I've long argued that finding out what people eat is the most intellectually challenging aspect of nutrition research. To put it bluntly, everybody lies. OK. We don't lie. We just can't remember or estimate portion sizes accurately.
For years, government agencies have gone to great trouble and done the best they can to get some reasonable idea of what Americans actually eat. They report the results as "What We Eat in America." The data may not be perfect (they almost certainly underestimate actual intake), but they are the best we have and always of great interest.
Every element of obtaining dietary intake information is fraught with error.
I always like to know what is going on with calories. The USDA's most recent data are from 2005-2006. These show that women on average consume 1785 calories a day, men 2638, and together 2157. These figures are based on intake reported for 24 hours and almost certainly underestimate real calorie intake by one-third or more. Compare these figures to calorie production, which is now 4000 per capita per day! (See Table 1). The truth undoubtedly lies somewhere in between, and all we can do is make good guesses.
USDA files its dietary intake reports under Products & Services. Its latest looks at intake of four nutrients: calcium, vitamin D, phosphorus, and magnesium. In comparison to dietary reference intakes (DRIs), Americans eat pretty well. The low magnesium intake makes me wonder if the DRI for that nutrient is too high, but I tend to be skeptical about such things.
Everything about these reports requires much careful interpretation, since every element of obtaining dietary intake information is fraught with error. Better methods would help a lot. If only we could figure out how to do this better. A challenge, indeed.