Principal Investigator

Judith Labiner-Wolfe, Ph.D
Consumer Studies Staff, Division of Social Sciences
Office of Regulations, Policy and Social Sciences
Center for Food Safety and Applied Nutrition
Food and Drug Administration
Email: [email protected]

---

Table of Contents

---

1. Summary

The Experimental Study of Carbohydrate Claims on Food Packages was designed to evaluate carbohydrate claims and related disclosures in terms of their effects on consumer understanding.

The experiment was conducted in March 2006 with over 9,400 respondents from a consumer Internet panel. Each respondent saw one product/label condition. Some respondents saw only a front panel, and others saw both a front panel and a Nutrition Facts label.

Analysis of the data shows that:

• Respondents who saw only the front panel with a claim that implied "low in carbohydrate," such as "Low Carb," "CarbConscious," or "1g Net Carb," perceived the product to be lower in carbohydrate and more helpful for weight management than those who saw the same product without a "low in carbohydrate" claim.
• Respondents who saw only the front panel often perceived different claims that implied "low in carbohydrate" as having a shared meaning. Respondents often rated products similarly whether they had "Low Carb," "CarbConscious," or "Net Carb" on the front.
• Respondents who saw only the front panel perceived products with a "Good Source of Carb" claim differently depending on the product on which it appeared.
• Adding a disclosure statement, such as "see nutrition information for fat content," did not consistently affect product perceptions when respondents saw only the front panel.
• Respondents who saw both the front panel and the Nutrition Facts label rated products with the same Nutrition Facts similarly and rated products with a more healthful Nutrition Facts more favorably than those with a less healthful Nutrition Facts.

2. Objective

The Experimental Study of Carbohydrate Claims on Food Packages was designed to evaluate carbohydrate claims and disclosure statements in terms of their effects on consumer understanding and consumers' ability to make appropriate product judgments for healthy dietary practices.

3. Study Design

The carbohydrate claims tested in the study were "Low Carb," "Good Source of Carb," "CarbConscious," "1g Net Carb," and "14g Net Carb." The study also included control labels, which did not bear any carbohydrate claim on the front panel. Where relevant, the study tested carbohydrate claims with and without the following disclosures on the front panel: (1) "see nutrition information for fat content;" (2) "see nutrition information for sugar content;" and (3) "not a low-calorie food."

Participants were exposed either to a front panel with or without a carbohydrate claim, or to both a front panel and corresponding Nutrition Facts label for one of three products. The respondents who saw both the front panel and Nutrition Facts label saw them side-by-side. The products in the study were a loaf of bread, a 12-ounce juice drink, and an individual frozen dinner. Figure 1 provides a picture of one front panel for the bread, juice drink and frozen dinner.

The Nutrition Facts labels were varied on the bread and dinner to create more and less healthful product profiles. Less healthful versions of the Nutrition Facts label listed more calories, fat and saturated fat, and less fiber than the more healthful versions. There was not a less healthful version of the juice drink because the product did not contain nutrients that could be manipulated, other than carbohydrates. The study included higher and lower carbohydrate versions for all three products.

The study questions asked respondents to rate the product shown to them on a series of seven-point Likert scales based on: purchase intent; perception of healthfulness; how high or low it was in several nutrients; and how likely it was to help someone to manage their weight, strengthen their bones, have more energy for sports, and eat more or fewer carbohydrates.

All participants were asked the same general series of questions. However, respondents who saw the juice drink were not asked to evaluate nutrients not in the product (e.g., fat, fiber, and protein). Only those who saw a claim that implied "low in carbohydrate" (which includes "Low Carb" "CarbConscious" and "Net Carb") were asked about the likelihood that the product would help someone to eat fewer carbohydrates. Only those who saw the "Good Source of Carb" claim were asked about the likelihood that the product would help someone to eat more carbohydrates.

The universe for this study was U.S. adult members of a consumer opinion panel belonging to the research firm Synovate. The panel consisted of over 600,000 U.S. households that were recruited by a variety of means and agreed to participate in Internet research studies. As an experimental study, this research was intended to help reveal causal relationships between label claims and consumer responses. The panel was not a nationally or locally representative sample and cannot be used to generate precise estimates of population parameters. The strength of the experimental study lies in its internal validity, on which meaningful estimates of differences across conditions can be produced.

In November 2005, the 600,000 panel members were emailed screening questions to collect information on diet status. The sample of respondents to the experiment was drawn from the pool of over 173,000 panel members who responded to the screening questions.

Data were collected in March 2006. The sample size of over 9,400 allowed for a minimum of 180 respondents per product and label condition. Each respondent saw a single product/label condition. A label condition was a front panel alone, with a Nutrition Facts label, or with a disclosure and/or a Nutrition Facts label. This study did not test every label condition with all three products.

Table 1 lists the label conditions and the products on which each appeared.

4. Analysis

The results from the experiment were analyzed using SAS software. The mean ratings on each outcome variable by product and condition were compared. Each outcome variable was also used as the dependent measure in Generalized Linear Models (GLM). The series of GLM evaluated the relationship between each outcome variable, the combination of the product and claim, and other variables that could affect respondent ratings. These other independent variables in the models were whether the respondent was male or female, diabetic, on a low-carb diet, used the Nutrition Facts for first-time purchases, consumed products like the one in the experiment, the respondent's age, the respondent's income, and interactions between the product and the condition, age and diet status, and age and income.

5. Main Findings

The analysis showed that without the Nutrition Facts label available, respondents who saw claims that imply "low in carbohydrate" on the front of food packages had some positive associations with these products that may or may not be accurate. When the Nutrition Facts label was available side-by side with the front panel, respondents made appropriate product judgments when evaluating products with claims that imply either fewer or more carbohydrates.

Findings from GLM analyses:

• Respondents who saw only the front panel rated food products with a claim that implied "low in carbohydrate" as lower in carbohydrate than those who saw the same product without a claim.
• Respondents who saw only the front panel rated food products with a claim that implied "low in carbohydrate" as more helpful for weight management than those who saw the same product without a claim.
• Respondents who saw only the front panel often provided similar ratings for food products with several different claims: "Low Carb" "CarbConscious," and "Net Carb."
• When the front panel only was available to respondents, the effect of the "Good Source of Carb" claim was product specific.
• Adding a disclosure statement did not consistently affect product perceptions when respondents saw only the front panel.
• Respondents who saw both the front panel and the Nutrition Facts label side by side rated products with the same nutrition profile similarly, regardless of a front panel carbohydrate claim. Products with a more healthful nutrition profile were rated as such, relative to those with a less healthful nutrition profile, regardless of the presence of a carbohydrate claim on the front panel.