**Food Consumption and Emotions at a Salad Lunch Bu**ff**et in a Multisensory Environment**

#### **Ulla Hoppu 1 , Sari Puputti 1 , Saila Mattila 1 , Marjaana Puurtinen <sup>2</sup> and Mari Sandell 1,3, \***


Received: 24 August 2020; Accepted: 19 September 2020; Published: 23 September 2020

**Abstract:** The food experience is multisensory and multisensory external stimuli may affect food choice and emotions. The objective of this study was to evaluate the effect of a multisensory eating environment on food choice, intake and the emotional states of the subjects in a salad lunch buffet setting. A total of 30 female subjects consumed a salad lunch twice in the multisensory laboratory. The two test conditions (control and multisensory condition with environmental stimuli) were randomized and the visits were scheduled one week apart. Subjects selected and ate a meal from a salad buffet including 14 food items and the intake of each item was weighed. They answered an online questionnaire about the meal and their emotional states (20 different emotion terms) after the lunch. There was no significant difference in the food consumption between the control and multisensory conditions. The subjects were very satisfied with their lunch for both study visits but the pleasantness of the eating environment was rated higher under the multisensory condition. In emotional terms, the subjects selected the term "happy" significantly more frequently under the multisensory condition compared with the control. In conclusion, the multisensory eating environment in this study was not related to food intake but may be associated with positive emotions. The effect of the eating environment on food choice and experience deserves further study with a larger study population in a real lunch restaurant setting.

**Keywords:** lunch buffet; vegetables; food intake; multisensory; emotion terms

#### **1. Introduction**

The workday lunch is an important part of Finnish food culture [1] and lunchtime salad buffets are common at restaurants and worksite canteens in Finland. Lunch is typically eaten around noon in Finland and on workdays the lunch break is half an hour. Healthy lunch choices, especially consumption of vegetables may promote public health [2] and, when replacing less sustainably produced food items in diet, also sustainable food system. Recently, different nudging techniques have been used to increase vegetable intake [3]. Lunch breaks spent in pleasant environments may be associated with positive emotions and further with wellbeing and recovery from stress [4]. Customers value peaceful eating environments but they may have different expectations for interior colors, background music and desired emotional sensations in relation to the restaurant menus [5].

Food perception is multisensory, integrating taste, smell, vision, touch and hearing. Food items have various internal sensory attributes. For example, the perception of vegetable quality combines many sensory characteristics [6,7]. External sensory stimuli, such as visual or sound, in the eating environment may modulate the multisensory experience [8]. The eating context, for example at home, a lunch restaurant or snack bar, may provide various external stimuli affecting food choices and

perception. Traditionally, sensory evaluations have been performed in standard sensory laboratory conditions. Recently, different multisensory, immersive or virtual reality applications have been tested in consumer research [9]. Most previous studies involved the evaluation of single foods or beverages in virtual or multisensory conditions such as cookies [10], coffee [11], beer [12] or non-alcoholic beer [13]. Zandstra et al. reported a consumer study of tomato soup comparing tasting sessions in three different contexts: in a laboratory, an immersive simulated café and a real café [14]. A virtual reality eating environment was used in consumer studies evaluating snack products and emotions [15] as well as chocolate products and emotions [16]. To our knowledge, food consumption at a salad buffet in a multisensory environment has not been studied previously.

The emotions elicited by different food products have also been a research focus recently but the entire eating situation has seldom been evaluated using emotional terms. Various methods have been used to evaluate emotions evoked by food experiences including questionnaires with emotion, mood or wellness terms [17,18]. Recently, new types of methods such as a language-independent graphical tool with emoji have been developed for the assessment of food-elicited emotions [19]. Emotion questionnaires have been used in addition to sensory tests to identify the differences between tested products and even to predict food choice [20,21]. Different eating environments have been associated with different emotions [22] and thus may be related to the consumer's experience of the meal.

The main aim of the present study was to evaluate the effect of a multisensory eating environment on food intake, especially vegetable and fruit intake, in a salad lunch buffet setting. A further aim was to compare subjects' reported emotional states under control and multisensory conditions.

#### **2. Methods**

#### *2.1. Subjects*

The subjects had previously participated in extensive sensory tests [23] and further taste testing in a multisensory laboratory. Invitations to participate in this study part were sent to 62 female subjects. Two sets of data were collected in this study: food consumption and emotions as well as eye-tracking data, which were used to record the lunch sessions in this study and are reported in more detail in another article (submitted manuscript). Celiacs and pregnant or breastfeeding women were excluded and subjects with smell hypersensitivity were not recommended to participate. Due to eye-tracking data collection, normal vision was required (below −1.0 diopter). Wearing contact lenses was allowed but wearing glasses was not permitted during food selection and eating. Wearing reading glasses was allowed when answering the online questionnaire. Food allergies and intolerances of the subjects were enquired about before the study visit and just before the meal. A total of 32 subjects attended the first visit but one did not attend for the second and one subject was excluded due to noncompliance with the study protocol (having lunch elsewhere and taking only a small portion of salad). Thus, 30 subjects attended the required two sessions and provided complete data for analysis. The study protocol was approved by the Ethical Committee of the University of Turku and all subjects provided written informed consent. The lunch was free for the subjects and no other compensation for participation was offered.

#### *2.2. Bu*ff*et Foods*

The lunch buffet included 14 different food items. The foods, their preparation and serving sizes are described in Table 1. Food items were ordinary foods generally included in lunch salad buffets in Finland. Food was selected based on visual appearance with mainly color pairs (red, green, orange, black, white, beige) so that they formed a colorful buffet. Foods also had different dominant taste qualities (salty, sweet, sour, bitter). Fresh vegetables and fruits were the main options. Two different lactose-free cheeses, chickpeas and peanuts were provided as protein sources. Pasta with two different sauces (pesto or aioli mayonnaise) was served to supply energy (carbohydrate and fat) for the lunch.



Food items were delivered weekly by the same local supermarket and the quality of the vegetables and fruit was carefully monitored daily. The food was prepared fresh daily in the kitchen of the sensory laboratory just before the session for each participant. After finishing the preparation, the serving trolley was kept in the cold storage room at +8 ◦C. The weights of the served and consumed amount of the foods were measured with a scale (Mettler Toledo PB3002-S, Mettler Toledo International Inc., Columbus, OH, USA), to 1 g accuracy. The foods were served in square-shaped 15 × 15 cm glass bowls. The bowls were placed on the serving trolley on three different levels (Figure 1). The order of the serving bowls was randomized for every subject.

**Figure 1.** Serving trolley with the buffet foods.

The serving sizes were selected based on similar volume appearance in the bowls and so that the subjects felt that they could take enough. Serving tools were ordinary tablespoons, except using a salad server for lettuce. The plates were white porcelain with a diameter of 22 cm. In addition to the salad buffet, rye and oat bread as well as margarine were served. Olive oil with a lemon flavor and French dressing were also offered. Water was served as a drink with the meal and coffee or tea with biscuits were offered after the meal.

#### *2.3. Multisensory Laboratory Conditions*

Multisensory conditions with different landscapes, sounds and odors were pilot-tested beforehand. Of the pilot-tested options, the forest landscape with birdsong and orange scent were selected for the multisensory condition and the other condition was a plain control. These two different conditions (control vs. multisensory) were randomized between the first and second study visit for each subject. Thus, all participants attended in both conditions but in a randomized order.

The multisensory laboratory equipment included an odor diffuser (Pump unit BB-200, @aroma GmbH, Berlin, Germany) and controlled illumination (five bulbs on the wall and three bulbs in a floor lamp, Hue, Philips, Amsterdam, Netherlands). The audio-visual multimedia system included an 80-inch Apple-tv (Apple Inc., Cupertino, CA, USA) and an audio system with two speakers (Genelec Oy, Iisalmi, Finland). In the control condition, the neutral room lighting system was used. In addition, there was no sound, no scent and no visual landscape on the screen. In the multisensory condition, there was a landscape of a pine forest and lake during summertime and bright lighting, matching the color tones of the landscape on the screen (Figure 2). The soundscape was birdsong in a Finnish summertime forest with various species of bird (recorded in Kortesjärvi in June). Orange scent (Orange Oil Sweet Brazil Pera, @aroma GmbH, Berlin, Germany) was diffused to the room with an odor diffuser for 30 s before a subject entered the room and then for 5 s in 3 min intervals until the subject had finished eating.

**Figure 2.** Lunch under a multisensory condition.

#### *2.4. Questionnaire*

Subjects also answered an online questionnaire (Webropol Oy, Helsinki, Finland) on an iPad (Apple Inc., Cupertino, CA, USA). Before the lunch, the questions asked about how hungry they felt (four options from not at all to very hungry) and the time (hours and minutes) since their previous meal. After the meal, the questions included how full they felt (four options, not at all to very full) and if they were satisfied with the salad meal (four options, not at all satisfied to very satisfied). The liking of the test environment in the multisensory room was also evaluated (9-point scale). The emotion terms were selected and modified from emotion questionnaires [24,25]. The term selection was pretested with Finnish consumers in a previous study setting focusing on ambient odors in the multisensory room. Altogether, 19 emotion terms, both positive and negative, as well as an open question option (something else) were presented and the subjects could choose as many options as they liked (check all that apply) based on how they felt at that moment. After the second study session, a few background questions (education, weight, height, how often they have salad for lunch) were asked.

#### *2.5. Procedure*

The test sessions were organized at the multisensory laboratory of the Functional Foods Forum (University of Turku, Finland) at usual lunch times in Finland. Subjects were asked to attend two study sessions at the same time of day (either at 10:45 a.m. or 12:30 p.m.) at least one week apart. The sessions lasted approximately 30–45 min. Session conditions (control vs. multisensory) in the multisensory room occurred in randomized order. Subjects were also instructed to have the same kind of breakfast on both study days and they were asked to avoid the use of scented cosmetic products before visits.

Subjects were first asked to view the trolley for 20 s while the researcher stood next to her. Next, the researcher left the room and closed the door and the subject collected a meal from the buffet. The subjects were instructed to take as much as they wanted and have all the foods at once. The researcher then removed the trolley from the room. Subjects were seated alone in the multisensory room, ate their meal at their own pace and knocked on the door when they had finished eating. After the meal they were served coffee or tea with biscuits. Subjects answered the online questionnaire while having coffee or tea.

The test session was recorded with a head-mounted eye-tracker (Tobii Pro Glasses 2, Tobii AB, Danderyd, Sweden). The subject wore the wireless eye-tracking glasses and gaze data were sent to a laptop in another room in live video format. This allowed the researcher to monitor the session behind a closed door and no other video cameras were needed. The subject knew that she was being monitored and that she herself was not visible on the video since it only recorded her first-person view of the laboratory. The eye-tracker was removed when the subject began to answer the online questionnaire. For this study, the eye-tracking recording was used for monitoring the session and calculating the time spent on eating the self-selected salad. The eating time was measured from the time the participant sat down to eat until she knocked on the door and let the researcher know she had finished. The detailed description of the eye-tracking methodology and analyses are reported in another article; this paper focuses on the food intake and emotional measures.

#### *2.6. Statistics*

The basic results of the intake of foods are presented as means (SD). The same subject attended under both conditions and the intakes in control vs. multisensory conditions were compared using the Wilcoxon signed-rank non-parametric test for repeated measurements. For paired nominal data, McNemar's test was applied. The statistical software used was IBM SPSS Statistics 26 (IBM Corporation, Armonk, NY, USA).

#### **3. Results**

The mean age of the participants (*n* = 30) was 53 years (SD 14 years) and their mean BMI was 26.8 kg/m<sup>2</sup> (SD 6.9). The educational background was high; 50% had a university education and 30% had a university education with an applied sciences degree. Half of the subjects (50%) reported having salad for lunch one to three times per week and 27% one to three times per month. There was no difference in the time since the previous meal for the two study visits: control of 3.9 (SD 2.1) vs. multisensory of 4.0 (SD 2.8) hours. The state of hunger did not differ significantly either as 77% in the control and 73% in the multisensory condition felt very or fairly hungry before the lunch.

The number of food items taken altogether from the salad bowls varied; the range was 7 to 14, mean 11 (SD 1.5) but we found no significant differences between the control and multisensory conditions. The mean (SD) weights of foods and the total portion weight in the control and multisensory conditions are presented in Table 2. The sum variables of the food groups (vegetables, fruit, cheese and pasta) were calculated and analyzed but the intakes of these groups did not differ between the two conditions. Because the total weight of the portions differed between subjects, proportions (%) of the foods in the total portion weight were calculated. However, no significant differences in these variables were observed between study conditions.


**Table 2.** Foods consumed (mean, SD grams) at different conditions (control vs. multisensory).

The multisensory condition included a forest landscape on screen, birdsong and orange scent. *p*-values all non-significant (Wilcoxon signed-rank test).

The eating time did not differ significantly between the two conditions (control 12.8 min vs. multisensory 13.0 min). Considering fullness after the meal, the proportions of responses were exactly the same for both conditions: 47% reported feeling very full and 50% fairly full. Contentment with the salad was also good for both conditions. In the multisensory condition, 83% were very satisfied with the salad compared with 77% in the control condition (non-significant difference). Liking ratings of the testing environment differed (Figure 3). Overall, the multisensory condition was significantly more pleasant than the control (*p* < 0.001).

**Figure 3.** Liking of the test environment.

The selected emotion terms under the control and multisensory conditions are presented in Figure 4. Most of the selected terms were positive for both conditions. Over two-thirds of the participants felt healthy in both test environments. No one stated that they felt stressed, cold or tired in either condition. Respondents chose the term "happy" more often in the multisensory condition (*n* = 13) compared with the control (*n* = 5); *p* = 0.02, McNemar's test. The subjects also tended to select the terms "relaxed" (*p* = 0.09) and "strong" (*p* = 0.07) more often for the multisensory condition.

**Figure 4.** Selection of the emotion terms (n) in different conditions.

#### **4. Discussion**

In the present study, no general effect of the multisensory environment on food choice or intake at a salad lunch buffet was observed. Therefore, changing individual food preferences and consumption patterns simply with external multisensory stimuli appears to be challenging. The selected foods, portion sizes and eating times were surprisingly similar for the same person under both conditions. This finding may reflect the overall stability of individual eating habits or a more situated tendency to repeat their first-time choices in the second session with exactly the same offerings. However, the multisensory condition was evaluated as very pleasant by the participants and positive emotional effects were reported based on the selection of emotion terms. In general, the feedback from participants regarding the whole experiment was positive; they valued the free, fresh and appealing buffet service and most were very satisfied with their meal.

Comparison with previous studies is challenging as we are not aware of studies using the same type of real-life but controlled lunch buffet settings in multisensory conditions. Previous studies used different study protocols, populations, buffet food selection and sensory primers. The buffet setting studies evaluated, for example, food choices of normal weight and overweight subjects [26]. Buffet meal intakes by different bitter taste sensitivity groups [27] or taste receptor genotype groups [28] were compared. In a multisensory study setting in Italy [29], consumers evaluated tomatoes and wild rocket in an immersive environment using countryside landscapes and sounds as well as natural herbs as olfactory cues. The liking scores were reported to be higher in the immersive environment compared with the traditional sensory laboratory setting [29]. In a self-service buffet setting, a priming experiment consisted of creating a leafy environment with green plants and an odor of herbs. The priming condition reduced the total energy intake [30]. Most previous studies have evaluated single foods or beverages in immersive conditions whereas this study provides new information about food consumption at a salad buffet in a multisensory environment.

Priming with food odor has been hypothesized to affect food selection but the results are controversial. Exposure to a fruity odor (pear) was found to increase the likelihood of selecting a fruity dessert [31]. Mors et al. [32] reported that priming with a bread or cucumber odor did not affect lunch choices but odor condition was associated with a self-reported positive mood. In the present study, exposure to the orange odor did not increase the selection or consumption of orange in the salad buffet; the trend seemed to be slightly the opposite. The auditory contribution to food perception was reviewed by Spence et al. [33]. Most previous studies have focused on the effect of music genre on the perception of a single flavor or food [34] but what kind of background sounds are most appropriate for lunch room conditions is not known. Different nature sounds, including birdsong, have been related to stress recovery and restorative benefits although restorative perceptions may vary between different bird species [35].

The buffet food selection in the present study was colorful and consumers were previously reported to value visually attractive and colorful salads [36]. External visual stimuli including colors of the food package, plates or cups may be associated with food perception [37]. Individuals differ in their associations of the color of liquid samples with taste qualities, pleasantness or healthiness [38]. In the present study, the tablecloth on the serving trolley was white while the color hue of the lighting and the color of the landscape were greenish in the multisensory condition. The color of the lighting may also have affected the color perception of food items offered in the salad buffet. Schifferstein et al. [39] reported that colored backgrounds affected the perceived attractiveness of vegetables but optimal background colors differed substantially for various vegetables. According to Hasenbeck et al. [40], yellow lighting increased the willingness to eat bell peppers. Because our buffet included food items with various colors, evaluating which colors of lighting would most effectively increase the attractiveness of vegetables and fruit was difficult. Complex landscape scenes present various colors and the effects of various pictures or scenes may be difficult to interpret. Investigating the effects of single sensory stimuli provides important information but in the multisensory context several aspects are combined. Real-life studies in restaurant settings also combine many sensory stimuli both in the food and in the environment. Therefore, the multisensory approach on consumer behavior and experience are challenging research topics.

Comparing study results focusing on emotions is difficult as the emotion terms vary and the results may be specific to the study population, study setting and the tested products. Few studies have related emotions specifically to the eating or meal situation [41,42]. In this study, the subjects selected the terms related to the whole meal situation and we do not know if their emotions were more related to the food eaten or the multisensory eating environment. We did not ask for opinions separately about different components (odor, lighting and sound) in the multisensory room. The reactions may be individual and some people may report unpleasant emotions associated with musical or pictorial stimuli [43]. Some consumers report adverse effects such as headaches related to fragrances [44]; thus, room odors should be used with caution. More research is needed on what kinds of sensory stimuli, as well as their combinations in different eating environments and with different consumer groups, can support pleasant eating experiences. Emotion questionnaires rely on self-reported subjective ratings of emotions and other measures would also be useful in food research settings. The review by Kaneko et al. [18] recommended combining various instruments, including physiological, behavioral and cognitive measures, for evaluating emotions evoked by food experiences.

The strength of this study was that real intake with real foods was studied and not just food pictures or fake food models [45,46]. In comparison with self-estimated portions of food intake in many nutrition studies, here the food intake was accurately weighed [47]. The same subjects attended two visits in a randomized order. Recording the sessions with a head-mounted eye-tracker overcame the need to set up external video cameras in the laboratory and we think this made the session monitoring feel less intrusive for the subjects. The possible limitation may be the short exposure time to the

multisensory environment while taking the food. However, we wanted to create a situation resembling the normal food selection phase in a lunch restaurant setting and not to have the subjects wait before selecting the food. The subjects were seated alone whereas in real restaurants there may be many other external stimuli present and other customers. Only women were included in this study and men may have different preferences for lunch buffet foods as well as greater energy needs, requiring larger portions. The number of subjects was rather small but comparable with other buffet setting studies [26]. Only one subject was attending at a time and the preparation, serving and weighing of various fresh food items was rather laborious and time consuming. In future, real lunch restaurant buffets with a larger study population including both sexes could be studied.

In conclusion, the multisensory room conditions in this study did not change the food intake of the subjects. Fresh, colorful and a varied vegetable selection at lunch is appealing and could promote the consumption of vegetables and sustainable eating habits. In addition to fresh vegetables and fruit, salad buffets usually include other components and thus the overall nutrient composition and healthiness of the lunch depends on individual consumer choices. A pleasant and relaxing ambience may elicit positive feelings and thus enhance meal satisfaction and wellbeing [48]. The promotion of positive eating situations among various consumer groups deserves further study.

**Author Contributions:** Conceptualization, U.H., S.P., M.P. and M.S.; methodology, U.H., S.P. and M.P.; investigation, U.H., S.P. and S.M.; writing—original draft preparation U.H.; writing—review & editing, S.P., S.M., M.P. and M.S.; supervision M.S.; project administration M.S.; funding acquisition M.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Academy of Finland, grant number 309408 and 318894.

**Acknowledgments:** Nora Logren is acknowledged for working as research assistant, Ilkka Heiskanen for allowing us to use the birdsong soundtrack from his personal soundscape recordings and the carpenter services of University of Turku for making the serving trolley. All the study subjects are also thanked for their participation. This study is part of the University of Turku strategic research profile on Biodiversity and sustainability.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Does the 'Mountain Pasture Product' Claim Affect Local Cheese Acceptability?**

**Isabella Endrizzi 1, \* , Danny Cliceri 2 , Leonardo Menghi 1,2,3 , Eugenio Aprea 1,2 and Flavia Gasperi 1,2**


**Abstract:** This paper aims to explore the impact of "mountain pasture product" information on the acceptability of local protected designation of origin (PDO) cheese produced from the raw milk of cows grazing in mountain pastures (P) or reared in valley floor stalls (S). A total of 156 consumers (55% males, mean age 41 years) were asked to evaluate their overall liking on a 9-point hedonic scale of four samples: Cheeses P and S were presented twice with different information about the origin of the milk (cows grazing on mountain pasture or reared in a valley floor stall). Demographics, consumer habits, and opinions on mountain pasture practice (MPP), attitudes towards sustainability, and food-related behaviours (i.e., diet, food waste production, organic food, and zero food miles products purchase) were recorded and used to segment consumers. The cheeses were all considered more than acceptable, even though they were found to be significantly different in colour and texture by instrumental analyses. In the whole consumer panel, the cheese P was preferred, while in consumer segments less attentive to product characteristics, this effect was not significant. External information had a strong effect: Overall liking was significantly higher in cheeses presented as "mountain pasture product", both in the whole panel and in consumer segments with different attitudes (except for those with a low opinion of MPP).

**Keywords:** mountain cheese; acceptability; conjoint analysis; external information; consumer segmentation; food sustainability

#### **1. Introduction**

Mountain dairies—which, in the alpine territories, are placed in contexts with a high naturalistic value, in most cases—play key roles in the promotion of local tourism, the preservation of biodiversity and the environment, and the maintenance of cultural and historical traditions [1]. They also find themselves in a position of increasingly seeking a compromise between production and conservation needs, as well as trying to respond convincingly to consumer requests regarding food safety and compliance with ethical farming principles. While animal husbandry, in general, has been subjected to criticism due to the excessive intensification and poor efficiency of production processes, mountain activity has gained growing interest from both tourists and consumers who associate this activity with greater sustainability, being able to combine production, environmental, and social needs on a small scale [2]. Nevertheless, mountain farming is less competitive and has higher costs than intensive production, with the consequent abandonment of such activities in the most remote areas in recent decades [3,4]. The EU has recognized, for some years, the need to prevent the abandonment of these mountain areas by focusing on the promotion and development of mountain food production as a way to promote sustainable

**Citation:** Endrizzi, I.; Cliceri, D.; Menghi, L.; Aprea, E.; Gasperi, F. Does the 'Mountain Pasture Product' Claim Affect Local Cheese Acceptability? *Foods* **2021**, *10*, 682. https://doi.org/10.3390/ foods10030682

Academic Editor: Antti Knaapila

Received: 5 February 2021 Accepted: 19 March 2021 Published: 23 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

development and to reduce the disadvantages of these areas [5]. With this objective, the EU has recently introduced a new labelling system for mountain products [6], which is an important step toward taking into account that there is a strong correlation between the perception of EU quality signs and the attitude towards food origin [7].

#### *1.1. Mountain Products and Consumer Perception*

The quality of dairy mountain products starts from the animals, which are moved from indoor feeding with conserved forage to fresh herbage (i.e., feeding on pasture), according to the traditional transhumance system [8]. Several studies have focused on the effects of summer grazing on dairy product characteristics, demonstrating that mountain products are different from indoor feeding products, in terms of sensory attributes [9–13], volatile organic compounds (VOCs)' profile [8,14–16], and fatty acid composition [11,17,18]. The quality of dairy mountain products is also a key aspect for consumers [19], for whom the production of high-quality dairy products is the most important ecosystem service among those provided by summer farms [20].

#### *1.2. Influence of Information and Attitudes*

When the quality of a food product is evaluated, we have to take into account that consumer perception is based both on intrinsic characteristics (mainly related to product physical properties) and extrinsic properties (i.e., any information provided to the consumer about the product) [21]. Among the external factors, those linked to product credibility, such as labels, brands, origin, organic and production method-related, health, and ethics, are those of high interest for product enhancement in the market [22].

For food products of animal origin, some studies have investigated the effect of information about the production method, demonstrating that information about cow grazing versus information about indoor system increased consumer preference for meat [23,24] and generated the highest willingness to pay or positive effect, in terms of liking milk [25–28]. Romanzin et al. [28] have reported similar results for cheese, even if the literature on this food matrix is scarce. A recent review on consumer perception, preferences, and behaviours regarding pasture products did not report any study on cheese [29].

External factors generate expectations about food products and influence the choices of consumers, having a role in both their perception and liking [30,31]. In spite of this, the magnitude of these effects on food choice depends on how well the consumer is informed, aware, and prepared towards the concepts associated with the external information transmitted [32–34]. Studies have demonstrated that consumers associate sustainability-related attributes, such as environmental friendliness, animal welfare, local, and small-scale production, with mountain pasture products [2,35]. Nevertheless, it is important to measure consumer engagement with respect to the mentioned aspects, in terms of knowledge, awareness, and attitude. Some examples in the literature have proposed self-assessed measures or questionnaires for estimating the individual environmental sustainability [36], the low-carbon consumption scale [37], and knowledge of food sustainability [38]. Poortinga and Darnton [39] developed a screening tool to segment consumers, according to their attitude towards environmental, economic, and social aspects of sustainability. The developed tool has been proven to be valid in differentiating the Welsh population, even though it lacks a domain that measures the aspect linked to food sustainability.

#### *1.3. Objective of the Study*

In this study, we explored the impact of "mountain pasture product" information on the acceptability of a local cheese, with the final aim to enhance the value of dairy products obtained from milk produced in mountain pastures, thus promoting the multiple positive externalities connected to them. In order to identify the profiles of consumers differently involved with the mountain pasture world and differently inclined towards environmental and food sustainability, we developed and proposed new questionnaires as screening tools for this precise purpose. Then, we verified whether consumer segments with different attitudes toward these quality characteristics experienced a different impact of external information about mountain pasture products.

#### **2. Materials and Methods**

#### *2.1. Cheese Samples*

A protected designation of origin (PDO) cheese (Puzzone di Moena) was the product chosen for the consumer test. Puzzone di Moena is a semi-hard cheese with characteristic washed rind, produced from raw bovine milk of animals reared in the mountain area of Trentino Alto Adige (Italy) from a minimum height of 1000 m (valley floor stalls) up to 2000 m of altitude (mountain pasture). Puzzone di Moena is generally sold at 100 days of maturation, but the ripening period varies from a minimum of 3 up to a maximum of 16 months. Puzzone di Moena "malga cheese" is a Slow Food presidium [40], sold with the label "Sapori di malga" (which means "Mountain hut flavours"). It is exclusively produced from alpine pasture milk during the summer pasturing period, from June to September.

The samples given to the consumers were obtained from two different wheels of Puzzone di Moena PDO produced in Predazzo dairy (Trento, Italy): One made from milk collected in valley floor stalls (S) and aged 100 days, and the other one from alpine pasture milk (P) and aged 200 days. The two cheeses were considered representative of the two types of Puzzone present in the local market: The mountain pasture product is sold with greater maturation, in order to enhance its distinctive characteristics (min 120 days) [41], while the valley bottom cheese is mostly sold with a minor seasoning (min 60 days) [42].

#### Sample Preparation

Cheese wheels were stored at 15 ◦C until the moment of portioning, which took place the day before the test. To obtain homogeneous samples, the whole wheel was first cut in half, then into two quarters. From each quarter, eight 1.5-cm thick slices were cut. From each slice, after removing the rind, a parallelepiped was obtained, which was further divided into 16 smaller parts (3 × 1.5 cm × 1.5 cm). The cheese pieces were then stored in vacuum-sealed containers at 10 ◦C until the next day. On the day of the test, each piece of cheese was placed in a transparent bio-plastic cup, covered with a lid, coded with a three-digit number, and stored at 15 ◦C until tasting.

#### *2.2. Physical and Rheological Properties*

For each cheese (P and S), 32 cheese parallelepipeds randomly selected from those cut were collected (one piece for each cheese slice) and submitted first to instrumental measurements of colour and then of texture characteristics. Colour measurements were recorded at room temperature on freshly cut cheese slices using a tri-stimulus CR-400 colorimeter supported by the CM-S100wSpectraMagicTM colour data software (Konica Minolta Sensing, Inc., Tokyo, Japan) and calibrated with a white standard plate. The L\*-, a\*-, and b\*-parameters of the CIEL\*a\*b\* colour space model (see [43]) describe visual lightness (as values increase from 0 to 100), redness to greenness (positive to negative values, respectively), and yellowness to blueness (positive to negative values, respectively) of the samples.

Texture properties were then measured using a TA-XT texture analyser, equipped with an acoustic envelope detector device (Stable MicroSystem Ltd., Godalming, UK). A 4-mm probe was used to compress the samples. Nine mechanical parameters were calculated from the recorded curves, following the method described by Costa et al. [44].

#### *2.3. Consumer Study*

The consumer test was conducted in the Trento Expo exhibition spaces (Trento, Italy), on the 16 and 17 March 2019, in "La Casolara 2019", the traditional Slow Food presidium fair dedicated to the best cheese and dairy production from all over the country, attracting not only local visitors.

The responses of 156 consumers were collected in a mobile sensory laboratory compliant with EN ISO standards 8589 [45], equipped with four mobile individual booths using the FIZZ 2.46A software (Biosystemes, Couternon, France).

The test consisted of an experiment that was evaluated in 'informed' conditions, combining conjoint analysis with the tasting of the two Puzzone di Moena PDO cheeses (described in Section 2.1). Each consumer received four cheese samples in total, according to a complete factorial design with two milk productions and two information levels. The two cheeses (P and S) were presented twice, each time with different external information: 'Produced from milk of cows reared on mountain pasture' (Claim\_P) or 'Produced from milk of cows reared in valley floor stalls' (Claim\_S). These two claims were submitted to consumers on the computer screen (Figure 1), just before tasting the sample. Consumers rated their overall liking of the four cheeses on a nine-point scale, from 1 = "Dislike extremely" to 9 = "Like extremely". The four samples were presented in a random and balanced order for each participant, who evaluated them under white light.

You are about to taste a cheese made from the milk of cows grazed in a mountain hut

You are about to taste a cheese made from the milk of cows kept in a valley floor stall

(**a**) (**b**)

**Figure 1.** Examples of the screen used in the conjoint study: (**a**) The information about mountain pasture cheese; and (**b**) the information about cheese made with milk from cows reared in valley floor stalls.

All subjects were not paid and voluntarily joined the test. Prior to participation, the experimental procedure was explained to all participants and written informed consent was obtained from each, according to the European Data Protection Regulation (UE 679/2016). The consumers were asked to pay attention and to carefully read all the instructions provided during the test. We provided participants with noise-proof earmuffs, in order to help them concentrate in the noisy fair environment, and asked them to follow a rinse procedure with water and unsalted crackers to avoid possible carry-over effects between the products tested.

#### *2.4. Questionnaires*

After tasting, by means of a series of questionnaires, participants provided information about a list of different topics, from socio-demographic data to self-reported behaviours and habits related to food, sustainability, and mountain pasture perception (Table 1). They were asked about their food diet, in order to identify the omnivore, flexitarian (people reducing or limiting their meat consumption), and vegetarian/vegan distribution in the panel, using 9 items adapted from De Backer and Hudders [46], which have already been used in Italian [47,48]. Consumers reported their percentage of weekly food waste, as well as organic and zero food mile products weekly purchased (<5%, 5–10%, 11–20%, 21–30%, 31–40%, >40%). To assess interest towards natural products, the Natural Product Interest

(NPI) sub-scale of the Health and Taste Attitude Scales (HTAS), developed by Roininen et al. [49] and validated by Saba et al. [50], was used. Participants rated their degree of agreement with a series of positive and negative statements on a 9-point scale (1 = totally disagree; 9 = totally agree), rather than the original 7-point scale, in order to be consistent with the other questionnaires submitted to the participants. In the Supplementary Materials, the original and the Italian version of the NPI sub-scale is reported (Table S1).

**Table 1.** Demographics, food behaviour questions, attitude questionnaires, and their relative acronyms, number of items, rating scale, response options, and references.


\* HTAS health and taste attitude scale.

#### 2.4.1. Attitude towards Sustainability and Food Sustainability

Furthermore, participants provided information regarding their attitude towards sustainability (ATS), rating their degree of importance, agreement, and concern on the response option expected by each statement of the 15-item Welsh screening tool for sustainability [39]. The original scale was back-translated in Italian by a native bilingual, following the procedure suggested by Brislin [51]. In this case, a 9-point scale (1 = not important at all/totally disagree/not concerned at all; 9 = extremely important/totally agree/extremely concerned, depending on the statement), rather than the original 6-point scale with the escape answer "don't know", was used. In the original Welsh questionnaire, there were no statements investigating sustainability, in terms of food consumption. Given the importance for a study like this to collect this information, a list of 18 positive and negative statements investigating attitudes towards local food, green restaurants, and domestic food waste were developed, in order to cover the food consumption sustainability domain (FCS; Table 2). In the Supplementary Materials, the original Welsh questionnaire, its translation in Italian (ATS), and the Italian version of FCS scale are reported (Tables S2 and S3).

**Table 2.** Mean (M) and Standard Deviation (SD) values for each statement of the Food Consumption Sustainability scale.


<sup>R</sup> Negative statements recoded for the final score calculation.

#### 2.4.2. Mountain Pasture Practice Values

Six further statements on the values and habits of mountain pasture practice (MPP) were developed, in order to explore the knowledge level and perception of the mountain pasture world and its products of consumers (Table 3). These statements were previously developed by a focus group of researchers involved in different fields related to mountain pasture and food (i.e., sensory, nutrition, chemistry, food technologies, animal husbandry, agricultural economics, and statistics). In the Supplementary Materials, the Italian version of the MPP scale is reported (Table S4).


**Table 3.** Mean (M) and Standard Deviation (SD) values for each statement of the Mountain Pasture Practice scale.

#### *2.5. Data Analysis*

Data analysis was performed using the STATISTICA v. 13.1 software (Dell Inc., Tulsa, OK, USA, 2016).

In order to confirm the differences between the two cheeses (P and S), the product effect (fixed factor) on colour and texture instrumental parameters was estimated using one-way analysis of variance (ANOVA). In all statistical tests, we consider a significant difference as *p* < 0.05 after Bonferroni correction.

#### 2.5.1. Analysis of Questionnaire Data

As a preliminary pre-treatment, the scores for the negative statements of NPI questionnaire [49], of the ATS scale [39], and of the developed scales on FCS and statements on MPP were reversed. For the NPI scale, the statements suggested by the authors were considered negative while, for the ATS scale, we considered four statements (8–11) of the original scale negative, as they were found to be opposite to the others in an explorative principal component analysis (PCA) map (data not shown). Following the same procedure, of the 18 statements developed for FCS, eight were considered negative (Table 2); while the MPP statements were all considered positive, being true statements extracted from the literature.

Subsequently, the internal validity for each scale was tested using the standardized Cronbach's alpha [58]. Cronbach's alpha values above 0.60 are considered acceptable and values above 0.70 are considered good to optimal [59]. Then, for each of the four scales, the sum scores were calculated for each participant by adding the score of each item, according to the procedure described by Roininen et al. [49]. Based on these scores, the participants were classified in three groups (low, moderate, and high interest/attitude), according to the 33rd and 66th percentiles.

A two-way analysis of variance (ANOVA) was carried out to test how gender and age (three age classes were considered: Age\_1, 18–30; Age\_2, 31–50; and Age\_3, 51–75) affected the sum scores of the attitude scales. One-way ANOVA was instead used to estimate the effect of food diet (omnivorous, flexitarian, or vegetarian/vegan), percentage of weekly food waste, organic, and zero food mile products weekly purchased (<5%, 5–10%, 11–20%, 21–30%, 31–40%, >40%) on attitude scale sum scores. For significant effects after Bonferroni correction (corrected *p* < 0.05), the Tukey–Kramer post-hoc honestly significant difference (HSD) test for unequal sample size was applied, whenever appropriate.

#### 2.5.2. Analysis of Conjoint Data

The liking data were analysed using a three-way ANOVA mixed model, considering both product and external information as main fixed factors and consumer as the random main factor, together with their second-order interactions. For significant effects after Bonferroni correction (corrected *p* < 0.05), the post-hoc HSD Tukey's test for multiple comparison was applied, whenever appropriate.

In order to identify which groups of people were more sensitive to intrinsic or extrinsic factors, the same ANOVA model was recalculated in sub-groups of consumers, identified by gender, age (Age\_1, 18–30; Age\_2, 31–50; Age\_3, 51–75 years of age), place of residence altitude (Alt\_1, >600; Alt\_2, 600–300; Alt\_3, <300 m a.s.l.; [60]), residence zone (Urb\_1, >150 inhab/km<sup>2</sup> ; Urb\_2, <150 inhab/km<sup>2</sup> ; [61]), interest towards natural products (NPI\_1, low; NPI\_2, moderate; NPI\_3, high), attitude towards sustainability (ATS\_1, low; ATS\_2, moderate; ATS\_3, high), attitude towards food consumption sustainability (FCS\_1, low; FCS\_2, moderate; FCS\_3, high), and attitude towards mountain pasture practice values (MPP\_1, low; MPP\_2, moderate; MPP\_3, high).

#### **3. Results**

#### *3.1. Instrumental Analysis*

Overall, five instrumental parameters—two for colour and three for texture—showed significantly different mean values in the two cheeses (Table 4). Cheese produced with pasture milk and a longer ripening period was more yellow, having a higher b\* index, whereas cheese produced with stall milk was lighter, having a higher L\* index. Three out of nine texture parameters showed significantly different mean values between the two cheeses. Cheese produced with pasture milk and a longer ripening period was harder, resistant, and more elastic, showing greater values for linear distance force (computation of the force curve length), delta force (difference between yield force and final force), and elasticity modulus, computed as the ratio between stress and strain.


**Table 4.** Instrumental characterisation of pasture (P) and stall (S) cheese: Means, standard deviations (in parenthesis), and *p*-values for colour and texture parameters.

\* Bonferroni corrected *p*-values.

#### *3.2. Consumer Panel Profile*

A total of 156 subjects (55% men) aged between 18 and 75 took part in the test (Table 5). From the analysis of socio-demographic data, it was found that the participants had a high level of education: 50% declared a secondary school and 31% a bachelor's/master's degree. Furthermore, 73% reported to live with their families, mainly in urban areas (60%), and 48% did not have children. With regard to lifestyle and behavioural habits, it emerged that consumers adopted an average healthy lifestyle: 63% claimed they had never smoked and 53% practiced sports up to twice a week.


**Table 5.** Percentage distribution of socio-demographic characteristics by gender of respondents recruited in the consumer study.

Regarding eating habits, the majority of the participants (69%) were omnivorous, 28% flexitarian (60% of them mainly lead a diet in which the consumption of meat was limited), and 3% vegetarian. As the tested product was of animal origin, vegans did not participate in the study.

With regard to attention to organic food, only 10% of the participants stated that organic products comprised more than 40% of their weekly shopping. Consumers, on the other hand, were attentive to the purchase of zero food miles products and to limiting food waste: The majority (76%) declared throwing away less than 5% of their weekly shopping.

A total of 44% of the consumer panel quite often organized excursions to the mountain huts (malga), mainly in Trentino province (87%). On these occasions, among the dairy products locally produced, they chose to buy fresh (33%) and mature cheese (33%), butter (16%), yogurt (10%), and milk (4%). The list of statements on mountain pasture practices and relative average scores and standard deviations are shown in Table 3. All statements

obtained a high average score, demonstrating how the consumers associated positive opinions with mountain pasture practices. The only exception was the second statement ("Both stable and pasture management have the same impact on climate change"), which divided the opinion of the participants. However, only 33% of participants were aware of the possibility of buying cheeses with the "Sapori di Montagna" label directly from the supermarket.

#### *3.3. Consumer Segmentation*

Before classifying the consumers, according to the scales, the internal validity of each scale was verified (Table 6). All scales were reliable, showing standardized Cronbach's alfas higher than 0.6 [59]. The NPI scale revealed a lower, but still comparable, internal validity, in comparison with that originally described by Roininen et al. [49] (Cronbach's alpha = 0.76). In Table 6, the percentages of people in the three groups (low, moderate, and high interest/attitude), calculated for each scale according to the 33rd and 66th percentiles, are reported.

**Table 6.** Means (M), standard deviations (SD), percentile cut-points (33rd and 66th), percentage of participants in each group (L = low attitude; M = moderate attitude; H = high attitude), and reliability of attitude scales: Natural product interest (NPI), attitudes toward sustainability in general (ATS), food consumption sustainability (FCS), and mountain pasture practice scale (MPP).


We found some associations with gender and age: Women (*p* = 0.033) and older respondents (*p* = 0.035) rated FCS items higher than men or younger respondents, respectively. Similar results for older consumers were obtained in the NPI and MPP knowledge scales (*p* < 0.05).

Additionally, those who showed a higher NPI declared a higher weekly purchase of organic and zero food miles products (*p* = 0.0004 and *p* = 0.006, respectively), and declared to be mainly flexitarians and vegetarians (*p* = 0.002). Participants who showed a higher attitude towards FCS declared less than 5% of weekly food waste (*p* = 0.003), more than 40% of weekly purchase of zero food mile products (*p* = 0.006), and limited or no meat consumption (*p* = 0.09).

In the first row of Table 7, the results of the ANOVA mixed model for the whole consumer panel are reported. The main factors of product and external information had significant effects on consumer liking scores. Among the interaction effects, only that between consumer and product (C × P) was significant; showing that, among all consumers, people with different liking for products were present. External information had the strongest effect (MS = 76.16), with products claimed as "mountain pasture product" (Claim\_P; M = 7.0, SD = 1.4) being statistically more preferred than those claimed as "valley floor stall product" (Claim\_S; M = 6.3, SD = 1.6), regardless of the cheese effectively tasted (Figure 2a). Product effect was the second strongest (MS = 42.06), with more seasoned and pasture cheese (P; M = 6.9, SD = 1.5) being statistically different and more preferred than the stall one (L; M = 6.4, SD = 1.6); see Figure 2b.



**Figure 2.** Effects of main factors in the conjoint study for the whole panel: (**a**) Information effect; and (**b**) Product effect. Bonferroni corrected *p*\* is reported.

The results of the ANOVA mixed model, recalculated for specific sub-groups of consumers identified by gender, age, altitude and urbanization of their place of residence, ATS, FCS, NPI, and opinion about MPP are reported in Table 7. Findings concerning the sub-groups of men, of people between 31 and 50 years old, of those who live in mountain areas, of those who had low ATS, high NPI, and positive opinions on MPP confirmed

those observed for the overall panel of consumers (first row of Table 7). The random main consumer effect was not significant for any of the sub-groups, except for the group of people with high FCS attitude.

The most interesting results concerned the main effects of product and external information. The effect of the information was significant for all consumer groups: Men and women, people with different ATS, and so on, always evaluated the cheeses presented with the claim "mountain pasture product" with higher liking scores. The only exception lay in those who had a low opinion of MPP, who were not influenced by this information. The product effect was significant for a few sub-groups of consumers, in which mountain pasture cheeses showed higher liking scores, in comparison with valley floor stall cheeses. Mountain pasture cheese was more appreciated in the group of men, people aged between 31 and 50 years old, and those who had a very positive opinion on mountain pasture practices, although the effect of information remained the most important (Table 7, in parenthesis). Furthermore, mountain cheese was also more appreciated by those who lived above 600 m a.s.l. than those at lower altitude, as well as by those who had higher FCS attitude and NPI than those who were less attentive to these food aspects: In these groups, the product effect was a more important factor than information (Table 7, in parentheses). Furthermore, living in a more urban area or having a different attitude towards the social, economic, and environmental aspects of sustainability did not influence the significant effect of information on mountain production. Surprisingly, in the group of those who were less attentive to sustainability aspects, the product effect was the most important factor, showing a significant preference for mountain pasture cheese. Furthermore, there were changes in the significance of interaction effects between consumer and design factors on liking, which varied depending on the sub-group. In any case, the significant presence of these effects meant that, even within consumer sub-groups, different opinions were possible, both in the evaluation of the product and of the information.

#### **4. Discussion**

#### *4.1. Mountain Cheese Acceptability*

In the present paper, the acceptability of typical local cheeses produced in the same dairy from either mountain pasture or valley floor stall milk was investigated. Overall, both cheeses achieved good consumer acceptability, obtaining average liking scores ranging from 6.0 to 7.3. These results confirm what previous studies have stated, even without a real product tasting, as the quality of dairy products is an important aspect for both tourists and local farmers in the perception of summer mountain farms [3,20]. The significant effect of the product found for the whole panel (first row of Table 7) demonstrated that pasture cheese was significantly preferred over the stall one. This is in line with the results of Romanzin et al. [28], who reported that consumers expressed a higher actual liking for mountain pasture Montasio cheese. However, the cheeses evaluated here had different ripening and were different (as demonstrated by instrumental analyses), both in terms of colour and texture, with the pasture cheeses being harder and more yellow. Colour and texture generally change with the aging of dairy products [62,63], even if the yellow colour in milk and cheese also highly depends on their carotenoid content, which is generally higher in spontaneous pasture [64]. Our finding confirms that mountain pasture dairy products are recognizable and distinguishable from valley floor stall products [17,18]. Nevertheless, product sensory differences due to different milk production were probably influenced by differences induced by maturation [65], as the pasture cheeses were ripened 100 days more than the stall ones.

#### *4.2. Information Effect*

For the whole consumer panel, cheese acceptability was influenced by external information about the milk used for cheese manufacturing, showing that the "mountain pasture product" claim generated higher liking scores. This effect remained true for all the consumer segments investigated, except for people with a less positive opinion of

mountain pasture practices (MPP1), who seemed not to be influenced by this information. Moreover, there were no effects due to gender, age, different area of residence, or different level of interest to natural products or different awareness of sustainable aspects (i.e., neither environmental nor food).

The strength and persistence of the external information effect on acceptability in the various groups of consumers surprised the authors, who expected a non-significance for the overall sample and significant effects in the groups of consumers who were more aware or sensitive to the information given [32,66]. This, in itself, is an important result; perhaps being due to the more detailed way in which the information was transmitted (i.e., a short sentence on the origin of the milk, accompanied by an image of the animal grazing or in the stable, depending on the level of information; Figure 1). Osburg et al. [67] reported, in fact, that more detailed information on the product increases both the purchase intention and the trust in eco-friendly products.

It is also possible that the image itself, in addition to reinforcing the concept already expressed in the claim, carries other concepts connected to the main information, such as animal welfare, naturalness, and sustainability issues that, when connected to food, also indicates a local, organic, and traditional product [68]. It is also known that visual imagery is an effective way to inform consumers and capture their interest, but also to increase the perceived product benefits [69]. Furthermore, the image linked to the mountain pasture cheese message used in this study was predominated by the colour green, which has been demonstrated to be associated with environmental friendliness and is the most effective for producing positive attitudes [70]. Previous studies that have examined consumer perception on verbal versus pictorial claims reached opposite conclusions on which modality is most effective [71,72]. Hence, further studies in this sense are necessary, in order to deepen the understanding of the message actually perceived by the consumer and to establish the most effective modality, passing from the evoked concept to a claim on packaging.

#### *4.3. Segmentation Effect*

The segmentation scales used in the present work—both those developed for this study and those developed by other authors—proved to be sufficiently reliable and with good external interpretability. The FCS scale, developed to measure attitudes towards sustainable food-related behaviours, was effective in assigning higher scores to people producing less domestic waste, purchasing a greater percentage of zero food miles and organic products, and people following a diet that limits or refuses meat consumption. The MPP scale, developed to investigate consumer opinions and knowledge about mountain pasture practices, was efficient in identifying consumers with different sensitiveness to external information about mountain pasture cheese.

Gender and age effects were found for the FCS, MPP, and NPI scales: Women and older consumers had higher scores with respect to food sustainability, mountain pasture practices aspects, and natural food interest. This finding was in line with our expectations: Women and older consumers are generally more attentive to the ecological aspects of food [36,73–75]. Cavaliere and Ventura [68], instead, argued that millennials are the most sustainable and environmentally friendly generation, even if their study did not include comparisons with groups of respondents aged over 30 years. The gender effect is also controversial. Muratore and Zarba [76] found that environmental aspects are more important to males, whereas other studies did not find any gender influence [77]. In our case, both genders were sensitive to information given; the difference was perhaps due to the different evaluative cues used by the two groups. Rahman et al. [78] saw that, in the evaluation of the sustainable aspects related to the production of garments, females were more sensitive to aspects of animal welfare, while males were more attentive to environmental aspects, such as air quality or the quantity of water used in production.

For mountain-related factors, those who lived in mountain areas (more than 600 m a.s.l) showed significant effects for both information and product factors; the latter was also the

most important factor in this group of people. This may be due to the fact that inhabitants of mountain areas, compared to those who live in the plains or hills, are more familiar with the mountain pasture products, recognizing their sensory characteristics and associating them as local and zero food miles products.

The segmentation between those who lived in urban and rural areas did not lead to differences: There was no significant preference for one cheese over the other, while information on the mountain pasture product had a significant effect on liking in both groups. These results are contrary to those found by Zuliani et al. [2], who revealed a gap between urban consumer conception of mountain farming and the actual farming practices. The authors would have expected a non-significant effect of information in this group of consumers, with less knowledge of farming practices. This aspect is in line with the results obtained by segmenting consumers on the basis of their opinions on mountain pasture practices: Those with a less positive opinion were not influenced by external information, while those with a moderately positive opinion were. Those who had a highly positive opinion of MPP were not only sensitive to information given but also recognized the mountain pasture product as a better product, even if the information factor remained the most important one.

It is well-known that consumers are willing to pay a premium price to support local and organic food [24,79–83]. In our findings, segmentations based on FCS and NPI—which investigate attitudes towards organic and local products—did not show any difference in terms of external information influence. Instead, there was a difference in terms of product influence: Those with a high FCS and NP attitude actually preferred mountain pasture cheese and, thus, were more attentive to the aspects related to the product, which was also the most important factor.

Consumer segments with different attitudes towards socio-economic and environmental sustainability showed the same sensitivity to information. People presenting moderate or high attitudes seemed less attentive to the products: Those less engaged with sustainability were those who preferred mountain pasture cheese over the stall one. These results can be partly explained by the fact that, although concern and awareness of environmental problems are decisive for individual choices, the correlation between these aspects and actual behaviour is weak [84]. Furthermore, possessing environmental values, being aware of environmental problems, and having a correct perception of one's own ecological footprint are necessary, but not sufficient, conditions to generate pro-environmental behaviour [19,36,85,86]; and, perhaps, not even to recognize these positive externalities in a food product.

#### *4.4. Limitations and Future Perspectives*

Our sample size of 156 subjects can be seen as a limitation. However, in the field of sensory sciences, the size of the consumer sample for an acceptability test is commonly recognised as adequate at around 100 consumers [87], even though the sample size from previously published research in the field commonly exceeds 100–120 subjects [88]. Thus, a consumer panel of 156 consumers can be considered an acceptable sample size, also because participants shared a core common feature (i.e., being potential consumers of dairy products), which represents the target demographic under investigation. In addition, our sample was balanced for gender and age classes.

All participants attended the cheese fair "La Casolara", which is both an advantage and a limitation for the purposes of this study. If, on one hand, it is possible to reach a large number of cheese consumers in a short time; on the other hand, despite the ability of these events to attract not only local visitors, the majority of the consumers came from the same region. The external validity of our findings would surely be enhanced by considering a more representative sample of consumers coming from various regions. Furthermore, the preference for local food products—and, thus, mountain pasture products—could be related to the regional ethnocentrism of consumers [89], due to the close connection between local traditional practices and regional origin. We tried to measure this connection

by classifying the subjects by area of origin, in terms of altitude and degree of urbanization, assuming a closer link for those who lived in a rural or mountain area.

Some of the variables included in this study were directly or indirectly measured using scales, while others were self-reported. It should be noted that the self-reported variables could be inaccurate, due to memory recall and subject bias [90]. Thus, in order to avoid the distortion of responses to different test instruction presentations, all the instructions were not given verbally but submitted on a screen, being the test anonymously administrated on a computer.

The study could be replicated by increasing the sample size and screening for regular cheese consumption. Future research could generalize the results obtained here into a national context, in order to identify the drivers promoting the introduction of these products into new markets. It would be interesting to repeat the test using other, more well-known types of cheese, such as parmesan, or some other types for which the sign of quality of "mountain product" is sought. Furthermore, the liking of other mountain products, such as butter or yogurt, which are sold directly in the mountain hut could be investigated. Finally, future research could also be dedicated to the study of the images and claims used to evoke the concept of valley floor stall and mountain pasture products, identifying the most effective ones in a large validation study.

#### **5. Conclusions**

The present study showed that the impact of "mountain product" information on the acceptability of local cheese generated an overall positive response in the consumers, who assigned it extra value. The importance of this extrinsic characteristic exceeded the intrinsic value of the tasted product, which, however, was globally recognized, even if to a lesser extent. The positive effect of information persisted even within groups of consumers with different socio-environmental characteristics and different levels of interest and attitude. This study also showed that the consumers generally associated positive opinions with mountain pasture practices; it was precisely the lack of this positive association that made the claim lose its effectiveness. Furthermore, the consumers who lived in mountainous areas, who had a high opinion of mountain pasture practice, and who were predisposed towards local and organic food products and sustainable food-related behaviours were able to identify mountain pasture cheese as a product of higher quality than the valley floor stall cheese. This study contributes to revealing that the foundations exist for mountain pasture products to become mainstream products for all consumers. Nevertheless, effort is needed to promote the product in places other than mountain huts or dairy vendors.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2304-815 8/10/3/682/s1. Table S1: List of the original statements of the Natural Product Interest domain of the Heath and Taste Attitude Scale (HTAS) (a) and the translation in Italian (b). Table S2: List of the original statements of the Welsh Sustainability Segmentation Screening Tool (a) and its adaptation in Italian (b). Table S3: List of the statements of the Food Consumption Sustainability scale in Italian. Table S4: List of the statements of the Mountain Pasture Practice scale in Italian.

**Author Contributions:** Conceptualization, I.E. and F.G.; Methodology, I.E., F.G., D.C., and L.M.; Software, I.E.; Validation, I.E., D.C., and L.M.; Formal Analysis, I.E.; Investigation, I.E., D.C., L.M., E.A., and F.G.; Writing—Original Draft Preparation, I.E.; Writing—Review & Editing, I.E., D.C., L.M., E.A., and F.G.; Supervision, I.E.; Project Administration, F.G.; Funding Acquisition, F.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research was funded by RDP 2014–2020, Autonomous Province of Trento, measure 16.1.1. and 16.1.2 (SMartAlp project, CUP C66D17000170008).

**Institutional Review Board Statement:** Ethical review and approval were waived for this study as the experimental protocol did not expose the participants to particular psychological, social or physical risks other than those conventionally defined as minimum. Participants voluntarily attended the study, provided their written consent, and had the right to withdraw the consent at any time and for any reason. Moreover, the present study did not include the collection or analysis of data that could directly be used to define the identity of participants, any risk of discomfort or inconvenience to participants, any risk of psychological distress to participants or their families, the involvement of vulnerable groups of participants whose capacity to consent to participation may be challengeable, and international data collection. The authors declare that the experimental protocol was conducted according to the guidelines of the Declaration of Helsinki.

**Informed Consent Statement:** Written informed consent was obtained from all subjects involved in the study, according to the European Data Protection Regulation (UE 679/2016).

**Data Availability Statement:** The data generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

**Acknowledgments:** The authors thank all the volunteers who participated in the test during the fair "La Casolara 2019", as well as Federica Penza, Jessica Zambanini, Emanuela Betta, Andrea Favaro, and Carine Pachoud for their contributions to data collection. We particularly thank to Heidi Hauffe and Johnathan Ezio Olandi, for their support in forward–backward translation method from English of the Welsh Sustainability Segmentation Screening Tool.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Consumer Acceptability and Sensory Profile of Sustainable Paper-Based Packaging**

**Stella Lignou \* and Omobolanle O. Oloyede**

Sensory Science Centre, Department of Food and Nutritional Sciences, University of Reading, Harry Nursten Building, Pepper Lane, Whiteknights, Reading RG6 6DZ, UK; bola.oloyede@reading.ac.uk **\*** Correspondence: s.lignou@reading.ac.uk; Tel.: +44-(0)118-378-8717

**Abstract:** Sustainability appears to be increasingly important to consumers. In order for companies to reach their sustainability targets and offer more environmentally friendly solutions to consumers, food producers and retailers have begun to change their packaging to more recyclable, bio-based and biodegradable packaging. This study evaluated the sensory characteristics of paper-based prototype packages developed for two product categories (biscuit and meat packages) using a trained sensory panel. Consumer liking, preference and purchase intent were assessed by 130 participants. For the biscuit packages, no significant differences were observed for the liking of any of the four dimensions assessed (appearance, design, feel or overall liking). However, consumer segmentation identified three relatively homogeneous groups of consumers exhibiting differences in the hedonic reaction to the three packages. For the meat packages, significant differences and preference were observed between the original and paper-based packages. For both categories, the purchase intent was low, indicating that further work needed to be done to improve several quality characteristics (e.g., design, size and strength of the package), which would lead to better consumer acceptability.

**Keywords:** paper-based packaging; sensory attributes; consumer acceptability; biscuit packages; meat packages

#### **1. Introduction**

Packaging is essential in providing adequate protection to foodstuffs during transport, distribution and storage, thus reducing food loss and waste. Packaging that has a comparatively low environmental impact as assessed by life-cycle assessment models can be considered to be sustainable packaging [1]. From a consumer point of view, a packaging design that evokes explicitly or implicitly the eco-friendliness of the packaging can be considered to be sustainable packaging [2]. Since sustainability appears to be increasingly important to consumers [3,4], market interest in alternative forms to plastic packaging has increased drastically in recent years [5]. In order for companies to reach their sustainability targets and offer more environmentally friendly solutions to consumers, food producers and retailers have started to change their packaging to more recyclable, bio-based and biodegradable packaging. As paper fulfils these requirements and is easily understood by consumers, there is a high market interest for paper-based solutions [5].

Paper and cardboard packaging were the most recycled packaging in the UK and Europe in 2018, with recycling rates of 74.4% and 82.9%, respectively [6]. This has motivated companies toward the use of paper-based packaging. In addition, recyclable materials generally give the impression that the packaging is environmentally friendly [7–9]. Paper/cardboard is associated with positive emotions and attributes such as trust, biological/natural [10], homely and fresh products [11]. It is generally preferred over plastic because plastic is associated with emotions and attitudes such as unnecessary, expensive or bad for the environment [11].

There is limited research looking at sustainable paper-based packaging while analytically exploring the sensory characteristics of the packaging and consumers' perceptions.

**Citation:** Lignou, S.; Oloyede, O.O. Consumer Acceptability and Sensory Profile of Sustainable Paper-Based Packaging. *Foods* **2021**, *10*, 990. https://doi.org/10.3390/ foods10050990

Academic Editor: Antti Knaapila

Received: 31 March 2021 Accepted: 28 April 2021 Published: 1 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Magnier & Schoormans [12] estimated the effects of visual appearance and verbal sustainability claims on purchase intent and found that consumer responses to the visual appearance and verbal sustainability claims of the package depended on their level of environmental concern. The study showed that consumers with low environmental concerns evaluated conventional-looking packages with a verbal sustainability claim more negatively. In a subsequent study, Magnier et al. [13] investigated the effect of packaging sustainability on consumers' perceived quality of three product categories and found a more positive perceived quality of a food product when it was packed in a sustainable rather than conventional way. Steenis et al. [14] showed how packaging sustainability influenced consumer perceptions, inferences and attitudes toward packaged products. They demonstrated that consumers often rely on misleading and inaccurate beliefs when judging packaging for sustainability. Most studies acknowledge how the expectations and responses of consumers vary based on the design (shape, orientation, alignment of graphical forms), branding, visual appearance, colour, verbal claims and quality of products [13,15–19].

Research has shown that consumers decide what to purchase based on extrinsic product characteristics and appearance [20]. Consumer perception of extrinsic product cues such as packaging material and brand name differs from intrinsic product cues such as aroma, flavour and texture [21]. Packaging and branding as extrinsic product cues have been shown to have an influence on how consumers evaluate food products [22] and can determine consumers' expectations [23]. Thus, it is important that careful attention is given to the design of a package because of its dual role: attracting consumers' attention and creating expectations of the sensorial properties of the product [21].

According to a recent systematic review by Ketelsen et al. [24], there were only two studies [25,26] focusing on consumers' affective liking of environmentally friendly packaging, so research in this area has been quite limited. The study conducted by Koenig-Lewis et al. [25] explored consumers' emotional and rational evaluations of proenvironmental packages for beverages. Sijtsema et al. [26] investigated consumers' perceptions of 'bio-based' products and found that while participants were unfamiliar with 'bio-based' as a concept, they associated the word 'bio-based' with both positive and negative sustainability attributes. Therefore, our study (a) evaluated the sensory characteristics of the newly developed paper-based packages for two product categories (biscuit and meat packages), as per Oloyede & Lignou [27] and (b) investigated consumer acceptability, liking and preference of the developed packages and also explored purchase intent.

#### **2. Materials and Methods**

#### *2.1. Materials*

Innovative, 3-dimensionally formed paper-based packages were developed for two product categories (biscuits/confectionery and meat/chilled products) using either 3D press forming or deep drawing technology.

#### 2.1.1. Biscuit Packages

Two paper-based prototypes were developed as an alternative to the traditional polyethylene terephthalate (PET) tray in flow wrap packaging for Strauss Ad Hazot chocolate-coated biscuits. A package of two formed cavities holding three cookies each were individually sealed and easily separable. Sample B1 had a smooth tray surface, whereas sample B2 had an embossed surface. Both versions (B1 and B2) were sealed with a printed lidding film, and trays were cut by twos, with individual trays connected to each other by perforation (Table 1).

#### **Table 1.** Biscuit and meat packages.

**Samples**

*Biscuit packages* B0: preformed polymer multicavity tray, polymer flow pack (horizontal)

*Meat packages* M0: preformed polymer tray with polymer lidding film

B1: form-fill-seal paper-based tray with paper-based lidding film

M1: preformed paper-based

tray with polymer lidding film

M2: form-sill-seal paper-based tray with polymer lidding film

B2: form-sill-seal paper-based tray with paper-based lidding film

M3: form-fill-seal paper-based tray with paper-based lidding film

º

º

#### 2.1.2. Meat Packages

Three paper-based prototypes were developed to replace an expanded polystyrene (EPS) tray for meat products for Colruyt Group. Sample M1 had an identical shape to the original tray and was formed by deep drawing, sample M2 was press formed with a smooth surface and less-steep side walls, and sample M3 was based on sample M2 with embossing in the bottom area and improved stiffness in the side walls. Samples M1 and M2 had a transparent polymer lidding film with the possibility to see the product, whereas sample M3 had a non-transparent paper-based lidding film (Table 1).

Life cycle assessment conducted on the paper-based trays with PET coating showed a lower environmental impact compared to plastic crystalline polyethylene terephthalate (CPET) trays and recycled plastic recycled polyethylene terephthalate (rPET) trays [21].

#### *2.2. Sensory Evaluation of the Packages*

º º Sensory evaluation was carried out using quantitative descriptive analysis (QDATM) to determine the sensory characteristics of the various prototype packages, and the characteristics were estimated quantitatively. A screened and trained sensory panel (*n* = 12; 11 female and 1 male) was used, and each member had a minimum of 1 years' experience with expertise in profiling techniques. The panellists received 5 h specific training (1 h per day) over a period of 5 days for each category of packages (biscuits and meat packages) (a total of 10 days for both categories). During the development of the sensory profile, the panellists were asked to describe the appearance and feel of the package and then open the package and describe the interior in order to produce as many descriptive terms as seemed appropriate. The terms were discussed by the panellists as a group, with the help of the panel leader, and this led to a consensus vocabulary of 15 and 16 attributes for biscuit and meat packages, respectively, as outlined in Tables 2 and 3 in Section 3. The quantitative sensory assessment was carried out in a temperature-controlled room (22 ◦C) under artificial daylight and in isolated booths, each equipped with an iPad. All panellists scored in duplicate for each sample in separate sessions (30 min each) over two days for each product category. Compusense Cloud Software (Version 21.0.7713.26683, Compusense, Guelph,

ON, Canada) was used to acquire the sensory data. In total, 7 samples were evaluated (3 biscuit packages and 4 meat packages in separated sessions). Samples, coded with three-digit random numbers, were provided in a monadic balanced order, with sample sets randomly allocated to panellists within each product category. Panellists were instructed to evaluate the appearance attributes first and then open the package and evaluate the remaining attributes related to the interior of the package. The intensity of each attribute for each sample was recorded on a 100-point unstructured line scale.


**Table 2.** Mean panel scores for sensory attributes of the three biscuit packages.

<sup>1</sup> Means not labelled with the same letters are significantly different (*p* < 0.05); means are from two replicate samples, measured on an unstructured line scale (0–100). <sup>2</sup> Fisher's least significance difference (LSD) at *p* = 0.05. <sup>3</sup> Probability, obtained from ANOVA, that there is a difference between the means.

**Table 3.** Mean panel scores for sensory attributes of the four meat packages.


<sup>1</sup> Means not labelled with the same letters are significantly different (*p* < 0.05); means are from two replicate samples, measured on an unstructured line scale (0–100). <sup>2</sup> Fisher's least significance difference (LSD) at *p* = 0.05. <sup>3</sup> Probability, obtained from ANOVA, that there is a difference between the means.

#### *2.3. Consumer Evaluation of the Packages*

The study was conducted at the Sensory Science Centre at the University of Reading (UK). One hundred and thirty people were recruited across the University of Reading and Berkshire area (male and female, aged 18 years and above, without allergies or intolerances to wheat, gluten and/or dairy). Consumers who took part in the qualitative part of the study [27] were not allowed to sign up. Participants attended a single, 45-min session. Samples were presented to the participants, and after observing the samples, they were asked to rate their liking (appearance, design, feel, overall) on a 9-point hedonic scale (where 1: dislike extremely, 5: neither like nor dislike, 9: like extremely) for all samples. They also indicated the appropriateness of attribute level on a 5-point Just-About-Right (JAR) scale for the following attributes: strength of the package (where 1: much too weak, 3: JAR and 5: much too strong) and naturalness (where 1: not much too natural, 3: JAR and 5: much too natural). Finally, consumers were asked to indicate their preference (ranking: most-preferred to least-preferred package for each category—biscuit or meat packages), purchase intent for the packages (5-point scale, where 1: definitely will not buy, 3: may or may not buy and 5: definitely will buy) and whether they regularly purchased or consumed biscuit or meat (pate) products. Participants were given the opportunity to leave additional comments after evaluating each package if they wanted to. In total, 7 samples were evaluated (3 biscuit packages and 4 meat packages in one session, but with a break between the two product categories). Samples were presented to consumers in a monadic balanced order using Williams design, with sample sets randomly assigned to consumers within each product category. The assessment took place in sensory booths as described in Section 2.2. Consumers were asked to not open the package during assessment. Data was collected using Compusense Cloud Software. The study was conducted in November 2019 and approved by the School of Chemistry, Food and Pharmacy Research Ethics Committee, University of Reading (study number: 51/19). Informed consent was obtained from all participants prior to the study.

#### *2.4. Statistical Analysis*

SENPAQ version 5.01 (Qi Statistics, Kent, UK) was used to carry out ANOVA of sensory panel data, wherein the main effects (sample and assessor) were tested against the sample by assessor interaction, with sample as a fixed effect and assessor as a random effect. For those attributes exhibiting significant difference in the one-way ANOVA, Fisher's least significant difference (LSD) test was applied to determine which sample means differed significantly (*p* < 0.05).

XLSTAT 2019.3.2 version (Addinsoft, Paris, France) was used to carry out the following analyses: (i) principal component analysis of the sensory panel data, (ii) one-way ANOVA (and Fisher's LSD test) for the consumer liking and purchase intent data (iii) analysis of the preference (ranking) data using Friedman's test; (iv) agglomerative hierarchical clustering (AHC) for overall liking and (v) penalty analysis of the JAR data for strength and naturalness attributes. In more detail, for the AHC, dissimilarity of responses was determined by Euclidean distance, and agglomeration using Ward's method (set to automatic truncation). For the penalty analysis, the influence of consumer perception of appropriateness of attribute level rating (JAR) on consumer liking was evaluated by calculating the mean drop in liking rating (scale 1–9) compared with mean liking of consumers that rated the attribute as JAR (JAR 3 on a 1–5 scale), determining whether this drop in liking score was significant.

#### **3. Results**

#### *3.1. Sensory Evaluation of the Packages*

#### 3.1.1. Biscuit Packages

Table 2 summarises the mean panel scores of the sensory attributes for the three samples (B0, B1 and B2). All 15 attributes were significantly different between the original package (B0) and the two prototypes (B1, B2). Discrimination, repeatability and consistency were checked for all assessors (Supplementary Data, Table S1). In terms of the appearance

attributes, B0 was evaluated as having a more complex design with more amount of text present on the packages because the two prototypes (B1 and B2) had no labels at the back of the package. B0 was quite slippery to hold and the sound of it was very noisy in comparison to B1 and B2. The colour of B0 was dark red and shiny, and the package was very rigid overall. After opening the packages, B0 had many more tears compared to B1 and B2, the inner lid was very shiny and the tray was very rigid, too. Panellists found B1 easier to hold but more slippery to hold compared to the B2 package. Both the B1 and B2 packages did not make any noise and had a matte outer package appearance. B1 was found easier to open compared to B0 and B2. Both B1 and B2 had very shiny inner trays but a less shiny inner lid compared to B0.

Principal component analysis was carried out on the correlation matrix of all samples and all attributes in order to graphically visualise the differences between the samples. The first two principal components accounted for all the variation in the data (Figure 1). The first axis (76.26%) mainly separated B0 from the two prototypes, whereas the second axis separated the two prototypes—B1 and B2 (23.74%). The majority of the attributes were positively correlated with the first axis and thus associated with B0. Important attributes included the complexity of the design, the amount of text, the brightness of the colour of the package, the noise of the package and the rigidity of the tray before and after opening the package. On the other hand, the two prototypes had a shinier inner tray and sharp edges. The B2 package had a rougher bottom surface, whereas B1 was easier to open.

**Figure 1.** Principal component analysis of biscuit packages (B0, B1 and B2) showing correlations with sensory attributes (codes on plot refer to sensory attribute codes in Table 2).

#### 3.1.2. Meat Packages

Table 3 summarises the mean panel scores of the sensory attributes for the four meat packages (M0–M4). All 16 attributes were significantly different among the original package and the three prototypes. It could be observed that the M0 was quite deep with a very shiny and rigid tray. The lid was quite tightly sealed on the top of the package, and overall, the package was quite stable when placed on a table. After opening the packages, M0 was quite easy to open but tears developed on the lid. The tray was still quite rigid even after removing the lid, and the inner tray was found to be less shiny compared to the other packages (M1–M3). The three paper-based prototypes (M1–M3) were less deep compared

to the original package, had a cream colour and were not shiny. M3 was quite rigid before opening the package and exhibited similar scores to M0 in terms of the tightness of the lid and stability when placed on a table.

Similar to biscuit products, principal component analysis was carried out in order to graphically visualise the differences between the meat packages. The first two principal components accounted for 96.3% of the variation in the data (Figure 2). The first axis mainly separated M1 and M2 from M0, whereas the second axis separated M3 from the rest of the packages. Attributes positively correlated with the first axis, and thus associated with the M0 package, were the rigidity of the tray before and after opening the package, the shininess of the outer tray, the depth of the package, the tightness of the lid and sitting of the tray on the table. On the other hand, attributes negatively correlated with the first axis and thus associated with the M1 and M2 packages were the difficulty of separating the barrier and the difficulty of opening the package as well as the ability to hold the package and the level of slipperiness when holding the package. Transparency of the lid attribute positively correlated with the second axis and was negatively correlated with M4 packages because the lid was not transparent at all.

#### *3.2. Consumer Evaluation of the Products*

Table 4 summarises the demographic data for the consumers. One hundred and thirty consumers evaluated the samples. A higher proportion of the consumers were female (72.3%), and the mean and median ages were 32.8 and 29, respectively. More than one-third of the consumers were working (36.9%), and 58.5% were students. In total, 47.7% of the consumers that took part were people connected with the food, nutrition or sensory sector. The largest ethnic group to participate were White British (40%). The majority of the participants consumed or purchased biscuits sometimes or frequently (78.5%), whereas for the meat packaging, and particularly for pate (as this was the meat product inside the package), only 33.8% of the participants consumed or purchased it sometimes or frequently.


**Table 4.** Consumer demographics and characteristics of consumer panel.

#### 3.2.1. Biscuit Products

The mean liking scores of the packages are presented in Table 5. The results show that there were no significant differences in the appearance, design, feel and overall liking for all the samples tested, with all results ranging between like slightly and like moderately. While consumers did not like any of the packages very much, the results showed that both original and new packages were liked at a similar level, which can be seen as a positive for the new paper-based packages, to some extent.

In order to identify relatively homogeneous groups of consumers, agglomerative hierarchical cluster analysis was conducted, and three clusters of consumers were identified (Table 6). Consumers in cluster 1 (40.8%) liked slightly the original package of the biscuits and less the paper-based packages (B1 and B2). Cluster 2 (50%), the largest cluster, liked all three samples, whereas cluster 3 (9.2%), did not like B0 but liked moderately the paperbased packages.

Consumers were also asked to rank the samples in order of overall liking with 1-most liked and 3-least liked (Table 5). The results from the Friedman's test showed that there was no significant difference in preference ranking of overall liking of all the three packages, a result that it is in agreement with the non-significant result obtained for overall liking.


**Table 5.** Liking scores, preference ranking and purchase intent for biscuit and meat packages.

<sup>1</sup> Means not labelled with the same letters are significantly different (*p* < 0.05); means are from 130 consumers on a 9-point hedonic scale (from dislike extremely to like extremely). <sup>2</sup> Mean rank (1: most preferred to 3: least preferred). <sup>3</sup> Measured on a 5-point scale (1: definitely will not buy to 5: definitely will buy).

**Table 6.** Overall liking of the biscuit packages for the clusters of consumers obtained from agglomerative hierarchical clustering.


<sup>1</sup> Means not labelled with the same letters are significantly different (*p* < 0.05); means are from 53 consumers for cluster 1, 65 consumers for cluster 2 and 12 consumers for cluster 3, respectively. The mean for overall liking is from 130 consumers.

Penalty analysis was used to relate JAR data to liking scores and explain drivers of liking in relation to strength and naturalness, and the results are presented in Table 7. There was no significant difference in the JAR strength of the packages, and all three packages were perceived very close to Just-About-Right (JAR = 3). However, a significant difference was observed for the JAR naturalness, with packages B1 and B2 considered closer to Just-About-Right compared to B0.

When the attributes are not at the optimum level for a consumer this may have an effect on the overall liking. The penalty analysis showed that for samples B1 and B2 there was a negative impact on the overall liking when the strength of the package was considered too low. Similarly, for naturalness, there was a significant drop in the liking of all the packages when the naturalness of the package was considered to be 'too little' by the consumers with B1 considered to be the least natural of all the packages.

Finally, consumers were asked about their purchase intent of these packages (5-point scale: 1-definitely will not buy, 2-probably will not buy, 3-might or might not buy, 4 probably will buy and 5-definitely will buy). The mean scores of the purchase intent for all three packages ranged between 3.3-3.6 (Table 5), and a significant difference was observed (*p* = 0.039), with consumers more likely to buy B0 than B2. There were no significant differences between B1 and B2 (*p* = 0.636) or B0 and B1 (*p* = 0.110). Additional comments on the packages provided by the participants were both positive and negative. Some examples of those comments are shown in Table 8.


**Table 7.** Mean Just-About-Right ratings and influence on overall liking ratings.

\* Represents a significant difference (*p* < 0.05) within a sample in overall liking compared with mean liking rating when the sample was considered Just-About-Right. Frequency (%) is the % of participants within each group.

**Table 8.** Examples of participants' comments (one positive and one negative comment) relating to the various packages.


#### 3.2.2. Meat Packages

The mean liking scores of the meat packages are presented in Table 5. As can be observed, there were significant differences in all four liking dimensions. The appearance, design and overall liking of M0 were significantly higher than all the paper-based packages. No significant differences were observed between M2 and M3 for appearance, design or overall liking. In terms of the liking of the feel, the feel of M0 was significantly more liked than M1 and M3, but not M2.

Similar to the biscuits, AHC results are presented in Table 9 for the meat packages. Consumers in cluster 1 (27.7%) slightly liked the original package of the meat and disliked moderately to slightly the transparent film paper-based packages (M1 and M2), whereas they disliked very much the non-transparent paper-based package (M3). Cluster 2 (53.8%), the largest cluster, liked slightly M0 and M2, followed by M3 and M1. Finally, cluster 3 (18.5%) disliked very much the paper-based packages with transparent film (M1 and M2) and neither liked nor disliked the other two packages.

**Table 9.** Overall liking of the meat packages for the clusters of consumers obtained from agglomerative hierarchical clustering.


<sup>1</sup> Means not labelled with the same letters are significantly different (*p* < 0.05); means are from 36 consumers for cluster 1, 70 consumers for cluster 2 and 24 consumers for cluster 3, respectively. The mean for overall liking is from 130 consumers.

When consumers were asked to rank their preference in terms of overall liking, significant differences (*p* < 0.0001) were observed (Table 5). M0 significantly differed from all the other packages and was the most preferred. On the other hand, M1 and M3 did not differ significantly and were the least preferred of all. This result was again in agreement with the overall liking results discussed earlier.

Significant differences in Just-About-Right strength and naturalness attributes were observed for the four packages (Table 7). In terms of strength, M0 was perceived just above Just-About-Right (JAR = 3), whereas for the other three samples, the strength of the packages were considered 'not too strong'. For the naturalness attributes, the M2 sample was close to Just-About-Right, whereas the naturalness of M0 was considered 'not too natural'. The penalty analysis showed that for samples M1 to M3, there was a negative impact on the overall liking when the strength of the package was considered too low. Similarly, for naturalness, there was a significant drop in the liking of all the packages when the naturalness of the package was considered to be 'too little' by the consumers, with M0 considered to be the least natural of all the packages.

Finally, in terms of purchase intent, the mean scores for all the paper-based packages ranged between 2.4–2.8 (Table 5), which implied that consumers did not generally like the design of those packages. On the other hand, the purchase intent for samples M0 was at 3.3, between 'might or might not buy' and 'probably will buy'. Similar to the biscuit packages, participants' comments on the packages were both positive and negative, and examples of those comments are shown in Table 8.

#### **4. Discussion**

The present study aimed to (1) explore the sensory characteristics of the new paperbased packages developed during the study for two product categories (biscuits and meat) in comparison to the original packages, as assessed by a trained panel and (2) evaluate consumers' liking and perceptions of the said packages. The findings from this study build on and contribute to existing knowledge on consumer opinions and reactions to paper-based packaging material [27].

For the biscuit packages, no significant differences were observed for the liking of any of the four dimensions (appearance, design, feel or overall liking); however, consumer segmentation identified three relatively homogeneous groups of consumers exhibiting differences in hedonic reaction for the three packages. Even though no significant preference was observed (*p* = 0.299), consumers in each cluster varied in their responses. Consumers

in cluster 2 (50%) "liked moderately" all three packages but seemed to "like significantly" more the new paper-based packages. Similarly, the paper-based packages were liked more by the consumers in cluster 3 (9.2%) who disliked the original package (B0). In a study conducted by Fernqvist et al. [11] exploring consumers' views on different aspects of fruit and vegetable packaging, the authors found that the design of the package was interpreted differently among participants. While some participants had a positive perception about the package, others had a negative opinion. Consumers were given the opportunity to add comments for the various packages, and it was clear that they appreciated the innovative packages of B1 and B2, and they loved the duo-pack design that meant a separation of the packages and that the consumption of a smaller portion was possible while keeping the other portion '*fresh, crisp and for longer*'. As expected, the paper-based packages had a more natural and sustainable feel when compared to the B0 package, and this was apparent from the Just-About-Right ratings and consumers' comments: '*it feels very natural*', '*it looks sustainable*', '*the packaging seems more natural and biodegradable*'. The results also demonstrated that there was a significant drop in the overall liking of the package when the naturalness was considered to be 'too little'. Prior research has shown that sustainability perceptions can be closely related to other benefits such as naturalness [13], which is a positive characteristic of sustainable packaging.

Focusing on the characteristics of the paper-based packages, it seemed that even though consumers liked the smoothness of the B1 bottom surface and its '*sustainable look and nice feel*', they thought the tray was not too rigid and was a bit fragile. This was confirmed from the sensory evaluation results, wherein trained panellists scored B1 significantly lower (43.7) for rigidity before opening the package compared to the original package (69.2), and also from the significantly lower score in terms of the JAR strength attribute. The perception of the rigidity of the package was further reduced to 30.8 after opening the package and removing the lid. On the other hand, the B2 tray had an embossed bottom surface, which consumers felt was 'easy to hold' and was seen as a positive characteristic. This was also confirmed by the trained panel wherein the level of perceived slipperiness was significantly lower (21.3) for B2 compared to the B0 and B1 samples. The perception of fragility may have had an effect on consumers' acceptability of the B1 package, as it may have been seen as a quality issue of the package that could affect its ability to protect its content.

There was also a cluster of consumers (cluster 1—40.8%) that significantly liked the original package compared to the new packages. These were consumers who preferred to go with what they were familiar with and were less keen to try new propositions. Some of the consumers in this group had comments such as '*love the compact design*', '*seems like the standard design so keen to buy*', '*I am familiar with this packaging*', '*it immediately reminds me of biscuits, which I like*'. Most consumers tend to be creatures of habit and unwilling to try new things, as found by Oloyede & Lignou [27]. In addition, consumers have an expectation of what the package design should be like and would generally be averse to trying designs that do not match the picture they have in their minds. Zhang et al. [19] reported that the design style or colour of the package of UHT milk was shown to have an influence on consumer attraction. The authors suggested that if consumers are more attracted to the design style or colour, their willingness to purchase will be higher. Ares and Deliza [17] showed that package shape and colour could have an impact on consumers' expected liking scores and their sensory expectations in a product category such as desserts, and similar results were demonstrated with this study. The relevance of package characteristics, in this case the shape of a standard biscuit package, had an effect on consumers' perception and acceptance and also on purchase intent. Consumers were more likely to buy the original package as earlier discussed.

Regarding the meat packages, significant differences were observed for appearance, design, feel and overall liking with subsequent significant preference of certain packages over others (*p* < 0.0001). In general, consumers liked the original package (M0) more than the paper-based packages (M1–M3); however, similar to the biscuits, consumer

segmentation identified three clusters of consumers with varying overall liking for the four packages, which was clear from the comments they added. Consumers in the largest cluster (cluster 2—53.8%) equally liked M0 and M2 when compared to M1 and M3. Consumers felt that the polystyrene of M0 '*evokes hygiene—associated with meat*'. They liked the feel of the packaging, how sturdy and deep it was and the fact that the lid on top was not in direct contact with the meat. This result agrees with the findings of Oloyede & Lignou [27], wherein focus group participants were worried about contamination due to the top lid touching the meat. This was also confirmed by the trained panel, who scored significantly higher the depth of this package (72.6) and the rigidity before and after opening the package (94.6 and 93.2, respectively) compared to the other three packages. The overall liking in the other two clusters was mainly driven by whether the top lid was transparent or not. For example, consumers in cluster 1 (27.7%) disliked very much M3, equally disliked M2 and M3 and liked slightly M0, whereas consumers in cluster 3 (18.5%) equally disliked very much M1 and M2 and neither liked nor disliked M0 and M3.

Interestingly, no matter the cluster, the M1 and M2 packages had very similar characteristics in general, which was confirmed from the sensory evaluation. Both samples had a smooth bottom surface that resulted in significantly higher perceived ability to hold, level of slipperiness and very low rigidity before and after opening the packages compared to M0 and M4. Some consumers liked this feel and stated that it '*looked very neat*'. The only differences observed between the two packages was the difficulty of opening the package and the difficulty in separating the inner barrier, with both receiving a higher rating for the M1 package. Observing the results for the M3 package, it seemed that on one hand, consumers preferred the embossed packaging tray over the non-embossed due to the touch and feel of the paper, the sturdiness and the fact that it made the packaging look more attractive (5.6 hedonic liking for cluster 2 and 5.0 for cluster 3); however, it was clear that for certain consumers, the lidding material and its transparency was crucial (2.1 hedonic liking for cluster 1), as consumers in general prefer to see the content of the packaging [28], and especially when the product is meat. Transparent packaging has been shown to increase willingness to purchase, expected freshness and expected quality in different food categories (cereal, boxed chocolates, dried pasta and fresh fish) [29]. Consumers mentioned that there was a minimalistic feel associated with M3, and they liked the fact that it was all paper and no plastic; however, they worried that the paper package might absorb moisture or meat blood/liquid with time. These findings agree with the study by Magnier and Crie [2], who found that eco-friendly packages, because of their simplicity, minimalism and lack of colours, are often perceived as less appealing.

The results show that the positive and negative perceptions regarding the paper-based packages had an effect on the overall liking of the products, which in turn affected the purchase intent. The mean scores of the purchase intent for all three paper-based packages ranged between 2.4–2.8, which is between 'probably will not buy' and 'might or might not buy', implying that consumers did not generally like the design of these packages.

There are a couple of limitations to this study. Given the limited duration of the project, there was insufficient time to completely develop the packages and include all the relevant information regarding the labelling of the products. The biscuit packages, other than the red cover which had the same graphics as the original package, had no further information on the nutritional profile of the content or any information regarding the recyclability of the actual package. For the meat packages, the situation was even more complicated because there was no information at all about the product. Previous research has shown that consumers' responses to either visual or verbal responses can vary depending on cognitive resources [30]; however, in our case, no cues were provided to the consumers. Future research with packages having all the relevant information needed by the consumers printed on the package would allow for better comparisons, not only of the design and feel of the package/material but also the appearance and the messages to be delivered to the consumers.

#### **5. Conclusions**

The results from the sensory and consumer evaluation of the new paper-based packages clearly demonstrated that these packages were a good example of how paper can be used as an alternative to plastic or foil for the development of packages in product categories such as confectionery or chilled products. In summary, consumers liked the sustainable nature of the paper-based packages; however, they found the trays (particularly for the meat packages) to be flimsy and not strong enough. For the biscuits, they liked the innovative design of the double pack but also loved the compact design of the original package, as this was more familiar to them and looked like a standard pack of biscuits. The results showed that while consumers were open to sustainable propositions, other quality characteristics were key aspects that must be addressed if sustainable packaging is to become a viable option. From an industry point of view, considerations have to be based not only on the sustainable nature of the packaging material but also on the design and size of the packaging. This is because design and size of package are more important factors influencing consumer choice than the sustainable character of the packaging material. Thus, further work needs to be done to improve several quality characteristics (e.g., design and size of package), which would lead to better consumer acceptability.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10.3 390/foods10050990/s1, Table S1: Summary table of assessor performance for biscuit and meat packages.

**Author Contributions:** Conceptualization, O.O.O. and S.L.; methodology, O.O.O. and S.L.; software, O.O.O. and S.L.; validation, O.O.O. and S.L.; formal analysis, O.O.O. and S.L.; investigation, O.O.O. and S.L.; resources, O.O.O. and S.L.; data curation, S.L.; writing—original draft preparation, S.L.; writing—review and editing, O.O.O. and S.L.; visualization, O.O.O. and S.L.; supervision, S.L.; project administration, S.L.; funding acquisition, S.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the EUROPEAN INSTITUTE OF INNOVATION & TECHNO-LOGY—EIT FOOD, grant number 19136: "InPaper: Innovative paper-based packaging technology and pack styles for food production".

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the School of Chemistry, Food and Pharmacy Ethics Committee of University of Reading (study number: 51/19 and date of approval: 14 November 2019).

**Informed Consent Statement:** Informed consent was obtained from all participants involved in the study.

**Data Availability Statement:** The data presented in this study is available on request from the corresponding author.

**Acknowledgments:** We would like to thank all our industrial partners for their invaluable insights during the project and in particular, Matthias Klauser and Johannes Rauschnabel from Syntegon Technology GmbH (formerly Bosch Packaging Technology) for working on the development of packages B1, B2, M2 and M3; Alexander Lenske and Marek Hauptmann from IVV-Fraunhofer for the development of the M1 package; and Veerle Carlier from Colruyt Group and Efrat Ben Hamo from Strauss Group for supplying the M0 and B0 products, respectively. We would also like to thank the volunteers who took part in the study.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


### *Article* **Consumer Awareness, Attitudes and Preferences towards Heritage Cereals**

#### **Karin Wendin 1,2, \* , Arwa Mustafa 1 , Tove Ortman <sup>3</sup> and Karin Gerhardt 3**


Received: 4 May 2020; Accepted: 2 June 2020; Published: 4 June 2020

**Abstract:** Interest in heritage cereals is increasing among consumers, bakeries and farmers, and the trends point towards the local production of crops and connect to sustainability. The most known variety is spelt, which has opened up for old landraces such as Oland wheat. Heritage cereals have shown a higher resilience than modern varieties and have the potential to supply the market with alternative products that have an attractive cultural background. Delicious and nutritious products based on heritages cereals have a growing market potential. Consumers' attitudes and preferences to different products are affected by factors such as age, gender and education. The aim of this study was to investigate and analyse different consumer groups' awareness, attitudes and preferences toward heritage cereals. The number of respondents who participated in this study and answered the web-based questionnaire was 434. It can be concluded that most consumers are aware of heritage cereals. Geographic background had an influence, while academic background did not. Bread and pasta are the most consumed products and are regarded as the most popular future products to be based on heritage cereals. The most essential factors in bread are taste and flavour, followed by freshness and texture. The origin of the cereal and its health aspects are important; women are more concerned about the origin than men, while older consumers are more concerned about health. Older consumers are also more willing to pay extra for heritage cereal than younger consumers.

**Keywords:** heritage cereals; consumer attitudes; preferences and awareness

#### **1. Introduction**

Today's consumer trends are moving more towards the local and regional production of crops (e.g., ancient or heritage crops), mainly due to a rising interest in sustainability [1]. It has been shown that the taste experience of a product is of the greatest importance to the consumers. Furthermore, product claims, such as ancient, natural, organic, or local, are the most likely to have a positive impact on the consumer's preference and/or choice [2–5]. High acceptability has, for example, been shown for breads containing Kamut or spelt [6].

Despite the numerous genetic and historical data on the origins of agricultural products, there is no universal definition for modern and older cereals [7]. Ancient cereals, according to Giambanelli et al. (2013) [8], are represented by populations of primitive cereals, which were not subjected to any modern breeding or selection processes (e.g., emmer, einkorn and spelt). What are today named as landraces were originated by farmers using natural selection, consequently saving various seed types year after year [9–11]. For convenience, in this paper the term "heritage cereals" is meant to include ancient cereals, landraces and older varieties.

The pursuit of higher yields and the industrialisation of agriculture over the past 150 years, meant that heritage cereals were lost from many parts of the world [11–13]. For future sustainability, there is a need to build up resilient agricultural systems [13,14]. Heritage cereals have shown more resilience to drought or other extreme weather circumstances than the modern varieties, which in turn might contribute to a robust agricultural system [15,16].

There is currently a trend of revived interest in heritage cereals from consumers, artisan bakeries and farmers [6]. Farmers of organic crops are interested in certain agronomic traits in heritage cereals, which makes them suitable for organic production [17–19]. Additionally, the fact that they are often sold at a premium price makes the old varieties highly attractive for farmers [1,20]. Heritage cereals might as well supply the market with new types of products that have an attractive cultural background and connection to authentic stories. Storytelling is highly important for heritage cereals and their growers and is an influential "trademark" [21]. Moreover, the demand for locally produced food is increasing [22,23], and alternative types of distributional and sale systems have gained ground, i.e., "short food supply chains", in which the heritage cereals fit well. These short supply chains aim to redefine the producer–consumer relationship in terms of providing knowledge of the origin of the food [24]. In the case of Sweden, several initiatives promoting a direct contact between producer and consumer have emerged, with examples such as "Farmer's market", "Local Food Nodes" and "REKO-rings".

Encouraging the production and consumption of heritage cereals is in line with the Swedish food strategy and the current government goal to increase organic food production [25]. Cereal-based food products constitute a large and central part of the human diet and ancient cereals are suggested to possess health-promoting effects due to their unique nutritional content. Thus, the development of delicious products based on these ancient cereals may enhance the large market potential as well as boost the consumption of whole grain [1,26,27].

Consumers' attitudes and preferences for different kinds of products may differ according to factors such age, gender, education level and geographic background. For example, in the case of fruits and berries, it has been shown that sustainability aspects are of higher importance to women—mainly to younger women—than to men. In the case of bread consumption, Sandvik et al. [2] pointed towards a more traditional consumption structure among Swedish consumers, however, a lower consumption of rye and whole-grain bread could be observed among younger consumers. This is in accordance with other studies showing that older consumer groups are more concerned with health aspects in comparison to younger consumer groups [28]. Consumers with a higher educational level are more aware of the health aspects and are more receptive to trends [29,30], however, knowledge about the level of impact from education is low. Geographical and cultural backgrounds are further factors that might have an impact on the consumers. Thus, it is of interest to study awareness, knowledge and attitudes towards heritage cereals among different groups of consumers in higher education arenas.

#### **2. Aim**

The aim of this study was to investigate and analyse consumers' awareness, attitudes and preferences towards heritage cereals. A further aim was to study whether consumers differing in academic and geographic backgrounds varied in the mentioned aspects while taking age and gender into account.

#### **3. Materials and Methods**

#### *3.1. Consumers*

Swedish consumers from two different academic institutes in Sweden were invited to answer a questionnaire concerning awareness, attitudes and preferences towards heritage cereals. The academic locations were the Swedish University of Agricultural Sciences (SLU), which is a university with disciplines focusing on primary agricultural production, and Kristianstad University (HKR), which is a university with a multidisciplinary focus. The participants had to be affiliated with one of the universities, either as a student or as an employee. Participation was anonymous and voluntarily. To gain enough data for reliable statistical calculations, a minimum number of 100 adult consumers from each academic location aged 18 years or older was aimed for during the recruitment process [31].

#### *3.2. Questionnaire*

The web-based questionnaire was launched during the month of April 2019. The software, Eye Question (version 3.9.7, Logic 8, Elst, The Netherlands) was used for the data collection. The survey contained the following areas of investigation: (a) consumers' awareness and consumption of heritage cereals; (b) consumers' attitudes towards heritage cereals; (c) consumers' preferences of future products with heritage cereals. The different areas of investigation are given in Table 1. The full questionnaire is provided in Supplementary File S1.



#### *3.3. Statistical Evaluation*

The collected questionnaire data were processed using descriptive and analytical statistics. Mean values and standard deviations were calculated. A multiple comparison test was performed by one-way ANOVAs in conjunction with Tukey's Post-Hoc Tests to compare groups of consumers. For observed frequency data, a chi-squared test was performed to determine the level of significant differences between the expected frequencies and the observed frequencies. For all statistical calculations, the significance level was set to *p* < 0.05. SPSS (Version 23, IBM, New York, NY, USA) was used throughout the calculations. The free software, Wordle (wordle.net, IBM Corporation, New York, NY, USA) was used to generate a word cloud out of the words used to illustrate which type of bread the study group consumed.

#### **4. Results**

#### *4.1. Consumers*

The total number of participating consumers in the questionnaire was 434, of which 311 were women, 117 men and 6 X (unidentified). Details about the participants are shown in Table 2. From the total study population, 323 participants were affiliated with SLU and 111 participants were affiliated with HKR. The SLU participants consisted of 227 women, 92 men and 4 X. Age group 1 consisted of 120 participants, age group 2 consisted of 108 while age group 3 consisted of 95. The HKR participants consisted of 84 women, 25 men and 2 X. Age group 1 consisted of 37 participants, age group 2 consisted of 42 and age group 3 consisted of 32. Since there was a low number of participants in gender group X, the resulting data from this group have not been taken into consideration.


– – – –

**Table 2.** Demographical distribution of the study population. Age in years.

#### *4.2. Consumers' Awareness and Consumption of Heritage Cereals*

To get insight about consumers' awareness concerning heritage cereal varieties, they were presented with different varieties of heritage cereals and asked to identify those that they were familiar with. Figure 1 presents the different varieties of heritage cereals and the frequencies of awareness within the different population sectors. Spelt was the most known variety among the different population sectors, while Halland wheat was the least known. No significant difference between the groups was shown in the awareness of spelt, while chi-squared tests showed that Halland wheat was significantly more (χ <sup>2</sup> = 5.98; *p* < 0.05) known to HKR than to SLU participants. Additionally, Oland wheat was significantly more (χ <sup>2</sup> = 9.47; *p* < 0.05) identified by the HKR participants, yet, it was the least known within age group 3. Furthermore, einkorn was significantly more (χ <sup>2</sup> = 5.97; *p* < 0.05) known to age group 1. Regarding the identification of varieties Kamut (χ <sup>2</sup> = 9.23; *p* < 0.05) and Halland wheat (χ <sup>2</sup> = 8.18; *p* < 0.05) there was significant difference in their recognition among the age groups, where they were the least known for age group 3. – – – *4.2. Consumers' Awareness and Consumption of Heritage*  To get insight about consumers' awareness concerning heritage cereal varieties, they were alland wheat was significantly more (χ land wheat was significantly more (χ more (χ amut (χ wheat (χ

varieties of heritage cereals. The category "others" included black oat, quinoa, buck **Figure 1.** Frequency data given in percentages for each group showing the awareness of the different varieties of heritage cereals. The category "others" included black oat, quinoa, buck wheat, dala wheat, spring wheat, millet, naked oat, and teff. Swedish University of Agricultural Sciences (SLU), Kristianstad University, Sweden (HKR).

The popularity of the consumption of cereal-based products was investigated by the rate of consumption of the products on weekly bases, as shown in Figure 2.

category "others" included couscous, rice cookies, crispbread, beer, pancakes, quinoa, **Figure 2.** The frequency of consumption given in percentages for each group of cereal based products per week. The category "others" included couscous, rice cookies, crispbread, beer, pancakes, quinoa, millet, seeds, gluten free, buck wheat, and bulgur.

lower in age group 3 compared to the younger age groups (χ significantly more consumed by men (χ < 0.05) among SLU participants (χ and within age group 3 (χ The responses to the question about the cereal products consumed at least once per week revealed that bread, followed by pasta, was the most consumed product among all categories of cereal-based products. Chi-squared tests revealed that the consumption of pasta was significantly lower in age group 3 compared to the younger age groups (χ <sup>2</sup> = 8.30; *p* < 0.05). Flakes and Muesli were significantly more consumed by men (χ <sup>2</sup> = 6.66; *p* < 0.05) among SLU participants (χ <sup>2</sup> = 4.56; *p* < 0.05) and within age group 3 (χ <sup>2</sup> = 6.30; *p* < 0.05). Figure 2 shows that more than 90% of the study population consumed bread. Figure 3 gives an indication of the most common types of bread consumed among the study population. Among the most widely consumed were crispbread, sourdough bread, sourdough dark bread, rye bread, whole grain and home-baked bread. category "others" included couscous, rice cookies, crispbread, beer, pancakes, quinoa, lower in age group 3 compared to the younger age groups (χ significantly more consumed by men (χ < 0.05) among SLU participants (χ and within age group 3 (χ

**Figure 3.** Illustration giving an indication of the popularity of bread types consumed. The figure is based on the qualitative data given in the questionnaire.

#### *4.3. Consumers' Attitudes towards Heritage Cereals*

Responses about the consumers' habits and attitudes when it comes to home baking compared to purchasing sites are illustrated in Figure 4, showing that purchasing at grocery stores was, according to one-way ANOVA, a significantly more common habit among the study population than home-baking or purchasing bread in a baker's store. There was no significance difference when comparing home-baking or purchasing at a bakery shop, nor was there significant different due to gender or academic institution. However, it was significantly more common to purchase bread at bakery shops in age groups 2 and 3 compared to age group 1 (*f* = 11.30; *p* < 0.05). in a baker's store. There was no significance difference when

Responses about the consumers' habits and attitudes when

– **Figure 4.** Mean and standard deviations showing trends of purchasing at different sites. Significant differences are indicated with different letters. The scale was 1–4, where 1 = always, and 4 = never.

Factors that govern consumers' preference Factors that govern consumers' preferences for products based on heritage cereals are presented in Table 3.

From Table 3 it can be inferred that Taste/Flavour and Freshness were the most important quality aspects of the bread. When comparing women to men, it was found out by one-way ANOVA that that texture (*f* = 8.82; *p* < 0.05), having the bread made with wholemeal flour (*f* = 14.54; *p* < 0.05) as well as the origin (*f* = 19.64; *p* < 0.05) of the cereal, were significantly more important factors to women than to men. On the other hand, the brand of the bread seemed to be the factor that was least thought about; however, age group 3 seemed to more concerned with this than the younger consumer groups.

When comparing the two institutions, it was found that bread features such as. its appearance (*f* = 4.32; *p* < 0.05), made by sourdough (*f* = 7.61; *p* < 0.05) and its freshness (*f* = 7.83; *p* < 0.05) were significantly more important to the participants from HKR compared to those from SLU.

When investigating bread attributes in relation to age, it was revealed that Odour/Aroma is significantly less important for age group 1 than for older age groups (*f* = 18.65; *p* < 0.05). Sourdough is significantly less important for age group 1 than for age group 3 (*f* = 5.58; *p* < 0.05). Wholemeal is significantly less importance for age group 2 than for age group 1 and age group 3 (*f* = 4.39; *p* < 0.05). The importance of the health aspects of the bread was significantly different among the groups, where it was the most important factor for age group 3 (*f* = 9.41; *p* < 0.05). Being an organic cereal was of significantly less importance for age group 1 than for age groups 2 and 3 (*f* = 12.94; *p* < 0.05). The importance of the price of the bread differed among the groups. Nevertheless, it was significantly of the most important to age group 1 and of the least importance to age group 3 (*f* = 17.27; *p* < 0.05). Freshness of the bread was significantly less important for age group 1 than for age group 2 and age group 3 (*f* = 12.13; *p* < 0.05).

In Table 3, the mean values and standard deviations are given as well as indications of significant differences.


**Table 3.** Mean and standard deviations showing the importance of essential characteristics of the bread. Significant differences are indicated with different letters. A 5-pointed scale was used where five was regarded as a very important factor, while one represented the least important factor.

Significant differences are indicated with different letters. SLU: Swedish University of Agricultural Sciences; HKR: Kristianstad University, Sweden; M: mean; std: standard deviation.

On the question: "May you consider purchasing bread or other products that are based on heritage cereals?", as many as 98.4% of the study population responded to consider purchasing bread products based on heritage cereals.

#### *4.4. Consumers' Preferences of Future Products with Heritage Cereals*

To explore the future willingness of the study population to purchase cereal products based on heritage cereals, they were presented with set of product categories and were asked which product or products they would consider purchasing. The categories are presented in Figure 5. When comparing gender, the chi-squared test showed that the willingness to purchase porridge was significantly higher among women than men (χ <sup>2</sup> = 7.45; *p* < 0.05). When comparing age groups, it was revealed that age group 3 was more likely to purchase bread and significantly less probable to purchase pasta when compared to younger age groups (χ <sup>2</sup> = 21.16; *p* < 0.05). A similar pattern was seen for the purchase of porridge (χ <sup>2</sup> = 10.47; *p* < 0.05), cooking cereals (χ <sup>2</sup> = 14.86; *p* < 0.05) and cookies (χ <sup>2</sup> = 9.57; *p* < 0.05). On the other hand, participants in age group 2 were significantly more likely to purchase flakes (breakfast cereals) (χ <sup>2</sup> = 7.73; *p* < 0.05) and flour (χ <sup>2</sup> = 9.56; *p* < 0.05). Bread was the most popular product to be considered purchasing. No differences were observed between SLU and HKR participants.

To be able to get a deeper understanding of the population preferences regarding the accessibility of heritage cereal products, the participants were asked about the location where they would prefer to

"Others" included the following options: Delivery to home/to work,

purchase products. The set of locations are presented in Figure 6. Grocery stores were the most popular site for purchasing, followed by the bakery. However, according to chi-squared analysis, there was no significant difference among group categories. Only some respondents chose the category "others". he category "others" includes

"Others" included the following options: Delivery to home/to work, **Figure 6.** Preference of purchasing sites for heritage cereal products in the future. The category "Others" included the following options: Delivery to home/to work, REKO-ring, on-line/online store, directly from producer/grower, café, farm shop, bake yourself and market.

The willingness to pay more for products based on heritage cereals was a common attitude in the different groups of the study population, being more pronounced in age group 3. The chi-squared test showed that age group 1 was significantly the least willing to pay more for products based on heritage cereals in comparison to the other groups (χ <sup>2</sup> = 9.89; *p* < 0.05), as shown in Figure 7. "Others" included the following options: Delivery to home/to work,

**Figure 7.** Frequency in percentages showing willingness to pay more for heritage cereal products.

#### **5. Discussion**

This study shows a great consumer interest in heritage cereals, where almost all consumers would consider purchasing bread or other products based on heritage cereals. This may be explained by the health trends and their relation to heritage cereals [32]. Furthermore, this great interest is well supported by respondents' abilities to identify different varieties of heritage cereals (e.g., more than 95% were aware of the variety spelt). This predominance could, to some extent, be explained by the fact that spelt has a very long history and was used as staple cereal thousands of years ago [33–35], and it has been shown that the acceptance of spelt is high among consumers [6]. Over the last few decades, spelt has become more commonly used in baking, and the addition of spelt flour during bread-making gives unique sensory characteristics to the bread (e.g., makes the bread stiffer as well as giving it a prolonged shelf life) [36]. About 40%–50% of the respondents were familiar with other varieties (e.g., emmer, Kamut and Oland wheat). The high percentage of awareness amongst the respondents could probably be explained by their academic background and that many of them belonged to agricultural and food studies departments. Robinson [32] explained European that consumer interest and awareness is influenced by mainstream media, which consequently has become the driving demand for flours from heritage cereals. Swedish consumers, however, are more likely to be influenced by social media and influencers such as Adam Arnesson (@ekobonden), Sebastien Boduet

(@sebastienboduet) and many more. It is inferred that respondents from the academic institute HKR had a higher awareness of Oland wheat and Halland wheat compared to SLU, which could be related to the geographic location of the academic institute, where HKR is situated closer to areas where these varieties are cultivated.

It is noteworthy that bread and pasta are the most consumed cereal products. These are also the products that the respondents indicated as most suitable for future heritage cereal products and which they were most willing to purchase. The phenomenal product recognition, in this case bread and pasta, is well known, and for food innovations a combination of recognition, quality, tradition and social approval are very important factors for consumer acceptance [37]. This could also explain the oldest group's lower interest in pasta.

This study showed that the most important factors for bread are taste and flavour. This is supported by rising consumer interest for better and more authentic flavours [32,38]. Freshness and texture are other important factors and, according to the respondents, they are more important than other factors, such as health factors, being organic and origin. This is in line with other studies that have reported the importance of flavour and other sensory attributes [28,39]. It was also established that health factors are more important to older consumers than to younger ones [29], an observation that is supported by the results from this study where the oldest consumer group regard health aspects as significantly more important than the younger group.

It is reported that Sweden has a fairly high consumption of organic products [40,41], thus it was surprising that the current study has signalled that the younger respondents regarded a cereal being labelled "organic" less important than the older consumer groups. "Locally produced" has recently been shown as more important to the consumers than "organic farming" [22], suggesting that organic farming would require more land than conventional farming and, in that respect, contribute more to climate change [42].

The word cloud illustration in Figure 3 points out wholegrain as a popular type of bread, which is in line with Kyrø, et al. (2012) [26] who reported on the consumption of bread in the Scandinavian countries during the 1990s, showing that rye contributed the most to the whole-grain intake: in Denmark about 70%; in Sweden about 50%; in Norway only about 20%. Furthermore, the total whole-grain consumption among different Swedish consumer groups were as follows: white-bread consumers had a mean total intake of 38 g/day; whole-grain bread consumers reported 45 g/day [2]. This supports the current study findings that the participants reported crispbread as the most consumed type of bread. Likewise, sourdough bread was reported to be commonly consumed; more common than white-bread and toast-bread. It should be highlighted that the consumers in the current study were affiliated with universities and, therefore, might have had a higher awareness about the role of whole-grain and sourdough for human health. The potential of sourdough to obtain healthier cereal products is becoming increasingly known [43].

In the current study, and based on the above discussion, it is evident that age is a critical factor. For instance, younger consumers are more aware of heritage cereals and different varieties than older consumers. This high-level of awareness is reflected by younger consumers showing a greater interest in natural agricultural products [44]. The younger group was also more sensitive to price and significantly fewer young respondents were willing to pay more for heritage cereals compared to respondents in the older consumer groups, which could be explained by differences in economical levels. Similarly, Hwang [45] showed that older consumers are more willing to pay a higher price for food when they are motivated to do so. This is in line with the results in this study where the older consumers were willing to pay a higher price for products based on heritage cereals than younger consumers. The same pattern could be seen in the habit of purchasing bread at the bakery, which was more frequent among the older consumers.

Another important factor for the older consumer group was the health aspects. Kraus et al. [30] found that food health aspects are of greatest importance for women and older consumer groups in studies on functional food. Gender difference was more obvious when studying the importance of consuming wholegrain products and knowledge about the origin of the cereal. In these assessments it was found out that women considered those as more important factors compared to men. This is also in line with Kraus et al., whose studies show that women are more concerned about nutritional aspects and consuming natural products than men. Moreover, women consumed less muesli and breakfast cereals than men, yet ate more porridge compared to men.

Regarding the groups differing in academic background, it could be seen that the sites' geographies seemed to influence the awareness of varieties of heritage cereals. The site of the academic institute seemed to have an impact, mainly on the awareness of different types of heritage cereals, which correlates with a study showing a high awareness of the importance of regional products among consumers [46]. Further, the results indicate that respondents affiliated with SLU consumed more muesli and flakes compared to HKR affiliates, while for respondents from HKR the factors appearance, sourdough and freshness were of higher importance than for those from SLU. These differences and the differences in awareness between academic sites could be due to the fact that HKR has a multidisciplinary focus with research and teaching within many subjects. Therefore, it is possibly more open to influences from a broader number of different disciplines compared to the agricultural focus at SLU.

Limitations: It should be noted that a limitation of this study was the unbalanced sample sizes of the consumer groups. The gender group X (unidentified sex) consisted of only six consumers, thus this group was too small to imply any relevant results and was, therefore, kept out of analysis. It should as well be noted there was an uneven sample sizes regarding men and women, as well as consumers belonging to the different academic institutes. To compare the frequencies of groups, the percentages of frequencies were calculated. It should be noted that a larger sample size of women is common in consumer studies [39,47,48]. Further, it should be noted that the two participating universities differ in size, where SLU is substantially larger than HKR.

#### **6. Conclusions**

It could be concluded that most consumers are aware of heritage cereals, where dinkle/spelt is the most well-known variety. Other varieties such as emmer, Kamut and Oland wheat were known by approximately 50% of the consumers. The geographic location of the academic institutions seemed to influence the awareness of heritage cereal varieties. The focus on academic background seemed to have only minor influence on attitudes towards heritage cereals. Bread and pasta are the most consumed products and are also regarded as the most potential future products that could be based on heritage cereals. With regards to bread, the most important factor is taste and flavour, followed by freshness and texture. Cereal origin and health aspects are of importance, however, women are significantly more concerned about the origin of the cereal than men, while older consumers are more concerned about health aspects of cereals and cereal-based products. Older consumers are also significantly more willing to pay more for products based on heritage cereal than younger consumer groups.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2304-8158/9/6/742/s1, File S1: Consumer Questionnaire.

**Author Contributions:** The authors K.W., A.M., T.O. and K.G. contributed to the manuscript according to the following: Conceptualization, K.W., K.G., T.O. and A.M.; Methodology, K.W.; Software, K.W.; Validation, K.W., K.G., T.O. and A.M.; Formal Analysis, K.W.; Investigation, K.W.; Resources, K.G. and K.W.; Data Curation, K.W. and A.M.; Writing—Original Draft Preparation, K.W.; Writing—Review and Editing, K.W., K.G., T.O. and A.M.; Visualization, K.W.; Project Administration, K.G.; Funding Acquisition, K.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** Formas, the Swedish research council for sustainable development, project no 2018-02393.

**Acknowledgments:** Laboratory technician Per Magnusson at Kristianstad University (HKR) is acknowledged for the administration of the web-based questionnaire.

**Conflicts of Interest:** The authors declare no conflicts of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Millennials' Consumption of and Attitudes toward Meat and Plant-Based Meat Alternatives by Consumer Segment in Finland**

**Antti Knaapila 1, \* , Fabienne Michel 2 , Kirsi Jouppila 1 , Tuula Sontag-Strohm <sup>1</sup> and Vieno Piironen 1**


**Abstract:** Millennials are considered the key generation with regard to the consumption of plantbased meat alternatives via flexitarianism. This study sought to characterize millennials' consumer segments based on their consumption of and attitudes toward meat and meat alternatives. We conducted an online survey on the hedonic tones of the associations evoked by meat and meat alternatives, consumption of such foods, and diet-related attitudes among a representative sample of Finnish millennials (*N* = 546, 59% women, age 20–39 years). Some 41% of respondents regularly ate plant-based meat alternatives, while 43% had tried such foods. We divided the respondents into six segments based on the hedonic tones of their meat vs. meat alternatives associations. The segments differed in terms of their consumption of meat alternatives and the underlying reasons why, importance of meat in meals, and Meat Commitment Scale scores. The segment that reported much more positive associations with meat than meat alternatives (~14% of the respondents) may prove resistant to interventions intended to reduce meat intake, whereas the segment that displayed the most positive attitudes toward meat alternatives (~18%) did not eat much meat. Thus, the four middle segments (totaling ~68%), whose associations' hedonic tones were close to each other, may be the best targets for future interventions designed to reduce meat consumption through the use of meat alternatives. To conclude, introducing a simple segmentation allowed us to identify consumer segments with large potential to reduce meat consumption.

**Keywords:** acceptance; consumer segmentation; flexitarian; meat analogue; meat substitute; online survey; plant-based protein; sustainability; vegan; vegetarian

#### **1. Introduction**

The need for more environmentally sustainable alternatives to meat (and especially to red and processed meat) is increasing due to planetary boundaries (i.e., global biophysical limits for safe operating space in, e.g., climate change, biosphere integrity, land-system change, and freshwater use [1]) limiting the capacity to produce more meat for the increasing global population [2]. In addition, while meat is an important source of nutrients, especially protein, heavy meat consumption may have adverse effects on human health (for a review, see [3]). The EAT–Lancet Commission on Healthy Diets from Sustainable Food Systems stated that the "transformation to healthy diets by 2050 will require substantial dietary shifts, including a greater than 50% reduction in global consumption of unhealthy foods, such as red meat and sugar" [2]. This goal will likely prove difficult to achieve, as global meat consumption (both the average per capita and total consumption) continues to rise [3].

Food products that are made of protein-rich nonanimal sources intended to resemble meat and that are used instead of meat are often referred to as meat analog(ue)s, meat

**Citation:** Knaapila, A.; Michel, F.; Jouppila, K.; Sontag-Strohm, T.; Piironen, V. Millennials' Consumption of and Attitudes toward Meat and Plant-Based Meat Alternatives by Consumer Segment in Finland. *Foods* **2022**, *11*, 456. https://doi.org/10.3390/ foods11030456

Academic Editor: Han-Seok Seo

Received: 9 December 2021 Accepted: 1 February 2022 Published: 3 February 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

substitutes, or meat alternatives. In the literature, these terms are generally used synonymously [4], although their definitions do sometimes differ among authors. The term meat analogue has been commonly used in recent reports on the production of such products using extrusion technology [5–11]. For instance, Kumar et al. [12] defined a meat analogue as "a food product that approximates the aesthetic qualities and/or chemical characteristics of certain types of meat. These are made from non-animal protein and their appearance and smell are very much similar to meat". Dekkers et al. [13] considered functionality alongside sensory properties and defined meat analogues as "products that can replace meat in its functionality, being similar in product properties/sensory attributes, and that can also be prepared by consumers as if they were meat". Moreover, the terms meat analogues and meat substitutes are often used to refer to products that more closely resemble meat in terms of their sensory properties than meat alternatives, a term that is used in a broader sense to refer to alternatives to meat. For example, Elzerman et al. [14] defined meat substitutes as "products that were developed to be eaten instead of meat" (e.g., vegetarian sausages and steaks) and meat alternatives as "other products that are often eaten as protein source in vegetarian meals, such as pulses and nuts". However, Choudhury et al. [15] regarded plant-based meat alternatives as a "sustainable source of proteins that can match the taste and texture, color, and nutritional profile of specific types of meat". Based on the previously mentioned studies, it appears that a consensus has not yet been reached concerning the terminology for these products.

Meat intake can be reduced in many ways and with proteins derived from many sources: using conventional vegetarian foods (e.g., pulses), hybrid meat products (containing both meat and plant-based ingredients) [16], and meat alternatives. The most commonly used alternative protein ingredients originate from plants (especially soy, pea and other legumes, oilseeds, and wheat), fungi (mycoprotein), insects, and algae (macroalgae and microalgae) [4,17,18]. In addition, cultured meat (in vitro meat) is regarded as an alternative to meat from livestock [4,17]. To distinguish among the different protein sources and so render the utilized term more precise, the source of the protein is sometimes included, for example, in plant-based meat alternatives. This term has been used to refer to commercial products in several recent reports, including some consumer studies [4,15,19–23]. Likewise, we used the term plant-based meat alternatives in the present study because it focused on respondents' orientations specifically toward plant-based alternatives to meat.

Plant-based proteins appear to be the most widely accepted meat alternatives/alternative proteins from the perspective of consumers [23,24]. Gómez-Luciano et al. [25] investigated the willingness to purchase three types of meat alternatives (plant-based proteins, cultured meat, and insects) on the part of consumers from four countries with dissimilar economic developmental statuses (the United Kingdom, Spain, Brazil, and the Dominican Republic) and found plant-based proteins to be the most preferred option. Similarly, Lundén et al. [26] reported Finnish consumers to prefer plant-based ingredients when compared with ingredients of insect or microbial origin.

Importantly, modern meat alternatives are targeted not only toward vegans and vegetarians but also toward flexitarians [15]. According to Dagevos [27], "a flexitarian abstains from eating meat occasionally without abandoning meat totally". He concluded that flexitarians are not a homogeneous group that follow a strict diet; rather, they represent a middle category between consumers who regularly eat meat and those who fully abstain from it [27]. In the absence of a strict definition of what flexitarian exactly means, it is understandable that Dagevos's review found the proportion of flexitarians to vary widely across studies, ranging from 11% to 66% [27]. Regardless of this variation, the number of flexitarians is likely to be substantially higher than the number of those who totally abstain from eating meat. Indeed, vegetarians and vegans represent only a low percentage of consumers in most countries [28], accounting for ~5% of consumers in the United States (2018) [29], 2.5% in France (2018) [30], and ~2% in Finland (2017) [31]. Therefore, flexitarianism is likely to make a substantial contribution to reducing meat consumption at the population level. However, flexitarians are a heterogeneous and rarely studied

group [27]. Thus, further research on both flexitarians and prospective flexitarians is required to successfully implement strategies for reducing meat consumption [32].

The millennial generation (or millennials, who are also referred to as Generation Y) are young(ish) adults who are considered to be more knowledgeable and concerned about environmental issues than older generations [33,34]. Therefore, millennials have been the target group in recent studies concerning food sustainability [35,36]. While there is no widely accepted definition of millennials, they are often considered to be people who reached adulthood during the early 21st century, that is, the people who were born during the 1980s and 1990s [37]. Millennials also represent an important consumer group because many are presently the parents of young children, and the parents' role is essential in terms of mediating the food consumption habits of their children [38].

Meat alternatives have the potential to grow from being niche products into mainstream ones [39]. According to the Food Sector Report by Smart Protein project [40], in Europe, the sales value of plant-based food increased by ~50% from 2018 to 2020. Yet, while the sales of plant-based meat alternatives are growing rapidly, in the United States, for example, they accounted for only around 1% of the value of all retail meat sales in 2019 [15]. In 2017, based on a review of 38 articles (published in 2004–2016) concerning consumers' sustainable protein consumption, Hartmann and Siegrist [41] concluded that consumer awareness of the environmental impacts of meat production and consumers' willingness to reduce meat consumption were, on average, low. Nevertheless, the market for plant-based meat alternatives is evolving rapidly, and many new companies producing meat alternatives have been founded in recent years. In fact, according to Choudhury et al. [15], more than half of all companies producing meat alternatives were founded in the last 10 years [15].

Onwezen et al. [23] recently conducted a systematic review of studies on consumer acceptance of alternative proteins. They found that the main product-related motives/barriers with regard to the use of plant-based meat alternatives stemmed from ethical, environmental, health, nutritional, and sensory aspects, in addition to familiarity/previous experiences of the products. Furthermore, the main psychological factors of relevance to the acceptability of meat alternatives were consumers' attitudes and beliefs regarding the products as well as food neophobia [23].

It is important to note, however, that the drivers and barriers concerning the use of plant-based meat alternatives are not the same for everyone, which means that an intervention that works for one consumer segment may not be effective for a different segment [42]. Therefore, it should prove useful to achieve the meaningful segmentation of consumers and then to investigate the differences among the segments.

Consumers can be classified simply based on whether or not they eat meat alternatives. Hoek et al. [43] reported that the key barriers for nonusers of meat alternatives were unfamiliarity with the products and their lower sensory attractiveness when compared with meat. To make meat alternatives more attractive to nonusers, the authors recommended improving the sensory quality and resemblance to meat, rather than highlighting ethical arguments, because such arguments only motivated heavy users of meat alternatives. The resemblance to meat was also identified as a desirable feature for meat alternatives by Michel et al. [20]. This feature appears to be especially important for light users of meat alternatives, as the desire for similarity decreased with increasing consumption frequency in the study by Hoek et al. [43].

Consumers can also be segmented by means of a multivariate data analysis of their responses to a set of questions. For instance, Lacroix and Gifford [44] identified three consumer groups using a latent profile analysis: "meat-reducers", "moderate-hindrance meat eaters", and "strong-hindrance meat eaters". Furthermore, Lemken et al. [42] searched for clusters within consumer data from Germany and New Zealand using a latent class analysis and identified five clusters in each country (three clusters were common to both countries, while two were unique for each country). Recently, Götze and Brunner [45] segmented a sample of Swiss consumers into six segments via a hierarchical cluster analysis. While the consumer groups included exclusive meat-eaters and meat-avoiders, the majority were found to lie between those extreme segments. In Finland, Niva and Vainio [46] recently studied consumers' past, current, and intended future consumption of beef, plant-based protein products, and insect-based products. Using latent class analysis they identified five clusters of consumers, two of which (totaling ~46%) were characterized by consuming both beef and plant-based protein products. The findings of these studies are in accordance with the results of Dagevos [27] and confirm the existence of a remarkable proportion of flexitarians.

The present study sought to characterize the consumption of meat and plant-based meat alternatives as well as to provide in-depth insights into the underlying motives in this regard among various consumer segments of millennials. Based on this knowledge, we further aimed to draw conclusions regarding the potential of the segments to replace meat with meat alternatives in their diet. To achieve these aims, we conducted an online survey among a representative sample of Finnish millennials. In Finland, plant-based meat alternatives are widely available in grocery stores (brands including PulledOats, Härkis, and Beanit), making it reasonable to run this survey in the country. The criteria for the different consumer segments were defined in such a way as to allow other researchers to replicate the segmentation in future studies.

#### **2. Materials and Methods**

#### *2.1. Overview*

We conducted an online survey that was jointly designed by all the authors, initially in English. The text of the survey was then translated into other languages as required to be used in Germany, Finland, France, and the United Kingdom. The first results of the survey conducted in Germany, France, and the United Kingdom have been reported by Michel et al. [47]. Here, we report results based on data collected in Finland. These data are being reported separately because in Finland we studied the millennials whereas in the other countries respondents' age range was wider (20–69 years, [47]) and because the questionnaire used in Finland differed somewhat from the questionaries used in the other countries. More specifically, the Finnish version included most but not all the parts of the original survey (e.g., the questions featuring pictures were excluded). The English version was translated into Finnish by four of the authors, who were all native Finnish speakers (A.K., K.J, T.S-S., and V.P.), and a research assistant from the University of Helsinki.

#### *2.2. Data Collection*

The required data were collected from millennials who lived in Finland. For this study, we decided to define millennials as people who were aged from 20 to 39 years at the time of the data collection (i.e., born in 1980–1999). Thus, we used age as the inclusion criterion for the study.

The nationality and ethnicity of the respondents were not probed in the survey. However, we assumed that virtually all the respondents were Finnish, as the invitations to the survey were only sent to people living in Finland and the text of the survey was solely in Finnish.

We employed a market research company (Taloustutkimus Ltd., Helsinki, Finland) to conduct the data collection in order to achieve a representative sample of millennials from among the general population of Finland. The company had its own online panel of preregistered volunteers, who were regularly invited to respond to surveys. Taloustutkimus was aware of the demographics of the registered panelists and, therefore, could invite defined samples from the panel to participate in survey studies. We provided the questions and response options for our survey to the company, which then collected responses from its online panel over the course of a week (20–26 November 2019) and provided us with data from 550 individuals.

The key concept featured in the survey was "meat alternative". However, at the time of the study, there was no established translation of this term in Finnish. We decided to

translate "meat alternative" into Finnish as "kasviproteiinituote", although the Finnish term refers to meat alternatives made solely of plant-based proteins (the Finnish words "kasvi", "proteiini", and "tuote" denote "plant", "protein", and "product", respectively) and so excludes other kinds of meat alternatives (such as those made of microbial proteins, whey, insects, or cultured meat).

The survey included both validated scales described in the prior scientific literature and additional questions designed specifically for this study. Lists of the questions/scales from the survey and their response options are presented in Tables 1 and 2, wherein they are grouped thematically. Table 1 includes questions related to diet and hedonic tone concerning meat and meat alternatives and their consumption, as well as drivers and barriers associated with their consumption. Table 2 contains questions derived from published scales measuring attitudes and food-related behavior. The text of the survey in Finnish is available in the (Supplementary Materials Table S1). The survey also included a few questions that were beyond the scope of the present study and, thus, are not reported here. The age and gender of the respondents were provided by Taloustutkimus from its registry.

**Table 1.** Survey questions 1–9: specific questions on diet, education, hedonic tone, consumption of meat and meat alternatives, reasons for use/nonuse, and importance of meat in meals and for guests.


<sup>1</sup> The Finnish translation of these questions is available in the (Supplementary Materials Table S1). <sup>2</sup> Education was the only demographical factor probed in the survey. The age and gender of the respondents were available from the register of the utilized market research company. <sup>3</sup> The order of presentation of Q3 and Q4 was randomized.

The questions related to the first associations with meat (Q3) and meat alternatives (Q4) were presented in a randomized order for each respondent. We placed these questions at the beginning of the survey in an effort to minimize the influence of the other items on the answers. After Q3 and Q4, we provided a definition of meat alternatives to be considered throughout the rest of the survey. It read as follows: "For the following questions, we refer to meat alternatives as commercially available plant-based convenience foods that can be used instead of meat. Examples are vegetarian sausages, veggie patties, or plant-based minced 'meat'". The remaining questions (Q5–Q14) were then presented.


#### **Table 2.** Survey questions 10–14: multi-item scales.

<sup>1</sup> The Finnish translation of these questions is available in the (Supplementary Materials Table S1). <sup>2</sup> A seven-point scale was used instead of the original six-point scale (from "Don't agree at all" [1] to "Fully agree" [6]) used by Dohle et al. [48]. <sup>3</sup> The Diet-Related Health Consciousness Scale by Dohle et al. [48] was partly based on the items from the Health Consciousness Scale by Schifferstein and Oude Ouphuis [53]. <sup>4</sup> One of the three scales developed by Lindeman and Väänänen [49], namely the Ecological Welfare Scale (including the subscales for Animal Welfare and Environment Protection), was used in this study. <sup>5</sup> The original three-item Natural Content Scale (part of the Food Choice Questionnaire) was complemented with a fourth item, " . . . is as little processed as possible".

#### *2.3. Data Analysis*

First, we cleaned the data of obvious errors. During the data cleaning, 4 out of 550 individuals (0.7%) were removed from the dataset due to providing inconsistent or otherwise doubtful responses. Therefore, we included answers from 546 respondents in our further analyses.

Second, the composite scores for the published multi-item scales (Table 2) were calculated according to the instructions in the original sources [48–52,54]. Thanks to the use of an electronic questionnaire, the data included no missing values (i.e., no missed responses). Cleaned data (*N* = 546) with the calculated scores are available in the (Supplementary Materials Table S2).

The data were analyzed statistically using the IBM SPSS Statistics version 27 software package (IBM, Armonk, NY, USA). We applied descriptive and analytical statistics to the data, and we used α = 0.05 as the criterion for statistical significance. The independent samples *t*-test, one-way and two-way analysis of variance (ANOVA), and Pearson's chisquared tests were also used as appropriate. The answer categories "Daily" and "More than once per day" for the questions concerning the eating frequency of both meat and meat alternatives (Q5) were combined into one category named "Daily" to increase the clarity of the results. This category implies eating a food item at least once per day.

Essentially, we classified the respondents into six groups based on the hedonic tone (valence: negative–positive) of their first associations with meat (Q3) and meat alternatives (Q4), as described below (in Section 3.2). In this paper, we refer to these groups of respondents as (consumer) segments.

We employed a two-way ANOVA using the respondents' gender and consumer segment as fixed factors (independent variables) in order to study the quantitative variables as appropriate. A full factorial model was run first and the significance of the gender×segment interaction was observed. If the interaction was nonsignificant, the interaction term was left out of the model and the results were reported based on the model including only the

main effects. Furthermore, if the main effect of the segment was significant, Tukey's post hoc test was applied to reveal which of the segments differed from the others.

#### **3. Results**

#### *3.1. Demographics and Diet*

The data (total *N* = 546) included more responses from women (322; 59.0%) than men (224; 41.0%). In comparison, according to official statistics concerning Finland [55], the gender distribution among 20–39-year-old Finns (at the end of 2019) was 48.4% women and 51.5% men [56].

The mean age of the respondents was 31.2 years and the age distribution was rather evenly distributed across 20–39 years (with the range defined by the inclusion criterion). The women respondents were, on average, a little younger than the men (30.6 vs. 32.0 years, respectively; *t*(504) = 3.04, *p* = 0.002). By contrast, the respondents' education, as measured by the number of years (including both school and professional education), did not differ between the genders (16.3 vs. 15.9 years, respectively; *p* > 0.05).

Among all the respondents, about two-thirds (67.2%) identified themselves as omnivores (agreeing with the statement "I eat all animal products"), while about one-third (32.8%) followed a diet that limited the consumption of animal products in one way or other. Following a limited diet in terms of the consumption of meat/animal-based products was more prevalent among the women than the men. Indeed, nearly half of the women (42.5%) but only about a fifth of the men (18.8%) followed a non-omnivorous diet, that is, identified themselves as either flexitarian, pescetarian, vegetarian, or vegan (Pearson's chi-square = 33.9, *p* < 0.001) (Table 3).


**Table 3.** Respondents' diet by gender.

Furthermore, approximately two-thirds of the non-omnivores (66.5%, corresponding to 21.8% of all the respondents) were either flexitarians or pescetarians, while the remaining third of the non-omnivores (33.5%, corresponding to 11.0% of all the respondents) were either vegetarians or vegans.

#### *3.2. Hedonic Tones of the First Associations with Meat and Meat Alternatives*

The hedonic tone (valence on a scale ranging from −5, "extremely negative", to 5, "extremely positive") of the first associations (words, images, or thoughts) spontaneously evoked when thinking about meat was, on average, close to neutral (1.1). Likewise, the average hedonic tone of the first associations with meat alternatives was close to neutral (1.0). No statistically significant difference was observed between the values (*t*(1090) = 0.61; *p* = 0.542). However, the individual differences in the ratings of the hedonic tones were large (SD 3.4 and 3.1 for meat and meat alternatives, respectively), implying that not all the respondents rated their associations as neutral.

#### 3.2.1. Hedonic Tone by Diet and Gender

The two-way ANOVA involving diet and gender as fixed factors showed no significant diet×gender interaction in terms of the hedonic tone of the first associations with either meat (F(4,536) = 1.52; *p* = 0.195) or meat alternatives (F(4,536) = 0.75; *p* = 0.560). This implied that within a given diet group, both genders provided similar ratings.

Diet had a significant main effect on the hedonic tones of the first associations evoked by both meat (F(4,540) = 191.1; *p* < 0.001) and meat alternatives (F(4,540) = 44.1; *p* < 0.001). Similarly, gender had a significant main effect in the case of both meat (F(1,540) = 7.6; *p* = 0.006) and meat alternatives (F(1,540) = 5.9; *p* = 0.015). The omnivores and men rated their first associations with meat as more positive (and those with meat alternatives as more negative) than the non-omnivores (i.e., flexitarians, pescetarians, vegetarians, and vegans) and women, respectively (Table 4).


**Table 4.** Hedonic tones of first associations with meat and plant-based meat alternatives (rated on a scale from −5 to 5) by diet and gender.

<sup>1</sup> The means among the diet groups (within a column) not sharing a common letter are significantly different (Tukey's test, *p* < 0.05).

Diet appeared to more clearly influence the respondents' hedonic responses to their first associations with meat than their first associations with meat alternatives. Although the overall means of the hedonic tones concerning meat and meat alternatives were similar, the difference between the means in the most extreme diet groups in terms of the hedonic tone associated with meat was 7.1 points (from −4.2 in vegans to 2.9 in omnivores), while it was only 4.0 points in the case of meat alternatives (from −0.1 in omnivores to 3.9 in vegans). Among the non-omnivorous diet groups, significant differences were observed in the average hedonic tone associated with meat but not that associated with meat alternatives (Table 4).

We observed a clear negative correlation between the hedonic tones associated with meat and meat alternatives, although the correlation was not strong (Pearson's r = −0.55, *p* < 0.01). Among the omnivores (the largest diet group) in particular, there was wide variation in the hedonic tone associated with meat alternatives (SD 2.9), although the mean was close to zero (neutral). Some omnivores may have had positive associations with both meat and meat alternatives, or alternatively, they may have regarded both neutrally. This led us to assume that it could prove useful to classify the respondents into segments based on the hedonic tones associated with both meat and meat alternatives (instead of using the hedonic tone associated with either meat or meat alternatives).

#### 3.2.2. Segmentation of the Respondents

We cross-tabulated the ratings of the hedonic tones of the first associations with meat and meat alternatives to identify potential clusters of respondents that could be used as consumer segments in further analyses. Indeed, a visual inspection of the crosstab suggested that the hedonic responses were clustered, not evenly distributed.

We identified six clusters, which we defined and labeled as follows: The most obvious clusters existed in the upper left corner of the crosstab (those respondents who had very positive associations with meat alternatives (Ma) but negative associations with meat, labeled "MaPos" and marked with dark green in Figure 1) and the lower right corner

(those who had very positive associations with meat but negative associations with meat alternatives, labeled "MeatPos" and marked with red in Figure 1). Furthermore, between these two extreme clusters in the corners, there were groups of respondents who slightly or moderately preferred their associations with meat (labeled "MeatPref" and marked with orange in Figure 1) or meat alternatives (labeled "MaPref" and marked with light green in Figure 1). However, there was also a cluster of respondents who reported positive associations with both meat and meat alternatives (labeled "BothPos" and marked with yellow in Figure 1). Finally, there was a cluster of respondents who did not report positive associations with either meat or meat alternatives, instead rating the associations with both as neutral or even slightly negative (labeled "NoPos" and marked with light grey in Figure 1).

**Figure 1.** Cross-tabulation of the hedonic tones (valence, on a scale from −5 to 5) of the first associations evoked by meat and plant-based meat alternatives and classifying the respondents into six consumer segments (marked with different colors). The numbers in the cells denote the counts of individual respondents who gave the respective combination of responses. Counts ≥10 are marked in bold to highlight the clustering (total *N* = 546 individuals).

The definition, size, and gender distribution of the formed consumer segments are summarized in Table 5. The size of the segments ranged from 58 (10.6%) to 129 (23.6%) individuals. The percentage of women in a segment increased with an increasing preference for meat alternatives (Table 5). By contrast, no difference in age (F(5,540) = 1.6; *p* = 0.158) or number of years in education (F(5,540) = 1.2; *p* = 0.314) was observed between the segments.

The omnivores represented the largest fraction in all the segments, except for the segment most positive with regard to meat alternatives (MaPos). Unsurprisingly, the segments that reported the associations with meat to have relatively more positive hedonic tones (MeatPos and MeatPref) consisted almost exclusively of omnivores. Yet, more than half of the respondents in the segments that did not exhibit a clear difference in terms of the hedonic tones (BothPos and NoPos) were also omnivores. Moreover, the omnivores even represented the largest diet group in the segment that reported a higher hedonic tone with regard to meat alternatives (MaPref), although this segment also consisted of a remarkable fraction of flexitarians and pescetarians (Table 6). The dominance of the omnivores in almost all the segments can be explained by the fact that the omnivores were also the overall largest diet group (67.2% of all respondents).


**Table 5.** Consumer segments based on the hedonic tones of the first associations with meat and plant-based meat alternatives.

<sup>1</sup> Values of the prevailing gender in a segment are highlighted in bold. <sup>2</sup> Relative size of a segment out of all 546 respondents.

**Table 6.** Diet by consumer segment.


<sup>1</sup> Consumer segments formed based on the hedonic tones of the first associations with meat and plant-based meat alternatives (see Figure 1 and Table 5). <sup>2</sup> Values of the largest diet group in a segment are highlighted in bold. Note that the majority of all respondents (67.2%) were omnivores.

#### *3.3. Consumption Frequency of Meat and Meat Alternatives and the Underlying Reasons Why* 3.3.1. Consumption

Meat, including various meat products (but not fish), was consumed on a daily basis by a third of the respondents (33.5%). By contrast, a fifth (20.5%) of the respondents reported eating meat never or only rarely. Notably, the remainder, that is, almost half of the studied millennials (46.0%), reported sometimes eating meat but abstaining from it at least one day per week. As expected, the segments that reported their associations with meat to have a more positive hedonic tone (Table 5) also consumed meat more frequently (Figure 2a).

**Figure 2.** Use frequencies of (**a**) meat (pork, poultry, beef, ham, sausages, etc.) and (**b**) plant-based meat alternatives (including vegetarian patties, soy, tofu, etc.) by gender and consumer segment. The number of individuals in a group is given in parentheses. For details concerning how the respondents were classified into segments, see Figure 1 and Table 5.

Plant-based meat alternatives were eaten daily by only about 11% of the respondents, although almost half of the respondents (45.5%) consumed them at least once a week. About two-thirds of the millennials (68.9%) ate meat alternatives at least once a month, whereas about one-third (31.0%) ate them rarely or never. As in the case of meat, the hedonic tone of the first associations with meat alternatives was reflected in how often such products were consumed (Figure 2b). These findings suggest that the hedonic tones of the first associations with meat and meat alternatives could be used to predict people's consumption of these food categories.

Next, we asked how many respondents consumed both meat and meat alternatives. Some overlap in terms of the consumption of these foods was expected because, in the case of both meat and meat alternatives, the majority of respondents reported eating them at least occasionally. In addition, we expected that some respondents consumed meat alternatives in an attempt to reduce their meat consumption (while not totally abstaining from eating meat), as 12.3% identified themselves as flexitarians (Table 3) and almost a

quarter (23.6%) reported positive hedonic tones with regard to the associations with both meat and meat alternatives (Table 5).

To investigate this issue, we cross-tabulated the consumption frequencies of meat and meat alternatives. This confirmed that almost half of the respondents (48.6%) ate both meat and meat alternatives at least once a month. Only meat (no meat alternatives) was eaten by 31.0%, while only meat alternatives (no meat) were eaten by 20.4% of the respondents. Notably, about a fifth of the respondents (20.4%) regularly ate (at least once a week) both meat and meat alternatives (Figure 3). The consumer segment that reported positive associations with both meat and meat alternatives (BothPos) represented the largest group among those who consumed both meat and meat alternatives at least once a month (37.7%) and those who consumed them on a weekly basis (36.0%).


**Figure 3.** Cross-tabulation of the consumption frequencies of meat (pork, poultry, beef, ham, sausages, etc.) in columns and plant-based meat alternatives (including vegetarian patties, soy, tofu, etc.) in rows. The percentages in the cells denote the proportion of respondents who responded with the combination represented by that cell (out of the total *N* = 546 respondents). Among all the respondents, 31.0% (red cells) consumed only meat, 20.4% (green cells) consumed only meat alternatives, and 48.6% (blue cells) consumed both meat and meat alternatives.

#### 3.3.2. Reasons for Eating and Not Eating Meat Alternatives

The question about why a respondent ate or did not eat plant-based meat alternatives was connected to a separate simple question concerning the consumption of meat alternatives. We first asked, "Do you eat plant-based meat alternatives?" (Q6), which had three response options. If the answer to Q6 was "Yes, on a regular basis", we then asked, "Why do you eat plant-based meat alternatives regularly?" (Q7a). If the answer to Q6 was "No" or "I have sampled meat alternatives but do not eat them on a regular basis", the next question was "Why do you not eat plant-based meat alternatives regularly?" (Q7b). Both questions concerning the reasons for eating/not eating meat alternatives were check-all-that-apply (CATA)-type questions with 7 (Q7a) and 12 (Q7b) predefined response options.

Approximately 4 out of 10 respondents (40.8%) reported eating plant-based meat alternatives on a regular basis. The regular consumption of meat alternatives was more common among the women (47.8%) than the men (30.8%) (X<sup>2</sup> (2) = 17.6; *p* < 0.001). The proportion of regular users of meat alternatives varied widely across the consumer segments (from 2.6% for MeatPos to 92.7% for MaPos) (Table 7).


**Table 7.** Overall consumption of plant-based meat alternatives by gender and consumer segment.

<sup>1</sup> The percentages (%) within a column indicate the proportion of a group who responded with a given answer. The most common response within each group is highlighted in bold. <sup>2</sup> The total number of individuals in a group is given in parentheses.

Environmental reasons were the most frequently cited motive for the regular consumption of meat alternatives among all the respondents (80.7%), followed by animal welfare reasons (64.6%) and health reasons (53.8%) (Table 8). There were some differences in motives between the genders. Notably, a larger proportion of women (59.7%) than men (33.3%) selected "I like the taste" as a reason for regularly eating meat alternatives.

**Table 8.** Reasons for eating plant-based meat alternatives regularly: percentage of regular eaters (40.8% of all respondents) offering a specific reason by gender and consumer segment.


<sup>1</sup> The percentages within a column indicate the proportion of a group who responded with a given answer (multiple answers possible). The response options were sorted from the most to the least frequent response among all the respondents. The two most frequent responses within each group are highlighted in bold. <sup>2</sup> The total number of regular eaters in a group is given in parentheses. <sup>3</sup> n/a, not applicable. Because only 2 out of 78 (2.6%) respondents in the MeatPos segment ate meat alternatives regularly, their responses are not shown.

The consumer segments differed in terms of their motives for eating meat alternatives. Environmental reasons were among the two most commonly mentioned reasons in all the segments, while they were the top motive for the MaPos, MaPref, and NoPos segments. Interestingly, the most frequently reported motive for the MeatPref and BothPos segments was "I like trying new foods".

Among those respondents who did not consume meat alternatives regularly, the most commonly cited reason for this behavior was "I do not like the taste of meat alternatives" (56.7%), followed by "Meat alternatives are too expensive" (51.4%) (Table 9). These two reasons were the top two reasons given by both the women and the men. However, in terms of the women, the third most commonly mentioned reason for not eating meat alternatives regularly was "I do not know how to cook meat alternatives", whereas for men it was "Meat alternatives are not a good replacement for meat".


**Table 9.** Reasons for not eating plant-based meat alternatives regularly: percentage of those who did not eat meat alternatives regularly (59.2% of all respondents) offering a specific reason by gender and consumer segment.

<sup>1</sup> The percentages within a column indicate the proportion of a group who responded with a given answer (multiple answers possible). The response options were sorted from the most to the least frequent response among all the respondents. The two most frequent responses within each group are highlighted in bold. <sup>2</sup> The total number of regular eaters in a group is given in parentheses. <sup>3</sup> n/a, not applicable. Because only 7 out of 96 (7.3%) respondents in the VegePos segment did not eat meat alternatives regularly, their responses are not shown.

The main reasons for not eating meat alternatives regularly also differed among the consumer segments. For the segments that reported a less positive hedonic tone with regard to meat alternatives (Meat Pos and Meat Pref), the top reason was clearly "I do not like the taste of meat alternatives". For the segments that reported a positive attitude toward meat alternatives (BothPos and MaPref) but who still do not eat such products regularly, the two most frequently mentioned reasons were "Meat alternatives are too expensive" and "I do not know how to cook meat alternatives". While the frequency of citing various reasons varied considerably among the segments in general, the reason "Meat alternatives are too expensive" was mentioned by a somewhat similar proportion of individuals in all the segments (42.1–56.1%).

#### *3.4. Status of Meat in Meals*

#### 3.4.1. Importance of Meat in Main Meals

We asked the respondents "How important do you consider meat to be for your main meal in the following situations?", that is, for a "typical weekday", "weekend", and "at a restaurant" (Q8, 7-point scale ranging from 1, "Not important at all", to 7, "Very important"). The mean rating for the importance of meat in a main meal was close to the midpoint of the scale and similar for the typical weekday (3.6), weekend (3.9), and at a restaurant (4.0) options.

The women considered meat in all of the given situations to be less important than the men did (indicating the significant main effect of gender). The mean importance ratings given by the women and men were 3.1 vs. 4.4 for meat in a main meal on a typical weekday (F(1,539) = 29.7; *p* < 0.001), 3.4 vs. 4.6 on the weekend F(1,539) = 25.9; *p* < 0.001), and 3.4 vs. 4.8 at a restaurant F(1,539) = 17.6; *p* < 0.001), respectively.

The consumer segments varied greatly in terms of their responses here. The main effect of the segment was significant for meat on a typical weekday (F(5,539) = 121.9; *p* < 0.001), on the weekend (F(5,539) = 127.2; *p* < 0.001), and at a restaurant (F(5,539) = 118.9; *p* < 0.001). As expected, the MeatPos segment rated the importance of meat in all the studied situations the highest, while the MaPos segment rated it the lowest.

#### 3.4.2. Difficulty of Thinking of a Vegetarian Main Course for Invited Guests

The responses to the question "How difficult is it for you to think of a vegetarian main course for invited guests?" (Q9, rated on an 11-point scale from 0, "Very easy", to 10, "Very difficult") varied widely among the respondents. The women regarded it as easier to think of a vegetarian main course for guests than the men (2.8 vs. 5.0, indicating a significant main effect for gender (F(1,539) = 18.5; *p* < 0.001). Similarly, the consumer segment had a significant main effect on the responses to this question (F(5,539) = 58.2; *p* < 0.001). As expected, among the various segments, the MaPos segment rated it the easiest to think of a vegetarian main course for guests (0.3), followed by the MaPref (1.9), BothPos (3.5), NoPos (3.7), MeatPref (5.6), and MeatPos (7.5) segments (the means of all the segments, except those of the BothPos and NoPos segments, differed from each other according to Tukey's test, *p* < 0.05).

#### *3.5. Diet-Related Attitudes*

Finally, we analyzed whether the genders and consumer segments differed in terms of their responses to the selected multi-item scales. All the scales showed good internal consistency as measured using Cronbach's alpha: diet-related health consciousness (0.77), ecological welfare concerns (0.90), importance of the natural content of foods (0.90), meat commitment (0.97), and food neophobia (0.89).

The women scored higher than the men in relation to the Ecological Welfare Scale (3.2 vs. 2.8, F(1,539) = 16.2; *p* < 0.001) and Natural Content Scale (2.8 vs. 2.6, F(1,539) = 22.1; *p* < 0.001). By contrast, the women scored lower than the men in terms of the Meat Commitment Scale (2.8 vs. 4.2, F(1,539) = 37.3; *p* < 0.001). No significant main effect of gender was observed with regard to scores for Health Consciousness Scale or Food Neophobia Scale (Table 10).

**Table 10.** Scores for the diet-related attitude scales by gender and consumer segment (means (M) and standard deviations (SD)).


<sup>1</sup> The total number of individuals in a group is given in parentheses. <sup>2</sup> The main effect of gender was significant for these variables (ANOVA, *p* < 0.05). <sup>3</sup> The main effect of segment was significant for all the variables (ANOVA, *p* < 0.05). The letters denote results of the Tukey's test, that is, which of the mean values (segments) are statistically different and which not (in a given variable, i.e., within a line). Lowest mean value has been marked with "a", next lowest with "b" and so on. The means between the segments (within the same row) not sharing a common lowercase letter differed (Tukey's test, *p* < 0.05).

According to the two-way ANOVA, the consumer segment had a significant main effect on the scores for all the attitude scales: Health Consciousness (F(5,539) = 6.6; *p* < 0.001), Ecological Welfare (F(5,539) = 19.3; *p* < 0.001), Natural Content (F(5,539) = 4.0; *p* = 0.001), Meat Commitment (F(5,539) = 179.3; *p* < 0.001), and Food Neophobia (F(5,539) = 3.5; *p* = 0.004). Tukey's test confirmed these results and classified the segments into different homogeneous subsets for all the variables except food neophobia. In the case of food neophobia, Tukey's test classified all the segments into the same homogeneous subset (*p* = 0.058) (Table 10).

The differences between the segments were the most obvious when it came to meat commitment and ecological welfare: the MaPos and MaPref segments were less committed to meat and more concerned about ecological welfare than the MeatPos and MeatPref segments. The scores from the scales measuring health consciousness, naturalness, and food neophobia did not reveal any systematic differences across the segments.

#### **4. Discussion**

#### *4.1. Hedonic Tones of the Associations with Meat and Meat Alternatives*

The millennials' mean hedonic tones evoked by the first associations with meat and meat alternatives were similar and slightly positive (1.1 vs. 1.0, respectively, on a scale ranging from −5 to 5). This finding conflicts somewhat with the findings of the studies reviewed by Onwezen et al. [23], who concluded that acceptance of alternative proteins was relatively low when compared with acceptance of meat. Our finding that the millennials' associations with meat alternatives were, on average, as positive as their associations with meat may reflect millennials' greater concern and knowledge regarding environmental issues when compared with older generations [33,34]. The millennials' orientation toward plant-based diets was also supported by the higher proportion of vegans and vegetarians in the present study (total 11.0%) when compared with the general population of 25–74-year-old Finns in 2017 (1.8%) [31] and 18–79-year old Finns in 2018 (6.7%) [46].

As expected, the women reported, on average, more positive associations with plantbased meat alternatives (and less positive associations with meat) than the men. However, the women rated their associations with meat alternatives as more positive than those with meat (1.5 vs. 0.4), which suggests that millennial women are, at least in countries such as Finland, a potential target group for plant-based meat alternative products.

It was also expected that the followers of diets that limited the consumption of meat (i.e., flexitarians, pescetarians, vegetarians, and vegans) would report negative associations with meat and positive associations with meat alternatives. Our findings confirmed that the vegetarians and vegans reported positive associations with meat alternatives more frequently, as did the flexitarians, although some plant-based meat alternatives on the market may resemble meat closely (to appeal to flexitarians). Interestingly, only 3.4% of those who did not regularly eat meat alternatives mentioned "Meat alternatives are too much like meat" as a reason why. This suggests that the plant-based meat alternative products sold in Finland do not resemble meat to such an extent that vegetarians are put off.

The individual variation in the hedonic tones of the associations with meat and meat alternatives was wide, which formed the basis for our segmentation procedure. As meat alternatives are by definition designed to replace meat in a person's diet, we assumed that it would be useful to study attitudes toward both meat and meat alternatives (not only toward one of them). Indeed, when cross-tabulating the ratings for the hedonic value of the associations with meat vs. meat alternatives, we observed that the ratings for meat and meat alternatives were not always simply opposite values (positive association with meat combined with negative association with meat alternatives, or vice versa), as some respondents reported positive (or neutral) associations with both meat and meat alternatives.

#### *4.2. Consumption of Meat and Meat Alternatives and the Underlying Reasons Why*

Both gender and consumer segment were associated with the consumption of meat and meat alternatives. The women and the consumer segments that reported more positive associations with meat alternatives (MaPref and MaPos) ate meat alternatives more frequently than the men and the consumer segments that reported more positive associations with meat (MeatPref and MeatPos). The men's greater preference for meat was in line with

the findings of prior research [46,57,58]. In Finland, according to the National FinDiet 2017 Survey [59], even 79% of men but only 26% of women ate more red and processed meat than the national guidelines recommend (500 g/week [60]).

The consumption of meat did not exclude the consumption of meat alternatives. This observation is consistent with the finding by Götze and Brunner [45] that meat alternatives can serve as a complementary component in one's diet. In a survey by Smart Protein project conducted in 2021 in adult consumers of 10 European countries (Austria, Denmark, France, Germany, Italy, Netherlands, Poland, Romania, Spain, UK), on average, 30% of the respondents identified themselves as flexitarians [24]. In the present study, around half of the respondents (48.6%) ate both meat and meat alternatives at least once per month. Thus, they can be referred to as flexitarians in a broad sense. However, a much lower proportion of respondents (12.3%) actually identified themselves as flexitarians. This could be because the description of a flexitarian in our questionnaire was strict ("I only rarely eat meat") and because the concept of flexitarian may still be largely unfamiliar to the general public. Nevertheless, the fraction of regular users of meat alternatives and those who had at least tried them totaled 84.0% in the present study (mean age 31.2 years). A slightly lower percentage (76%) of somewhat older respondents (mean age 57.4 years) had tried meat alternatives in the study by Götze and Brunner [45].

Environmental reasons represented the top motive for eating meat alternatives regularly. Among the women (and the respondents overall), the second most commonly cited motive was animal welfare, whereas among the men it was health reasons. A similar set of reasons, that is, "ecological welfare" and "health" (together with "sensory appeal"), were found to be the top food choice motives for using meat substitutes in the study by Hoek et al. [43]. These results suggest that many consumers regard meat alternatives as healthy. However, the nutritional value of novel plant-based meat alternatives may not always be as high as thought, because some products can, for example, contain high amounts of saturated fat and sodium. For further discussion on nutritional aspects of meat alternatives, see the review by Tso et al. [61] and commentary by Tso and Forde [62]. Of course, the quality of the diet as a whole is more important than its single items, also when considering replacement of animal-based foods in a diet [63]. For example, results from a clinical intervention study by Päivärinta et al. [64] indicated that replacing part of the animal-based proteins with plant-based proteins in a Nordic diet increased fiber intake, improved fat quality, and benefited blood lipoprotein profile.

In the present study, the drivers of consumption differed between the consumer segments, similar to the situation in previous studies [43,65]. Interestingly, for the MeatPref and BothPos segments, the most frequently given reason for regularly eating meat alternatives was "I like trying new foods". Although food neophobia (i.e., reluctance to try new foods) has been frequently identified as one of the barriers to the consumption of alternative proteins [23,46,66], it may be more important in relation to certain other kinds of meat alternatives, such as insects and cultured meat, than plant-based meat alternatives [23]. Moreover, in the present study, the consumer segments did not differ significantly (according to Tukey's test) in terms of their Food Neophobia Scale scores.

With regard to the barriers to consumption, the most commonly cited reasons for not eating meat alternatives regularly were "I do not like the taste of meat alternatives" and "Meat alternatives are too expensive". Taste being given as a reason for not eating meat alternatives is consistent with previous findings by, for example, Hoek et al. [43] (for a review, see [67]). Similarly, price was identified as a top barrier toward eating plant-based products in flexitarians in the survey by Smart Protein project [24]. Likewise, price being given as a barrier is in line with the conclusion by Michel et al. [20] that meat alternatives must be offered at competitive prices if they are to have a good chance of replacing meat. However, the frequencies of citing reasons differed considerably between the consumer segments. For example, among those who did not use meat alternatives regularly despite reporting relatively positive associations with them (from the MaPref and

BothPos segments), one of the most frequently mentioned reasons for nonuse was "I do not know how to cook meat alternatives".

#### *4.3. Consumer Segments*

In contrast to the present study, Lemken et al. [42] and Niva and Vainio [46] used latent class analysis and Götze and Brunner [45] hierarchical cluster analysis and multiple variables to segment consumers from different countries (Germany/New Zealand, Finland, and Switzerland, respectively) and studied adults of all ages. Despite the clear differences between these studies and the present investigation, they all ended up with a similar number of segments (5–6/population) with comparable features. All four studies identified a consumer cluster firmly oriented toward eating meat. Lemken et al. [42] termed the consumer group resembling our "MeatPos" segment the "meat only" cluster; Niva and Vainio [46], "established beef lovers"; and Götze and Brunner [45], the "uncompromising meat-eaters". Similarly, all the authors identified a segment strongly devoted to meat alternatives/legumes. The majority of individuals in the former type of segment were men, while the majority in the latter were women [42,45,46].

Most respondents in the present study (68.1%), similar to the situation in the studies by Lemken et al. (55.7% in Germany and 57.3% in New Zealand) [42], Niva and Vainio (53.8%) [46], and Götze and Brunner (67.6%) [45], were classified into the middle groups/segments, whose attitudes toward meat/meat alternatives and/or their consumption were not as extreme as those in the two segments described above. The middle segments arguably exhibit the highest potential to reduce their meat consumption by replacing it with meat alternatives. The segments with the strongest orientation toward meat may prove resistant to interventions intended to reduce meat consumption, while the segments that report the strongest avoidance of meat may not need to reduce their consumption. Therefore, the middle segments could be the best targets for interventions aiming to reduce meat consumption with the help of plant-based meat alternatives.

#### *4.4. Limitations*

The present study focused on millennials (20–39-year-old individuals). No respondents from other age groups were studied. Thus, we cannot directly compare millennials to consumers from other generations in the same population. Likewise, we only included respondents from one country/culture (Finland) in our study. However, we compared our results with those of relevant prior studies conducted in other countries and with wider age ranges of respondents [42,45]. Furthermore, we have allowed access not only to our results but also to our questionnaire (Table S1) and data (Table S2) to enable other researchers to utilize them in future studies.

Most of the questions in our survey were derived from published and validated multiitem scales (Table 2). However, among the scales, a validated translation was only available in Finnish for the Food Neophobia Scale [68]. Nevertheless, four of the present authors, who were all native Finnish speakers and experts in the field of food sciences, proofread the translations of the other scales. Yet, we acknowledge the need for further validation of these scales in the Finnish language and culture.

#### **5. Conclusions**

Our survey data, which were obtained from a representative sample of Finnish millennials, suggest that the hedonic tones of the first associations with meat vs. plant-based meat alternatives (positive-negative) are not unidimensional; rather, they are two-dimensional phenomena that can be used for easy consumer segmentation. The hedonic tone associated with meat alternatives was opposite to that associated with meat for some respondents, albeit not for all of them. In fact, some people think positively about both meat and meat alternatives, while other consumers are neutral concerning both food categories. Our classification of consumers was performed based on their responses to two simple questions, and it led to six segments. This allowed us to distinguish not only people who exclusively

promote meat or vegetarian diets but also those who have positive attitudes toward both meat and meat alternatives. These respondents were mostly flexitarians or omnivores who consumed meat alternatives because they liked to try new foods, in addition to environmental reasons. Thus, this consumer segment was considered the best target group for behavioral interventions designed to replace meat consumption with the consumption of meat alternatives.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods11030456/s1, Table S1: Survey questions (in English and Finnish), Table S2: Data (responses to the online survey).

**Author Contributions:** Conceptualization, A.K., F.M., V.P., K.J., and T.S.-S.; methodology, F.M. and A.K.; formal analysis, A.K.; investigation, A.K., F.M., V.P., K.J., and T.S.-S.; resources, V.P.; data curation, A.K.; writing—original draft preparation, A.K.; writing—review and editing, A.K., F.M., V.P., K.J., and T.S.-S.; visualization, A.K.; project administration, V.P.; funding acquisition, V.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the European Institute of Innovation and Technology (EIT), a body of the European Union, under Horizon 2020, the EU Framework Programme for Research and Innovation (EIT Food activity number 19121, *Tasty Texture-Tailored Fibre/Protein-Rich Vegetarian Health Power Food & Novel Extrusion Technology Platform for the Manufacture of Such*, 3TexVegS+H), and by the Strategic Research Council of the Academy of Finland (project number 327698, *Legumes for Sustainable Food System and Healthy Life*, Leg4Life).

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki and followed the ethical principles concerning sensory and consumer research at University of Helsinki, approved by the University of Helsinki Ethical Review Board in the Humanities and Social and Behavioural Sciences (Statement 46/2016).

**Informed Consent Statement:** Informed consent was obtained digitally from all respondents involved in the study.

**Data Availability Statement:** The data that have been analyzed for this article (anonymized responses to the online survey) are available as (Supplementary Materials Table S2).

**Acknowledgments:** We thank all the respondents to the online survey. We also thank Lotta Kemppinen for her contribution to the translation of the questionnaire into Finnish.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analysis, or interpretation of the data; in the writing of the manuscript; or in the decision to publish the results.

#### **References**

