Tuesday, May 25, 2010

Not too skinny, not too fat, but just right

Last winter, Half Sigma pointed to an online FAQ where it was revealed that commercial turkeys have, through selective breeding, developed breasts too large to allow for natural breeding to occur. Consequently, they must be artificially inseminated. He wondered if there was something applicable to humans in this:
People have also been, allegedly, getting fatter. I think we need to look into whether there’s a genetic explanation. Are fat people having more children than thin people? It’s well established that married people weigh more than single people, and traditionally married people have children and single people don’t. Maybe people are getting fatter for the same reason that turkeys have gotten too fat to mate. It’s all in the breeding.
In 2004, GSS interviewers were asked to assess respondents' weight and place each into one of four categories. The following table shows the mean number of children by weight class. To exclude those whose reproductive life is still ahead of them, only respondents who were at least 40 years old at the time of the interview were included:

Really Heavy2.11

This provides a pretty crude measure of heftiness, but it doesn't support the assertion that humans are getting fatter because the corpulent have more kids than healthy folks do. Those who are noticeably slender, a designation that applies to less than 7% of the population (probably indicating skinniness to the point of apparent emaciation), aren't popping out a lot of kids, either.

I find the table above to be encouraging!


Jokah Macpherson said...

Go average folks!

This is one of those situations where if you control for sex, you get astonishing results. It doesn't surprise me there aren't many skinny men because the ones that do exist hardly reproduce at all.

silly girl said...

He, he.

I don't know how they define it, but one bmi chart I found defined female bmi of 25 as healthy weight. That is 150 lbs at 5'5'' 30th %ile for a 40 year old. I would call that fat to look at.

123 lbs at 5'5'' is the 7th%ile for a 40 year old. I would call that normal.

Similarly even the 2nd %ile at 40, 115 lbs on a small frame is just thin. The same person at age 16 would be at the 29th %ile, clearly normal range.

By the numbers, fat is average. To look normal and healthy at that age is a much lower percentile.

Calculator height, weight, age, bmi, percentiles:


silly girl said...

Okay, if that weren't depressing enough, the 50%tile for 40 year old women 5'5'' is 162 lbs, bmi 27 and officially overweight.

So, by the chart AE posted, 50% of the women are in the heavy to very heavy category officially. Probably 50%-100% of the women in the average category are also fat. Leaving the nominally thin at about 10%-max 20% of women in the thin category. That is if you still consider a 40 year old 136 lbs 5'5'' to be "thin".

Jokah Macpherson said...

Silly girl,

As far as I know, the GSS variable is just the inverviewer's assessent of the subject's weight using their own peepers, which is probably more honest than an objective standards-based measurement like BMI that doesn't take into account individual body types.

Audacious Epigone said...

Silly Girl,

BMI always seems to me to overstate the putative corpulence of taller people and understate it for shorter people. I get "overweight" with a BMI of 27, but no one assessing me would describe me as such. As Jokah says, though, the GSS question is based on interviewer assessments, and is consequently (even though it's aproximate and not standardized) probably more useful than BMI.

silly girl said...

My real point was the percentiles. Since the actual distribution of thin women in the overall population is known, it seems that the thin folks are slowly losing their already small stake. If more thin women were interviewed for the GSS than are normally representative, it doesn't change their actual distribution. The interviews reveal the number of kids per woman by weight class not the distribution of those weight classes.

kurt9 said...

I think the BMI index is bogus. A much more useful standard would be body fat percentage, which is measured reasonably well by electrical resistivity by an instrument cheap and easy enough to use that gym employees routinely use it. The fact that the medical community is too lazy to switch from BMI to body fat percentage measurement despite the ease and low cost measurement equipment demonstrates the intellectually bankruptcy of the medical industry.

TGGP said...

Robin Hanson highlighted a study suggesting that heavier folks have more kids.

Ron Guhname said...

I looked at the 2004-2006 MIDUS Study that weighed people. The correlation between current weight and number of children is .02. Nothing there.

Anonymous said...

"The correlation between current weight and number of children is .02. Nothing there."

However, the distribution of the overweight folks remains such that the vast majority are what most would call fat.