Introduction

Soy beverages can be obtained from whole soybeans, by aqueous extraction or by suspension of isolated soy protein in water, and they are usually sold in carton packs after being submitted to the UHT (Ultra High Temperature) technique. Due to the increasing number of product launches in this segment, there is a difference between the information printed on the packages. Amid the COVID-19 crisis, the global soymilk market, estimated at $11.6 billion in the year 2020, is expected to reach a revised size of $16.5 billion by 2027 (Research and Markets, 2021b). Of the soymilk formulation options, the best-selling is called plain or original and accounted for over 52.7% of the market share in 2016 (Research and Markets, 2021a).

As packaging is key in product purchase selection and decision (Della Lucia et al., 2007), the information on the package and the consumer´s expectation in relation to the product (Deliza et al., 2000; Dantas et al., 2005) can be assessed subjectively using qualitative tests such as the focus group technique (^{Dutcosky, 2013}).

The focus group is a method of exploratory nature, with the goal of providing ideas about preferences, incentives and barriers to certain behaviors (van Kleef, van Trijp Luning, 2005a). It allows participants to explain motivations and reasons for their attitudes, preferences, and perceptions in a roundtable session conducted by an impartial moderator (Della Lucia; Minin; Carneiro, 2010; Dutcosky, 2013). The factors and levels determined in the focus group can be used in other research methods such as the conjoint factor analysis.

Conjoint factor analysis (CFA) is commonly used to get information on the effect of different attributes on the consumer preference and/or purchase intent. It uses the scores assigned for the different product and service versions and the knowledge of factors/levels that constitute each one of these versions, to determine the contribution or the value of factorial levels (Carneiro et al., 2003).

Therefore, the objective of the present research was s to identify the impact of the attributes of soy beverage packages original flavor on product purchase intent, and to propose a package that can influence this purchase positively.

Materials and Methods

Participants were informed about the procedures, as registered in the National System of Ethics in Research (Certificate of Presentation for Ethical Appreciation n° 0142.0.268.000-09).

2.1 Definition of attributes and their levels

To define the attributes and levels applied during the conjoint analysis (CA), focus group sessions were conducted, as suggested by Della Lucia, Minim and Carneiro (2010). During the interview, five packages of Brazilian plain soymilk, from different brands, were presented sequentially to the participants. The packages were Tetra Pak Brik Squareline® type carton (square bottom), with a plastic lid with an open-and-close system and aluminum seal (Tetra Pak® pull tab system with recap). In the supplementary material (Annex 1- Table 1) there is a description of the five packages used in the research. The focus group consisted of 7 sessions of approximately 90 minutes, in which 41 individuals participated, selected and characterized from questionnaires collecting personal data, frequency of soy drink consumption and reading of food product labels, in addition to habits of purchase. The groups were divided according to the characteristics of the participants, so that the sessions represented their opinion, and information of interest to the product was obtained. Participants were encouraged by the moderator to express opinions. In order to standardize the conduct of the sessions, we used some questions, made without order specificity, according to the context and group dynamics: a) what did you think of this package?; b) what attracts your attention about this packaging and label?; c) Do nutritional information and ingredients influence your purchase intention?; d) Do you like to see the expression “0% cholesterol, 0% lactose” or “lactose-free, cholesterol-free” on packaging?; e) What do you mean by original flavor?; f) How to inform on the label that this product is fortified/enriched?; g) Which nutrient do you believe is most important to be highlighted on the label of this product?; h) Would you like to see any other information on the label of this product?; i) Would you like to see any changes in terms of packaging design (shape, material, cover)?; j) Would you like to see any changes in terms of label design (colors, figure, font)?

At the end of each session, all packages were evaluated together and each participant made their own general evaluation, justifying their criteria. In each session, the data obtained through the answered questionnaires and the opinions of the participants were recorded, analyzed and discussed considering the words used by consumers, the context of the question and the specificity of the answers and, when possible, represented by the use of percentage values.

2.2 Conjoint analysis: experimental design, packaging development and evaluation of results

From the data collected through the focus group, 8 packaging proposals for plain soymilk were elaborated, corresponding to a fractional factorial design with four factors and two levels (2^4-1). The following factors were used: a) brand, b) type of cup illustration: cup with product projection coming out of the cup, in the form of drops and splashes (denoted in this work as splash) or, cup filled with product, c) presence/absence of nutritional claims (more nutritious, with calcium and vitamins) and, d) background color (light/dark), as shown in Table 1. The images were assembled using the PowerPoint® and Paint® programs, defining as a packaging proposal only the image of the front panel.

Table 1: Experimental design of plain soymilk. 

For the ACF, 86 participants were used, in sessions with a variable number of participants, according to their availability. The sessions were conducted in a classroom containing a computer connected to multimedia projection equipment, at the Federal Institute of Paraná (Umuarama/Brazil). At each session, the order of presentation of the images was randomized and the images coded by 3 digits randomized.

Participants were asked to behave as if they were in a supermarket and needed to buy plain soymilk, and that, for each image presented, they wrote down the sample number and registered the respective purchase intention. Initially, the 8 proposed packages were presented on the same slide for 15 seconds, so that participants could have an overview of the products. Subsequently, packages coded with 3-digit numbers were presented sequentially for 30 seconds, using a 10-second interval of white screen between each image. The order of presentation was randomized in each session. The sample number was noted for each packaging image, and the purchase intention of each participant was evaluated on a scale. To assess purchase intention, a 7-point structure scale was used, anchored at the lower end with the expression “definitely not buy” and at the upper end “I would definitely buy”.

The results were tabulated as grades (1 to 7) for each package evaluated, in a table of double entry of consumers versus packages and analyzed using the SAS program (SAS INSTITUTE INC, 1996) as suggested by Carneiro et al. (2005). For the assessment of purchase intention, preference coefficients were calculated for all consumers, in an individual analysis, according to Corso and Benassi (2015), through an additive model, where the general assessment of preference (purchase intention ) was formed by the sum of the contributions at each of the levels of each factor (packaging attributes).

Subsequently, consumers were grouped into groups through cluster analysis, using the linkage average method and the Euclidean distance as a measure of similarity. Based on the estimated preference coefficients for the groups, the relative importance of the variables on the purchase intention of consumers in each group was evaluated. Clusters formed by a very small number of consumers, that is, that did not represent groups, were repositioned in other larger clusters that presented similar preference behavior. The Fischer and Tukey tests were applied to the results in order to verify if there was a significant difference between the coefficients for each attribute within the group (p<0.05).

Results and discussion

3.1 Focus group

Consumers were predominantly women (79%); the participants had between 40 to 69 years of age (59%), with high educational attainment (22% with graduate degrees and 24% with post-graduate degrees) and with incomes between one and ten times the Brazilian minimum wage (91%). The majority of the participants (42%) was responsible for purchasing food for their homes. Participants (90%) bought soy beverages with juice, plain or light plain, and the average consumption of soy beverages by the team was 1.34 liter/ month.

Most participants (61%) always or often read the labels of the products they consumed and reported in the questionnaire that the attributes price, brand, shelf life, and nutritional information were observed on the product label. According to Wardle, Parmenter, and Waller. (2000), Cowburn and Stockley (2005), Ares et al. (2008b), Karamravan, Koohi and Abbaszadeh (2014), nutritional knowledge can have an influence on eating behavior, influencing consumer choices

Studies show a relation between soybean-based food consumption and nutritional knowledge level (Wansink; Westgren; Cheney 2005). In relation to the characteristics observed at the moment of purchase, most consumers mentioned: a) price (more than 70%), followed by: b) shelf-life (58%), c) brand and d) presence of nutritional claims (47% and 41%, respectively).

According to participants, product brand was related to both previous positive purchasing experiences and preferences for the background color of the main panel. Similarly, Ares et al. (2011) verified that, for Spanish consumers, the commercial brand of a yogurt was a relevant factor.

Another factor mentioned was the preference, of some participants. for packages with a lighter background or, of other ones, for packages with a darker background. Another characteristic reported as attractive was type of illustration showing a glass filled with the product printed on the package´s main panel. Some participants mentioned as positive the presence of a glass filled with the product and others showed preference for the illustration showing a glass being filled with the beverage, with product splashes. According to Ares et al. (2008a), food companies need to differentiate their products to make them more attractive to consumers. This way, different illustrations and colors used on the packages may promote interest in the acquisition of the product by the consumer. A summary of the main participants´ comments is presented as supplementary material (Annex 2- Table 2).

There were several reports on the preference for products with nutritional claims since most participants purchased soy beverages for health reasons, which for them “are much healthier than milk drinks”.

Since most participants mentioned the attribute "lower price" and the fact that the product's expiration date is a legal requirement, we decided to vary following factors into two levels: (1) brand, (2) illustration, (3) presence of nutritional claims and, (4) package´s background color.

3.2 Factorial conjoint analysis

The group of participants were balanced in gender (51% women); they were young (85% between 15 and 35 years of age) and were responsible for purchasing food for their homes (98%) and frequently read product labels (63%).They had a high level of education (43% with a college degree), and income of up to five times the Brazilian minimum wage (67%).

After the visual assessment of the images by the participants and data analysis, data from one participant who had assigned the same grade to all packages were removed, as suggested by Carneiro et al. (2003). The remained data were assessed individually and showed p> 0.15 e R²>0.66.

Participants were grouped by similarities and dissimilarities into three (3) main groups with 48, 31 and 6 participants, respectively. Table 2 presents the results of the aggregated analyses for each group.

Table 2: Result of aggregated analyzes for each group. 

One aspect previously raised during the focus group sessions and cited again during the CFA was that product price was a relevant aspect and that purchase decision was based on low product values.

In general, all groups (Table 2) pointed out as positive the presence of nutritional claims on the package, differentiated by certain visual characteristics, which would possibly influence purchase decision. CrillyMoultrie and Clarkson(2004) show that there is a major influence of the package on the purchase decision process, since it attracts consumers' attention to products, conveying messages about the product in a graphic form (Moskowitz et al., 2009) serving as product recognition, as well as information on lifestyle (Van Dam; Van Trijp, 1994).

For Group 1, representing 56.5% of the assessors, purchasing intent was affected significantly by the presence of a nutritional claim, the brand and the main panel with a light background. As for Group 2, the characteristics that made greater impact on purchasing intent was the presence of a claim (74.31%), plus the brand and the filled-up glass illustration (17.36% and 7.29%, respectively). Purchase intent in Group 3, representing 7% of the participants, was affected by all the attributes studied. In this group, however, type of glass was the most important factor and the filled-up glass illustration made a positive impact on purchase intent, as well as the less known brand.

Other aspects that may be taken into account in the analysis are that participants were usual consumers of soy beverages and that the images generated contained visual elements similar to those found in commercial samples, i.e., these elements were already known, consciously or unconsciously, by the participants, raising expectations regarding the products. In a review on the subject, Deliza and MacFie (1996) comment on the expectations raised by the consumers´ previous experiences with the product, information on the label, packaging characteristics and the product itself, mainly its appearance. Therefore, it is highly relevant for the first experience with the product to be positive, meeting all the sensorial and hedonic expectations of the consumer.

Most participants considered the presence of nutritional appeals an important decision-making factor in purchasing soybean-based beverages. Thus, although consumers had an interest in consuming functional food products, the development of persuasive health allegations and adequate marketing strategies is very difficult, as pointed out by van Kleef, van Trijp e Luning(2005b). Therefore, the adequacy of packaging and product to meet consumers´ expectations may enhance the success of a given product (Chung et al., 2011). According to Ares et al. (2008a) the inclusion of information on the source of ingredients may raise negative expectations in relation to the flavor of soy beverages, and special attention should be given to the most adequate way to include this information in the label. Thus, when consuming food, not only health factors will be perceived, but, as in any conventional food, factors such as sensorial quality, price and convenience will also influence consumers´ attitudes, which are segmented according to their attitudes and preferences. The identification of these segments may allow us to target different types of products to each segment (Ares et al., 2010b).

Conclusion

By associating qualitative information from the focus group and the joint factor analysis, some parameters stood out as being important for the purchase decision process of original flavored soy beverages, although the weight of these attributes is different depending on the consumer.

The presence of some nutritional claim is considered an important aspect for the originally flavored soybean-based beverage acquisition process. The “glass containing the bottled beverage” illustration, the tone of the package background and the brand are factors that segment consumers during the purchase of original flavored soybean beverages.

In general, the package with the presence of a nutritional claim, a filled-up glass and a light background has a higher probability of increasing purchase interest by consumers.

Based on this information, it is possible to conclude that the objective of different package layouts for original flavored soybean-based beverages is to meet the needs of several consumer segments thus justifying the presence of several information and layout options on the product´s packages.