Thursday, August 7, 2008

Media Misinterpretations

The New York Times just published an article called "The Overflowing American Dinner Plate", in which they describe changes in the American diet since 1970, the period during which the obesity rate doubled. Bill Marsh used USDA estimates of food consumption from 1970 to 2006. Predictably, he focuses on fat consumption, and writes that it has increased by 59% in the same time period.

The problem is, we aren't eating any more fat than we were in 1970. The US Centers for Disease Control NHANES surveys show that total fat consumption has remained the same since 1971, and has decreased as a percentage of calories. I've been playing around with the USDA data for months now, and I can tell you that Marsh misinterpreted it in a bad way. Here are the raw data, for anyone who's interested. They're in easy-to-use Excel spreadsheets. I highly recommend poking around them if you're interested.

The reason Marsh was confused by the USDA data is that he confused "added fats" with "total fat".  While total fat intake has remained stable over this time period, added fats have increased by 59%. The increase is almost exclusively due to industrially processed seed oils (butter and lard have decreased). Total fat has remained the same because we now eat low-fat cuts of meat and low-fat dairy products to make up for it!

Another problem with the article is it only shows percent changes in consumption of different foods, rather than absolute amounts. This obscures some really meaningful information. For example, grain consumption is up a whopping 42%. That is the largest single food group change if you exclude the misinterpreted fat data. Corn is up 188%, rice 170%, wheat 21%. But in absolute amounts, the increase in wheat consumption is larger than corn or rice! That's because baseline wheat consumption dwarfed corn and rice. We don't get that information from the data presented in the article, due to the format.

So now that I've deconstructed the data, let's see what the three biggest changes in the American diet from 1970 to 2006 actually are:
  • We're eating more grains, especially white wheat flour
  • We're eating more added sweeteners, especially high-fructose corn syrup
  • Animal fats from milk and meat have been replaced by processed seed oils
Wheat + sugar + processed vegetable oil = fat and unhealthy. Sounds familiar, doesn't it?


Ross said...

Hey Stephan,

Interesting analysis. You should be the one writing for the NYT :) Perhaps you could write a response to the article?

Stephan Guyenet said...

Ross, you read my mind. I just submitted a letter to the editor about it.

AngloAmerikan said...

The New York Times chart is a nice rendering of the data visually. Thanks very much for the analysis - it makes much more sense now. It didn't occur to me think about how a smaller percentage could equate to a far greater quantity of one product.

dislyxec said...

Which spreadsheets and which columns are you looking at to determine that total fats have remained (relatively) constant?

From what I see both grains and "added fats" have increased significantly.

Added sugars has also increased a lot (and changed from mostly beet/cane to corn-based), but the last four years noted (2003-2006) show a slight downtick.

Is there data here that shows the meat consumed is now leaner?

Jeff said...


Excellent post. The world is so complex and there is so little time, it is easy to be misled. Your comment about wheat increasing being misleading is exactly the type of thing we commoners would just glaze over and all of us miss the real point. You are doing an excellent service with your blog. Please keep up the excellent work and posts. I can't tell you how thankful I am.


Linda said...

Excellent analysis!

Stephan Guyenet said...


Glad you're looking at the data! The USDA data do not address actual total fat consumption. The confusion arises because the "total fat" column does not actually refer to total fat intake. It only cherry picks certain sources.

If you want to get a feel for the decrease in milkfat consumption, look at the dairy numbers. You'll see that whole milk has been replaced by lowfat milk in the time period we're talking about. I know it's bizarre, but that is not included in the "total fat" figure they present. They don't address meat fat in the data.

You see, those data only refer to consumption of specific types of foods. They make no effort to estimate the fat content of meat, which has changed over the last 30 years due to fatophobic consumers. For example, they track beef sales but not the fat content of the meat that's being sold. That's why you can't use their data to calculate total fat consumption. Actually you might be able to do it, but you'd have to have other information to supplement it and a lot of time on your hands.

Stephan Guyenet said...

Jeff and Wifezilla,

Thank you, I'm glad you enjoyed it.

dislyxec said...

Yeah, I understand how the "total fats" column is actually "added fats", and thus their data does not actually have any total fat consumption of the diet.

So, I agree that you're right to take exception to the 59% more fat consumption misinterpretation, but I think saying that actual total fat has remained the same is possibly misleading, since it isn't backed by the data provided.

I'm apt to believe it's true, since red meat has decreased and poultry increased a lot, poultry typically being leaner than red meat.
I also agree from observation that leaner cuts of meats are being consumed (and that people tend to trim more fat off of their cuts), but the data linked do not seem to have that information.

Stephan Guyenet said...


Yes, you're right about the USDA data not addressing total fat intake. The data I cited that fat consumption has not changed comes from the CDC NHANES survey, which did assess total fat consumption.

dislyxec said...

Ah cool, thanks.