Savage Spawn by Jonathan Kellerman Read Free Book Online Page B

aggression, violence, and/or criminality and attempting to relate those characteristics to various measurements of early childhood disruption. Numerous studies have been carried out on the relationship between criminality and abuse, divorce and marital separation, and the intrusion of violence into the child’s life, either directly or through the media.
    Family breakdown, as exemplified by divorce and separation, is clearly related to a host of psychological problems in children, including school problems, truancy, alcoholism, and drug use, all of which are often precursors of criminal behavior. Childhood aggression seems to predict alcoholism and drug use in adolescence; in turn, substance abuse seems to predict adult criminality, especially when combined with parental alcoholism and drug use (5, 6, 31–33).
    Seems
to, because statistics are valid only for groups and are mathematically irrelevant when making predictions about individuals. Statistics can also be monkeyed with easily by experts in order to support whatever conclusions the researcher wants them to buttress.
    One example of this is the misuse of a statistical concept commonly bandied about in the popular press:
correlation,
which is a mathematical expression of coexistence between two or more variables.
Correlation is not, by itself, causation.
Correlations simply state probability associations. A positive correlation means that when X is present, Y is more likely to be present. A negative correlation means the presence of X is more probable when Y is absent.
    Correlations
can
be causal—smoking cigarettes is both correlated with and a cause of lung cancer. Putting a gun to one’s head and pulling the trigger is highly correlated with, and causally linked to, infliction of a fatal wound.
    But consider the correlation between blond hair and blue eyes. Both traits coexist at higher-than-chance rates, but neither trait brings about the other. Similarly, itching and wheezing resulting from exposure to an allergen are concomitant symptoms with a shared cause, but they lack a causal connection.
    Another poorly understood statistic is the
power
of a correlation—how much variability within a group the correlation actually explains. Mathematically, this is obtained by squaring the correlation. By point of example, let’s say we study the relationship between freckles and red hair in a group of people and come up with a correlation of +.60. This means that red-haired people tend to be freckled and vice versa (without saying anything about causality). However, there will also be plenty of red-haired people without freckles as well as freckled people without red hair (statisticians call this “scatter”). Squaring .60 (.60 X .60), we obtain .36, or 36 percent, which means that slightly over a third of the group can be classified as red-haired and freckled. That may still be important, but it’s a long way from the majority status implied by the .60 correlation itself. Failure to understand the correlation-squared index is one reason nonscientists are often overly impressed by scientific data.
    Given all these caveats, let’s look at more of the data on environment and violence, criminality, and psychopathy.
    Several studies of violent children, most of them based on small samples, indicate an extremely high degree of family chaos and child abuse in the backgrounds of kids who kill (9–13). Paternal absence is common, as are disproportionately high rates of fathers with a history of imprisonment and/or drug and alcohol abuse and/or psychopathy.
    Fathers with a
specific
constellation of personality traits—high reactivity to stress, alienation, and aggression—seem to keep cropping up as progenitors of violent kids (32). The importance of paternal input makes sense when we remember that boys learn how to handle aggression from their fathers—or, when fathers are absent, from surrogate males, mainly peers, such as fellow