behavior, these studies provide the most satisfactory way of quantifying the strength of environmental effects. Other types of studies that have provided useful information include adoption studies; studies of various classes of relatives, using path models; and experimental studies, including experiments of nature, in which individuals are exposed to changing environments.
As Plomin and Bergman (1991) pointed out, the fact that a variable is labeled as “environmental” does not mean that its effects are environmentally mediated; many supposedly “environmental” measures actually index genetic, as well as environmental, influences. This point is best seen with the many risk factors that are based on someone's personal characteristics or behavior, such as low intelligence, aggressiveness, and behavioral inhibition. These involve a genetic component as well, and it is necessary to determine whether the psychiatric risk reflects genetic or environmental mediation.
Individual differences in environmental effects. One of the most important recent findings from behavioral genetics research with twins has been the demonstration that, for many aspects of normal and abnormal psychosocial development, nonshared environmental effects tend to be substantially more important than shared ones in quantifying variability in outcomes (Plomin and Daniels, 1987). Shared risk factors that apply to the family as a whole, such as family discord or poverty, are indeed relevant, but they are likely to impinge on different children in the same family to varying degrees or in different ways, that is, to have nonshared effects. If we are to understand how such environmental risk factors operate, we must investigate the processes on a person-specific basis and not simply assume a uniform impact.
Relative differences between siblings in how they are treated in a home may be more important than the absolute level of this treatment in the home (Dunn and Plomin, 1990). That is, what appears to be a shared risk factor within the family may actually be an unshared risk factor, with its concomitant unshared effect, for, perhaps, only one child in the family. For example, it may be that it is less important whether parents respond to their children in a generally warm, or strict, or harsh fashion than whether one child in the family is consistently dealt with less warmly, or more strictly, or more harshly than his or her brothers and sisters. In other words, scapegoating or favoritism may be the operative risk factor (Boer and Dunn, 1992).
Attention needs to be paid to shared and unshared risk and protective factors and shared and unshared effects outside, as well as inside, the family. The body of evidence that school influences have effects on children's behavior and scholastic attainments (Maughan, in press;