Maternal work hours and Australian childrens mental health and - - PDF document

SLIDE 1

1

Maternal work hours and Australian children’s mental health and behaviour: exploring differences by maternal education

Meg Kingsley and Ann Evans, School of Demography, Australian National University

Abstract

Evidence suggests parents’ work is a social determinant of children’s socio-emotional development. While this influence can be positive, recent trends in the Australian labour market, such as the destandardisation of work hours, are incompatible with strategies to reconcile work and family. Evidence suggests both long and short work hours may affect children’s wellbeing; however previous research is inconsistent and may indicate the relationship depends on family characteristics. This study examines the relationship between mothers’ work hours and children’s mental health and behaviour over ages 4-15, and examines the moderating effect of maternal education. It uses Waves 1-6 of data from the Kindergarten cohort in the Longitudinal Study of Australian Children and employs mixed-effects multi-level growth curve modelling. The results indicate that younger children of mothers working between two and three days per week have lower levels of socio-emotional and behavioural problems, however by adolescence, children of mothers working full time have lower levels of problems. However, the relationship between mothers’ work hours and children’s socio- emotional difficulties differs depending on the educational attainment of the mother, suggesting family background may moderate the links. The findings imply longer term impacts of changes in education and the deregulation of the Australian labour market.

Notes

This research is supported by an Australian Government Research Training Program (RTP) Scholarship. This paper uses unit record data from Growing Up in Australia, the Longitudinal Study of Australian

• Children. The study is conducted in partnership between the Department of Social Services (DSS), the

Australian Institute of Family Studies (AIFS) and the Australian Bureau of Statistics (ABS). The findings and views reported in this paper are those of the author and should not be attributed to DSS, AIFS or the ABS.

SLIDE 2

2

Introduction

Understanding the factors that influence childhood mental health and behaviour is important as it impacts adult health and wellbeing, as well as the wellbeing of children. Emotional and behavioural problems in childhood are related to psychopathology in adulthood (Repetti 2002) and non-cognitive traits have been shown to influence both social and economic success later in life (Heckman et al. 2006). Research suggests children’s emotional and behavioural development is sensitive to family stress (Khanam & Nghiem 2016), and a factor that may influence family stress and children’s mental health and behaviour, is parental work

• conditions. For parents, family and work are interconnected domains, and parental work can shape family

processes and child wellbeing. Increasing the workforce participation of mothers of young children is a current aim of both welfare and gender equality policies in Australia, and parental employment can have a positive influence on children’s development. However, the policy and cultural context of Australia’s work-care regime has been shaped by the male breadwinner/female caregiver dichotomy (Mahon et al. 2016) and viewing care and work family reconciliation as individual choices (Pocock et al. 2013; The Work + Family Policy Roundtable 2016). As a result, Australia’s family policy environment encourages female employment and parenthood to be reconciled using a mix of part-time work, periods out of the workforce with re-entry in a different job, and private sector childcare (Rendall et al. 2013), resulting in inequalities in the labour market and the experience of the work-family interface. At the same time that some Australians are working long hours, the prevalence of part-time work and underemployment is increasing (Australian Bureau of Statistics 2016a). Destandardisation and individualisation of work hours has been accompanied by growing insecurity regarding paid work, unsocial work hours and casualisation for some workers (Beck 2000; Germov 2011). As noted by the OECD (2012), this instability in the labour market is at odds with long-term strategies to reconcile work and family, and might be detrimental to children’s wellbeing. Evidence suggests that short or long work hours are associated with difficulties managing work and family life and may be detrimental to the health and wellbeing of both parents and children. However, previous research on the relationship between parents’ work hours and children’s mental health and behaviour is inconsistent. Emerging theories suggest that these inconsistencies highlight the role that family context has in the relationship between work and family. Li et al (2012) suggest that the link between parental work conditions and children’s wellbeing depends on a number of contextual factors that influence familial resources; however it is not currently well understood how the impact of parental work arrangements on children varies between different families. Socio-demographic characteristics, such as parental educational attainment, are associated with different employment conditions, so some parents may be more likely to experience difficulties managing the work-family interface. In addition, the advantages and disadvantages of job characteristics are not evenly distributed across groups (Presser & Ward 2011). This paper extends previous Australian research by using longitudinal data to explore the moderating effect of maternal education levels

• n the relationship between mothers’ work hours and children’s socio-emotional wellbeing.

SLIDE 3

3

Theoretical Orientation

This study builds on previous research on the relationship between parents’ work hours and children’s wellbeing that has been guided by Bronfenbrenner’s theory of child development (Bronfenbrenner & Crouter 1982; Bronfenbrenner & Morris 2006; Parke 2004). This theory positions parents’ work as an important part

• f the exosystem in which children grow and develop because it can influence family resources, processes,

time, and emotional exchanges (Li et al. 2012). Theories emerging from the USA (Williams 2013, Williams & Boushey 2010) suggest that socio-demographic groups in society may experience difficulties managing the work-family interface differently, and the effect of working hours on parents and children may be context-dependent. Williams argues the risk that long work hours will trigger work-family problems depends

• n the background characteristics of workers. Longer working hours are concentrated amongst people of

higher socio-economic position (McGinnity & Calvert 2009, OECD 2012, Strazdins et al 2011) and Williams writes that for higher educated people, working long hours create work-family problems. Longer working hours of mothers may influence poor outcomes because they contribute to role

• strain. According to role strain theory, people fulfil different roles in their lives, such as parent and

employee, but strain occurs when the duties, expectations, norms and behaviours associated with the different roles conflict. This theory asserts demands and resources associated with participation in the work

• r family domain directly affect role quality and performance in the other domain (Voydanoff 2005). Role

strain associated with spending long hours in the workplace may result in psychological overload and lower levels of happiness for parents (Aryee 1992; Cooklin et al. 2016; Frone et al. 1992; Greenhaus & Beutell 1985; Higgins and Duxbury 1992; Lautsch & Scully 2007), which then impacts on children’s socio-emotional wellbeing. Lautsch and Scully (2007) argue while reducing work hours is posited as a solution to work-family balance issues in middle-class families (e.g. OECD 2012), shorter working hours can create stress in disadvantaged families because income is not high enough to provide for children. Williams (Williams et al 2013; Williams & Boushey 2010) suggests that for lower educated workers, around-the-clock availability for work and schedule inflexibility create tension. In addition, underemployment may be a source of difficulty for lower skilled workers. Shorter work hours may be related to children’s development through the relationship between underemployment, income and family stress. Family stress theory argues that income affects parenting because financial hardship affects’ psychological wellbeing, which shapes their parenting style and their child’s development (Smith & Brooks-Gunn 1997; Yamauchi 2010; Yeung et al. 2002). Recent research suggests that family stress theory is relevant to children’s socio-emotional and behavioural

• utcomes (Khanam & Nghiem 2016). However, part-time work may also be related to children’s wellbeing

through the psychological distress associated with poorer quality work, or an elevated burden of responsibility for child care and housework, leading to role strain as described above.

SLIDE 4

4

Related Evidence

Evidence suggests that longer work hours for parents may be linked to poorer socio-emotional and behavioural outcomes for children. For example, Strazdins et al (2011) found that mothers’ work hours were significantly related to the socio-emotional wellbeing of Australian preschool aged children, such that mothers in full-time employment reported having children with more emotional and behavioural difficulties. Supporting this finding, Hadzic et al (2013) found more behavioural problems in children when mothers worked full-time or were not in paid employment relative to those employed part-time. However, other researchers have not found an effect of maternal work hours on children’s wellbeing. For example, Cooksey et al (1997) and Johnson et al (2013) did not find an effect of mothers’ work hours on children’s emotional or behavioural problems. Baxter et al (2012) investigated the relationship between short part-time hours at the family level and children’s socio-emotional outcomes between ages 4-5 and 10-11. They found children in short part-time hours families had poorer socio–emotional outcomes than children in families with long part-time or full-time hours of employment. However, Baxter et al found the association between short parental work hours and poorer child outcomes was no longer statistically significant once measures of financial wellbeing were included in the regression models, suggesting family context may moderate the links between short work hours and children’s wellbeing. Part-time work may not only be related to poorer outcomes for some workers because of an association with lower incomes. Underemployment has been found to be related to depression, independent

• f other socio-demographic variables (Dooley et al 2000). Additionally, female part-time workers have been

found to experience high levels of stress and time pressure, perhaps due to higher levels of unpaid work (Pocock et al 2007, Rose et al 2013). Baxter and Chesters (2011) found no difference in perceptions of work- family balance between women employed part-time and those employed full-time. The contrasting evidence may suggest that other factors affect whether mothers’ work hours are related to children’s development. The

• utcomes work hours may depend on the socio-economic status of workers and the voluntariness of the

arrangement (Lautsch & Scully 2007). The conditions associated with longer or shorter work hours may influence outcomes. Part-time workers are more likely to work in poor quality jobs than full-time employees: they are less likely to have paid leave entitlements, a say over their start and finish times, guaranteed minimum hours of work, regular weekly hours or be able to choose when holidays are taken (Australian Bureau of Statistics 2013; Charlesworth et al 2011), which may create stress for parents and affect children. In Australia, parents in less advantaged households have been found to have poorer quality jobs compared with those in more advantaged households. Australian research found that 4-5 year old children of parents with poor quality jobs showed more emotional and behavioural difficulties (Nicholson et al 2012; Strazdins et al 2010). In addition, the quality of part-time work may not be evenly distributed in the population, with lower educated workers in part-time employment potentially in poorer quality jobs than higher educated workers in part-time

• employment. These associations have implications for multigenerational transmission of disadvantage.

SLIDE 5

5

The Present Study

Based on previous evidence, the influence of maternal work hours on children’s socio-emotional outcomes is unclear and the relationship may depend on socio-demographic factors of the parent. This study investigates the relationship between mothers’ work hours and children’s mental health and behaviour as they age and explores differences by maternal education. The study addresses the following research question: How does the education level of mothers affect the relationship between work hours and children’s mental health and behaviour as children age?

Method

Data

The data source for the analyses is Waves 1-6 of Growing Up in Australia: the Longitudinal Study

• f Australian Children (LSAC). LSAC is a large scale, nationally representative Australian study of

children and families that follows the experiences and wellbeing of two cohorts of children and their families, from infancy to adolescence (Australian Institute of Family Studies 2015). LSAC

• btains the perspectives of mothers and fathers, and collects information on a broad range of

influences on child and family wellbeing. The children in LSAC were aged 0–1 years (the B cohort) and 4–5 years (the K cohort) at the first wave of the study in 2004. Around 5,000 children in each cohort participated in Wave 1. The families are visited once every two years when they are interviewed and direct observations and assessments are conducted (Australian Institute of Family Studies 2015). The analyses in this study are based on data from the K-cohort (n=4,893 in Wave 1), who were aged 14-15 in Wave 6 (n=3,537). Thus, the analyses utilise data for 10 years of the child’s life, from preschool age to mid-adolescence. This study is based on an unbalanced panel. There are 24,444 child-year observations for the outcome measure. Mothers’ employment status is missing for 1 per cent of these observations. After accounting for sample loss due to missing data

• n the control variables, the primary analytic sample is 23,676 child-year observations.

Measures

Children’s Socio-emotional and Behavioural Problems

Children’s socio-emotional and behavioural outcomes were measured at each wave of LSAC using the Strengths and Difficulties Questionnaire (SDQ). The SDQ can be completed by parents and teachers of children aged 4–16 and by youth aged 11–16 (Goodman et al. 2010). These analyses utilise the SDQ data provided by the parent who knows the child best (usually the mother), as there are higher rates of missing data for the teacher reports. Total scores can range from zero to 40, with higher scores indicating more

• problems. The mean total score in the sample is 8.0. Children’s SDQ scores are treated as continuous

variables in the analyses. Following exploration of the relationship between children’s SDQ scores and age between the ages of 4 and 15, a cubic slope for age is allowed for.

SLIDE 6

6

Children’s Age

A key aim of this study is examining effects on children as they age. Child’s age in years is included as a continuous variable in the models. Child’s age is centred such that the initial status refers to age 4, allowing an interpretable intercept within the range of data collection. Child’s age is interacted with mothers work hours to test whether the effect of works hours changed over time.

Mothers’ Employment Status and Work Hours

Mothers’ usual weekly work hours were assessed at each wave of LSAC. They are treated as a categorical variable in order to include parents who were unemployed or not in the labour force at the time of the survey. The categories are not in the labour force (NILF), unemployed (UE), 0-14 hours/week, 15-24 hours/week, 25-34 hours/week, 35-44 hours/week, and 45 or more hours per week. These categories enable examination

• f the effects of different levels of part-time work, given the majority of mothers in the sample work

part-time.

Mothers’ Education Level

Given the relationship between socio-economic status and work hours, and socio-economic status and children’s outcomes, mothers’ highest level of educational attainment at the time of the survey is included as a categorical variable. The categories were: Did not complete year 12, Completed year 12, Certificate or Diploma, and Bachelor Degree or higher.

Covariates

The models control for child and family variables, as well as mothers’ characteristics. A small proportion of children have a single mother in the home (13 per cent in Wave 1, increasing to 16 per cent by Wave 6). While the models control for whether a partner is present, and if so, his work hours, other characteristics of fathers are not included so that children of single mothers can be included in the models. Other work variables are not included in the models so that the analyses can include those parents who were not working. Based on previous literature, a number of control factors that may affect children’s wellbeing are included. The factors included in the models in this study are those that most substantially improved the amount of variance explained relative to a simple OLS linear regression model that included only mothers’ work hours. The control variables are described below, with descriptive statistics presented in the results section. Continuous covariates are centred at their grand mean, so that the reference child represents a realistic

• scenario. The base category for categorical variables was determined by considering the most common

category in the data set.

Child variables. Child’s gender is included in the models and the reference category is boys.

Children’s health status may affect their mental health and behaviour, as well parents’ level of employment and mental health; therefore, whether the child has any medical conditions or disabilities that had lasted or are likely to last for six months or more was assessed at each wave and included as a categorical variable. Whether the child was breastfed until at least six months of age is included as a time-invariant categorical variable, given evidence breastfeeding is associated with later childhood socio-emotional and behavioural

• utcomes (e.g. Lind et al. 2014, Rochat et al. 2016).

SLIDE 7

7

Family variables. Whether there is a non-biological parent living in the household is included as a

categorical variable at each wave. For single parent families where the parent is biologically related to the child, the answer is coded to ‘No’. There were 42 children who lived with a single mother who they are not biologically related to. The number of resident children in the household at each wave is included as a categorical variable (1 child, 2 children, 3 or more children). Whether there is an infant (child aged less than

• ne year old) in the household at each wave included as a categorical variable. The number of children, or an

infant, in the household may affect parents’ work hours, as well as the behaviour of the children in the

• household. Equivalised household income is included as a continuous variable to control for the relationship

between work hours and income. This is calculated by combining the income of both parents in two-parent households, or including only the mother’s income in single-parent households, and adjusting for the number

• f people in the household (Strazdins et al. 2010). Income at each wave is adjusted to the June 2014

Consumer Price Index (CPI) (Australian Bureau of Statistics 2014). This variable is included in its logged form in the analyses.

Mother variables. Mothers’ age is included as a continuous variable at each wave. Whether the

mother has a partner and, if so, the partner’s usual weekly work hours is included in the model as a categorical variable at each wave. The categories for partner’s employment are: no partner, partner not working, partner working part-time (1-34 hours/week), partner working full-time (35-49 hours/week) and partner working longer hours (over 50 hours per week). Whether the mother has any medical conditions or disabilities that had lasted or were likely to last for six months or more was assessed at each wave and included as a categorical variable.

Empirical Approach

As a panel survey, in the LSAC dataset there are two basic levels: each child, and each occasion at which they are assessed. Whilst some researchers pool the observations across children and waves for analysis, this fails to account for the dependence or correlation of units that belong to the same cluster. Multi-level modelling accounts for observations that are clustered by child, and therefore likely to have correlated errors. Growth curve models, a type of multilevel model, explicitly model the shape of trajectories of individual

• ver time and how these trajectories vary by covariates, and randomly (Rabe-Hesketh & Skrondal 2012).

Theoretically, socio-emotional and behavioural trajectories are likely to vary across children. Growth curve models are also flexible and efficient for unbalanced and variably spaced longitudinal data1, and predictors can be time-invariant or time-varying (Singer & Willett 2003). For these reasons, estimation of the association between maternal work hours and children’s socio-emotional and behavioural trajectories is conducted using mixed effects multi-level growth curve modelling.

1 Children are not all the same age at each wave (e.g. some children were just over 4 at Wave 1, while others were

nearly 6) and the time between waves varies for each child.

SLIDE 8

8 Analyses are estimated with Level 1 as age (i.e. within individual effects) and Level 2 as children (i.e. between individual effects). The general growth curve model, for the repeatedly measured variable 𝑧𝑢𝑗

• f individual i at occasion t, may be expressed as follows:

𝑧𝑢𝑗 = (𝛾1 + 𝜂1𝑗 ) + (𝛾2 + 𝜂2𝑗 )𝑌𝑢𝑗 + 𝜗𝑢𝑗 where 𝛾1 is the mean intercept; 𝛾2 is the mean slope; 𝜂1𝑗 is a random cluster intercept for child i, the deviation of child i's intercept from the mean intercept 𝛾1; 𝜂2𝑗 is a random slope for child i, the deviation of child i's slope from the mean slope 𝛾2; 𝑌𝑢𝑗 is a level 1 predictor for occasion t for child i; and 𝜗𝑢𝑗 is the individual error term. Whether parental work hours are related to children’s socio-emotional and behavioural trajectories was evaluated using a sequence of statistical models. First, examination of the unconditional means model indicated that around two-thirds (64 per cent) of the variance in SDQ scores is due to differences across children, with the remaining proportion attributable to differences within children themselves. Second, examination of the unconditional growth models indicate that the difference in within-child variation in SDQ scores attributable to age is 21 per cent. Next, examination of the extent to which within-child variation is associated with mothers’ work hours, after accounting for age, is conducted by adding maternal work hours to the model. After accounting for age, the difference in within-child variation in the outcome measure attributable to mothers’ work hours is 2 per cent. Given the research interest in understanding the effect of hours on children as they age, an interaction between children’s age and parents’ hours is included in the base model. Then child, family and mother factors are added to the model to test whether any effects of work hours remains once covariates are accounted for. In a further step, an interaction between mothers’ work hours and education level is tested. In the third model, a three-way interaction between mothers’ work hours, education level and children’s age examines whether the effect of work hours on children’s trajectories differs depending on the education level of the mother. For clarity, only the results of the last three models are presented in the Results section. The models allow for a random intercept and a random slope for age. These effects allow for between-child variability in initial levels of children’s difficulties and in the way they change as they age. Only the lower-order (linear) term for age used in the fixed part of the model is allowed to vary randomly between children. This is deemed to be a reasonable approach in the literature (Rabe-Hesketh & Skrondal 2012). The significance of the random effects in each model are systematically examined by comparing nested models and conducting likelihood-ratio tests in STATA (Hamilton 2013). To evaluate model fit, three goodness of fit indices are used: the deviance statistic (-2 log-likelihood), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). Smaller numbers on all three measures indicate a better model fit.

SLIDE 9

9

Results

Descriptive Analyses

The distribution of the sample by the outcome, predictor and confounding variables is summarised in Table

• 1. The table suggests a cubic relationship between children’s age and their socio-emotional and behavioural

problems, and further exploration confirms this relationship. Between Waves 1 and 3 average SDQ scores decrease, while they rose in Wave 4, before declining again in early adolescence. The family variables demonstrate changes in family structure that occur as the study child ages, such as an increasing proportion children with a non-biological parents in the household, and increasing proportion of children living in households with more children. More mothers enter the workforce and increase their weekly work hours over the waves of the

• survey. In Wave 1 the most common category is NILF, followed by 15-24 hours per week. By Wave 6, the

most common category is 35-44 hours per week. The proportion of mothers who are single parents increased slightly between Wave 1 and Wave 6. Reflecting the gendered distribution of work hours in Australia (Charlesworth et al. 2011), the majority of mothers’ partners’ work full-time hours, followed by longer hours, and this is consistent across waves. Maternal educational attainment improved over the waves of the

• survey. While this indicates mothers completed higher qualifications over time, it may also reflect attrition

from the study of lower education mothers. The proportion of children who were breastfed to six months may also reflect this.

Table 1. Distribution of sample variables by wave for children with complete outcome, work hours and socio-demographic variables (23,676 observations) Wave 1 n=4843 2 n=4268 3 n=3731 4 n=3884 5 n=3709 6 n=3243 Child variables SDQ score (mean) 9.35 7.90 7.50 7.82 7.36 7.01 Age (mean in years) 4.7 6.8 8.8 10.8 12.9 14.9 Gender (% boy) 51.0 51.2 50.1 51.1 51.1 50.3 Medical condition (% yes) 20.4 14.1 8.3 8.6 4.5 4.9 Breastfed at 6 months (% yes) 57.3 58.2 59.6 60.0 60.2 61.8 Family variables Presence of non-biological parent (% yes) 3.1 3.8 6.6 7.5 8.6 9.3 Infant in household (% yes) 9.1 5.6 3.4 2.1 1.4 0.7 Number of children (%) 1 child 11.1 8.7 8.3 8.0 9.2 9.0 2 children 48.8 45.9 45.5 44.9 44.8 49.9 3 or more children 40.1 45.4 46.3 47.1 46.0 41.0 Equivalised weekly household income (CPI adjusted, mean in $) 842 946 1046 1100 1141 1216 Mother variables Weekly work hours (%) 13.9 16.9 20.1 21.7 24.01 25.7 NILF 38.6 29.5 22.7 20.8 17.6 15.6 Unemployed 3.7 3.1 2.3 2.5 2.7 2.3 0-14 15.7 16.2 13.9 11.0 8.7 8.6 15-24 17.5 20.0 22.2 22.0 20.9 19.0 25-34 8.9 11.8 15.3 17.5 18.3 18.5 35-44 10.5 13.4 16.3 18.1 21.3 23.3 45+ 5.0 6.1 7.3 8.1 10.4 12.8

SLIDE 10

10

Table 1. Distribution of sample variables by wave for children with complete outcome, work hours and socio-demographic variables (23,676 observations) Wave 1 n=4843 2 n=4268 3 n=3731 4 n=3884 5 n=3709 6 n=3243 Age (mean) 34.6 36.9 39.0 41.1 43.2 45.3 Partner’s employment status (%) No partner 13.4 14.1 13.5 14.5 15.7 16.2 Partner not working 5.5 4.7 3.9 5.3 5.3 5.6 Partner working PT hours (1-34) 5.6 5.7 4.7 5.6 5.8 5.9 Partner working full-time (35-49) 41.5 42.2 43.1 44.6 43.3 44.4 Partner working longer hours (50+) 34.0 33.4 34.8 29.9 29.9 27.9 Educational attainment (%) Did not complete Year 12 21.3 18.4 15.8 14.6 12.5 10.8 Completed Year 12 15.0 13.2 11.9 10.8 9.8 8.9 Certificate or Diploma 35.2 38.6 40.8 41.5 42.7 45.1 Bachelor degree or higher 28.6 29.8 31.6 33.2 35.1 35.2 Medical condition (% yes) 26.3 13.0 8.8 8.7 11.8 13.5 Source: Author’s calculations based on LSAC Wave 6 data release.

Growth Curve Analyses

Table 2 shows selected results of the growth curve models (full model results are provided in the appendix). The main effect of children’s age indicates that problems decrease as children age, however, the

higher order terms are significant, indicating there is a cubic relationship between socio-emotional problems and age, as suggested by the descriptive results.

Table 2. Selected results of growth-curve models for children’s socio-emotional and behavioural problems from ages 4 to 15 Model 1 Model 2 Model 3 Fixed effects  SE  SE  SE Intercept 10.45* 0.12 9.74* 0.17 9.61* 0.20 Children’s age

• 1.08*

0.06

• 0.93*

0.06

• 0.88*

0.06 Children’s age squared 0.17* 0.01 0.16* 0.01 0.16* 0.01 Children’s age cubed

• 0.01*

0.00

• 0.01*

0.00

• 0.01*

0.00 Mothers’ work hours NILF (ref.)

• Unemployed
• 0.38

0.26

• 0.53*

0.26

• 0.62

0.39 0-14

• 0.50*

0.15

• 0.33*

0.14

• 0.28

0.23 15-24

• 0.64*

0.14

• 0.41*

0.14

• 0.21

0.22 25-34

• 0.51*

0.17

• 0.29+

0.17

• 0.18

0.27 35-44

• 0.42*

0.16

• 0.17

0.17

• 0.15

0.26 45+

• 0.63*

0.22

• 0.30

0.22

• 0.63

0.37 Children’s age × work hours NILF (ref.)

• Unemployed

0.00 0.04 0.02 0.04 0.04 0.06 0-14 0.01 0.03 0.00 0.03

• 0.02

0.04 15-24 0.01 0.02 0.00 0.02

• 0.05

0.04 25-34

• 0.01

0.03

• 0.02

0.03

• 0.05

0.04 35-44

• 0.02

0.02

• 0.03

0.02

• 0.06

0.04 45+ 0.02 0.03 0.00 0.03 0.01 0.05 Mothers' educational attainment Less than Year 12 0.57* 0.13 1.13* 0.23 Completed Year 12

• 0.20

0.15 0.18 0.27 Certificate or Diploma (ref.)

• Bachelor degree or higher
• 0.64*

0.12

• 0.99*

0.26 Mothers’ education level × work hours Did not complete year 12 NILF (ref.)

• Unemployed

0.26 0.65

SLIDE 11

11

Table 2. Selected results of growth-curve models for children’s socio-emotional and behavioural problems from ages 4 to 15 Model 1 Model 2 Model 3 0-14

• 0.35

0.40 15-24

• 1.05*

0.40 25-34

• 1.09*

0.49 35-44

• 0.85+

0.48 45+

• 0.25

0.72 Completed Year 12 NILF (ref.)

• Unemployed

0.21 0.80 0-14 0.35 0.45 15-24

• 0.14

0.43 25-34

• 0.47

0.51 35-44 0.19 0.49 45+ 0.84 0.76 Certificate or Diploma (ref.)

• Bachelor Degree or higher

NILF (ref.)

• Unemployed

0.19 0.78 0-14 0.13 0.36 15-24 0.23 0.34 25-34 0.66+ 0.39 35-44 0.60 0.39 45+ 1.08* 0.50 Mothers’ education level × work hours × children’s age Did not complete year 12 NILF

• 0.09*

0.04 Unemployed

• 0.20+

0.12 0-14

• 0.04

0.07 15-24 0.05 0.05 25-34 0.03 0.07 35-44

• 0.03

0.06 45+

• 0.10

0.10 Completed Year 12 NILF

• 0.18*

0.05 Unemployed

• 0.18

0.14 0-14

• 0.12+

0.07 15-24

• 0.04

0.06 25-34

• 0.05

0.07 35-44

• 0.08

0.06 45+

• 0.20+

0.10 Certificate or Diploma (ref.)

• Bachelor Degree or higher

NILF 0.00 0.04 Unemployed

• 0.05

0.12 0-14 0.01 0.05 15-24 0.01 0.04 25-34

• 0.03

0.04 35-44

• 0.01

0.04 45+

• 0.05

0.06 Socio-demographic & health controls No Yes Yes Variance components  SE  SE  SE Level 1 error (within individual) 2.89 0.02 2.88 0.02 2.87 0.02 Level 2 (between individual) Initial status 4.46* 0.06 4.05* 0.06 4.04* 0.06 Rate of change 0.38* 0.01 0.38* 0.01 0.38* 0.01 Model fit statistics Deviance 134672.96 131064.47 131000.11 AIC 134713.00 131138.50 131152.10 BIC 134874.80 131437.10 131765.50 Source: Author’s calculations based on LSAC Wave 6 data release. Notes: *p<0.05, +p<0.10. Results based on mixed-effects multi-level linear regression in STATA 12 SE; random effects vary by child with unstructured residuals. Continuous variables are centred.

SLIDE 12

12 In Model 1, relative to mothers outside the labour force, any level of maternal employment is related to significantly lower levels of problems for children. In particular, working 15-24 hours per week, or 45 or more hours per week, is related to the lowest level of difficulties relative to being outside the labour force. While there is evidence to suggest that maternal work hours were related to the initial level of children’s difficulties, the relationship between work hours and the rate of change is not significant, suggesting that mothers’ work hours is not significantly related to the trajectory in problems over time. In Model 2, the effect sizes of the main effect are attenuated once the covariates are added and the three longest working hours categories are no longer significantly associated with fewer problems relative to being outside the labour force. This suggests that background characteristics may explain part of the relationship between maternal work hours, especially long work hours, and children’s socio-emotional behavioural problems. For example, household income and maternal education are significant predictors of children’s level of socio-emotional and behavioural problems. Longer working hours are concentrated amongst people of higher socio-economic position (McGinnity & Calvert 2009, OECD 2012, Strazdins et al. 2011), and the association between the longest work hours category and better outcomes for children may be confounded by parents’ educational attainment or income levels. Once background characteristics are accounted for, of the employed categories, working 15-24 hours per week is associated with the lowest level

• f initial problems relative to children of mothers who are outside the labour force. Working 35-44 hours per

week is associated with the smallest reduction in problems relative to being outside the labour force. In Model 2, mothers’ unemployment is associated with the lowest level of problems, however, caution should be used when interpreting this result due to small cell sizes for this group (see Table 1). To test whether maternal education levels affect the relationship between work hours and children, an interaction between work hours and education levels is tested for both the main effect and the rate of

• change. Once the interaction between work hours and education level is accounted for, the main effect of

work hours is no longer significant for any category. It is also interesting to note that the main effect of education is strengthened once the relationship between work hours and education is accounted for. With respect to the interaction, some combinations of work hours and education levels are significantly different from the base categories. In particular, the results indicate that work hours are differently related to the level

• f children’s problems in both the lowest and highest levels of educational attainment relative to Certificate
• r Diploma level qualifications. In addition to differences by maternal education in the effect of mothers’

work hours on the level of children’s problems, Model 3 also indicates that the trajectories of change are different. Figure 2 displays the predicted levels and change in problems for children of mothers with different combinations of educational attainment and work hours who were otherwise average.

SLIDE 13

13

Figure 1. Predicted SDQ scores by mothers’ work hours, mothers’ education level and children’s age Source: Author’s calculations based on LSAC Wave 6 data release. Note: Chart display results holding all other variables at their means.

SLIDE 14

14 For children of mothers with less than Year 12, being outside the labour force is associated with higher difficulties in children of any age compared with working long hours. In contrast, for those with Bachelor degrees or higher, the results indicate that until early adolescence, children of mothers with a Bachelor degree or higher have more problems if their mother works longer hours. With respect to part-time work, for children of mothers who had not completed Year 12, very short hour jobs are associated with higher levels of difficulties compared with the other employed groups. However, for those with Bachelor degrees or higher, part-time hours are associated with lower levels of difficulties in children.

Variance Components and Model Fit

The variance components are presented in the middle section of Table 2 as standard deviations and can be interpreted on the same scale as the fixed effects. The random effect for the intercept indicates that initial levels of problems are diverse in the sample. The random effect of children’s age suggests there is some variability in the trajectories of problems between children. Adding the socio-demographic controls explains a small proportion of the variance in children’s initial status, but does not explain variability in the effect of children’s age. Adding the interaction between maternal education and work hours only explains a small proportion in variability in children’s initial status.

Discussion

This study extends previous Australian research by examining the relationship between mothers’ work hours and children’s mental health and behaviour over a 10 year period, and examining the role of maternal education in moderating the effects. The results initially indicated there was a statistically significant relationship between mothers’ work hours and children’s problems, such that children of mothers who worked 15 to 24 or 45 or more hours per week had the lowest initial level of problems. However, the results changed once confounders were accounted for, indicating that background factors were contributing to the effect of long work hours. The findings of Model 2, that part-time work is related to better socio-emotional well-being for young children, supports previous research (Strazdins et al 2011; Hadzic et al 2013); however, the results of this study suggested that the relationship between maternal work hours and children’s wellbeing as they aged differs depending on the education level of the mother. In this study, the results indicated mothers’ longer working hours are associated with an increase in difficulties for children in some families, while short part-time hours jobs for mothers may be related to socio-emotional and behavioural problems in children of less educated mothers. Young children of mothers who were more highly educated had more problems if the mother worked longer working hours. In contrast, preschool aged children of mothers who had not completed high school had more problems if their mother worked shorter hours. These findings are somewhat consistent with Baxter et al (2012). The finding that part-time jobs may be related to poorer child outcomes for some children may highlight the importance of choice regarding work hours, as well as the timing and quality of part-time jobs for lower educated workers, consistent with other research (Nicholson et al 2012; Strazdins et al 2006, 2010). The differences in the effects of mothers being outside the labour force by education level may also be associated with the role of choice and opportunity in not working or looking for work, and associated distress. Mothers with lower

SLIDE 15

15 levels of educational attainment may perhaps be discouraged job seekers or those who would like to go to work but for whom it is not financially viable because of child care costs. However, the higher rates of difficulties associated with the NILF category across most groups, may be also related to stress associated with responsibilities and circumstances other than work – such as caring for others, being in education or having a disability. Overall, this study provides some support for Williams’ (Williams et al 2013, Williams & Boushey 2010) hypothesis that longer working hours creates difficulties in more advantaged families. The results suggest work-family interventions that target work hours would need a nuanced approach. Differences in children’s wellbeing by mothers’ work hours and educational attainment may be associated with the distress associated with lack of choice and opportunity regarding changes in work hours. Researchers in the USA (Lautsch & Scully 2007) have argued that working fewer hours is not a universal solution to work-family balance issues and the characteristics of the parents and family need to be taken into account. Additionally, Australian researchers (Wooden et al 2009) have noted that it is the desire to change work hours that is related to wellbeing, not actual hours worked. Part-time work may be related to a desire for more hours, while long hours may be related to a desire for less, and it is this mismatch that is related to wellbeing, not actual hours worked. The current findings have implications for supporting employed mothers to work the number of hours they desire in good quality jobs — to improve the mental health of both mothers and their

• children. The present analyses have shown the distribution of under- and over-employment by maternal

education level may have implications for children. This implies there may be longer term impacts of the deregulation of the Australian labour market that are not evenly distributed in the population. In the context

• f the rise in part-time jobs and underemployment in Australia (Australian Bureau of Statistics 2016a), these

effects may be concerning. In considering the findings of this paper it should be noted the analyses are based on a general population sample and the differences in children’s wellbeing by maternal work hours were small. There are several other limitations of this study that should be considered. Firstly, there is potential bias in parental reports of children’s emotional and behavioural problems (Johnson et al 2013). Additionally, although longitudinal data was utilised, this study examined concurrent relationships, and causal interpretation cannot be placed on the results. Importantly, the variance components of the multi-level modelling approach indicated variability in children’s initial socio-emotional difficulties, and in their trajectories, that was unaccounted for. This indicates that the models do not account for the full-range of factors that could affect the relationship between maternal work hours and children’s socio-emotional difficulties, and the mechanisms linking work hours and children’s problems in different families requires further investigation. The differences in the effects of short hour jobs may highlight the role of other characteristics of part-time work that differ for women with different levels of educational attainment, that have also found to be associated with parental stress and difficulties for children, but have not been explicitly examined in this

• study. Deeper exploration of socio-economic differences in the relationship between work and children

requires an explicit examination of other work characteristics such as job quality and work schedules. Future

SLIDE 16

16 analyses should also consider differences in these effects between other socio-demographic groups; in particular the role of family structure and partner factors require further exploration. In conclusion, this study extends previous Australian research by examining the relationship between parental work hours and children’s mental and health and behaviour over a 10 year period, and examining the role of maternal education in shaping the effect. The study showed that children’s socio-emotional health is associated with mothers’ working hours, however the nature of this relationship differs depending on characteristics of the mother. Underemployment may adversely affect children of lower-educated mothers, while long work hours may affect children of higher educated mothers. The findings imply there may be longer term impacts of changes in the Australian labour market that are unevenly distributed in the population.

SLIDE 17

17

Appendix: Full model results

Table 3. Full results of growth-curve models for children’s socio-emotional and behavioural problems from ages 4 to 15 Model 1 Model 2 Model 3 Fixed effects  SE  SE  SE Intercept 10.45* 0.12 9.74* 0.17 9.61* 0.20 Children’s age

• 1.08*

0.06

• 0.93*

0.06

• 0.88*

0.06 Children’s age squared 0.17* 0.01 0.16* 0.01 0.16* 0.01 Children’s age cubed

• 0.01*

0.00

• 0.01*

0.00

• 0.01*

0.00 Mothers’ work hours NILF (ref.)

• Unemployed
• 0.38

0.26

• 0.53*

0.26

• 0.62

0.39 0-14

• 0.50*

0.15

• 0.33*

0.14

• 0.28

0.23 15-24

• 0.64*

0.14

• 0.41*

0.14

• 0.21

0.22 25-34

• 0.51*

0.17

• 0.29+

0.17

• 0.18

0.27 35-44

• 0.42*

0.16

• 0.17

0.17

• 0.15

0.26 45+

• 0.63*

0.22

• 0.30

0.22

• 0.63

0.37 Children’s age × work hours NILF (ref.)

• Unemployed

0.00 0.04 0.02 0.04 0.04 0.06 0-14 0.01 0.03 0.00 0.03

• 0.02

0.04 15-24 0.01 0.02 0.00 0.02

• 0.05

0.04 25-34

• 0.01

0.03

• 0.02

0.03

• 0.05

0.04 35-44

• 0.02

0.02

• 0.03

0.02

• 0.06

0.04 45+ 0.02 0.03 0.00 0.03 0.01 0.05 Children’s gender (ref. = boy)

• 1.38*

0.12

• 1.38*

0.12 Child has a serious medical condition (ref. = no)

• 1.37*

0.09 1.37* 0.09 Child breastfed until 6 months (ref. = yes)

• 1.09*

0.12 1.08* 0.12 Non-biological parent in household (ref. = no)

• 1.14*

0.16 1.13* 0.16 Number of children in household 1

• 0.04

0.14 0.03 0.14 2 (ref.)

• 3 or more
• 0.33*

0.09

• 0.34*

0.09 Infant in the household (ref. = no)

• 0.29*

0.12 0.30* 0.12 Equivalised weekly household income (log)

• 0.30*

0.06

• 0.30*

0.06 Mothers' age

• 0.06*

0.01

• 0.06*

0.01 Partner’s employment status No partner

• 0.97*

0.12 0.95* 0.12 Partner not working

• 0.07

0.14 0.05 0.14 Partner working part-time (1-34)

• 0.06

0.12 0.05 0.12 Partner working full-time (35-49) (ref.)

• Partner working longer hours (50+)
• 0.03

0.07

• 0.03

0.07 Mother has serious medical condition (ref. = no)

• 0.25

0.08 0.24* 0.08 Mothers' educational attainment Less than Year 12

• 0.57*

0.13 1.13* 0.23 Completed Year 12

• 0.20

0.15 0.18 0.27 Certificate or Diploma (ref.)

• Bachelor degree or higher
• 0.64*

0.12

• 0.99*

0.26 Mothers’ education level × work hours Did not complete year 12

• NILF (ref.)
• Unemployed
• 0.26

0.65 0-14

• 0.35

0.40 15-24

• 1.05*

0.40 25-34

• 1.09*

0.49 35-44

• 0.85+

0.48 45+

• 0.25

0.72 Completed Year 12

• NILF (ref.)
• Unemployed
• 0.21

0.80 0-14

• 0.35

0.45 15-24

• 0.14

0.43

SLIDE 18

18

Table 3. Full results of growth-curve models for children’s socio-emotional and behavioural problems from ages 4 to 15 Model 1 Model 2 Model 3 25-34

• 0.47

0.51 35-44

• 0.19

0.49 45+

• 0.84

0.76 Certificate or Diploma (ref.)

• Bachelor Degree or higher
• NILF (ref.)
• Unemployed
• 0.19

0.78 0-14

• 0.13

0.36 15-24

• 0.23

0.34 25-34

• 0.66+

0.39 35-44

• 0.60

0.39 45+

• 1.08*

0.50 Mothers’ education level × work hours × children’s age Did not complete year 12

• NILF
• 0.09*

0.04 Unemployed

• 0.20+

0.12 0-14

• 0.04

0.07 15-24

• 0.05

0.05 25-34

• 0.03

0.07 35-44

• 0.03

0.06 45+

• 0.10

0.10 Completed Year 12

• NILF
• 0.18*

0.05 Unemployed

• 0.18

0.14 0-14

• 0.12+

0.07 15-24

• 0.04

0.06 25-34

• 0.05

0.07 35-44

• 0.08

0.06 45+

• 0.20+

0.10 Certificate or Diploma (ref.)

• Bachelor Degree or higher
• NILF
• 0.00

0.04 Unemployed

• 0.05

0.12 0-14

• 0.01

0.05 15-24

• 0.01

0.04 25-34

• 0.03

0.04 35-44

• 0.01

0.04 45+

• 0.05

0.06 Variance components  SE  SE  SE Level 1 error (within individual) 2.89 0.02 2.88 0.02 2.87 0.02 Level 2 (between individual) Initial status 4.46* 0.06 4.05* 0.06 4.04* 0.06 Rate of change 0.38* 0.01 0.38* 0.01 0.38* 0.01 Model fit statistics Deviance 134672.96 131064.47 131000.11 AIC 134713.00 131138.50 131152.10 BIC 134874.80 131437.10 131765.50

SLIDE 19

19

References

Aryee, S. (1992). Antecedents and outcomes of work-family conflict among married professional women: Evidence from Singapore. Human Relations, 45(8), 813-837. Australian Bureau of Statistics. (2013). Working Time Arrangements, Australia, November 2012. Cat. No. 63420DO002_201211. Australian Bureau of Statistics, Canberra. Australian Bureau of Statistics. (2014). Consumer Price Index, Australia, June 2014. Cat. No. 6401.0. Australian Bureau of Statistics, Canberra. Australian Bureau of Statistics. (2016a). Labour Force, Australia, Detailed - Electronic Delivery, February

• 2015. Cat. No. 6291.0.55.001. Australian Bureau of Statistics, Canberra.

Australian Institute of Family Studies (2015). Longitudinal Study of Australian Children Data User Guide – November 2015. Australian Institute of Family Studies, Melbourne. Baxter, J. & Chesters, B. (2011). Perceptions of work-family balance: How effective are family-friendly policies? Australian Journal of Labour Economics, 14(2), 139-151 Baxter, J., Gray, M., Hand, K., & Hayes, A. (2012). Parental joblessness, financial disadvantage and the wellbeing of parents and children. Department of Families, Housing, Community Services and Indigenous Affairs Occasional Paper No. 48. Commonwealth of Australia, Canberra. Beck, U. (2000). The brave new world of work. Polity Press, Cambridge. Bronfenbrenner, U., & Crouter, A. C. (1982). Work and family through time and space. Families that work: children in a changing world. Sheila B. Kamerman and Cheryl D. Hayes (eds.). National Academy Press, Washington, DC. Bronfenbrenner, U., & Morris, P. A. (2006). The bioecological model of human development. Handbook of child psychology. DOI: 10.1002/9780470147658.chpsy0114 Charlesworth, S., Strazdins, L., O'Brien, L., & Sims, S. (2011). Parents' jobs in Australia: Work hours polarisation and the consequences for job quality and gender equality. Australian Journal of Labour Economics, 14(1), 35-57. Cooklin, A. R., Dinh, H., Strazdins, L., Westrupp, E., Leach, L. S., & Nicholson, J. M. (2016). Change and stability in work–family conflict and mothers' and fathers' mental health: Longitudinal evidence from an Australian cohort. Social Science & Medicine, 155, 24-34. Cooksey, E. C., Menaghan, E. G., & Jekielek, S. M. (1997). Life-course effects of work and family circumstances on children. Social Forces, 76(2), 637-665. Dooley, D., Prause, J., & Ham-Rowbottom, K. A. (2000). Underemployment and depression: longitudinal

• relationships. Journal of Health and Social Behavior, 41(4), 421-436.

SLIDE 20

20 Frone, M. R., Russell, M., & Cooper, M. L. (1992). Antecedents and outcomes of work-family conflict: testing a model of the work-family interface. Journal of Applied Psychology, 77(1), 65. Germov, J. (2011). Working 24/7: The new work ethic. In J. Germov and M. Poole (Ed.), Public sociology: An introduction to Australian society (2nd edition) (pp. 397-420). Allen & Unwin, Crows Nest. Goodman, A., Lamping, D. L., & Ploubidis, G. B. (2010). When to use broader internalising and externalising subscales instead of the hypothesised five subscales on the Strengths and Difficulties Questionnaire (SDQ): data from British parents, teachers and children. Journal of Abnormal Child Psychology, 38(8), 1179-1191. Greenhaus, J. H., & Beutell, N. J. (1985). Sources of conflict between work and family roles. Academy of Management Review, 10(1), 76-88. Hadzic, R., Magee, C. A., & Robinson, L. (2013). Parental employment and child behaviors: Do parenting practices underlie these relationships? International Journal of Behavioral Development, 37(4), 332-339. Hamilton, L. C. (2012). Statistics with Stata: version 12. Cengage Learning, Boston. Heckman, J. J., Stixrud, J., & Urzua, S. (2006). The effects of cognitive and noncognitive abilities on labor market outcomes and social behavior. Journal of Labour Economics, 24(3), 411-482. Higgins, C. A., Duxbury, L. E., & Irving, R. H. (1992). Work-family conflict in the dual-career family. Organizational Behavior and Human Decision Processes, 51(1), 51-75. Johnson, S., Li, J., Kendall, G., Strazdins, L., & Jacoby, P. (2013). Mothers' and fathers' work hours, child gender, and behavior in middle childhood. Journal of Marriage and Family, 75(1), 56-74. Khanam, R., & Nghiem, S. (2016). Family Income and Child Cognitive and Noncognitive Development in Australia: Does Money Matter? Demography, 53(3), 597-621. Lautsch, B. A., & Scully, M. A. (2007). Restructuring time: Implications of work-hours reductions for the working class. Human Relations, 60(5), 719-743. Li, J., Johnson, S. E., Han, W. J., Andrews, S., Kendall, G., Strazdins, L., & Dockery, A. (2014). Parents’ nonstandard work schedules and child well-being: A critical review of the literature. The Journal of Primary Prevention, 35(1), 53-73. Lind, J. N., Li, R., Perrine, C. G., & Schieve, L. A. (2014). Breastfeeding and later psychosocial development of children at 6 years of age. Pediatrics, 134(Supplement 1), S36-S41. McGinnity, F., & Calvert, E. (2009). Work-life conflict and social inequality in Western Europe. Social Indicators Research, 93(3), 489-508.

SLIDE 21

21 Mahon, R., Bergqvist, C., & Brennan, D. (2016). Social Policy Change: Work–family Tensions in Sweden, Australia and Canada. Social Policy & Administration, 50(2), 165-182. Nicholson, J. M., Strazdins, L., Brown, J. E., & Bittman, M. (2012). How parents' income, time and job quality affect children's health and development. Australian Journal of Social Issues, 47(4), 505.

• OECD. (2012). The Future of Families to 2030, OECD Publishing.

http://dx.doi.org/10.1787/9789264168367-en Parke, R. D. (2004). Development in the family. Annual Review of Psychology, 55, 365-399. Pocock, B., Skinner, N., & Williams, P. (2007). Work, life and time: the Australian work and life index

• 2007. Centre for Work+ Life, University of South Australia, Adelaide.

Pocock, B., Charlesworth, S., & Chapman, J. (2013). Work-family and work-life pressures in Australia: advancing gender equality in “good times”? International Journal of Sociology and Social Policy, 33(9/10), 594-612. Presser, H., & Ward, B. (2011). Nonstandard work schedules over the life course: a first look. Monthly Labor Review, 134(7), 3-16. Rabe-Hesketh, S., & Skrondal, A. (2008). Multilevel and longitudinal modeling using Stata. STATA press. Rendall, M., DeRose, A., Kreute, F., Lappegard, T., Reeder, L., Ronsens, M., & Toulemon, L. (2013). Employment impacts on entry to parenthood in different family policy regimes. HILDA conference 2013 paper. Melbourne Institute of Applied Economic and Social Research, University of Melbourne, Melbourne. Repetti, R. L., Taylor, S. E., & Seeman, T. E. (2002). Risky families: family social environments and the mental and physical health of offspring. Psychological Bulletin, 128(2), 330. Rochat, T. J., Houle, B., Stein, A., Coovadia, H., Coutsoudis, A., Desmond, C., & Bland, R. M. (2016). Exclusive breastfeeding and cognition, executive function and behavioural disorders in primary school-aged children in rural South Africa: a cohort analysis. PLoS Medicine, 13(6), 1-30. Rose, J., Hewitt, B., & Baxter, J. (2013). Women and part-time employment Easing or squeezing time pressure? Journal of Sociology, 49(1), 41-59. Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and event

• ccurrence. Oxford University Press, New York.

Smith, J. R., & Brooks-Gunn, J. (1997). Correlates and consequences of harsh discipline for young children. Archives of Pediatrics and Adolescent Medicine, 151(8), 777–786.

SLIDE 22

22 Strazdins, L., Clements, M. S., Korda, R. J., Broom, D. H., & D’Souza, R. M. (2006). Unsociable work? nonstandard work schedules, family relationships, and children’s well‐being. Journal of Marriage and Family, 68(2), 394-410. Strazdins, L., Lucas, N., Shipley, M., Mathews, B., Berry, H. L., Rodgers, B., & Davies, A. (2011). Parent and child wellbeing and the influence of work and family arrangements: a three cohort study. FaHCSIA Social Policy Research Paper, (44). Commonwealth of Australia, Canberra. Strazdins, L., Shipley, M., Clements, M., Obrien, L. V., & Broom, D. H. (2010). Job quality and inequality: Parents’ jobs and children’s emotional and behavioural difficulties. Social Science & Medicine, 70(12), 2052-2060. The Work + Family Policy Roundtable. (2016). Work Care & Family Policies – Election Benchmarks 2016. Women + Work Research Group, The University of Sydney, NSW, Australia. Voydanoff, P. (2005). Social Integration, Work‐Family Conflict and Facilitation, and Job and Marital

• Quality. Journal of Marriage and Family, 67(3), 666-679.

Williams, J. C., & Boushey, H. (2010). The three faces of work-family conflict: The poor, the professionals, and the missing middle. http://dx.doi.org/10.2139/ssrn.2126314 Williams, J. C., Blair‐Loy, M., & Berdahl, J. L. (2013). Cultural schemas, social class, and the flexibility

• stigma. Journal of Social Issues, 69(2), 209-234.

Wooden, M., Warren, D., & Drago, R. (2009). Working time mismatch and subjective well‐being. British Journal of Industrial Relations, 47(1), 147-179. Yamauchi, C. (2010). Parental investment in children: Deferential pathways of parental education and mental health. Economic Record, 86(273), 210–226. Yeung, W. J., Linver, M. R., & Brooks-Gunn, J. (2002). How money matters for young children’s development: Parental investment and family processes. Child Development, 73(6), 1861–1879.