4.2  Family influences

From the vantage point of adulthood, perhaps only 10% to 20% of the variation in individual outcomes can be causally attributed to differences in family and home environment, at least within the range of circumstances most commonly studied (Nechyba, McEwan and Older-Aguilar 1999). It is possible that in more extreme circumstances a stronger effect of immediate family and home characteristics could be found.

Many studies find a correlation between parents’ education and their children’s incomes, but as in the case of parental and children’s education, not all studies agree (Levine 1999, Murnane, Willet and Levy 1995). The orthodox view of parental education is that it influences the level of educational attainment of the next generation and that mothers’ education is more influential than fathers’ in this respect, for example, because of the greater amount of time that mothers are likely to spend at home informally improving their children’s educational skills (Haveman and Wolfe 1995). Results from a study of identical female and male twins, that enabled the existence of unmeasured heritable traits and marital sorting to be taken into account, showed that the positive relationship between children’s and mothers’ schooling in their US dataset is solely due to the correlation between mothers’ unobserved endowments and their schooling (Behrman and Rosenzweig 2002).

Overall, parental education per se appears to have little causal significance for outcomes in the next generation. Rather, the ability and willingness of parents to provide cognitive stimulation for their children (which may not necessarily be related to their level of education) may be more important (Lefebvre and Merrigan 1998).

Family income is often cited as a powerful determinant of children’s outcomes with the expectation that raising incomes, particularly of the poorest families with children, would make a major impact on outcomes. There is a consistent correlation in a large number of studies between household income and child and adult outcomes in the next generation (see for example Boggess, Corcoran and Jenkins 1999). Yet the best evidence indicates that the effect of family income on children’s outcomes is complex, correlated with other variables likely to have an independent effect, and likely to be non-linear and mediated through other variables. The effect of low income may be greatest for young children, and for cognitive development and school readiness. However, the evidence indicates that even as much as a doubling of parental income would increase children’s cognitive test scores by amounts unlikely to have any practical impact on their later trajectories (Mayer 2002).

Parental employment is perhaps the most important route by which family income can be raised. However, the benefits of extra income (both for parents and for children) need to be weighed against the effects on children’s development (Ruhm 2000). Effects of parental employment may vary by the age of the child at which the parent is employed, by the child’s gender, by the gender of the parent employed, and may also be manifest at differing stages during the child’s development. Some effects evident at one point may fade over time.

A number of studies show a correlation between early maternal employment and poorer behavioural and cognitive outcomes during early childhood for most children (see Nechyba et al 1999).[16][17] However, a New Zealand study found that at age 10 there is no association between current maternal employment and children’s competencies (Wylie, Thompson and Lythe 2001).

Parental employment in the very early years and in adolescence appears to have the ability to harm children’s development, even when the potential effects of the extra income earned are taken into account. A study using data from the National Longitudinal Survey of Youth (NLSY) that controls for unobserved characteristics concludes that substantial cognitive gains accrue to children whose mothers stay home for at least two to three years after giving birth (Ruhm 2000). Working during the second and third years appears to have less favourable or more deleterious consequences when the mother is also employed in the first year.[18]

Data from the CHDS show positive but weak associations between mother’s current work and reading scores between ages 8 and 13 (Barker and Maloney 1999). However, the results suggest that mothers’ work does not have a direct causal effect, but is rather a proxy for other characteristics of the mother or family that independently have positive effects on test-scores.

Overall, the evidence suggests that parental employment, particularly during early infancy, and also in adolescence, does have the potential to harm children, even when the effects of extra income earned are taken into account. The effects are likely to be ameliorated by the availability of affordable good quality childcare and out of school supervision. Some evidence suggests that negative effects are stronger for children from two-parent, and more affluent families – but this needs to be seen in a context where such children on average have significantly better outcomes than children from lone parent families.

The impact of parental separation, divorce, lone parenthood, co-habitation and remarriage on children has been addressed in a large number of studies. In general, being raised in a lone parent or step-family (as compared with being raised in a family with both biological parents) or in an “out of family” placement, is correlated with poorer outcomes on a wide range of measures.

Research from longitudinal studies suggests that on average children in lone parent families – regardless of the mother’s age – have higher probabilities of a number of negative outcomes. They include psycho-social distress and behavioural disorders at all ages, academic underachievement and early school-leaving, health problems, accidental injury, early transitions into adult behaviour (such as sexual activity, entering the workforce, leaving home, childbearing) and low income, lower occupational status and unemployment during adulthood (Baker, Pryor and Shirley 2000, Haveman and Wolfe 1993).

Moreover, remarriage (with one step-parent) does little to change children’s school or behavioural outcomes, compared to being in a lone parent family (despite home environments in step-families being similar to those where two biological parents are present) (Baker et al 2000, Nechyba et al 1999). Children whose biological mother is cohabiting with a non-biological father do worse on average than children in lone parent families in terms of cognitive, behavioural and psychological outcomes (McLanahan 1997). Conversely, children with a widowed parent do as well on average as children brought up with two biological parents although they tend to share some of the same material disadvantages as children in other lone parent families (McLanahan and Sandefur 1994, Rodgers and Pryor 1998).

Data from the CHDS shows that children reared in lone parent families due to separation or divorce are at risk of greater health, educational and adjustment difficulties, but much of the risk reflects not the absence of a parent, but rather a series of social and contextual features that are more common in disrupted families and often present before separation rather than being a consequence of separation (Fergusson 1998, Fergusson, Lynskey and Horwood 1994d).

A mounting body of research suggests that the relationship between lone parenthood and adverse outcomes is not primarily a direct causal one, and that much of the increased risk of poor outcomes amongst children in lone parent families is explained by other factors associated with, but not exclusive to, lone parenthood (Baker et al 2000, Landy and Kwan Tam 1998). In general, factors that have an influence on the probability of divorce, lone parenthood, cohabitation and remarriage may independently have an effect on children’s outcomes. Analyses that do not control for these factors will produce biased estimates of the effect of family structure on outcomes.

Many international reviews and meta-analyses have shown that there are negative consequences for children from parental separation that endure into adulthood (Amato 2000, 2001, Amato and Keith 1991). These adverse impacts appear whether family disruptions occur at birth, in early childhood or in adolescence (McLanahan and Sandefur 1994). Short-term distress at the time of separation is common, but this usually fades with time. Long-term adverse outcomes typically apply only to a minority of children experiencing the separation of their parents (Rodgers and Pryor 1998).

Much of the increased risk of poor outcomes associated with separation and divorce may be due to disruption in children’s living arrangements, involving, for instance, loss of social networks and change in school, often in more disadvantaged neighbourhoods (Nechyba et al 1999). These disruptions help explain why children in step-families are at similar risk of adverse outcomes as children in lone parent families after divorce.

A study using data from the DMHDS found children exposed to parental disagreement about discipline, a number of residence changes and changes in step-parents during the primary school years, are more likely to become antisocial and delinquent in adolescence (and that this relationship is not a result of earlier behavioural problems causing parental disagreement, and changes in parents) (Henry, Caspi, Moffitt and Silva 1996). Children in foster care, particularly multiple successive placements, are also likely to be further at risk of adverse outcomes, as a result of disruption to social relations and schooling.

Overall, disrupted family structures are not themselves a problem – there are many diverse family structures that can function well for parents and children alike – but the instability of relationships, households and families that accompanies change is a great threat to the well-being of children Pryor and Rodgers (1998).

Several studies present a pessimistic picture of the likelihood of positive adult outcomes for the children of teenage mothers. Pregnancies to teenagers carry a higher frequency of miscarriage, stillbirth, prematurity, antenatal complications and neonatal death than pregnancies in the general population (Fergusson et al 2001, Wolfe, Wilson and Haveman 2001). Pre-school age children of teenage mothers tend to show more behaviour problems, including higher levels of aggression and lower impulse control, than their peers born to older mothers and in adolescence experience higher rates of delinquency, grade failure, early school leaving and unemployment than their peers born to older mothers (Coley and Chase-Lansdale 1997). The children of women who have their first child as a teenager are 2.6 times more likely themselves to become parents as teenagers than the children of older mothers (Jaffee, Caspi, Moffitt, Belsky and Silva 2001).

However, most children born to teenage mothers do well avoid negative outcomes (Jaffee et al 2001). The key question is whether negative child outcomes occur because of factors resulting from early parenthood per se or whether they can be attributed to the pre-pregnancy characteristics associated with the young mother (Fergusson et al 2001).

A study (Jaffee et al 2001)using the DMHDS found that once maternal characteristics and family circumstances were controlled, the effect of teen childbearing per se on outcomes was statistically insignificant (Jaffee et al 2001). Analysis based on the CHDS suggests that poor outcomes for children are likely to be not so much a reflection of the youth of the mother, but the product of a selective process whereby the women who are least well-equipped tend to give birth at a younger age (Fergusson and Woodward 1999). A study using data from the NLSY in the USA found that the negative effect of early motherhood on the academic performance of their children was almost entirely explained by pre-birth individual and family background factors of teen mothers themselves (Levine, Pollack and Comfort 2000).

A number of studies have attempted to overcome problems of the overestimation of the negative effects of teenage childbearing arising from unmeasured effects that jointly determine poor outcomes in adulthood and teenage pregnancy by comparing sisters’ outcomes, cousins’ outcomes and the outcomes of teenagers who miscarry with those who do not (see for example Hotz, Williams McElroy and Sanders 1999). They suggest substantially smaller effects of teenage childbearing per se on outcomes for mothers and their children than do more typical studies. Also, due to the early onset of health problems and restricted economic opportunities, teenage pregnancies may be desirable in some sub-populations, and indeed may be a rational choice for some, typically poor minority, girls (Geronimus 1997).

There is a lively debate on the consequences of early childbearing. Overall, the evidence indicates that growing up with a mother who first gave birth as a teenager can be a risk factor for a range of negative child and adult outcomes. However, the impact of teenage pregnancy as a cause of subsequent poor outcomes is more modest than simple correlations suggest. Although most of the problems experienced by teenage mothers can be attributed to pre-existing factors, there is still a substantial degree of disadvantage attributable to having given birth as a teenager (UNICEF 2002).

There is evidence that children from difficult or severely dysfunctional families disproportionately suffer severe substance abuse, high levels of depression and anxiety and suicide and run away from home (CSR Incorporated 1997, Hider 1998).

Some studies suggest a low association between parenting practices and child outcomes, while others have identified that positive parenting practices can significantly reduce the odds of poor outcomes such as emotional disorders and aggressive behaviour (Pryor and Rodgers 2001). Although there is general consensus that authoritative (high control/high warmth), parenting is optimal for facilitating positive child development, there is a debate in the literature regarding the culture-specific and context-specific nature of parenting styles. Poor parental supervision, and the parents’ rejection of the child are moderately strong predictors of subsequent delinquency by the child.

The maltreatment of children in the form of sexual or physical abuse and neglect can have short and long-term consequences for well-being. Children who suffer maltreatment during infancy and early childhood can develop emotional, behavioural and cognitive problems that have repercussions throughout their lives.

New Zealand research, replicated in many international studies, suggests that children who experienced severe or harsh parental punishment practices had one-and-a-half to four times higher rates of conduct problems, substance abuse, depression, anxiety and violent crime in early adulthood, than those whose parents did not use physical punishment (Fergusson 1998, Jenkins and Keating 1998, Kelley, Thornberry and Smith 1997). The higher rates of problems among abused children appear to reflect the consequences of generally compromised childhoods rather than the specific effects of the abusive treatment (Fergusson 1998).

Children who have been sexually abused exhibit a variety of problems both in the short term and later on as adults (Briere and Elliott 1994, Oddone, Genuis and Violato n.d.). The CHDS findings on sexual abuse suggest that, independently of the socio-economic context, exposure to child sexual abuse is causally linked to later mental health and adjustment problems in some children (Fergusson 1998). This conclusion accords with recent research in basic brain science that brains are sculpted by experiences (Teicher 2002).

Although maltreatment is a serious threat to the short and long term well-being and development of children and adolescents, in some cases children do not appear to suffer significant effects (Masten and Coatsworth 1998).

4.3  Community influences

Poor families and those from particular ethnic groups may face limitations on the communities which they can easily choose to live in. Communities, which differ in terms of social capital, economic conditions, material resources and infrastructure, social norms and peers may create an avenue through which parental socio-economic characteristics affect child outcomes (Brooks-Gunn and Duncan 1997, Corak and Heisz 1998, McCulloch and Joshi 2000, Phipps 1999). However, most of the studies on community influences emanate from the United States, and may have limited application in New Zealand, where neighbourhoods may be less distinct from one another.

Community and neighbourhood effects on outcomes are hard to isolate and measure, but are typically modest, and appear to relate most directly to antisocial behaviour and crime rather than educational attainment outcomes (Burgess et al 2001, Nechyba et al 1999).. They are probably less important than family effects, which are, in turn, less important than individual effects. For young children, wider social influences such as neighbourhood effects operate through their impact on the family (Zuckerman and Kahn 2000). However, neighbourhood effects become more important as the child gets older (Nechyba et al 1999).

The most important contribution an older child may make to his or her own well-being is through his or her choice of peers (Nechyba et al 1999). Once in school, peer groups may matter more than family for child development (Harris 1995). Peer groups are particularly influential in adolescence (Hawkins, Herrenkohl, Farrington, Brewer, Catalano, Harachi and Cothern 2000). Many studies have shown that antisocial peer groups play an important part in the development of deviancy and violence (US Surgeon General 2001). However, the evidence of peer effects is difficult to interpret because of unobserved variables and the number of different peer groups – such as school, class, and friends – that may be relevant to children’s outcomes. Poor parenting may lead children to associate with a negative peer group, rather than the peer group itself leading a child into problematic and antisocial behaviour (Eccles 1999).

Individual and community levels of education have a strong association with a range of indicators of adult well-being (Treasury 2001a).[19] The effectiveness of schooling in raising skills and capabilities may therefore be particularly important in improving adult well-being. Schools may account for between 8% and 19% of the variation in cognitive and educational outcomes for pupils (Cuttance 1998). New Zealand research gives a much lower estimation of the extent to which schools account for the variation in individual achievement levels (Harker and Nash 1996).[20] Teacher quality may be at least as important as schools for improving achievement.

The evidence that schools affect outcomes appears fairly unequivocal. The key question is why they do so. It is difficult to give an adequate account of the skills that schools produce and the way in which these skills are rewarded in the labour market, and contribute to outcomes in other areas.

Schools play an important part in the development of cognitive skills such as literacy and numeracy that are important for a range of outcomes (Bowles, Gintis and Osborne 2001, Machin 2001). New Zealand analysis of data from the International Adult Literacy Study shows that literacy and numeracy skills appear to have an additional effect on other outcomes (Maré and Chapple 1999). There also appears to be a two-way relationship between behaviour and reading ability in childhood (Rowe 2002). Findings from the DMHDS suggest that behavioural problems pre-date reading problems, and that reading failure exacerbates problem behaviour (McGee, Williams, Share, Anderson and Silva 1986).

In addition to cognitive skills, schooling may also influence other traits of pupils that have independent effects on earnings (Bowles et al 2001). However, the relationship between traits and labour market outcomes is complex, and may not be casual.

Wider features of the socio-economic environment such as the degree of income inequality in a society are correlated with the average level of outcomes such as health (Keating and Hertzman 1999, Kennedy, Kawachi, Glass and Prothrow-Smith 1998, Taylor and Smith 1997, Wilkinson 1996).

However, recent research indicates that this association is unlikely to be causal. The relationship tends to disappear in studies that control for a fuller range of individual characteristics associated with the outcome of interest (Judge and Paterson 2001, Sturm and Gresnez 2002). Recent research in the USA confirms the same, well-known income-health gradient, but finds that differences in mortality are associated with high school educational attainment rather than regional income inequality (Muller 2002). There is similar doubt about the assumed relations between average educational outcomes and the degree of socio-economic dispersion in a country (OECD 2001a).

The level of social capital has also been found to be correlated with a number of negative outcomes (Kawachi, Kennedy, Lochner and Prothrow-Smith 1997, Runyan, Hunter, Socolar, Amaya-Jackson, English, Landsverk, Dubowitz, Browne, Bangdiwala and Matthew 1998). However, the effects on outcomes attributable to social capital disappear once a full range of individual family characteristics were controlled for in a multi-variate model (Fields and Smith 1998).