. This may mean the regional and child-level characteristics are capturing more of the variation than they would, had family-level been available for inclusion in the analyses.

Table 8: Multivariate analyses of which children are in early childhood education in the year before full-time schooling in each state and territory, AEDI
  NSW ( OR) Vic. ( OR) Qld ( OR) SA ( OR) WA ( OR) Tas. ( OR) NT ( OR) ACT ( OR)
Locational factors
Remoteness (ref. = Outer regional areas) 1.00 1.00 1.00 1.00 1.00 1.00 1.00  
Major cities 0.85*** 1.18* 1.26*** 1.04 1.35*** n.a. n.a. n.a.
Inner regional areas 1.00 0.89 1.06 0.94 1.09 1.97*** n.a.  
Remote or very remote areas 0.71*** 0.40*** 0.78*** 0.74 0.85 0.80 0.86  
Socio-economic status of region (ref = Most disadvantaged, bottom 20%) 1.00 1.00 1.00 1.00 1.00 1.00 n.a. n.a.
Middle advantage, middle 60% 1.13** 1.05 0.72*** 1.00 0.68*** 1.62***   1.11
Most advantaged, top 20% 2.15*** 1.21* 1.18 1.22 0.77* 1.33   1.00 a
Indigenous child 0.48*** 0.35*** 0.67*** 0.57*** 0.66*** 0.84 0.60** 0.53
Non-English speaking background 0.42*** 0.36*** 0.52*** 0.42*** 0.56*** 0.25*** 0.47*** 0.61**
Special health care needs 0.88* 0.46*** 0.51*** 0.92 0.77* 0.56* 0.86 0.65
Constant 8.35*** 16.90*** 5.46*** 17.79*** 9.39*** 8.23*** 14.86*** 19.06***
Sample size 80,277 56,713 47,034 14,814 25,291 5,306 2,796 3,975

Note: In these analyses, enrolment in any preschool or LDC was counted as being in ECE. See Appendix E for further information about the variables listed. a As no ACT areas were classified as being in the most disadvantaged category, the reference category here was changed to the most advantaged areas. * p < .05; ** p < .01; *** p < .001

Source: AEDI (2009)

With the AEDI data, all of the variables included explain a significant amount of the variation in rates of participation in ECE, with lower participation in more remote areas; higher participation in more advantaged areas (although "moderate" advantage had lower participation than the most disadvantaged areas); lower participation in the eastern states; and lower participation among Indigenous children, NESB children and children with special health care needs. For NSPCCC, lower participation rates were apparent for the more remote areas, while higher participation rates were apparent when parents had higher levels of educational attainment and were employed part-time or full-time. An unexpected finding is that the association with age of child revealed lower participation in ECE among the older children (remembering these are all children in the year before full-time school). Using LSAC, children in outer regional areas have significantly lower participation rates than those in major cities. (The difference was not statistically significant for remote areas.) Participation rates were lower for children in the eastern states, and for Indigenous children. As in the NSPCCC, participation rates were higher when parents had higher education levels and when they were employed. In these data, participation rates were only significantly higher when the primary carer was employed part-time.

The multivariate analyses conducted for states and territories separately, using the AEDI, revealed some consistent (or near consistent) findings (Table 8). For example, lower participation rates in ECE were apparent in most states and territories for Indigenous children (the exceptions being Tasmania and the ACT). In all states and territories, NESB children had lower participation rates. There was more variation across states and territories in respect to how ECE participation varied according to the remoteness of the area and the disadvantage of the area.

These findings are discussed further in the sections that follow.

5.3 Locational factors

Children's participation in ECE is likely to be dependent upon the ECE options available in the child's community, which are likely to vary across states/territories in Australia. These different patterns of ECE participation are apparent in the earlier presented information (Section 2), as well as in tables derived from the three surveys: the NSPCCC (Table C1), LSAC (Table C2) and AEDI (Table C3).

Given the different options of ECE delivery across Australia, when considering locational factors, we would ideally incorporate information about the range of ECE services (the number, type and costs) within the locality of the child, to ascertain to what extent the supply of services affects the uptake of those services. As this local area information was not available for inclusion in these analyses, some state-based analyses have been included as an alternative approach, to model the different ECE systems in place across different jurisdictions. As described above, in the multivariate analyses, a broader indicator of eastern versus other state/territories is used.

Within states/territories of Australia there is also likely to be variability in the ECE options available to families. In Australia, regional variation is often considered in respect to two measures of location - the remoteness of regions (measured in terms of distance to service centres) and the socio-economic status of regions. We focus on these two locational factors in the following subsections.

Remoteness

Differences in ECE participation according to remoteness were discussed in subsection 4.2. Existing research, and consultations in this project, led to the expectation of finding lower levels of ECE participation in the more remote areas of Australia. The data examined here support this. According to each of the data sources examined, as remoteness increased, children were less likely to be in ECE prior to starting their first year of full-time schooling (Table 9).

Table 9: Remoteness of location and percentage of children not participating in early childhood education in the year before full-time schooling, AEDI, NSPCCC and LSAC
  Major cities (%) Inner regional (%) Outer regional (%) Remote or very remote (%) Australia (%) Sample size (N)
AEDI 10.0 10.7 12.9 19.0 10.8 236,253
NSPCCC 16.0 19.4 23.8 26.6 17.9 1,637
LSAC 6.2 6.6 11.3 12.1 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates.

Source: AEDI (2009); NSPCCC (2009); LSAC (2008)

In the multivariate analysis, differences between major city areas and remote/very remote areas were statistically significant in AEDI and NSPCCC (see Table 7). For example, according to the AEDI, the odds ratio of 0.55 for remote/very remote areas indicates that in these remote areas, the odds of children participating in ECE was 0.55 that of children living in major city areas of Australia. Differences were also apparent when comparing major city areas to inner regional areas and outer regional areas, although the difference between major city areas and inner regional areas was only statistically significant in the AEDI data. In LSAC, the difference in ECE participation was only statistically significant in comparing major city areas to outer regional areas; however, that study is not designed to be representative of families living in remote areas of Australia.

Looking at the AEDI state/territory analyses (Table 8), there was some variation across state/territories in regard to the association between remoteness and ECE. In NSW, children living in major city areas were less likely to be in ECE than were children in outer regional areas. In all other states/territories, being in a major city area was associated with relatively high (or at least equal) rates of ECE access when compared to other regions. It was in the three eastern states that the differences for remote areas were most apparent.

It is worth noting that the multivariate analyses take account of other characteristics, including Indigenous status of children, and socio-economic status of regions. As remoteness areas vary in terms of these characteristics, when considering outcomes in particular regions, it is important to be mindful also of these factors, which are discussed separately in this section. Nevertheless, these analyses indicate that, overall, and in most states (but especially the three eastern states), remoteness was associated with lower levels of access to ECE.

Socio-economic status of regions

In our previous discussion of variation in ECE according to the socio-economic status of regions, we noted that some research has found lower rates of ECE participation in areas of greater financial disadvantage. However, we expect that this picture may be quite complex, as in some disadvantaged areas, there may actually be targeted provision of services, including ECE, to help address the needs of low-income families. Further, even within disadvantaged regions we expect there to be considerable heterogeneity of families, and like parents in other regions of Australia, there will be those who ensure their children participate in ECE, whether that ECE is provided locally or outside their region of residence.

For these analyses, the socio-economic status of the region in which children live was available for the AEDI (although not for those living in NT) and for LSAC. These were based on the SEIFA score of the community in AEDI and the Statistical Local Area (SLA) in LSAC, and in both datasets was measured using the SEIFA score of relative disadvantage.

According to the AEDI data, children living in regions with a relatively high socio-economic status were the most likely to participate in ECE in the year prior to their first year of full-time schooling (Table 10).

Table 10: Socio-economic status of region and percentage of children not participating in early childhood education in the year before full-time schooling, AEDI and LSAC
  Most disadvantaged (bottom 20%) (%) Moderate advantage (middle 60%) (%) Most advantaged (top 20%) (%) Australia (%) Sample size (N)
AEDI a 14.0 12.3 8.0 10.8 233,412
LSAC b 12.0 6.0 3.4 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates. a In the AEDI data, the SEIFA classification was not available for NT, so while the Australia total for AEDI includes NT, the NT data are not included in the SEIFA categories. b LSAC data are based on the SEIFA index of disadvantage from the 2006 Census, matched to SLAs.

Source: AEDI (2009); LSAC (2008)

In the multivariate analyses (Table 7), taking account of some other characteristics of children (including family characteristics for NSPCCC and LSAC; but not for the AEDI, as these characteristics were not available with this dataset), the odds of participating in ECE was higher in the most advantaged regions than in the least advantaged regions (e.g., using AEDI, an odds ratio of 1.24 indicates that the odds of children being in ECE in the most advantaged regions were 1.24 times that of the odds of being in ECE in the least advantaged regions). Interestingly, these analyses showed that compared to the least advantaged regions, the likelihood of participating in ECE was actually lower for children in the middle category of socio-economic disadvantage. This highlights that the association between the socio-economic status of regions and ECE participation is not a straightforward one.

Further, Table 8 shows that it is also important to examine these data by state/territory. According to the AEDI, differences between the lowest and highest socio-economic status regions are especially apparent in NSW, and such differences are not apparent in all states/territories. In fact, in the multivariate analyses of WA, it is children in regions with higher socio-economic status who had the lower rates of participation in ECE prior to their first year of full-time schooling, compared to those in regions of lower socio-economic status. Comparing the lowest and the middle categories of socio-economic status, we see that children living in regions classified as being in the middle have a lower likelihood of participating in ECE if living in Queensland or Western Australia, but a higher likelihood of participating in ECE if living in NSW or Tasmania. These findings portray a complex and mixed pattern in relation to the association between regional level disadvantage and ECE participation.

Using the LSAC data, at the national level, the overall percentage in ECE did not vary with socio-economic status of the region in the multivariate analysis, after taking account of family-level characteristics. While not statistically significant, the odds ratios were consistent with Table 10, which shows, overall, somewhat higher rates of non-participation in ECE in the more disadvantaged regions.

It should be noted that in analysing socio-economic status of regions using the AEDI data, we were not able to also take account of the socio-economic status of families. To some extent, then, associations attributed here to regional effects may actually represent some effects of family characteristics. The LSAC analyses included information on the socio-economic status of families (for example, income, employment, single parenthood), as well as of the region. This may account for the different findings from the two datasets.

5.4 Socio-economic status of families

In this subsection, consideration is given to a range of variables that capture family socio-economic status. In particular, the focus is on family income, family employment, single versus couple parenthood, and parental education. Analyses of ECE according to these characteristics allows us to consider whether rates of access are lower for children whose parents are on lower incomes, are jobless, are single parents or have relatively low levels of educational attainment. As discussed in subsection 4.2, and as with the above analyses of socio-economic status of regions, we do not expect these associations to be straightforward. This is especially so given that these different measures of socio-economic status are likely to be linked in some way. Therefore, it may be difficult to disentangle which factors have the greatest effects on family decision-making with regard to ECE participation. Overall, though, we expect to find lower rates of participation in ECE in more financially disadvantaged families.

The analyses use LSAC and NSPCCC data, as family characteristics are not available in the AEDI.

Family income

To examine family income and child participation in ECE, the income of each family in LSAC and NSPCCC was ranked from lowest to highest, and then each sample was divided into three groups: those with relatively low incomes (in the bottom 20%), those with relatively high incomes (in the top 20%) , and those who make up the middle 60% of family incomes.

Table 11 shows that a higher percentage of children of lower income families were not in ECE, when compared to families with moderate or higher incomes.

Table 11: Family income and percentage of children not participating in early childhood education in the year before full-time schooling, NSPCCC and LSAC
  Lower incomes (bottom 20%) (%) Middle incomes (middle 60%) (%) Higher incomes (top 20%) (%) Australia (%) Sample size (N)
NSPCCC 23.4 14.3 11.6 17.9 1,637
LSAC 13.5 5.8 2.4 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates. The Australia total includes families with missing information about income.

Source: NSPCCC (2009); LSAC (2008)

The multivariate analyses presented in Table 7 show no significant associations between being in ECE and family income. However, these models also include parental employment and education, which will be strongly associated with parental income. If these models are re-estimated without parental education and employment, the results show significantly lower levels of access to ECE by children in lower income families (results are shown in Table C4).

In later analyses of the LSAC data (Figure 3), associations between parental income and participation in ECE are considerably stronger for families in the eastern states, compared with the vast majority of children in other states/territories in ECE, regardless of family income.

Parental employment

As discussed in subsection 5.1, when children were not in early education, parents very often report that this was because a parent is available to care for children. In NSPCCC, 59% of parents of children not in ECE gave a response related to the importance of home care for children or the availability of a parent. When disaggregated by the employment status of the primary carer for children not in early education (usually the mother), these reasons were given by 34% of full-time employed primary carers, 54% of part-time employed primary carers and 63% of not-employed primary carers. This indicates that reasons for child non-participation in ECE do vary according to the primary carer's employment status. Among all families, however, there was a great deal of variability in reasons for non-participation, regardless of parental employment status.

When analysed using NSPCCC and LSAC, there were some clear differences in child participation in ECE according to the employment status of the primary carer. If the primary carer was not employed, children were less likely to be in ECE. This was particularly apparent in the NSPCCC. Differences were statistically significant in the multivariate analysis (Table 7) as well as the descriptive analyses (Table 12).

Table 12: Employment status of primary carer and percentage of children not participating in early childhood education in the year before full-time schooling, NSPCCC and LSAC
  Not employed (%) Employed part-time (%) Employed full-time (%) Australia (%) Sample size (N)
NSPCCC 24.3 13.4 8.8 17.9 1,637
LSAC 10.9 3.7 7.1 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates.

Source: AEDI (2009); LSAC (2008)

According to the LSAC data, associations between parental employment and rates of access to ECE were most apparent in the eastern states. This is shown in Figure 4, in the later analyses of types of ECE.

Single versus couple parents

We also noted in subsection 4.2, that concerns related to the ECE participation of children in single- rather than couple-parent families were not particularly apparent in the literature or the stakeholder discussions.

In the current analyses of LSAC and NSPCCC, differences in rates of participation in ECE between children of single- and couple-parents also were not apparent in the multivariate analysis. In addition, looking at the overall differences in these samples in Table 13, there was little difference in the percentage of children not in ECE; that is, these data do not provide any evidence that if the primary carer is a single parent, children are at particular risk of missing out on ECE.

Table 13: Primary carer's relationship status and percentage of children not participating in early childhood education in the year before full-time schooling, NSPCCC and LSAC
  Single parent (%) Couple parent (%) Australia (%) Sample size
% children not in ECE N
NSPCCC 17.8 17.9 17.9 2,637
LSAC 9.8 6.7 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates.

Source: AEDI (2009); LSAC (2008)

Parental education

A key objective of ECE is to improve children's readiness for school. This is likely to be particularly beneficial for children who are not exposed to early learning activities, such as reading, in the home. One indicator of early learning activities in the home is parental education (Barnett & Yarosz, 2007). Therefore, it is important to consider to what extent children who have parents with relatively low levels of education are accessing ECE.

The association between parental education and early learning in the home is apparent if analysed with the LSAC sample: among those children whose primary carer had incomplete secondary education, 19% were not read to in the past week, compared to 2% not having been read to in the past week when the primary carer had a bachelor degree or higher.

Low parental education is also likely to be strongly associated with other risk factors for children's learning, such as financial disadvantage. In the LSAC data, when the primary carer had incomplete secondary education, 36% of those families had an income within the bottom 20% of the income distribution for families in the sample, compared to 10% of families in which the primary carer had a bachelor degree or higher.

Table 14 shows rates of participation in ECE by the highest level of education achieved by the primary carer, using NSPCCC and LSAC. These analyses show that children of primary carers with lower levels of education were the least likely to be in ECE and this is confirmed in the multivariate analysis.

Table 14: Primary carer's level of education and percentage of children not participating in early childhood education in the year before full-time schooling, NSPCCC and LSAC
  Incomplete secondary education (%) Secondary education or diploma/ certificate (%) Bachelors degree or higher (%) Australia (%) Sample size (N)
NSPCCC 23.4 18.1 15.8 17.9 2,637
LSAC 12.8 7.2 2.6 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates.

Source: LSAC (2008); NSPCCC (2009)

In the context of the home learning environment, it is also useful to consider the association between reading in the home and children's participation in ECE. This is possible using LSAC. Of the 213 children in the sample who were not in any form of early childhood education, 16% were not read to in the last week, 28% were read to on 1 or 2 days, 26% were read to on 3, 4 or 5 days and 31% were read to on 6 or 7 days. This compares to the overall sample average of 6% not read to in the last week, 17% read to on 1 or 2 days, 27% read to on 3, 4 or 5 days and 49% read to on 6 or 7 days. It is therefore important to note that many children without formal early childhood education are also likely to not be getting high levels of early learning opportunities at home; that is, their parental care in the year before their first year of full-time schooling will not always be a good substitute for formal ECE in relation to getting the children school-ready.

5.5 Indigenous children and families

The relatively low participation in ECE by Indigenous children is well documented (see subsection 4.2). Table 15 shows clear differences in rates of participation in ECE for Indigenous children when compared to non-Indigenous children. For Indigenous children, rates of non-participation in ECE were 21% in the AEDI, 26% in LSAC, and 31% in the NSPCCC. For non-Indigenous children, rates of non-participation were much lower, at 10% (AEDI), 6% (LSAC) and 18% (NSPCCC).

Table 15: Indigenous status and percentage of children not participating in early childhood education in the year before full-time schooling, AEDI, NSPCCC and LSAC
  Not Indigenous (%) Indigenous (%) Australia (%) Sample size (N)
AEDI 10.3 21.0 10.8 236,284
NSPCCC 17.6 30.6 17.9 1,637
LSAC 6.1 26.2 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates.

Source: AEDI (2009); NSPCCC (2009); LSAC (2008)

The multivariate analysis of preschool access (Table 7) shows that, holding other characteristics of children and families constant, Indigenous children had lower rates of preschool enrolment when analysed using AEDI and LSAC. Statistically significant differences were not observed in NSPCCC, even though Table 15 shows relatively high levels of non-participation among Indigenous children for this sample. (The non-significance may be related to the small number of Indigenous children in the sample.)

In the state/territory-specific multivariate analyses of AEDI (Table 8), Indigenous children had lower rates of participation in ECE in all but two states - Tasmania and ACT.

The overall lower rates of participation by Indigenous children are consistent with the findings of Biddle (2007), who analysed preschool participation of 3-5 year old children using the 2001 Australian Census. He found that within this broader age group, after taking account of a range of family and child characteristics, Indigenous children had lower participation rates in preschool. Biddle undertook additional analyses by estimating models for preschool participation separately within the Indigenous population and within the non-Indigenous population. This allowed comparison of the two models to see whether predictors of preschool participation (such as lower parental education or remoteness) had the same effect on participation for Indigenous and non-Indigenous children. Biddle found that living in a remote area, and living in a household with low income and lower levels of education in the family had a stronger negative effect on children's participation in preschool for Indigenous compared to non-Indigenous children.

This approach has been replicated here using the AEDI, with the multivariate analyses of ECE participation done separately for Indigenous and for non-Indigenous children. Table 16 shows that a number of the factors that explain lower rates of participation in ECE by non-Indigenous children are not statistically significant for Indigenous children. Specifically, among Indigenous children, preschool participation did not vary according to whether English was a second language nor according to whether the child was identified as having special needs; however, these characteristics were related to lower rates of participation among non-Indigenous children. Among Indigenous children, those in remote parts of Australia had relatively low rates of enrolment in ECE, as was also true for non-Indigenous children; that is, lower rates of participation were apparent for Indigenous children, especially Indigenous children in remote areas. There were also state-level differences that were consistent with findings for non-Indigenous children. The SEIFA results were somewhat difficult to interpret for Indigenous children, as the SEIFA classification was not available for those living in the NT, and Indigenous children are over-represented in this state.

Table 16: Multivariate analyses of which Indigenous and non-Indigenous children are in the year before full-time schooling, AEDI
  Non-Indigenous ( OR) Indigenous ( OR)
Eastern states (NSW, Vic., Qld) (ref. = All others) 0.79** 0.61***
Locational factors    
Remoteness (ref. = Major cities) 1.00 1.00
Inner regional areas 0.89*** 1.08
Outer regional areas 0.74*** 0.96
Remote or very remote areas 0.45*** 0.64***
Socio-economic status of region (ref. = Most disadvantaged, bottom 20%) 1.00 1.00
Middle advantage, middle 60% 0.94 0.79**
Most advantaged, top 20% 1.32*** 0.94
SEIFA not available (Northern Territory) 1.88*** 0.91
Non-English speaking background 0.42*** 1.00
Special health care needs 0.69*** 0.91
Constant 12.46*** 6.97***
Sample size 226,016 10,190

Note: Refer to Appendix B for important notes regarding these estimates.

Source: AEDI (2009)

5.6 Non-English speaking background families

This section examines the extent to which children's enrolment in ECE varies when they are from a non-English speaking background. For this study, whether or not the child is NESB provides the best indicator available for cultural and linguistic diversity. The actual indicator of NESB in the report varies according to the source of the data. The AEDI used information on whether children had English as a second language. LSAC used information on the main language the children spoke at home. NSPCCC used information on the main language the survey respondent spoke at home, since similar information was not collected in respect to the children. Note that this only captures ethnicity in very broad terms, which may not be sufficient for examining issues for children from particular cultural or ethnic groups - that would require a more detailed study (Wise & Da Silva, 2007).

Table 17 shows some evidence of NESB children having somewhat lower rates of participation in ECE. This is most apparent in the AEDI data, but somewhat lower rates of participation in ECE are also apparent for these children in LSAC and NSPCCC.

Table 17: NESB status and percentage of children not participating in early childhood education in the year before full-time schooling, AEDI, NSPCCC and LSAC
  English-speaking (%) NESB (%) Australia (%) Sample size (N)
AEDI 9.6 19.0 10.8 236,284
NSPCCC 17.8 20.5 17.9 1,637
LSAC 6.5 10.4 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates.

Source: AEDI (2009); NSPCCC (2009); LSAC (2008)

The multivariate analyses (Table 7) found that, holding other characteristics of children and families constant, NESB children had relatively low rates of ECE participation when analysed using AEDI, but not using NSPCCC and LSAC. In the state/territory-specific models of AEDI (Table 8), NESB children had lower ECE participation in each state/territory.

Additional models were estimated to look at whether the factors predicting ECE participation differed for children according to whether they were from a non-English speaking background (Table 18). The model estimated specifically for NESB children lacked statistical significance for the characteristics of children being Indigenous or having special needs. These characteristics were only statistically significant for children who were not NESB.

Table 18: Multivariate analyses of which NESB and ESB children are in early childhood education in the year before full-time schooling, AEDI
  NESB ( OR) ESB ( OR)
Eastern states (NSW, Vic., Qld) (ref. = All others) 0.73*** 0.78***
Locational factors    
Remoteness (ref. = Major cities) 1.00 1.00
Inner regional areas 0.91 0.90***
Outer regional areas 0.81*** 0.76***
Remote or very remote areas 0.76*** 0.46***
Socio-economic status of region (ref. = Most disadvantaged, bottom 20%) 1.00 1.00
Middle advantage, middle 60% 0.94 0.90**
Most advantaged, top 20% 1.20*** 1.29***
SEIFA not available (Northern Territory) 0.96 1.88***
Indigenous child 0.96 0.46***
Special health care needs 0.99 0.64***
Constant 5.63*** 12.91***
Sample size 28,817 207,389

Note: Refer to Appendix B for important notes regarding these estimates.

Source: AEDI (2009)

5.7 Children with disabilities or special health care needs

The next analyses examine differences in participation in ECE for children with disabilities or special health care needs. The identification of children in this category varied across sources. The AEDI used the indicator of children having special needs, LSAC uses the indicator of children having special health care needs and the NSPCCC did not include an indicator of health or disability.

According to the AEDI, children with special needs were somewhat less likely than other children to have been in ECE prior to starting their first year of full-time schooling (Table 19). This indicator was statistically significant in the multivariate analysis (Table 7), although in the state/territory analyses (Table 8), it was not statistically significant for SA, NT and ACT. This difference was not reflected in the LSAC data. The different findings from each of these sources may reflect the different indicator variables used to identify children with special (health care) needs.

Table 19: Special needs/health care status and percentage of children not participating in early childhood education in the year before full-time schooling, AEDI and LSAC
  No special (health care) needs (%) Has special (health care) needs (%) Australia (%) Sample size (N)
AEDI 10.6 15.0 10.8 236,284
LSAC 7.4 5.4 7.1 3,005

Note: Refer to Appendix B for important notes regarding these estimates.

Source: AEDI (2009); LSAC (2008)

As with other analyses, within AEDI and LSAC, children with special needs are likely to be a very diverse group in relation to the nature and severity of their health care needs. To fully understand the issues for ECE participation for these children, such characteristics would need to be examined. It is beyond the scope of this study to look into such details more comprehensively.

5.8 Summary

Section 5 has presented analyses of children's participation in ECE to help understand which children are missing out on ECE, and to help identify particular barriers to these children's inclusion in ECE.

The reports by parents about the non-participation by children in ECE provided some indication that a small proportion of children are missing out because of barriers - or perceived barriers - in regard to cost, availability, accessibility or appropriateness. However, parents of non-attending children were more likely to say their children were not in ECE because of reasons related to the availability of a parent to care for children, or related to a belief in parental care of children. While this suggests some degree of choice by these parents, it warrants further attention, preferably with a different research methodology that would allow the decision-making process to be explored more fully. This would be particularly useful in regard to more disadvantaged and vulnerable families.

The analyses presented here confirm the expectations of the stakeholders and also the findings reported in the literature, that children missing out on ECE are more often represented among disadvantaged families, and among children who are perhaps in greatest need of ECE in respect of preparing children for school. The groups of children who stood out in these analyses as being less likely to be participating in ECE were Indigenous children and children from NESB families. Children from socio-economically disadvantaged families were also less likely to participate in ECE than those from socio-economically advantaged families. Children living in remote areas had the lowest levels of participation in ECE, compared to those living in major city areas, and some variation was also apparent according to the disadvantage of regions. However, the findings with regard to geographic location were not apparent when the socio-economic status of families was also taken into account.

The patterns of ECE participation for different groups of children appear to vary across the states/territories of Australia, which may reflect the different systems of ECE delivery. In particular, more variation in rates of ECE participation was evident in the eastern states.

These analyses were based on three main datasets - the AEDI, NSPCCC and LSAC. Within each dataset there were some measurement issues, which meant that to analyse ECE participation, the most reliable approach was to consider any participation in preschool (kindergarten) or LDC to be ECE. While each dataset had its own particular set of issues (as documented in Appendix B), undertaking similar analyses with these three sources of information has provided more support to the findings, especially those that consistently arose from all three.

Most of the findings presented here were consistent with expectations, although some suggest that further research may be useful in helping to disentangle how different factors affect family decision-making regarding child participation in ECE. In particular, more research on factors related to family income, employment and parental education levels, and how they intersect with decisions about ECE would help in the understanding of the issues of more vulnerable families.

Section 6 extends the descriptive analyses presented in this section to look more closely at children's participation in the different types of ECE, and how this varies by some of the characteristics already examined here.