A monocausal theory of the aggregate divorce rate?

The divorce rate is almost perfectly predictable from the proportion of men to women in the population. In 1920, when the annual divorce rate was eight per one thousand married women, there were 104 males for every female. By 1980 the tables had turned, and there were only ninety-five males for every one hundred females, and the divorce rate had risen to twenty-three per one thousand married women. The sex ratio has remained virtually unchanged for the past thirty years and is paralleled by a steadily high divorce rate. The correlation betwee the population sex ratio (or number of males per one hundred females) and the divorce rate at four-year intervals between 1896 and 1992 was -.91, indicating that changes in the number of men relative to women accounts for 83 percent of the changes in divorce rate.

From Nigel Barber’s The Science of Romance who, on this claim, cites his own Why Parents Matter.

Can this be true? If it is such a neat fact, why have I never heard it before?

A good bit of web searching yielded surprisingly little enlightenment. One Amazon reviewer writes:

On page 150, Barber refers to a 1983 book purporting that the increasing divorce rate was due to a shortage of marriagable men. That was true in 1983. Barber fails to note that by 1987 the marriageable male/female ratio reversed, and we’re now in a “women shortage” era.

Point well taken, we should consider the sex ratio of marriageable people, not the overall sex ratio. More significantly, the general rate of American divorce is rising over time. The sex ratio has been moving in favor of more women as well, at least until recently, perhaps because survival out of childhood depends less on parental discretionary investments, noting that many parents prefer boys to girls. So the correlation may be spurious, rather than representing causality.

The fact is intriguing, but so far it is not a good monocausal theory of the divorce rate.