5
Consumer Use and Understanding of Front-of-Package Labeling Systems
This chapter reviews the studies that have examined front-of-package (FOP) labeling systems in applied and experimental settings. This chapter provides a general overview of the types of studies in the current literature and the methodological strengths and limitations of each. It also distinguishes the literature that has been published in peer-reviewed journals from information from industry, marketing, and government sources, i.e., applied and other settings of marketing research. This primarily includes recent consumer research conducted by the U.S. Food and Drug Administration (FDA), the UK government, and the Grocery Manufacturer’s Association and International Food Information Council (GMA-IFIC). The committee noted that while there are a number of theoretical, purchase intent studies, studies examining consumer purchase behavior in the grocery store setting are quite limited in both number and scope. In assessing the evidence and deriving its conclusions about consumer use and understanding of FOP systems the committee included information from applied marketing research, which provided insight into various nutrition rating systems and symbols that was not available in the peer-reviewed literature, particularly into what type of FOP symbols consumers might use. Information provided by FDA, GMA-IFIC, and other industry sources (discussed below) also provided such insight and, along with the committee’s expert judgment, served to complement the interpretation of the peer-reviewed evidence. The following discussion describes in further detail the committee’s approach to prioritizing and interpreting the available evidence.
The committee reviewed the literature for a number of FOP systems described in Front-of-Package Nutrition Rating Systems and Symbols: Phase I Report (IOM, 2010), placing an emphasis on published field experimental studies (described below). When reviewing these studies, the committee paid particular attention to the influence of various FOP systems on consumer choice of products and additional outcomes, such as perceptions of the healthfulness of products. The studies either indirectly or directly provided evidence on other factors important to examining FOP systems, including: (a) the extent to which simpler compared to more complex systems are more influential, and (b) the relative influence of FOP systems that highlight or frame nutrition content in only positive terms (e.g., an indicator for being high in nutrients). In general, the data do not exist to compare the effects of every potential FOP system against all other possible options. However, the committee determined that a number of important conclusions could still be drawn.
The following section discusses the literature on consumer use of and preferences for FOP symbol systems. The committee limited its review to literature that directly examined FOP symbol systems, was published in the United States and Europe, and covered a search period from January 2000 to June 2011. The committee included studies published prior to 2000 at its discretion. Chapter 4 discusses the literature on use of the Nutrition Facts panel (NFP), and Chapter 6 discusses the literature on consumer response to aspects of labeling, including health claims, package clutter, and related themes. Appendix D provides a complete description of the committee’s approach to its review of peer-reviewed published literature.
Types of Front-of-Package Symbol System Studies Examined
The committee used a hierarchy to categorize studies of FOP symbol systems, ranging from studies that are most likely to provide the best insight into how consumers might respond to a particular FOP symbol system to those that provide a lesser quality of evidence or are associated with greater uncertainty. Studies published in the peer-reviewed literature are at the top of the hierarchy. Within this group, field or natural experiments were given the greatest weighting. Field experiments examined implementation of an FOP symbol system in “real-world” settings and assessed their effects with objective outcomes such as changes in sales data. Although these studies are limited in number, their results are most likely to reveal how FOP systems might influence consumer choice if implemented.
Peer-reviewed studies reporting randomized designs provided the next level of evidence. These studies randomized subjects to view one (or several) variants of a FOP label, either in a research space, outside a supermarket, or online, and compared reactions across the experimental conditions on a variety of outcomes, including consumer choices, perceptions of product healthfulness, and overall preferences for FOP systems. Table 5-1 summarizes examples of field experiments and randomized design studies.
The committee then considered the applied marketing research literature that was either sought out or provided to the committee. Because it had not been subjected to a peer-review process, this work, positioned at the lower tier of the hierarchy, was given substantially less weight in the committee’s deliberations.
Peer-Reviewed Field Experiments
The strongest evidence to demonstrate how an FOP system will operate in a real-world shopping environment comes from actually implementing the system on supermarket shelf tags or products, and observing via sales data the impact that it has on consumer food choice and purchase decisions. As the name implies, field experiments are conducted in natural settings, which may not allow for full control of the environment, but allow for a full examination of how consumers make choices in a natural or real-world setting, with all of the inherent time, cost, and other pressures. They also provide a realistic sample—consumers go to the grocery store as part of their usual shopping routine. The outcome of interest in field experiments is sales, which is a key outcome under examination by the committee. Whether there are correlations between food product sales and patterns of total food consumption and subsequent outcomes such as levels of obesity has yet to be determined. As such, studies of this type are needed to better understand the scale and scope of the effects of FOP systems on consumer behavior.
The committee identified four studies that examined differences in sales after introduction of shelf tag–based FOP systems. In the first study, Levy et al. (1985) examined whether the prominent but relatively simple display of low/reduced dietary components (according to current dietary guidance: sodium, calories, fat, and cholesterol) on a shelf tag would increase sales as a result of “shelf salience.” This program was aimed to be “more promotional than rationally persuasive.” Ten comparison stores were matched with 10 control stores. Although the results were inconsistent across product categories, the investigators found that on average sales of the labeled products were 4
to 8 percent higher than sales of unlabeled products in comparison stores. The outcome showed an impact of the system on sales and by implication consumer appeal.
In the second study, Sacks et al. (2009) examined the impact of the UK-based traffic light (TL) system on a small subset of products in a store. Red, yellow, or green TLs were posted on “ready meals” (already packaged and chilled meals) and freshly made but pre-packaged sandwiches. Four weeks after the introduction of the labeling system, sales of the sandwiches had not changed. Sales of “ready meals” increased by a small amount, but there was no differential increase in the sales of healthier versus less healthy items. A major limitation of the study is the examination of a very small subset of products with only one FOP system. But, the system as implemented did not encourage consumers to choose healthier products for these product categories. The system was more complex than that of Levy et al. (1985) and included both positive (green for low) and negative (red for high) valence. The study did not provide evidence that allowed the committee to differentially examine the influence of these various factors.
The third study examined the sales of a single product, popcorn, in a field experiment in five grocery stores in the East Bay area of California. Berning et al. (2011) labeled popcorn products differently in the stores, but, if applicable, labels indicated whether the products were low in nutrients to avoid, e.g., “low fat” or “low calorie.” Popcorn products not meeting FDA standards for being low in a nutrient were not labeled with nutrition information, and “control stores” did not label any of the popcorn products with nutrition information. Compared to the “no nutrition labels” condition, when the labels were present there was a decrease in overall sales of the labeled, healthier products and a non-significant but similar in magnitude increase in the sales of the less healthy products. Again, this study was conducted on a single product category, and the findings differ from those of Levy et al. (1985), who found that a similar labeling system was effective for a variety of product categories. In addition, the product tested by Berning et al. (2011) is considered a “treat” or luxury rather than a necessity, which may have influenced how the shelf tag information was perceived by consumers.
In the fourth study, Sutherland et al. (2010) examined the change in sales for all products after the Guiding Stars system was implemented in Hannaford stores (see Chapter 4 for additional details). Looking at sales for the 8-month period after the system was introduced and comparing to the previous year’s sales to control for seasonality, the investigators found no change in the number of products labeled with stars over time, but there was a slight increase in the purchases of products labeled with stars. The greatest increase in sales appeared to be for 1-star products, although there was some increase for 2- or 3-star products. The study did not report changes in overall sales, so the committee was unable to ascertain the relative importance of an overall increase in sales and/ or substitutions between the starred (healthier) and non-starred (less healthy) products. An exception is additional data about ready-to-eat cereals, for which there was no evidence that healthier products were substituted for less healthy products. Unfortunately, there was no control group in this study, so it was not possible to ascertain whether sales changed in stores that did not introduce nutrition labeling. This study was funded by Guiding Stars program, and the investigators received compensation from Guiding Stars.
In summary, the evidence from field experiments is limited and inconsistent. The studies were not set up to truly test one FOP symbol system over another, and there is no evidence from the four studies to show one system as being superior to others. However, there is evidence that FOP and shelf tag systems can have some influence on consumer purchases. Of the systems studied, the TL-based system, which was the most complex, appears to have been the least effective in influencing consumer choices. Levy et al. (1985) reported consumer response to a simple shelf tag system, and Sutherland et al. (2010) reported consumer response to a simple and ordinal FOP system. However, the latter’s conclusions cannot be validated because the study was not controlled.
Overview of Peer-Reviewed Studies That Are Not Field Experiments
Non-field experiments do not allow for drawing conclusions about consumer behavior as easily as do field experiments, but they have their advantages. Because they are easier to undertake, non-field experiments can simultaneously examine various FOP systems and can easily control external factors, which allows for focusing on the actual variables of interest. The many issues inherent in non-field experimental studies are discussed below.
Non-field experiments are generally performed in one of three distinct settings. Common approaches are to
TABLE 5-1 Examples of Peer-Reviewed Studies Evaluating Front-of-Package (FOP) Systems
Peer-Reviewed Field Experiments | ||||
Intervention or | ||||
Source | Study Questions Study Design | Conditions | Outcomes Assessed | Summary of Results |
Levy et al., 1985 | What is the effect of a Quasi-experimental | A media campaign | Food products by | Overall, the nutrition |
nutrition information study using matched | introduced the nutrition | category were subdivided | information program had a | |
program on market comparison grocery | program to consumers, | into brands with and | positive effect on purchases | |
shares of selected stores in two | and informational | without a low/reduced | of products included in the | |
shelf-labeled food metropolitan cities | guides were available | dietary component shelf | program. The size of the | |
products low in | in stores and 1,600 | tag | effect varied widely between | |
sodium, calories, | selected products in 23 | food categories but generally | ||
cholesterol, and fat? | categories were labeled | Sales were tracked in | followed the same order of | |
with shelf tags | each participating store | magnitude as that due to | ||
for 2 years | price level or trend effects | |||
during the study period. | ||||
Sales-based market shares | Shoppers, especially those | |||
were measured as the | on special diets, reported | |||
percentage of store unit | using the shelf tags to | |||
sales by category | choose products. | |||
Sacks et al., 2009 | What impact has Comparison of weekly | Introduction of TL | Change in sales compared | A small sample of products |
the Food Standards product sales as a | labeling to ready meals | in the 4 weeks before and | over a short period of time | |
Authority-percentage of total | and sandwiches in the | after introduction of TL | showed that sales of ready | |
recommended TL sales by food product | United Kingdom | labels | meals increased following | |
labeling format had on category before and | introduction of TL labeling. | |||
food sales in a major after TL labeling | Percentage change in | |||
UK supermarket chain? | sales by product category | Sales of sandwiches did not | ||
after introduction | change significantly after TL | |||
of TL labeling was | labels were introduced. | |||
compared with the | ||||
relative healthiness of the | There was no association | |||
products | between the healthiness of | |||
the products and change in | ||||
sales. | ||||
Berning et al., 2011 | What is the effect of Comparison of product | Positive nutrition labels | Sales of labeled vs. | During the intervention |
nutrition labels that sales of popcorn with | affixed to grocery store | unlabeled popcorn were | period, sales for healthy | |
highlight specific nutrition shelf tag | shelves below boxes | analyzed from scanner | popcorn decreased while | |
positive nutrition labeling compared to a | of microwave popcorn | data | sales for unhealthy popcorn | |
standards on microwave dummy variable | in five chain stores in | increased (only significant | ||
popcorn sales? | California | at the 80 percent confidence | ||
level). |
Sutherland et al., | What is the effect Natural experiment | Purchasing data | Data on purchase changes | The Guiding Stars program |
2010 | of a comprehensive from a cross-sectional | was collected and | was reported for ready-to- | was found to be effective |
storewide supermarket sample of higher | assessed for a 2-year | eat cereals | at bringing about changes | |
point-of-purchase socioeconomic | period from the study | in food purchasing | ||
nutrition intervention shoppers at a major | population | behavior immediately | ||
using shelf-label 3-tier grocery chain; no | after intervention with | |||
star icons (Guiding comparison group | incremental improvement up | |||
Stars) on food and | to 2 years later. | |||
beverage choices? | ||||
Peer-Reviewed Non-Field Experiments | ||||
Borgmeier and | 1. Which signpost Choice experiment of a | Task 1: Pairwise | Consumers’ ability to | Signpost labels were more |
Westenhoefer, 2009 | label format enables convenience sample in | comparison of foods | differentiate healthier | helpful than no labels for |
consumers to Germany | to identify healthier | from less healthy | identifying healthier foods | |
identify healthier | choices | products by food label | and the MTL performed | |
from less healthy | format | better than the %GDA | ||
products? | Task 2: Simulated | and simple tick systems. | ||
shopping trip to test | Influence of food labels | However, even if a food’s | ||
2. What is the impact | food choice | on food choice and | perceived healthiness is | |
of food labels on | quality of diet | influenced by signpost | ||
food choice and | Experimental | labeling, it is unlikely to | ||
quality of diet? | conditions consisted | have a major impact on | ||
of four different label | actual food choice. | |||
formats: | ||||
–simple healthier | Higher sodium intake was | |||
choice tick | associated with higher | |||
–MTL | education in the TL and | |||
–% GDA | colored GDA conditions, | |||
–color % GDA | while the simple tick was | |||
Control = no label | associated with lower | |||
education. | ||||
For all conditions, the | ||||
average daily intake for | ||||
fat, saturated fat, sugar, | ||||
and sodium was above | ||||
recommended amounts. |
Peer-Reviewed Field Experiments | ||||
Intervention or | ||||
Source | Study Questions Study Design | Conditions | Outcomes Assessed | Summary of Results |
Feunekes et al., 2008 | 1. How do different Choice experiment in | Participants randomly | Perceived differences | Stars and smiley faces |
FOP labeling geographically and | assigned to three of | in healthiness between | showed the greatest | |
formats differ in culturally diverse | six labeling formats | healthier and less healthy | differences in perceived | |
helping consumers European populations | rated them on liking, | products | healthiness, followed by TL. | |
differentiate | comprehension, | |||
between healthier | credibility, and | Purchase intentions | Summary information was | |
and less healthy | perceived healthiness | not as trusted as other | ||
options? | formats. | |||
Labeling formats: | ||||
2. What effect does | Study: 1 Checkmark; | No differences were found | ||
labeling format | summary (1-7); stars, | in purchase intent between | ||
have on decision- | smiley faces; TL with | labeling formats. | ||
making when | words (no numbers); | |||
taking into account | circle, with numbers | Less time was spent making | ||
the shopping | and colors | decisions with simpler | ||
environment? | information formats. | |||
Study 2: Checkmark; | ||||
stars; multiple | Official endorsements tend | |||
checkmarks; % GDA | to increase the credibility of | |||
with and without | a labeling format. | |||
additional information | ||||
Kelly et al., 2009 | Which FOP labeling Choice experiment of a | Participants responded | Ability to discriminate | The TL system was the |
system is most effective convenience sample in | to questions about | healthier from less | most effective for assisting | |
in assisting consumers Australia | mock packages | healthy products by | consumers to identify | |
to make healthier, more | representing healthier | label condition and food | healthier products and make | |
informed food choices? Face-to-face shopping | and less healthy | product category | comparisons quickly and | |
center interviews with | options using three | easily. | ||
randomly selected | food product categories | Food choice preferences | ||
shoppers | (cereal, savory snacks, | There is a disjuncture | ||
frozen meals) and four | between FOP systems | |||
label conditions (TL; | initially perceived to | |||
TL + overall rating; | be easiest to use and | |||
monochrome % GDA; | consumers’ actual ability to | |||
color % DI) | interpret the systems. |
Gorton et al., 2009 | 1. What is the ability | Intercept survey of | Participants were | Survey responses were | The ability to estimate |
of shoppers from | shoppers recruited from | randomly selected and | analyzed to determine | nutrient content using the | |
different racial/ | supermarkets in New | geo-coded. Survey | participants’ use of | nutrition label was similar | |
ethnic groups to use | Zealand | questions included | nutrition labels, reasons | across ethnic and income | |
nutrition labels? | nutrition label | for non-use of labels, | groups, but the ability to | ||
information, special | basic understanding | use nutrition information to | |||
2. What are shoppers’ | dietary requirements, | and interpretation of | determine healthfulness of a | ||
preferences from | and socio-demographic | label information, and | food showed wide variation | ||
among four different | data | preference from among | among ethnic groups. | ||
label formats? | four label formats | ||||
(nutrition information | |||||
panel, TL, MTL, and | |||||
% DI). Responses were | |||||
assessed by racial/ethnic | |||||
group, i.e., Maori, Pacific | |||||
Islander, Asian, or New | |||||
Zealand European | |||||
Balcombe et al., | Are consumers willing | Choice experiment | Survey instrument | Consumers’ WTP for | Respondents were strongly |
2010 | to pay for reductions | based on a factorial | designed with 24 | products chosen from a | averse to red TL labels. |
in the various nutrients | design | choice sets (4 six- | mix of goods in a virtual | ||
as indicated by the | choice stets) | shopping basket | Price estimates for moving | ||
TL system, i.e., fat, | Randomly distributed | from red to green were | |||
saturates, sugar, and | questionnaire mailed to | Nutrients were | greater than moving from | ||
salt, in terms of a | UK households | characterized by TL | amber to green. | ||
basket of shopping? | labels | ||||
Men showed overall lower | |||||
WTP than women. | |||||
Households with children | |||||
and respondents with higher | |||||
education showed higher | |||||
WTP than those without | |||||
those attributes. |
Peer-Reviewed Field Experiments | ||||
Intervention or | ||||
Source | Study Questions Study Design | Conditions | Outcomes Assessed | Summary of Results |
Dunbar, 2010 | 1. What is the ability Choice experiment in | Study 1: Food choice | Consumers’ ability to | Participants given a food |
of consumers to a city marketplace in | based on food product | select a healthy food | product name only made | |
choose between the UK | name only or food | product; | the quickest selections; | |
alternate products | name with GDA label | Efficiency of product | however, when either a | |
in the context of | choice; | task-based interface or GDA | ||
composing a meal? | Study 2: –Defined task | Consumers’ ability to | information was introduced | |
(select a meal with <1 | compose a meal using | they made healthier choices. | ||
2. What is the quality | gram of salt) using a | label information; | Only the task-based interface | |
of consumer choices | label tuned to GDA | Quality of consumer | allowed participants to make | |
and the efficiency | label | food product selection | selections as quickly as the | |
with which choices | –Select the healthiest | based on nutrition label | food name only. | |
are made? | food product using the | information | ||
GDA label | ||||
Research on Consumer Preference | ||||
van Kleef et al., | 1. What is the extent Analysis of previously | Varied functional food | Participants rated all | Participants preferred |
2005 | to which consumers collected data from | concepts were offered | mini-concepts on four | physiology-based health |
perceive health a choice experiment | as a set of 100 claim- | dependent measures: | benefits over psychology/ | |
claims appropriate using Dutch shoppers | carrier “mini-concepts” | uniqueness, attractiveness, | behavior-based benefits. | |
with specific food selected on the basis of | credibility, and intention | Claims were best received | ||
products? their consideration of | to try | when attached to products | ||
health aspects of food | with a positive health image | |||
2. How are consumer when shopping | and health claim history. | |||
responses to health Data was collected in a | No evidence was found to | |||
claims affected market research facility | support the superiority of | |||
by alternative | enhanced function claims | |||
communication | over disease risk-reduction | |||
formats? | claim formats. |
Berning et al., 2010 | What are shoppers’ | Choice experiment | Color images of | Analyses of survey | Preference for high |
preferences for | using an intercept | shelf labels displayed | responses estimated | prominence nutrition | |
nutrition information | survey of shoppers | beneath a picture of a | shoppers’ preferences for | information correlated with | |
provided on grocery | outside a national | food product | three sources of nutrition | high prominence unit price | |
store shelf labels? | grocery store chain in | Images included | information: price, unit | and high prominence price | |
three different locations | variations in price | price, (displayed from | information. | ||
information, unit | low to high prominence), | ||||
price, and nutrition | and nutrition information | Preference for low | |||
information | (not present or low to | prominence nutrition | |||
high prominence) | information correlated with | ||||
low prominence unit price | |||||
and high prominence price | |||||
information. | |||||
Positive consumer | |||||
preferences for provision | |||||
of nutrition information on | |||||
grocery store shelves suggest | |||||
that stores and shoppers can | |||||
benefit from provision of | |||||
shelf-label information. |
engage participants in person, as they leave an actual supermarket, or to bring them into a laboratory-based setting. In both cases, participants are shown the FOP labels in person—either in a picture or on an actual product. The third setting is online, i.e., over the Internet, in which case participants view FOP labels via a computer monitor. A limitation to all of these approaches is that participants may not make the choice that they would in an actual in-store shopping environment. In fact, they likely take more time to choose than they would in an actual environment, are not considering price, and are aware that their responses are being examined as part of a research study.
The studies reviewed below employed a large range of FOP symbol systems. This variability provides insight into differences among FOP symbol types but makes comparisons across studies more difficult. Moreover, the simulated food packages are often much simpler than the actual food packages, that is, the FOP symbol is often shown on a plain package stripped of all other information and marketing normally found on products.
Several sample-based considerations could influence the committee’s interpretation of the results. First, the samples are often “convenience samples,” or individuals who are not randomly or otherwise selected. Often, such samples consist of individuals who are leaving supermarkets, arguably the individuals from whom investigators would be most likely to gather data on responses to FOP systems. However, it is possible that not all stores in a geographic area are sampled or not all consumers consent to a survey, which may exclude a large cross-section of the population, spanning children and other vulnerable groups, such as low-income and certain racial/ethnic groups.
Non-field studies utilize several distinct outcome measures. Of the studies examined, the outcomes most relevant to how consumers might respond to an FOP symbol system are those where consumers make an actual hypothetical choice among products or note their intent to purchase a particular product. However, when Wansink and Ray (1992) compared measures of brand attitude and consumption intention, they found that attitude toward a product was a weak predictor of consumption. Aikman et al. (2006) examined relationships between perception of healthfulness of foods, attitudes, and eating behavior and found that consumers are either not aware or do not use nutrition information when making decisions about the healthfulness of foods, and beliefs about the healthfulness of foods are not related to the frequency of consumption.
Less telling, but also potentially important, are consumers’ abilities to choose the healthier product from two or more products. The committee examined consumer preferences for various FOP symbol systems. Chocarro et al. (2009) and Barreiro-Hurle et al. (2010) determined from their studies of consumer label use, nutrition knowledge, and consumer food choices that knowledgeable consumers are more likely than other consumers, particularly price-conscious consumers, to choose healthier foods from among a variety of product options.
Analysis of Evidence from Peer-Reviewed Studies That Are Not Field Experiments
Several experimental laboratory studies were initiated to provide evidence regarding which FOP label format is best understood by consumers. Borgmeier and Westenhoefer (2009) conducted a randomized experimental study of 420 consumers to determine how well different FOP nutrition labels worked. The researchers considered four label formats: simple tick, TL format, monochrome Guideline Daily Amount (GDA), and color-coded GDA. There was also a control condition for which no nutrition information was provided. The simple tick was similar to the Smart ChoicesTM icon (see Phase I report, Table S-1) that was used briefly in the United States. Participants were asked to complete two tasks. First, they were asked to select the most healthful product in each pair from 28 pairs of food products. Then, in a simulated shopping experience, they were asked to select all the foods and drinks that they would consume during the next day.
Results from the paired comparison task indicate that the TL format was associated with the highest percentage of correct choices. That is, consumers correctly identified the most healthful product (24.8 out of 28 pairs) when nutrition information was presented via a TL system. However, the different food label formats did not influence consumers’ ultimate choice of foods. In all experimental groups, the average daily intakes for fat, saturated fat, sugar, and sodium were above the recommended daily amounts. Thus, although the TL system helped consumers identify the most healthy food options, it did not influence consumers’ actual choices. The results of both tasks were consistent across different demographic segments (Borgmeier and Westenhoefer, 2009).
Feunekes et al. (2008) conducted a similar study. This two-part study examined the influence of eight FOP nutrition labeling formats that differed in complexity, from relatively simple to more complex, in terms of the
amount and type of information provided. Whereas the simpler formats provided an interpretation of the overall healthfulness of the product, the more detailed formats provided judgments of the healthfulness of each nutrient. Participants, selected from four European countries, evaluated healthy and less healthy foods from the same product category.
In the first part of the study, six labeling formats were used: Healthier Choice Tick, Health Protection Factor, smiley faces, stars, Multiple Traffic Light (MTL), and Wheel of Health. The Healthier Choice Tick was a single tick and was only present on the healthier product in the pair. Stars, smiley faces, and the Health Protection Factor provided “grades” for the products. A product could be awarded one to five stars or smiley faces. The Health Protection Factor was derived from the system used to rate sunscreen lotions; products were assigned a number from 1 to 7, with higher numbers indicating a healthier product. The MTL presented information on five key nutrients (energy, total fat, saturated fatty acids, sugar, and salt). Like other iterations of the MTL, each nutrient was given a score of low (green), medium (amber), or high (red), which was indicated by both color and text. The Wheel of Health was based on a system used by the UK retailer Sainsbury’s. The label provided the exact amount of the five key nutrients in a pie-chart format, with each slice of the pie colored green (low), amber (medium), or red (high). Participants evaluated the different labeling formats for their ease of understanding. Results indicated that participants found all the formats easy to understand, relatively credible, and likeable and that participants were significantly better able to differentiate between the healthy and unhealthy products when the simpler, graded smiley faces and stars formats were used. Also of interest were the effects of label endorsement: Participants reported that the labels were significantly more credible when endorsed by international or national organizations (Feunekes et al., 2008).
In a second study in the same paper, the investigators introduced two additional different label formats, a multiple choice tick and a GDA format. As in the first study, they found that all formats helped consumers better differentiate between healthy and unhealthy products. However, consumers took the longest time to evaluate the products when the GDA format was presented. Consequently, the investigators concluded that simpler FOP labeling formats such as Healthier Choice Tick or stars may be more effective in helping consumers make healthier choices (Feunekes et al. 2008).
Kelly et al. (2009) conducted a similar study in Australia to examine the effects of format on consumers’ evaluations of FOP labeling systems, by examining a TL system and a variation of the TL system in which an overall rating of the product was also included. They also tested a monochrome Percent Daily Intake (% DI) and a modified % DI (M-% DI) labeling format. In these formats, the percentage dietary contribution from energy, protein, total fat, saturated fat, total carbohydrate, sugar, fiber, and sodium for an average adult was presented. In the modified % DI format, the indicator color for each nutrient (green, amber, or red) was presented in addition to the number. Most participants (90 percent) believed that consistent FOP labeling across all food products would be the easiest to understand. Furthermore, they were best able to identify the healthier product when presented with the TL system ranking levels of total fat, saturated fat, sugar, and sodium as either high, medium, or low and assigned a red, amber, or green color-code, respectively. Participants had the most difficulty differentiating products when the M-% DI format was used; 64 percent identified the healthier option, compared to 81 percent with the TL system.
Although the above studies focus on hypothetical choices of products, their findings are similar to those of other research examining outcomes such as understanding the various FOP labels. For example, Gorton et al. (2009) surveyed consumers shopping in a supermarket in New Zealand to assess their understanding of different FOP labeling schemes. In this study, shoppers were presented with a series of questions that assessed preference for and understanding of four nutrition label formats: MTL, simple traffic light (STL), the mandatory Nutrition Information Panel (NIP), and % DI. Consumers were best able to understand the STL and the MTL formats. This study, however, did not examine the influence of consumer preference for purchasing one food over the other in response to labeling format.
Balcombe et al. (2010) conducted a survey-based choice experiment to examine consumer response to the UK TL system. The investigators used as an outcome willingness to pay (WTP) for reductions in fat, saturated fat, sugar, and salt in food products. They found that participants had a very strong preference, reflected in their WTP, for avoiding a market basket containing foods with “red” lights.
Dunbar (2010), in a two-part study, examined consumers’ ability to choose between alternative products in
the context of composing a meal and assessed the quality of consumers’ food choices as well as the efficiency with which choices are made. In the first study, participants recruited from a city marketplace in the UK were asked to choose the healthiest product from among a selection of products displaying either product name only or the GDA panel plus the product name. In the second study, participants were given additional instructions for a specific task and a new condition using a label that also included a task (called a task-based interface). They were asked to choose a product that (1) could be used to make a meal that was low in salt and (2) was the overall healthiest choice. The participants in the first study made faster decisions when given the product name only but made better overall choices when given the GDA label with nutrition information. Interestingly, participants given the GDA label did not significantly reduce the levels of salt chosen in the “pick the healthiest” task. Participants in the second study made faster decisions when given a simplified rather than a more complex GDA text label. Furthermore, they were significantly better able to reduce the amount of salt in their selections with the simplified text label compared to the product name only. There was, however, no significant improvement in making healthier choices with the simplified text label when the label included the “pick the healthiest” task.
Research on Consumer Preference
One study examined factors that influence consumers’ preference for labeling formats. Using a face-to-face interview of shoppers outside a national grocery store chain, Berning et al. (2010) examined shoppers’ preferences for nutrition information provided on grocery store shelf labels. In the choice experiments, color images of shelf labels were displayed below a picture of a food product. The images presented variations of price information, unit price, and nutrition information (total fat, saturated fat, calories, cholesterol, sugar, and sodium either with “low” or “high” prominence). The results showed positive consumer preferences for the provision of nutrition information on grocery store shelf labels and suggested benefits for both stores and shoppers from the provision of shelf-label nutrition information. These benefits include alignment of incentives between stores (providing nutrition information may increase sales) and shoppers (seeking nutrition information on products) and identification of shopper product preferences following the introduction of shelf tag information for certain products in stores.
As noted previously, the committee gave only minimal weight to information from applied marketing sources, the type of research at the lowest level of the evidence hierarchy. However, the committee determined that including this type of evidence as a component of the totality of evidence to consider was important because it provided additional insight into how consumers perceive and may use FOP labeling. The Phase I report outlined plans for Phase II by describing a multifaceted approach that would include information from relevant consumer behavior literature, experts from relevant fields, and research on FOP undertaken by FDA (IOM, 2010). In lieu of undertaking its own research, the committee could use already existing research on the usability of labels by population subgroups. Information provided the committee in a public workshop (see Appendix F) as well as applied research identified by the committee formed the core of this evidence.
Research from FDA
The committee reviewed research performed by FDA that directly examined FOP food labeling using an approach similar to the laboratory-based experiments described above (Lin and Levy, 2010). In the first of two studies, 2,424 subjects in an online convenience panel were randomized to see a number of different FOP systems—the National Facts panel (NFP), the Smart Choices symbol, a TL system, and a Nutrition Highlights system, similar to the Nutrition Keys1 program developed by the Grocery Manufacturers’ Association (GMA). The selection of healthier products was the key outcome, and the results varied depending on which food category was examined. There were no statistically significant differences for snacks, and for meals the NFP had the highest percentage
1 Now called “Facts up Front.”
of correct responses. For cereals, the NFP and TL system tended to perform better, although results were not consistently statistically significant across all comparisons. The committee concluded, therefore, that differences in results by product category call into question the extent to which these findings can be generalized.
Study 2 examined 4,901 participants, also from an online convenience sample. In this study, a larger number of FOP systems were examined, and the outcome was the consumer choice between two products—either the healthier or less healthy product. Decisions among product categories were broken down into hard versus easy choice, with and without the presence of a health cue. When the choices were easy, there were no differences among the FOP systems in choosing a healthy product when the health cue was present, and only small differences without this prompt. For more difficult choices, in general the NFP was most helpful to consumers in choosing a healthy product, which the authors attributed partially to its overall familiarity. No statistically significant differences emerged with the other FOP system.
Research from the United Kingdom
In a study initiated by the Food Standards Agency comparing different TL systems, the British Market Research Bureau (BMRB, 2009) found that consumers’ levels of comprehension of different FOP labels were generally high. Of the two labels with the highest overall levels of comprehension, one combined text (the words high, medium, and low), TL colors, and a % GDA and the other combined text and TL colors. The investigators concluded that the coexistence of a variety of different FOP label formats in the marketplace can be confusing to consumers.
Research from Industry or Stakeholder Groups
Industry and stakeholder groups provided the committee with information about sales data that directly related to testing of various FOP systems and that addressed key questions. This information included shelf tag as well as FOP systems data. The American Heart Association provided data from a field-based study that highlighted the Heart Check program on shelf tags of products participating in the program.Chapter 4.
The GMA-IFIC used an online survey to test comprehension, communication, and interpretation of a potential FOP nutrition information system (Smith-Edge and Hildewine, 2010). The key findings identified from this survey were that increasing the amount of nutrition information on the front of packages strengthened consumers’ comprehension and comfort level with the information; consumers viewed the NFP less often when they were asked to find specific information that was available on the FOP; consumers who were provided with calories plus negative and positive nutrients were more likely to agree that the FOP nutrition information was helpful with decision-making and understanding than those provided with calories only; and across all labeling systems tested and for all product categories, a majority of consumers were able to select products considered to be healthier.
2 Consumer Marketing Research, Heart-Check Mark. Submitted by Dennis Milne, American Heart Association, October 15, 2010.
3 Information Research, Inc., Letter to the Committee on Examination of Front-of-Package Nutrition Rating Systems and Symbols. Submitted by Annette Maggi, NuVal, March 2010.
Findings
Overall, the evidence regarding the effects of FOP systems is not comprehensive. From the limited set of real-world studies reviewed, no single system emerged as the absolute “best.” Looking at the evidence more broadly, when comparing across studies or when comparing multiple systems within the same study, some limited evidence emerges that shows the simpler systems to be more effective in encouraging healthier choices. The lack of research on children and vulnerable populations is noteworthy. In addition, as noted in Chapter 4, when food choice is constrained by economic considerations, healthier food choices will likely receive little attention if they are not affordable.
Given the paucity of evidence from the peer-reviewed literature, the additional evidence obtained from applied market research, including that of FDA as well as the British Market Research Bureau and food manufactures, provided additional insight, not available from other sources, into how consumers perceive and use FOP labeling.
Conclusions
The committee concluded that research on FOP symbol systems is limited. No single FOP symbol system is supported by evidence of its superiority to all others, and FOP systems alone as currently developed do not show consistent evidence of dramatically influencing consumer choice. However, there is some limited evidence that FOP systems that are simple and easy to understand more effectively encourage choices of healthier products, particularly in the real-world settings where consumers make decisions quickly in a larger context of choices.
Aikman, S. N., K. E. Min, and D. Graham. 2006. Food attitudes, eating behavior, and the information underlying food attitudes. Appetite 47:111-114.
Balcombe, K., I. Fraser, and S. DiFalco. 2010. Traffic lights and food choice: A choice experiment examining the relationship between nutritional food labels and price. Food Policy 35:211-220.
Barreiro-Hurle, J., A. Gracia, and T. de-Magistris. 2010. Does nutrition information on food products lead to healthier food choices? Food Policy 35:221-229.
Berning, J. P., H. H. Chouinard, K. C. Manning, J. J. McCluskey, and D. E. Sprott. 2010. Identifying consumer preferences for nutrition information on grocery store shelf labels. Food Policy 35:429-436.
Berning, J. P., H. H. Chouinard, and J. J. McCluskey. 2011. Do positive nutrition shelf labels affect consumer behavior? Findings from a field experiment with scanner data. American Journal of Agricultural Economics 93:364-369.
BMRB (British Market Research Bureau). 2009. Comprehension and Use of UK Nutrition Signpost Labelling Schemes. London: British Market Research Bureau.
Borgmeier, I., and J. Westenhoefer. 2009. Impact of different food label formats on healthiness evaluation and food choice of consumers: A randomized-controlled study. BMC Public Health 9:184.
Chocarro, R., M. Cortinas, and M. Elorz. 2009. The impact of product category knowledge on consumer use of extrinsic cues—A study involving agrifood products. Food Quality and Preference 20:176-186.
Dunbar, G. 2010. Task-based nutrition labelling. Appetite 55:431-435.
Feunekes, G. I. J., I. A. Gortemaker, A. A. Willems, R. Lion, and M. van de Kommer. 2008. Front-of-pack nutrition labelling: Testing effectiveness of different nutrition labelling formats front-of-pack in four European countries. Appetite 50:57-70.
Gorton, D., C. Ni Mhurchu, M.-H. Chen, and R. Dixon. 2009. Nutrition labels: A survey of use, understanding and preferences among ethnically diverse shoppers in New Zealand. Public Health Nutrition 12:1359-1365.
IOM (Institute of Medicine). 2010. Front-of-Package Nutrition Rating Systems and Symbols: Phase I Report. Washington, DC: The National Academies Press.
Kelly, B., C. Hughes, K. Chapman, J. C.-Y. Louie, H. Dixon, J. Crawford, L. King, M. Daube, and T. Slevin. 2009. Consumer testing of the acceptability and effectiveness of front-of-pack food labelling systems for the Australian grocery market. Health Promotion International 24:120-129.
Levy, A. S., O. Mathews, M. Stephenson, J. E. Tenney, and R. E. Schucker. 1985. The impact of a nutrition information program on food purchases. Journal of Public Policy & Marketing 4:1-13.
Lin, C-T. J., and A. Levy. 2010. Food and Drug Administration Front of Pack Consumer Research. Presented at Workshop on Consumer Behavior Research and Front of Package Nutrition Rating Systems and Symbols: What Do Consumers Know, Understand and Use? Institute of Medicine. Washington, DC.
Sacks, G., M. Rayner, and B. Swinburn. 2009. Impact of front-of-pack “traffic-light” nutrition labelling on consumer food purchases in the UK. Health Promotion International 24:344-352.
Smith-Edge, M., and R. Hildewine. 2010. International Food Information Council Foundation Front of Pack Labeling Consumer Research Project. Presented at Workshop on Consumer Behavior Research and Front of Package Nutrition Rating systems and Symbols: What Do Consumers Know, Understand and Use? Institute of Medicine. Washington, DC, October 26.
Sutherland, L. A., L. A. Kaley, and L. Fisher. 2010. Guiding Stars: The effect of a nutrition navigation program on consumer purchases at the supermarket. American Journal of Clinical Nutrition 9(Suppl):1090S-1094S.
van Kleef, E., H. C. M. van Tripp, and P. Luning. 2005. Functional foods: Health claim-food product compatibility and the impact of health claim framing on consumer evaluation. Appetite 44:299-308.
Wansink, B., and M. L. Ray. 1992. Estimating an advertisement’s impact on one’s consumption of a Brand. Journal of Advertising Research 26:9-16.
This page intentionally left blank.