Ah yes, the turkey, the gravy, the stuffing and who could forget the pumpkin pie? As an aspiring obesity researcher I can't help but be fascinated by food intake patterns on holidays such as thanksgiving. These holidays are a good example of how people's dietary patterns are more closely related to the patterns of others than they are to the individual - the thanksgiving feast is common to millions, but it's not everyday we sit down and stuff ourselves full of turkey.
My fascination with food and holidays extends to our ability to measure peoples' food intake patterns - something this field is not terribly good at. In fact, a study just published in the Public Library of Science journal, PLOS One, by Edward Archer has identified some major flaws in the reporting of dietary intake in the US' National Health and Nutrition Examination Survey (NHANES)- the largest survey of the dietary habits of Americans. They reported that throughout the 39-year history of the NHANES the majority of respondents reported physiologically implausible dietary intakes - essentially, they were under-reporting how many calories they ate compared to how many they burn. Their conclusion? "The ability to estimate population trends in caloric intake and generate empirically supported public policy relevant to diet-health relationships from U.S. nutritional surveillance is extremely limited." It's thought that this effect is largely due to measurement error (the tools aren't good enough) and response bias - people who know they're being observed will change their behavior (see Hawthorne effect), or in this case, misreport their behavior.
While I may be disappointed with our measurement techniques for food intake I take solace in the fact that other researchers have it worse than I - namely the quantum physicists. Observer effects exist everywhere in science - altering the path of an electron by measuring it with a photon, the measurement of the momentum or position of a particle where improvement in one leads to reductions the other, and of course the observer-expectancy effects in clinical trials. Even when measuring the air pressure in a tire, some of the air must be let out by the observer! The quantum physicists, however, have to put with a nasty concept called the wave function collapse - this is where, simply put, the type of measurement being used on a particular system affects the outcome or end state of the system. Consider the quantum Zeno effect, where the very existence of a system relies upon it's being measured (or else otherwise decaying).
There just doesn't seem to be a good way around the overall systematic bias of measuring food intake - just like the physicists measuring the momentum and position of a particle, there are trade-offs in the different measurements of food intake. I'm not sure that we'll ever obtain a true record of the diet of a population (prospective or retrospective), and we may be forced to reduce measurement bias wherever possible, but ultimately accept its existence and its effect on our outcomes of interest. Research in this area may forever be required to add the phrase 'interpret with caution' to every paper or book published. I should note that some advancements have been made - recent tools involve taking pictures of food, sometimes with a smart phone, for assessment of serving size and composition. These tools have a long way to go, and they may introduce other forms of bias, but they seem promising for energy balance research.
However, even if measurement error is reduced to null, or an acceptably low level, there will always be physiological differences in the food we consume, digest, and store. Rob Dunn notes several of these in his recent article (Scientific American, September issue, 'Everything you know about calories is wrong'), including variation in: the 'appetites' of gut bacteria, rates of absorption of different foods and alteration in these rates due to co-consumption of other foods (think interaction effects), and proportion of digestion of many foods (nuts and legumes in particular). There are other factors that influence the number of calories we consume and store as well, such as the preparation of food which can partially digest some compounds making it easier for the body to digest them (and thus requiring less energy from the body to do so), and the immune response that some foods prompt in the body - something that has not been considered from an energy balance perspective.
I suppose all scientists - be they measuring the particles of the universe or those in a slice of a pizza - have to put up with observer effects. For many of us, learning to live and work with these effects is difficult to swallow - but alas the observer effects are here to stay, giving us something to chew on over the holidays.