Looking at Divorce, III

If my parents divorce, does the separation make me more or less likely to trust Mom? How about Dad? People in general? On these questions, the current weight of clinical and social science evidence is clear: More divorce, less trust. When the very thing that I most want and need to trust - my parents’ love for one another and, by extension, for me - is compromised or taken away, I will find it harder, not easier, to trust other people and the world around me.

But a recent study from Valerie King and Alan Booth of Pennsylvania State University seeks to challenge this idea. These scholars want us to know, as the USA Today headline about their study recently put it, that “Children’s trust in adults survives divorce.” According to the study, children of divorce “are no worse off in terms of trust than kids from intact families.” So what gives? Here’s their main point: “It is not the divorce itself, but the relationship with parents after the split that is the key to trust in adult life, says Valerie King . . . The message, King says, is that while parental divorce doesn’t affect a child’s ability to develop intimate relationships as an adult, his relationship with his parent does.”

My curiosity aroused, I got and read the study itself. Here is how the authors summarize their findings, based on telephone interviews with parents and their adult offspring: “although parental divorce and divorce proneness are negatively associated with trusting parents in adulthood, these affects largely disappear once the quality of the early parent-child relationship is controlled. Trust of parents, intimates, and others is strongly linked to positive early parent-child relations regardless of divorce. The one exception is the remaining pervasive negative effect for trusting dads. Similarly, close relationships with parents mediate the negative effect of early parental divorce on trusting others and some of the negative effects of divorce proneness on trusting intimates. Thus, contrary to current suggestions that parental divorce may set the stage for inhibiting the development of trust among offspring, especially in regard to their own intimate relationships and more generally to trusting others, our findings reveal little support for a strong negative role of parental divorce if parents can maintain good relationships with their children.” 

These are remarkable sentences, deserving of careful translation into plain English. First, let’s consider the short sentence buried in the middle: “The one exception is the remaining pervasive negative effect for trusting dads.” What this means is that children of divorce, as a direct result of the divorce itself, tend not to trust their fathers. Indeed, this is the study’s single clearest and arguably most important finding. To its credit, the USA Today story did report that, for children of divorce, trust in fathers “falters over time.” But if King and Booth themselves had chosen to underscore this finding, instead of rhetorically demoting it to something approaching a footnote, merely the “one exception” to their main argument, the USA Today headline would likely, and more accurately, have announced: “Children of divorce mistrust their fathers.”

But what about that main argument? Looking at the rest of their summary, the tell-tale phrase is that the problems they discovered “largely disappear” when “the quality of the early parent-child relationship is controlled.” What can this mean? It does not mean that the actual problems experienced by the children of divorce somehow “disappear.” The study clearly shows, and the summary tersely admits, that children of divorce do suffer from a measurable loss of trust in others. 

No, what “disappears” are certain coefficients in the authors’ statistical tables. They “disappear” because King and Booth contrafactually assume - simply insert numbers that suggest, without any evidentiary basis - that parent-child relationships in divorced families are just as strong as parent-child relationships in intact families. That’s called “controlling” for the “quality” of parent-child ties. By playing this game of make-believe, the authors can statistically cancel out (make “disappear”) the previously measured effects of parental divorce on children’s trust in others. If you are competent with numbers, it’s an easy trick.

It’s also a corrupting trick. Consider an analogy. The negative effects of flooding largely disappear once the impact of water damage is controlled. Here’s another. The negative effects of father absence largely disappear once family income and male role models are controlled. (Unfortunately, I’m not making this last one up; it’s common enough in the academic literature on fatherhood.) But in real life, of course, flooding tends to cause water damage. Father absence tends to cause lower income and the loss of a male role model. And parental divorce tends to cause deteriorations in parent-child relationships. In real life, these consequences cannot be “controlled” and therefore their effects on people do not “disappear.” 

When done properly, “controlling” for particular variables can help scholars to estimate with some precision the different ways in which a phenomenon expresses itself. Here are examples of how estimates based on such controls might properly be expressed. Flooding your home is typically harmful to the furniture, in part through the measurable mechanism of water damage. Father absence is typically harmful for children, in part through the measurable mechanisms of lowering family income and removing a male model from the home. And parental divorce usually makes children less trusting, in part through the measurable mechanism of worsened parent-child relationships. What is intellectually fraudulent is for King and Booth to suggest that identifying one measurable mechanism through which a problem is expressed somehow makes the problem itself  “disappear.” As a matter of logic, this is a simple point. How can King and Booth, and so many other people with Ph.D.’s, fail to grasp it?

If we strip away the fog and nonsense, here is what King and Booth found. Children of divorce are less trusting of others, in large part because divorce seems typically to lead to worsened parent-child relationships. In those atypical cases in which parent-child relationships after divorce remain as strong, or almost as strong, as parent-child relationships in intact families, children seem to be as trusting, or almost as trusting, as children in intact families with respect to mothers and to others outside the family. With respect to fathers, the experience of divorce appears to be so influential that even a strong father-child bond after divorce is not enough to prevent children of divorce from losing trust in their fathers.

If King and Booth had said it this way, as opposed to the fog-and-nonsense way that they chose, I don’t know what the USA Today headline would have declared. But it obviously would not have declared that “Children’s trust in adults survives divorce.” Meanwhile, for those intent on showing that divorce does not harm children, and willing to overlook the corruption of social science methodology, this study is a godsend.

Sources: Karen S. Peterson, “Children’s trust in adults survives divorce,” USA Today, April 25, 2000. Valerie King and Alan Booth, “Family Instability and Interpersonal Trust,” paper for the annual meeting of the Population Association of America (Los Angeles, CA: March 2000).

First published Winter 2001.