Daughters do not cause divorce

Young girl stuck in the middle of arguing parents.

(Inside Science) -- Good news for families: fewer fathers than previously thought may be divorcing wives because they give birth only to daughters.

Back in 2003, an economist named Gordon B. Dahl described a trend hidden in divorce data. He found that parents with a first-born girl are five percent more likely to divorce than a couple with a first-born boy.

“It’s a well documented pattern, if relatively small,” said Florencia Torche, an associate professor of sociology at New York University in New York City, who was not involved in the study. “But it’s significant. It’s both statistically real and socially important,” she continued.

For the last decade or so, sociologists have pondered the question: do daughters cause divorce? A number of theories have come and gone. Some have hypothesized that women feel more comfortable raising girls on their own, so are less likely to stick around in an unhappy relationship if they only have daughters. But the prevailing theory has been that men probably prefer sons and will instinctively work harder at maintaining the family unit for a boy. 

New research published in the journal Demography this month turns such previously held conjectures on their head.

“Daughters may be a product of stressful relationships rather than the other way around,” said study co-author Amar Hamoudi from Duke University in Durham, North Carolina. 

Stress, such as that caused by arguments within an unstable relationship, has biological implications with hormonal reactions. Hamoudi said that biological data, along with data he collected from a computer-generated model, suggests female fetuses may be hardier than their male counterparts.

In other words, the female survival advantage could already be at play inside the womb. This would mean that a female fetus is less likely to miscarry from the stress that a mother may experience from a pre-existing unhappy relationship.

“Something like half of fertilizations don’t make it to live birth, a large fraction of which is lost in the first few weeks,” explained Hamoudi. “When biologists look at early pregnancies, they observe 120 male embryos for every 100 girls. But at birth we see only 105 for every 100.”

This research reinvigorates the debate, said Torche, because it adds a new suggested cause to the pattern rather than merely confirming its existence. She said it’s not surprising that it has taken so many years to come up with a biological cause because it’s a daunting task to sift through all the potential interacting factors.

“The issue that researchers face in general is that they don’t observe all the variables,” Torche said.

Hamoudi generated a simulation using data points on a number of variables including divorce rates and the gender ratio of the children. He then conducted a sensitivity analysis, which is one way of examining correlations and causality; it runs the simulation many times over. Each time, the variables are altered within what’s deemed to be a reasonable range, so each of the parameters Hamoudi factored into his model were considered at their extremes.

The Duke researchers, which included Hamoudi and his co-author Jenna Nobles, a sociologist at the University of Wisconsin-Madison, found the association involving divorce and the discrepancy between gender ratios at the embryonic stages and birth stuck throughout the sensitivity analysis. 

“This increases my confidence in their findings,” said Torche. “It seems very likely that what they’ve found is real and accounts for a significant proportion of the association. I would say that it could contribute to about half of the divorce rate, perhaps even all of it.”

Hamoudi is quick to point out that he hasn’t technically disproved any previous theories, be it a father's sexism or a mother’s apprehension of raising a son alone.

“We’ve shown that there’s a plausible alternative to a very provocative pattern. It’s important to remember that our conclusions are only as good as the data and the data on early pregnancy is surprisingly sparse.”

While Torche may agree that Hamoudi’s results are statistically significant and that the association is not sheer chance, it equally cannot be proven that the correlation isn’t simply caused by another factor that Hamoudi failed to include in his simulation. It’s near impossible to say what causes the pattern without leaving any room for doubt.

“This can’t be claimed to be an absolute truth, but I think it’s very likely,” said Torche

Reprinted with permission from Inside Science, an editorially independent news product of the American Institute of Physics, a nonprofit organization dedicated to advancing, promoting and serving the physical sciences.