Research and evaluation activities in USDA: from the Food Surveys Research Group, Beltsville Human Nutrition Research Center, Agricultural Research Service - Regular Items

Family Economics and Nutrition Review, Spring, 2001

Do People Really Know How Much They Eat?

How well are people able to report amounts of foods they eat? This question forms the basis of research conducted to improve the accuracy of information about the types and amounts of foods eaten by Americans. This information is collected in periodic USDA food consumption surveys that date back to the 1930's. The most recent nationwide dietary survey conducted by USDA is the 1994-96, 1998 Continuing Survey of Food Intakes by Individuals (CSFII).

The USDA has been a leader in the development of methods to help people more accurately recall and report what they eat. Research supported by USDA in the early 1980's demonstrated that, for dietary recalls administered specifically in the home, common household measuring utensils are potentially more helpful than abstract three-dimensional food models.

Food preferences and sizes of food portions have changed over the years. Ethnic foods and one-dish meals have grown in popularity; typical portions served in food establishments are reported to be increasing. For example, larger- or super-sized servings of beverages and french fries have become a mainstay on menus in fastfood restaurants. Many food choices today come in a variety of sizes-from bite-size to fun-size to king-size. Larger portions and the use of multiple servings during a meal add to the complexity, and emphasize the necessity, of collecting amounts accurately. At the same time, foods being consumed in smaller portions by weight-and health-conscious people must also be estimated accurately.

Food Model Booklet

As portions and preferences are changing, so are the methods used to conduct nationwide food surveys. In-home collection of in-person dietary data is not the only means of gathering this type of information about Americans. The Food Surveys Research Group (FSRG) has been conducting research on collecting dietary data by telephone. Telephones provide an economical and valid means for gathering such data if response rates can be maintained by notifying respondents in advance about participating in the study. As part of the research program to adapt new collection methods, FSRG has developed an easy-to-use Food Model Booklet that incorporates high-tech graphics and research on how people perceive quantities and on what is and is not helpful.

The Food Model Booklet contains 30 life-size pictures ranging from two-dimensional drawings of typical household food containers (e.g., glasses, cups, and bowls) to more creative drawings of geometric shapes (e.g., amorphous mounds, wedges having a moveable arrow to denote sizes, concentric circles, and a grid). In previous CSFII surveys, interviewers used measuring cups and spoons and a ruler to help survey respondents recall portion sizes of foods they ate. With the addition of the Food Model Booklet, respondents now have more ways to estimate amounts, thus making it easier for them and more accurate. It also allows for data collection via the telephone.

USDA has placed a priority on improving portion-size estimation aids through an 1890 Capacity Building Grant that was under the direction of Dr. Sandria Godwin at Tennessee State University in collaboration with Dr. Edgar Chambers IV at Kansas State University. The research has focused on identifying the cognitive strategies people use in reporting food amounts and on developing novel measurement aids. Aids were found to help in formulating memory recall and in setting boundaries for visualization and comparison. The research has shown that people prefer aids that are similar in size and shape to actual household containers--a finding that FSRG has implemented in developing the Food Model Booklet. The research also has demonstrated that estimates of relative size (i.e., small, medium, and large) are effective for some difficult-to-measure foods.

Development of the Food Model Booklet involved graphic design and computer power precision to draw the two-dimensional models to life-size depiction and accurate volume capacity. But development did not stop there. An accuracy test was conducted with 264 adults to assess how accurately people estimated amounts of 17 commonly eaten foods such as: spaghetti, pizza, cola, meatloaf, and potato salad. The participants estimated the amount of each food by using a two-dimensional model from the Food Model Booklet and typical measuring aids (including measuring cups and spoons and a ruler). The results: Overall, people estimated serving sizes as well with the Food Model Booklet as with other measuring aids. And, estimates improved by about one-third with the "mound" models, compared with the measuring cups.

The Food Model Booklet is but one answer to ensuring that respondents can accurately report how much they eat. Dietary recalls were administered in a nationwide pilot study of nearly 800 individuals of all ages. The Food Model Booklet was used as one of the measurement aids in addition to measuring cups and spoons and a ruler. Only 17 percent of the food portions reported were estimated by the assistance of the Food Model Booklet. Most food portions were reported with descriptions such as "large, medium, thin slice, and can of soda," as well as cups or spoons. In many cases, the name of the item, such as Big Macs or Fig Newtons[R], is directly associated with a known weight, so only the number eaten needs to be reported. Also, many foods come prepackaged with labels showing the quantities. Hence the marketplace is another important source for determining portion sizes.