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