Monday, April 4, 2011

Pink or Blue? Does it Matter? I say No!

The organization published a press release today - New research: Moms keep going until they have a girl!  You can read it at GeneaPress.

The Blog has more information on their site in Global Research: Pink or blue – does a baby’s gender matter?  This article has the statistics from which the press release draws conclusions.

Based on the statistics provided, the main conclusion was:

"The most profound statistic was that families with 2 boys were almost 23% more likely to have another child than families with a boy and a girl, indicating a strong will amongst parents to produce a girl." 

That was probably concluded based on the worldwide statistic for families with "Male-Male" children being 27.9% of the families with three or more children, divided by the average 22.65% for families with "Female-Male" and "Male-Female."  Of course, using that methodology, the statistics show that families with "Female-Female" children that have a third child are 19% more likely to have a third child also.

The number of worldwide families surveyed with the first two children being "male-male" is 172,048, and the number of families with the first two children being "female-female" is 165,851.  The number of families with a male and a female (in no particular order) are 279,255.

I think the analysis and the conclusion are flawed, for these reasons: 

*  The survey was taken using family trees submitted by members, and was not obtained from vital statistics provided by countries around the world. 
*  The members, and their family tree entries, may not match the demographics in age, ethnicity, or income level that is representative of the world.
*  The families surveyed already had at least three children, not two children.  
*  We don't know how many families in the family tree system had only one, or only two children.  We really need a measurement that includes families that had only two children - say how many "male-male" and how many "female-female," so we can determine what percentage actually had another child.  We don't know how many "tried" or "didn't try" to have another child after two children. 
*  The survey period of 1990 to 2010 seems too recent for me.  I would have selected 1980 to 2000, rather than 1990 to 2010, in order to permit families to "finish" their child-bearing and complete their families.

On the other hand, we can say that, using the statistics comparing families with at least three children and at least four children, that:

*  There were 617,154 families in the study with three or more children, and 135,257 families with four or more children, so only 21.9% of the families in the study with three or more children had at least one more child.
*  The 165,851 families that started out with "female-female" produced 35,364 children (21.3%), and had a third female 53.9% of the time
*  The 172,048 families that started out with "male-male" produced 36,797 children (21.4%), and had a third male 56.2% of the time.
*  The 279,255 families with a "male-female" or a "female-male" produced 63,096 children (22.6%).
*  Of the third children, 67,912 were male, and 67,345 were female.  That's a 1% advantage of male births over females.
*  Look at that - the percentages of families having a third child are actually a bit higher for families with a "female-male" or "male-female" in the first two children.  That's a completely opposite conclusion from the study.  And, I think, it is a fairly drawn conclusion.

My conclusion is that the statistics are interesting, but that the study is flawed, and that the conclusions, while they may seem right to readers, cannot be fairly drawn without additional data.  My guess is that genetics plays a part in all of this.

Nice try, MyHeritage.  No cigar from Genea-Musings, who loves problems like this!


Celia said...

- no cigar at all! Thanks for your detailed clear discussion of the statistics/analysis. You're always so easy to read, Randy - a pleasure!

Geolover said...

Your analysis is great, but there's an additional factor. The treebies whose trees were surveyed may not have known of children who died young, so total number of families having produced X number of children has an additional question. It would be numerically small, given the chose time frame, but still . . . .

Cindy Bergeron Scherwinski said...

A terrific explanation Randy!