• Summary
• Downloads
• Explanatory Notes
• Related Information
• Past Releases

APPENDIX 5: PREDICTORS FOR BEING OUT OF THE LABOUR FORCE


IDENTIFYING PREDICTORS FOR BEING OUT OF THE LABOUR FORCE

Logit model was used to identify the factors underlying labour force participation. The probability of a person being out of the labour force was modelled as a function of individual and household characteristics.

The data used for the analysis was from the 2014–15 National Aboriginal and Torres Strait Islander Social Survey. The sample comprised Aboriginal and Torres Strait Islander persons aged 15–64 years. Replicate weights were used to maintain the representativeness of the sample.

Average marginal effects are reported instead of coefficients, as they are easy to interpret. The average marginal effect measures the change in the probability of a person being out of the labour force when the value of an independent variable changes by a unit. For example, if the value given for sex is 1 when female and 0 when male, the average marginal effect associated with sex is computed as follows:

1. Treat the person as though they were female, regardless of what the person’s sex actually is. Leave the value of all the other independent variables as they are (see Table 1.1 below for the full list of variables included in the modelling).

2. Compute the probability that this person (if she were female) would be out of the labour force.

3. Repeat steps 1 and 2 but this time as if the person were male.

4. Take the difference between the two predicted probabilities. The difference just computed is the marginal effect for one individual.

5. Repeat steps 1, 2 and 3 for everyone in the sample.

6. Compute the average of all the marginal effects.

Average marginal effects are expressed as a value between 0 and 1. A negative average marginal effect means the probability of being out of the labour force falls as the value of the variable increases. A positive average marginal effect means the probability of not being in the labour force rises as the value of the variable increases. In this example, the positive average marginal effect for sex is 0.149, which means the probability of being out of the labour force is higher for females than males. For the sake of interpretation, the estimated value of the average marginal effect (0.149) should be multiplied by 100 (0.149*100) and expressed in percentage points (15 percentage points).

Given that our model included more than one predictor variable, when we interpret the marginal effect associated with sex, we assume that two randomly chosen individuals in the sample have the same characteristics (as represented by the variables included in the model) except one of them is male and the other is female. Then we say ‘For the same age, social marital status, level of education, long-term health condition, disability status, etc, females are on average 15 percentage points more likely than males to be out of the labour force’. For the sake of brevity, we can also say ‘Females are 15 percentage points more likely than males to be out of the labour force’. The marginal effects associated with the other predictor variables should be interpreted in a similar way.


VARIABLES USED FOR IDENTIFYING PREDICTORS FOR BEING OUT OF THE LABOUR FORCE

The following table indicates which variables were used when estimating average marginal effects.


Table 1.1 Variables used for identifying predictors for not being in the labour force

Variable	Categories
Labour force status	1 0	Not in the labour force (reference) In the labour force
Sex	1 0	Female (reference) Male
Age		Continuous variable taking values in the range of 15 to 64 years
Age squared		Continuous variable
Remoteness	1 0	Lives in remote area (reference) Lives in non-remote area
Social marital status	1 0	Not living in a couple relationship (reference) Living in a couple relationship
Dependent children	1 0	Lives in a family with dependent child/ren (reference) Lives in a family with no dependent child/ren
Household size		Continuous variable taking values in the range of 1 to 10(a)
Main language spoken at home	1 0	An Australian Indigenous language (reference) English
Lives on homelands	1 0	Lives on homelands/traditional country (reference) Does not live on or does not recognise homelands/traditional country
Participation in cultural activities	1 0	Did not participate in selected cultural activities in last 12 months(b) (reference) Participated in selected cultural activities
Involved in cultural events	1 0	Involved in selected events, ceremonies or organisations in last 12 months(c) (reference) Not involved in selected events, ceremonies or organisations
Diagnosed with long-term health condition	1 0	Diagnosed with a long-term health condition (reference) Not diagnosed with a long-term health condition
Disability status	1 0	Has profound/severe/moderate/mild core activity limitation or a schooling/employment restriction (reference) Has no specific limitation or disability
Social contact	1 0	Participated in selected sporting, social or community activities (reference) Did not participate in selected sporting, social or community activities
Carer	1 0	Did not care for a person with disability, long-term health condition or old age in last four weeks (reference) Cared for a person with disability, long-term health condition or old age
Social support	1 0	Able to get support in time of crisis from outside household (reference) Not able to get support
Unfair treatment	1 0	Has not had unfair treatment in last 12 months because Aboriginal and/or Torres Strait Islander (reference) Has had unfair treatment
Transportation	1 0	Can't get to/sometimes or often have difficulty getting to places needed (reference) Can easily get to places needed
Household financial resources	1 2 3	Household members could raise $2,000 in an emergency in a week (reference) Household members could not raise $2,000 in a week Not known if could raise $2,000 in a week
Incarceration	1 0	Has been incarcerated in lifetime (reference) Has never been incarcerated
Education level(d)	1 2 3 4 5	Year 12 (reference) Year 9 or below Year 10 or 11 Advanced diploma/diploma/Certificate III/IV Bachelor degree or higher
Studying	1 0	Not currently studying (reference) Currently studying

RESULTS OF MODELLING

The following tables provide the average marginal effects and percentage point differences for all people aged 15–64 years (Table 1.2), by sex (Table 1.3) and by remoteness (Table 1.4).

Table 1.2 shows the average marginal effect associated with remoteness is not statistically significant. This implies that living in a non-remote or remote area is not significantly associated with labour force participation (based on the characteristics included in the model). From a methodological perspective, it makes little sense to then split the sample into non-remote and remote areas. However, from a policy perspective, we are not only interested in the effect of remoteness by itself. We are also interested in identifying the factors underlying not being in the labour force within non-remote and remote areas, and in whether the importance of those factors varies by remoteness. For this reason, we have split the sample into non-remote and remote and run separate regressions, the results of which are in Table 1.4.


Table 1.2 Predictors for not being in the labour force, Aboriginal and Torres Strait Islander persons aged 15–64 years

	Total persons
Variable	Average marginal effect	Percentage point difference	Standard error
Sex	0.149(a)	14.9(a)	0.021
Age	0.001	0.1	0.001
Remoteness	0.022	2.2	0.025
Social marital status	0.109(a)	10.9(a)	0.022
Dependent children	0.043	4.3	0.028
Household size	0.012	1.2	0.008
Main language spoken at home	0.048	4.8	0.031
Lives on homelands	0.023	2.3	0.021
Participation in cultural activities	0.023	2.3	0.024
Involved in cultural events	-0.026	-2.6	0.026
Diagnosed with long-term health condition	0.040	4.0	0.024
Disability status	0.174(a)	17.4(a)	0.026
Social contact	0.079	7.9	0.063
Carer	0.038	3.8	0.022
Social support	-0.032	-3.2	0.040
Unfair treatment	0.047(a)	4.7(a)	0.021
Transportation	0.070(a)	7.0(a)	0.025
Household financial resources
Could not raise $2,000 in a week	0.124(a)	12.4(a)	0.025
Unsure if could raise $2,000 in a week	0.113(a)	11.3(a)	0.051
Incarceration	0.157(a)	15.7(a)	0.037
Education level(b)
Year 9 or below	0.138(a)	13.8(a)	0.038
Year 10 or 11	0.081(a)	8.1(a)	0.038
Advanced diploma/diploma/Certificate III/IV	-0.071(a)	-7.1(a)	0.035
Bachelor degree or higher	-0.181(a)	-18.1(a)	0.050
Studying	0.017	1.7	0.031
Number of observations		4 620

Table 1.3 Predictors for not being in the labour force by sex, Aboriginal and Torres Strait Islander persons aged 15–64 years

	Males				Females
Variable	Average marginal effect	Percentage point difference	Standard error		Average marginal effect	Percentage point difference	Standard error
Age	0.002	0.2	0.002		0.001	0.1	0.001
Remoteness	0.065	6.5	0.035		-0.022	-2.2	0.032
Social marital status	0.168(a)	16.8(a)	0.038		0.050	5.0	0.026
Dependent children	-0.005	-0.5	0.041		0.105(a)	10.5(a)	0.038
Household size	0.013	1.3	0.014		0.012	1.2	0.009
Main language spoken at home	0.034	3.4	0.048		0.042	4.2	0.040
Lives on homelands	-0.005	-0.5	0.032		0.030	3.0	0.029
Participation in cultural activities	0.041	4.1	0.036		0.005	0.5	0.030
Involved in cultural events	-0.011	-1.1	0.041		-0.031	-3.1	0.033
Diagnosed with long-term health condition	0.038	3.8	0.034		0.038	3.8	0.028
Disability status	0.210(a)	21.0(a)	0.034		0.125(a)	12.5(a)	0.034
Social contact	0.140	14.0	0.098		-0.060	-6.0	0.095
Carer	0.020	2.0	0.034		0.043	4.3	0.030
Social support	0.003	0.3	0.053		-0.070	-7.0	0.059
Unfair treatment	0.047	4.7	0.028		0.041	4.1	0.029
Transportation	0.074(a)	7.4(a)	0.035		0.059	5.9	0.037
Household financial resources
Could not raise $2,000 in a week	0.094(a)	9.4(a)	0.031		0.144(a)	14.4(a)	0.031
Unsure if could raise $2,000 in a week	0.112	11.2	0.058		0.087	8.7	0.066
Incarceration	0.110(a)	11.0(a)	0.038		0.306(a)	30.6(a)	0.073
Education level(b)
Year 9 or below	0.057	5.7	0.055		0.177(a)	17.7(a)	0.050
Year 10 or 11	0.035	3.5	0.054		0.100(a)	10.0(a)	0.048
Advanced diploma/diploma/Certificate III/IV	-0.075	-7.5	0.052		-0.073	-7.3	0.046
Bachelor degree or higher	-0.213(a)	-21.3(a)	0.090		-0.217(a)	-21.7(a)	0.078
Studying	-0.089	-8.9	0.048		0.080(a)	8.0(a)	0.040
Number of observations	1 968				2 652

	Non-remote				Remote
Variable	Average marginal effect	Percentage point difference	Standard error		Average marginal effect	Percentage point difference	Standard error
Sex	0.145(a)	14.5(a)	0.025		0.156(a)	15.6(a)	0.033
Age	0.002	0.2	0.001		-0.004	-0.4	0.002
Social marital status	0.127(a)	12.7(a)	0.028		0.048	4.8	0.030
Dependent children	0.050	5.0	0.038		-0.014	-1.4	0.034
Household size	0.016	1.6	0.013		0.007	0.7	0.008
Main language spoken at home	-0.005	-0.5	0.066		0.092(a)	9.2(a)	0.035
Lives on homelands	0.038	3.8	0.027		-0.029	-2.9	0.034
Participation in cultural activities	0.012	1.2	0.028		0.081(a)	8.1(a)	0.034
Involved in cultural events	-0.023	-2.3	0.029		-0.031	-3.1	0.043
Diagnosed with long-term health condition	0.046	4.6	0.030		0.023	2.3	0.034
Disability status	0.163(a)	16.3(a)	0.027		0.148(a)	14.8(a)	0.039
Social contact	0.112	11.2	0.079		-0.087	-8.7	0.114
Carer	0.022	2.2	0.027		0.082(a)	8.2(a)	0.035
Social support	-0.044	-4.4	0.049		0.014	1.4	0.053
Unfair treatment	0.053(a)	5.3(a)	0.025		0.014	1.4	0.034
Transportation	0.076(a)	7.6(a)	0.031		0.056	5.6	0.037
Household financial resources
Could not raise $2,000 in a week	0.117(a)	11.7(a)	0.029		0.129(a)	12.9(a)	0.035
Unsure if could raise $2,000 in a week	0.103	10.3	0.056		0.112	11.2	0.080
Incarceration	0.152(a)	15.2(a)	0.047		0.126(a)	12.6(a)	0.046
Education level(b)
Year 9 or below	0.094(a)	9.4(a)	0.045		0.213(a)	21.3(a)	0.051
Year 10 or 11	0.062	6.2	0.045		0.109(a)	10.9(a)	0.051
Advanced diploma/diploma/Certificate III/IV	-0.075	-7.5	0.044		-0.069	-6.9	0.053
Bachelor degree or higher	-0.226(a)	-22.6(a)	0.072		-0.130	-13.0	0.137
Studying	0.020	2.0	0.038		-0.006	-0.6	0.048
Number of observations			2 739				1 881