Risk and Protective Factors Can Be Found in Multiple Contexts

One of the earliest and most replicated findings from the empirical literature is that risk and protective factors are found at multiple levels of the social ecology, or the relationship between humans and their environments, from biological and psychological characteristics of the individual to the family and the community (Rutter, 1987; Werner and Smith, 1982, 1992; Luthar, 2003; Crews, Bender, et al., 2007). For example, a synthesis of 18 meta-analytic reviews of risk and protective factors for children found that the strongest risk factors for internalizing and externalizing problems include comorbid internalizing or externalizing problems, family environment stress (e.g., divorce, single parenting), corporal punishment, lack of bonding to school, delinquent peers, and poor peer relations (Crews, Bender, et al., 2007).

One implication of the multilevel nature of risk and protection is that high-risk groups can be identified on the basis of their individual, family, or community indices of risk. Similarly, preventive interventions can be developed to change risk and protective factors across levels of the social ecology (Maton, Schellenbach, et al., 2004; Sandler, Ayers, et al., 2004). Possible interventions might include psychosocial programs to directly strengthen family and child protective factors or reduce risk factors; programs in social settings that affect child development, such as preschools, schools, and social welfare agencies; or policy-level changes, such as funding prevention services or directly increasing a family’s access to resources.

Risk and Protective Factors Tend to Be Correlated and to Have Cumulative Effects

Risk factors tend to be positively correlated with each other and negatively correlated with protective factors. Thus, some young people have multiple risk factors, and those with multiple risk factors are less likely to have protective factors. For example, in a five-state sample of 6th through 12th graders, those who were in the highest quintile on a cumulative measure of risk factors were likely to be in the lowest quintile on the measure of protective factors (Pollard, Hawkins, and Arthur, 1999).

Furthermore, the presence of multiple risk or protective factors tends to strengthen the prediction of disorder or positive development (Rutter, 1979; Sameroff, Gutman, and Peck, 2003; Goodyer and Altham, 1991; Fergussson and Horwood, 2003). The effect of cumulative risk and protective factors is also found in studies of populations exposed to a common risk factor, such as poverty, parental substance abuse, parental mental illness, or parental divorce (Wyman, 2003; Roosa, Sandler, et al., 1988; Sandler, Wolchik, et al., 1986). In one analysis, for example, although no single risk factor had a strong relation to disorder or positive development, the accumulation of risk factors across family, parent, peers, and community had a substantial effect in predicting multiple problem outcomes (Sameroff, Gutman, and Peck, 2003). When compared with those with eight or more risk factors, youth with three or fewer risk factors had significantly better odds of showing psychological adjustment and self-competence and not showing problem behavior.

One implication of cumulative risk and protection is that preventive interventions may be more effective when they target multiple risk and protective factors rather than just one. In some cases, mediational analysis from an experimental trial can identify which risk and protective factors are responsible for program effects (e.g., Tein, Sandler, et al., 2006). In other cases, when multiple risk and protective factors are targeted, the trial can be designed specifically to test whether components intended to change specific risk and protective factors have additive effects to improve preventive impact (West and Aiken, 1997).

Some Risk and Protective Factors Have Specific Effects, But Others Are Associated with Multiple MEB Disorders and Problem Behaviors

A common finding in the study of major risk factors is that each is associated with an increased likelihood for multiple problem outcomes (e.g., Shanahan, Copeland, et al., 2008; Kessler, Davis, and Kindler, 1997). A rigorous test of the specific versus the general effects of risk factors would require a prospective longitudinal study in order to ensure that the risk factors arise before the onset of disorders and to understand what earlier factors may have contributed to the appearance of a risk factor at a given time (e.g., unemployment leading to parental depression). It would also be necessary to assess a comprehensive set of risk factors and use a meaningful approach to classify them into distinct categories or dimensions. Looking at these effects across meaningful subgroups, such as gender or developmental period, would also be important.

A major analytic issue is whether the associations between the risk factors and multiple disorders are due to the direct effects of these risk factors or to confounding variables that are associated with both the risk factors and with the disorders. One possibility is that the associations between risk factors and multiple disorders could be accounted for by the covariance between risk factors. To test for confounding with other risk factors, studies would need to examine the effects of a given risk factor while controlling for the effects of all associated risk factors. The other possibility is that a risk factor is related to a particular disorder independently of its relations to other disorders. Because childhood disorders are highly comorbid (see Chapter 2), it would be necessary to test the effects of risk factors on disorders while controlling for the effects of other comorbid disorders.

Many studies have attempted to tease apart specific versus general effects of childhood risk factors on mental, emotional, and behavioral problems. One review of more than 200 empirical studies published between 1987 and 2001 found few consistent relations between adverse outcomes and five risk factors: exposure to violence, abuse, divorce/marital conflict, poverty, and illness (McMahon, Grant, et al., 2003). The reviewers also note that serious methodological limitations across the studies precluded drawing strong conclusions from the existing literature.

Several epidemiological studies have found some evidence for associations between specific risk factors and disorders when controlling for the effects of other risk factors (Kessler, Davis, and Kindler, 1997; Shanahan, Copeland, et al., 2008, Cohen, Brook, et al., 1990). Although cross-sectional studies (e.g., Shanahan, Copeland, et al., 2008) have found specific associations with one disorder or disorder domain, they are not able to address the direction of effects between risk factors and disorder or the mechanisms that link risk factors and disorder. An eight-year prospective longitudinal study found that three risk factors—parental mental illness, a mother–stepfather home, and maternal inattention—were significant predictors of more than one problem outcome when controlling for all other risk factors (Cohen, Brook, et al., 1990). Other risk factors had specific effects. For example, somatic risk, social isolation, and lax rules had a specific relation with internalizing problems; parental sociopathy and power-assertive punishment had specific effects on externalizing problems; and neighborhood crime and residential instability had significant relations with substance abuse.

An implication of the findings on specific versus general effects of risk and protective factors is that evaluations of interventions with groups at risk for multiple mental, emotional, and behavioral problems should be designed to detect effects on multiple problem outcomes. For example, parental divorce is associated with other risk factors, such as interparental conflict, parental mental health problems, and harsh and inconsistent parenting. It is also associated with multiple problem outcomes, including substance abuse problems, internalizing and externalizing problems, and academic problems. The potential for multiple benefits from preventive interventions increases the likelihood that they will reduce the burden of disorder on the affected individuals and be a cost-effective investment for society.

Another implication is that preventive interventions should be based on as clear an understanding as possible of the relations between the targeted risk factors and the outcomes of concern. Identification of a risk factor that is specifically associated with some disorders, after the effects of other risk factors and comorbid disorders have been accounted for, increases confidence that it is potentially a causal factor and that modifying that risk will lead to a reduction in the rate of onset of that specific disorder. Prevention strategies that are targeted to high-risk groups would require an understanding of the pathways of risk and protective processes that lead to specific disorders in the risk group and identifying the potentially modifiable processes.

Risk and Protective Factors Influence Each Other and Mental, Emotional, and Behavioral Disorders Over Time

Research in developmental psychopathology (Cichetti and Toth, 1992; Masten, 2006) and resilience (Luthar, 2003) has described multiple models—main effect, moderational, and mediational models—by which risk and protective factors influence each other and the development of emotional and behavior problems over time. Design of prevention interventions should be based on a solid theory grounded in one of these models. In main effect models, risk factors are related to higher levels of disorder, and protective factors have a counterbalancing relation to lower levels of disorder; these effects are often cumulative across multiple risk and protective factors (Rutter, 1979; Sameroff, Guttman, and Peck, 2003; Sandler, Ayers, and Romer, 2001).

In moderational models, a protective factor may reduce the relation between a risk factor and disorder, or a vulnerability factor may exacerbate the relations between a risk factor and disorder. In looking at the relations between stress and psychopathology, for example, such variables as intelligence or academic achievement and positive family environment moderated the adverse effects of stress (Grant, Compas, et al., 2006). Significant interactions have also been found between cumulative indices of risk and protective factors related to community, school, family, and individual variables (e.g., opportunities for school involvement, family attachment, and social problem-solving skills) in predicting substance use, delinquency, and school problems (Pollard, Hawkins, and Arthur, 1999). The aggregated protection measure reduced the odds for a problem outcome when the score on the aggregated risk measure was high, but not when the score was low. However, Grant, Compas, and colleagues (2006) point out that it is difficult to draw general conclusions concerning moderators of stressors because available studies test effects across different stressful situations and different outcomes and use different measures to operationalize the constructs. A clear conceptual framework is needed to integrate findings across putative protective factors. For example, the moderating role of cognitive variables in the effects of stress on depression is established because of well-established theoretical models of depression.

In mediational models, a chain of events is hypothesized in which the effects of risk or protective factors operate through their effects on another risk or protective factor, which in turn affects the development of mental, emotional, and behavioral problems. Particularly when there is a temporal relation among the factors, mediational models provide a powerful approach to looking at the pathways for the development of disorder over time (Cole and Maxwell, 2003; MacKinnon, 2008). Parenting has been found to be a mediator of the effects of multiple risk factors, including poverty, parental divorce, parental bereavement, and parental mental health problems (Sampson and Laub, 1993; Simons, Johnson, et al., 1996; Kwok, Haine, et al., 2005; Grant, Compas, et al., 2003; Wolchik, Wilcox, et al., 2000; Wyman, Sandler, et al., 2000), and multiple interventions have been designed to strengthen parenting (see Chapter 6).

Both moderational and mediational models also show that risk and protective factors in one context (e.g., the family) may influence or be influenced by factors in other contexts. A meta-analysis found that the effects of poverty on children’s internalizing and externalizing problems was partially mediated through its effects to impair effective parenting (Grant, Compas, et al., 2003), whereas other studies found that the effects of poverty are mediated through neighborhood or school variables (Gershoff and Aber, 2006).

Similarly, biological factors may mediate the effects of psychosocial risk and protective factors, and conversely psychosocial risk and protective factors may moderate the effects of biological risk factors. For example, Davies, Sturge-Apple, and colleagues (2007) found that diminished cor-tisol reactivity mediated the relationship between children’s exposure to interparental conflict and the development of externalizing problems over a two-year time period. Research has also identified interactions between genetic and environmental factors to predict disorder (see Chapter 5). However, intervention trials that include genetic or other biological information in the design or analysis of the trial are just beginning to emerge. For example, Brody, Kogan, and colleagues (2008) tested whether a psychosocial intervention to improve parent–child communication of parents in rural poor African American families would moderate the effect of genetic risk due to the presence of a specific variant of the serotonin transporter gene on their children’s initiation of alcohol use, binge drinking, marijuana use, and sexual intercourse in early adolescence. As predicted, they found that program participation moderated the relationship between genetic risk and high-risk behavior; youth at genetic risk who did not receive the intervention had significantly greater increases in risk behavior initiation (1.91 versus .90 on the risk index) from pretest than youth at genetic risk who were assigned to the program.

In addition to identifying direct or indirect associations between risk and protective factors and outcomes of interest, it is also important for the development of interventions to understand the processes by which these effects occur (Luthar and Cicchetti, 2000). Conceptually, several different mechanisms have been proposed. Rutter (1987), for example, discussed five processes by which the effects of risk factors could be reduced: (1) by altering the experience of the risk factor (e.g., by coping); (2) by altering exposure to the risk factor (e.g., by parental monitoring of child involvement with antisocial peers); (3) by averting negative chain reactions (e.g., when harsh parenting leads to child oppositional behavior, which leads to increased conflict); (4) by strengthening protective factors (e.g., self-esteem, adaptive control beliefs); and (5) by turning points, which change the total context and provide new opportunities for development (e.g., moving from institutional care to a positive school environment). Each of these processes may be targeted by preventive interventions.

Evaluations of mediators and moderators of preventive interventions also enable the development of more efficient prevention programs (Brown and Liao, 1999). For example, a meta-analysis of school-based programs for the prevention of problem behavior, such as substance use and delinquency, found that program effects on two theoretical mediators (bonding to school and school achievement) were related to effects on problem behaviors (Najaka, Gottfredson, and Wilson, 2001). These findings provide guidance that future school-based prevention programs could usefully include components to promote these mediating factors.

Depression

The incidence of depression is rare in children through age 6 and low prior to puberty; it increases as young people reach adolescence, with 5 percent of adolescents in a given year experiencing clinical depression and as many as 20 percent having had a clinical episode sometime during their adolescence, rates similar to those found in young adults (Angold and Costello, 2001). Around age 13, depression becomes about twice as common among girls than boys (Angold and Costello, 2001). This changing picture means that prevention programs need to be appropriate for specific developmental periods, taking into account age and gender differences in the mechanisms leading to depression.

Interventions to prevent depression in young people have primarily focused on three risk factors: parents with mood disorders, a depressogenic cognitive style, and elevated levels of depressive symptoms or a history of depression. Across studies, the rates of depression in adolescents with a depressed parent are three to four times higher than rates in those with non-depressed parents (Beardslee, Versage, and Gladstone, 1998). The mechanism is not understood but is likely to involve a combination of genetic and psychosocial influences, including poor parenting, high family stress, and conflict (Garber, 2006; Riley, Valdez, et al., 2008).

For children of depressed parents, preventive interventions have been developed to promote multiple family-level protective processes and to help children cope effectively (Beardslee, Gladstone, et al., 2003). Beardslee and Podorefsky (1988) specifically examined resilience in this population and identified three characteristics in the children: the capacity to accomplish age-appropriate developmental tasks, the capacity to be deeply engaged in relationships, and the capacity for self-reflection and self-understanding. Specifically, the youngsters understood that their parents had an illness, that they were not to blame and were not responsible for it, and that they were free to go on with their own lives. Correspondingly, the researchers found that a commitment to parenting despite depression characterized the parents of resilient children. These resilience characteristics were built into their preventive intervention strategy, illustrating the connection between understanding risk and resilience and developing preventive interventions.

A depressogenic cognitive style is marked by a tendency to ruminate and to see the world without optimism and as not in one’s control (Abramson, Alloy, et al., 2002; Kaslow, Abramson, and Collins, 2000). The results of intervention trials to modify depressogenic cognitive styles have been promising in terms of reducing depressive symptoms and disorder. In some cases, improvement in depressive cognitive mediators accounted for program effects to reduce depression (Clarke, Hornbrook, et al., 2001; Gilham, Reivich, et al., 1995). Clarke’s program included both the risk factor of depressogenic cognitive style and the presence of parental depression.

A high level of depressive symptoms is an important risk factor for the onset of the disorder. In addition, depression is a recurrent disorder, with more than half of those who experience an initial episode experiencing a recurrence. Indicated prevention programs for those with symptoms or a history of depression have focused on changing processes thought to be related to the development of depression, such as depressogenic cognitive styles, explanatory style, interpersonal problem solving, and optimistic thinking (see Garber, 2006, for a review).

Anxiety Disorders

Anxiety tends to begin at an early age and to be chronic (McClure and Pine, 2006; Silverman and Pina, 2008). In the Great Smoky Mountains Study, for example, the mean age of onset was about 7 years old for specific phobias, separation anxiety, and social phobia and about 10 years old for agoraphobia, panic, and obsessive-compulsive disorder (Costello, Egger, and Angold, 2004). Research to identify specific protective, risk, or maintaining factors has been limited (e.g., Craske and Zucker, 2001; Donovan and Spence, 2000; Hudson, Flannery-Schroeder, and Kendall, 2004; Shanahan, Copeland, et al., 2008). The factors identified are generally related to the individual, the family, or school and peers.

Although most of the factors associated with anxiety are implicated in other MEB disorders as well, some are more specific. A child’s temperament, specifically behavioral inhibition (characterized by irritability in infancy, fearfulness in toddlerhood, and shyness in childhood), has been found to be associated with an increased vulnerability to anxiety disorders (e.g., Biederman, Rosenbaum, et al., 1993). Similarly, anxiety sensitivity (a predisposition to fear anxiety-related sensations arising from the belief that these sensations are signs of physical, psychological, or social harm; Reiss, 1991; Reiss and McNally, 1985) also appears to be a specific risk factor for anxious symptoms (e.g., Reiss, Silverman, and Weems, 2001).

In the family, parents with anxiety disorders are more likely to have children who are at increased risk for anxiety disorders than their non-anxious counterparts (e.g., Rosenbaum, Biederman, et al., 1993). It also appears that anxious children are more likely than their nonanxious counterparts to have anxious parents (e.g., Last, Hersen, et al., 1987; Turner, Beidel, and Costello, 1987). Some of this association is likely to be due to shared genes or inheritable temperamental styles (e.g., behavioral inhibition). Children also learn anxious reactions via parental modeling and reinforcement of anxious behaviors (e.g., Barrett, Dadds, and Rapee, 1996; Rapee, 2002). Parents of anxious children are typically more controlling and intrusive than parents of children without clinical anxiety (Hudson and Rapee, 2001; Muris and Merckelbach, 1998), and parental overcontrol and intrusiveness seem to reinforce child inhibition (Rapee, 2001). Attachment style may influence anxiety in children as well (e.g., Erikson, Sroufe, and Egeland, 1985; Sroufe, Egeland, and Kreutzer, 1990). One study identified an anxious-resistant attachment style in infancy as a predictor of anxiety disorders in young adulthood (Warren, Huston, et al., 1997).

Prevention programs have typically targeted children who are at high risk for anxiety due to parental anxiety disorders (Bienvenu and Ginsburgh, 2007), behavioral risk factors for anxiety disorders (e.g., behavioral inhibition; Rapee, Kennedy, et al., 2005), or environmental risk factors (e.g., witnessing community violence; Cooley, Boyd, and Grados, 2004). Prevention programs also have targeted prodromal youth (Dadds, Spence, et al., 1997) and asymptomatic youth (Barrett, Farrell, et al., 2006). Studies are still needed to clarify both the mechanisms by which a prevention program achieves its effects and models of anxiety disorder development (Kellam, Koretz, and Moscicki, 1999).

Schizophrenia

The diagnostic criteria for schizophrenia and other psychotic disorders in the schizophrenia spectrum are undergoing reexamination and revision (Tsuang and Faraone, 2002), but the current diagnostic measurements have sufficient reliability to permit a clear study of risk factors and the developmental course. The incidence of schizophrenia and other psychotic disorders accelerates dramatically during adolescence and young adulthood. Because risk factors have been identified from the prenatal period through young adulthood, opportunities for prevention span these life stages.

Family history can be an important predictor of schizophrenia, and there is strong evidence that genetic factors increase the risk for schizophrenia, with multiple genes operating and interacting in complex ways (Erlenmeyer-Kimling, Rock, et al., 2000; Gottesman, 1991; Owen, O’Donovan, and Harrison, 2005; Tsuang and Faraone, 1994). Having one affected parent conveys a lifetime risk 5 to 15 times that of the general population; having two parents with schizophrenia conveys a nearly 50 percent risk (Bromet and Fennig, 1999). Thus, youth who have an affected first-degree relative are an important potential target group for selective intervention. However, this strategy would not be sufficient, as 90 percent of cases of schizophrenia do not have a family history (Brown and Faraone, 2004; Faraone, Brown, et al., 2002).

An important identified risk factor for schizophrenia is obstetric complications, which convey twice the risk of that in the general population. These complications are sufficiently common that reducing them would have the potential for reducing the overall population rate of schizophrenia (Geddes and Lawrie, 1995). Malnutrition (Susser, Neugebauer, et al., 1996), hypoxia, or infection (Pearce, 2001) are thought to have an adverse effect on the neurodevelopment of the fetus. Thus, ensuring good prenatal care (and reducing maternal rubella infections in developing countries) for all expectant mothers is a universal prevention strategy for schizophrenia to be investigated. Another selective strategy might be aiming supportive interventions to those born with obstetric complications.

Screening for developmental difficulties through multiple stages of life may be appropriate among children born with obstetric complications or whose family history suggests high risk (Brown and Faraone, 2004). Data from studies of high-risk groups suggest that nearly all those with affected family members who later have a diagnosis of schizophrenia had attention problems in childhood as well as diagnoses and difficulties in meeting the important task demands at successive stages of life (Mirsky, Yardley, et al., 1995; Weiser, Reichenberg, et al., 2001).

Identification of the prodromal stage of schizophrenia may present an opportunity to intervene (McFarlane, 2007). Indeed, there are a number of trials currently under way that use low-dose atypical antipsychotics, often in combination with family-focused psychosocial interventions, to prevent the onset of a first episode of psychosis in adolescents and young adults with prodromal symptoms (see Chapter 7). Another promising line of research involves identification and potential intervention among youth and young adults who have underlying signs and symptoms suggesting a genetic liability for schizophrenia without full manifestation of symptoms. The term “schizotaxia” represents a nonpsychotic construct with signs of brain abnormalities and some degree of cognitive, neuropsychological, and social impairment. Such a constellation of negative symptoms and neuropsychological deficits is common among unaffected first-degree relatives of those with schizophrenia (Faraone, Biederman, et al., 1995; Faraone, Kremen, et al., 1995). Particularly relevant for prevention is some evidence that schizotaxia symptoms among adults are ameliorated with low-dose resperidone (Tsuang and Faraone, 2002). Despite major challenges in nosology and ethical considerations regarding labeling and intervention among young people, this line of research holds promise as a strategy for preventing schizophrenia.

Substance Abuse

Substance abuse and dependence tend to emerge in mid-to-late adolescence and to be more common among boys. Substance abuse is greater among young people who experience early puberty, particularly among girls. It is widely accepted that children of drug and alcohol abusers are more likely to develop substance abuse problems (Mayes and Suchman, 2006; Hawkins, Catalano, and Miller, 1992). Considerable evidence supports that a genetic vulnerability to abuse may be conferred at birth, and that this vulnerability may be most significant in relation to the transition from drug use to dependence later in life (Mayes and Suchman, 2006).

During childhood, risk for substance abuse is higher for those who have a difficult temperament, poor self-regulatory skills, are sensation seeking, are impulsive, and do not tend to avoid harm. Children who have early persistent behavior problems are also more likely to develop a substance use problem (Hawkins, Catalano, and Miller, 1992). Furthermore, substance abuse is also often comorbid with anxiety, depression, and attention deficit hyperactivity disorder (Mayes and Suchman, 2006; Hawkins, Catalano, and Miller, 1992; Sher, Grekin, and Williams, 2005). Evidence suggests that parents who form warm, nonconflictual relationships with their children, provide adequate monitoring and supervision, and do not provide models of drug use help protect their children from developing substance use disorders.

During middle childhood and into adolescence, peers play an increasingly important role in children’s psychological functioning. Children who associate with deviant or drug-using peers or who are rejected by peers are more likely to develop substance use problems (Mayes and Suchman, 2006; Hawkins, Catalano, and Miller, 1992; Sher, Grekin, and Williams, 2005). Peers create norms and opportunities for substance use (Mayes and Suchman, 2006; Hawkins, Catalano, and Miller, 1992; Sher, Grekin, and Williams, 2005) and influence attitudes toward substance use. Children and adolescents who have a low commitment to school (Hawkins, Catalano, and Miller, 1992) or experience school failure are more likely to abuse substances. And healthy peer groups and school engagement appear to be protective.

Children and adolescents with more access and availability to alcohol and drugs are more likely to use them (Mayes and Suchman, 2006; Hawkins, Catalano, and Miller, 1992). There is also evidence that child and adolescent substance use is affected by societal norms about use. Norms can be conveyed by laws, perception (or misperception) of peer use, enforcement, taxation, and/or advertising (e.g., alcohol).

Adolescent use of coping strategies involving behavioral disengagement, tendency toward negative emotionality, conduct disorder, and antisocial behavior increase the risk for substance abuse. For both children and adolescents, early drug use predicts later drug use.

In young adulthood, different risk factors appear to represent different pathways to substance abuse. There is consistent evidence of elevated substance abuse, particularly of alcohol, among those attending college, the same group that had lower use in adolescence (Brown, Wang, and Sandler, 2008). This suggests that dormitory life and the fraternity/sorority system, with their lack of parental oversight and consistent exposure to peer models, may create powerful norms encouraging use (Brown, Wang, and Sandler, 2008). For those who do not attend college, antisocial behavior and lack of commitment to conventional adult roles appear to be pathways to abuse.

Underage Drinking

Although not all those who drink in their youth develop substance abuse or substance dependence, underage drinking has received significant public health attention, given the prevalence of drinking among those under the legal drinking age, problematic drinking patterns, and their deleterious effects. A brief discussion of factors related to underage drinking provides an illustration of the developmental aspects of a problem behavior of significant public health concern and similarities with the trajectory of some MEB disorders. The likelihood of serious alcohol dependence as an adult is greatly increased the earlier that young people start drinking (Grant and Dawson, 1997; Gruber, DiClemente, et al., 1996).

Almost one-third of young people between the ages of 12 and 20 report recent drinking, with the majority engaging in binge drinking (five or more drinks), when they drink. Although at lower rates than those in older age groups, drinking is reported by youth as young as age 12, with patterns of heavy drinking increasing with age (National Research Council and Institute of Medicine, 2004b). After age 25, rates of overall drinking, as well as rates of frequent and heavy drinking, steadily decline.

Alcohol use by children and adolescents is influenced over the developmental course by genetics, family, peers, neighborhood, and broader social contexts through norm development, alcohol expectancies, and availability (see the review by Zucker, Donovan, et al., 2008). Risks are apparent as early as ages 3 to 5 years, when children develop the understanding that adults drink alcoholic beverages and learn norms about its use (e.g., men drink more than women).

Children whose parents are drinkers are more likely to be drinkers, and their own drinking correlates well with their perception of their parents’ drinking. This may occur because parents model drinking and help children develop positive expectancies about the effects of alcohol. Children are also exposed to positive images of alcohol use from television and movies. Among adolescents, positive alcohol expectancies are related to initiation of alcohol use.

As children grow older, peer influences become stronger. Peers provide opportunities for modeling of and encouragement for alcohol use. Media and peer culture depicts drinking as a positive part of social life. Adolescents who associate with alcohol-using peers encourage continual use and can be resistant to change. In addition, adolescents tend to overestimate their peer’s drinking, which leads to heavier drinking to conform to the perceived norm.

Public policy in the form of drinking-age laws and their enforcement also influences alcohol use. Lowering the drinking age is associated with increases in teen drunk driving and teen traffic fatalities, while raising it is associated with less teen drunk driving (Wagenaar and Toomey, 2002; National Research Council and Institute of Medicine, 2004b). A higher drinking age (and its enforcement) may decrease underage drinking because it limits access to alcohol, but also by communicating social norms against drinking generally and underage drinking specifically (Hawkins, Catalano, and Miller, 1992). In addition, alcohol consumption decreases with price increases from taxation, particularly among young people with less disposable income (Coate and Grossman, 1988; National Research Council and Institute of Medicine, 2004b).

The risk factors for underage drinking suggest that prevention efforts can be formulated to influence the availability of alcohol, norms about alcohol, and alcohol use expectancies. Limiting media exposure of even young children may decrease normative perceptions of drinking and decrease the development of positive alcohol expectancies (National Research Council and Institute of Medicine, 2004b). Within the family, interventions may be designed particularly around limiting exposure to models of excessive drinking in the home, at family events, and through media sources. Family-based efforts may also target adolescents by monitoring exposure to alcohol-using peers and involvement in alcohol-related activities.

Risk Factors Associated with Multiple Disorders

Negative life events at the family, school or peer, and community levels have been associated with multiple psychopathological conditions, such as anxiety, depression, and disruptive disorders (see Craske and Zucker, 2001; La Greca and Silverman, 2002). Similarly, social support and problem-solving coping appear to have broad protective effects (e.g., Pina, Villalta, et al., 2008).

Studies using nationally representative samples and studies of diverse ethnic, gender, and age groups have found that behavior problems involving serious antisocial behavior, substance use (cigarettes, alcohol, drugs), and risky sexual behavior have common risk and protective factors across developmental stages and across multiple levels of the social ecology, including individual genetic factors, dysfunctional parent-child interactions, and poverty. They also often occur together in adolescence (Biglan, Brennan, et al., 2004).

There appears to be an interrelated set of developmental factors in which earlier risk (or protective) factors increase the likelihood of later ones and in which earlier manifestations of problem behaviors increase the likelihood of later risk factors and problem behaviors (Biglan, Brennan, et al., 2004). Furthermore, early developmental tasks result in developmental competencies during childhood (e.g., verbal fluency) or deficits (e.g., insecure attachment) that can be risk or protective factors at later developmental stages. For example, difficult temperament, which is biologically determined, affects the parenting an infant receives, which in turn affects development of early attachment.

Under one model of the development of a set of problem behaviors—antisocial behavior, high-risk sex, academic failure, and substance use—early family conflict was found to lead to poor family involvement, which later leads to poor parental monitoring and associating with deviant peers (Ary, Duncan, et al., 1999). Both poor monitoring and association with deviant peers lead to higher levels of problem behaviors.

A multiyear retrospective study of the effects of adverse childhood experiences or childhood trauma (psychological, physical, or sexual abuse, witnessing violence against the mother, living with household members who were substance abusers, mentally ill or suicidal, or incarcerated) identified strong graded relationships between these experiences and a range of negative outcomes in adulthood. Adult outcomes associated with these childhood experiences included alcoholism and alcohol abuse, depression, drug abuse, and suicide attempts. The likelihood of multiple health risk factors in adulthood were greater when multiple types of negative childhood exposures were experienced (Felitti, Anda, et al., 1998). An analysis specific to mental health outcomes identified a significant relationship between an emotionally abusive family environment and the level of adverse experience with negative mental health outcomes (Edwards, Holden, et al., 2003).

Poverty. By whatever index used, poverty is a highly prevalent risk factor for children in the United States. In 2007, 18 percent of all U.S. children lived in families with incomes below 100 percent of the federal poverty line; the percentage was higher among ethnic minorities (10 percent of white children, 28 percent of Latino children, and 35 percent of African American children) (U.S. census). However, this measure does not fully capture the proportion of families who do not have sufficient resources to meet their basic needs for housing, child care, food, transportation, health care, miscellaneous expenses, and taxes. The Economic Policy Institute estimated that more than 2.5 times the number of families with incomes at or below the federal poverty line do not have sufficient budgets to meet their basic needs independent of outside subsidies (Boushey, Brocht, et al., 2001). Families who live in poverty or near poverty continually need to make trade-offs between necessities. For example, 65 percent of families with household incomes between 100 and 200 percent of the federal poverty line experienced at least one serious hardship during the prior year, including food insecurity, lack of health insurance, or lack of adequate child care (Boushey, Brocht, et al., 2001).

Poverty is a risk factor for several MEB disorders and is associated with other developmental challenges. Poor children show difficulties with aspects of social competence, including self-regulation and impulsivity (Takeuchi, Williams, and Adair, 1991), and abilities associated with social-emotional competence (Eisenberg, Fabes, et al., 1996). Furthermore, poverty has been found to be associated with a wide range of problems in physical health, including low birth weight, asthma, lead poisoning, and accidents, as well as cognitive development. Poor children are also more likely to experience developmental delays, lower IQ, and school failure (Gershoff, 2003; Brooks-Gunn and Duncan, 1997).

Gershoff, Aber, and Raver (2003) describe three pathways by which poverty affects child development. With the parent investment pathway, the relations between poverty and children’s cognitive development is mediated by the quality of the home environment, which is represented by the amount of cognitively stimulating material in the home (e.g., books, CDs) and how often parents take their children to stimulating places, such as museums and libraries. With the parent behavior and stress pathway, the parents are considered to be under high levels of stress because of their economic difficulties and the occurrence of stressful life events for which they have insufficient resources to cope effectively. Parental stress leads to increased levels of parental depression and interparental conflict, which in turn lead to problems in parenting, including withdrawal from the children, hostility, more frequent use of corporal punishment, and at extreme levels maltreatment. Each of these factors has been found to relate to higher levels of internalizing and externalizing problems in children.

The third pathway involves the neighborhood and community in which poor families are more likely to live. Poor neighborhoods and schools are less likely to have the resources that promote healthy child development and are more likely to be settings that expose children to additional risk factors, such as violence and the availability of drugs and alcohol. Disentangling the effects of the neighborhood and the family is difficult, but there is evidence that many of the factors associated with poor neighborhoods and schools are associated with multiple mental, emotional, and behavioral problems for children (Gershoff and Aber, 2006). More research is needed to tease out these effects and, most importantly, to identify factors that may protect children from the negative effects of living in high-poverty neighborhoods (Roosa, Jones, et al., 2003).

Gershoff, Aber, and Raver (2003) also describe policy- and program-level interventions that may be effective in reducing the negative effects of poverty on children. Their model illustrates interventions to change each of the pathways that lead to adverse outcomes. Parent-directed human capital enhancement policies at the federal and state levels are designed to aid families through programs for job training and education to increase parents’ skills and earning capacity and programs to encourage young women to postpone childbearing so that they can stay in school and obtain better jobs. Income support programs, such as the Earned Income Tax Credit and the Child Support Enforcement Program, are designed to increase the economic self-sufficiency of families. Programs also offer in-kind support, including supplemental child nutrition (e.g., Special Supplement Food Program for Women, Infants, and Children), health insurance for children, and high-quality child care. Parent-directed programs are designed to aid children by enhancing parents’ own well-being and their ability to provide a healthy childrearing environment. Two-generation programs are designed with multiple components to assist both parents and children. For example, Early Head Start focuses on improving child development, family development, and staff and community development. Finally, child-directed programs include providing additional funds for high-poverty schools and for after-school programs in poor neighborhoods.

A natural experiment found that increases in family income and income-related resources were followed by a reduction in both psychiatric and behavioral symptoms in children (Costello, Compton, et al., 2003; see also Chapter 6).

Family Dysfunction and Disruption. With the family as the primary setting for child development from birth through childhood and adolescence, it is not surprising that dysfunction in family relations, particularly parent–child relations, is associated with multiple mental, emotional, and behavioral problems, including those described above. Many risk factors (e.g., poverty, parental mental illness) influence mental, emotional, and behavioral problems and disorders through their effects on parent–child relations (Grant, Compas, et al., 2003; Riley, Valdez, et al., 2008). The discussion here focuses on two broad categories of risk factors that are related to dysfunctional family relations and that provide opportunities for preventive intervention: child maltreatment, which represents the extreme manifestation of family dysfunction, and disruptions in family structure, which create serious challenges to healthy family functioning.

Child Maltreatment. Maltreatment of children by primary caregivers is one of the most potent risk factors for mental, emotional, and behavioral problems, and it has been found to be associated with other serious risk factors, such as poverty and parental mental illness. Protective factors include children’s positive relationship with an alternative caregiver, positive and reciprocal friendships, and higher internal control beliefs (Bolger and Patterson, 2003).

The prevalence of child maltreatment in the United States is unclear. One estimate places it at 1.2 percent of children in 2004 (National Child Abuse Data System). Hussey, Chang, and Kotch (2006) report that 11.8 percent of adolescents report physical neglect, 28.4 percent report physical assault by a parent or caregiver, and 4.5 percent report sexual abuse by a parent or caregiver sometime before they reached the sixth grade. In the National Longitudinal Study of Adolescent Health (Add Health), which includes a nationally representative sample of adolescents, each form of maltreatment was associated with multiple health problems, including depression, substance use, violence, obesity, and poor physical health (Hussey, Chang, and Kotch, 2006). The majority of these associations remained significant after controlling for such demographic variables as family income, age, gender, ethnicity, parent education, region, and immigrant generation (Hussey, Chang, and Kotch, 2006).

In a recent empirical examination in the National Comorbidity Study (Molnar, Buka, and Kessler, 2001), one of the largest and most methodologically sound studies, childhood sexual abuse was reported by 13.5 percent of the women and 2.5 percent of the men. Significant associations were found with 14 mood, anxiety, and substance abuse disorders among women and 5 disorders among men. The analysis controlled for other adversities, including divorced parents, parental psychopathology, parental verbal and physical abuse, parental substance use problems, and having dependents for women.

The lifetime rate of depression was 19.2 percent for those with no childhood sexual abuse and 39.3 percent for those who had experienced abuse (odds ratio = 1.8; Molnar, Buka, and Kessler, 2001). Rates of dysthymia, mania, and posttraumatic stress disorder were also significantly higher for sexually abused women but not for men. The impact of childhood sexual abuse was especially strong for those who had no other adversities; their odds for depression were 3.8 (95 percent confidence interval). For those who reported 5 or more adversities, the odds of depression were 1.7 (95 percent confidence level). There was some evidence that chronic sexual abuse led to higher rates of some disorders (Molnar, Buka, and Kessler, 2001).

Parental psychopathology, especially among mothers, was the most significant family adversity associated with abuse (Molnar, Buka, and Kessler, 2001) and warrants further investigation. However, finding high rates of disorder with abuse but no other risk factors emphasizes the importance of the negative effects of abuse. The persistence of negative effects of child maltreatment is seen in studies that assess functioning across periods of development. For example, the Virginia Longitudinal Study of Child Maltreatment found that of 107 maltreated children who were followed from middle childhood through early adolescence, fewer that 5 percent were functioning well consistently over time (Bolger and Patterson, 2003).

Understanding the factors that influence the linkage between child maltreatment and problem outcomes starts by distinguishing different levels of abuse. In particular, abuse that starts early and is chronic is linked with pervasive and persistent problems across domains of functioning. Children abused in infancy show difficulties in areas that include affect regulation (e.g., high negative affect, blunted affect), hypervigilance, emotional lability, disruptions in their attachment relations, and self-system deficits (e.g., more negative self-representations) (Ialongo, Rogosch, et al., 2006).

The most effective approach to reducing the effects of maltreatment is to prevent its occurrence. Because of the pervasive mental, emotional, and behavioral problems for which maltreated children are at risk, programs that prevent abuse have the potential to avert multiple disorders and promote healthy development across multiple domains of functioning. There is evidence, for example, that a home visiting program for economically poor, single parents has been effective in reducing the occurrence of child abuse (Olds, 2006; see Box 6-1) and that a population-level approach to strengthening parenting reduces rates of abuse in the community (Prinz, Sanders, et al., 2009). Interventions are also aimed at mitigating the impact of abuse after it has occurred. Several randomized trials with maltreated children demonstrated that infant and preschool psychotherapy and a home visiting program were successful in markedly reducing rates of insecure attachment (Ialongo, Rogosch, et al., 2006). Other program models have demonstrated success to improve maltreated children’s relationships with foster parents (Fisher, Gunnar, et al., 2000) and with well-functioning peers (Fantuzzo, Sutton-Smith, et al., 1996).

Family Disruption. Family disruption can occur for many reasons, including separation or divorce, the death of a parent, and incarceration of a parent. The committee focused on parental divorce and bereavement because they have been the subject both of considerable research and of preventive trials.

The rate of divorce in the United States increased from the 1950s through the 1970s and then stabilized or decreased somewhat over the following decades (Bramlett and Mosher, 2002; U.S. Census Bureau, 2005). However, the official divorce rate underestimates the rate of marital disruption, which may occur as separations that do not become divorces or as disruptions of households with unmarried parents (Bramlett and Mosher, 2002). It is estimated that 34 percent of children in the United States will experience parental divorce before reaching age 16 (Bumpass and Lu, 2000). Children can experience a wide range of other stressors following divorce, such as loss of time with one or more parents, continuing interparental conflict, and parental depression (Amato, 2000). Evidence suggests that effective child coping or interpretation of these stressors, quality of parenting received from both parents, and level of interparental conflict is related to postdivorce adjustment (e.g., Kelly and Emery, 2003; Sandler, Tein, et al., 2000).

Death of a parent (i.e., parental bereavement) occurs to 3.5 percent of U.S. children before age 18 (U.S. Social Security Administration, 2000). The effect of parental death on surviving children rises to national concern particularly when rates increase due to such national disasters as the terrorist attacks of September 11, 2001, war, and such epidemics as HIV.

Following parental divorce, children are at increased risk for multiple mental, emotional, and behavioral problems, including physical health problems, elevated levels of alcohol and drug use, premarital childbearing, receiving mental health services, and dropping out of school (Troxel and Matthews, 2004; Furstenberg and Teitler, 1994; Hoffmann and Johnson, 1998; Goldscheider and Goldscheider, 1993; Hetherington, 1999). Meta-analyses of studies conducted through the 1990s have shown that problems have not decreased (Amato and Keith, 1991a; Amato, 2001). McLanahan’s (1999) analysis of 10 national probability samples revealed school dropout rates of 31 percent and teen birth rates of 33 percent for adolescents in divorced families versus 13 and 11 percent, respectively, for adolescents in nondivorced families. Adults who were exposed to parental divorce as children have been found to be more likely to divorce and to have an increased risk for mental, emotional, and behavioral problems, including clinical levels of mental health problems, substance abuse, and mental health service use (Chase-Lansdale, Cherlin, and Kiernan, 1995; Kessler, Davis, and Kindler, 1997; Maekikyroe, Sauvola, et al., 1998; Rodgers, Power, and Hope, 1997; Zill, Morrison, and Coiro, 1993; Amato, 1996).

Children who experience parental bereavement appear more likely to experience mental, emotional, and behavioral problems, such as depression, posttraumatic stress disorder, and elevated mental health problems for up to two years following the death (Worden and Silverman, 1996; Geresten, Beals, and Kallgren, 1991). These risks appear to remain after controlling for other risk factors, such as mental disorder of the deceased parent (Melhem, Walker, et al., 2008). Research has shown mixed findings concerning the mental, emotional, and behavioral problems of bereaved children when they reach adulthood (Kessler, Davis, and Kindler, 1997). However, two prospective longitudinal studies supported increased risk of depression in adult women who experienced parental bereavement as children (Reinherz, Giaconia, et al., 1999; Maier and Lachman, 2000).

Although family disruption is associated with multiple MEB disorders and problems, the majority of children who experience these major stress-ors adapt well. The most consistent predictive factors are interparental conflict and the quality of parenting by both the mother and the father (Kelly and Emery, 2003; Amato and Keith, 1991b). Parent–child relations that are characterized by warmth, positive communication and supportiveness, and high levels of consistent and appropriate discipline have consistently been related to better outcomes following divorce (Kelly and Emery, 2003; Amato and Keith, 1991b). High-quality parenting from both the custodial parent (usually the mother) and the noncustodial parent (usually the father) is related to lower levels of child internalizing and externalizing problems (King and Sobolewski, 2006). But interparental conflict is one of the most damaging stressors for children from divorced families. Conflict often precedes the divorce and is associated with lasting child problems following the divorce (Block, Block, and Gjerde, 1988). In some families, conflict continues long after divorce, which is particularly destructive when children are caught in the middle (Buchanan, Maccoby, and Dornbusch, 1991). Recent research has found that high-quality parenting from both parents related to lower child mental health problems even in the presence of high interparental conflict (Sandler, Miles, et al., 2008).

Several factors have been found to influence outcomes for children who experience parental bereavement. Among parentally bereaved children who had signed up for an intervention program, four factors distinguished bereaved children who had clinical levels of mental health problems from those who did not: positive parenting by the surviving caregiver, lower mental health problems of the surviving parent, the coping efficacy of the child, and children’s appraisals of how much recent stressful events threatened their well-being (Lin, Sandler, et al., 2004). Other factors, such as coping efficacy, control beliefs, postbereavement stressful events, and children’s fears that they will be abandoned by the surviving caregiver, have been associated with mental health outcomes for bereaved children (Wolchik, Tein, et al., 2006).

An interesting focus of research has investigated the pathways that lead from family disruption due to divorce or bereavement, along with other commonly co-occurring biological and social risk factors, to adult depression. One analysis of longitudinal data on female twins, siblings, and unrelated women found support for three pathways to the development of depression (Kendler, Gardner, and Prescott, 2002). In an internalizing pathway, genetic risk leads to neuroticism, which in turn leads to early-onset anxiety disorder, and these three influences each lead to episodes of major depression. In an externalizing pathway, conduct disorder and substance misuse lead to depressive disorder. In an adversity pathway, early childhood exposure to a disturbed family environment, childhood sexual abuse, and parental loss lead to low educational attainment, lifetime trauma, and low social support, which in turn lead to four adult risk factors (marital problems, difficulties in the past year, dependent stressful events, and independent stressful events), which in turn lead to an episode of major depression. All three pathways include contributions from genetic factors and interconnections among family adversity, externalizing problems, and later adult adversities.

A prospective longitudinal study, the National Collaborative Perinatal project, also considered timing in an examination of the association between family disruption (divorce or separation before age 7), low socioeconomic status, and residential instability and the onset of adult depression (Gilman, Kawachi, et al., 2003). The effect of low socioeconomic status in childhood on depression risk persisted into adulthood, but the effects of family disruption and residential instability were specific to early-onset depression. Early-onset depression is of special concern because it carries with it a poorer prognosis of increased recurrence and, in some studies, more severe depressions.

Community and School Risk Factors

Most prevention research has focused on risk and protective factors at the level of the individual and the family, but there is increasing recognition that child development is powerfully affected by the broader social contexts of schools and communities (Boyce, Frank, et al., 1998). Risk factors, such as victimization, bullying, academic failure, association with deviant peers, norms and laws favoring antisocial behavior, violence, and substance use, are linked primarily with neighborhoods and schools. For example, poor and ethnic minority children in particular are frequently exposed to violence in their neighborhoods and schools. Among 900 low-income, primarily minority adolescents in New York City in 2002–2003, rates of exposure to violence of various kinds were high: someone offering or using drugs (70 percent), someone beaten or mugged (51 percent), someone being stabbed (17 percent), someone being shot at (14 percent), and someone being killed (12 percent) (Gershoff, Pedersen, et al., 2004). Many also reported being the victim of violent acts, such as being asked to sell or use drugs (35 percent), having their home broken into (18 percent), being beaten up (13 percent), and being threatened with death (9 percent). Much of the exposure to violence occurs either at school or on the way to school (DeVoe, Peter, et al., 2003; Bell and Jenkins, 1991; Richters and Martinez, 1993; Gershoff, Aber, and Raver, 2003).

Exposure to violence is associated with children’s development of various mental health problems, particularly posttraumatic stress disorder, anxiety, depression, antisocial behavior, and substance use (Jenkins and Bell, 1994; Gorman-Smith and Tolan, 1998). A reciprocal relation exists between academic achievement and mental health outcomes, in which mental health problems adversely affect academic achievement (Adelman and Taylor, 2000), and poor academic achievement is related to the development of multiple problem behaviors (e.g., substance abuse, antisocial behavior) as well as teenage pregnancy and low occupational attainment (Dryfoos, 1990).

The growing empirical research on characteristics of neighborhoods and schools that are linked with problem development as well as positive youth development has implications for the development and evaluation of prevention and promotion interventions. Gershoff, Aber, and Raver (2003) propose that another dimension of schools and neighborhoods that may affect the development of child mental, emotional, and behavioral problems is the degree to which they provide settings that support healthy development. They characterize neighborhood disadvantage as the absence of settings that provide opportunities for healthy child development—settings for learning (e.g., libraries), social and recreational activities (e.g., parks), child care, quality schools, health care services, and employment opportunities. For schools, disadvantage can be assessed as lower per student spending, a high percentage of children from families in poverty, a higher number of inexperienced and academically unprepared teachers, a high student-to-teacher ratio, and school size being either too large or too small. Each of these characteristics of neighborhoods and schools has been linked with mental, emotional, and behavioral problems of children. Although it is difficult to disentangle the causal effects of neighborhood and school disadvantage from the effects of factors in families and children who live in disadvantaged neighborhoods, research has found that neighborhood disadvantage was associated with higher internalizing and externalizing problems over and above the genetic contribution (Caspi, Taylor, et al., 2000) and that an experimental study found that children whose families were moved from a disadvantaged neighborhood had a lower rate of arrest for a violent crime than those who remained in a high-poverty neighborhood (Leventhal and Brooks-Gunn, 2003).

Similarly, the strongest environmental association related to schizophrenia is urbanicity (Krabbendam and van Os, 2005), although the relation with social class is also strong. It appears that living in urban environments during childhood affects later development of schizophrenia, even if there is a move to less urban environments later in life (Pederson and Mortensen, 2001). This relationship is therefore not fully explained by the “drift” hypothesis, in which those who are developing schizophrenia move to urban settings. There are a few hypotheses that are being pursued to explain this relationship, including increased stress and discrimination against minorities, lack of social capital and other resources in impoverished communities, and gene–environment interactions.

Another way in which the community influences child development is through the norms, values, and beliefs of the residents. For example, collective efficacy, a concept developed by Sampson, Raudenbush, and Earls (1997), refers to “shared beliefs in a neighborhood’s conjoint capability for action to achieve an intended effect, and hence an active sense of engagement on the part of residents.” It provides the informal social controls that counteract antisocial behavior and has been found to be related to levels of community violence (Sampson, 2001). Peer norms favoring the use of drugs, antisocial behavior, or belonging to gangs are also powerful neighborhood factors that contribute to problem behaviors.

Hawkins and Catalano (1992) proposed the construct of bonding to school, community, and family as key in explaining the development of substance use and antisocial behavior. Positive bonds consist of a positive relationship, commitment, and belief about what is healthy and ethical behavior. Positive bonds to a group develop from having the opportunity to be an active contributor, having the skills to be successful, and receiving recognition and reinforcement for their behavior.

In school, students’ relationships with their peers and teachers and the social climate in the classroom have a powerful effect on their development of mental, emotional, and behavioral problems as well as their development of age-appropriate competencies. For example, aggregate-level student-perceived norms favoring substance use, violence, or academic achievement are related to antisocial behavior. For boys with elevated levels of externalizing problems, being in a first grade classroom with high aggregate levels of behavior problems has been found to be associated with a marked increase in the odds of having serious externalizing problems when they reached the sixth grade (Kellam, Ling, et al., 1998). But some teacher characteristics are related to lower levels of mental, emotional, and behavioral problems for students. These include using classroom management strategies with a low level of aggressive behavior, having high expectations for students, and having supportive relations with students.

Programs promoting classroom and school procedures that encourage prosocial behavior, academic achievement, or increased positive bonding to school have important implications for children’s healthy development. For example, use of a group contingency to promote prosocial behavior in first grade students has been found to reduce aggressive behavior in first grade (Dolan, Kellam, et al., 1993) and through middle school (Muthén, Brown, et al., 2002). The effects persisted with a reduction 13 years later in the rate of diagnosis of alcohol and illicit drug abuse or dependence (Kellam, Brown, et al., 2008). Also, for the subgroup of boys who started first grade with high levels of aggressive behavior, this intervention reduced the rate of antisocial personality disorder (Petras, Kellam, et al., 2008) and mental health service use (Poduska, Kellam, et al., 2008).

Structural and policy changes can reduce risk associated with the transition to senior high school (Seidman, Aber, and French, 2004). This transition is associated with a decline in academic performance as well as an increase in delinquency, depression, suicidal thoughts, and substance use. However, policy changes, such as reduced school size, that create smaller working units with more supportive relations with teachers and peers have been shown to reduce this risk (Felner, Brand, et al., 1993).