Health and Hope Part
Chapter 14 Do Labels Make a Difference?
Rather than repeating the detailed factor by factor discussion, we can look at Table 14.2 to get a sense of the “ bigger picture. ” There are six results that tell us something about ethnicity and labels.
1. There are some differences by ethnicity. For example, Asians look for no preservatives, no artifi cial sweeteners, and organic. It ’ s clear that Asian respon- dents are more focused on “ health ” and “ good for you ” characteristics, and look for these in the food labels.
2. Afro - Americans focus more on number of ounces, anti - oxidants, organic, and lack of sweeteners than do Whites, but not dramatically so. The difference is only about 8 – 10% more Afro - Americans than Whites focus on these label factors.
3. Hispanics provide an interesting fi nding in terms of the effects of acculturation. Those Hispanics who were born in the U.S. show patterns similar to Whites. However, Hispanics who immigrated to the U.S. show patterns that are quite different. These fi rst - generation Americans, who are acculturating, show far greater interest in the label factors than do the Hispanics who were born in the U.S. However, the change is not in freshness or the number of calories. With those factors, there is not much effect of acculturation. The effect comes with the other label factors, the less important ones. The biggest one is antioxidants. This is relatively unim- portant to Hispanics born in the U.S. (36%) but much more important to Hispanics who are acculturating (55%).
People Vary When It Comes to What ’ s Important on a Label
We just saw that there are differences in the importance of different label factors. We were able to look at the different groups and see some variation among groups. A more insightful analysis can come from taking a granular view. In this view, we look at the distribution of ratings on the 9 - point scale for all 1,022 respondents. The question we should ask is very simple: Does each of these factors show most of the respondents clustering around the mean, or is there a distribution?
The answers won ’ t be particularly surprising. Look at Figure 14.1 . We see two types of patterns. The fi rst pattern shows most of the respondents clustering around a small region on the 9 - point scale. We see this for the label factor “ fresh. ” We see this pattern but a little less striking for the label factors “ ingredients ” and “ calories, ” where an increasing number of respondents rate the factor high.
We see a different pattern for the other label factors. The importance ratings distribute. For example, when it comes to organic, we see no clear pattern or, for that matter, when it comes to antioxidants.
We conclude from this graphic analysis that there are different types of respondents, with different mind - sets. The most important fi nding is that almost everyone thinks that the role of the label is to tell one about product freshness and perhaps about ingredients and calories. The remaining factors are idiosyncratic, depending upon the individual ’ s particular interest in health and wellness.
What Interest in Ingredients Drives a Person to Look at Label Factors?
We conclude this section on labels with an attempt to discover a linkage between what ’ s important to a respon- dent in terms of “ ingredients in a food, ” and what the person looks for in a label.
The same respondents who rated importance of label factors also rated the importance of different ingredients in foods that they buy. The respondents were not asked whether they looked for information about these ingredients, but rather simply to rate the impor- tance to them of the ingredient in store - bought foods.
As one might expect, there is variation among respondents as to what is important (Grunert, Brunson, Bredahl, and Bech, 2001 ; Saba, 2001 ). However, we are more interested here in linking what people think is important to what they look for on the label. In Table 14.3 , see the correlations between what ’ s impor- tant in terms of ingredients (columns) and what people look for on the label (rows). The numbers in the body of the table are Pearson correlations, the Pearson R statistic. This statistic tells us the strength of a linear relation between the importance rating of fi ber (specifi c feature) and the importance of the ingredient (general factor looked for).
1. Our results suggest that if a person is interested in any of the ingredients (all columns), then the person will look at the ingredient list, and most likely the number of calories, antioxidants, and lack of preservatives.
2. Being interested in calories as “ an ingredient ” translates primarily to being interested in label statements about ingredients and number of calories, respectively.
3. The remaining ingredient interests translate to spe- cifi c interests in label factors. For example, being interested in “ sugar ” translates to thinking that “ natural ” is important on a label. Or being interested
1 3 5 7 9 FRESH 1 3 5 7 9 PRESERVATIVE 1 3 5 7 9 ORGANIC 1 3 5 7 9 SWEETENERS 1 3 5 7 9 NATURAL 1 3 5 7 9 SERVINGS 1 3 5 7 9 CALORIES 1 3 5 7 9 OUNCES 1 3 5 7 9 ANTIOXIDANT 1 3 5 7 9 INGREDIENT
Figure 14.1 Distribution of the ratings for importance on the 9 - point scale. Each label factor was rated in importance by 1,022 respondents.
The Pearson R or linear correlation ranges between a low of − 1 (perfect inverse relation), to a middle value of 0 (no linear relation at all), to a high of +1 (perfect linear relation). With 1,022 “ cases ” or observations, it is highly unlikely to achieve a high Pearson R if there is no underlying relation ( Box, Hunter, and Hunter, 1978 ).
We have boldened in those correlations above 0.40. We can feel confi dent that those correlations denote a high degree of association between what ingredient is important to the respondent when shopping, and what the respondent looks at when inspecting the label.
Chapter 14 Do Labels Make a Difference? 143
food or beverage. With the increasing interest by government agencies on the best way to communicate healthfulness to consumers through food labels (Goldberg, 2000 ), we now move to a study that addresses that problem.
The origin of the study may be of some interest. In the summer of 2007 ILSI (International Life Sciences, Inc.) of Southeast Asia sponsored a three - day sympo- sium on nutritional challenges in Southeast Asia. The senior author of this book attended and presented a paper. At the end of the conference, he went out for a tour of Singapore with a number of the other participants. During a refreshment break, the group visited a store, looked at some of the products, and talked to some locals. Coming out of that was the realization that it was possible to use some of these scientifi c methods to explore responses to labels, worldwide. No one at the conference was able to articulate exactly what a Singaporean consumer needed to see on the label. And so was born the experiment — to identify what particular messaging on the label would attract consumers to feel that the food was healthy. The actual experiment we report here was to be the fi rst of a series of studies, conducted for a number of foods, both inside the U.S,. as well as in Southeast Asia and Western Europe. We report on the results of the fi rst study that emerged from the trip. So, in a way, this half of the chapter is our fi rst fruit, in which we look at the specifi c messaging on the label.
in calories translates to “ number of servings ” being important on a label.
Summing Up
This fi rst study suggests to us that respondents fi nd dif- ferent factors important when they read the labels. It shows differences in label factors by age more than by gender. It also reveals that ethnicity is important, with Asians far more interested than Whites, Afro - Americans, and U.S. – born Hispanics. Hispanics who immigrated to the U.S. differ from U.S. – born Hispanics, with the former, emigrating group, more sensitive to factors such as artifi cial sweeteners. U.S. – born Hispanics look more similar to Whites. Finally, respondents are very similar to each other when it comes to freshness but differ sub- stantially on the importance of other label factors, such as organic. It ’ s important to most respondents to under- stand product freshness from reading the label. However, people differ from each other in the importance they attribute to “ organic. ”
What Label Information “ Works ”
Now that we have explored what is important “ in general, ” let us look at how respondents react to specifi c messaging. We are still going to work at the “ general ” level, rather than at the level of a specifi c
Table 14.3 How interest in specifi c ingredients (columns) correlates with what is stated to be important when one reads a food label (row). The numbers in the body of the table are Pearson correlations (R - statistic).
Fiber Protein Sodium Vitamins Trans
Fat
Sugar Saturated
Fat
Carbohydrates Calories
Ingredient (in general) 0.53 0.50 0.49 0.51 0.53 0.55 0.53 0.48 0.45 Number of Calories 0.48 0.47 0.47 0.38 0.54 0.55 0.57 0.50 0.83 Antioxidants 0.56 0.54 0.51 0.60 0.51 0.46 0.47 0.46 0.36 Lack of Preservatives 0.50 0.48 0.53 0.48 0.50 0.48 0.47 0.44 0.35 Natural 0.41 0.45 0.44 0.50 0.39 0.42 0.38 0.37 0.26 Ounces 0.38 0.37 0.36 0.37 0.33 0.35 0.35 0.38 0.37 No Sweeteners 0.38 0.39 0.43 0.40 0.39 0.37 0.36 0.33 0.20 Number of Servings 0.37 0.39 0.32 0.31 0.35 0.34 0.33 0.33 0.46 Organic 0.38 0.38 0.37 0.36 0.35 0.35 0.32 0.32 0.27 Fresh 0.24 0.23 0.25 0.24 0.23 0.25 0.21 0.24 0.20
We begin with the nature of food claims, the topic that concerns the Food and Drug Administration in the U.S. The FDA has established language guidelines for char- acterizing the level of nutrients in food (see Table 14.4 ). The terms “ good source of ” or “ excellent source of ” can only be used to describe protein, vitamins, minerals, dietary fi ber, or potassium. Good sources of a given nutrient provide 10 – 19% of the daily value (DV) per reference amount, whereas excellent sources provide 20% or more. Conversely, such terms as “ free, ” “ low, ” or “ reduced/less ” apply to calories, total or saturated fat, sodium, sugars, or cholesterol. These, too, are based on reference amounts or on 50 g if the reference amount is small. Clearly, consumers cannot deal with many of these issues. Most consumers pay attention to words like “ light, ” “ fat free, ” etc. They do not pay attention to the legal defi nitions.
Now that we know the legal defi nitions, let ’ s see how consumers respond to different test labels. For
Table 14.4 Nutrient content claims as defi ned by the Food and Drug Administration (Source: Drewnowski et al. 2008 , unpublished manuscript)
Term Nutrient content per reference amount Good source of;
contains; provides
Contains 10 – 19% of the DV. Used to describe protein, vitamins, minerals, dietary fi ber, or potassium, but not carbohydrate.
Excellent source of; high; rich in
Contains 20% or more of the DV. Used to describe protein, vitamins, minerals, dietary fi ber, or potassium, but not carbohydrate. More; added;
extra; plus
Contains 10% or more of the DV. May only be used for protein, vitamins, minerals, dietary fi ber, or potassium,
High potency Describes individual vitamins or minerals that are present in food at 100% or more of the RDI per reference amount.
Free; zero; no; without
Calories < 5; total fat < 0.5 g; saturated fat
< 0.5 g; trans fat < 0.5 g; cholesterol < 2 mg;
sodium < 5 mg; sugars < 0.5 g (per reference amount and per labeled serving) Low; little; few;
low source of
Calories < 40 * ; total fat < 3 g * ; saturated fat
< 1 g †
; cholesterol < 20 mg * ; sodium
< 140 mg * ; sugars: not defi ned
Reduced/less; lower; fewer
Calories > 25% less; total fat > 25% less; saturated fat > 25% less; cholesterol > 25% less; sodium > 25% less; sugars > 25% less †
Light Percent reduction for both fat and calories
must be stated
* Per 50 g if reference amount is small
†
With < 15% calories from saturated fat
†
Relative to appropriate reference food. May not be termed “ low. ”
this experiment, we selected different phrases that would ordinarily appear on the package. This approach makes sense — we are investigating how nutrition labeling affects consumer responses, so we work in the realm of what is legal, rather than exploring close in versus far out ideas.
In order to explore a reasonably wide range of elements, we looked at six silos, each of six elements, which comprise a total of 36 elements. The elements appear in Table 14.5 . Note that we have divided the set of elements into three sets of silos. The fi rst set comprises 12 elements that are desirable. We wanted to look at two ways to express these nutrients, because it wasn ’ t clear whether or not the way of expressing a positive nutrient made a difference. No one really knew whether, for instance, there would be much diff- erence between two similar elements with different emphases, such as “ This product is a good source of protein ” versus “ This product is an excellent source of protein. ”
Our second pair of silos, C and D, covered the nutri- ents that the FDA advises limiting. Finally, the third pair of silos, E and F, covered emotional messages that aren ’ t technically part of the nutrition label itself, but could be put on the package as additional “ wellness - related ” information.
We recruited the respondents to participate from a local American panel, specializing in e - mail recruiting of respondents. The key to the entire study comes from correctly orienting the respondents. The respondents received an email invitation, which told them just a little about the study, not enough to tell them exactly what was being tested, but enough information to intrigue them. Of course, it is necessary to reward people for participa- tion. To make this study, and other studies like it, more cost effective, we ran a sweepstakes so that respondents who participated had a chance to win a cash prize. These incentives make the respondents feel that their time and efforts are appreciated.
Respondents who chose to continue (i.e., participate), clicked on the embedded link in the e - mail invitation. This led them to the orientation page that told them about the study. The orientation page set up the respon- dent ’ s expectations about the study, while providing relatively little detail about the stimuli, the rationale for the study, etc. The goal of the orientation page is to make sure the respondents understand what is expected of them. Here is what we used for this study. If you read it carefully you will see that there is a great deal of
Chapter 14 Do Labels Make a Difference? 145
Table 14.5 Elements tested in the nutrition label project and their impact values for the total panel
Additive constant 22 Silos A and B — Desirable nutrients A1 This product is a good source of protein. 17 A2 This product is an excellent source of protein. 16 A3 This product is a good source of vitamin C. 9 A4 This product is an excellent source of vitamin C. 11 A5 This product provides vitamin A. 6 A6 This product is high in vitamin A. 9 B1 This product contains fi ber. 9 B2 This product is high in fi ber. 14 B3 This product contains calcium. 8 B4 This product is rich in calcium. 11 B5 This product provides iron. 4 B6 This product is rich in iron. 5 Silos C and D — Nutrients to limit C1 This product is low in total fat. 7 C2 This product is fat free. 10 C3 This product contains little saturated fat. 3 C4 This product has no saturated fat. 12 C5 This product is a low source of cholesterol. 7 C6 This product has zero cholesterol. 11 D1 This product is low in total sugar. 8 D2 This product is sugar free. 9 D3 This product is low in added sugar. 4 D4 This product is free of added sugar. 10 D5 This product is a low source of sodium. 9 D6 This product is without sodium. 12 Silos E and F — Emotional messages and reassurance E1 Wholesome food that gives you more nutrition per bite 4 E2 Is a good way to balance your diet to keep it nutrient
rich
6
E3 Is a total nutrient package with more nutrients than calories
4
E4 Naturally packed with nutrients for better health 6 E5 Meets your daily nutrient needs without too many
calories
5
E6 Puts more nutrient power on your plate 7 F1 Is a great way to enjoy your healthy lifestyle 4 F2 Takes the stress out of healthful eating 4 F3 Lets you eat well to live well, starting today 3 F4 You and your family can eat right, for life. 5 F5 Be at your best — enjoy good taste and good health. 7 F6 You can trust the nutrition label to guide smart eating. 3
general information, but nothing that “ gives the game away. ”
Providing nutrition information on food and drink product labels is an important way of conveying the message about diets and health to the consumer. It is important that such information accurately refl ect the nutrient composition of the product in a simple manner that is easily understood by the consumer. However, nutrition labels do not always get the right message across. You are invited to review a selection of messages and rate each product on a “ healthfulness ” scale.
Participants were told that they would be presented with nutritional information for a hypothetical food product. Based on this information, they were asked to use a 9 - point category scale, to rate whether, in their opinion, the product was 1 = least healthy or 9 = most healthy. Each respondent evaluated 48 test concepts, comprising small combinations of the elements. Throughout this book, we use the same type of approach — experimental design, so we will simply point out the highlights of the “ method, ” rather than go into detail.
We ran the study with 320 U.S. consumers. The data speak for themselves. See Table 14.5 . Let ’ s interpret the results. The data you will see comes from an analysis of the ratings. The rating scale (1 = least healthy to 9 = most healthy) was converted to the binary scale as we did for the above - mentioned attitude research on labels. That is, ratings of 1 – 6 for a test concept were converted to 0, ratings of 7 – 9 were converted to 100. We then developed a model for each of the 320 respondents showing how every one of the 36 elements in Table 14.5 “ drives ” the less healthly/more healthy response.
With that information in mind, let ’ s now look at the data in a bit more depth. We don ’ t have to probe very deeply — the results are fairly straightforward and, occa- sionally, surprising. Keep in mind that we did not specify the particular food. We are only looking at the general response to these test “ nutrition labels. ”
1. The additive constant is 22. The additive constant is the percent of respondents who would rate the concept 7 – 9 if no elements were present. Clearly this constant is an estimated, theoretical value, but it is good as a baseline indicator (Green and Krieger, 1991 ). Without any additional information, 22% or about one in four respondents would say that the product is “ more healthful. ” That fi nding provides us with a sense of
the coherence of the segments, it is not easy to predict membership in the segments. The segmentation “ tells a story. ”
Let ’ s look at this story for labels. We segmented the respondents by the pattern of their impact values. Now to see the results, which appear in Table 14.6 . Rather than showing all of the elements by the segments, which could generate a very large table, we have summarized the results by choosing the highest and lowest perform- ing elements for each segment.
One of the goals of the segmentation was to search for a group of respondents who could be classifi ed as “ health sensitive. ” This segmentation may be especially important for specifi c categories for functional/enriched foods that will benefi t from favorable profi les and health claims.
The segmentation generated three groups that could be interpreted. The respondents in a segment show similar profi les of impacts. That is, the people may differ widely by age, gender, income, even stated interest in health. Yet these people show similar patterns of responses to what they consider to be “ healthy. ” We see