Bryan Caplan  

Human Sex Ratio Doesn't Run in Families

Balance of Payments... What If...
I have three children, all sons.  On a gut level, I feel like there's a systematic factor at work.  It must be my overpowering machismo, right?  But it looks like my gut is dead wrong.  At least in the NLSY, sex ratio doesn't run in families.  If you simply look the frequencies of BB, BG, GB, and GG...
For the first two children, families heterogeneous by sex were more likely than expected in an independence constant probability model.
If you estimate a full model with a gender-biased stopping rule (i.e., if you get GG, try again), the pure chance model checks out:
In this model-fitting exercise, there was a straightforward and unambiguous answer to the sex composition question when parameters for parental stopping behavior separated that process from the sex probabilities. The answer to the question, "Does having boys or girls run in the family," was clearly "No."
Bottom line: If you want girl, marrying someone from a family with a low male/female ratio won't help you.  Try sperm sortation instead.

Comments and Sharing

COMMENTS (8 to date)
Bob Murphy writes:

I'm really surprised Bryan that you of all people would have had a "gut feeling" about something for which there is a 25% chance (i.e. 3 kids of same sex).

I still suspect it's your parenting style that made them turn out all boys.

RL writes:

"Bottom line: If you want girl, marrying someone from a family with a low male/female ratio won't help you"

Damn! And I let the otherwise perfect girl go out of my life solely based on this now-I-see fallacious reasoning!! (I recall our parting: "It's not you, darling; it's your family's high male/female ratio...")

David C writes:

In regards to sperm sortation, how does anybody prevent the same problems that are currently happening with adoption? Won't this cause a large increase in the female population in the US, possibly resulting in widespread polygamy? Oh wait, I'm a guy. I'm supposed to be in favor of this.

Problem 2:
What about the 9% of Johns who should have been Janes? What about the 24% of Janes who should have been Johns? When parents normally roll the dice, it's 50/50 and nobody needs to know. But with this procedure, there's a record out there, and the child will have access to it. How does any parent go about telling their kid that they were a mistake?

Tracy W writes:

David C - don't many parents now have kids that were a mistake? Lots of people around nowdays were born because of a contraception failure. Before modern contraception, lots of people were conceived before the wedding, or without a wedding, which was, at least in the West, a mistake back then.

Or take the case where one person is widowed and then remarries and has kids with the new spouse, what does that imply to the kids about their existance? (Assume that the widower did not kill their late spouse in self-defence).

paul writes:

"it takes a man to make a man" ... ; )

David C writes:

Tracy W,
A kid doesn't necessarily need to know about a contraception failure. But I'll grant that there are similar circumstances already around. My question is why increase the number of people in said circustances.

caveat bettor writes:

Bryan, I only skimmed the article at the link you provided, but it seems to me that the conclusions are drawn ACROSS the significant sample, not WITHIN FAMILIES. Let me know if that is incorrect.

elliott sober writes:

Your article reports the best fitting model of the class of models you considered. I wonder what the results would be if you compared models that differed in their number of adjustable parameters, and didn't just look at the best fitting model (which is bound to be pretty complex), and instead used a model evaluation criterion like AIC (the Akaike information criterion). What would the bottom line be then?


Comments for this entry have been closed
Return to top