4261.6 - Educational outcomes, experimental estimates, Tasmania, 2006-2013  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 28/07/2014  First Issue
   Page tools: Print Print Page Print all pages in this productPrint All

This document was added or updated on 02/10/2015.

AGAINST THE ODDS: FACTORS INFLUENCING CHILD DEVELOPMENT IN TASMANIA - RESULTS OF REGRESSION ANALYSIS

This appendix presents the results of logistic regression analysis to assist with better understanding the extent to which particular personal, family and household characteristics impact on the likelihood of children from low or high socioeconomic groups being classified as developmentally vulnerable in their first year of schooling. The regression models demonstrate the influence of selected variables from the ABS Census of Population and Housing and the National Early Childhood Education and Care Collection on Australian Early Developmental Census (AEDC) measures. The results are further discussed in Against the Odds: Factors influencing child development in Tasmania.

For the purposes of this analysis, low socioeconomic status for a child was defined as being part of a household with a weekly household income between $1 and $599, where no parent was employed and the highest educational qualification of any parent in the household was Year 11 or below. High socioeconomic status for a child was defined as being part of a household with a weekly household income of $2000 or more, where at least one parent was employed and at least one parent had a Bachelor Degree or above.


LOGISTIC REGRESSION ANALYSIS

Logistic regression is a popular and widely used statistical technique for modelling categorical events with two or more outcomes. The event for this analysis was developmental vulnerability and the outcomes were a child being developmentally vulnerable on one or more domains versus not being developmentally vulnerable on any domain.

The results of logistic regression models are generally expressed in terms of the odds of the event. The models in this Appendix measure the effects of different factors on the odds of a child being developmentally vulnerable. The effect of each factor is in relation to a comparator coming from the same variable. For example, the odds of males being developmentally vulnerable compared with females. In the tables below, the significance of a result is indicated by one or more asterisk(*). Where a result does not include an asterisk, the odds ratio is not significant and should be interpreted with caution.

An odds ratio of one means that there is no difference between the factor and the likelihood of the occurrence of an event. An odds ratio of more than one indicates an increased likelihood of being classified as developmentally vulnerable, in relation to the comparator. For example, in Table 1 below, males from lower socioeconomic households were 2.5 times more likely to be developmentally vulnerable on one or more domain than females. In comparison, an odds ratio of less than one indicates a decreased likelihood of being developmentally vulnerable. For example, children from lower socioeconomic households who were read to very regularly were 90% (1 - 0.10 = 0.90) less likely to be developmentally vulnerable on any domain than children read to somewhat regularly.

Whether a child's parent(s) or caregiver(s) were actively engaged with the school in supporting their child's learning was highly correlated with whether the child was read to, so could not be included in the same regression model. Therefore, separate regression models were run substituting parental engagement for the regularity with which a child was read to, with results in Tables 2 and 4 below.

Putting multiple variables into a logistic regression model means the effect of any single factor takes into account the effects of other variables and factors. It is important to note that there are likely other factors that influence early childhood development which were not explored in this analysis, such as students' inherent ability, family cohesion, teacher quality and community support.

Earlier regression models also included tenure type and the number of days a child had been absent from school. Neither of these factors were found to have a significant relationship with developmental vulnerability on the AEDC and were subsequently removed.



TABLE 1. ODDS RATIO OF BEING DEVELOPMENTALLY VULNERABLE ON ONE OR MORE DOMAINS FOR CHILDREN FROM LOWER SOCIOECONOMIC HOUSEHOLDS
CharacteristicsOdds ratio

Whether child was regularly read to, or encouraged in their reading, at home
Not read to regularly5.31***
Read to somewhat regularly (comparison group)-
Read to very regularly0.10***
Family type
Couple family0.75
Lone parent (comparison group)-
Sex
Females (comparison group)-
Males2.53*
Mother's birth age (a)
15 to 19 0.76
20 to 24 (comparison group)-
25 to 290.48
30 to 341.14
Over 351.83
Age started school
Younger than 5.5 years0.53
5.5 – less than 6 years (comparison group)-
6 years or older0.45
Number of children in family
One 1.97
Two (comparison group)-
Three0.83
Four or more0.39
Hours enrolled in preschool per week
1 to 10 hours0.43
11 to 14 hours (comparison group)-
15 hours or more1.34
Indigenous Status
Non-Indigenous (comparison group)-
Aboriginal and Torres Strait Islander0.58
Remoteness
Inner Regional (comparison group)-
Outer Regional, Remote or Very Remote0.68

*** Significantly different effect to the comparison group (p<0.001)
** Significantly different effect to the comparison group (p<0.01)
* Significantly different effect to the comparison group (p<0.05)
(a) Includes natural and adoptive mothers.
Source: Integrated Tasmanian Education and ABS Census Dataset.


TABLE 2. ODDS RATIO OF BEING DEVELOPMENTALLY VULNERABLE ON ONE OR MORE DOMAINS FOR CHILDREN FROM LOWER SOCIOECONOMIC HOUSEHOLDS(a)
CharacteristicOdds ratio

Whether parents were engaged with the school
Parents were not engaged3.70**
Parents were somewhat engaged (comparison group)-
Parents were very engaged0.09***

*** Significantly different effect to the comparison group (p<0.001)
** Significantly different effect to the comparison group (p<0.01)
* Significantly different effect to the comparison group (p<0.05)
(a) This logistic regression model also included all of the variables listed in Table 1, except for 'Whether child was regularly read to, or encouraged in their reading, at home'. However the other results were similar to those presented in Table 1 and are therefore not presented in this table.
Source: Integrated Tasmanian Education and ABS Census Dataset.


TABLE 3. ODDS RATIO OF BEING DEVELOPMENTALLY VULNERABLE ON ONE OR MORE DOMAINS FOR CHILDREN FROM HIGHER SOCIOECONOMIC HOUSEHOLDS(a)
CharacteristicsOdds ratio

Whether child was regularly read to, or encouraged in their reading, at home
Read to not regularly or somewhat regularly (comparison group)-
Read to very regularly0.07***
Sex
Females (comparison group)-
Males1.89*
Mother's birth age (b)
20 to 290.90
30 to 34 (comparison group)-
Over 350.68
Age started school
Younger than 5.5 years1.17
5.5 – less than 6 years (comparison group)-
6 years or older1.75
Number of children in family
One 2.23
Two (comparison group)-
Three0.78
Four or more1.02
Hours enrolled in preschool per week
1 to 10 hours2.63
11 to 14 hours (comparison group)-
15 to 19 hours0.86
20 hours or more1.32
Remoteness
Inner Regional (comparison group)-
Outer Regional, Remote or Very Remote1.60

*** Significantly different effect to the comparison group (p<0.001)
** Significantly different effect to the comparison group (p<0.01)
* Significantly different effect to the comparison group (p<0.05)
(a) Family type and Indigenous status were not included in this model because almost all of the high socioeconomic children were from couple families and non-Indigenous.
(b) Includes natural and adoptive mothers. There were no children in this group whose mother's age was 15-19 years olds at the time of the child's birth.
Source: Integrated Tasmanian Education and ABS Census Dataset.


TABLE 4. ODDS RATIO OF BEING DEVELOPMENTALLY VULNERABLE ON ONE OR MORE DOMAINS FOR CHILDREN FROM HIGHER SOCIOECONOMIC HOUSEHOLDS(a)
CharacteristicOdds ratio

Whether parents were engaged with the school
Parents were not engaged or somewhat engaged (comparison group)-
Parents were very engaged0.11***

*** Significantly different effect to the comparison group (p<0.001)
** Significantly different effect to the comparison group (p<0.01)
* Significantly different effect to the comparison group (p<0.05)
(a) This logistic regression model also included all of the variables listed in Table 3, except for 'Whether child was regularly read to, or encouraged in their reading, at home'. However the other results were similar to those presented in Table 3 and are therefore not presented in this table.
Source: Integrated Tasmanian Education and ABS Census Dataset.