Risk and Resilience in Six- and Ten-Year-Old Children

1. Introduction

We know from much of the work in developmental psychopathology over the last thirty years that when children live in adverse environments they are much more likely to develop emotional and behavioural problems than when they live in less stressful environments. One of the important observations, however, about children living in adverse circumstances is that some children seem to be able to withstand even high levels of adversity, without their functioning becoming impaired. In this report our aim was to create an index of adversity so that we could identify the children who were most at risk in their development. We were then in a position to examine the factors in the lives of children that contributed to more optimal development. The National Longitudinal Study of Children and Youth (NLSCY) provides an excellent opportunity to examine risk and resilience in different age groups of children.

The data that are presently available from the NLSCY are cross-sectional. Questions relating to risk and resilience in children are most optimally addressed using longitudinal data. A child whose development is not compromised at one point in time, may be developing an adaptation strategy that leaves them vulnerable to future adversity. Similarly, a child who shows symptomatic behaviour at one point in time may be struggling to adapt to difficult stresses in his or her environment, such that at future points in their development they are more able to withstand adversity (Masten and Coatsworth, 1998). The data presented in this cross-sectional analysis, however, allow for an initial formulation and examination of issues related to environmental adversity and factors in children's lives that enable them to cope effectively with such adversity.

1.1 Adversity and Children's Development: Factors That Are Associated With Increased Psychological Disorder in Children

Through epidemiological studies it has been possible to establish those factors that are associated with increased psychiatric problems in children. Rutter et al.(1975) using data from the Isle of Wight study found that parental psychiatric disorder, large family size, overcrowding in the home, marital conflict/divorce, and parental criminality were all factors that increased the likelihood that children would show psychiatric disorder.

Since this early work many studies have replicated the increased risk of psychiatric disorder that is associated with these environmental risks. Marital conflict and particularly the open expression of anger and hostility have been found to increase the likelihood that children will show externalizing problems (Cummings and Cummings, 1988; Jenkins and Smith, 1991). Divorce shows a similar association with children's psychopathology (Cherlin, Furstenberg, Chase-Lansdale, Kiernan and Robins, 1991). Several different kinds of psychiatric disorder among parents have been shown to be associated with an increased risk of psychiatric disorder among children. Parental depression (usually maternal) is most strongly associated with internalizing outcomes in children, but it is also associated with an increased risk in externalizing problems (Hammen, Burge, Burney and Adrian, 1990). Parental alcohol abuse has been shown to be associated both with internalizing and externalizing outcomes in children (von-Knorring, 1991).

Other environmental factors have also been found to be associated with poor developmental outcomes in children. When children are born to teenaged mothers they are more likely to have emotional and behavioural problems in their early childhood (Hetherington, 1997). Poverty and low socio-economic status play an important role in how children manage over the course of their development (Dodge, Pettit and Bates, 1994; Sampson and Laub, 1994) . Using data from the NLSCY study, Offord and Lipman (1996) reported an inverse relationship between income adequacy and behavioural problems. As income decreased, behavioural problems increased. Children who show cognitive impairment, even to a relatively small degree, are more likely to show emotional and behavioural problems than children who do not show such compromised cognitive development (Rutter, Tizard and Whitmore, 1970; Werner and Smith, 1982; Anderson et al., 1989). Hostility in the parent child relationship repeatedly emerges as a very strong predictor of emotional and behavioural problems in children. Dodge, Bates and Pettit (1990) investigated the relationship between harsh discipline and aggression among children in the general population. Six months before children entered school their parents were interviewed about the frequency and severity with which they gave physical punishment to their children. On entering school the children's aggression with peers was assessed. Frequency and severity of harshness predicted aggression in school. Such relationships have been reported in several different studies (Simons, Whitbeck, Conger and Chyi-In, 1991; Sternberg et al., 1993).

Although for all these factors there is strong empirical evidence to show they are associated with poor outcomes in children, it may be that such effects occur when multiple risks occur together. Rutter (1979) reported that the presence of an isolated adverse factor in a child's life did not raise the risk of disorder but it was only when risks occurred together that children showed an increase in behavioural problems. He found that children with one risk were not more likely than children with no risks to show disorder. Five percent of children with two risks showed serious disorder compared to 20 percent of children with four or more risks. Sameroff, Seifer, Bartko (1997) have also found that it is when several risks occur together that development is most compromised.

The NLSCY data set provides an excellent opportunity to examine whether children with single risks in their lives show more emotional and behavioural problems than children who are not exposed to any such risks. It also allows us to examine how rates of disturbance increase with the number of risks in a child's life. Are there levels of risk above which most children show disorder? These questions have important policy implications for service delivery. If high levels of disturbance are only evident when children are exposed to multiple risks, there may be a greater argument for targetting interventions to those children experiencing multiple risks in their lives. Such a pattern would suggest that children can cope adequately with small amounts of stress, but that as stress increases they reach a point beyond which they can not respond adaptively.

1.2 Resilience in Children

Factors have been identified in the lives of children and adults that help them to withstand adversity and develop positive patterns of coping (Rolf et al., 1990; Luther, 1993; Stouthamer-Loeber, Loeber, and Farrington, 1993; Werner, 1993). Rutter (1987) has stressed the importance of making distinctions between different kinds of processes that are associated with better coping in children. He points out that risk research carried out over the last fifty years has told us about factors in children's lives that compromise development. Early resiliency studies simply found the opposite of risk research. When the negative end of the factor is found to be absent in a child's life, the child is resilient. For instance we know that having a poor parent child relationship is a risk in a child's life. It has been argued that having a good parent child relationship when the child is under stress makes for 'resilience'. Rutter suggests that for the study of resiliency to tell us anything different from what the study of risk has revealed, we have to examine the statistical interactions between the risk variable and the putative protective factor. A protective factor is a variable that is found to have a different effect on outcomes at different levels of risk. As Rutter describes: "The essential defining feature... (of a protective factor) is that there is a moderation of the person's response to the risk situation" (Rutter, 1987, p. 317). When a putative protective factor operates to lower disturbance in children exposed to high levels of risk, and has little effect on children at low risk, it tells us something specific about how children in the risk circumstance cope with that risk. He argues that protection involves not the evasion of risk (as in the absence of a risk factor) but in the successful coping with problematic events. We are interested in the processes that are involved in children experiencing high levels of risk being able to negotiate these risks so that development is not impaired.

1.3 The Role Of Social Relationships

One of the consistent factors to be associated with protection, or people's ability to cope with stressful circumstances, is the social support that a person has available to them. For adults experiencing major and traumatic life events, the likelihood of experiencing a depressive breakdown is significantly reduced if the person has someone in their lives in whom they can confide (Cohen and Wills, 1985; Brown and Harris, 1978; Brown et al., 1986). For children the quality of social relationships has also been found to be important in helping to buffer the child from environmental adversity (Werner, 1995). For instance Jenkins and Smith (1990) found that children experiencing high levels of marital disharmony showed lower levels of psychopathology if they had a close relationship with an adult outside the nuclear family, usually a grandparent. We also found that a close relationship with a sibling was associated with lower levels of disturbance in children exposed to marital conflict but not associated with disorder among children who were not exposed to such stress (Jenkins, 1992). Cicchetti and Nurcombe (1997) found that social relationships were protective for low income children. Interestingly, they also found that more compromised youngsters, those who were from low income homes and were maltreated, showed less evidence of being protected by positive social relationships. These children were more protected by attributes of their personalities.

The way in which social support operates to lessen risk in children and adults is not entirely which the person can talk about problems appears to be beneficial and may relate to developing a metacognitive structure that allows the individual to think differently about adverse events in their lives such that the events do not elicit such distress reactions (Gottman, Fainsilber Katz and Hooven, 1996; Oatley and Jenkins, 1996). Or it may be that social support provides the individual with a social role through which an identity is constructed and maintained (Durkheim, 1905; Oatley and Bolton, 1985; Scheff, 1997). The individual feels part of a community, connected to others and engaged in cooperative action. When children's environments are negative this sense of being part of a community and engaged in cooperative and meaningful activity may be more difficult to achieve.

Three sources of extra-parental affectionate relationships will be examined using NLSCY data: the presence of good sibling relationships, friendships and a close relationship with a teacher. The hypothesis to be tested is that children experiencing high levels of adversity will show low levels of psychopathology in the presence of one or more affectionate relationships. A statistical interaction is hypothesized between the environmental risk variable and the presence of an affectionate relationship. As part of these analyses we will examine whether having one affectionate relationship is as effective in helping a person cope as having several affectionate relationships.

1.4 Teacher Monitoring and Class Climate

The second kind of buffering effect to be examined is a school effect. High quality schooling has been found to be associated with lower risk of disorder in children. Some of the types of schooling factors associated with the lowering of risk are having teachers who set and monitor work, provide children with opportunities to behave responsibly and to be praised for this, as well as a positive social climate in the classroom and the school (Mortimore and Stoll, 1988; Rutter, Maughan, Mortimore, and Ouston, 1979).

School processes that contribute to resilience may also be explained by creating an environment in which children can experience a sense of belonging and cooperation. High risk children may feel a sense of lack of connection in aspects of their lives when home environments are characterized by hostility, depression, alcohol abuse, etc. Within an organized and supportive school a sense of cooperation and coherence may be achieved.

Or it may be that the protective effect of schools really lies in their ability to provide more of a monitoring and supervisory function for children who are not experiencing such structures elsewhere in their lives. One of the high risk factors for children in developing externalizing disorder is a lack of parental monitoring and supervision (Patterson, Dishion, and Reid, 1993) and this has also been found to be associated with lower school achievement (Ho Sui-Chu and Willms, 1996). It is possible that when parents do not provide this monitoring and supervisory function for children that teachers and the classroom setting can provide this function.

Here the role of the school in providing a cooperative social environment and the supervisory and monitoring function in high risk children has been examined.

1.5 General Issues In Resilience Research

Luther (1993), Cohen and Wills (1985) and Rutter (1987) have distinguished between different kinds of resilience effects. One kind of protective factor exists when the presence of the putative protective factor is associated with reduced levels of psychopathology in children living in very adverse circumstances but the factor is not associated with levels of psychopathology in non-stressful environments. This is assessed by the presence of a statistical interaction between the index of risk and the putative protective factor, and the absence of a significant relationship between the protective factor and the outcome variable in the low risk group (Cohen and Wills, 1985). This type of pattern can be seen in Figure 1. We shall refer to putative protective factors that combine with a risk variable in this way as "Model A-Protective high risk". Sometimes a putative protective factor is associated with reduced levels of psychopathology in children at high and low risk, but the association is much stronger in the high risk groups than the low risk group. With this kind of pattern the association between the putative protective factor and the outcome variable is significant in the low risk group, and the interaction between the risk factor and the putative protective factor is also significant (Cohen and Wills, 1985). See Figure 2 for a graphical depiction of this pattern that we refer to as "B-Protective high and low risk".

Figure 1. Protective Factor Model A: Protective at high risk only

Figure 2. Protective Factor Model B: Protective at low and high risk

The final pattern to be described is not a protective pattern. It is generally referred to as a main effect model. The "putative protective factor" is associated with psychopathology both when children are living in adverse circumstances and when they are not (see Figure 3). This type of pattern is indicated by a significant main effect for the putative protective factor, but no significant interaction between the risk and protective factor. This is the type of factor that Rutter refers to as the opposite of risk (see "Resilience in children"). The reason that this is not generally referred to as a protective factor is that its absence is associated with psychopathology irrespective of whether the child is experiencing high risk. For an example of such a factor in relation to risk, Jenkins and Smith (1990) found that the quality of the mother-child relationship was associated with psychopathology when children lived with parents who had highly conflictual marriages and when they did not live in such stressful environments.

Figure 3. No protection: Main effects model