One of the researchers I most admire is the American sociologist Paul Amato. Over the years he has written some of the most interesting papers on divorce, its predictors, and its effects.

It was his analysis of the “Marital Instability Over the Life Course” study—which followed a sample of hundreds of parents and their children for twenty years—that showed how it’s not the end of high-conflict marriages that injures children but the end of low-conflict ones. (Half of U.S. divorces are low-conflict.)

“Mommy and daddy don’t love each other any more” may make sense to the parents. But it’s wholly unintelligible to a child who doesn’t see it coming.

Amato’s latest study—published recently in the Journal of Marriage and Family—looks at two huge data sets that each followed some ten thousand children with married parents, one through elementary school, and the other from eighth grade through high school.

Some of these children experienced the divorce of their parents along the way. A much smaller group experienced the death of one or both parents.

In this study, Amato and Christopher Anthony set out to explore how parental divorce and death influenced the children as they grew up.

Instead of simply comparing children with stably married parents to children whose parents had divorced or died during the observation period, Amato and Anthony used each individual child as his/her own control.

The statistical technique they used is called “fixed effects.” Suffice it to say, this method accounts not just for the obvious individual-level factors that can cause both parental divorce and children’s problems—like poverty or ill health—but also for any other individual-level influences that are hard or impossible to measure (e.g., personality).

In other words, among kids who experience parental divorce or death, a fixed effects model shows whether the event led to a change in the individual child’s outcomes relative to his or her own expected outcomes, not just compared to children who don’t experience parental death or divorce.

So what you get at the end is as close as social science can come to the pure residual influence of divorce or death on its own, above and beyond any other influences.

The main finding from the study is that both divorce and death consistently influence children in a negative way, although the overall effect size in each case is quite small.

Specifically, following parental divorce, children tend to become worse at reading, at maths, in their approach to learning, in their interpersonal skills, in their self-control, and in the way they internalise and externalise problems.

Following parental death, only maths and internalising tend to deteriorate. The evidence for a causal effect—in both cases—is strong.

Across all of these findings, children’s most common response to the divorce itself was to show no change.

The reason why the overall effect is negative is that the proportion of children who did worse after divorce was slightly higher than the proportion of children who did better.

For example, internalised problems increased for 24 percent of children and reduced for 10 percent; 22 percent of children got worse at maths whereas 16 percent got better.

Interestingly, it’s not just divorce and death that have this effect. There are similar negative effects found in other studies where a parent leaves the household for other reasons—such as incarceration, migrating overseas for work, or being deployed overseas in the military.

It might be tempting to see all this as no big deal. Most kids cope. Some do better even if more do worse. Whether parental conflict was high or low before the divorce undoubtedly plays a role in this.

But there’s one further important variation. And this is new. Across all of these findings—maths, reading, internalising or externalising problems—divorce had an especially marked negative effect when the parents were more at risk of divorce in the first place, whether due to their education level, ethnicity, or socioeconomic status.

The children who do worst after divorce have parents who are least able to buffer its effects.

This might help explain why politicians and policy-makers tend to hold a relaxed view of family breakdown.

Maybe they look at their own experience and their network of friends and can’t see the problem. But in families with fewer resources, when parents are less well-equipped to buffer its effects, divorce remains devastating for children.

Harry Benson is Research Director of the UK-based Marriage Foundation. This article was first published on the Institute for Family Studies blog.