Abstract
This research finds that how a firm conveys a food’s organic nature through an organic label impacts consumers’ evaluation of that food. Consistent with previous research, adding organic labels to foods is detrimental to evaluations of vice (but not virtue) foods, but simple changes to the structure of the organic label attenuate the negative effect of such labels on evaluations of vice foods. Specifically, whereas product-level organic labels (e.g., “organic burrito”) result in lower evaluations of vice foods, ingredient-level organic labels (e.g., “burrito with all organic ingredients”) do not. No effect of organic label structure is found for virtue foods. The authors draw on theories of feature-based categorical typicality and fluency to suggest one psychological process by which organic label structure can impact consumers’ evaluations of vice foods.
Although the percentage of overall U.S. food sales that are organic remains relatively low (only 5%–6%; Reiley 2019), the organic food industry is booming; it accounted for over $50 billion in sales as of 2018 (Organic Trade Association [OTA] 2019), an increase from $3.6 billion in 1997 (OTA 2016). Given the size and rapid growth of this industry, understanding how consumers evaluate organic foods is of great importance. This importance is reflected in the multitude of articles in marketing journals that examine this topic (Fernqvist and Axelson 2013; Loureiro and Lotade 2005; Loureiro, McCluskey, and Mittelhammer 2002; Shafie and Rennie 2012; Van Doorn and Verhoef 2011, 2015). If a general takeaway can be found in these articles, it is that consumers’ evaluations of and preferences for organic foods are nuanced and driven by a myriad of factors that can either be specific to the individual or pertain to the food they are evaluating.
In contrast, our research focuses on how organic foods are labeled and, more specifically, examines the consequences of applying an organic label at the product level (e.g., “organic burrito”) versus at the ingredient level (e.g., “burrito with all organic ingredients”). Given the nature of these two types of labels, the scope of this research is inherently limited to composite foods (e.g., a deli sandwich), which are processed foods comprised of multiple ingredients that are sometimes, but not always, processed themselves (e.g., cooked and sliced deli meat, cheese, olive oil, lettuce, tomatoes) and combined to form the composite food. Simply put, one cannot apply a label to the ingredients of a food if that food cannot be deconstructed into constituent ingredients, as would be the case with milk, for instance.
The main findings presented in this article are twofold. First, adding organic labels to foods is detrimental for evaluations of vice (but not virtue) foods, which is consistent with previous research (Van Doorn and Verhoef 2011, 2015). However, the generally negative impact of organic labels on evaluations of vice foods is qualified by another factor. Whereas product-level organic labels lead to significantly lower evaluations of vice foods relative to nonorganic vice foods, ingredient-level organic labels do not. Organic label structure does not significantly influence evaluations of virtue foods.
In short, the findings presented in this research demonstrate that consumers do not universally reject vice foods that are organic. In reality, consumers are quite sensitive to how the organic nature of the vice food is revealed to them through the organic label. In what follows, we first delve deeper into consumers’ evaluations of organic foods, with a particular focus on previous findings regarding consumers’ responses to organic–vice and organic–virtue foods. We then discuss the distinction between product- and ingredient-level organic labels in greater detail and present evidence that this distinction is neither contrived nor artificial by examining a large sample of labels appearing on real-world organic foods. We next draw on theories of feature-based categorical typicality and fluency to suggest one psychological process by which organic label structure can have an impact on consumers’ evaluations of vice (but not virtue) foods. This, in turn, is followed by our empirical package, which demonstrates the effect of organic label structure on consumers’ evaluations of organic foods. We close with a discussion of the implications of these findings, as well as the limitations of this work.
How Organic Label Structure Can Impact the Evaluation of Organic Foods
Consumers’ Responses to Organic Foods
To date, research on consumers’ responses to organic foods has largely drawn the broad conclusion that consumers tend to respond positively to organic foods because of the perceived health and ethical benefits such foods offer (Fernqvist and Axelson 2013; Loureiro and Lotade 2005; Shafie and Rennie 2012). However, organic foods can be off-putting to price-sensitive consumers (Shafie and Rennie 2012) and uncompelling to consumers skeptical of their benefits (Loureiro, McCluskey, and Mittelhammer 2002).
Of particular interest for the current research, researchers have shown that consumers reliably prefer vice foods (e.g., beer, salty snacks, pizza, burritos) less when the foods are organic than when they are not (Van Doorn and Verhoef 2011, 2015). To be clear, vice foods are those that provide immediate pleasure (e.g., tastiness) but have long-term negative consequences (e.g., weight gain), whereas virtue foods are less immediately pleasurable but have greater long-term benefits (Wertenbroch 1998). Van Doorn and Verhoef (2011, 2015) argue that consumers’ diminished preference for vice foods that are organic (vs. not) is the consequence of a liability effect: the perceived “wholesomeness” of being organic reduces the expected hedonic rewards (enjoyment and pleasure) that consumers anticipate to receive from consuming the vice food (i.e., an unhealthy = tasty heuristic). It is important to emphasize that Van Doorn and Verhoef’s findings do not imply that making vice foods organic will reduce the overall amount of vice foods consumed (i.e., it is not the case that consumers will choose healthier virtue foods over organic vice foods). Their findings merely indicate that consumers will choose nonorganic over organic vice foods.
A noteworthy characteristic of Van Doorn and Verhoef’s (2011, 2015) research, and research on organic foods in general, is that the research question focuses on differences between foods that are organic versus not organic. Although such questions are undeniably fundamental to understanding consumers’ evaluations of organic foods, they neglect to consider important differences that might arise as a function of how a firm conveys a food’s organic nature to the consumer. If decades of work on framing effects have taught us anything, it is that how something is said is as important as what is said. Accordingly, here we focus on how a firm conveys a food’s organic nature through its organic label structure and, in turn, how the specific organic label structure affects consumers’ evaluations of that food.
Organic Label Structure
At present, the U.S. Department of Agriculture (USDA) has approved four distinct labeling categories for organic products: (1) “100% organic,” (2) “organic,” (3) “made with organic [fill in ingredient here],” and (4) specific ingredient lists. 1 So, labels such as “100% organic,” “all organic,” and “made with all organic ingredients” are all within the legal ambit of the definition so long as the product’s true nature conforms to the label’s claims. However, although the presence of an organic label is strictly regulated, the manner in which those labels are structured (i.e., using either product- or ingredient-level labels) is at the discretion of the firm. Product-level organic labels precede the composite food’s name (e.g., “organic burrito,” “organic burger,” “organic salad”), whereas ingredient-level organic labels follow the composite food’s name and refer to its ingredients (e.g., “burger with all organic ingredients”).
This distinction between product- and ingredient-level organic labels is neither artificial nor contrived. This was confirmed by an analysis of new product introductions in the United States and Canada between 2010 and 2012 obtained from Mintel’s Global New Products Database. 2 Of the 38,831 new products introduced across 353 categories, we identified 2,683 sold under an organic label. Of these, 1,439 (54%) made the claim at the product level (e.g., “Back to Nature Organic Shortbread Cookies” and “Earth Balance Organic Soy Nog”). The other 1,244 (46%) products made an organic claim at the level of individual ingredients but not at the product level (e.g., “Ozery Bakery’s Crispy Pitas made with organic wheat” and “Arrowhead Mills Pancake and Waffle Mix made with organic sprouted grain”). Although firms regularly use both product- and ingredient-level organic labels, it is noteworthy that the vast majority of ingredient-level labels refer to one or a few ingredients rather than all of the ingredients in the product. Perhaps this reflects compliance with the FDA regulations discussed previously, but it also highlights that marketers of completely organic food products, which could be given “all organic ingredients” labels, are currently ignoring the option of using ingredient-level labels in lieu of product-level labels. The current research suggests that doing so may be a mistake for many food manufacturers.
Why Might Organic Label Structure Influence Evaluations of Organic Vice Foods but Not Organic Virtue Foods?
Although manufacturers and retailers use both product- and ingredient-level organic food labels, it is not clear, a priori, if such decisions have an impact on consumers’ evaluations of those foods. Consumers’ evaluations of foods can be complex and based on numerous inputs, including social norms (Herman, Roth, and Polivy 2003), inferences (Parker and Lehmann 2014), and personal tastes. Then again, simply describing ground beef as being “80% lean,” as opposed to “20% fat,” results in significantly more favorable evaluations of that meat (Levin, Schneider, and Gaeth 1998; Levin and Gaeth 1988). So, food labels are both easily manipulated by firms and consequential for consumers’ evaluations of the foods to which they are applied. The negative impact of product-level (but not ingredient-level) organic labels on evaluations of vice but not virtue foods found in the previously discussed research might partially be explained using the combination of two streams of literature: studies examining (1) feature-based categorical typicality and (2) fluency.
Feature-based categorical typicality
A comprehensive discussion of the psychological principles of categorization and its consequences on mental representations, judgments, and preferences is outside the scope of the current work. For our purposes, it suffices to say that firms use labels to disambiguate category membership. This is often the case in consumer goods and services categories in which marketers wish to convey information about their offerings. A particularly ubiquitous example is that of brands and their products. Each brand is a category and each product a member of that category (Parker et al. 2018). Firms apply brand names to products so that consumers will use category-based induction (Rein, Goldwater, and Markman 2010) to infer qualities of the considered product (e.g., “Since this is a Nike product, it must be high quality.”).
Critically, although labels disambiguate membership, consumers still regularly differentiate members of categories on the basis of their typicality (Hampton 2007). For instance, although Levi’s sells both jeans and sweaters, consumers consider jeans to be far more typical of the brand despite knowing the brand offers both products (Reinholtz, Bartels, and Parker 2015). The process by which typicality is assessed can be quite complex, but in general it is done by comparing the features (be they concrete, abstract, or symbolic) of the members with the features one expects of members of that category (Rosch and Mervis 1975). The more congruent a member’s features are with those expected of the category (i.e., the greater the feature overlap) the more typical the object is considered of the category. As we discuss in detail subsequently, we argue that organic composite vice (virtue) foods have features that render them as being highly atypical (typical) of organic foods.
Fluency
The importance of perceived category typicality in consumers’ attitudes toward and evaluations of available options has been previously established (Loken and Ward 1990). Simply stated, consumers have a strong tendency to find more typical members of a category more appealing (Landwehr, Labroo, and Herrmann 2011; Loken and Ward 1990; Mandler 1982; Meyers-Levy and Tybout 1989; Veryzer and Hutchinson 1998). One reason this occurs is because atypical category members are difficult to understand (i.e., are conceptually disfluent; Alter and Oppenheimer 2009), especially when considered in relation to the category and the features that are expected of that category. Previous research has found that consumers tend to evaluate conceptually disfluent stimuli more negatively due to the negative affect arising from disfluent processing (Reber, Winkielman, and Schwarz 1998; Seamon et al. 1995; Winkielman and Cacioppo 2001). For instance, consumers hold less favorable attitudes toward a brand when its product is advertised in a conceptually disfluent manner (Lee and Labroo 2004) and when its employees behave in a manner inconsistent with the brand’s positioning (Sirianni et al. 2013). Conversely, consumers like numbers that are more easily processed and, consequently, brands with those numbers in their names (King and Janiszewski 2011). So, if vice (virtue) foods are (not) considered atypical of organic foods, vice (virtue) foods labeled as organic will (not) be disfluent and will (not) be evaluated less positively.
Why organic label structure might influence perceived typicality and fluency
Placing an organic label on a food product establishes it as a member of the category “organic foods.” Consequently, the features expected of organic foods are the basis for assessing the typicality of that food, which will in turn at least partially serve as a basis for evaluating that food product. Accordingly, it is critical to know what features consumers expect of organic foods to better understand which foods they will consider more (or less) typical, thus receiving them more (or less) favorably when the foods are labeled as organic.
Fortunately, there already exists evidence that “healthy” is an expected feature of organic foods (Pew 2016; Van Doorn and Verhoef 2011, 2015). Researchers have also found that consumers expect organic foods to be fresh, natural, unprocessed, and farm-raised/grown (Grzybowska-Brzezinska et al. 2017; Hilverda, Jurgens, and Kuttschreuter 2016; Moser, Raffaelli, and Thilmany-Mcfadden 2011). Two of our own independent studies (presented in the Web Appendix, pp. 2–6) yielded similar findings. Thus, we can say with confidence that the feature “healthy” is the most expected feature of organic foods, to the point of dominating other features in assessing the typicality of foods in the organic food category. Of course, vice foods are inherently unhealthy, whereas the opposite is true of virtue foods. So, vice foods should be considered quite atypical of organic foods, whereas the opposite should be true of virtue foods.
Many of the other features expected of organic foods—fresh, natural, unprocessed, and farm-raised/grown—suggest a general proclivity to expect that organic foods have the feature of being “raw” or “unaltered” from their original state (i.e., unprocessed). So, it stands to reason that ingredients will, on average, be considered quite typical of organic foods, whereas composite foods will be considered quite atypical. We make this inference because as one transitions from composite foods to the ingredients that comprise them, one also transitions from more processed and altered foods to more raw and unaltered foods.
With a better understanding of the features that consumers expect of organic foods, it follows that organic labels would lead to less favorable evaluations of vice (but not virtue) foods. This should be especially true of composite vice foods, which are neither healthy nor raw/unaltered foods. So, applying an organic label to composite vice foods at the product level should place strong downward pressure on consumers’ evaluations of those foods. In contrast, applying the same product-level organic label to composite virtue foods should be far less damaging to consumers’ evaluations of those foods given their high degree of healthiness.
Extending this logic, placing the organic label at the ingredient level versus the product level applies the “organic foods” category to the ingredients as opposed to the composite product. Since ingredients should be considered more typical of the organic foods category, ingredient-level organic labels should be less disfluent, particularly in the case of vice composite foods. If so, ingredient-level organic labels should attenuate the negative impact of placing an organic label on vice foods. This effect should be less evident for virtue foods, the evaluations of which are less affected by organic labels in general. Formally,
The four studies presented next find the predicted effect of organic label structure on consumers’ evaluations of foods, as well as important boundaries of that effect. Studies 1a and 1b find that product-level organic labels result in more negative evaluations of composite vice foods relative to ingredient-level labels (1a and 1b) and nonorganic vice foods (1b). Study 2 demonstrates that the impact of organic label structure is more apparent for vice than virtue foods, which suggests that at least a portion of the effect is driven by perceptions of categorical typicality and the consequently experienced (dis)fluency. Study 3 finds that the negative impact of product-level organic labels on vice foods is attenuated after consumers view an image portraying the ingredients of the focal food. We close with a discussion on the implications and limitations of our findings.
Empirical Evidence
Studies 1a and 1b: The Impact of Product- Versus Ingredient-Level Organic Labels
Studies 1a and 1b demonstrate the basic effect of organic label structure (ingredient- vs. product-level) on evaluations of composite vice foods by measuring the likelihood of ordering a dish at a restaurant. Both studies contrast ingredient- and product-level organic labels, but Study 1b additionally examines a control condition in which the dish does not have an organic label (i.e., is a nonorganic composite vice food) to better understand how label structure influences preferences for organic versus nonorganic composite vice foods.
Study 1a: The Role of Organic-Label Structure on Evaluations of Burgers
Method
Burgers were the focal food in this study. We asked 195 Amazon Mechanical Turk (AMT) participants to imagine they were at a restaurant. A pretest with 63 AMT participants verified that burgers are considered a composite vice food (see Web Appendix, p. 9).
We randomly assigned participants to either the ingredient-level condition, which presented them with a dish called “Our Famous Burger (w/ all organic ingredients),” or the product-level condition, which presented them with a dish called “Our Famous Organic Burger.” Participants then indicated the likelihood with which they would order this dish on a 1 (“very unlikely”) to 10 (“very likely”) scale.
Results
The likelihood of ordering the dish was significantly higher in the ingredient- (vs. product-) level condition (Mingred = 7.18, SD = 2.42 vs. Mprod = 5.79, SD = 2.80; F(1, 193) = 13.75, p < .001; the complete analysis of variance (ANOVA) table is reported in the Web Appendix, p. 12).
Study 1b: Labeling Affects the Relative Evaluations of Organic Versus Nonorganic Vice Foods
Method
Participants in Study 1b (201 AMT workers) evaluated burritos and fajitas. The same pretest described in Study 1a found that burritos and fajitas are considered composite vice foods (see Web Appendix, p. 9). We randomly assigned participants to one of six conditions in a 2 (food type: burrito vs. fajitas) × 3 (organic label structure: nonorganic vs. ingredient-level vs. product-level) between-subjects design. Although food type was a between-subjects factor, this we manipulated this factor solely for the purposes of examining the robustness and generalizability of the effect (i.e., we did not expect this factor to generate any significant main or interaction effects). We presented participants in the nonorganic conditions with a dish called “Our Famous Burrito [Fajitas].” Those in the ingredient-level conditions were presented with a dish called “Our Famous Burrito [Fajitas] (w/ all organic ingredients).” Those in the product-level conditions were presented with a dish called “Our Famous Organic Burrito [Fajitas].” Participants indicated the likelihood with which they would order the dish on a 1 (very unlikely) to 10 (very likely) scale.
Results
We analyzed participants’ likelihood of ordering a dish using a 2 (food type) × 3 (organic label) between-subjects ANOVA. As expected, there was neither a main effect of food type (burrito vs. fajitas) on likelihood of ordering (F < 1, n.s.) nor an interaction between food type and organic label structure (F < 1, n.s.). However, the main effect of label structure (nonorganic vs. ingredient-level vs. product-level) was significant (F(2, 195) = 6.15, p < .003). Thus, we collapsed across food type for the remaining analyses.
Consistent with Study 1a, the likelihood of ordering the organic dish was significantly higher when an ingredient- (vs. product-) level organic label was used (Mi ngred = 6.69, SD = 2.24 vs. Mprod = 5.25, SD = 2.66; F(1, 198) = 11.58, p < .001; Figure 1). Moreover, the likelihood of ordering the nonorganic dish was significantly higher than the likelihood of ordering the organic dish when a product-level organic label was used (Mnonorg = 6.36, SD = 2.42 vs. Mprod = 5.20; F(1, 185) = 7.75, p < .01; the complete ANOVA table is reported in the Web Appendix, p.12). Finally, the likelihood of ordering a nonorganic dish versus an organic dish when an ingredient-level organic label was used was not significantly different (F < 1, n.s.).
Study 1b. Likelihood of ordering. Notes: Error bars represent standard errors.
Discussion of Studies 1a and 1b
The likelihood of ordering the focal dish was significantly influenced by the type of organic label used across three different composite vice foods (burgers, burritos, and fajitas). For all three foods, participants were more likely to purchase a dish that was described with an ingredient-level (vs. product-level) organic label. Moreover, Study 1b found that using product-level organic labels for foods diminished their appeal relative to nonorganic foods (consistent with the findings of Van Doorn and Verhoef 2011, 2015). However, this negative effect was eliminated when the organic product had an ingredient-level label.
Study 2: The Relative Impact of Organic Label Structure for Vice and Virtue Foods
We have speculated that perceptions of feature-based categorical typicality and consequent experiences of (dis)fluency may partially explain the influence of organic label structure on consumers’ evaluations of food. This would imply that the structure of organic labels (product- vs. ingredient-level) should have a more meaningful impact on evaluations of vice than virtue foods. Specifically, shifting the organic label from describing the composite vice food (a product-level label) to its ingredients (an ingredient-level label), which are typically associated with the features “raw” and “unaltered,” should reduce perceptions of atypicality and feelings of disfluency and result in significantly more positive evaluations of that vice food. In contrast, composite virtue foods are definitionally healthy (the dominant feature expected of organic foods) and, thus, the evaluations of such foods should be relatively insensitive to the presence and placement of organic labels.
Method
We randomly assigned participants in Study 2 (201 AMT workers) to one of four conditions in a 2 (organic label structure: ingredient-level vs. product-level) × 2 (food: vice vs. virtue) between-subjects design. As in Studies 1a and 1b, we asked participants to imagine they had found a specific dish on a restaurant’s menu. Those in the vice conditions were told they had found nachos, whereas those in the virtue conditions were told they had found a mixed salad. The same pretest described in Study 1a found that nachos are considered a composite vice food, whereas mixed salads are considered a composite virtue food (see Web Appendix, p. 9).
Participants in the product-level conditions saw options labeled “organic nachos” and “organic mixed salad.” Those in the ingredient-level conditions saw options labeled “nachos with all organic ingredients” and “mixed salad with all organic ingredients.” Participants then responded to three dependent measures: (1) how likely they were to purchase the food (1 = “very unlikely,” and 9 = “very likely”), (2) how appetizing the food was (1 = “not at all,” and 9 = “very much”), and (3) how appealing the food was (1 = “not at all,” and 9 = “very much”). Finally, as an attention check, we presented participants with all four versions of the organic labels (2 products × 2 label structures) used in the study and asked to choose the one they had seen earlier in the study.
Results
We dropped 19 participants that indicated seeing the incorrect combination of food and organic label from the following analysis (final N = 182). Including these responses does not substantively change the following results (the omnibus ANOVA results for the reduced and full samples, as well as the composite score means and standard deviations for the full sample, are included in the Web Appendix, p. 13). We expected that the organic vice food (nachos) would be significantly more likely to be ordered, rated more appetizing, and rated more appealing when the organic label was applied at the ingredient-level rather than at the product-level (consistent with what we found in Studies 1a and 1b). In contrast, we anticipated that the effect of label structure would be less pronounced for the virtue food (salad). The three dependent measures (order, appetize, and appeal) were strongly correlated (α = .96) and, thus, averaged into a single composite evaluation score. This composite score was submitted to a 2 (organic label structure: ingredient-level vs. product-level) × 2 (product: vice vs. virtue) between-subjects ANOVA.
There was a significant main effect of label structure (product- vs. ingredient-level) on composite evaluation scores (F(1, 178) = 15.12, p < .001), but no significant effect of food type (salad vs. nachos; F = 1.38, p > .24). More importantly, the expected interaction between label structure and food type emerged (F(1, 178) = 6.23, p < .02; Figure 2). Planned contrasts revealed that participants evaluated the nachos significantly more positively when they were described with an ingredient- (vs. product-) level organic label (Mingred = 7.16, SD = 1.77 vs. Mprod = 5.10, SD = 2.68; F(1, 178) = 20.84, p < .001). Evaluations were also more positive for the mixed salad when it was described with an ingredient- (vs. product-) level organic label, but this effect was far less pronounced and did not approach significance (Mingred = 5.98, SD = 1.93 vs. Mprod = 5.53, SD = 2.13; F < 1, n.s.).
Study 2. Composite evaluation scores. Notes: Error bars represent standard errors.
Discussion
Study 2 found that ingredient- (vs. product-) level organic labels increased evaluations of both vice and virtue foods, but this effect was significantly stronger for vice foods. The overall pattern of results is also consistent with the process we suspect is at least partially driving the effect. Specifically, “healthy” is the dominant feature expected of both organic and virtue foods. Consequently, the evaluations of virtue foods are relatively unaffected by whether the organic label is applied to the composite food (i.e., at the product level) or to its constituent ingredients (i.e., at the ingredient level). In contrast, vice foods are inherently unhealthy and, thus, applying an organic label to the composite food (i.e., at the product level) forces that food—which is neither healthy nor raw/unaltered—into the category of “organic foods,” which is defined by the features “healthy” and “raw/unaltered.” However, applying the organic label to the constituent ingredients of the vice food, which share the feature of being raw/unaltered with organic foods, should reduce the disfluency and consequently significantly offset the negative impact of the organic label. Of course, many other factors, such as the overall visceral appeal of the food (e.g., many people find nachos more enticing than salads), influence such evaluations, and that can also be seen in the results.
Study 3: Moderating the Effect with Food Imagery
Thus far, we have found that composite vice foods with product-level organic labels are less appealing than nonorganic vice foods (Study 1b) and composite vice foods with ingredient-level organic labels (Studies 1–2). We have also found that organic label structure is significantly less impactful for virtue foods (Study 2). Study 3 aims to identify a managerially controllable factor that can moderate the documented effect of organic label structure on food evaluations.
To identify a factor that might moderate the documented effect, we first reasoned that consumers are likely to naturally represent composite foods, be they vice or virtue, as the foods are presented to them (i.e., as unitary foods), which would be consistent with the concreteness principle (Slovic 1972). In other words, when thinking of a burrito, one tends to think of the burrito as a whole rather than mentally deconstructing it into its constituent ingredients. For reasons we have discussed, we speculate that when an organic label is applied to the “whole” vice food (i.e., the composite vice food), the food is placed in the “organic foods” category, of which it is an atypical member and, consequently, disfluent. But this reasoning suggests that mentally deconstructing the food might attenuate the effect found for vice foods, as doing so would focus the consumer more on the ingredients, which are more typical of organic foods. In short, it is possible to exogenously spark the type of thinking that consumers appear to engage in when they are presented with an ingredient-level organic label. Study 3 uses food imagery to help participants mentally deconstruct the composite food, and we expected this to moderate the documented effect of organic label structure on evaluations of vice foods.
Method
We randomly assigned 431 AMT participants to one of four conditions in a 2 (organic label structure: ingredient-level vs. product-level) × 2 (food image: constructed vs. deconstructed) between-subjects design. We asked all participants to imagine they were in a restaurant. Participants in the ingredient-level conditions were presented with a dish called “Burrito (w/ all organic ingredients).” Those in the product-level conditions were presented with a dish called “All Organic Burrito.”
All participants were additionally shown one of two images of a burrito. We showed participants in the deconstructed conditions a picture of an open burrito in which the ingredients were visible and lying on the tortilla. In contrast, we showed participants in the constructed condition a picture of a closed burrito (both images are provided in the Web Appendix, p. 11). Then, as in Study 2, we asked participants to indicate how likely they were to order the food (1 = “very unlikely,” and 9 = “very likely”), how appetizing the food was (1 = “not at all,” and 9 = “very much”), and how appealing the food was (1 = “not at all,” and 9 = “very much”). We found these measures to be highly correlated (α = .94) and, thus, we averaged them into a single composite evaluation score. Next, two attention-check questions asked participants what type of food they had previously evaluated and the name of the restaurant they had ostensibly visited.
Results
We dropped 46 people that gave incorrect answers to one or more of the attention check questions mentioned previously from the following analysis (final N = 385). Although including these responses does change the omnibus ANOVA results, it does not substantively change the overall pattern of results (i.e., the planned contrasts described subsequently are quite similar; see the Web Appendix, p. 14, for complete details).
We submitted the composite evaluation score to a 2 (organic label structure: ingredient-level vs. product-level) × 2 (food image: constructed vs. deconstructed) between-subjects ANOVA that revealed significant main effects for both label structure (F(1, 382) = 11.22, p < .001) and food image type (F(1, 382) = 13.98, p < .001). These main effects were qualified by a marginally significant interaction (F(1, 382) = 3.61, p < .058; Figure 3). Planned contrasts revealed that the food image factor did not significantly affect composite evaluations when participants were presented with an ingredient-level organic label (Mdeconstructed = 6.38, SD = 1.68 vs. Mc onstructed = 6.00, SD = 2.12; F(1,382) = 1.69, p > .20). In contrast, evaluations of the burrito with a product-level organic label were significantly higher when it was accompanied by an image of a deconstructed (vs. constructed) burrito (Mdeconstructed = 6.08, SD = 1.98 vs. Mconstructed = 4.91, SD = 2.21; F(1, 382) = 16.11, p < .001).
Study 3. Composite evaluation scores. Notes: Error bars represent standard errors.
Discussion
Replicating the previous studies, Study 3 participants that were left to their own devices had far more negative evaluations of the vice food when it had a product-level organic label than when it had an ingredient-level organic label. However, when we explicitly showed participants the composite dish decomposed into its constituent components, the negative impact of the product-level organic label was significantly reduced. Using a subtle but realistic manipulation, the results of this study show that mentally deconstructing the dish only moderates evaluations in the product-level label condition, suggesting that consumers naturally represent composite foods as unitary foods, not their constituent ingredients. When an organic label is applied to such “whole” vice foods, those foods are atypical of consumers’ organic food concepts. In contrast, ingredient-level labels focus the consumer on the ingredients of the composite food, which are more typical of (i.e., share more features with) organic foods. Thus, mental deconstruction had no influence on evaluations when an ingredient-level label was applied to the vice food.
General Discussion
The organic food industry is massive and growing at an impressive rate. Consequently, it has received a great deal of attention in the marketing literature. This research has largely drawn the broad conclusion that consumers tend to respond positively to organic foods because of the perceived health and ethical benefits they offer (Fernqvist and Axelson 2013; Loureiro and Lotade 2005; Shafie and Rennie 2012). However, this general conclusion has been qualified by a number of findings. For instance, organic foods can be off-putting to price-sensitive consumers (Shafie and Rennie 2012) and uncompelling to consumers skeptical of their benefits (Loureiro, McCluskey, and Mittelhammer 2002). Of particular interest for the current research, vice foods (e.g., beer, salty snacks, pizza, burritos) have been shown to be reliably less preferred when they are organic than when they are not (Van Doorn and Verhoef 2011, 2015).
Building on these findings, we find that consumers’ responses to organic vice foods are quite nuanced and surprisingly sensitive to seemingly simple changes in the structure of organic labels placed on foods. Product-level organic labels appear before the composite food’s name (e.g., “organic burrito,” “organic burger,” “organic salad”), whereas ingredient-level organic labels appear after the composite food’s name and refer to the food’s ingredients (e.g., “burger with all organic ingredients”). There are two main findings in our research. First, adding organic labels to foods is more damaging to evaluations of vice rather than virtue foods, which is consistent with previous research (Van Doorn and Verhoef 2011, 2015). However, whereas product-level organic labels lead to significantly lower evaluations of vice foods relative to nonorganic vice foods, ingredient-level organic labels do not Organic label structure does not significantly influence evaluations of virtue foods.
We speculate that the impact of product- and ingredient-level organic labels can be explained by understanding consumers’ organic food concepts (i.e., their internal mental representation of the real-world category of organic foods). First, previous research and two studies reported in the Web Appendix (pp. 2–6) found that consumers consider “healthy” the most defining feature of organic foods but also expect organic foods to be “raw/unaltered” (e.g., grown on a farm, fresh, simple). From this, one can predict that foods that are neither healthy nor raw/unaltered will be highly discrepant from the organic food category and, thus, disfluent and poorly received. Composite (i.e., multi-ingredient) vice foods possess neither of these features. Thus, applying a product-level organic label to a vice food (e.g., “Organic Burger”) diminishes evaluations of that food relative to a nonorganic version, as previous research has documented (Van Doorn and Verhoef 2011, 2015). However, merely shifting the organic label to the ingredient level (e.g., “Burger with all organic ingredients”) does not result in significantly diminished evaluations of that same vice food. This, we argue, occurs because ingredients are, in most cases, thought to be “raw/unaltered” components of a composite food and, therefore, are far less discrepant with the organic food category. An alternative account suggesting that consumers are more skeptical of product- (vs. ingredient) level organic labels (i.e., they believe them to be less truthful, factual, accurate, honest, and informative) was examined in a study reported in the Web Appendix (p. 15) but received no empirical support.
We found evidence of the impact of organic label structure on the evaluations of vice foods in four studies. In all four, the evaluations of vice foods were significantly less positive when participants were given a product- (vs. ingredient-) level organic label. In addition, Study 1b found that evaluations of vice foods with ingredient-level organic labels were not significantly lower than those of nonorganic vice foods, Study 2 found that the impact of organic label structure is not significant for virtue foods, and Study 3 found that helping consumers deconstruct composite vice foods into their constituent ingredients through imagery offsets the negative impact of being organic, much like applying an ingredient-level organic label does.
In summary, these results suggest a rosier picture for organic vice foods than that suggested by previous findings (Luchs, Naylor, Irwin, and Raghunathan 2010; Raghunathan, Naylor, and Hoyer 2006; Van Doorn and Verhoef 2011, 2015). It may very well be the case that producers can profitably sell organic composite vice foods by either (1) using ingredient- (vs. product-) level organic labels or, failing that, (2) inducing consumers to mentally deconstruct those foods when a product-level label is applied.
As discussed, the USDA has defined four labeling categories for organic products: (1) “100% organic,” (2) “organic,” (3) “made with organic [fill in ingredient here],” and (4) specific ingredient lists. This means that labels like “100% organic,” “all organic,” and “made with all organic ingredients” all fall within the legal ambit of the definition as long as the product’s true nature conforms to the label’s claims. However, although the USDA strictly regulates the presence of organic labels, the firm has discretion regarding the manner in which such labels are structured (i.e., using either product- or ingredient-level labels). In addition, the Food and Drug Administration (FDA) offers guidelines regarding how food products should be labeled (FDA 2013). Among the long list of guidelines, the FDA states that the “statement of identity” (i.e., the name of the food) (1) must stand out (p. 7) and (2) should be appropriately descriptive and not misleading (p. 7). When using the term “misleading,” the FDA means that the food cannot be labeled such that consumers would understand it to be something different from what it would be if it were given its “common name.” Neither product- nor ingredient-level organic labels would be considered misleading as long as they conform to the USDA’s standards. Interestingly, although the FDA has very specific guidelines for when, how, and where information regarding nutrients, juice percentage, serving size, and ingredients should be placed, no guidance is offered for organic labels. Moreover, the majority of the FDA’s guidelines pertain to the legally necessary salience and truthfulness of the information as opposed to factors that might affect how consumers process and use that information. At a minimum, the current work suggests that policy makers such as those at the FDA and USDA should consider how label structure (organic and other)—and not just the presence, salience, and truthfulness of information on the label—influences consumers’ decisions when applied at the product versus ingredient level.
More broadly, this work speaks to the power of framing (Tversky and Kahneman 1981) on consumers’ decisions. For example, Kroese, Marchiori, and Ridder (2016) found that merely guiding people to healthier choices using environmental cues resulted in more healthy choices than trying to convince them to make healthier choices. Similarly, researchers have shown that firms can encourage healthy choices (in terms of portion control) by varying the size of serving plates, package sizes, lighting and layout, and color and convenience (Wansink 2010). Other research finds that message framing can influence people's perceptions of risk and how that perception affects their attitudes, intentions, and behavior regarding nutrition (Aldridge 2007). Our research points to the potential advantages of thinking about public policy interventions that consider framing in broader terms. The current research found that framing—in terms of label structure—that increases or decreases the congruence of vice foods with the organic food category significantly affects attitudes toward those foods (whether or not organic vice foods are objectively “healthier” than nonorganic vice foods is a complex and unanswered question on which future research might shed light). It is possible that similar opportunities exist for framing within other public policy domains to increase or decrease the congruency of specific products with specific labels/categories and, thus, encourage consumers to make “better” choices.
As with all research, this work has its limitations. For instance, in the interest of conducting rigorously controlled experiments to examine the link between organic label structure and the evaluations of foods, we chose to rely on studies that presented participants less information and less noise than would be common during most real-world decisions regarding food. Despite resulting in strong internal validity, this may call into question the generalizability and ecological validity of our results. However, our confidence in the generalizability of these results is bolstered by the reliability of the results across a variety of different foods in our studies and because we have provided evidence of firms using similar manipulations in packaged goods introduced in the United States and Canada. We would encourage firms and academics alike to test our theories in noisier, real-world contexts. Although many factors influence food choices, we believe the structure of organic labels will affect choices even in such contexts.
Finally, it does not appear that the two identified features most expected of organic foods—that they are healthy and raw/unaltered—are combined additively when assessing the typicality of organic foods. At least in the paradigm examined here, using the label manipulations that were of focal interest, it seems that a food needs to have at least one or the other feature to be considered fairly typical of organic foods and that adding the second feature does not substantially increase its perceived typicality (whereas having neither feature significantly decreases its perceived typicality). This implies that organic foods might represent a particularly unique and challenging type of category called a disjunctive category (Rips and Collins 1993). Members of such categories tend to have some but not all the features that are considered central to consumers’ concept of that category. This revelation may open many questions about organic foods in particular and consumer products in general. We hope that future research addresses these interesting new directions.
Supplemental Material
Supplemental Material, JPPM922873_Web_Appendix - How Product Type and Organic Label Structure Combine to Influence Consumers’ Evaluations of Organic Foods
Supplemental Material, JPPM922873_Web_Appendix for How Product Type and Organic Label Structure Combine to Influence Consumers’ Evaluations of Organic Foods by Jeffrey R. Parker, Iman Paul, Ryan Hamilton, Omar Rodriguez-Vila and Sundar G. Bharadwaj in Journal of Public Policy & Marketing
Footnotes
Associate Editor
Maura Scott
Guest Editor
Beth Vallen
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Note
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
