SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. BACKGROUND
  4. STUDY 1
  5. STUDY 2
  6. STUDY 3
  7. STUDY 4
  8. GENERAL DISCUSSION
  9. REFERENCES

The authors address some implications of recent legislation that will require calorie labeling for national chain restaurants. Drawing from the health halo and information disclosure literatures, the potential positive consumer outcomes associated with the disclosure of calorie (only) and additional nutrient information are examined. Results across four studies show that while most consumers underestimate calorie levels of restaurant menu items, the degree of underestimation is substantially greater for sodium. The provision of sodium content levels for menu items, in addition to calorie information, influences purchase intentions and choices of consumers with high health risk levels, but has little effect on other consumers. Reducing Americans' average daily intake of sodium (currently 3,400 mg) to the recommended level of 2,300 mg could eliminate 11 million cases of high blood pressure and prevent 92,000 annual deaths (Palar and Sturm 2009). Therefore, the results have potentially significant implications for consumer health and welfare and the restaurant industry.

Nearly one-third of the calories in the standard American diet come from foods that add calories with little nutritional benefit—soft drinks, sweets and desserts, alcoholic beverages, and salty snacks (Block 2004). Therefore, it is not surprising that the prevalence of childhood and adult obesity, diabetes, and other chronic diseases has increased dramatically in the last 30 years. The unhealthiness of the American diet is the result of a complex mix of social, marketing, environmental, product, and individual (e.g., affective preferences) factors (Seiders and Petty 2004; Wansink 2010). Consequently, food consumption–related issues have been investigated from the perspective of many different disciplines. Within the field of marketing, much of this research has focused on the highly visible and well-publicized “obesity crisis.” Given the strong relationship between the overconsumption of food and obesity, it is not surprising that calories have been the focus of many of these studies. However, America's recent preoccupation with calories has a potential downside—negative nutrients that are associated with the development of several serious health conditions may not always be receiving the level of attention they deserve.

In this research, we propose and test halo and anchoring effects across nutrition attributes in four studies. Extensive prior research has shown that the perception of one attribute of an individual can strongly influence how other attributes of that individual are perceived (Roe, Levy, and Derby 1999). Recent research has found that such cognitive biases also extend to food. That is, consumers often seem to behave as though healthy foods have “halos” that extend to other characteristics of the food (Roe, Levy, and Derby 1999; Chandon and Wansink 2007). For example, some consumers may infer that an avocado is a “good diet food” because it is a fresh fruit perceived as healthy; however, avocados are not (relatively) low in calories.

Health halo effects are typically examined for single attributes or positively correlated attributes. Consequently, inferences influenced by halo effects appear relatively consistent across attributes. For example, Chandon and Wansink (2007) convincingly show that consumers underestimate calorie levels of healthy foods to a greater extent than food presumed to be less healthful. This underestimation also influences consumers' choice of complementary food and beverages. The current research extends this work by addressing how interattribute, category-based inferences can also vary in their degree of inaccuracy (Hastak and Mazis 2011). For example, we consider how the health positioning of a category (grilled chicken vs. fried chicken) may influence consumers' perceptions of nutrient attributes (e.g., sodium). We then address how effective various information disclosures are at minimizing these effects.

Within this context, we assess differences in health halo effects for calories versus sodium for a broad variety of restaurant menu items. Sodium is of particular concern for several reasons. The average daily intake of sodium for US consumers is 3,400 mg, almost 50% above the maximum recommended level and more than twice the level for the 50% of the US population at greatest risk. The US government has responded by issuing a call to action to reduce sodium intake (Centers for Disease Control and Prevention (CDC) 2009, 2013). Excessive sodium intake increases the risk of hypertension, and hypertension is among the known risk factors correlated with stroke and coronary heart disease. In fact, the link between hypertension and heart disease is stronger than the links between heart disease and smoking, high cholesterol, or obesity (Liebman 2010; Howlett et al. 2012). Reducing Americans' average daily intake of sodium to the recommended level of 2,300 mg could potentially eliminate 11 million cases of high blood pressure, prevent 92,000 annual deaths, reduce strokes by 66,000 per year, and save over $18 billion annually in health care costs (Palar and Sturm 2009; Liebman 2010).

The primary sources of salt in our diet (77%) are restaurant and processed food items (Food and Drug Administration (FDA) 2012; Howlett et al. 2012), and calories consumed away-from-home now account for nearly 50% of food expenditures. One way in which health-conscious consumers may reduce sodium and calorie intake is to avoid or decrease their consumption of these high calorie and high salt content foods. While calorie and sodium information is required on the Nutrition Facts panel of most packaged foods (and therefore easily accessible for motivated consumers), for restaurant foods this is not the case. Traditionally, nutrition information has not been available at the point of purchase for restaurant chains, but diners will soon notice some major changes in their shopping environment. The US Patient Protection and Affordable Care Act (Public Law 111-148), signed into law on March 23, 2010, requires the provision of calories on menus, menu boards, and drive through menu boards for all restaurant chains with twenty or more outlets nationally.1 The provision of sodium information, however, is not mandated. Thus, because sodium information will only be available upon request, it seems unlikely to be frequently accessed or to have any major influence on product evaluation processes at the point of purchase (Wootan and Osborn 2006; Howlett et al. 2012). This may be problematic since many restaurant items often contain extremely high amounts of sodium; some items contain two to three times more than the daily recommended level (Center for Science in the Public Interest (CSPI) 2009).

A policy storm is amassing as the public health community, consumer advocates, restaurants and food marketers, medical professionals, and policy makers at local, state, and federal levels formulate strategies to address what many view as a developing health crisis (Institute of Medicine (IOM) 2010; CDC 2013). In a complete “information disclosure world,” national legislation might have mandated the disclosure of sodium levels, as well as calorie content, on menus and menu boards. Since this is not the case, this complex set of health, regulatory, and competitive circumstances provides an opportunity to address several key issues regarding the impact of health halos. Answers to these questions will help better inform the decision making processes of marketing managers throughout the food industry, in addition to concerns of the public health and policy communities. Across four studies, we consider the following questions:

  1. Drawing from literatures that address health halo effects and anchoring and adjustment biases, how accurately do consumers estimate sodium levels of different restaurant items across more and less healthful food categories (salads, hamburgers) and how does this accuracy compare to the accuracy of calorie content estimates?
  2. Relative to estimated sodium intake, what is the actual sodium intake across food categories? How do these differences compare to the differences between estimated and actual calorie consumption?
  3. How does (a) the disclosure of calorie information affect inferred sodium levels and influence purchase intentions for more and less healthful products? and (b) the disclosure of sodium information in conjunction with calorie information influence product evaluations and purchase intentions?

In summary, this research extends prior research by addressing how (misleading) health halos can differentially affect perceptions of calorie and sodium attributes and their role in consumers' product evaluations in the absence of more complete information disclosure. Within the context of both quick service and table service restaurant environments, we also develop a better understanding of how various types of nutrition information disclosures will influence misleading health halos and consumers' product evaluations. Results suggest that the “devil is in the details” concerning the effectiveness of restaurant chain disclosures. We hope that this research will contribute to more effective consumer sodium reduction initiatives and help consumer welfare advocates better understand the implications of the changing regulatory and competitive environments.

BACKGROUND

  1. Top of page
  2. Abstract
  3. BACKGROUND
  4. STUDY 1
  5. STUDY 2
  6. STUDY 3
  7. STUDY 4
  8. GENERAL DISCUSSION
  9. REFERENCES

During the past several years, prior research has highlighted the fact that consumers underestimate calorie levels of meals served at both fast food and table service restaurants (e.g., Wansink and Chandon 2006a, Chandon and Wansink 2007; Tangari et al. 2010; Howlett et al. 2009). Since the overconsumption of calories is directly associated with the development of obesity, it is not surprising that academic researchers, consumer advocate organizations, and public policy makers have expended considerable effort addressing consumers' awareness and understanding of calories. Research demonstrating that consumers often substantially underestimate the calorie levels of away-from-home foods has been a primary justification for the assertion that calorie labeling on menus and menu boards is necessary to help consumers make informed choices.

Far less attention has focused on consumer awareness of the high sodium levels of many away-from-home foods. The relative paucity of studies addressing consumers' sodium-related knowledge is intriguing given both the negative health implications of excessive sodium intake and the high sodium levels of many foods prepared outside the home. Oftentimes, compared to calories, the sodium found in restaurant meals constitutes a higher percentage of the total RDA (2,300 mg). For example, Subway six-inch sandwiches, on average, have contained approximately 12% of the recommended calorie allowance and 52% of the recommended sodium allowance. This large percentage difference is relatively common across restaurant meals and has raised considerable concern among health care professionals and public health advocates (CSPI 2009).

Health Halos and Anchoring Effects for Calories and Sodium

While prior literature has addressed how both inter- and intra-attribute inferences can mislead consumers (Hastak and Mazis 2011), little research has addressed how perceptions of a superordinate attribute (i.e., a “healthy” product) may differentially bias inferences of specific subordinate food attributes (calories and nutrients). Our first hypothesis is drawn from two interrelated streams of research. First, there is a substantial literature regarding consumers' inferences about missing attributes and how other internal and external information may bias estimates. This research has shown that in the absence of specific attribute information, most consumers will construct their estimates from both internal and external cues (e.g., Kardes, Posavac, and Cronley 2004; Chandon and Wansink 2007; Ross and Creyer 1992). This suggests that factors such as product or firm positioning are likely to have a significant influence on the value of the inferred information. In fact, recent research has repeatedly shown that identifying a product as “low fat” or “organic” can lead consumers to underestimate its calories and this affects consumption levels and choices (Wansink and Chandon 2006b). Recent health halo literature has typically focused on effects on calorie perceptions, while ignoring other subordinate nutrients, such as sodium.

Given no extrinsic information regarding sodium or calorie content levels, consumers are likely to make inferences based on the perceived general healthfulness of the product category of the restaurant item. A “health halo” will be associated with meal categories perceived as relatively healthy (Roe, Levy, and Derby 1999; Chandon and Wansink 2007). This halo effect will affect both the underestimation of sodium and calorie levels across food items that vary in perceived healthfulness. Another type of health halo was explored by Chernev (2010). He showed that the context within which a specific food was evaluated influenced calorie estimates. For example, a hamburger accompanied by a side of three celery sticks was judged lower in calories than the same hamburger presented in isolation.

Prior literature on anchoring effects also provides a helpful theoretical foundation for this research. This body of research demonstrates that easily accessible anchors (Tversky and Kahneman 1974) influence many numeric estimates made by consumers under conditions of uncertainty. This effect has been consistently demonstrated across a variety of decision domains, such as value judgments (Johnson and Schkade 1989) and probability judgments (Hogarth and Einhorn 1992). Furthermore, anchoring effects occur regardless of whether the anchor is externally provided or internally generated.

In general, we expect that in control conditions when no explicit external information is provided, consumers will be less knowledgeable about sodium level than calorie content. While knowledge of calorie levels may be rather low, knowledge regarding sodium is likely to be much lower. The media focus on “calories” is significant. Popular magazines, talk shows, advertising, product packaging all devote significant attention to weight loss, calorie-counting, and similar information and programs. All this publicity is likely to have raised consumer awareness and understanding of calories, at least relative to sodium. Thus, forming an accurate estimate of a food's sodium level is likely to be a difficult task. Given this, consumers are likely to rely on superordinate attribute information such as the overall perceived healthfulness of the meal category (e.g., hamburger and fries vs. salads vs. grilled chicken sandwiches) as relevant information when estimating subordinate attributes such as sodium (or calories).

However, objective sodium levels for restaurant items are probably less related to category healthfulness perceptions than are calories. Sodium/salt is often used as a flavor enhancer when calorie or fat levels are naturally low or reduced, and adding salt has no effect on calories or other negative nutrients (fat or saturated fat). Of course, calories and negative nutrients such as sodium are naturally linked through the size of the item; as the size of the item increases, both calories and sodium levels generally increase. On balance, however, adjustments from a given anchor point due to the health positioning of a category are more likely to be biased and not consistent with objective sodium levels. While calories will often be underestimated (Chandon and Wansink 2007; Burton et al. 2006), we propose that the underestimation of sodium levels will be greater and much more substantial for healthful categories. We predict:

H1a: Sodium levels of restaurant menu items will be underestimated to a greater extent than calorie levels.

H1b: A greater number of consumers will underestimate sodium levels as compared to calorie levels.

H2: Compared to calories, the discrepancy between estimated and actual sodium levels will be larger (smaller) for menu categories perceived as more (less) healthful.

Effects of Disclosures on Consumer Estimations and Product Evaluations

The general purpose of an information disclosure is to provide information that may benefit consumers during evaluation and choice processes. Disclosures clarify information and reduce misleading impressions (Andrews, Burton, and Kees 2011; Hastak and Mazis 2011). In fact, the primary purpose of the Nutrition Facts panel (NFP) was to reduce consumer confusion and to assist consumers in maintaining healthy dietary practices (NLEA 1990; Burton, Garretson, and Velliquette 1999; Hieke and Taylor 2012). Disclosures are particularly useful when consumers lack information or may be misled by other information that is either omitted from, or included, on packaging or promotional material (cf. Hastak and Mazis 2011). In the case of restaurant menu items, where nutrition information generally has not been easily accessible, recent findings show that consumers generally underestimate calories. These incorrect inferences often may result in suboptimal choices, at least for some segments of consumers (e.g., Chandon and Wansink 2007; Howlett et al. 2009; Tangari et al. 2010). For example, restrained eaters may end up consuming more calories than expected.

The prominent display of calories may reduce the potentially misleading effects of a health halo in several different ways (Andrews, Netemeyer, and Burton 1998). First, the disclosure provides relevant information (at least for some consumer segments) that is directly accessible at the point of purchase. This should encourage the retrieval of other potentially relevant information from memory (e.g., “I need to watch my weight”; “there are negative health effects of being overweight”). Second, the mere presence of an information disclosure may be perceived as providing additional diagnostic information that consumers should consider in making product evaluations (Burton, Howlett, and Tangari 2009). Third, disclosures that provide highly specific information may broaden the cognitive frame of reference for evaluations of food options (e.g., beyond attributes of price, size, perceived taste, etc.), and reduce inappropriate generalizations. For these reasons, calorie disclosures appear to have potential benefits for many consumers as noted in Public Law 111-148 (p. 455) which mandates calorie labeling at chain restaurants.

In Studies 1–4, we examine the effects of health halos and anchoring processes on the relative (mis)estimation of calorie and sodium levels for a variety of fast food and table service items that vary in perceived healthfulness. In Studies 2–4, we examine how calorie and nutrition disclosures influence the relative accuracy of calorie and sodium estimates.2 Also of interest is the degree to which disclosures help remedy the misleading effects of health halos, an issue with significant public policy and consumer welfare implications.

STUDY 1

  1. Top of page
  2. Abstract
  3. BACKGROUND
  4. STUDY 1
  5. STUDY 2
  6. STUDY 3
  7. STUDY 4
  8. GENERAL DISCUSSION
  9. REFERENCES

Method

Study 1 focuses on consumers' perceptions of a variety of fast food menu items. The data collection focused on several well-known restaurant chains (e.g., McDonald's, Burger King, Subway) and 20 specific items from four food categories generally perceived as more (salads, grilled chicken sandwiches) and less healthful (hamburgers and fries, fried chicken sandwiches). Study participants were 102 undergraduate students (average age = 22; 59% male) who voluntarily participated for extra credit. They all reported dining at fast food restaurants in the past year, and the average frequency was more than 100 times. Participants also reported high awareness of the restaurant chains used in the study.

Both within and across the meal healthfulness categories, there was substantial variance in calorie and sodium levels. Objective calorie and sodium levels for the menu items, obtained from corporate websites, are shown in Table 1. Participants were given information on serving sizes and brief item descriptions, similar to information typically presented on a menu board. They also were provided with the recommended daily levels for calories and sodium found in the bottom portion of the NFP (e.g., the NFP indicates that daily sodium intake should be less than 2,400 mg). Because restaurants will be required by the US Patient Protection and Affordable Care Act to provide information that will help consumers interpret the nutrition information disclosure, in our study participants are provided with the daily values shown in the bottom part of the NFP (i.e., the level recommended by the FDA). This daily value information serves to minimize the differences in consumers' initial awareness of recommended calories and sodium levels. In this repeated-measures design, for each meal item (no drinks were included), participants estimated both calories and sodium levels. The accuracy of these calorie and sodium estimates was calculated by subtracting the objective values (obtained from corporate websites) from the estimated values provided by participants. The accuracy percentage value was computed by dividing the accuracy difference score by the objective value to determine the level of under(over)estimation of sodium and calories. We also calculated the percentage of respondents under(over)-estimating calories and sodium for each item.

Table 1. Study 1: Differences in the Accuracy of Consumers' Estimates of Sodium and Calories for 20 Restaurant Items
  SodiumCalories
Quick Service Restaurant Menu ItemsRestaurantMean Sodium EstimatesActual Sodium LevelsAccuracy% of Consumers UnderestimatingMean Calorie EstimatesActual Calorie LevelsAccuracy% of Consumers Underestimatingp-Value for Differences in %'s
  1. Note: Numbers in bold are averages for the four meal categories and the overall mean for all 20 items. “Accuracy” is the difference between consumers' sodium (calorie) estimates and the objective levels obtained from websites. The percentages (shown in parentheses) are the mean difference divided by the objective level. The “% of Consumers Underestimating” is the percentage of participants who underestimated (rather than overestimated) the sodium (calorie) levels for each item. The final column showing the “p value for differences in %'s” compares the mean percentage accuracy for sodium vs. the mean percentage accuracy for calories.

Totals across all items 6371676−1039 (−62%)90548754−206 (−27%)67<.0001
Hamburger and fries 8481418−570 (−40%)78810998−188 (−19%)68<.0001
Triple Whopper w/cheese & med. friesBK11042180−1076 (−49%)8910931590−497 (−31%)85<.0001
Double Quarter Pounder w/cheese & med. friesMcDonald's9301600−670 (−42%)808901120−230 (−21%)74<.0001
Hamburger & med. friesBK7641150−386 (−34%)82626650−24 (−4%)57<.0001
Hamburger & med. friesMcDonald's595740−145 (−20%)616326302 (0%)55<.0001
Fried/less healthy chicken sandwich 67522801605 (70%)965931096503 (46%)89<.0001
Tendercrisp chicken sandwich & med. friesBK7012320−1619 (−70%)956121140−528 (−46%)94<.0001
Premium crispy chicken bacon club sandwich & med. friesMcDonald's7232080−1357 (−65%)956531040−387 (−37%)89<.0001
Crispy chicken ranch sandwich & friesChili's7923080−2288 (−74%)1006671580−913 (−58%)98<.0001
Chicken Caesar sandwich & kettle chipsPanera6002460−1860 (−76%)994971070−573 (−54%)94<.0001
Chicken & bacon ranch sandwich (6 in.) & Lay's Classic chipsSubway5611460−899 (−62%)93534650−116 (−18%)69<.0001
Grilled/healthy chicken sandwich 60918301221 (67%)94510748238 (32%)85<.0001
Tendergrill chicken sandwich & med. friesBK6721800−1128 (−63%)95553810−257 (−32%)87<.0001
Premium grilled chicken sandwich & med. friesMcDonald's7061330−624 (−47%)81561750−189 (−25%)80<.0001
Grilled chicken sandwich & friesChili's6722970−2298 (−77%)100575920−345 (−38%)90<.0001
Oven roasted chicken breast (6 in.) & Lay's classic chipsSubway5011300−799 (−61%)94416530−114 (−22%)79<.0001
Chicken salad sandwich & kettle chipsPanera4941750−1256 (−72%)99445730−285 (−39%)88<.0001
Salads 4871215728 (60%)8936731354 (17%)32<.0001
Tendergrill chicken garden salad w/fat free ranchBK5041460−956 (−65%)9635830058 (19%)34<.0001
Bacon ranch salad w/low fat ItalianMcDonald's5361030−494 (−48%)87355200155 (78%)10<.0001
Veggie delight salad w/fat free ItalianSubway354810−456 (−56%)8623195136 (143%)5<.0001
Classic cafe salad w/reduced sugar Asian sesame vinaigrettePanera369580−211 (−36%)77273170103 (61%)17<.0001
Tendercrisp chicken garden salad w/ranchBK5251730−1205 (−70%)97462590−128 (−22%)70<.0001
Bacon ranch salad w/crispy chicken w/ranchMcDonald's6351680−1045 (−62%)935225202 (0%)58<.0001

Scores for each item within each of the four meals categories (i.e., hamburger and fries meals, fried/nongrilled chicken sandwiches, salads, and grilled/baked chicken sandwiches) are shown in Table 1. In addition to the estimates, we gathered information on perceived healthfulness of each item (“This food item is healthy”; 1 = Strongly Disagree to 7 = Strongly Agree). As anticipated, a within subjects analysis revealed significant differences in the perceived healthfulness of the four categories of items (F(3, 300) = 213.1, p < .001). Contrasts showed significant differences between the two less healthful categories shown at the top of the table (hamburgers and fried/less healthy chicken sandwiches); these items were perceived as significantly less healthful (M = 2.25; t = −20.1, p < .001) than the items shown at the bottom of the table (salads and grilled/more healthful chicken sandwiches; M = 3.82). As shown in Table 1, mean objective calorie levels are also lower for the two more healthful categories.

Results

For the 20 items shown in Table 1, while objective calorie and sodium levels are both underestimated, estimated sodium levels are much less accurate. The actual, average sodium level across the items is 1,676 mg (about 73% of the recommended daily amount); the average estimate is more than 1,000 mg lower. Calorie estimates were 206 calories lower than actual levels. In terms of percentages, the level of underestimation for sodium is 62%, substantially higher than for calories (27%, t = 17.73, p < .001). In addition, across the items, the percentage (90%) of consumers who underestimated sodium levels is significantly greater (z = 3.96; p < .001) than the percentage who underestimated calories (67%). These results support H1a and H1b.

Perceived meal healthfulness moderates these differences (F = 375.8; p < .001). A plot of the interaction for accuracy across the more and less healthful meal categories is shown in Figure 1. Across items in the more healthful categories (i.e., salads and grilled chicken sandwiches), calories are slightly underestimated (−17%). However, sodium is underestimated by more than 50% in both the two healthful (−64%) and two less healthful (−58%) categories. Note in Table 1 that for salads, which are lower in calories, consumers overestimate calorie levels (54 more calories), but they continue to substantially underestimate sodium (728 mg or 60% of total). Figure 1 shows that the difference between the accuracy of consumers' estimates for calories and the accuracy of consumers' estimates for sodium is larger (smaller) for the more (less) healthful foods. This pattern of findings is consistent with H2, which predicted that compared to calories, the discrepancy between estimated and actual sodium levels would be larger (smaller) for menu categories perceived as more (less) healthful.

image

Figure 1. Study 1: The Moderating Role of Perceived Meal Healthiness on Accuracy Level of Estimates of Calories and Sodium

Note: The plot shows percentage underestimation of meal item calories and sodium. Items perceived as less healthy include hamburgers and fried chicken sandwiches and French fries. Items perceived as more healthy include grilled/baked chicken sandwiches and salads.

Download figure to PowerPoint

Discussion

This initial study revealed an intriguing pattern of results. Consistent with health halo and anchoring effects, consumers underestimate sodium to a greater extent than calories. In particular, when food is perceived to be more, rather than less healthful, underestimation of sodium level is quite large. Since the objective sodium levels of many of these meals (M = 1,676 mg) are quite high, the potential for the overconsumption of sodium seems quite likely.

While useful in identifying these initial perceptions for a variety of menu items, findings from this pilot study do not address actual consumer choices and consumption. To address this limitation, Study 2 combines several forms of data collection to replicate and extend these findings to actual fast food purchases. From a consumer welfare perspective, this allows us to compare actual and recommended sodium intake levels. In addition, we extend tests of the hypotheses regarding the relative difference between calorie and sodium underestimation to actual food choices. Given Study 1 findings, we anticipate that sodium consumption will be underestimated to a greater extent than calories. Extending the findings of Chandon and Wansink (2007), the health halo effect for sodium consumption should dominate, and calories and sodium underestimations will be higher for the more, rather than less healthful meals. We also address how one form of a nutrition disclosure may affect evaluations.

STUDY 2

  1. Top of page
  2. Abstract
  3. BACKGROUND
  4. STUDY 1
  5. STUDY 2
  6. STUDY 3
  7. STUDY 4
  8. GENERAL DISCUSSION
  9. REFERENCES

Procedures, Participants, and Measures

In the initial phase of data collection, participants kept a diary of their fast food restaurant visits, including purchase price, specific foods and drinks consumed, and overall meal satisfaction. During this phase of the study, there was no mention of specific nutrients such as fat, sodium, or calories to avoid any potential reporting bias. After the diary collection was completed, participants rated each meal occasion in terms of taste, value, estimated calorie and sodium levels, heart disease risk, and likelihood of repurchase.

In the third phase of the data collection, participants accessed restaurant websites to obtain actual calorie and sodium levels of each recorded meal. Accuracy levels were computed by determining the difference between the estimated and actual values. Several days after obtaining these objective calorie and nutrient levels from corporate websites, participants re-rated the food items. Thus, ratings were obtained before and after the collection of objective nutrient information. The responses from these three stages of data collection were merged to create a data set of more than 400 meals in the following fast food meal categories: hamburger-based meals, submarine or deli meat sandwiches, pizza, chicken, and Tex-Mex. On the basis of a 9-point scale, submarine sandwich meals were perceived as significantly more healthful (M = 5.96; p < .001) than chicken or Tex-Mex meals (M's = 4.1 and 4.2). In turn, these three food categories were perceived as more healthful (p < .001) than hamburgers and pizza (M's = 3.2 and 3.1). While males (n = 76; 51%) and females (n = 73; 49%) were equally represented in the sample, males consumed 60% of the fast food meals. The sample was comprised of undergraduate students at a major state university who participated in the study for extra credit. The ages of participants ranged from 19 to 39, and the median number of fast food meals recorded in the one-week diary was five.

Primary measures included estimated calorie and sodium levels of the fast food purchases and the accuracy of these estimates. As is in Study 1, respondents were once again provided with the recommended daily values for calories and sodium found in bottom portion of the NFP. Using the objective nutrient values, individual-level “accuracy” scores were calculated by subtracting objective levels from estimated levels. Also, for each fast food meal purchase, single item measures employing a nine-point response scale (endpoints of very unfavorable (“1”) and very favorable (“9”)) assessed calorie, sodium, and meal repurchase intentions at two points in time (i.e., several days before and several days after objective information was obtained from corporate websites). Specifically, respondents were asked to rate how unfavorable or favorable the calorie and sodium levels were for each meal they consumed. They were also asked how unfavorable or favorable their intentions were to repurchase the same meal. For the measure of likelihood of heart disease, participants were asked if the products would increase/decrease the likelihood of developing heart disease if they were included as a regular part of their diets (Kozup, Creyer, and Burton 2003). Endpoints were “would increase the likelihood” (“9”) and “would decrease the likelihood” (“1”).

Results

Table 2 shows participants' calorie and sodium estimates, objective values obtained from corporate websites, and computed accuracy levels across the five different meal types. Several patterns are evident. While the underestimation of calories may be considered substantial (−270 calories, 28% less than actual level), it is substantially less than the underestimation of sodium levels (−1,150 mg, 59% below actual level). Tests of differences between the calorie and sodium underestimation percentages are significant for the total sample (t = 13.6, df = 405, p < .001), as well as for all five meal types (t-values range from 3.67 to 9.12, p < .001). These findings support the results found in Study 1.

Table 2. Study 2 Reported Meal Consumption: Effects of Meal Type on Meal Estimates and Accuracy for Calorie and Sodium Levels of Actual Fast Food Purchases
 Total Meals (n = 406)Hamburger-based meal (n = 110)Pizza-based meal (n = 50)Submarine Sandwich-based meal (n = 93)Fried chicken/chicken nugget-based meal (n = 99)Mexican food-based meal (n = 54)F-Value
  1. Note: Actual levels are based on objective values obtained from fast food restaurant chain websites. The accuracy level score represents the estimated number of calories (sodium) for the purchased meal minus the actual calorie (sodium) level calculated from nutrition information on the restaurant's website. Negative numbers indicate that the number of calories and sodium in the meal is underestimated by the consumer. The numbers in parentheses are the absolute accuracy level divided by the actual objective level.

  2. a

    Numbers shown for sodium levels are in milligrams.

  3. ap < .001; bp < .01; cp < .05; dp < .10.

Calories
Meal estimate mean6837927565456566795.09a
Actual value mean9531,1011,15483284187512.64a
Accuracy level−270 (−28%)−309 (−28%)−398 (−35%)−287 (−34%)−186 (−22%)−196 (−22%)2.09d
Sodiuma
Meal estimate mean7989398306587448223.12c
Actual value mean1,9481,7642,3682,1151,7651,9825.48a
Accuracy level−1,150 (−59%)−825 (−47%)−1,537 (−65%)−1,456 (−69%)−1,021 (−58%)−1,160 (−59%)7.18a

While the extent to which sodium levels are underestimated is noteworthy, so is the magnitude of the objective values. The mean levels across these fast food meal types range from 1,764 to 2,368 mg, values representing from 77% to 103% of the recommended daily value (2,300 mg). The mean sodium level across all meal types is 1,948 mg, 85% of the recommended daily limit. In contrast, the objective calorie levels are 48% of the recommended daily value for a 2,000 calorie diet.

We also examined the relationship between the actual levels of calories and sodium for the meals that were purchased by the participants in the study. The correlation between the level of calories and sodium is .71 (p < .001), indicating that approximately 50% of the variance in the actual meal sodium levels can be explained by the calorie level of the purchased fast food meals. (Much of this relationship between calories and sodium can probably be explained by meal size; as the size of the meal purchased increases, so does the actual levels of both calories and sodium.) The correlation between the estimated levels of calories and sodium is very similar to that of the actual levels (r = .68; p < .001).

To test the hypothesis related to effects of the disclosure of calories and nutrient levels, we performed a mixed MANOVA with meal type as a between-subjects factor and a repeated-measure factor that consisted of evaluations both before and after exposure to the objective meal nutrition information. The four dependent measures included perceived calorie and sodium evaluations, heart disease risk perceptions, and meal repurchase intentions. Results show a main effect of the disclosure on each of the four dependent variables with the strongest effects for decreases in sodium evaluations (F(1, 444) = 49.5, p < .001, η2 = .10), and repurchase intentions (F(1, 444) = 38.2, p < .001; η2 = .08). The main effect of the disclosure was also significant for calorie perception (F(1, 444) = 7.1, p < .01, η2 = .02) and heart disease (F(1, 444) = 11.5, p < .01, η2 = .03). However, there was also a moderating effect of meal type on the effect of the disclosures for the sodium evaluation (p < .05; η2 = .03) and heart disease perception (p < .01; η2 = .05). As shown in Figure 2, sodium evaluations are less favorable across all meal categories after exposure to objective nutrition information; however, this effect is strongest for the submarine sandwiches (the meal category considered most healthful) and Tex-Mex meals (intermediate in perceived healthfulness). The plot for heart disease (middle of the figure) shows a similar pattern of effects. Specifically, heart disease risk perceptions appear to increase most dramatically after the disclosure for submarine sandwiches. The interaction for repurchase intentions does not reach statistical significance (p > .10, η2 = .01); as shown in the bottom of Figure 2, there is a significant reduction after the disclosure for all meal types.

image

Figure 2. Study 2: The Moderating Role of Meal Type on Effects of a Nutrition Disclosure on Consumer Sodium Evaluations, Heart Disease Perceptions, and Meal Repurchase Intentions

Download figure to PowerPoint

Discussion

For frequent restaurant patrons, it is easy to understand how average daily sodium intake can exceed the recommended intake level. Compared to calories, there is considerably less information presented in the media about sodium. Thus, it is not surprising that, consequently, sodium tends to be underestimated to a greater extent than calories. In general, both the magnitude of the objective sodium levels for these various meal purchases and the level of underestimation of sodium will be of interest to the health community and policy makers. In addition, the results demonstrate that the provision of objective information decreases consumers' repurchase intentions across all types of meals. Again, the halo effect for sodium appears more substantial than that for calories. When the halo is disconfirmed by a nutrition disclosure, the negative effects seem particularly pronounced for submarine and deli meat sandwiches that are perceived as more healthful than other types of fast food meals.

While these Study 1 survey findings and Study 2 quasi-experimental results are intriguing, the prior studies did not address a key public policy concern that is highly relevant for restaurant chains. National legislation will require fast food menu boards to present only calorie information. While sodium information must be available upon request, prior studies suggest that few consumers bother to request information that is not disclosed in highly accessible locations and formats (Roberto, Agnew, and Brownell 2009; Wootan and Osborn 2006). Thus, in this next study we use a between-subjects experiment in which we examine effects of calorie information (vs. a no calorie control) on sodium meal accuracy levels, calorie perceptions, and intent to purchase the meal. As shown in the prior study, calorie level may not necessarily provide diagnostic information that is beneficial for sodium estimation. For example, a submarine sandwich with processed meats may have a lower level of calories but high amounts of sodium. Given no explicit presentation of sodium levels, we expect the calorie disclosure on the menu to interact with the specific meal item in affecting the calorie perception and intent, but not to affect the sodium estimates and accuracy levels.

H3: Compared to a control disclosure condition, a calorie disclosure will (a) have little effect on absolute sodium level estimates across differing meal items and (b) decrease purchase intentions for less healthful products but will have little effect for more healthful products.

STUDY 3

  1. Top of page
  2. Abstract
  3. BACKGROUND
  4. STUDY 1
  5. STUDY 2
  6. STUDY 3
  7. STUDY 4
  8. GENERAL DISCUSSION
  9. REFERENCES

Method

A web-based survey was administered to a panel of adult consumers. Two versions of menu board stimuli were used; one presented brief descriptions of the items along with objective calorie information (calorie condition) while the other only provided the item descriptions (control condition.) Other than the calorie provision manipulation, the menus were identical and included four items from two fast food restaurants. For each restaurant, two paired items varied in size/nutrition levels but were similar in taste. The menus included: (1) Burger King (BK) Whopper with cheese, large fries, and large regular drink (1,550 calories), (2) BK Whopper Junior, medium fries, and medium diet drink (730 calories), and (3) Subway (SW) 12-inch Turkey Sub Sandwich with cheese, Classic Lays Potato Chips, and large regular drink (1,080 calories), (4) SW 6-inch Turkey Sub Sandwich with cheese, Baked Lays Potato Crisps, and medium diet drink (480 calories).3 Thus, within each restaurant, items had similar taste expectations but differed in terms of calorie and nutrient levels. Presentation of the stimuli was counterbalanced for both the restaurants and the menu items and did not affect the results. Prior usage and dining frequency were similar for both restaurants; more than 95% of the participants reported dining at BK or SW in the past. There were 239 participants (53% female; 47% male), and the mean age was 47 years.

Dependent measures of interest included estimates of absolute sodium levels (in mg.) for each of the meals (i.e., respondents were asked to estimate the amount of sodium in milligrams they believed to be in each meal), and a 7-point measure of intent to purchase using not likely (“1”) and very likely (“7”) as scale endpoints.

Results and Discussion

We performed a mixed MANOVA with calorie provision as a between-subjects factor (present or absent) and a repeated-measure factor that consisted of the four meal items. Dependent measures included estimates of sodium level and purchase intention. As anticipated, there was an effect of the calorie provision by item interaction for purchase intentions (F = 4.99, p < .01). However, there was no effect of the interaction of calorie disclosure and item on sodium estimates. Plots are shown in Figure 3.

image

Figure 3. Study 3: The Role of a Menu-Based Calorie Disclosure on Sodium Estimate Accuracy and Purchase Intention

Note: Sodium estimates are measured in milligrams.

Download figure to PowerPoint

As shown in the top of Figure 3, the extent to which absolute levels of sodium are underestimated vary substantially across items (p < .01) and range up to more than 1,800 mg (81% of daily value). Note that the disclosure of calories has essentially no effect on the estimates of milligrams of sodium and the resulting level of inaccuracy, supporting H3a. Results indicate that (bottom of Figure 3) the provision of calorie information has a negative influence on purchase intentions for the BK Whopper meal (F(1, 385) = 7.57, p < .01) and a marginal negative effect on purchase intentions for the Subway 12-inch meal (p = .07). As predicted, calorie provision has no effect on purchase intentions for either of the two more healthful meals (p > .10). These results provide partial support for H3b.

The pattern of results suggests that while calorie disclosure can have some effect on consumers' purchase intentions, it does nothing to help consumers make more accurate estimates of the sodium levels of popular fast food items. Sodium levels are again substantially and consistently underestimated.

While Studies 1, 2, and 3 focused more on popular fast food meals, table service restaurants also serve a significant segment of the population. Unfortunately, many of these foods have high calorie and sodium levels. Even when calorie levels may be relatively modest, sodium is often quite high (CSPI 2009). Thus, while much of the prior research on health halos, anchoring, and calorie under/overestimation has focused on fast food items, extending this research to table service restaurants provides an opportunity to explore boundary conditions of the health halo effects. Will evidence of a stronger health halo for sodium than calories also be found for table service restaurant foods? Will the provision of an extended disclosure that includes sodium influence product evaluations and purchase intentions? Does a consumer's hypertension diagnosis status differentially influence purchase intentions when a sodium disclosure is added to a calorie disclosure, and are any effects consistent across products? These questions are addressed in Study 4.

STUDY 4

  1. Top of page
  2. Abstract
  3. BACKGROUND
  4. STUDY 1
  5. STUDY 2
  6. STUDY 3
  7. STUDY 4
  8. GENERAL DISCUSSION
  9. REFERENCES

Study 4, in addition to the calorie information disclosure assessed in Study 3, includes a more comprehensive nutrition information condition. In this expanded disclosure, sodium and saturated fat are presented in addition to calories. This additional nutrient information is not required by the national menu labeling law, but is required in cities such as Philadelphia. Given these alternatives to required consumer disclosures, we examine the influence of these different menu labeling conditions on nutrient estimates and purchase intentions across various menu items from major table service chains (e.g., Applebee's).

Drawing from tenets of accessibility-diagnosticity (Feldman and Lynch 1988), previous research has indicated that consumers diagnosed with hypertension perceive sodium as more personally relevant and pay more attention to sodium content than to other nutrients (Howlett et al. 2012). Diagnosed consumers will be more likely to have information regarding effects of sodium and information relating to sodium and saturated fat (CDC 2009, 2011) should be more diagnostic in influencing purchase intentions. Thus, hypertensive consumers should be more attentive to product cues related to sodium, such as levels of sodium presented in the context of a menu disclosure, because this information is personally relevant and thus should influence their subsequent purchase decisions. In contrast, such cues should be less relevant to nonhypertensive consumers, and be given far less consideration when deciding what menu items to purchase. As sodium levels increase, hypertensive consumers should become more strongly affected by the disclosure, resulting in an enhanced impact on purchase intentions. This suggests a three-way interaction between hypertensive status, the nutrition disclosure, and the product sodium level.

H4: A nutrition disclosure including sodium will reduce purchase intentions for hypertensive consumers and this reduced effect is more pronounced for higher sodium products. For nonhypertensive consumers, the nutrition disclosure including sodium will have little effect on purchase intentions, when compared to a calorie-only disclosure.

Method

Because the likelihood of hypertension increases with age, we conducted an experiment using a nationwide panel of adult consumers over the age of 45. There were a total of 114 participants with an average age of 56 and the sample was equally divided between males and females (50% male; 50% female). As in the third study, participants completed a web-based survey that included menu-based stimuli. There were three different menu conditions, a control condition with no calorie or nutrient information, a calorie-only condition, and a more comprehensive nutrition information condition that included calorie, saturated fat, and sodium levels. All menu conditions included descriptions of four actual full service restaurant meals from chains that will be affected by the national labeling law. Calorie and nutrient levels were drawn directly from corporate websites. For each item there was a more detailed description that described what was included with the meal. The items included: (1) grilled chicken sandwich (1,240 calories, 2,510 mg sodium), (2) hamburger and fries (1,170 calories, 1,900 mg sodium), (3) Memphis dry rub baby back ribs (full rack) (1,180 calories, 4,720 mg sodium) (4) a marinated chicken breast dish (1,230 calories, 4,390 mg sodium). Note that the calorie levels of the menu items were essentially equivalent, varying by seventy calories or less. The sodium levels varied across the items such that the fiesta lime chicken and Memphis dry rub baby back ribs dishes had much higher sodium levels compared to the other items and were consistent with levels reported by the chains. The order of the items presented on the menu was counterbalanced; order did not influence the results. Participants also were asked whether they have ever been told by a health professional that they have high blood pressure. Those that responded positively were recorded as being diagnosed with hypertension.

The dependent measures included the percentage accuracy of estimates, purchase intentions, and product choice. Percentage accuracy estimates were calculated as in prior studies. Two purchase intention measures were used; correlations between the items ranged between .95 and .98 for the four menu items. Finally, participants were asked, “If you were dining out and all these menu items were available, which one item would you choose?”

Results

Nutrient Estimates

Percentage accuracy estimates for calorie and sodium were determined using the same procedure used in Study 1. There was a main effect of the menu condition (F = 32.88, p < .001), indicating that participants were more accurate in their estimates as they received more nutrient information; however, there was a three-way interaction between menu condition, nutrient, and meal (F = 3.65, p < .01). Plots for calories and sodium are shown in Figure 4. As expected, calorie estimates were close to accurate when participants were given the menu with the calorie information (estimates range from −9% to +9%). (See the top portion of Figure 4.)4 For calorie estimates in the control condition, there are large differences in accuracy for the different menu items (even though objective values are similar). The chicken menu dishes appear to benefit from a health halo related to the healthier perception of chicken and this leads to biased calorie estimates. When participants receive calorie information, estimates are adjusted and there is less variance across the menu items. However, note that when participants were given the menu with full nutrition information, for three out of the four meal items, participants actually overestimate calories (all p's < .05), compared to the calorie-only condition.

image

Figure 4. Study 4: Percentage Accuracy Estimates across Attributes and Menu Nutrient Disclosure Conditions

Note: Calorie and sodium estimates are measured in absolute values (e.g., sodium in mg.). The percentage values are the absolute accuracy levels (estimated number of calories (sodium) for the purchased meal minus the actual calorie (sodium) level (from information on the restaurant's website) divided by the actual levels found on websites).

Download figure to PowerPoint

As expected, sodium estimates were more accurate when participants received the menu with more comprehensive information, compared to the control and calories-only condition. When comparing the plots, we find that for calories in the control condition there tends to be large differences in accuracy for the different menu items. Estimates are adjusted in the conditions where participants receive nutrient disclosure information and there is less variance across the menu items. In contrast, for sodium, in the control and calorie-only condition, sodium underestimation is more substantial (40–80%) and exceeds calorie underestimation for all products. Similar to the findings in the prior study, the provision of calorie information does not have any positive benefit on the accuracy of the sodium estimates.5 Results again support H1.

Purchase Intentions

For purchase intentions we performed a mixed ANOVA with hypertension status, menu labeling condition as the between-subjects factors and the four meals as a within-subjects factor. There was the predicted meal item by menu condition by hypertension status interaction (F = 3.59, p < .01) for purchase intentions. As shown in Figure 5, for hypertensive consumers, there are differences across the menu labeling conditions for the two items highest in sodium, the fiesta lime chicken (F = 6.31, p < .01) and the ribs meal (F = 5.01, p < .01). Contrasts show that the extended nutrition disclosure reduces purchase intentions for both the higher sodium fiesta lime chicken (p < .01) and the ribs meal (p < .01), relative to the disclosure condition containing only calories and the control condition. In contrast, for consumers not diagnosed with hypertension there is not a difference in purchase intentions across the menu labeling conditions (for hamburger and fries meal, p = .058; all others p > .20).

image

Figure 5. Study 4: Effects of Hypertensive Status and Menu Nutrient Disclosure Conditions on Purchase Intentions

Download figure to PowerPoint

Meal Choice

To examine meal choice, the two items that were higher in sodium were combined into one group and the items that were very low in sodium were combined into a second group. Consumers diagnosed with hypertension were more (less) likely to choose the items that were lower (higher) in sodium when they received the calories, saturated fat, and sodium menu condition (χ2 = 7.05, p < .05). When they were exposed to the extended nutrition disclosure, 78% chose a lower sodium product compared to 42% in the calories only or control menu labeling conditions (z = 2.67, p < .01). Conversely, there was no difference in choice (p > .20) across the menu labeling conditions for those that were not diagnosed with hypertension.

GENERAL DISCUSSION

  1. Top of page
  2. Abstract
  3. BACKGROUND
  4. STUDY 1
  5. STUDY 2
  6. STUDY 3
  7. STUDY 4
  8. GENERAL DISCUSSION
  9. REFERENCES

Integrating literature on anchoring processes, health halos, and the role of disclosures in informing consumers when complete attribute information is not available, we address issues related to the mandated disclosure of calories in restaurants (Public Law 111-148). Implications of the pattern of findings for marketing and retail chain management, public health and policy concerns, and consumers are considered.

Overview of Findings

Study 1 suggests that the recent concern regarding the low level of awareness of the high calorie levels of many restaurant items (that helped lead to calorie disclosure legislation) is dwarfed by consumers' (lack of) awareness of sodium levels. Similarly, Study 2 and Study 4 show that this underestimation of sodium relative to calories extends to fast food purchases and full service restaurant items. In both studies, sodium levels are underestimated by a greater percentage of consumers and by greater percentage amounts than are calories, but there are differences across food type categories. Sodium levels of meal types viewed as healthier (e.g., salads in Study 1; sub-sandwiches in Study 2) are underestimated at a more substantial level, suggesting a greater “health halo” effect for sodium than for calories. Because there is less information in memory to make reasonable inferences about sodium levels, halos based on healthy perceptions of food categories or restaurants appear to be amplified to a greater degree than calories. Study 4 suggests that adding other negative nutrients to menu labels may actually influence how consumers estimate calories as well as the other nutrients. In addition, in the fourth study we analyzed consumers that should be more aware of the consequences of sodium (i.e., those diagnosed with hypertension). We found that extended disclosures (beyond calories alone) had large effects on purchase intentions for high sodium products but there was little impact for nonhypertensive consumers. This finding has some important implications for long-term consumer welfare; given that the negative effects of excessive sodium consumption appear to be cumulative, failure to attend to sodium content levels may potentially create significant future health problems for many consumers (Lichtenstein et al. 2006; Howlett et al. 2012).

Implications for Public Policy and Consumer Health

The focus of the new restaurant labeling initiative (Public Law 111-148) only on calorie disclosure has raised some concerns among the public health community. Their concern seems warranted. Our findings show that while consumers have a difficult time estimating calories, their judgments of sodium content are much worse, and health halo effects appear much stronger for sodium. Providing calorie information (alone) provides little help. Some public health officials are concerned because excessive sodium intake is a significant health problem that is finally beginning to be addressed more seriously by health and public policy communities. Recent reports suggest that reducing the daily intake of sodium from the current average of 3,400 mg to the highest recommended level of 2,300 mg6 has the potential to prevent 92,000 deaths and 66,000 strokes per year in the United States (Palar and Sturm 2009). While often passing under the radar of consumer awareness and impact on evaluations, the high sodium content levels of many away-from-home foods have at least occasionally put the restaurant industry on the defensive. For example, the Center for Science in the Public Interest filed a class action suit against a restaurant (Denny's) in 2009, accusing the firm of “deceptive and unconscionable commercial practices in its sales of meals containing alarmingly large and undisclosed amounts of sodium” (Shapley 2009). Following the lead of the British government, which initiated a program to reduce sodium consumption in 2003, there are now similar initiatives in Canada, Australia, New Zealand, and much of Europe. The Institute of Medicine (IOM 2010) recently made recommendations about various ways to reduce salt intake and noted that an enhanced understanding of factors that affect consumer awareness and behavior relative to sodium is needed. This recommendation is reinforced by the Study 4 results that show that purchase intentions and choice are strongly influenced by a disclosure that includes sodium for those with hypertension, but there is little effect for nonhypertensive consumers.

Consistent with this IOM objective to increase “understanding and awareness” of sodium levels consumed from restaurant foods, Public Law 111-148 has offered substantial flexibility to the FDA. They could consider various options for making sodium information available to consumers, particularly for table service restaurants where menus offer some flexibility and supplemental pamphlets disclosing information that is more comprehensive could be made available on tables. Of course, sodium could be required on menus and menu boards, but adding information to drive-thru's and menu boards where space is limited would lead to significant operational challenges. Certainly, the above programs used to create greater awareness of sodium issues, in general, could be helpful in sensitizing consumers to the sodium issue for restaurant fare, and perhaps this will increase the requests for sodium disclosure information regardless of how it is made available.

Implications for Marketers

The new restaurant labeling initiative (Public Law 111-148; p. 455) focuses only on calorie disclosures. Our findings suggest that the perceived healthfulness of the food category interacts with information provision to influence calorie perceptions and purchase intentions, but calorie disclosures do not affect sodium evaluations or estimation accuracy. These findings suggest that because sodium and calories are not always highly correlated, restaurants will be able to continue to use high levels of sodium to enhance or maintain the flavor of reduced calorie levels. (A full day's worth of sodium contains essentially no calories; salt can be added to foods with no impact on calories, fat, or other negative nutrients.) While sodium and other nutrients must be made available upon request from the restaurants, research suggests that it is unlikely to be accessed by consumers, particularly if calorie information is available (Roberto, Agnew, and Brownell 2009). This suggests that restaurants may offer new and modified products that reduce the disclosed calorie levels but may increase sodium levels without being directly considered or noticed by most consumers. Consumers will infer sodium levels from internal information such as the healthfulness of the category and the disclosed calorie information. They are likely to be affected by the product/category health halo, and continue to substantially underestimate sodium level. While this offers an opportunity for marketing and retail management, as noted below it raises some important issues that should be considered by the health community and regulatory agencies (i.e., FDA) concerned about sodium consumption.

Conceptual Implications for Future Consumer Welfare Research

Prior research has not fully explored boundary conditions that determine when health halos will have the strongest biasing effects. Health halos are more likely to have similar effects for correlated attributes most directly related to a superordinate attribute of the “healthfulness” category positioning. The perceived healthfulness of a food category is generally related to calorie and fat perceptions (IFIC 2009; Schmidt 2009). Calorie and fat levels in restaurant foods are highly correlated because a gram of fat contains more than twice as many calories as a gram of protein or carbohydrates. Conversely, the attributes of sodium and calories are not likely to be as strongly correlated because sodium contains no calories and thus can be added or taken away from a product without influencing calorie level. For example, in Study 2 the correlation between the actual levels of calories and sodium for the meals purchased was .71, indicating that the calorie level would explain about 50% of the variance in the sodium level. In contrast, the correlation between calorie and fat levels was .86. Thus, the stronger positive relationship between the superordinate concept (healthful category positioning) and the subordinate attribute of calories should lead to somewhat less biased inferences about calorie levels than for sodium levels. The nature of the underlying objective relationships between the superordinate concept and the subordinate attributes helps determine the level of the resulting bias. Such considerations extend research on inferences based on interattribute comparisons and health-related schemas (Hastak and Mazis 2011), and offer guidance for conceptually determining when disclosures should be most useful (from an informational perspective) in correcting misleading consumer inferences.

Limitations and Additional Future Research

This research is subject to several important limitations, and findings suggest some additional future research. Two of the four studies employed a student sample of consumers generally under 30 years of age, and this may impact the generalizability of findings due to lesser interest in nutrition in general, and sodium in particular. While sodium tends to have a cumulative effect over the lifetime, and is thus relevant for children, adolescents, and young adults, it certainly will not be a focal concern for many young adults without concerns about hypertension or future health. However, we suspect that the underestimation of sodium is not influenced by age, a belief that is consistent with the level of underestimation across items found for the adult samples in Study 3 and Study 4. Sodium knowledge is simply not high for consumers and levels are not integrated into evaluations without additional priming or promotion. Future research may consider the inferred relationship between unhealthiness and taste with respect to chain restaurant foods that vary in their levels of sodium vis-à-vis calories (Raghunathan, Naylor, and Hoyer 2006).

Another consideration relates to the levels of sodium and the way it was measured. Since sodium is measured in milligrams, the sodium values often appear somewhat higher than calories for the items used in the studies. The use of these larger numbers may explain some small portion of the estimation error reported by study participants. Future studies might compare findings if sodium is presented and assessed in grams versus milligrams.

We also acknowledge that all the studies occurred outside of a restaurant environment, where situational and contextual cues may affect responses (Burton and Kees 2012). Because there are markets in which labeling currently exists (e.g., New York City, Seattle, Philadelphia), as well as control markets without labeling, it would be intriguing to conduct quasi-experimental field studies focusing on consumer search and awareness of sodium and calorie perceptions and consumption (Balasubramanian and Cole 2002). Some recent research shows that calories consumed from fast food have decreased recently, and it would be interesting to track similar changes in sodium (Hellmich 2013). Also, given that the FDA has some flexibility in the specific regulations it proposes associated with national menu and menu board labeling initiatives (i.e., Public Law 111-148), studies that addressed whether calorie labeling has had an impact on sodium consumption, in comparison to markets requiring more explicit information on sodium in disclosures (e.g., Philadelphia), would be of substantial interest. Of course, as mandated calorie labeling is introduced into the national market, this will offer many opportunities for field studies that address sodium awareness and evaluations across chains that make the supplemental sodium information available in differing disclosure formats.

REFERENCES

  1. Top of page
  2. Abstract
  3. BACKGROUND
  4. STUDY 1
  5. STUDY 2
  6. STUDY 3
  7. STUDY 4
  8. GENERAL DISCUSSION
  9. REFERENCES
  • Andrews, J. Craig, Scot Burton, and Jeremy Kees. 2011. Is Simpler Always Better? Consumer Evaluations of Front-of-Package Nutrition Icons. Journal of Public Policy & Marketing, 30 (Fall): 175190.
  • Andrews, J. Craig, Richard G. Netemeyer, and Scot Burton. 1998. Consumer Generalization of Nutrient Content Claims in Advertising. Journal of Marketing, 62 (October): 6275.
  • Balasubramanian, Siva K. and Catherine Cole. 2002. Consumers' Search and Use of Nutrition Information: The Challenge and Promise of the Nutrition Labeling and Education Act. Journal of Marketing, 66 (3): 112127.
  • Block, Gladys. 2004. Foods Contributing to Energy Intake in the US: Data from NHANES III and NHANES 1999–2000. Journal of Food Composition and Analysis, 17 (3–4): 439447.
  • Burton, Scot, Elizabeth Creyer, Jeremy Kees, and Kyle Huggins. 2006. Attacking the Obesity Epidemic: The Potential Health Benefits of Providing Nutrition Information in Restaurants. American Journal of Public Health, 96 (9): 16691675.
  • Burton, Scot, Elizabeth Howlett, and Andrea Tangari. 2009. Food for Thought: How Will the Nutrition Labeling of Quick Service Restaurant Menu Items Influence Consumers' Product Evaluations, Purchase Intentions, and Choices? Journal of Retailing, 85 (3): 258273.
  • Burton, Scot and Jeremy Kees. 2012. Flies in the Ointment? Addressing Potential Impediments to Population-Based Health Benefits of Restaurant Menu Labeling Initiatives. Journal of Public Policy and Marketing, 31 (Fall): 232239.
  • Burton, Scot, Judith Garretson, and Anne Velliquette. 1999. Implications of Accurate Usage of Nutrition Facts Panel Information on Food Product Evaluations and Purchase Intentions. Journal of the Academy of Marketing Science, 27 (Fall): 470480.
  • Center for Science in the Public Interest. 2009. Heart Attack Entrées and Side Orders of Stroke. http://cspinet.org/new/pdf/cspirestaurantsaltreport.pdf.
  • Centers for Disease Control and Prevention. 2009. Application of Lower Sodium Intake Recommendations to Adults—United States, 1999–2006. http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5811a2.htm.
  • Centers for Disease Control and Prevention. 2011. Legal and Policy Resources on Public Health “Winnable Battles”: Sodium Reduction. http://www2a.cdc.gov/phlp/winnable/sodium_reduction.asp.
  • Centers for Disease Control and Prevention. 2013. World Salt AwarenessWeek. http://www.cdc.gov/features/sodium/
  • Chandon, Pierre and Brian Wansink. 2007. The Biasing Health Halos of Fast Food Restaurant Health Claims: Lower Calorie Estimates and Higher Side-dish Consumption Intentions. Journal of Consumer Research, 34 (3): 301314.
  • Chernev, Alexander. 2010. The Dieter's Paradox. Journal of Consumer Psychology, 21 (2): 178183.
  • Feldman, Jack M. and John G. Lynch. 1988. Self-Generated Validity and Other Effects of Measurement on Belief, Attitude, Intention, and Behavior. Journal of Applied Psychology, 73 (3): 421435.
  • Food and Drug Administration. 2012. Lowering Salt in Your Diet. http://www.fda.gov/ForConsumers/ConsumerUpdates/ucm181577.htm.
  • Hastak, Manoj and Michael Mazis. 2011. Deception by Implication: A Typology of Truthful but Misleading Advertising and Labeling Claims. Journal of Public Policy & Marketing, 30 (Fall): 157167.
  • Hellmich, Nancy. (2013). Adults Gobbling Fewer Calories from Fast Food. USA Today. http://www.usatoday.com/story/news/nation/2013/02/20/fast-food-intake-drops/1932871/
  • Hieke, Sophie and Charles R. Taylor. 2012. A Critical Review of the Literature on Nutritional Labeling. Journal of Consumer Affairs, 46 (1): 122156.
  • Hogarth, Robin M. and Hillel J. Einhorn. 1992. Order Effects in Belief Updating: The Belief-Adjustment Model. Cognitive Psychology, 24 (1): 155.
  • Howlett, Elizabeth, Scot Burton, Andrea Tangari, and My Bui. 2012. Hold the Salt! Effects of Sodium Information Provision, Sodium Content, and Hypertension on Perceived Cardiovascular Disease Risk and Purchase Intentions. Journal of Public Policy & Marketing, 31 (Spring): 418.
  • Howlett, Elizabeth H., Scot Burton, Kenneth Bates, and Kyle A. Huggins. 2009. Coming to a Restaurant Near You? Potential Consumer Responses to Nutrition Information Disclosure on Menus. Journal of Consumer Research, 36 (October): 494503.
  • IFIC. 2009. Consumer Sodium Research: Concern, Perceptions and Action. http://www.foodinsight.org/Content/6/FINAL-IFIC-Sodium-Consumer-Research-Report-8-14-09.pdf.
  • Institute of Medicine. 2010. Strategies to Reduce Sodium Intake in the United States (Report Brief). http://www.iom.edu/media/Files/Report%20Files/2010/Strategies-to-Reduce-Sodium-Intake-in-the-United-States/Strategies%20to%20Reduce%20Sodium%20Intake%202010%20%20Report%20Brief.ash
  • Johnson, Eric J. and David A. Schkade. 1989. Bias in Utility Assessments: Further Evidence and Explanations. Management Science, 35 (4): 406424.
  • Kardes, Frank R., Steven S. Posavac, and Maria L. Cronley. 2004. Consumer Inference: A Review of Processes, Bases, and Judgment Contexts. Journal of Consumer Psychology, 14 (3): 230256.
  • Kozup, John, Elizabeth H. Creyer, and Scot Burton. 2003. Making Healthful Food Choices: The Influence of Health Claims and Nutrition Information on Consumer's Evaluations of Packaged Food Products and Restaurant Menu Items. Journal of Marketing, 67 (April): 1934.
  • Lichtenstein, Alice, Lawrence J. Appel, Michael Brands, Mercedes Carnethon, Stephen Daniels, Harold A. Franch, Barry Franklin, Penny Kris-Etherton, William S. Harris, Barbara Howard, Njeri Karanja, Michael Lefevre, Lawrence Rudel, Frank Sacks, Linda Van Horn, Mary Winston, and Judith Wylie-Rosett. 2006. Diet and Lifestyle Recommendations Revision 2006: A Scientific Statement from the American Heart Association Nutrition Committee. Journal of the American Heart Association, 114 (1): 8296.
  • Liebman, Bonnie. 2010. Shaving Salt, Saving Lives. Nutrition Action Health Letter, 37 (3): 16.
  • Nutrition Labeling and Education Act. 1990. Public Law 101-535 (104): Stat. 2355.
  • Palar, Kartika and Roland Sturm. 2009. Potential Societal Savings from Reduced Sodium Consumption in the U.S. Adult Population. American Journal of Health Promotion, 24 (1): 4957.
  • Raghunathan, Rajagopal, Rebecca Walker Naylor, and Wayne D. Hoyer. 2006. The Unhealthy = Tasty Intuition and Its Effects on Taste Inferences, Enjoyment, and Choice of Food Products. Journal of Marketing, 70 (4): 170184.
  • Roberto, Christina A., Henry Agnew, and Kelly D. Brownell. 2009. An Observational Study of Consumers' Accessing of Nutrition Information in Chain Restaurants. American Journal of Public Health, 99 (5): 820821.
  • Roe, Brian, Alan S. Levy, and Brenda M. Derby. 1999. The Impact of Health Claims on Consumer Search and Product: Evaluation Outcomes: Results from FDA Experimental Data. Journal of Public Policy & Marketing, 18 (Spring): 89105.
  • Ross, William T. and Elizabeth H. Creyer. 1992. Making Inferences about Missing Information: The Effects of Existing Information. Journal of Consumer Research, 19 (1): 1425.
  • Schmidt, David B. 2009. United States Consumers' Perception and Use of Nutrition and Health Information on Food Labels. IFIC, http://www.nutritionthailand.or.th/upload/docpath/S56_1.pdf.
  • Seiders, Kathleen and Ross D. Petty. 2004. Obesity and the Role of Food Marketing: A Policy Analysis of Issues and Remedies. Journal of Public Policy and Marketing, 23 (Fall): 153170.
  • Shapley, Dan. 2009. Should Moons Over My Hammy Come with a Warning Label? http://www.thedailygreen.com/healthy-eating/eat-safe/dennys-sodium-47072402.
  • Tangari, Andrea Heintz, Scot Burton, Yoon-Na Cho, Elizabeth Howlett, and Anastasia Thyroff. 2010. Weighing in on Fast Food Consumption: The Effects of Meal and Calorie Disclosures on Consumer Fast Food Evaluations. Journal of Consumer Affairs, 44 (Fall): 431462.
  • Tversky, Amos and Daniel Kahneman. 1974. Judgment under Uncertainty: Heuristics and Biases. Science, 185 (4157): 11241131.
  • U.S. Patient Protection and Affordable Care Act. 2010. Public Law: 111–148.
  • Wansink, Brian. 2010. From Mindless Eating to Mindlessly Eating Better. Physiology & Behavior, 100 (5): 454463.
  • Wansink, Brian and Pierre Chandon. 2006a. Meal Size, Not Body Size, Explains Errors in Estimating the Calorie Content of Meals. Annals of Internal Medicine, 145 (5): 326332.
  • Wansink, Brian and Pierre Chandon. 2006b. Can “Low Fat” Nutrition Labels Lead to Obesity? Journal of Marketing Research, 43 (4): 605617.
  • Wootan, Margo G. and Melissa Osborn. 2006. Availability of Nutrition Information from Chain Restaurants in the United States. American Journal of Preventive Medicine, 30 (March): 266268.
  1. 1

    The FDA continues to finalize the specific guidelines for the national labeling of restaurant items, and there will be extended time allowed for implementation in restaurants. The specific guidelines are expected to go into effect during 2014.

  2. 2

    In Studies 1–4, sodium estimates are measured in actual milligrams. Moreover, when the terms “underestimation” or “accuracy” are used, they refer to the difference in study participants' estimates and the absolute values of the item measured (e.g., sodium level in mg or calorie level found on the restaurant chains' websites).

  3. 3

    While meal sodium levels were not disclosed on either version of the menu, they ranged from a low of 1,180 mg (approximately 50% of the recommended daily value) for the Whopper Junior meal to a high of 2,580 (greater than 100% of the daily value) for the 12-inch turkey Subway sandwich meal.

  4. 4

    Participants responded to the estimate questions near the end of the survey (after choice and purchase intentions dependent variables) and did not have the menu directly available when responding to the estimation questions. Placing the estimate questions after these more general dependent variables prevents responses to these questions focusing on sodium and calories from potentially influencing the choice and intention measures. Given the placement on the survey, some errors in estimates are anticipated even when the calorie and or sodium information was available on the menu.

  5. 5

    There also was a main effect of hypertensive status on accuracy level (p < .05). Across all four meal options, hypertensive consumers underestimated sodium levels by a greater percentage than did nonhypertensive consumers.

  6. 6

    For more than 50% of the adult population in the United States, the most recent recommendation is 1,500 mg of sodium.