Why Nutritionists have a Hard Time Beating Obesity

[joshwriting]

Portion sizes growing with American waistlines

Food servings are bigger than 20 years ago, but most unaware, study says

All well and good, right?

A 1994 informal survey found that the standard plate size in the restaurant industry grew in the early 1990s, from 10 inches to 12.

“That holds 25 percent more food,” Schwartz said. “That really makes a difference in how much our plates can hold and how much we eat from them.”

New math?

Comments

[ratesjul.livejournal.com]

Hmmm

(pi)*5^2 = 78.5398...
(pi)*6^2 = 113.0973...

78.5398/113.0973 = .694

The bigger plate would seem to be 1.44 times the area of the other.

Emperical measures?

[murasaki99.livejournal.com]

Hmm, was the nutritionist talking 25% by weight or volume? If you take the 12-in. plate as 100%, the 10-in. plate was 83% of that. So the plates grew by only 17%? Assuming I remember my really basic stuff correctly.

Of course some stuff like commercial muffins are huge compared to my standard muffin pan. I think one commercial muffin = 4 home-made ones.

[whatbox.livejournal.com]

I'm grateful that you are not the only person I know who'd read that and of all the implications, notice the math error immediately.

My personal filters only picked up, "No wonder trying to eat properly is still so discouraging. God, the food really is increasing in quantity."

[dpolicar]

A couple of restaraunts I go to regularly are of the huge-portions variety. What I actually find is that when the portion size crosses the threshold between "too large" and "absurd", I am far more likely to eat half of it (which is an entirely plausible meal) and take half of it home to eat some other time.

Not that this changes either the general consequence of larger portion sizes, or the math, but I thought I'd mention it.

Incidentally, the analysis above isn't taking into account the difference in plate-rim-width between a 10" and a 12" plate. The ratio of effective plate area would be
(PI*[(10-n1)/2]^2)/(PI*[(12-n2)/2]^2)
= [(10-n1)/(12-n2)]^2

Granted, if we want the result to be 1.25 then...
SQRT(1.25) = (10-n1)/(12-n2)
SQRT(1.25)*(12-n2) = 10-n1
SQRT(1.25)* n2 - n1 = SQRT(1.25)* 12 - 10
1.12 * n2 - n1 ~= 3.42

...which seems remarkably implausible.