Lost in processing? Perceived healthfulness, taste and caloric content of whole and processed organic food

1 The "organic" claim explicitly informs consumers ab out the food production method. 2 Yet, based on this claim, people often infer unrela ted food attributes. The current 3 research examined whether the perceived advantage o f organic over conventional 4 food generalizes across different organic food type s. Compared to whole organic 5 foods, processed organic foods are less available, familiar and prototypical of the 6 organic food category. In two studies (combined N = 258) we investigated how both 7 organic foods types were perceived in healthfulness , ta te and caloric content when 8 compared to their conventional alternatives. Partic ipants evaluated images of both 9 whole (e.g., lettuce) and processed organic food ex emplars (e.g., pizza), and reported 10 general evaluations of these food types. The associ ati n of these evaluations with 11 individual difference variables – self-reported kno wledge and consumption of organic 12 food, and environmental concerns – was also examine d. 13 Results showed that organically produced whole food s were perceived as 14 more healthful, tastier and less caloric than those produced conventionally, thus 15 replicating the well-established halo effect of the organic claim in food evaluation. 16 The organic advantage was more pronounced among ind ividuals who report being 17 more knowledgeable about organic food, consume it m ore frequently, and are more 18 environmentally concerned. The advantage of the org anic claim for processed foods 19 was less clear. Overall, processed organic (vs. con ventional) foods were perceived as 20 tastier, more healthful (Study 1) or equally health ful (Study 2), but also as more 21 caloric. We argue that the features of processed fo o may modulate the impact of the 22 organic claim, and outline possible research direct ions to test this assumption. 23 Uncovering the specific conditions in which food cl aims bias consumer’s perceptions 24 M AN US CR IP T AC CE PT ED ACCEPTED MANUSCRIPT WHOLE AND PROCESSED ORGANIC FOOD 2 and behavior may have important implications for ma keting, health and public-policy 1 related fields. 2


Procedure and Measures 9
Individuals were invited (e.g., institutional email, social network websites) to 10 collaborate on a Qualtrics web survey about the perception and evaluation of food 11 images. By clicking on a hyperlink, individuals were directed to a secure webpage in 12 which they were told that we were conducting a consumer psychology study aiming 13 to explore how people evaluate different types of food products. They were also 14 informed about the expected duration of the study (approximately 10 minutes). 15 Participants were told that in their particular case all the food products that 16 would be presented were organic. They were further informed that their task was to 17 evaluate each image of an organic food product by comparing it to its conventional 18 counterpart in three dimensions: healthfulness (1 = Less healthful that conventional to 19 7 = More healthful than conventional); taste (1 = Less tasty that conventional to 7 = 20 More tasty than conventional) and caloric content (1 = Fewer calories that 21 conventional to 7 = More calories than conventional). Participants were asked to 22 answer as quickly as possible, although there was no time limit to complete the task. 23 They were also informed that there were no right or wrong answers, and that all data 24 M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT would be treated anonymously. Participants provided consent to collaborate in the 1 study by checking the "I agree" option. 2 Before starting their task, participants provided demographic information (age, 3 gender, education level, occupation). Next, each participant completed 32 trials (i.e., 4 the full set of stimuli) presented in random order. In each trial, the food image was 5 presented on the top center of the screen. To emphasize that the product was organic, 6 the sentence "This organic product is..." was presented below each image, followed 7 by the three rating scales (in random order in each trial). 8 The general evaluations about both organic food types were assessed using the 9 same set of three evaluative dimensions (presented in random order): "In my opinion, 10 whole organic foods (e.g., fruits, vegetables, etc.) are...", and "In my opinion, 11 processed organic foods (e.g., sweets, ready-to-eat meals, etc.) are...". These overall 12 evaluations of each organic food type were presented in different pages. 13 We also assessed individual differences regarding participants' self-reported 14 knowledge about organic food ("How do you rate your knowledge about organic 15 food?", 1 = Very low; 7 = Very high), and their frequency of organic food 16 consumption ("How often do you eat organic food?", 1 = Rarely; 7 = Frequently). 17 Additionally, we asked participants to complete the New Environmental Paradigm 18 scale (NEP; Dunlap et al., 2000). This scale comprises 15 items about environmental 19 concerns (e.g., "Humans are seriously abusing the environment"; 1 = Strongly 20 disagree, 5 = Strongly agree) and presented good reliability in our study (α = .70). 21 Finally, participants were also asked about their diet (absence of "specific 22 diet", macrobiotic, vegetarian, vegan, gluten free, other) and then they were thanked 23 and debriefed . 24 Results and Discussion 25 M A N U S C R I P T

A C C E P T E D ACCEPTED MANUSCRIPT
First, we present the descriptive results regarding the evaluation of organic 1 versus conventional foods. Second, we examine the impact of food type (whole vs. 2 processed) on ratings of both exemplars and general evaluations in the three 3 evaluative dimensions. These results are summarized in Table 1. Third, we present 4 additional analysis examining associations with individual differences. Finally, we 5 summarize the results of the normative ratings of food exemplars. 6

Evaluations of Organic versus Conventional Food 7
To examine differences in the evaluations of organic food versus conventional 8 food, we compared mean ratings against the scale midpoint (a mean score of 4 in a 9 given dimension indicates that a given organic food is equated to its conventional 10 alternative). 1 11 Exemplars. Mean ratings on each dimension were calculated for each food 12 type by averaging the 16 whole foods exemplars and the 16 processed foods 13 exemplars (see Table 1, Exemplars Evaluation columns). 14

17
1 Because higher scores on the calories dimension represented a negative evaluation of the food items (i.e., more calories than their conventional counterparts), ratings for this dimension were reversed, so that higher scores indicate fewer calories than the conventional counterparts. By doing so, higher scores in all dimension indicate more positive evaluations of each food type.

3
Participants evaluated the exemplars of both whole and processed organic 4 food as more healthful and tastier than their conventional alternative, all p < .001. 5 Whereas whole organic foods were perceived as having fewer calories than 6 conventional alternatives, p < .001, processed organic foods were perceived as having 7 more calories than conventional foods, p = .013. 8 General Evaluations. The general pattern of findings for the general 9 evaluations replicated that of the evaluation of the exemplars. Whole and processed 10 organic foods were perceived as more healthful and tastier than their conventional 11 counterparts, all p < .017. Whole organic foods were evaluated as having fewer 12 calories than whole conventional foods, p < .001, whereas processed organic foods 13 were seen as equally caloric as their conventional option, p = .182 (see Table 1,  14 General Evaluation columns). 15 Overall, the organic nature of both whole and processed foods was perceived 16 as advantageous in healthfulness and taste, for both exemplars and general 17 evaluations. A similar advantage was observed for calories, but only for whole foods. 18

Additional Analyses 3
To further examine if the advantage of organic foods was associated with 4 individual differences, we explored the role of participants' self-reported knowledge 5 about organic food and the frequency of their organic food consumption, as well as 6 their environmental concern (see Schuldt & Hannahan, 2013) in the reported 7 evaluations (for each dimension and food type). 14 Note. (r) = reversed rating (i.e., higher ratings indicate fewer calories). Correlations for whole foods 15 appear below the diagonal, and correlations for processed foods appear above the diagonal.

16
Self-reported knowledge about organic food was positively correlated with 17 frequency of consumption of organic food. For whole foods, these two variables, as 18 well as environmental concerns, were positively associated with taste ratings of both M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT exemplars and general evaluations. In addition, environmental concerns were 1 positively associated with healthfulness ratings. For processed foods, self-reported 2 knowledge and frequency of consumption were positively associated with taste 3 ratings of exemplars, although to a weaker extent. Overall, the results suggest that 4 individual differences are associated with perceived advantages of organic food over 5 their conventional alternatives, especially for whole foods. 6 Regarding the association between evaluative dimensions, for whole foods 7 taste and healthfulness were always positively associated, regardless of being 8 exemplars or general evaluations. For processed foods, healthfulness was positively 9 associated with calories for both exemplars and general evaluations, that is, the fewer 10 the perceived calories, the higher the healthfulness ratings. Moreover, there was a 11 negative association between calories and taste, but only for exemplars, that is, 12 exemplars perceived as having more calories were rated higher in taste. Taste and 13 healthfulness were only positively associated for general evaluations. 14 Interestingly, ratings in the same evaluative dimension were correlated in both 15 exemplars and general evaluations, for both whole and processed foods suggesting 16 some convergence between both measures. 17

Normative Ratings 18
We also present descriptive results by food exemplar according to the three 19 evaluative dimensions (see Appendix A). Based on the confidence intervals, organic 20 exemplars were categorized as "less", "equal" or "more" than conventional ones in 21 The entire set of whole organic food exemplars was categorized as more 1 healthful and tastier than their conventional counterparts. The majority of these 2 exemplars were also perceived as having fewer calories (87.5%) than their 3 conventional alternatives. The only exceptions were two food items perceived as 4 equally caloric when compared to their conventional alternatives (12.5%; i.e., bananas 5 and potatoes). All processed organic foods were perceived as tastier, and either 6 equally (56.3%; e.g., croissants) or more healthful (43.8%; e.g., spaghetti with tomato 7 sauce) than their conventional equivalents. Half of these food items were perceived as 8 more caloric (50.0%; e.g., hamburger), whereas the remaining were rated as equally 9 (37.5%; e.g., ham sandwich) or less caloric (12.5%; e.g., muesli bar), than their 10 conventional counterparts. 11 In short, the results from Study 1 indicated that organic food exemplars -both 12 whole and processed -were judged as healthier and tastier than their conventional 13 counterparts. Whole organic food exemplars were perceived as less caloric, whereas 14 the processed exemplars were perceived as more caloric than the conventional 15 alternatives. General evaluations of healthfulness, taste and calories of both organic 16 food types replicated these findings. Taken together, our results suggest a more 17 positive impact of the organic claim for whole (vs. processed foods) and that the 18 perceived advantage of this type of foods over conventional ones is associated to 19 individuals' self-reported knowledge, frequency of consumption and environmental 20 concerns. 21 However, the stimuli set in the current study was not optimal given that whole 22 foods included exclusively exemplars of fruits and vegetables, whereas processed 23 foods included mostly exemplars based on other ingredients (e.g., meat and grain-24 based food). Therefore, the difference between subsets was not solely the whole 25 versus processed nature of the exemplars, but overlapped with the food categories 1 represented. 2

Study 2 3
The primary goal of the current study was to replicate Study 1 by using stimuli 4 that are more varied and balanced throughout food types. Specifically, the subset of 5 whole foods now also includes exemplars such as meat or fish, and the processed 6 foods subset includes exemplars that are fruit or vegetable-based. Besides expanding 7 the variety and number of food exemplars (60 vs. 32 as in Study 1), the new stimulus 8 set always depicts packaged food products to ensure that the most salient difference 9 between sets is their whole or processed nature. The design included two factors manipulated within-participants: 2 (Food 21 type: Whole foods; Processed foods) x 3 (Dimension: Healthfulness; Taste; Calories). 22

Materials 23
The food images (n = 60) were selected from the webpage of an international 24 grocery retailer that included nutritional information. A panel of three judges 25 FOOD 21 discussed and agreed on the familiarity of the food product depicted in each image. 1 All branding and nutritional information labels were removed using Photoshop, but 2 the original product identification was kept (e.g., "chocolate chip muffins"). Half of 3 the images depicted packaged whole foods (M actual calories = 78.50, SD = 62.37) and 4 included fruits (e.g., apples, grapes, n = 10), vegetables (e.g., lettuce, potatoes, n = 5 10), and fish and meat (e.g., salmon fillets, raw pork steaks, n = 10). The fruit and 6 vegetables subsets matched the products used in Study 1 (four new products were 7 added). The remaining images depicted packaged processed foods (M actual calories = 8 191.23, SD = 102.71) and included sweets (e.g., ice-cream, cake, n = 10) and meals 9 (e.g., frozen pasta, pizza, n = 10). When selecting meal exemplars, we aimed to 10 systematically vary the assortment, by including vegetable-and meat-based options 11 (e.g., "vegetable lasagna" and "cheese and tomato pasta", as well as "beef lasagna" 12 and "cheese and bacon pasta"). Likewise, the assortment of sweets also included fruit-13 based options (e.g., "strawberry sundae" and "lemon sorbet"). A new subset of 14 processed foods was added to match the type of items included in the whole foods 15 categories -i.e., fruits, vegetables and meat/fish (e.g., canned fruit, canned vegetable 16 soup and canned tuna, n = 10). All food items were presented in color against a white 17 background (540 × 540 pixels, see Appendix B for the full description of the stimuli 18 used). 19

Procedure and Measures 20
Participants were invited to the psychology laboratory to collaborate on a 21 survey about perception and evaluation of food (using Qualtrics). Informed consent, 22 instructions and measures were identical to Study 1, with the exception of the number 23 of trials (60 in Study 2). Each session took on average 15 minutes. 24

Results and Discussion
Data were analyzed as in Study 1: evaluation of organic versus conventional 1 foods; impact of food type on ratings of both exemplars and general evaluations in the 2 three evaluative dimensions; additional analysis examining whether the evaluation of 3 organic foods was associated with individual differences; and normative ratings of 4 food exemplars. 5

Evaluations of Organic versus Conventional Food 6
Exemplars. Mean ratings on each dimension were calculated for each food 7 type (i.e., average of 30 whole foods and 30 processed foods exemplars). As in Study 8 1, higher scores correspond to more positive evaluations in the three dimensions (i.e., 9 more healthfulness, tastier and fewer calories) and the advantage of organic food over 10 conventional food was assessed by comparing mean ratings against the scale midpoint 11 (see Table 3, Exemplars Evaluation columns). 12 counterparts, all p < .001. However, for processed food the only advantage of organic 1 over conventional food occurred at the taste level, p < .001. Processed organic 2 exemplars were rated as having more calories than their conventional alternatives, p 3 =.009, and as equally healthful, t < 1. 4 General Evaluations. The pattern found for general evaluations of whole 5 organic foods replicates results from Study 1, i.e., more healthful, tastier and less 6 caloric than their conventional counterparts, all p < .001. Processed organic foods 7 were rated as being as healthful and tasty as conventional food, both p > .141, and as 8 having higher caloric content, p = .005 (see Table 3, General Evaluation columns). 9

Additional Analyses 19
As in Study 1, we examined the associations between individual variables and 20 in the reported evaluations (for each dimension and food type, see Table 4). 21 Note. (r) = reversed rating (i.e., higher ratings indicate fewer calories). Correlations for whole foods 2 appear below the diagonal, and correlations for processed foods appear above the diagonal. As in Study 1, ratings in the same evaluative dimension were correlated in 4 both exemplars and general evaluations, for both whole and processed foods 5 suggesting some convergence between measures. 6

Normative Ratings 7
The pattern regarding whole foods was similar to Study 1, such that all 8 exemplars were categorized as more healthful and as tastier than conventional food 9 (see Appendix B). The majority of whole organic foods was also perceived as having 10 fewer calories than the conventional alternative (86.7%). Processed organic foods 11 were categorized as equally (66.7%; e.g., boxed chicken wrap) or more healthful 12 (30.0%; e.g., strawberry jam) than their conventional alternative. These food items 13 were also categorized as tastier (90.0%; chocolate chip muffins), and the remaining 14 (e.g., instant mashed potatoes) as equally tasty to conventional ones. Processed 15 organic food exemplars were categorized as equally (46.7%; e.g., canned tropical fruit 16 salad) or as higher in calories (46.7%; chilled pepperoni pizza) than conventional 17 alternatives. The exceptions were two exemplars categorized as having fewer calories 18 (i.e., canned mashed peas and lemon sorbet). 19

General Discussion 20
Claims presented on food labels -such as "organic" -influence how 21 consumers perceive and behave towards a given food product (for a review, see 22 considering that the whole food exemplars were objectively low in calories, which 18 could have constrained the impact of the organic claim. Moreover, our results suggest 19 that the perceived advantage of whole (vs. processed) organic food seems to be more 20 pronounced among individuals that report to be more knowledgeable about organic 21 food, consume organic food more frequently and are more environmentally 22 concerned. 23 The advantage of the organic claim for processed foods is less clear. Overall, 24 organic (vs. conventional) processed foods were perceived as tastier, as more (Study 25 M A N U S C R I P T

A C C E P T E D ACCEPTED MANUSCRIPT
WHOLE AND PROCESSED ORGANIC FOOD 27 1) or equally healthful (Study 2), but as more caloric than conventional alternatives. 1 The few studies examining the impact of organic claims according to food type do not 2 report systematic effects across evaluative dimensions. For example, Ellison and 3 colleagues (2015) reported the impact of the organic claim on taste evaluations for a 4 whole food product (but not for a processed food product), and on healthfulness 5 evaluations for a processed food product (but not for a whole food product). However, 6 in that study only a single exemplar of each food type was used (strawberries and 7 cookies), whereas in our studies we included a broader set of exemplars (16 or 30 8 exemplars of each food type). Nonetheless, in our studies, the organic claim was 9 introduced simply by referring to the food products as organic. This generic claim is 10 usually applied to whole foods, but not to processed foods. It is possible that the claim 11 needs to be more specific in the case of processed foods, for instance focusing on the 12 production method of the ingredients they include. For example, in Schuldt and 13 Schwarz's (2010) study, Oreo cookies were not described as fully organic, but rather 14 as made with organic flour and organic sugar. Future studies should explore this 15 possibility. 16 The impact of the organic claim for whole foods seems to be robust. 17 Therefore, we think it is worthwhile to further explore which features of processed for highly processed products. The discrepancies found in the evaluations of 1 processed foods between our two studies seem to support this idea. A main difference 2 between our studies is that in Study 2 all food products were packaged. Packaging 3 may be perceived as counteracting the sustainable nature of organic products, and 4 thus have a detrimental impact on consumer choice, at least for organic fruits and 5 vegetables (van Herpen, Immink, & van den Puttelaar, 2016, Study 1). In our studies, 6 packaging did not seem to affect the evaluation of whole foods (e.g., presenting 7 apples in a plastic bag did not change how apples were perceived in terms of 8 healthfulness, taste and caloric content). Still, for processed food exemplars, 9 packaging may have increased the perception of their level of processing. For 10 example, in Study 1 a pasta dish was presented on a plate, whereas Study 2 presented 11 a box of a frozen pasta meal. Therefore, it is possible that participants perceived the 12 latter as more processed than the former, and were less likely to be influenced by the 13 organic claim. 14 In future studies, instead of using a dichotomous categorization of food type 15 processed (e.g., pasteurization and wrapping to preserve or increase food 2 accessibility, such as milk and fresh meat), to ultra-processed (e.g., ready to eat 3 products with little or no preparation, such as desserts and frozen meals). 4 Alternatively, considering that individuals hold expectations regarding the naturalness 5 of different product types (Smith, Barratt, & Selsøe, 2015), researchers can use 6 stimuli (e.g., images of food) pre-tested regarding their perceived level of 7 transformation (see Foroni et al., 2013). Moreover, future studies could also assess 8 beliefs regarding whole and processed foods (e.g., European Food Information 9 Council, 2016), as well as regarding organic food, and examine if such individual 10 differences modulate the impact of the food production claims. It would also be 11 interesting to replicate our work manipulating food type between-participants, to 12 make the comparison between whole and processed foods less salient. Doing so 13 would discard the potential contribution of task demands to the current findings. 14 The main contribution of our work relates to the systematic examination of the 15 role played by food type on organic food evaluation. Overall, our findings show that 16 the perceived advantages of organic over conventional food are stronger for whole 17 than for processed foods, and are more prominent in individuals that report being 18 more knowledgeable, consume organic food more often, and are environmentally 19 concerned. By providing subjective norms of a diverse set of food exemplars, the 20 current work also offers practical implications for researchers interested in 21 investigating the impact of organic claims on food perception and behavior.