Advertisement

Single-Larger-Portion-Size and Dual-Column Nutrition Labeling May Help Consumers Make More Healthful Food Choices

Published:January 23, 2013DOI:https://doi.org/10.1016/j.jand.2012.11.006

      Abstract

      Background

      The Food and Drug Administration is considering changes to the Nutrition Facts label to help consumers make more healthful choices.

      Objective

      To examine the effects of modifications to the Nutrition Facts label on foods that can be listed as having 1 or 2 servings per container, but are reasonably consumed at a single eating occasion.

      Design

      Participants were randomly assigned to study conditions that varied on label format, product, and nutrition profile. Data were collected via an online consumer panel.

      Participants/setting

      Adults aged 18 years and older were recruited from Synovate's online household panel. Data were collected during August 2011. A total of 32,897 invitations were sent for a final sample of 9,493 interviews.

      Intervention

      Participants were randomly assigned to one of 10 label formats classified into three groups: listing 2 servings per container with a single column, listing 2 servings per container with a dual column, and listing a single serving per container. Within these groups there were versions that enlarged the font size for “calories,” removed “calories from fat,” and changed the wording for serving size declaration.

      Main outcome measures

      The single product task measured product healthfulness, the amount of calories and various nutrients per serving and per container, and label perceptions. The product comparison task measured ability to identify the healthier product and the product with fewer calories per container and per serving.

      Statistical analyses performed

      Analysis of covariance models with Tukey-Kramer tests were used. Covariates included general label use, age, sex, level of education, and race/ethnicity.

      Results

      Single-serving and dual-column formats performed better and scored higher on most outcome measures.

      Conclusions

      For products that contain 2 servings but are customarily consumed at a single eating occasion, using a single-serving or dual-column labeling approach may help consumers make healthier food choices.

      Keywords

      NUTRITION INFORMATION IS REQUIRED ON MOST packaged foods and this information must be provided in a specific format as defined by the Food and Drug Administration (FDA) in the Code of Federal Regulations.
      Food labeling: Mandatory status of nutrition labeling and nutrient content revision, format for nutrition label. 58 Federal Register 2079.
      The main features of the label format include the serving size and number of servings per container, the number of calories per serving, the amount per serving in grams (or micrograms) of key macronutrients, and a percent daily value to help consumers understand how much 1 serving of the particular food contributes to a daily diet. Research has shown that most consumers use labels as least some of the time
      • Choiniére C.J.
      • Lando A.
      and that when presented with Nutrition Facts (NF) labels, consumers can make more healthful choices.
      • Ollberding N.J.
      • Wolf R.L.
      • Contento I.
      Food label use and its relation to dietary intake among US adults.
      • Temple J.L.
      • Johnson K.
      • Recupero K.
      • Suders H.
      Nutrition labels decrease energy intake in adults consuming lunch in the laboratory.
      In response to the continued high levels of obesity in the United States,
      • Flegal K.M.
      • Carroll M.D.
      • Kit B.K.
      • Ogden C.L.
      Prevalence of obesity and trends in the distribution of body mass index among US adults, 1999-2010.
      • Ogden C.L.
      • Carroll M.D.
      • Kit B.K.
      • Flegal K.M.
      Prevalence of obesity and trends in body mass index among US children and adolescents, 1999-2010.
      the FDA has been considering changes to the food label to help consumers eat a more healthful diet and maintain a healthy weight. In 2003, FDA established an internal Obesity Working Group. As a result of the ideas suggested by the group, FDA issued two Advance Notices of Proposed Rulemaking in 2005 requesting comments on format changes to the NF label. One Advance Notice of Proposed Rulemaking requested comments on whether and, if so, how to give greater emphasis to calories on the NF label, including by removing “calories from fat,”
      Food labeling: Prominence of calories. 70 Federal Register 17008.
      and the other requested comments on whether and, if so, how to amend the agency's serving size regulations.
      Food labeling: Serving sizes of products that can reasonably be consumed at one eating occasion; updating the reference amounts customarily consumed; approaches for recommending smaller portion sizes. 70 Federal Register 17010.
      Under current regulations,
      Food labeling: Mandatory status of nutrition labeling and nutrient content revision, format for nutrition label. 58 Federal Register 2079.
      there is some discretion for determining serving sizes for packaged foods that contain >1 serving but that are reasonably consumed in a single eating occasion. Examples of these types of products include a 20-oz soda and a “grab bag” of chips. The serving size amounts for these foods are based on reference amounts customarily consumed (RACC) and are defined for a variety of food categories.
      Food labeling: Mandatory status of nutrition labeling and nutrient content revision, format for nutrition label. 58 Federal Register 2079.
      When the product contains between 100% and 200% of the RACC, it must be labeled as a single serving, but when it contains more than 200% of the RACC, the food may be labeled as 1 serving if it can be reasonably consumed in a single eating occasion. Products that have large RACCs (100 g or 100 mL or larger) and that contain between 150% and 200% of the RACC may be labeled as 1 or 2 servings. For example, the RACC for soda is 240 mL (8 fl oz). A soda container with >480 mL (16 fl oz) may be labeled as 1 serving if it can be reasonably consumed in a single eating occasion, but otherwise must be labeled as 2 or more servings.
      Recent research has suggested that environmental factors, including the size of packages, may significantly influence how much consumers eat.
      • Wansink B.
      Environmental factors that increase the food intake and consumption volume of unknowing consumers.
      Consumers may not fully consider serving size information when using food labels and may, therefore, make inappropriate conclusions about the nutrient content of a packaged food.
      • Mohr G.S.
      • Lichtenstein D.R.
      • Janiszewski C.
      The effect of marketer-suggested serving size on consumer responses: The unintended consequences of consumer attention to calorie information.
      • Pelletier A.L.
      • Chang W.W.
      • Delzell J.E.
      • McCall J.W.
      Patients' understanding and use of snack food package nutrition labels.
      Two labeling remedies have been suggested to make the actual nutrient content for these products more obvious to consumers: a dual-column approach that highlights that there is >1 serving per container by presenting nutrition information for 1 serving and for the entire container, and changing the labeling rules to require these products to declare nutrition information for the entire container (ie, define the entire container as a single serving). Dual-column labeling formats and labeling these types of products as a single serving were rated favorably in a focus group study.
      • Lando A.M.
      • Labiner-Wolfe J.
      Helping consumers make more healthful food choices: Consumer views on modifying food labels and providing point-of-purchase nutrition information at quick-service restaurants.
      Other research has also suggested that dual-column or single-serving labeling may help consumers make more informed choices.
      • Mohr G.S.
      • Lichtenstein D.R.
      • Janiszewski C.
      The effect of marketer-suggested serving size on consumer responses: The unintended consequences of consumer attention to calorie information.
      • Antonuk B.
      • Block L.G.
      The effect of single serving versus entire package nutritional information on consumption norms and actual consumption of a snack food.
      However, no quantitative study has directly examined how these two remedies compare to alternative remedies. Our study presents the results of an online experiment evaluating the effect of format changes on products that have 1 or 2 servings per container but that are customarily consumed at a single eating occasion.

      Methods

       Study Design and Procedures

      The experiment used a 10 (label format)×2 (product category)×2 (nutritional profile per category—one fixed as more healthful than the other) between-subjects design. The 10 labeling formats shown in Figure 1 can be classified into three groups: listing 2 servings per container with a single column (“two servings, single-column”), listing 2 servings per container with a dual column (“two servings, dual-column”), and listing 1 serving per container (one serving, single-column). There were five formats in the two servings, single-column grouping: 1) current NF label “control”; 2) current label, but without “calories from fat”; 3) current label, without “calories from fat” and with the font for calories enlarged; 4) changed wording for serving size declaration to emphasize that there were two servings per container and removed calories from fat; and 5) dual listing for calories, in which the calories per serving and calories per container were declared, but the remaining nutrients were declared only per serving, and calories from fat was removed. The two servings, dual-column grouping consisted of three similar label formats: 6) all information (ie, calories, weight amounts, and percent daily values [%DVs]) for a single serving and for the full container appeared in separate columns; 7) same as the previous dual column but without “calories from fat”; and 8) a dual column in which only the calories and %DVs for a single serving and for the full container appeared in separate columns (without “calories from fat”). There were two label formats in the one serving, single-column grouping, both of which labeled the product as having a single, large serving: 9) like the control label, but without “calories from fat”; and 10) like the control label, but without “calories from fat” and with the font for calories enlarged. The two product categories were a frozen meal and a “grab bag”–sized bag of chips. Within each product category there were two products, one that was more healthful and one that was less healthful. The nutrient profiles of the products were based on frozen meal and chip products found on the market. The formats and nutrition profiles are shown in Figure 1 and Table 1, respectively.
      Figure thumbnail gr1
      Figure 1Label formats shown to participants in a study to examine whether modifications to the Nutrition Facts label can help consumers make more healthful choices. Full Nutrition Facts labels were shown to participants but have been truncated in this Figure below the gray line to save space. Arrows did not appear on the labels shown to participants, but have been added here to emphasize changes.
      Table 1Nutrition profiles of the four food products shown to participants in a study to examine whether modifications to the Nutrition Facts label can help consumers make more healthful choices
      NutrientChips AChips BFrozen Meal AFrozen Meal B
      g or mg%DV
      %DV=percent daily value.
      g or mg%DVg or mg%DVg or mg%DV
      Calories140180220300
      Calories from fat60804080
      Total fat810101557913
      Saturated fat15282104.523
      Trans fat0000
      Cholesterol (mg)00001553010
      Sodium (mg)1205240102401060025
      Total carbohydrate19625735123010
      Dietary fiber31016624416
      Sugars2378
      Protein22911
      Vitamin A5252
      Vitamin C00204
      Calcium022015
      Iron2086
      a %DV=percent daily value.
      Participants were invited by e-mail to a website to complete the study online. After accessing the website, participants were randomly assigned to an experimental condition and were asked to complete two sets of tasks, which were followed by a set of questions about participants' use of food labels in general and demographics. For the first task set (“Single Product Task”), participants viewed one product NF label (randomly assigned from 40 possible combinations of product×nutrition profile×format) and answered a series of questions about the information shown on that NF label. For the second task set (“Choice Task”), participants were shown a pair of product NF labels (either for two frozen meals or for two bags of chips). One label showed the more healthful nutrition profile, and the other showed the less healthful profile. Depending on the experimental condition assigned, the label formats for both products could be the same or could be different. The purpose of the Choice Task was to explore how participants' ability to compare products might be affected by the NF label format modifications in two scenarios: when comparing labels of the same format, and when comparing labels with different formats. For the latter scenario, representative label formats (label numbers 2, 3, 7, 9, and 10 in Figure 1) were selected based on input provided from nutrition experts within FDA. The order of the Single Product Task and Choice Task was counterbalanced, as was the screen position (right or left) where the more healthful product appeared in the Choice Task. In addition, the product category viewed in the Single Product Task and Choice Task was counterbalanced: participants who saw a NF label for a frozen meal in the Single Product Task saw NF labels for chips in the Choice Task, and vice versa.
      Before conducting the experiment, two sets of six in-person cognitive interviews and two sets of 50 online pretests were conducted to ensure that the study questionnaire and stimuli were understandable, that respondents could provide informative answers, and that the overall study time averaged about 15 minutes. Cognitive interview participants were selected from a list maintained by the interview facility, and pretest participants (who were not included in the final study) came from the ePanel described below.

       Participants

      Participants for this study were recruited from Synovate's Global Opinion Panels, Internet ePanel, a commercially available online research panel of 1 million households and 2 million individuals. This panel was chosen because it covers a wide range of US consumers and has been used before for similar types of experimental studies.
      • Labiner-Wolfe J.
      • Lin C.T.J.
      • Verrill L.
      Effect of low-carbohydrate claims on consumer perceptions about food products' healthfulness and helpfulness for weight management.
      • Lin C.T.J.
      How do consumers interpret health messages on food labels?.
      Panel members are volunteers who are aged 18 years or older, mostly recruited though online marketing programs and referrals from existing members. Synovate maintains the panel on an ongoing basis by monitoring for and removing overused households, poor-returning households, and households from which surveys are returned as undeliverable. Panelist demographic and contact information are routinely updated throughout the year. Panel members typically participate in 12 to 14 studies per year. Although not paid for their participation in specific surveys, panel members are offered incentives in the form of sweepstakes entries (where winners can receive cash prizes between $10 and $500 in a 1-month period) and a redeemable points reward program (where 1,000 points=$1) for their participation.
      The survey program incorporated multiple levels of data security. First, the website at which the survey could be accessed was secured and only allowed respondents with a valid ID and passcode to enter. Upon entering, participants were encouraged to complete the survey in a single session, but were permitted to exit before finishing and to re-enter the survey (again, only with their ID and passcode) at the point where they left off. However, once participants completed the questionnaire, the passcode could not be re-used, nor could the entered data be altered. The website was firewalled so that respondents could enter only the survey for which their ID and passcode were assigned.
      Study invitations were sent out in cycles and were targeted to panelists to produce a sample that reflected the US population. By periodically examining the incoming sample for demographic criteria, additional e-mails could be broadcast in a staggered manner to increase the likelihood that the final sample matched the 2010 US census for sex, age, education, and race/ethnicity, region, income, and household size. Data collection was carried out during August 2011. A total of 32,897 study invitations were sent for a final sample of 9,493 completed interviews for a completion rate of 28.8%. The study protocol was approved under exempt review by the FDA Institutional Review Board.

       Measures

      There were four main outcome measures for the Single Product Task, all employed in or adapted from prior research.
      • Choiniére C.J.
      • Lando A.
      • Levy A.S.
      • Fein S.B.
      Consumers' ability to perform tasks using nutrition labels.
      • Levy A.S.
      • Fein S.B.
      • Schucker R.E.
      More effective nutrition label formats are not nececessarily preferred.
      • Levy A.S.
      • Fein S.B.
      • Schucker R.E.
      Performance characteristics of seven nutrition label formats.
      The first measure was participants' overall rating of product healthfulness within the product category (chips or frozen meal) on a 5-point scale (1=not at all healthy, 5=very healthy) (“Assume you were going to eat a frozen meal [chips], how healthy of a choice would this frozen meal [chips] be?”). The second measure (“total correct”) was the number of correct responses to eight questions about the nutrient content of the product (eg, “How many grams of total fat are in one serving of these chips?” and “How many grams of dietary fiber are in the whole container of these chips?”). An index was calculated by adding one point for each correctly answered question (range=0 to 8; Cronbach's α=.84). The third measure focused on a single item from the eight nutrient content questions, namely, whether participants correctly identified the number of total calories per container. We analyzed this item independently to determine how the label formats specifically affected participants' ability to determine the number of calories associated with consuming an entire container of food. The fourth measure was a five-item scale assessing participants' evaluation of the label format itself. The items in this label perception index asked participants to rate (1=not at all, 5=very) how useful, trustworthy, helpful, and the like they considered the label format to be (Cronbach's α=.82).
      Participants could answer “Don't Know” to any of the five perception questions. Such responses were omitted in constructing the perception index; therefore, the number of questions answered could be less than five.
      There were three main measures for the Choice Task. The first measure was based on participants' responses to the question, “Based on what you can see on the labels, if you wanted to buy the healthier product, which of these two products would you select?” If participants chose the product with fewer calories, less fat and sodium, and more positive nutrients, then the answer was considered correct; otherwise, the answer was considered incorrect. The second measure asked participants to identify which product in the pair had the fewest calories per container, and the third measure asked participants to select the product with fewer calories per serving.
      Because previous research has shown that individual differences in general label use and demographics can influence label use in a specific situation
      • Ollberding N.J.
      • Wolf R.L.
      • Contento I.
      Food label use and its relation to dietary intake among US adults.
      • Lin C.T.J.
      • Lee J.Y.
      • Yen S.T.
      Do dietary intakes affect search for nutrient information on food labels?.
      participants were asked how often they use the NF label when deciding to buy a food product (1=never, 4=often). Demographic variables included: age, sex, level of education, and race/ethnicity.

       Statistical Analysis

      Four analysis of covariance models (ANCOVAs) were estimated for the Single Product Task and three ANCOVAs for the Choice Task. Each dependent variable was modeled as a function of label format, product (chips or frozen meal), nutrition profile (more or less healthful for the Single Product Task), and covariates (ie, general label use, age, sex, level of education, and race/ethnicity). A Tukey-Kramer test set at the 0.05 level was used to compare the adjusted means for each label format. The analysis was conducted in SAS 9.2 (1999, SAS Institute, Inc).

      Results

       Single Product Task

      Table 2 shows the sample characteristics. More than 70% of participants report using the NF label either sometimes or often. This is similar to the results from the 2008 FDA Health and Diet Survey.
      • Choiniére C.J.
      • Lando A.
      Table 2Descriptive sample characteristics of subjects (N=9,493) in a study to examine whether modifications to the Nutrition Facts label can help consumers make more healthful choices
      CharacteristicResponse
      Mean±SD
      SD=standard deviation.
      Age (y)46±15.5
      Body mass index28.5±7.1
      n%
      Sex
      Female4,65149.1
      Race/ethnicity
      White, non-Hispanic6,23965.7
      Black/African American1,12411.8
      Hispanic/Latino1,31713.9
      Asian, American Indian, Pacific Islander6036.4
      Other/no answer2102.2
      Education
      Less than high school2793.0
      Completed high school3,75639.8
      Some college2,55827.1
      Completed college2,84330.1
      Label use frequency
      Never8218.7
      Rarely1,68017.9
      Sometimes3,11333.1
      Often3,78840.3
      a SD=standard deviation.
      The ANCOVA results for the four outcome measures are shown in Table 3. In the Single Product Task, the only label format modification that appeared to consistently affect healthfulness ratings was the modification that defined the entire container as a single serving (formats 9 and 10 in Figure 1): participants who saw products labeled as having 1 serving per container rated the product less healthful than participants who saw products labeled with any other format, including the current label control, that indicated the product had 2 servings per container (P<0.05). Among the label formats indicating 2 servings per container, there were not many statistically significant different mean scores on healthfulness. However, the two servings, dual-column label formats had slightly lower healthfulness scores than label formats that did not show the nutrition information for the entire package, although most of these differences were not statistically significant.
      Table 3Mean ratings of healthfulness, label perception, total correct index, and percent correct calories per container given by subjects in a study to examine whether modifications to the Nutrition Facts label can help consumers make more healthful choices
      Label formatHealthfulness rating
      Healthfulness rating ranges from 1(not at all healthy) to 5(very healthy).
      (n=8,660)
      Total correct index
      Total correct index is based on eight questions. For each question the respondent could get it correct or incorrect. The index is the sum of the correct answers and can range from 0 to 8.
      (n=9,219)
      Percent correct total calories
      Percent correct for total calories is based on being able to correctly identify the number of calories per container.
      (n=9,219)
      Label perception index
      Label perception index is based on five questions each with a 5-point scale. The index is divided by the number of questions answered such that it ranges from 1 (low perception) to 5 (high perception).
      (n=9,098)
      A. Two servings, single-column formats
      1. Current label2.91
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      4.43
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      63
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.65
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      2. Remove calories from fat2.85
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      4.52
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      67
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.63
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3. Enlarge calories and remove calories from fat2.91
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      4.59
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      67
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.62
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      4. Emphasize 2 servings per container and remove calories from fat2.90
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      4.69
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      71
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.66
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      5. Dual listing for calories only and remove calories from fat2.83
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      4.84
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      84
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.77
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      B. Two servings, dual-column formats
      6. Dual column—g and %DV
      %DV=percent daily value.
      columns
      2.75
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      5.08
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      83
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.85
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      7. Dual column—g and %DV columns and remove calories from fat2.77
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      5.07
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      85
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.83
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      8. Dual column—%DV only and remove calories from fat2.70
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      4.85
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      85
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.81
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      C. One serving, single-column formats
      9. Remove calories from fat2.29
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      5.34
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      83
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.70
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      10. Enlarge calories and remove calories from fat2.28
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      5.37
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      82
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      3.79
      For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      a Healthfulness rating ranges from 1 (not at all healthy) to 5 (very healthy).
      b Total correct index is based on eight questions. For each question the respondent could get it correct or incorrect. The index is the sum of the correct answers and can range from 0 to 8.
      c Percent correct for total calories is based on being able to correctly identify the number of calories per container.
      d Label perception index is based on five questions each with a 5-point scale. The index is divided by the number of questions answered such that it ranges from 1 (low perception) to 5 (high perception).
      e %DV=percent daily value.
      wxyz For each column, means with the same superscript letter (w, x, y, z) are not significantly different from each other, and means with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      In contrast to the findings for healthfulness ratings, more than one format modification appeared to produce differences in the two measures of accuracy. The one serving, single-column formats and some of the two servings, dual-column formats performed the best. On the broad index of eight nutrient content questions, the one serving, single-column formats and the two serving, dual-column formats that had dual columns for both grams and for %DV (formats 6 and 7 in Figure 1) had the highest accuracy. The two servings, dual-column format that had dual columns only for %DV (format 8) and the format with the dual listing for calories only (format 5) had the next highest accuracy. Formats 1 through 4, which were all two servings, single column formats, had the lowest scores. Among these formats, the control label format (format 1) had the lowest accuracy.
      The pattern is similar when looking at the narrower measure of the percentage that correctly identified the number of calories for the total package. The one serving, single-column formats, the entire set of dual-column formats, and the dual listing for calories only (format 5) did equally well, and all performed better than the remaining two servings, single-column formats (formats 1-4). Among the two servings, single-column formats, the format that emphasized that there were 2 servings per container (format 4) scored significantly higher than the control label format (P<0.05).
      On label perception ratings, dual-column formats, including the format with a dual listing for calories only (format 5), were rated more positively than the other two servings, single-column formats (P<0.05).

       Choice Task

      In addition to evaluating a single product, participants were asked to compare NF labels for a more and less healthful version of the same product (Choice Task). They could have seen either the same or different label formats for both product versions. When participants compared products labeled with the same format, label format did not have a significant effect on participants' ability to determine the more healthful of the two products (F=1.76, P>0.05), or which one had fewer calories per container (F=1.19, P>0.05). Across all experimental conditions, the vast majority of respondents correctly identified the more healthful product (88%) and the one with fewer calories per container (90%). On the other hand, label format did affect participants' accuracy in selecting the product with fewer calories per serving (F=3.64, P<0.01). Post hoc comparisons showed that participants who saw two servings, dual-column labels were slightly less accurate on average when reporting calories per serving (about 93% correct) than participants who saw more traditional single-column formats (about 96% correct).
      When participants compared products labeled with different formats (ie, mixed-format comparisons), other differences were observed (results shown in Table 4). Because there were no significant differences between comparisons involving formats 2 and 3 or between comparisons involving formats 9 and 10, their scores were combined for this analysis. For each of the three mixed-format comparisons (two servings, single-column vs one serving, single-column; two servings, dual-column vs one serving, single-column; and two servings, dual-column vs two servings, single-column), there was an “easy” and “hard” version of that comparison (see Figure 2). The task was considered “easy” when a participant could rely on the calorie declaration without considering the number of servings per container to derive the correct answer in all cases (ie, the number of calories per serving for the more healthful product was less than both the number of calories per serving and per container for the less healthful version). Conversely, the task was considered “hard” when looking at calories without considering the number of serving per container could lead to the wrong answer.
      Table 4The three outcome measures for the Choice Task involving the comparison of different label formats in a study to examine whether modifications to the Nutrition Facts label can help consumers make more healthful choices
      Label formatPercent correct healthier (n=4,627)Percent correct fewer calories per container (n=4,627)Percent correct fewer calories per serving (n=4,627)
      Easy comparisons
      The task was considered “easy” when looking at the calories line without considering the number of serving per container would lead to the correct answer; that is, the number of calories per serving for the healthier product was less than both the number of calories per serving and per container on the unhealthy version.
      Two servings, single-column vs one serving, single-column86
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      86
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      89
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      Two servings, dual column vs one serving, single-column81
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      80
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      87
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      Two servings, dual-column vs two servings, single-column81
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      84
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      89
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      Hard comparisons
      The task was considered “hard” when looking at the calories line without considering the number of serving per container could lead to the wrong answer.
      Two servings, single-column vs one serving, single-column44
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      45
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      74
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      Two servings, dual-column vs one serving, single-column51
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      64
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      75
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      Two servings, dual-column vs two servings, single-column75
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      68
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      81
      For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      a The task was considered “easy” when looking at the calories line without considering the number of serving per container would lead to the correct answer; that is, the number of calories per serving for the healthier product was less than both the number of calories per serving and per container on the unhealthy version.
      b The task was considered “hard” when looking at the calories line without considering the number of serving per container could lead to the wrong answer.
      wxyz For each column, percentages with the same superscript letter (w, x, y, z) are not significantly different from each other, and percentages with different letters are significantly different. A Tukey-Kramer test set at the 0.05 level was used. The following covariates were included in the models: product name and healthfulness profile, age, sex, race, education, and label use before purchase.
      Figure thumbnail gr2
      Figure 2Examples of “easy” and “hard” comparisons for the Product Choice task in a study to examine whether modifications to the Nutrition Facts label can help consumers make more healthful choices.
      As shown in Table 4, for all three dependent measures, participants did significantly better on the “easy” comparisons than on the “hard” comparisons (P<0.05). There were very few differences between the label format comparisons for the “easy” comparisons. However, for the “hard” comparisons, the presence of a dual-column format in a mixed label format comparison increased the percentage of participants who could correctly identify the more healthful product or the one with fewer calories per container (P<0.05). For example, in the “hard” experimental conditions where both label formats were a single-column format and the more-healthful product was labeled as containing 1 serving per container, whereas the less-healthful product was labeled as containing 2 servings per container, only 44% of participants could correctly identify the more healthful product. This percentage increased to 51% (P<0.05) when a dual-column label was compared with the one serving, single-column format and to 75% (P<0.05) when a dual-column label was compared to a two servings, single-column label. Similarly, only 45% of participants could correctly identify the product with fewer calories per container when the two servings, single-column formats were compared with the one serving, single-column formats. This increased to 64% and 68%, respectively (P<0.05 for both), when the dual-column format was compared with the one serving, single-column formats and the two servings, single-column formats.
      Among the “hard” comparisons, participants were better at choosing the product with fewer calories per serving than at identifying the product with fewer calories per container or the more healthful product. The former task may have been relatively easier than the other two tasks because all labels prominently displayed the number of calories per serving and multiplying the number of servings by 2 was not necessary to obtain the correct answer. When participants had to identify the product with fewer calories per serving, only the “hard” comparison involving the dual-column format vs the two servings, single-column format resulted in higher scores (81% correct compared with 74% and 75% correct). In other words, there was no difference in the percent that could correctly identify the product with fewer calories per serving between the “hard” comparisons involving the one serving, single-column format. This may be because for the “hard” comparisons with the one serving, single-column format, the label with the fewer calories per serving was always the “less-healthy” product and the product with more calories per the entire package. This contradiction between which product had fewer calories per serving and which had fewer calories per container may have confused some participants.

      Discussion

      Using an online experimental design, this research evaluated participants' ability to use and their preference for nine modified nutrition labels and a current label control. Because the focus was on testing changes to serving size declarations and the effects of emphasizing calories on the label, all products tested could contain 1 or 2 servings, but were generally considered types that can reasonably be consumed at a single eating occasion. The main findings are that for single product evaluations, single serving per container labeling and dual-column formats generally performed better and scored higher on the label perception index than two servings, single-column formats, including the control label. In tasks where nutrition information for two products is displayed in the same label format, participants were very accurate in determining the healthier product and the one with fewer calories per container, regardless of the label format. When nutrition information was displayed in different formats, no performance differences were found for the “easy” comparisons where participants could identify the lower-calorie product without considering the number of servings per container. However, the presence of a dual-column format improved scores for “hard,” mixed-format comparisons. Another main finding is that enlarging the font size for calories and removing “calories from fat” did not independently affect label usability as measured in this study.
      These findings about the beneficial effects of one serving, single-column and two servings, dual-column labeling complement those of other studies.
      • Mohr G.S.
      • Lichtenstein D.R.
      • Janiszewski C.
      The effect of marketer-suggested serving size on consumer responses: The unintended consequences of consumer attention to calorie information.
      • Lando A.M.
      • Labiner-Wolfe J.
      Helping consumers make more healthful food choices: Consumer views on modifying food labels and providing point-of-purchase nutrition information at quick-service restaurants.
      • Antonuk B.
      • Block L.G.
      The effect of single serving versus entire package nutritional information on consumption norms and actual consumption of a snack food.
      Lando and Labiner-Wolfe
      • Lando A.M.
      • Labiner-Wolfe J.
      Helping consumers make more healthful food choices: Consumer views on modifying food labels and providing point-of-purchase nutrition information at quick-service restaurants.
      found in focus group research that participants repeatedly expressed displeasure with labels that listed ≥2 servings per container for products that they thought they would eat in a single eating occasion. Regardless of their math skills, many stated that doing calculations on food labels was not an interest or priority.
      • Lando A.M.
      • Labiner-Wolfe J.
      Helping consumers make more healthful food choices: Consumer views on modifying food labels and providing point-of-purchase nutrition information at quick-service restaurants.
      Antonuk and Block
      • Antonuk B.
      • Block L.G.
      The effect of single serving versus entire package nutritional information on consumption norms and actual consumption of a snack food.
      found in an experimental study that nondieting participants ate less of a multiple-serving snack food when the food was labeled with a dual-column vs a single-column format. They speculate that dual columns may act as a contextual cue to highlight the number of servings per container and the amount consumed if the entire package is eaten and, hence, reduce consumption. Similarly, in our study, dual-column labeling may have made serving size more evident, especially when participants compared products that used different serving sizes to declare products' nutrition information.
      Similar to findings of Mohr and colleagues,
      • Mohr G.S.
      • Lichtenstein D.R.
      • Janiszewski C.
      The effect of marketer-suggested serving size on consumer responses: The unintended consequences of consumer attention to calorie information.
      our study found that labeling a food as having 1 serving per container (instead of 2 servings) caused participants to rate the products as being less healthful, whereas dual-column labeling did not have this effect. Thus, if one of the goals of improving nutrition labeling is to emphasize the relative healthfulness of various foods and perhaps decrease the rate at which individuals exceed their daily calorie needs, then labeling foods customarily consumed in a single eating occasion as a single serving may be more effective than dual-column labeling.
      There are a number of plausible reasons why enlarging the font in which calories is displayed or removing “calories from fat” from the label did not improve participants' comprehension of the label over the current label. These changes may not have been noticed.
      • Lando A.M.
      • Labiner-Wolfe J.
      Helping consumers make more healthful food choices: Consumer views on modifying food labels and providing point-of-purchase nutrition information at quick-service restaurants.
      Or, even if they were noticed, they may not have helped cue respondents to notice the number of servings per container, which was the key for getting many of the questions correct. Participants who saw format 4, which emphasized serving size by changing the words and using boldface type (format 4), did somewhat better than those in the control label format group in assessing the total calories per container but not on other tasks. Perhaps because the placement of the “2 Servings Per Container” was directly under “Calories in 1 cup serving,” this format was helpful in cueing participants that calories would need to be doubled. Nevertheless, this label format still required participants to do the math to determine the calorie and nutrient content of a container-sized portion.
      This research has a number of strengths and limitations. In addition to the randomized, controlled, experimental design, the study included a large sample size that permitted many label format modifications to be compared simultaneously and allowed for the detection of small differences between the label formats. Also, by using a national consumer panel, a diverse range of individuals was included. The study was limited by the nature of showing people NF labels on a computer screen as opposed to showing them actual food labels. The labels shown on the computer screen were probably larger and easier to read than NF labels on actual packaged foods. Therefore, changes such as enlarging calories may not have had the same effect as on smaller packages where poor visibility could adversely affect comprehension. Also, the NF labels were the only information provided to participants while they answered questions about these products. In more realistic shopping or eating occasions, consumers may rely on other information in addition to, or in lieu of, the NF label, so it is uncertain how these label modifications would affect actual purchase and eating behavior. Moreover, although we found dual-column labeling to be effective, this type of labeling could perform worse on smaller-sized packages due to increased clutter. Additional research would be needed to determine how the various label modifications might perform in a more realistic setting, such as when consumers are comparing multiple products in a grocery store. Finally, although reflective of US census data on most demographics, the study sample had slightly higher levels of education than the general public. Therefore, comprehension of these alternative label formats among lower-educated consumers, especially those with limited literacy and numeracy skills, may need further examination in future research.

      Conclusions

      We found that on products that contain 2 servings but are customarily consumed at a single eating occasion, using a single serving per container or a dual-column labeling approach may help consumers make healthier food choices.

      References

      1. Food labeling: Mandatory status of nutrition labeling and nutrient content revision, format for nutrition label. 58 Federal Register 2079.
        1993 (codified at 21 CFR §101.9)
        • Choiniére C.J.
        • Lando A.
        2008 FDA Health and Diet Survey topline frequencies.
        (Updated November 28, 2011. Accessed April 27, 2012)
        • Ollberding N.J.
        • Wolf R.L.
        • Contento I.
        Food label use and its relation to dietary intake among US adults.
        J Am Diet Assoc. 2010; 110: 1233-1237
        • Temple J.L.
        • Johnson K.
        • Recupero K.
        • Suders H.
        Nutrition labels decrease energy intake in adults consuming lunch in the laboratory.
        J Am Diet Assoc. 2010; 110: 1094-1097
        • Flegal K.M.
        • Carroll M.D.
        • Kit B.K.
        • Ogden C.L.
        Prevalence of obesity and trends in the distribution of body mass index among US adults, 1999-2010.
        JAMA. 2012; 307: 491-497
        • Ogden C.L.
        • Carroll M.D.
        • Kit B.K.
        • Flegal K.M.
        Prevalence of obesity and trends in body mass index among US children and adolescents, 1999-2010.
        JAMA. 2012; 307: 483-490
      2. Calories count: Report of the Working Group on Obesity.
        (Published March 12, 2004. Updated November 10, 2011. Accessed April 27, 2012)
      3. Food labeling: Prominence of calories. 70 Federal Register 17008.
        2005
      4. Food labeling: Serving sizes of products that can reasonably be consumed at one eating occasion; updating the reference amounts customarily consumed; approaches for recommending smaller portion sizes. 70 Federal Register 17010.
        2005
      5. Food labeling: Mandatory status of nutrition labeling and nutrient content revision, format for nutrition label. 58 Federal Register 2079.
        1993 (codified at 21 CFR §101.12)
        • Wansink B.
        Environmental factors that increase the food intake and consumption volume of unknowing consumers.
        Annu Rev Nutr. 2004; 24: 455-479
        • Mohr G.S.
        • Lichtenstein D.R.
        • Janiszewski C.
        The effect of marketer-suggested serving size on consumer responses: The unintended consequences of consumer attention to calorie information.
        J Mark. 2012; 76: 59-75
        • Pelletier A.L.
        • Chang W.W.
        • Delzell J.E.
        • McCall J.W.
        Patients' understanding and use of snack food package nutrition labels.
        J Am Board Fam Pract. 2004; 17: 319-323
        • Lando A.M.
        • Labiner-Wolfe J.
        Helping consumers make more healthful food choices: Consumer views on modifying food labels and providing point-of-purchase nutrition information at quick-service restaurants.
        J Nutr Educ Behav. 2007; 39: 157-163
        • Antonuk B.
        • Block L.G.
        The effect of single serving versus entire package nutritional information on consumption norms and actual consumption of a snack food.
        J Nutr Educ Behav. 2006; 38: 365-370
        • Labiner-Wolfe J.
        • Lin C.T.J.
        • Verrill L.
        Effect of low-carbohydrate claims on consumer perceptions about food products' healthfulness and helpfulness for weight management.
        J Nutr Educ and Behav. 2010; 42: 315-320
        • Lin C.T.J.
        How do consumers interpret health messages on food labels?.
        Nutrition Today. 2008; 43: 267-272
        • Levy A.S.
        • Fein S.B.
        Consumers' ability to perform tasks using nutrition labels.
        J Nutr Educ. 1998; 30: 210-217
        • Levy A.S.
        • Fein S.B.
        • Schucker R.E.
        More effective nutrition label formats are not nececessarily preferred.
        J Am Diet Assoc. 1992; 92: 1230-1234
        • Levy A.S.
        • Fein S.B.
        • Schucker R.E.
        Performance characteristics of seven nutrition label formats.
        J Public Policy Mark. 1996; 15: 1-15
        • Lin C.T.J.
        • Lee J.Y.
        • Yen S.T.
        Do dietary intakes affect search for nutrient information on food labels?.
        Soc Sci Med. 2004; 59: 1955-1967

      Biography

      A. M. Lando is a consumer science specialist, Center for Food Safety and Applied Nutrition, Food and Drug Administration, College Park, MD.
      S. C. Lo is a consumer science specialist, Center for Food Safety and Applied Nutrition, Food and Drug Administration, College Park, MD.