Abstract
Nutritional information on packaging is becoming increasingly important in the food industry. Currently, labels are seen not only on the back of the packaging but also on the front. As there are many versions of front-of-pack labels (FoPLs), the research aims to determine which helps consumers the most in making decisions about which food to include in a healthier lifestyle. Nutri-Score, Guideline Daily Amount (GDA) and Multiple Traffic Lights (MTL) FoPLs on cereals were compared using eye-tracking (ET) and choice-based conjoint analysis (CBCA). CBCA was used to assess consumer preferences and the labels and products were also ranked. Based on the results, GDA type FoPL proved to be the most useful based on conjoint analysis, ranking and the analysis of ET parameters. This label helped participants the most in choosing the product that best fits into a healthier lifestyle. The Nutri-Score label, which offers little information on a product's nutritional content, was not favourably received by the Hungarian sample, who preferred more detailed FoP labels.
Introduction
Types of front-of-pack (FoP) labels
The complex habit of choosing what to eat is influenced by a number of interrelated elements (Köster, 2009). In the modern developed world, a wide variety of foods may be purchased with little to no effort, and there are many possibilities within each food group (Rozin, 2005). In this context, food labels are crucial for capturing consumers' attention and providing them with information that shapes expectations and influences purchase decisions. Customers utilize specific information from labels to make decisions when choosing from a variety of options for a particular product (Mawad et al., 2015). Due to their role as a major contributor to the global rise in the incidence of overweight, obesity and non-communicable diseases, unhealthy eating habits have emerged as a topic of significant concern (WHO, 2013).
It is well-known that the primary function of food packaging is to protect food during transport, distribution and storage, but it can and should also be used for communication with consumers (Kuti et al., 2021). There are two basic ways that food labels convey information to consumers: linguistic signs, such as information about ingredients, brands, manufacturers, and nutritional facts, or signs based on aesthetics: colors, shapes or pictures (Smith et al., 2010). It has been shown that consumers pay signifcant attention to price and expiry dates on packaging, but they also increasingly focus on nutritional information (Wyrwa and Barska, 2017). This is why the importance of so-called front-of-pack labels (FoPLs) has increased, both for food safety reasons and to promote a healthier lifestyle. These labels can be found on the front of the package and indicate the level or quantity of key nutrients using texts or colours, or even just a symbol to indicate the healthiness of the product (Hodgkins et al., 2012). FoP labels are communication elements that summarize nutritional information on the packaging in a detailed way (Godden et al., 2023). However, these labels are not uniform; they have different appearances on the packaging, making them more difficult for consumers to interpret, and their use is not compulsory (Cowburn and Stockley, 2005; Sharf et al., 2012; Gabor et al., 2020). There are now several types of FoPLs available for producers and they are categorized into the following three categories: directive, semi-directive and non-directive FoPLs (Hodgkins et al., 2012). Non-directive labels only state the amount of key nutrients in the food but do not include any health claims. Examples include the Nutrition Facts Panel (NFP) and Guideline Daily Amounts (GDA) (Bonsmann et al., 2010; Benito et al., 2013; Orquin et al., 2020). Semi-directive labels provide information on the nutritional content of a product, but use colours to classify each nutrient (low, medium and high) to help consumers make informed decisions when choosing a product. The best example of this type of label is the Multiple Traffic Light (MTL) label, which uses three colours for classification: green, yellow and red (Hawley et al., 2012; Roberto et al., 2012; Temple, 2020). Directive FoP labels do not cover the nutritional content of the product, so no quantitative information can be obtained from these labels. However, they reduce the cognitive effort of consumers by providing simple symbols. This way, consumers do not need to engage in a complex evaluation process or develop a more thorough understanding of the indications on the label. One of the most widely known labels is Nutri-Score, which classifies products on a scale from A to E and assigns a colour to each letter, making it easier for the consumer to interpret. The letters correspond to colours and healthfulness: A – dark green: healthiest; B – light green; C – yellow; D – orange; E − red: least healthy. It provides information on the nutritional value of the product based on predefined scoring criteria (Chantal and Hercberg, 2017; Egnell et al., 2020; Hercberg et al., 2022; Godden et al., 2023). Previous studies have shown that Nutri-Score is the most popular and understandable front-of-pack label for consumers (Talati et al., 2019; Egnell et al., 2020; Aguenaou et al., 2021). Some research shows that the GDA label is very effective (Crosetto et al., 2016; Deliza et al., 2019). There are also studies that show that MTL performed better than GDA (Watson et al., 2014; Arrúa et al., 2017; Gorski Findling et al., 2018; Khandpur et al., 2018).
Effects of different nutrition labels on consumer choices
A number of studieshave been carried out in laboratory conditions to test FoPLs, and the findings suggest that these labels help consumers to choose healthier foods (Dubois et al., 2020). Gabor and colleagues (2020) conducted a study using an eye-tracker to examine three nutrition labels on chocolate bars. Their results showed that the Nutri-Score label required the least visual attention, followed by MTL, then GDA, and the same order of healthiness perception was obtained. In an other study on biscuits and cookies, participants were asked to select the product with the label they would buy. Results suggest that the nutrition warnings were effective in attracting consumers' attention, but the product itself was the most important driver in their choice (Tórtora et al., 2019). However, it is interesting to note that FoPLs can not only be applied to packaged foods but can also be used on other items, such as menus. In this way, FoPLs can provide additional information on the healthiness of meals offered by a restaurant. This is illustrated by a 2016 study conducted by Reale and Flint. In their research, they created three menus that incorprated three different ways of presenting the calorie content of foods, including categorizing foods into three groups (red, yellow and green) using the MTL principle. There was no difference found between the labels in terms of attention, but the use of colours led participants to choose lower calorie foods. A 2019 study by van den Akker and colleagues tested Nutri-Score and MTL labels on cereals. Their results showed that Nutri-Score promoted the choice of a healthier alternatives. However, for those who followed a health-conscious lifestyle, no effect was observed on portion size or choice (van den Akker et al., 2022).
Conjoint analysis
Luce and Tukey (1964) laid the foundations for the conjoint analysis (CA) method, which allows researchers to study consumer preferences. Choice-based conjoint analysis has several advantages, as formulated by Almli and Næs (2018).
Using choice-based conjoint analysis (CBCA), researchers can analyze and estimate consumers' purchase behaviour and preferences, which result from decisions between several brands or items using a variety of stimuli. This method suggests that buyers derive benefits from particular combinations of features of goods rather than from just the goods. The conclusions are then interpreted as “part-worth utilities” (Raghavarao et al., 2010). The major advantage of CBCA is that different part-worth utilities of products can be evaluated, and specific product characteristics can be directly applied in conjoint analysis, for example, for a brand-specific attribute. In this way, the consideration of sub-values can show the impact of product characteristics on consumers' product choices (Meyerding and Merz, 2018). The most significant drawback is that it is not possible to determine whether a particular product feature is uninteresting to the participant or simply does not capture their attention. However, by supplementing the analysis with an eye-tracker measurement, this problem can be eliminated.
The purpose of CA is to determine which combination of characteristics of a product or service is most likely to influence a respondent's decision. CA seeks to understand the relative preference for individual attributes and combinations of attributes (McFadden, 1973). This is achieved by decomposing preference levels for each attribute based on the overall ratings of alternative products or services. Choice-based conjoint analysis expresses preferences through choices rather than rankings (Radler et al., 2020). Several studies have used conjoint analysis to examine preferences for front-of-pack labels, including Drewnowski et al. (2010). Anabtawi et al. (2020) and Godden et al. (2023). The aim of Anabtawi and his colleagues was to investigate the relative influence of macronutrients on the perceived healthiness of products using traffic light labelling (TLL). Their research concluded that sugar was the most important macronutrient, with red labelling having a significantly higher effect than green (Anabtawi et al., 2020). Drewnowski and colleagues aimed to estimate the relative contribution of the declared amounts of different nutrients to consumers' perceptions of the overall “healthiness” of foods by examining nutritional content claims. Their findings indicate that perceptions of healthiness are determined by the presence of protein, fibre, vitamin C and calcium, as well as the absence of saturated fat and sodium. Furthermore, total sugars and added sugars were found to have lower utility and contributed less to healthiness (Drewnowski et al., 2010).
Eye-tracking
The purpose of thesenses is to connect us with our environment. The eye is one of the most important sense organs, as it receives about 80% of the information we receive from the environment. It is also known that a quarter of the human brain is dedicated to visual image processing, leading to a signifincant amount of knowlegede being accumulated about this organ (Zurawicki, 2010). Due to this ever-growing body of knowledge, marketers are gaining a better understanding of how consumers react to certain stimuli from their surroundings (Zurawicki, 2010). As the first impression of a product is formed through the eyes, it is crucial for a product to attract visual attention. Consequently, visual attention is becoming increasingly important in gaining a deeper understanding on consumer decision-making (van der Laan et al., 2015). Choosing a product is a complex process; therefore, to gain more insight into it, it is essential to understand what attracts the eyes, what sustains visual attention and how each factor influences the decision-making process (Navalpakkam et al., 2012).
Eye tracking has been extensively used for a long time to examine people's visual attention, with the most popular approach being pupil-centered corneal reflectance (PCCR). Eye-trackers record participants' eye movements, including where they gaze and how long they focus on a particular area, where their gaze first lands and how long they glance at a given location. In order to define these variables, eye-trackers are able to measure eye direction and point, eye presence recognition, eye location, eye identity, eyelid closure, and pupil dilation and size (Santos et al., 2015).
Eye-tracking measures and conjoint analysis
In the food industry, CA is an effective tool for studying product packaging, since consumers often select products based on their packaging, making its optimal design critically important (Silayoi and Speece, 2007; Widaningrum, 2014; Ares et al., 2016).
Through the use of conjoint analysis, it is possible to understand differences in consumer preferences for product attributes. This method has been used in a number of studies on FoPLs, as demonstrated by Godden et al. (2023). Their research examined four product attributes - nutrition labelling, presence/absence of a nutrition claim, brand and price – focusing on how FoP labelling influenced participants' food choices. The results indicated that a segment of consumers was inclined towards healthier food choices when exposed to the Nutri-Score, while another segment tended to choose unhealthy options. They concluded that consumers have varying preferences for labels and that labels do not guide all consumer segments towards healthier alternatives (Godden et al., 2023).
In 2014, Ares and colleagues used eye-tracking and conjoint analysis to measure different variations of yoghurt labels. Their study aimed to assess the impact of rational and intuitive thinking styles on consumer decision-making and information processing. The findings revealed that consumers who relied predominantly on rational thinking styles searched for information on packaging to a greater extent compared to consumers who followed intuitive thinking. Thus, studies of this nature can be valuable for determining the impact of thinking styles on food choices, which can greatly assist in designing communication strategies to change dietary habits (Ares et al., 2014).
The aim of this study is to deepen the understanding of the relationship between attention and choice, with a special focus on the effect of FoPLs on food choices. By combining eye-tracking and choice -based conjoint analysis, inferences will be drawn about the relationship between product and FoPL based on the results of CBCA and the eye-tracking parameters under investigation.
This study aims to answer the following questions:
Which FoP label is preferred by consumers?
Which FoP label receives most visual attention?
Which product receives most visual attention?
Do the different methods of analysis yield consistent results?
Materials and methods
Location
The measurements were conducted at the Buda Campus of the Hungarian University of Agriculture and Life Sciences, in a room of approximately 20 m2 room located in the Sensory Laboratory of the Institute of Food Science and Technology. A table with a computer was placed in the centre of the room. Lighting was provided by an LED panel (6,500 K, 1,600 lm) mounted on the ceiling directly above the table.
Participants
The primary study included a total of 33 participants. Due to unacceptably poor eye sampling quality (80%), three participants were excluded from the study. The recruited participants were all Hungarian university students. Table 1 provides detailed demographic data on the participants.
Demographic profile of the participants, n = 30 (%)
| Gender | Male | 47.0 | |
| Female | 53.0 | ||
| Place of living | Male | Large city | 27.0 |
| Small town | 13.0 | ||
| Rural | 7.0 | ||
| Female | Large city | 30.0 | |
| Small town | 17.0 | ||
| Rural | 6.0 | ||
| Education | Male | Graduate | 7.0 |
| Undergraduate | 40.0 | ||
| Female | Graduate | 20.0 | |
| Undergraduate | 33.0 |
Eye-tracker and software
Information on the eye-tracking procedure is provided in accordance with the guidelines set by Fiedler et al. (2020). Eye movements were tracked using the Tobii Pro Nano (Tobii Pro AB, Danderyd, Sweden). This eye-tracker is discreet, compact, allows for some head movement and does not interfere with the participant's natural behaviour. The software called Tobii Pro Lab v.1.171 (Tobii Pro AB, Danderyd, Sweden) was used to display the timelines. During eye-tracking, near-infrared light is emitted onto the face of the participants' faces but it is reflected only by their eyes. The reflected light is recorded by a camera, and a 3D eye model technique is used to compute the gaze point and position of the eyes. Image processing algorithms are used to detect distinctive features in the participants' eyes and their reflection patterns. The software generated and extracted a sizable amount of data, including variables such as first fixation duration (FFD), total fixation duration (FD) and fixation count (FC). The optimal viewing angle is 65°, and the recommended distance between the camera and the participants' eyes is 60–65 cm.
Process
As the initial step in the measurement process, participants were requested to sit down in front of the computer in a comfortable position and were instructed to remain still during the test. Next, the measurement process was explained. Before starting the measurements, calibration of the eye-tracker was completed. The timelines (Fig. 1) were initiated only after a satisfactory calibration result. The participants were first presented with a screen containing instructions. After reading the content, participants advanced to the following slide by pressing a key on the keyboard. Next, participants were shown two products with different FoPLs on a subsequent slide, and were asked to select the product they preferred, without time limit. Once they made their choice, they pressed a key on the keyboard, at which point the mouse cursor appeared on the screen. Using the cursor, participants indicated their selected product. To proceed to the next slide, participants pressed a key on the keyboard again. A fixation cross was presented for two seconds between each choice set. Each participant completed decisions on five product pairs. At the end of the process, participants completed a brief questionnaire that collected demographic data and feedback on the products and the FoPLs.
The applied two-alternative (binary) forced choice task. Participants looked at the products without a time limit and as soon as they made their decisions, a decision confirmation screen appeared where they clicked on the chosen product. (Product photos are blurred on purpose)
Citation: Progress in Agricultural Engineering Sciences 20, 1; 10.1556/446.2024.00132
Visual stimuli
Participants were presented two products at a time, along with their corresponding front-of-pack labels (NS, GDA, MTL). Each participant visually inspected five pairs of images, with the order of the products randomized. Product photos of breakfast cereals were used, including both healthier and less healthy alternatives. The five products are shown is Fig. 2. Three FoPLs (NS, GDA, MTL, see Fig. 3) were used for each of the five products, resulting in a total of 15 product variations to be tested. Participants were shown a fixation cross on the screen for two seconds before each photo was exhibited to ensure that their gaze was fixed on the center of the screen prior to viewing the poduct images An LG W2452V-PF 24″ Full HD LCD monitor with a resolution of 1,366 × 768 was used to display the visual stimuli. The separation between areas of interests (AOIs) was increased to prevent overlap. We applied an Identification by Velocity Threshold (I-VT) filtering method that used interpolation between gaps (75 ms), noise reduction (median), a velocity threshold of 30°/s, merged adjacent fixations (<0.5°) between fixations (<75 ms), and discarded short fixations (<60 ms).
Products tested in the measurement. (Product photos are blurred on purpose)
Citation: Progress in Agricultural Engineering Sciences 20, 1; 10.1556/446.2024.00132
Examples of the three front-of-pack labels used in the measurement (GDA, NS and MTL, respectively)
Citation: Progress in Agricultural Engineering Sciences 20, 1; 10.1556/446.2024.00132
Questionnaire
In addition to demographic data (e.g. gender, age, place of living), the questionnaire also included questions about the products and labels featured in the survey. The questionnaire was created using Google Forms (Google LLC, California, USA).
Participants were asked to rank the products in the study (1 - healthiest, 5 - least healthy) both before and after the eye-tracking measurement. They also rated the products based on the perceived healthiness. Regarding the labels, participants were asked whether the labels provided sufficient information for making a product choice (1 - not at all, 5 - very much information).
The ranking results of the products were analysed using the Kruskal-Wallis test, while the evaluations of the products and labels were analysed using one-way analysis of variance (ANOVA).
Conjoint analysis
Conjoint cards were created using the R-project (ver. 4.2.1) (R Development Core Team, 2023) package
Results and discussion
In this section, the results of the eye-tracking measurement are presented first. Next, the data obtained from the ranking and additional questions in the questionnaire are analysed. The results of the conjoint analysis are then presented. Finally, the results obtained from the different methods are compared.
Eye-tracking measures
Figure 4 presents a heatmap generated from the eye-movements of the participants. Red areas indicate higher visual attentention, while cooler colors represent lower attention. The green spot between the two stimuli represents the starting point, as the fixation cross was placed in the centre of the screen.
An example of a heat map produced using data from the eye-tracking survey. (Product photos are blurred on purpose)
Citation: Progress in Agricultural Engineering Sciences 20, 1; 10.1556/446.2024.00132
Table 2 presents the results of one-way analysis of variance (ANOVA). Univariate tests indicate that the product had a significant effect on FD, FC, FFD and DD parameters. As shown in the table, the P3 product attracted the least visual attention based on the FD value, which is also true for the FC and DD parameters. The table clearly shows that P1 and P4, and P2 and P3 products belong to the same group, with the two groups being significantly different from each other. Within the groups, two and three products, respectively, received almost the same visual attention. Based on the eye-tracking parameters, P1 and P4 products received the most visual attention.
Results of the ANOVA for the products for the visual parameters
| FD | FC | TTFF | FFD | DD | DC | |
| P1 | 2.900b | 6.509ab | 1.838a | 0.327a | 2.931b | 3.421a |
| P2 | 2.379ab | 5.930ab | 1.008a | 0.244a | 2.244ab | 3.404a |
| P3 | 2.204ab | 5.281a | 1.422a | 0.318a | 2.302ab | 3.109a |
| P4 | 3.093b | 7.444b | 1.172a | 0.252a | 3.248b | 3.651a |
| P5 | 1.542a | 5.169a | 1.497a | 0.288a | 1.623a | 2.763a |
| Pr > F | 0.003* | 0.012* | 0.227 | 0.032* | 0.003* | 0.225 |
Bold and * indicates effect of a significant level of P < 0.05. TTFF: Time To First Fixation; FFD: First Fixation Duration; FC: Fixaton Count; FD: Fixation Duration; DC: Dwell Count; DD = Dwell Duration.
For the label, a significant effect was observed for the FD, FC, DD and DC (Table 3). For those parameters where a significant difference was observed, the GDA label received the most visual attention in all cases. However, for each of the eye-tracking parameters, no significant difference was found between the GDA and MTL labels. After statistical analysis of all parameters, it ws evident that the NS label received the least visual attention. This finding confirms that the low information content of the NS label requires less visual attention, whereas the GDA and MTL labels require more time to process their greater information content and thus receive more visual attention.
Results of the ANOVA for the labels for the visual parameters
| FD | FC | TTFF | FFD | DD | DC | |
| GDA | 4.004b | 7.441b | 2.134a | 0.663a | 4.130b | 3.549b |
| MTL | 3.355b | 7.074b | 2.110a | 1.164a | 3.442b | 3.495b |
| NS | 1.605a | 4.087a | 2.006a | 0.334a | 1.584a | 2.796a |
| Pr > F | <0.0001* | <0.0001* | 0.847 | 0.477 | <0.0001* | 0.014* |
Bold and* indicate that the effect is statistically significant at the P < 0.05 level. TTFF: Time To First Fixation; FFD: First Fixation Duration; FC: Fixaton Count; FD: Fixation Duration; DC: Dwell Count; DD = Dwell Duration; GDA = Guideline Daily Amount; MTL = Multiple Traffic Light; NS = Nutri-Score.
The results for the choice data are presented in Table 4. Significant differences were observed between the chosen and not chosen alternatives for the parameters FD, FC, DD and DC. Based on the eye-tracking parameters, it is evident that considerably more visual attention was paid to the product that was ultimately chosen by the participants.
Comparison of the chosen and not chosen alternatives based on the eye-tracking parameters
| FD | FC | TTFF | FFD | DD | DC | |
| Chosen | 6.557b | 14.087b | 3.223a | 1.378a | 6.736b | 7.447b |
| Not chosen | 4.245a | 10.400a | 3.710a | 0.611a | 4.291a | 5.640a |
| Pr > F | <0.0001* | <0.0001* | 0.143 | 0.167 | <0.0001* | <0.0001* |
Bold and* indicate that the effect is statistically significant at the P < 0.05 level. TTFF: Time To First Fixation; FFD: First Fixation Duration; FC: Fixaton Count; FD: Fixation Duration; DC: Dwell Count; DD = Dwell Duration.
These results are consistent with previous research, such as the research of Gabor et al. (2020), where the order of visual attention was the same: the GDA label received the most visual attention and Nutri-Score received the least. Similarly, the research of Fenko et al. also supports our results, as they examined directive and semi-directive labels and found that the directive label received the least visual attention (Fenko et al., 2018). The finding that there is a correlation between product choice and visual attention to products, i.e. that participants tend to choose the products they look at more frequently and for longer periods of time, is also supported by several previous studies in (Krajbich and Rangel, 2011; Reutskaja et al., 2011; Ares et al., 2014; Danner et al., 2016).
Results of the questionnaire
Label and product rating
The questionnaire asked participants to rate the front-of-pack labels. The results of the evaluation are presented in Table 5. The data were analysed using one-way analysis of variance (ANOVA). The results clearly show that the Nutri-Score label was the least preferred by participants in terms of whether the label provides sufficient information when making food choices. In contrast, the MTL and GDA labels received the same rating. A significant difference was found between the three labels.
Results of the rating of the products including P-values
| Name of label/product | Value | Pr > F |
| GDA | 4.2000b | <0.0001 |
| MTL | 4.2000b | |
| NS | 2.667a | |
| P4 | 3.767b | |
| P1 | 3.467b | |
| P3 | 2.267a | |
| P2 | 1.667a | |
| P5 | 1.663a |
Abbreviations: P = product, GDA = Guideline Daily Amount label, MTL = Multiple Traffic Light label, NS = Nutri-Score label.
Participants were also asked to rate the products, with the results also shown in Table 5. The highest-rated product was P4, followed by P1. A significant difference was observed between the five products.
The results from previous studies vary, but in most cases, Nutri-Score is the most preferred label by participants due to its simplicity, ease of understanding and ability to assist decision-making (e.g. Talati et al., 2019). However, there are also studies, similar to the current research, where Nutri-Score does not perform best. For example, the results of Mazzú et al. showed that NutrInform Battery was more effective, a label that shows a high similarity with GDA (Mazzù et al., 2021).
However, it should also be considered how the research task is presented to the participants. In the present study, participants were asked to rate the labels based on which one provided the most information for selecting a product that best fits into a healthier lifestyle. In most studies, however, the task involves choosing the label that is most appealing, e.g. (Feunekes et al., 2008). Furthermore, it shoud be noted that some FoPLs are only effective for certain consumer segments, for which no Hungarian data has been published, to our knowledge.
Ranking of products before and after eye-tracking measurement
Before the eye-tracking measurement began, participants were asked to rank the products in the questionnaire. The results of the pre-measurement ranking are shown in Fig. 5. As illustrated in the graph, the P1 and P4 products received the highest ratings, consistently occupying the top ranks. The P3 product, with few exceptions, ranked in the middle, while P2 and P5 ranked last. According to the results of the Kruskal-Wallis test, P1 showed a significant difference from P2, P3 and P5, while P4 did not. The P2 product showed a significant difference from two products: P1 and P4. The P3 product showed no significant difference, except when compared to the P2 product. These results clearly demonstrate that the products form two distinct groups: one group consists of P1 and P4, and the other group includes P2, P3 and P5.
Ranking of the products before the measurement, analyzed by the Kruskal-Wallis test (Kruskal-Wallis test: H (4, N = 150) = 94.39978 P = 0.000)
Citation: Progress in Agricultural Engineering Sciences 20, 1; 10.1556/446.2024.00132
After the eye-tracking measurement, the ranking was repeated. By this time, the participants were more familiar with the products and their nutritional values, allowing for an examination of whether their perception of the healthiness of the products had changed. Figure 6 shows the results of the ranking after the eye-tracking measurement. P1 and P4 continued to occupy the top two places, and P3 remained in the middle; however, more participants rated the product as less healthy than previously. The perception of P2 has changed in a positive direction. The results of the Kruskal-Wallis test indicate that only one new significant difference emerged: the difference between P3 and P5 became significant. The ranking of the products within the two groups did not change compared to the results before the eye-camera measurement.
Ranking of the products after the measurement, analyzed by Kruskal-Wallis test (Kruskal-Wallis test: H (4, N = 150) = 101.0793 P = 0.000)
Citation: Progress in Agricultural Engineering Sciences 20, 1; 10.1556/446.2024.00132
Results of Multiple Correspondence Analysis (MCA)
The questionnaire also asked participants which of the listed nutritional values (energy, fat, saturated fat, sugar, protein, salt, vitamins and minerals) they consider important when looking at a label when shopping. The responses were analysed using Multiple Correspondence Analysis (MCA). The results of the MCA are presented in Fig. 7. The figure shows that for participants who consider the nutritional value of sugar important, fat and protein content are also important. Similarly, for those who prioritize vitamin content, saturated fat and mineral content are also important, but not energy content. On the other hand, participants who do not consider the amount of sugar in food important tend not to care about the amount of protein, salt or fat content either. However, it is interesting to note that for participants who do care about energy content, the amount of vitamins, saturated fat and protein is not important.
Results of the Multiple Correspondence Analysis (MCA) on nutritional values. 1 means that the nutrient was marked as important, while 0 means that the nutrient was not marked as important. The plot displays the row and column profiles as points in a two-dimensional space. The points are plotted based on their contributions and cosines to the first two dimensions of the analysis
Citation: Progress in Agricultural Engineering Sciences 20, 1; 10.1556/446.2024.00132
Results of the choice-based conjoint analysis
Table 6 shows the part-worth utility values obtained from the CBCA. The results show that the highest part-worth utility value for the front label is GDA (L3), followed by MTL (L2). For the products, the highest part-worth utility value is shown for P1, followed by P4. In terms of products, P3 ranks last.
Results of the conjoint analysis for labels and products
| L1. | L2. | L3. | P1. | P2. | P3. | P4. | P5. |
| −0.002909 | 0.825416 | 1.577450 | 1.051825 | −1.940835 | −3.0524418 | 0.825378 | −2.469662 |
Bold indicates the highest rated product and label in the CBCA analysis.
Abbreviations: L = label, P = product. L1 = Nutri-Score (NS) label, L2 = Multiple Traffic Light (MTL) Label, L3 = Guideline Daily Amount (GDA) label.
Comparison of the results of different methods
As a final step, the results of the different methods were compared for both labels and products, as illustrated in Table 7. The conjoint analysis, ranking and eye-tracking parameters show similarities. For all labels, except for one eye-tracking parameter (FFD), the GDA (Guideline Daily Amount) FoPL was ranked first. The Nutri-Score label was considered the least informative by the participants. In terms of products, P1 was considered the healthiest based on the conjoint analysis, while P4 was ranked as the healtiest product based on the ranking and analysis of most eye-tracking parameters. These two products also ranked second in the alternative evaluations. P5 was considered by respondents to be the food least compatible with a healthy lifestyle.
Comparison of label and product results across all methods, depicted as ranks
| Rank method | Part-worth utilities | Ranking score | FC | FD | DC | DD |
| 1 | L3 | L3 | L3 | L3 | L3 | L3 |
| 2 | L2 | L2 | L2 | L2 | L2 | L2 |
| 3 | L1 | L1 | L1 | L1 | L1 | L1 |
| 1 | P1 | P4 | P4 | P4 | P4 | P4 |
| 2 | P4 | P1 | P1 | P1 | P1 | P1 |
| 3 | P2 | P3 | P2 | P2 | P2 | P3 |
| 4 | P5 | P2 | P3 | P3 | P3 | P2 |
| 5 | P3 | P5 | P5 | P5 | P5 | P5 |
Abbreviations: L = label, P = product. L1 = Nutri-Score (NS) label, L2 = Multiple Traffic Light (MTL) Label, L3 = Guideline Daily Amount (GDA) label.
Conclusion
Based on the results, the Hungarian sample preferred informative FoP labels, and they lack a favourable attitude towards labels such as Nutri-Score, which provide limited information on the nutritional content of products. The Nutri-Score label received less visual attention than GDA or MTL, likely due to its lower information content, which may lead to faster information processing and decision-making process. As a limiting factor, it is important to note that the Nutri-Score label is still in the early stages of introduction in Hungary, making it less familiar to participants, and there have been few information campaigns to raise awareness about the label. While it might have been beneficial to inform participants about the different FoP labels at the start of the study, the goal was to simulate a real-life shopping situation, where participants would not typically receive such information. In future studies, it would be beneficial to conduct the measurement with a larger sample size and to cluster the respondents by label use, differentiating between health-conscious shoppers and of non-health-conscious shoppers. A further limiting factor may be the age group of participants. Therefore, we believe that further measurements should include a broader range of age groups and individuals more representative of Hungarian consumers.
The findings of the study provide clear answers to the research questions. For the question “Which FoP label is preferred by consumers?”, the Guideline Daily Amount (GDA) label emerged as the most preferred, as indicated by both the conjoint analysis and eye-tracking results. Regarding “Which FoP label receives more visual attention?”, the GDA label received the highest level of attention, followed closely by the Multiple Traffic Lights (MTL), with Nutri-Score (NS) receiving the least. In answer to “Which product receives more visual attention?”, products P1 and P4 consistently captured the most visual attention and were ranked as the healthiest options. Lastly, for the question “Do the different methods of analysis yield consistent results?”, both methods—conjoint analysis and eye-tracking—showed consistency, with GDA proving to be the most effective in aiding consumer decisions.
Author contributions
Conceptualization, D.Sz. and A.G.; methodology, D.Sz., O.F., A.G.; software, D.Sz.; validation, O.F.; formal analysis, D.Sz.; investigation, D.Sz.; resources, A.G.; data curation, A.G.; writing—original draft preparation, D.Sz.; writing—review and editing, D.R., E.B.; visualization, D.Sz.; supervision, D.R. and O.F.; project administration, A.G.; funding acquisition, A.G. All authors have read and agreed to the published version of the manuscript.
Institutional review board statement
Ethical review and approval were waived for this study as research data has been robustly anonymized, such that the original providers of the data cannot be identified, directly or indirectly, by anyone.
Data availability statement
The data presented in this study are available on request from the corresponding author.
Conflicts of interest
The authors declare no conflict of interest.
Acknowledgments
DSz thanks the support of the Doctoral School of Economic and Regional Sciences, Hungarian University of Agriculture and Life Sciences. AG thanks the support of the János Bolyai Research Scholarship of the Hungarian Academy of Sciences. AG thanks the support of the National Research, Development, and Innovation Office of Hungary (OTKA, contracts No FK 137577 and K 134260). Supported by the ÚNKP-22-3-I New National Excellence Program of the Ministry for Innovation and Technology from the source of the National Research, Development and Innovation Fund.
References
Aguenaou, H., El Ammari, L., Bigdeli, M., El Hajjab, A., Lahmam, H., Labzizi, S., Gamih, H., Talouizte, A., Serbouti, C., El Kari, K., Benkirane, H., El Berri, H., AL-Jawaldeh, A., and Yahyane, A. (2021). Comparison of appropriateness of Nutri-Score and other front-of-pack nutrition labels across a group of Moroccan consumers: awareness, understanding and food choices. Archives of Public Health, 79(1): 1–13, https://doi.org/10.1186/s13690-021-00595-3.
Almli, V.L. and Næs, T. (2018). Conjoint analysis in sensory and consumer science: principles, applications, and future perspectives. In: Methods in consumer research, volume 1: new approaches to classic methods, Vol. 1. Elsevier Ltd, https://doi.org/10.1016/B978-0-08-102089-0.00019-4.
Anabtawi, O., Swift, J.A., Hemmings, S., Gertson, L., and Raaff, C. (2020). Perceived healthiness of food items and the traffic light front of pack nutrition labelling: choice-based conjoint analysis and cross-sectional survey. Journal of Human Nutrition and Dietetics, 33(4): 487–495, https://doi.org/10.1111/jhn.12741.
Ares, G., Mawad, F., Giménez, A., and Maiche, A. (2014). Influence of rational and intuitive thinking styles on food choice: preliminary evidence from an eye-tracking study with yogurt labels. Food Quality and Preference, 31(1): 28–37, https://doi.org/10.1016/j.foodqual.2013.07.005.
Ares, G., Arrúa, A., Antúnez, L., Vidal, L., Machín, L., Martínez, J., Curutchet, M.R., and Giménez, A. (2016). Influence of label design on children’s perception of two snack foods: comparison of rating and choice-based conjoint analysis. Food Quality and Preference, 53: 1–8, https://doi.org/10.1016/j.foodqual.2016.05.006.
Arrúa, A., MacHín, L., Curutchet, M.R., Martínez, J., Antúnez, L., Alcaire, F., Giménez, A., and Ares, G. (2017). Warnings as a directive front-of-pack nutrition labelling scheme: comparison with the Guideline Daily Amount and traffic-light systems. Public Health Nutrition, 20(13): 2308–2317, https://doi.org/10.1017/S1368980017000866.
Benito, A., Basulto, J., Babio, N., Vicent, P., and Lo, L. (2013). Adolescents’ ability to select healthy food using two different front-of-pack food labels : a cross-over study. 17(6): 1403–1409, https://doi.org/10.1017/S1368980013001274.
Bonsmann, S., Celemín, L., and Grunert, K. (2010). Food labelling to advance better education for life. 14–19, https://doi.org/10.1038/ejcn.2010.204.
Chantal, J. and Hercberg, S. (2017). Development of a new front-of-pack nutrition label in France: the five-colour Nutri-Score. Public Health Panorama, 03(04): 712–725.
Chapman, C, Bahna, E, and Alford, J.E.S (2022). Tools for choice modeling, conjoint analysis, and MaxDiff analysis of best-worst surveys_. R package version 0.0.0.9082.
Cowburn, G. and Stockley, L. (2005). Consumer understanding and use of nutrition labelling: a systematic review. Public Health Nutrition, 8(1): 21–28, https://doi.org/10.1079/phn2004666.
Crosetto, P., Muller, L., and Ruffieux, B. (2016). Helping consumers with a front-of-pack label: numbers or colors? Journal of Economic Psychology, 55: 30–50, https://doi.org/10.1016/j.joep.2016.03.006.
Danner, L., De Antoni, N., Gere, A., Sipos, L., Kovács, S., and Dürrschmid, K. (2016). Make a choice! Visual attention and choice behaviour in multialternative food choice situations. Acta Alimentaria, 45(4): 515–524, https://doi.org/10.1556/066.2016.1111.
Deliza, R., de Alcántara, M., Pereira, R., and Ares, G. (2019). How do different warning signs compare with the guideline daily amount and traffic-light system? Food Quality and Preference: 103821, https://doi.org/10.1016/j.foodqual.2019.103821.
Drewnowski, A., Moskowitz, H., Reisner, M., and Krieger, B. (2010). Testing consumer perception of nutrient content claims using conjoint analysis. Public Health Nutrition, 13(5): 688–694, https://doi.org/10.1017/S1368980009993119.
Dubois, P., Albuquerque, P., Allais, O., Bonnet, C., Bertail, P., Combris, P., Lahlou, S., Rigal, N., Ruffieux, B., and Chandon, P. (2020). Effects of front-of-pack labels on the nutritional quality of supermarket food purchases: evidence from a large-scale randomized controlled trial. SSRN Electronic Journal, https://doi.org/10.2139/ssrn.3567974.
Egnell, M., Galan, P., Farpour-Lambert, N.J., Talati, Z., Pettigrew, S., Hercberg, S., and Julia, C. (2020). Compared to other front-of-pack nutrition labels, the Nutri-Score emerged as the most efficient to inform Swiss consumers on the nutritional quality of food products. Plos One, 15(2): 1–18, https://doi.org/10.1371/journal.pone.0228179.
Fenko, A., Nicolaas, I., and Galetzka, M. (2018). Does attention to health labels predict a healthy food choice? An eye-tracking study. Food Quality and Preference, 69(March): 57–65, https://doi.org/10.1016/j.foodqual.2018.05.012.
Feunekes, G.I.J., Gortemaker, I.A., Willems, A.A., Lion, R., and van den Kommer, M. (2008). Front-of-pack nutrition labelling: testing effectiveness of different nutrition labelling formats front-of-pack in four European countries. Appetite, 50(1): 57–70, https://doi.org/10.1016/j.appet.2007.05.009.
Fiedler, S., Schulte-Mecklenbeck, M., Renkewitz, F., and Orquin, J.L. (2020). Guideline for reporting standards of eye-tracking research in decision sciences. PsyArXiv.
Gabor, A.M., Stojnić, B., and Ban Ostić, D. (2020). Effects of different nutrition labels on visual attention and accuracy of nutritional quality perception – results of an experimental eye-tracking study. Food Quality and Preference, 84(December 2019): 103948, https://doi.org/10.1016/j.foodqual.2020.103948.
Godden, E., Thornton, L., Avramova, Y., and Dens, N. (2023). High hopes for front-of-pack (FOP) nutrition labels? A conjoint analysis on the trade-offs between a FOP label, nutrition claims, brand and price for different consumer segments. Appetite, 180(October), https://doi.org/10.1016/j.appet.2022.106356.
Gorski Findling, M.T., Werth, P.M., Musicus, A.A., Bragg, M.A., Graham, D.J., Elbel, B., & Roberto, C.A. (2018). Comparing five front-of-pack nutrition labels’ influence on consumers’ perceptions and purchase intentions. Preventive Medicine, 106(August 2017), 114–121, https://doi.org/10.1016/j.ypmed.2017.10.022.
Hawley, K.L., Roberto, C.A., Bragg, M.A., Liu, P.J., Schwartz, M.B., and Brownell, K.D. (2012). The science on front-of-package food labels. 16(3): 430–439, https://doi.org/10.1017/S1368980012000754.
Hercberg, S., Touvier, M., & Salas-salvado, J. (2022). The Nutri-Score nutrition label. 92: 147–157.
Hodgkins, C., Barnett, J., Wasowicz-kirylo, G., Stysko-kunkowska, M., Gulcan, Y., Kustepeli, Y., Akgungor, S., Chryssochoidis, G., Fernández-celemin, L., Storcksdieck, S., Gibbs, M., and Raats, M. (2012). Understanding how consumers categorise nutritional labels : a consumer derived typology for front-of-pack nutrition labelling q. Appetite, 59(3): 806–817, https://doi.org/10.1016/j.appet.2012.08.014.
Khandpur, N., de Morais Sato, P., Mais, L.A., Bortoletto Martins, A.P., Spinillo, C.G., Garcia, M.T., Urquizar Rojas, C.F., and Jaime, P.C. (2018). Are front-of-package warning labels more effective at communicating nutrition information than traffic-light labels? A randomized controlled experiment in a Brazilian sample. Nutrients, 10(6): 1–15, https://doi.org/10.3390/nu10060688.
Köster, E.P. (2009). Diversity in the determinants of food choice: a psychological perspective. Food Quality and Preference, 20(2): 70–82, https://doi.org/10.1016/j.foodqual.2007.11.002.
Krajbich, I. and Rangel, A. (2011). Multialternative drift-diffusion model predicts the relationship between visual fixations and choice in value-based decisions. Proceedings of the National Academy of Sciences of the United States of America, 108(33): 13852–13857, https://doi.org/10.1073/pnas.1101328108.
Kuti, B., Horacsek, M., Szakos, D., and Kasza, G. (2021). Regulation of nutrition labeling of foods in the European Union and Hungary: a historical review from the beginning to the present day. Élelmiszervizsgálati Közlemények, 67(1): 3281–3292.
Luce, R.D. and Tukey, J.W. (1964). Simultaneous conjoint measurement: a new type of fundamental measurement. Journal of Mathematical Psychology, 1(1): 1–27, https://doi.org/10.1016/0022-2496(64)90015-X.
Mawad, F., Trías, M., Giménez, A., Maiche, A., and Ares, G. (2015). Influence of cognitive style on information processing and selection of yogurt labels: insights from an eye-tracking study. Food Research International, 74: 1–9, https://doi.org/10.1016/j.foodres.2015.04.023.
Mazzù, M.F., Romani, S., Baccelloni, A., and Gambicorti, A. (2021). A cross-country experimental study on consumers’ subjective understanding and liking on front-of-pack nutrition labels. International Journal of Food Sciences and Nutrition, 72(6): 833–847, https://doi.org/10.1080/09637486.2021.1873918.
McFadden, D. (1973). Conditional logit analysis of qualitative choice behavior. Frontiers in econometrics: 105–143.
Meyerding, S.G.H. and Merz, N. (2018). Consumer preferences for organic labels in Germany using the example of apples – combining choice-based conjoint analysis and eye-tracking measurements. Journal of Cleaner Production, 181: 772–783, https://doi.org/10.1016/j.jclepro.2018.01.235.
Navalpakkam, V., Kumar, R., Li, L., and Sivakumar, D. (2012). Attention and selection in online choice tasks. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 7379 LNCS: 200–211, https://doi.org/10.1007/978-3-642-31454-4_17.
Orquin, J.L., Bagger, M.P., Lahm, E.S., Grunert, K.G., and Scholderer, J. (2020). The visual ecology of product packaging and its effects on consumer attention. Journal of Business Research, 111(January): 187–195, https://doi.org/10.1016/j.jbusres.2019.01.043.
R Development Core Team (2023). R: a language and environment for statistical ## computing. R Foundation for Statistical Computing, Vienna, Austria, https://www.R-project.org/. R Foundation for Statistical Computing.
Radler, B.T., Shaw, B., and Gorby, T.A. (2020). Using choice-based conjoint analysis for creating effective outreach messages to promote oak management. Journal of Forestry, 118(4): 419–432, https://doi.org/10.1093/jofore/fvaa018.
Raghavarao, D., Wiley, J., and Chitturi, P. (2010). Choice-based conjoint analysis: models and designs. In: Choice-based conjoint analysis: models and designs, https://doi.org/10.1201/9781420099973.
Reale, S. and Flint, S. (2016). The impact of menu label design on visual attention , food choice and recognition : an eye tracking study. Food Choice and Recognition, 31: 328–340, https://doi.org/10.1111/joss.12216.
Reutskaja, E., Nagel, R., Camerer, C.F., and Rangel, A. (2011). Search dynamics in consumer choice under time pressure: an eye-tracking study. American Economic Review, 101(2): 900–926, https://doi.org/10.1257/aer.101.2.900.
Roberto, C.A., Bragg, M.A., Schwartz, M.B., Seamans, M.J., Musicus, A., Novak, N., and Brownell, K.D. (2012). Facts up front versus traffic light food labels. AMEPRE, 43(2): 134–141, https://doi.org/10.1016/j.amepre.2012.04.022.
Rozin, P. (2005). The meaning of food in our lives: a cross-cultural perspective on eating and well-being. Journal of Nutrition Education and Behavior, 37(SUPPL. 2), https://doi.org/10.1016/S1499-4046(06)60209-1.
Santos, R.D.O.J.dos, Oliveira, J.H.C.de, Rocha, J.B., and Giraldi, J.D.M.E. (2015). Eye tracking in neuromarketing: a research agenda for marketing studies. International Journal of Psychological Studies, 7(1): 32–42, https://doi.org/10.5539/ijps.v7n1p32.
Sharf, M., Sela, R., Zentner, G., Shoob, H., Shai, I., and Stein-Zamir, C. (2012). Figuring out food labels. Young adults’ understanding of nutritional information presented on food labels is inadequate. Appetite, 58(2): 531–534, https://doi.org/10.1016/j.appet.2011.12.010.
Silayoi, P. and Speece, M. (2007). The importance of packaging attributes: a conjoint analysis approach. European Journal of Marketing, 41(11–12): 1495–1517, https://doi.org/10.1108/03090560710821279.
Smith, V., Møgelvang-Hansen, P., and Hyldig, G. (2010). Spin versus fair speak in food labelling: a matter of taste? Food Quality and Preference, 21(8): 1016–1025, https://doi.org/10.1016/j.foodqual.2010.05.016.
Talati, Z., Egnell, M., Hercberg, S., Julia, C., and Pettigrew, S. (2019). Food choice under five front-of-package nutrition label conditions: an experimental study across 12 countries. American Journal of Public Health, 109(12): 1770–1775, https://doi.org/10.2105/AJPH.2019.305319.
Temple, N.J. (2020). Front-of-package food labels : a narrative review. Appetite, 144(March 2019): 104485, https://doi.org/10.1016/j.appet.2019.104485.
Tórtora, G., Machín, L., and Ares, G. (2019). Influence of nutritional warnings and other label features on consumers’ choice: results from an eye-tracking study. Food Research International, 119(October 2018): 605–611, https://doi.org/10.1016/j.foodres.2018.10.038.
van den Akker, K., Bartelet, D., Brouwer, L., Luijpers, S., Nap, T., and Havermans, R. (2022). The impact of the nutri-score on food choice: a choice experiment in a Dutch supermarket. Appetite, 168: 105664, https://doi.org/10.1016/j.appet.2021.105664.
van der Laan, L.N., Hooge, I.T.C., De Ridder, D.T.D., Viergever, M.A., and Smeets, P.A.M. (2015). Do you like what you see? The role of first fixation and total fixation duration in consumer choice. Food Quality and Preference, 39: 46–55, https://doi.org/10.1016/j.foodqual.2014.06.015.
Watson, W.L., Kelly, B., Hector, D., Hughes, C., King, L., Crawford, J., Sergeant, J., and Chapman, K. (2014). Can front-of-pack labelling schemes guide healthier food choices? Australian shoppers’ responses to seven labelling formats. Appetite, 72: 90–97, https://doi.org/10.1016/j.appet.2013.09.027.
WHO (2013). World health report: Research for universal health coverage. Geneva: WHO. August 2013. https://www.who.int/publications/i/item/9789241564588.
Widaningrum, D.L. (2014). The importance of take-out food packaging attributes: conjoint analysis and quality function deployment approach. EPJ Web of Conferences, 68: 1–7, https://doi.org/10.1051/epjconf/20146800036.
Wyrwa, J. and Barska, A. (2017). Packaging as a source of information about food products. Procedia Engineering, 182: 770–779, https://doi.org/10.1016/j.proeng.2017.03.199.
Zurawicki, L. (2010). Neuromarketing. Exploring the brain of the consumer. Springer, New York.
