[From Proceedings of the Human Factors and Ergonomics Society 41st Annual
Meeting. Copyright 1997 by the Human Factors and Ergonomics Society, P.O. Box
1369, Santa Monica, CA 90406-1369 USA; 310/394-1811, Fax 310/394-2410,
An Evaluation of Nutrition Label Quantifier Terms: Consumer Perceptions vs. FDA Definitions
Marc L. Resnick, Industrial and Systems Engineering, Florida International University, Miami, Florida
The FDA has regulated food label nutrition claims such as "reduced sodium" and "low fat" since it began enforcing the Nutritional Label and Education Act in 1994. However, there has yet been no quantitative evidence that the FDA's definitions of these quantifier terms correspond to consumers' perceptions of what the terms mean. This study investigated three common quantifier terms used on food labels (reduced, low and free) in relation to four dietary components (fat, sugar, sodium and cholesterol). Subjects were shown food labels that were modified to contain every combination of these quantifiers and dietary components and asked to report how much of the component was present in one serving of the food item. The results show that consumers' understanding of these terms do not correspond with the FDA's regulations. For the seven combinations for which the FDA has a specific numerical definition, subjects significantly overestimated all seven. Furthermore, the perception of each quantifier term was affected by the dietary component being modified, the food item being evaluated and the subject doing the evaluating. These results suggest that in order for food labels to provide accurate information for consumers, the definitions of these quantifier terms must be adjusted.
Two trends in the lifestyles of U.S. consumers over the past decade have created a fervent interest in food labels. The market for packaged and prepared food has expanded significantly, increasing the variety of food products that require labels. Many of these products contain ingredients that are not easily predicted by the average consumer. At the same time, consumers have become more nutritionally conscious. Because of these trends, the labels of packaged food products became filled with nutritional claims that used quantifier terms such as 'reduced,' "low" or "free" very loosely. To eliminate the confusion surrounding the definition of these terms, the FDA developed a set of standards that regulated the use of many nutritional claims. These standards mandated that all processed food products covered by the Nutrition Labeling and Education Act of 1990 (NLEA) contain a label with information about several dietary components, including fat, cholesterol, sodium, sugar and others. These "Nutrition Facts" labels contain both the amount of each component and the % Daily Value assuming a 2000 calorie per day diet. However, many consumers do not read nutritional food labels (Feick, Herrmann and Warland, 1986; Kessler, 1991). Even those who do read them often do not have the expertise to comprehend many aspects of the labels. Simple nutritional claims such as "low fat" on the front of the food label are more likely to be read and can significantly affect consumer purchasing decisions.
In acknowledgment of this reality, the NLEA also defined certain quantifier terms, such as "reduced," "low," and "free," for use on food labels (see Table 1). However, these labeling guidelines created an additional set of problems that were not considered in 1994. There was no research verifying whether the FDA's definitions of these quantifier terms corresponded to consumers' perceptions of the contents of food products labeled with these terms.
As evidenced by a study conducted in the food industry (Opinion Research Corporation, 1990) the perceptions of many quantifier terms, such as "low cholesterol," are misunderstood by a significant proportion of the population. This study also discovered a wide interpersonal variation in the perceptions of quantifier terms. In fact, the American Dietetic Association recommended that the term "reduced" be removed from the NLEA because of the potential for misunderstanding (ADA, 1990). Wogalter, Kalsher and Litynski (1996) tested the relative perceptions of consumers regarding eight quantifier terms. A significant result of this study was that two quantifier terms that are treated equally by the FDA, "no" and "free," are perceived as significantly different by consumers. A limitation of this study, however, is that it did not measure the quantitative difference between any of the quantifier terms. The practical differences between the quantifier terms were not evaluated.
Russo et al (1986) reported that consumers are more likely to be affected by nutrition labels that relate information regarding negative dietary components, defined as nutrients such as fat and cholesterol of which most people need to consume less compared to positive dietary components, defined as nutrients such as vitamins and minerals of which most people need to consume more. The most common examples of negative dietary components are fat, cholesterol, sodium and sugar. Feick et al. (1986) suggested that the high costs associated with negative health outcomes (diabetes, heart attack, atherosclerosis) increase the rewards of reading the labels and mitigate the time costs of spending the time learning how to use the labels and reading them. In order for consumers to reap the rewards of this information, it must be presented in a format that is understandable and consistent. The NLEA standardized the presentation of nutritional information. This study investigated the effectiveness of these NLEA standards.
The objective of this study was to investigate consumer perceptions of three common quantifier terms (reduced, low, and free) as seen on actual food labels. The study measured the quantity of four dietary components (fat, cholesterol, sugar and sodium) that consumers perceived as present in food products when modified by these quantifier terms. These values are compared to FDA standards to determine the appropriateness of the standards and suggest areas for adjustment.
Fifty subjects from the greater Miami metropolitan area were recruited for participation in the study, twenty eight females and twenty two males. Subjects' ages ranged from nineteen to sixty four. All subjects had lived in the U.S. for at least ten years and spoke English fluently.
Stimulus Materials The labels of four nationally-sold canned soups were digitized. Using Microsoft PowerPoint, an additional image was added to the soup labels. For each label, thirteen images were created, corresponding to a complete factorial pairing of three quantifier terms (reduced, low and free) with four dietary components (fat, sugar, sodium and cholesterol) and one control label with no added image. The images were placed in the same location on each can label using the same font size and color. An example is shown in Figure 1.
Four booklets were created, each containing sixteen labels. The sixteen labels corresponded to the twelve modified labels and four copies of the control, one for each dietary component, and rotated by soup type. Each pairing was presented to the subject only once to prevent subjects from referring to earlier answers when responding. Each label was presented on its own page of the booklet. The order of presentation of stimuli within each booklet was counterbalanced. At the bottom of each page, a survey question was printed. The question asked the subject to report how much of the dietary substance noted in the label was present in one serving of the soup (bottom of Figure 1). On the answer sheet, the U.S. Recommended Daily Values for each dietary component, based on a 2,000 calorie per day diet, were listed as a reference (from USFDA, 1995).
Participants were provided with a booklet and answer sheet. The first page of the answer sheet contained five demographic questions which collected data on the subjects' age, gender, language preference and nationality. The second page repeated the questions from the survey booklet. Subjects were instructed to examine each label in the booklet and answer the question on the bottom of the page. They responded with the amount (by weight) of the dietary component they perceived to be present in one serving of the soup. The units were provided for the subject on the booklet and on the answer sheet. At the conclusion of the survey, subjects were debriefed. The time for each subject to complete the survey averaged ten minutes.
Summaries of the subjects' responses for cholesterol, fat, sodium and sugar are shown in Figures 2 through 5 respectively. The FDA definitions of "low" and "free" are marked on the Figures. These results show that subjects are clearly unaware of either the FDA definitions or of their enforcement on food labels. Subjects judged the content of fat and sodium in the "low" soups well above the FDA maximums of 5g and 150 mg respectively. With the exception of the minestrone soup, cholesterol was also significantly overestimated, compared to the FDA limit of 20mg. In the NLEA, the FDA definition of "free" requires a food product to have only physiologically inconsequential amounts of a component. This corresponds to less than 0.5g of fat and sugar, 2mg of cholesterol, and 5mg of sodium. Subjects consistently estimated levels of these dietary components as higher than the FDA limits.
From the responses to the control labels, it is apparent that subjects had different perceptions of the contents of the four different flavors of soup. This was expected because of the differing expectations for meat and vegetable based products. However, the beef and barley soup was actually perceived as having the least amount of cholesterol. There was also considerable variation between subjects for each flavor, with coefficients of variation ranging from 0.7 for the sugar content of the beef and barley soup to 1.6 for the sodium content of the vegetable soup.
The effects of the modifier terms varied greatly, both with respect to the different flavors of soup and the dietary components. Subjects reported that fat free indicated between 0 and 20g of fat, compared to a percent daily value of 65g. At the other extreme, responses to "cholesterol free" ranged from 0 to 140mg compared to a percent daily value of 300mg. In other words, some subjects perceived that "fat free" could indicate up to 30% of the Daily Recommended Value of fat and 45% of the Daily Recommended Value of cholesterol. This is significantly above the actual FDA regulations. Results for "low fat" ranged from 1g per serving to 60g per serving.
The differences between estimates of dietary content for soup flavors were also very high. Per serving estimates of fat content for "fat free" beef and barley soup ranged from 0g to 4g while estimates for "fat free" cream of chicken ranged from 0g to 20g. Subjects' perceptions may be affected by the word "cream" in the title, since cream in general has a lot of fat. These kinds of considerations seem to have had a greater effect on subjects' perceptions than the FDA regulated quantifier terms.
There are several findings of this study that have critical ramifications for food labeling and the NLEA definitions of quantifier terms. Consumers are clearly unaware of the definitions of these quantifiers. Furthermore, the definitions do not correspond to consumers' natural assumptions about the respective quantities indicated. When a consumer sees a product with a label such as "reduced cholesterol," she may make an incorrect assumption about the contents of the product. This can lead to significant health consequences.
The large variation in the subjects' perceptions of the contents of the soups without added quantifiers suggests that consumers are not aware of the content of dietary components in common food items. Thus claims on food labels should be in absolute rather than relative terms. The current NLEA definition of "reduced" is 50% less than the common content of that component in that food product. But if consumers are not aware of the common product's content, the claim of reduced will not be meaningful. On the other hand, if the definition consistently means "less than 5g," then consumers need only learn the definition once. Of course these results also indicate that this initial learning has not occurred. The FDA needs to conduct more educational initiatives if food labels are to be useful. Otherwise, labels will have to be quantitative, stating the specific amount of the dietary component (such as "only 3g of fat per serving").
Subjects' varying interpretations of quantifiers such as "cholesterol free" with different soup flavors also suggests a lack of trust in food labels in general. Consumers are habituated to pre-NLEA food labels that made dubious claims regarding food content. One subject estimated the contents of one serving of "low fat" beef and barley soup as 60g of fat, almost an entire day's daily value. Without trust in the veracity of labels, even consistent definitions would not solve the problem.
A final result that bears consideration is the fact that consumers' estimates of many of the dietary components had little foundation in logic. It seems peculiar that any consumer could think that one serving of a soup labeled "fat free" could have 30% of the entire daily recommended amount of fat. Though this was the extreme case for "fat free," there were several instances of unbelievable responses. Post-test surveys indicated that subjects were not intentionally trying to give bad answers, but were simply uneducated about nutrition.
There are two initiatives that must be undertaken for the goals of the NLEA to be effective. First, consumers must be made aware of the FDA definitions of quantifier terms. From a human factors perspective, it would be best to match FDA definitions with consumers' natural perceptions of the terms' meanings. Then, subjects would not have to independently learn the definitions. However, it appears that there is no consistent perception in this regard. Therefore, the FDA must embark on an educational campaign to educate consumers on the FDA definitions. If consumers are generally unaware of the dietary contents of many packaged foods, such as this study indicates, then these definitions must be constructed in absolute terms such as "no more than 5g" rather than relative terms such as "50% less than a serving of the normal kind." Finally, the existence of mandatory standards must be publicized, so that the misinformation of past labels does not continue to desensitize consumers to current labels.
American Dietetic Association (1990). ADA's response to FDA's cholesterol labeling rule. Journal of the American Dietetic Association, 90, 1520.
Feick L.F., Herrmann R.O. and Warland R.H. (1986) Search for nutrition information: A probit analysis of the use of different information sources. Journal of Consumer Affairs, 20, 173-192.
Foulke J.E. (1992) Wide Sweeping FDA Proposals to Improve Food Labeling. FDA Consumer, Jan-Feb, 9-13.
Kessler D.A. (1991) Building a Better Food Label. FDA Consumer, Sept, 11-13.
Kurtzweil P. (1994). "Nutrition Facts" to Help Consumers Eat Smart. U.S. FDA Press Release. Washington D.C.
Opinion Research Corporation (1990). Food labeling and nutrition: What Americans want - Summary of findings. National Food Processors Association. Washington D.C.
Russo J.E., Staelin R., Nolan C.A, Russell G.J. and Metcalf B.L. (1986) Nutrition information in the supermarket. Journal of Consumer Research, 13, 48-70.
U.S. Food and Drug Administration. (1995) Nutrition and Your Health: Dietary Guidelines for Americans. Fourth Edition. U.S. FDA/Center for Food Safety and Applied Nutrition.
Wogalter M.S., Kalsher M.J. and Litynski D.M. (1996) Influence of Food Label Quantifier Terms on Connoted Amount and Purchase Intention. Proceedings of the Human Factors and Ergonomics Society 40th Annual Meeting. Human Factors and Ergonomics Society. Santa Monica, CA. p385-389.