Desktop version

Home arrow Psychology arrow Parenting and theory of mind


Determining Causality

Correlation and causality.

One of the truisms that psychology students learn early in their studies—and then often relearn and relearn—is that correlation does not imply causality. There is probably no topic for which it is more important to keep this principle in mind than the topic of parenting. The great majority of parenting studies are correlational. They are correlational because the independent variable of interest, namely parental behavior, is not experimentally manipulated or controlled; rather it is simply measured. The outcome of interest, child behavior, is also measured, and the analyses then look for a correlation between the two sets of scores. A positive correlation is compatible with the hypothesis that the parental behavior contributes to the child outcome. The problem, however—and the reason that such a causal conclusion cannot be drawn with certainty—is that a positive correlation is also compatible with three other explanations.

One is that the causal direction is the reverse—thus not from parent to child but from child to parent. Suppose we find, for example (as research in fact does find), that relatively high use of physical punishment by parents correlates with relatively high levels of aggression in children. It may be that physical punishment promotes aggression (perhaps, for example, by providing a model of aggressive behavior). But it may also be that aggressive children elicit physical punishment from their parents. Such children may in general be difficult to control, and the parent may have tried and abandoned milder disciplinary strategies before resorting to the use of physical punishment.

The parent-to-child and child-to-parent models each posit a single causal direction. A third possibility is that the causality runs in both directions—that over time the parent’s behavior affects the child and the child’s behavior affects the parent in a back-and-forth, reciprocal fashion. Thus physical punishment may indeed promote aggression; the high level of aggression, however, then provokes still more punishment, which in turn leads to still more aggression, and so on. Many researchers believe that such a “transactional” model (Sameroff, 2009) may apply to the majority of parent-child correlations.

The three possibilities considered to this point address what is known as the “directionality issue”: We assume that there is a causal relation between parent and child; the question is in what direction the causality runs. A final possibility, generally labeled the “third factor issue,” is that there is no causal relation at all between parental practice and child outcome; both are caused by some third factor or set of factors. In the case of physical punishment and aggression, for example, it may be that parents who are high in the use of punishment are also high in whatever the genes are that underlie aggression; they pass these genes on to their children, and this is the reason that the children are high in aggression.

Although shared genes are not the only possible third-tactor explanation for parent-child correlations, they may be the most generally applicable one. Parents contribute only some of the child’s environment, but they contribute all of the child’s genes. A major argument in J. R. Harris’s book (1998) The Nurture Assumption is that most parent-child correlations from childrearing research are better interpreted as genetic than as environmental in origin.

Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >

Related topics