« The Economic Naturalists's Field Guide | Main | How our macro book differs »
Fast Food, Fat Food
Catherine Rampell at Economix posts this interesting chart showing the relationship between the "time the average person in a given country spends eating and that country’s obesity rate (as measured by the percentage of the national population with a body mass index higher than 30)."
Posted by Alex Tabarrok on May 6, 2009 at 07:10 AM in Food and Drink | Permalink
Comments
Obviously, Obama should require all McDonald's to install more comfortable handicap accessible chairs.
Posted by: Andrew at May 6, 2009 7:19:05 AM
There is no correlation if you only look at Europe.
Posted by: a at May 6, 2009 7:22:22 AM
Does anything happend if you just remove the two obvious outliers (USA and Mexico) ?
Posted by: Mathieu P. at May 6, 2009 7:27:07 AM
Demographics is the first thing that comes to mind looking at this chart. Isn't BMI just based on some kind of height to weight ratio? Mexicans generally have different body types than Swedes. Maori are freaking huge; watch a rugby match sometime. If we adjust for this stuff, is there any relationship at all?
Posted by: josh at May 6, 2009 7:36:40 AM
I don't like the BMI as an indicator of obesity. I personally have a BMI of 29 (not quite 30 but still I was told I was in the overweight range) yet I am hardly overweight. At least I don't think I so; I am undoubtedly under significant bias but my reasoning stems from the following: I am 22, a member of my university's lightweight rowing team, 5'11'', 157 lbs, work out more once a day every day, follow a strict diet and have less than 7% body fat. Yet, my BMI places me in the overweight (just barely not in the obese) category; I would beg to differ.
Likewise my brother has an even high BMI than I, yet he weighs the same and is two inches taller. He is a lightweight rower at his college as well. During my BMI test(?) I was told that because I was an athlete my reading would be skewed upwards. If I understand correctly then, if you workout regularly it will read as obese, and if you don't workout at all it will read as obese. Who then doesn't read as obese?
Posted by: Let them eat Thomas Paine at May 6, 2009 7:37:08 AM
I'm with Mr. Paine, 'cuz last time I was out at the mall I just couldn't believe how fit everyone had become.
Posted by: Tomas at May 6, 2009 7:50:39 AM
Let them eat, if you are 157 pounds and 5'11", then your BMI is 22. It's weight in kilos, divided by the square of your length in meters. If your brother weighs the same but is taller, than his BMI must be lower.
Posted by: Zamfir at May 6, 2009 7:56:48 AM
It's true that BMI doesn't account for muscle - but I doubt the extremely-slim but heavily muscled population counts for much of the total.
Posted by: MattC at May 6, 2009 8:04:30 AM
@Thomas Paine, your BMI is 22, not 29.
Also, Maoris are quite a small percentage of New Zealanders, and Mexicans *are* a lot fatter than Swedes.
Posted by: jonm at May 6, 2009 8:05:18 AM
The chart is somewhat interesting. The trend line plotted, however, is not at all interesting. It's a use of statistics to obscure. As noted, the trend is almost entirely dependent upon one or two outliers.
Posted by: John Thacker at May 6, 2009 8:10:15 AM
Interesting. It must have been a poorly done test then; we had it done at a health expo. But I specifically remember them saying that athletes do poorly on BMI tests.
Posted by: Let them eat Thomas Paine at May 6, 2009 8:11:00 AM
Actually, looks at four horizontal groupings.
1. USA and Mexico
2. Previous Commonwealth countries; UK, Australia, Canada, New Zealand.
3. Central European and Scandinavia
4. Developed Oriental, Japan and Korea
Group them that way, and what do you have? Not much.
David C.
Posted by: David C. at May 6, 2009 8:17:20 AM
The biggest problem I see with this (and many similar) charts is that the correlations are done among groups (in this case, countries) rather than individuals. I'd be much more impressed to see the scatter plot of "time spent eating" vs. BMI for individuals, across nations, to see what correlations there are.
Posted by: Alex R at May 6, 2009 8:23:11 AM
Maori are around 15% of NZ and I think you may have hurt their feelings, there.
Of course Mexicans are fatter than Swedes; I'm questioning the implication that time spent eating is a causal factor.
I'm wondering why Tyler posted this since it seems pretty obviously worthless. Is there some meta-reason to post it?
Posted by: josh at May 6, 2009 8:28:37 AM
One issue with doing it for individuals is that people are probably terrible at remembering how much time they spend "eating". The only way to do it would be in some sort of study where participants have to monitor their "eating time", but that has all sorts of other problems. That's probably why the country data was given.
Posted by: Andy at May 6, 2009 8:57:59 AM
Thomas Paine:
You're right on. I need to get to -2% body fat to get out of the overweight grouping. I'll let you know how that goes
Posted by: Tom at May 6, 2009 9:25:39 AM
@a: There is no correlation if you only look at Europe.
UK is top left en France is bottom left. I suspect the correlation is still well above zero (significance might be a different matter).
@josh: It's Alex though.
Posted by: JSK at May 6, 2009 9:26:59 AM
@Thomas Paine
A quick and dirty way to calculate your BMI is: BMI Calculation: [weight ÷ (height in inches)*(height in inches)] x 703
And, if your brother ways the exact same as you, but is taller, then his BMI has to be lower because height is squared and in the denominator. It's weight that pulls it up, as weight is in the numerator of the calculation.
But, it could be they give you a different test for body fat than this BMI measure, which is what a lot of people use when they only data that they can get their hands on is height and weight. I suspect that a nutritionist who is conducting experiments on people will be more likely to use other measures.
Posted by: anon at May 6, 2009 9:36:24 AM
But Andy, making the measurement for a group of people has all the problems of making it for an individual, and the additional problems of making sure your sample is representative of the group.
Posted by: Cyrus at May 6, 2009 9:37:23 AM
The chart doesn't prove there is not a correlation between counter-intuitive variables.
Posted by: Andrew at May 6, 2009 10:18:51 AM
Given the confusion evident in the comments about how to calculate BMI, perhaps the graph is simply a one-dimensional plot of "maladeptness at using metric measurements."
It would correlate well with common stereotypes... Asians at the bottom, because they are good at math; continental Europeans next, because they have used metric longest; former British empire above, because they only switched in the past few decades; Mexico higher because they still use a mix of metric and SAE in commerce; the USA highest as the only major SAE holdout.
Posted by: Bob Knaus at May 6, 2009 10:35:07 AM
@Andrew - double negative, argh! Does that mean the chart does prove it?
Posted by: anon at May 6, 2009 10:35:53 AM
Lets also see some unemployment stats for the relevant time period?
Posted by: rb at May 6, 2009 11:00:55 AM
BMI is complete BS. I am 5'10", weight 190 pounds.
I can also bench press my weight and run a mile in 7 minutes.
According to the BMI chart, I am overweight.
Two simple questions: Can an "overweight" person bench press his weight AND run a mile in 7 minutes?
Posted by: Dave at May 6, 2009 11:39:06 AM
How on earth can someone spend 160 minutes eating a day???
Posted by: Stan B at May 6, 2009 12:26:45 PM
