Analysis of the moderating effect of front-of-pack labelling on the relation between brand attitude and purchasing intention

Depósito de Investigación de la Universidad de Sevilla https://idus.us.es/ This is an Accepted Manuscript of an article published by Elsevier in Journal of Business Research, 2021 Vol. 122, on January, available at: https://doi.org/10.1016/j.jbusres.2020.08.062 Copyright 2020 Elsevier. En idUS Licencia Creative Commons CC BY-NC-ND 1 ANALYSIS OF THE MODERATING EFFECT OF FRONT-OF-PACK LABELLING ON THE RELATION BETWEEN BRAND ATTITUDE AND PURCHASING INTENTION Abstract Nutritional labelling is an instrument to generate eating patterns. The establishment of means of support to the nutritional labelling via a front-of-pack (FOP) labelling which facilitates the consumers’ comprehension and improves their choices is encouraged. The aim of this work is to establish the impact that FOP labelling has on the relation between brand attitude and purchase intention, as well as to determine the effect of gender on these relations. Two surveys have been done. The first (129 responses) analyses the relation between brand attitude and purchase intention without the presence of FOP labelling. The second (111 respondents) presents the FOP labelling. Multigroup analysis establishes the existence of differences in the relation between brand attitude and purchase intention due to the existence of FOP labelling in the case of the male sample. The effectiveness of FOP labelling is conditioned by the consumers’ brand attitude and by the impact of gender. Keywords: front-of-pack labelling; brand attitude; purchase intention; gender. 1. Introduction. With numerous consumers suffering from health problems because of a poor diet, the analysis of nutrition-related elements becomes relevant. An approach normally used to face such problems is to offer information concerning the nutritional content of foods (Ikonen et al., 2019). Faced with the need for consumers to receive this information before making decisions, nutritional labelling has arisen as an instrument which seeks to educate, enable and empower them, disseminating easy-to-read nutritional characteristics that help to elaborate healthier diets, having a significant effect on the choice of foods (Bauer and Reisch, 2019; Brambila-Macias et al., 2011; Gómez et al., 2017; Grunert and Wills, 2007). As a response to consumers’ uneven use and comprehension of nutritional labelling, public and private initiatives (manufacturers and distributors) have developed different versions of Front-of-Pack (FOP) labelling with the aim of increasing the ease of use of nutritional information and of helping consumers in their shopping process via the simple presentation of information which supplements back nutritional labelling (Bauer and Reisch, 2019; Gómez et al., 2017; Ikonen et al, 2019; Newman et al., 2018; Smed et al., 2017). FOP can be reductive or interpretative. Reductive labelling has a basic quantity, or key elements, of specific information of the back nutritional panel with little interpretation. Interpretative labelling (e.g. NutriScore), carries out a general evaluation of the healthiness of the product and offers a simplified and interpretative compiling of the nutritional information which can help consumers in their evaluation (Ikonen et al, 2019; Newman et al., 2014, 2018). In spite of the benefits indicated, these FOP labels are criticised for excessively simplifying nutritional information, shifting it to a unique indicator which could confuse the consumer and prompt a wrong evaluation, and limiting its capacity to lead consumers towards healthier choices (Ikonen et al., 2019; Newman et al., 2018). 2 However, despite the fact that many countries are betting on them, there is no clear understanding of the way in which interpretative labels can be effective, since there are factors that frustrate the objective of guiding consumers to choose healthier foods (Vizcaíno and Velasco, 2019). Among which are that consumers ignore nutritional information because the visual environment acts as a barrier to attending said information (Orquin et al., 2020), that other labels with “natural” characteristics counteract the impact of graphic labels (Davis and Burton, 2019); or that familiarity and trust towards the brand protect the product from being evaluated as unhealthy, even in the presence of an interpretive FOP (Vizcaíno and Velasco, 2019). The nutritional information must be provided bearing in mind that decisions regarding to food are influenced not only by the informational content but also by preferences, attitudes and intentions towards the foods (Bauer and Reisch, 2019; Bucher et al., 2016; Gómez et al., 2017; Limbu et al., 2019). Given that attitude is the most important predictor of purchase intention, it must be considered whether the labelling has the impact which is supposed in this relation (Kozup et al., 2003; Rana and Paul, 2017; Woo and Kim, 2019). Bearing in mind that authors state that declarations about the healthiness of products do not moderate the relation between the perceptions of the foods and the purchasing intention (Cermin et al., 2019; Davis and Burton, 2019). Studies acknowledge the differing role of gender in response to marketing strategies (Meyers-Levy and Loken, 2015; Salnikova and Grunert, 2020). Within the theories developed to explain the differences in behaviour between the different genders in information processing, the selectivity hypothesis postulates that women tend to process data more comprehensively and have a lower threshold from which they capture information. The ignorance of the factors that can affect the impact of NutriScore on the attitude and intention to buy food, as well as defining in a clear way the role which play gender in this relationship is the gap that this paper aims to cover. Bearing the previous elements in mind, the work goes deeply into the knowledge about the impact that FOP labelling has on food purchase attitude and intention. Likewise, it analyses the role played by gender, facing the controversy as to the role which this variable develops in the relation between FOP labelling, attitudes and food purchase intentions. Several works establish that the gender influences the search for and use and comprehension of labelling information (Grunert and Wills, 2007; Gomes et al., 2017). Section 2 develops the theoretical framework. It presents the theories from which the current work begins, as well as the hypotheses established. Section 3 describes the method, variables, constructs and metrics, and the description of the sample and sample processing. Sections 4 and 5 reveal and discuss the results. Section 6 sets out the study’s conclusions and main contributions. 2. Theoretical framework. 2.1. The relationship between attitude and purchase intention. Two are the most widely used models to explain consumer behavior, the theory of reasoned action (Ajzen and Fishbein, 1980; Fishbein and Ajzen, 1975) and the theory of planned behavior (Ajzen, 1991). The theory of reasoned action (Ajzen and Fishbein, 1980; Fishbein 3 and Ajzen, 1975) states that the intention of behavior (immediate antecedent of the behavior), is a function of the beliefs that developing a certain behavior will lead to a certain result. Fishbein & Ajzen (1975) divide antecedent behavioral beliefs into behavioral and normative beliefs. Behavioral beliefs are the underlying influence on the attitude of developing behavior, while normative beliefs influence the subjective norm of the individual regarding developing behavior. According to the theory of planned behavior (Ajzen, 1991), intentions to perform behaviors can be predicted from attitudes toward behavior, subjective norms, and perceived behavioral control; and these intentions, along with perceptions of behavioral control, explain actual behavior. In other words, attitudes are linked to behavioral intention. Attitudes are derived from beliefs accessible in memory and guide consumer behavior, and may be affected by the context in which they are expressed (Ajzen and Fishbein, 2000). Attitude is an antecedent of intention, as well as the degree to which an individual has a previous valuation of possible behaviour concerning any shopping situation (Ajzen, 1991). The attitude associated with a brand is a global evaluation based on reactions to stimuli or beliefs related with it, being a central component between relational exchanges which includes cognitive, affective and behavioural intentions. Attitudes are stable and lasting predispositions, being considered the most important predictor of the consumer’s behaviour. A positive attitude towards a brand influences a customer’s purchase intention (Foroudi et al., 2018; Kudeshia and Kumar, 2017; Olsen et al., 2014; Schivinski and Dabrowski, 2016). The purchase intention is linked with actions which are foreseen or planned for the future, and is the predisposition to transform beliefs and attitudes towards a product into actions (Kudeshia and Kumar, 2017; Schivinski and Dabrowski, 2016; Woo and Kim, 2019). Attitude is the most important predictor of the intention of buying healthy food (Kozup et al., 2003; Rana and Paul, 2017; Woo and Kim, 2019). 2.2. FOP labelling in the relationship between attitude and behavioural intention Statements relative to the healthiness of the products can positively influence their evaluations and choices (Berry et al., 2017; Chrysochou and Grunert 2014). Although attitudes are relatively long-lasting, they can be modified through marketing actions which emphasise the product’s specific attributes (Olsen et al., 2014). The informational elements (information and labels) of the package, stimuli linked to intrinsic or extrinsic characteristics of the products, are the most explicit way of communicating healthiness (Festila and Chrysochou, 2018; Salnikova and Grunert, 2020). In foods, labelling is one of the essential instruments for consumers to be able to find out about their content, enabling making appropriate decisions based on information related to their healthiness (Kumar and Kapoor, 2017). Food labelling influences the purchase through objective and subjective consumption indications (Deng and Srinivasan, 2013). The subjective signs of the labelling, such as nutritional declarations, can influence the amount that a person adopts as reasonable consumption, or the pleasure or guilt linked with it (Bauer and Reisch, 2019). The attitudes of consumers towards products can improve when there are favourable nutritional declarations on the package. The effect of this certification is greater when it is done through third parties, as consumers trust the evaluation of independent agents more (Bezawada and Pauwels, 2013; de-Magistris et al., 2017; Scarpato et al., 2017). According 4 to this, it could be stated that the existence of favourable nutritional declarations leads to positive attitudes towards the product and boosters the purchase intention (Berry et al., 2017; Cermin et al., 2019; Chrysochou and Grunert, 2014; Kozup et al., 2003; Rana and Paul, 2017; Olsen et al., 2014). FOP nutritional labelling, given its easy interpretation, influences consumers’ purchasing intentions (Gómez et al., 2017; Newman et al., 2014), a statement which is verified for interpretative labels, such as Nutriscore (Ikonen et al., 2019). Since the literature indicates that the existence of interpretative labelling (such as NutriScore) influences the attitude and intention to buy food, it is appropriate to propose the first working hypothesis in the following terms: H1: the presence of FOP labelling moderates the relation between brand attitude and purchase intention. 2.3. The selectivity hypothesis and the impact of the FOP in the relationship between attitude and purchase intention Studies acknowledge the differing role of gender in response to marketing strategies (Meyers-Levy and Loken, 2015; Salnikova and Grunert, 2020). Within the theories developed to explain the differences in behaviour between the different genders in information processing, the selectivity hypothesis postulates that women tend to process data more comprehensively and have a lower threshold from which they capture information. This makes women more inclined to detect, elaborate and employ information that is less accessible and more relevant when making evaluations. For their part, men are more selective data processors and have a greater trust in heuristics of less effort, which often depend on relevant signals or that activate developed and easy access preconceptions (Meyers-Levy and Lokem 2015). A corollary to the selectivity hypothesis is that men are more likely to simplify food purchasing decisions through the use of heuristics (Mead and Richerson, 2018; Meyers-Levy and Loken, 2015). Consumers place more trust in memorybased information than in context-based information. When memory-based information is more accessible, contextual information will not be used to make judgments (Ahluwalia and Gürhan-Canli, 2000; Menon et al., 1995; Vizcaíno and Velasco, 2019). The heuristics applied to decision-making involves making rapid evaluations of the healthiness of food based on different elements of the product (Bui et al., 2017; Mead and Richerson, 2018), with men being more prone to its use (Meyers-Levy and Loken, 2015). Gender significantly influence the search for, evaluation, use and comprehension of information related to the food included in the labelling (Cheng and Liu, 2019; Grunert and Wills, 2007; Oliveira et al., 2018; Skubic et al., 2019; Gomes et al., 2017). In fact, women prefer foods with labelling, while men prefer foods without them (Smed et al., 2017; Skubic et al., 2019). Women use nutritional labels to a greater extent than men, being more concerned about the nutritional content of foods and this influences their purchase intention more (Adasme-Berríos et al., 2016; de-Magistris and Gracia, 2016; Kumar and Kapoor, 2017). The effectiveness of the message’s setting, as well as that of the visual elements could differ, therefore, with sociodemographic characteristics, such as gender (Bauer and Reisch, 2019; Puccinelli et al., 2013). This leads to the following hypothesis: H2: the presence of FOP labelling does not moderate the relation between brand attitude and purchase intention for males. 5 H3: the presence of FOP labelling moderates the relation between brand attitude and purchase intention for females. 3. Methods 3.1. Constructs and variables The items used to measure the constructs incorporated into the model come from Kudeshia and Kumar (2017) and from Schivinski and Dabrowski (2014), 5 items being used to measure brand attitude and 3 for purchase intention. To avoid the valuation of the FOP influencing the assessment, each questionnaire presented five products of the brand being analysed, one for each valuation (from A to E). Likewise to determine the impact of the presentation and inclusion of FOP on the package, two questionnaires was designed. One questionnaire was elaborated considering NutriScore evaluations and the other one without them (control group). The use of a control group to see the effectiveness of labels has been widely used (Bauer et al., 2013; Cermin et al., 2019; Chrysochou and Grunert, 2014; Deng and Srinivasan 2013; Hasford et al., 2019). The questionnaire was completed with questions which enable presenting the profile of the sample: gender, age, frequency of consumption of the category of the product, and size of the home municipality. The Nutriscore valutions of the products employed were obtained from https://es.openfoodfacts.org/. Insert here figure1 3.2. Sample and data The fieldwork was done between 11/11/2019 and 4/12/2019, links to the quesionnaires being given to university students. 129 people answered the questionnaire in which the FOP labelling was not presented and 111 that which presented FOP labelling. In the questionnaire which presented FOP, an explanation was given of its meaning to overcome any possible ignorance of the respondents. Nutriscore was opted for as it is the type of summary interpretative FOP, that is already being used in Spain and will be mandatory from the last quarter of 2020. The table below (Table 1) shows the profile of the sample by gender, age, frequency of consumption and size of home municipality. The sample has a balanced gender distribution, an average age close to 25 years old and a majority of respondents who consume yoghurts 4 days a week and live in cities of more than 100,000 inhabitants. Insert here table1 To respond to the proposed hypotheses, six structural models were used which analysed the relation between brand attitude and purchase intention: (1) global sample without FOP (GloNoFOP); (2) global sample with FOP (GloInFOP); (3) male sample without FOP (MalNoFOP); (4) male sample with FOP (MalInFOP); (5) female sample without FOP (FemNoFOP); (6) female sample with FOP (FemInFOP). Structural Equation Models through Smart PLS3.2.8 were employed for the analysis of the results via the application of 6 the PLS Consistent algorithm (Ringle et al., 2015). Likewise, RCommader was used to do the T and F tests. 4. Results 4.1. Analysis of the measurement model We began the analysis of the measurement model with the study of the reliability of the indicators. As the table below (Table 2) shows, all the indicators surpass 0.707 for their corresponding constructs, except for the models exclusively aimed at women, where the indicator AM2 had to be eliminated in its corresponding models ("This brand has a good reputation") as in neither of the two cases did it reach the level required. Insert here table2 As to the reliability of the constructs, they all surpass the strict criteria for the Cronbach alpha (0.8/0.9), as well as for the Dijkstra-Henseler indicator (rho_A) and composite reliability (0.7). Likewise, the models have convergent validity as they are over the required threshold for the AVE (0.5). Finally, as approximate fit measurements of the model, the SRMR values are, in all cases, less than 0.08 (Table 3). As can be seen, values below the threshold correspond in the six models: CloNoFOP (0.044); GloInFOP (0.035); MalNoFOP (0.029); MalInFOP (0.012); FemNoFOP (0.049); FemInFOP (0.048). Insert here table3 The table below (Table 4) shows that the models surpass the criteria to establish discriminant validity. So, although the model MalInFOP has a value over that established as appropriate (0.85-0.90), its discriminant validity can be accepted based on the confidence interval as none of the intervals includes the value 1. Insert here table4 4.2. Analysis of the structural model Firstly, the possible collinearity of the models based on the VIF value is ruled out. Likewise, the R2 surpasses the value required in all the cases, approaching a substantial value (0.65) in five of the models. The f2 exceeds the threshold of 0.35. Finally, the significant result ofthe relation between brand attitude and purchase intention in the six models is pointed out (Table 5). Insert here table5 As the table above indicates, the relation between brand attitude and purchase intention is significant, both based on the p-value and on the confidence intervals. 4.3. Analysis of the interaction effect Since the NutriScore value (A to E) works as an interval scale, we began by studying the interaction effect. Hence, the valuation (from A to E, codified as 1 to 5) which the FOP 7 label presented was taken as the moderator variable. Through the present analysis, it can be established whether the different values of the moderating variable influence the relationship between attitude and purchase intention. The interaction effect was carried out to measure the impact of FOP on the relation between brand attitude and purchase intention. The interaction effect of the presence of FOP in the relation analysed for the three models which include it was studied (GloInFOP, MalInFOP, FemInFOP). Table 6 shows that in no model does the presence of the labelling moderate the relation (t statistic below 1.63). Insert here table 6 4.4. Multigroup analysis Multigroup analysis was used to establish if there is a difference between the models. The models’ hypotheses with various groups which seek the comparison of intensity between them can be tested when statistically comparing the paths of the structural models for each sample. Thus, the models with and without NutriScore were compared for the global sample, men and women. According to this procedure (t-Student test), the result for the comparisons in the global models (t= -0.8206) and in the women’s models (t=1.0847) show the absence of a moderator effect. There is a difference for the case of the men’s model (t= -3.1096). According to these results, the presence of FOP labelling causes differences in the management model of men in the brand attitude/food purchase intention relation. 5. Discussion The results confirm, in line with previous works, the influence of brand attitude on food purchase intention (Kozup et al., 2003; Rana and Paul, 2017; Woo and Kim, 2009), this relation being significant in all the models elaborated. Attitudes are therefore reaffirmed as stable and lasting predispositions, being considered the most important predictor of behaviour (Forouidi et al., 2018; Kudeshia and Kumar, 2018; Schivinski and Dabrowski, 2016). The results establish that the mere presence of FOP nutritional labelling does not moderate the relation between brand attitude and purchase intention. Through the interaction effect, it has been seen that in the global sample (t-statistic = 1,046), as well as in men (t-statistic = 0.335) and women (t-statistic = 1,260), the NutriScore values (1 to 5 ) do not affect the strength of the relationship between attitude towards the brand and purchase intention. Likewise, when comparing this relationship based on the presence or not of NutriScore, we observe that there are no significant differences (t = -0.8206) either. In other words, neither the assessment presented by NutriScore, nor the mere presence of said labeling, affects the relationship between brand attitude and purchase intention, so H1 is not being accepted. These results position us in line with previous works which question the influence of FOP labelling to lead consumers towards healthier choices (Ikonen et al., 2019). A possible cause of this absence of impact is the fact that interpretative systems, such as is the case of Nutriscore, are more useful when carrying out a comparative process between different options (Newman et al., 2014). Likewise, another reason for the absence of effect of the NutriScore values in the relationship between the attitude towards the brand and the 8 purchase intention is that we find a sample with a high rate of consumption of the products studied. It should be remembered that familiarity and trust towards the brand protect the product from being evaluated as unhealthy based on the presentation of an interpretative FOP (Vizcaíno and Velasco, 2019). After verifying the non-existence of an effect of FOP labelling on the relation between brand attitude and purchase intention, we went deeper into the impact that gender can have on them. In this line, the selectivity hypothesis postulates that women and men differ in information processing (Meyers-Levy and Lokem, 2015). According to the results obtained, the presence of interpretive nutritional FOP does not affect the relationship between the attitude towards the brand and the intention to buy yogurts in the case of women, thus accepting H2. Therefore, it is verified how women are more likely to detect and use more relevant information in their evaluations (Meyers-Levy and Loken, 2015). In this way, if, as the literature has exposed, women read nutrition labels to a greater extent (Adasme-Barrios, et al., 2016; de-Masgritis and García, 2016; Kumar and Kapoor, 2017), said labeling does not affect the way in which the attitude towards the brand and the purchase intention are related, perhaps motivated by the relevant information indicated before. In the present case, and considering the results of the multigroup analysis, it is noted that the presence of FOP labelling moderates the relation between brand attitude and food purchase intention for the case of the male sample, H3 being accepted. In other words, men simplify decisions by using heuristics, supported by signals from the packaging to inform food purchase decisions (Mead and Richerson, 2018; Meyers-Levy and Loken, 2015). Thus, it is verified that the impact of the visual elements differs based on gender (Bauer and Reisch, 2019; Puccinelli et al., 2013). This fact places us in line with the selectivity hypothesis, since the difference in the response to gender-based marketing strategies has been verified (Meyers-Levy and Loken, 2015; Salnikova and Grunert, 2020). This is accompanied by the greater propensity of men to simplify decisions through the use of heuristics (Mead and Richerson, 2018; Meyers-Levy and Loken, 2015). 6. Conclusions and contributions 6.1. Conclusions At a time when there are debates concerning the most effective FOP nutritional labelling system, it is especially relevant to contribute evidence about the impact that this has on consumers. This is why the present work has been developed with the aim of analysing the impact that this labelling has on the relation between brand attitude and purchase intention. Firstly, it is verified that the presence of FOP labelling does not modify the relation between brand attitude and food purchase attitude. Yet, gender-based analysis has been incorporated with the aim of going deeply into this effect. Hence, it is established that the behaviour of this relation referred has an uneven gender-based behaviour as there are significant differences motivated by the presentation of FOP labelling for the male sample.