1
Should brands fear the traffic light format?
An investigation of the effects of the color-coded and the monochrome GDA nutrition label on (subjective) label comprehension, label liking, brand credibility, brand attitude and purchase
intention, and the moderating effect of nutrition knowledge.
Student Dewy Gerritsen 10423486
Supervisor Dr. G.J. de Bruijn
Master’s Thesis Persuasive Communication
Graduate School of Communication June 26 2014
2
ABSTRACT
The aim of this study was to explore the different effects of the monochrome GDA and the color-coded (traffic light) GDA label on label comprehension, label liking and the ultimate effects on brand credibility, purchase intention and the attitude toward the brand that presented one of the labels. The set up of this study is based on two streams of research: consumer decision making and attitude formation and change. In a self-administered survey, participants evaluated either the monochrome or the color-coded GDA food label on
subjective (perceived) and objective (actual) comprehension, perceived healthfulness and label liking, and evaluated the brand in terms of brand attitude, brand credibility and purchase intention. The moderating role of nutrition knowledge was also taken into account to
determine whether the effects between label type and comprehension differ when consumers have high or low nutrition knowledge. Results reveal an overall preference for the color-coded GDA label. More specifically, the color-color-coded GDA label is rated significantly higher for perceived label comprehension, and is significantly more liked than the monochrome version. There was found no moderating effect for nutrition knowledge on the relationship between label type and comprehension. Additionally, participants who saw the product in combination with a color-coded GDA label were more likely to purchase the brand in the future than participants who saw the monochrome GDA label. Moreover, the results indicate a significant positive relationship between subjective label comprehension and label liking. In turn, a more positive label attitude shows to have a positive effect on brand attitude, brand credibility and purchase intention. Altogether the current study provides valuable and remarkable new insights on label type preference, and what effect this ultimately has on the brand showing the specific label.
3
INTRODUCTION
Although food packages are nowadays filled with health logo’s, guidelines of daily values and nutrition tables, all this information does not seem to enable the consumer to make deliberate food choices. In a survey performed by the Dutch documentary ‘Altijd wat’ (Sommer, 2011), Dutch consumers stated that they felt the need to bring a calculator and encyclopedia to the supermarket, to make some sense of all the nutrition information presented on food packages. Consumers seem to call for clear, easy to interpret nutrition labels, which enable them to make product comparisons in shopping situations in which they are generally limited in time and opportunity to process nutrition information (Feunekes, Gortemaker, Willems, Lion, & Van Den Kommer, 2008). Is there a food label format that meets these needs?
A potential solution to devious and difficult to interpret food labels, could simply be the addition of traffic light colors. Research has shown, that traffic light food label formats are better understood across all socioeconomic groups, than the current widely applied
monochrome Guideline Daily Amount (GDA) label (Malam, Clegg, Kirwan, McGinigal, Raats, Shepherd, & Dean, 2009). Moreover, adding traffic light colors to nutrition labels, generally has a positive effect on the evaluation of the label (Grunert & Wills, 2007; Malam et al., 2009) and 65% of the Dutch consumers would vote in favor of the introduction of this traffic light system (Sommer, 2011). Despite these positive evaluations, on June 17 2010 the European Parliament rejected the introduction of the traffic light system in the EU (Euractiv, 2010). The Corporate Europe Observatory, a NGO, stated that just before the voting in the European Parliament started, the food and beverage industry invested more than one billion euro’s in a lobby campaign to block the introduction of the traffic light system and to promote the monochrome GDA system in the EU (Corporate Europe Observatory, 2010). This
campaign, which has been described as ‘the biggest campaign seen in Brussels in recent years’, seems to be the result of a fear amongst food brands that consumers will tend to avoid products with amber and red label elements (Euractiv, 2010). But is this fear a founded fear?
A study that could be a tentative indication that traffic light colors should not be predominantly feared by brands, is the study performed by Sacks, Rayner and Swinburn (2009), which examined the changes in sales of ready meals and sandwiches in the UK after the introduction of the traffic light format. These products differentiated in their healthiness and showed different assemblies of green, amber and red elements. The results showed that whereas sandwiches did not significantly decrease or increase in sales, the sales for
ready-4 meals even increased with 2.4% in the four weeks after the introduction of the traffic light label. Moreover, of equal importance is the fact that there was no association found between product (un)healthiness and change in sales (Sacks et al., 2009).
Of course there can be several unnamed factors that could be accountable for these effects. However, to my opinion the outcome is quite remarkable and deserves further investigation. While reading about the traffic light format, the sense arose that if traffic light colors actually are evaluated more positively than the labels that are currently presented by brands , these positive evaluations could well extend to the product, or ultimately even to the brand and purchase intentions consumers have. Previous research has found that food
information on packages that affects nutrition perceptions, extends to overall product attitudes and purchases intentions (Kozup, Creyer, & Burton, 2003; Andrews, Burton, & Kees, 2011). However, to my knowledge, no research has been done regarding these effects, let alone how these effects differ between the widely applied monochrome GDA nutrition label and traffic light formats. The current study therefore aims to explore differences in (perceived)
comprehension for the monochrome GDA and the traffic light GDA label, and what effect this ultimately has on label liking, brand credibility, purchase intentions and brand attitude consumers experience. Also, the moderating role of nutrition knowledge will be taken into account, to determine whether the effects between label types differ amongst consumers with high or low nutrition knowledge. As mentioned, beside the outcomes on brand attitude and purchase intention , the differences for brand credibility between the monochrome and traffic light GDA label will be examined. Studies on the effects of brand credibility on consumer brand choice and customer loyalty (Sweeney & Swait, 2008; Baek, Kim, & Yu, 2010) show multiple positive influences of brand credibility: it enhances brand loyalty, positive word of mouth and reduces switching behavior. All together, the results on these outcome variables (brand attitude, brand credibility and purchase intention) can be of great relevance for brands in the FMCG market that are already presenting the monochrome label, for who establishing a strong, long term brand-consumer relationship is desired and often even crucial.
5
THEORETICAL BACKGROUND
The specific labels that will be examined in this study are the monochrome GDA (Figure 1) and the color-coded GDA label (far right format in Figure 2).
Monochrome GDA format
I will start off with the most common label, the monochrome GDA (Figure 1), which is already presented on food and drink packages by numerous major brands such as Coca Cola, Maggi, Quaker, Snickers and Van Gilse. The monochrome GDA label presents five nutrients (calories, sugar, fat, saturated fats, salt) and their content levels: in absolute values and in
percentages of the recommended daily intake quantities within one’s diet. The percentage
reference quantities are intended to serve as benchmarks for consumers and are based on European directives (Food and Drink Federation [FDA], 2012). Therefore, the application of the GDA label is relatively consistent amongst all kinds of food and drink products. In the current study, this ‘one-colored’ format will be referred to as the monochrome GDA, in line with Aschemann-Witzel, Grunert, Trijp, Bialkova, Raats, Hodgkins and Koenigstorfer (2013).
Figure 1. An example of the monochrome GDA label.
Traffic light formats
The ‘traffic light system’ is a variously applied system for color coding nutrient amounts, but is consistent regarding the colors that are applied: green, amber and red (see Figure 2 for several examples of this format). Previous studies have examined an awful lot of different assemblies and designs of the traffic light system on their use, comprehensiveness and liking (Feunekes et al., 2008; Grunert, Wills, & Fernández-Celemín, 2010; Van Herpen, Hieke, & Van Trijp, 2014). Therefore, when comparing traffic light based systems, the results are not clear-cut. What we do learn from these previous studies are three important things: consumers
6 tend to avoid cognitive load when interpreting nutrition information; traffic light systems seem to outperform monochrome GDA systems when it comes to simplicity; and consumers generally like simplification of nutrition information (Grunert & Wills, 2007). However, studies on liking of food labels, surprisingly show that simple traffic light formats (see far left image in Figure 2), which present nutrients in combination with color, but lack of reference information such as absolute numbers or percentages of daily values, were mostly less liked (Unilever, 2006).
Simple traffic light Multiple traffic light Traffic light wheel Color-coded GDA*
Figure 2. Examples of traffic light formats. Note. * Format used in the current study.
According to Grunert and Wills (2007) this finding can be explained by three factors: there is too little information, consumers want to know what the colors mean and how they should be interpreted; they experience restriction in their freedom of choice, because ‘healthfulness choices’ have already been made for them; and consumers experience the simple traffic light as being “too didactic” (Grunert & Wills, 2007, p.392). In line with these findings, the qualitative study on behalf of the Food Standards Agency (FSA) (Malam et al., 2009) found that the hybrid, color-coded GDA (which combines traffic lights with absolute values and percentages) outperformed the simple traffic light format on preference, which seems to be in line with the general opinion of Dutch consumers (Somers, 2011). In the findings of their literary research, Grunert and Wills explain these effects by stating that “the color coding provides the simplicity (since the processing of the label can be limited to the colors, ignoring the rest), but the GDA information gives “reassurance by providing numbers that one could go back to for verification, giving an impression of more transparency and less paternalism” (Grunert & Wills, 2007, p. 392). Furthermore, because of this straightforward and
7 outperform the monochrome GDA on comprehension amongst all socio-economic groups (Malam et al., 2009). Based on the fact that many brands already show a monochrome GDA format, this study will specifically compare the monochrome GDA and the color-coded GDA label. Given the findings of previous studies (Malam et al., 2009) it is generally expected that:
H1. Label liking and label comprehension will be higher for the color-coded GDA label than for the monochrome GDA label.
Nutrition information processing, decision making and attitude formation
In examining the effects of both labels, the fundamentals of this theoretical framework are based on the conceptual model Grunert et al. (2010) apply in analyzing the effects of nutrition information on consumers, which is an adaptation of the hierarchy of effects model proposed by Grunert and Grunert (1995). This model was developed by incorporating two streams of research on consumer decision making (Peter, Olson, & Grunert, 1999; Solomon, 2006), and on attitude formation and change and behavior (e.g. use of the label and purchase intentions) (McGuire 1969; Petty & Cacioppo 1986; Eagly & Chaiken 1993). This model describes that in order for nutrition labels to have any effect, through exposure, consumers first must
perceive the label. Subsequently, consumers try to interpret the label and through deep or
superficial levels of processing, they reach some level of understanding. Then, when a certain degree of comprehension is established, evaluation of the label follows. Finally, it is assumed that the evaluation of the label can in turn affect overall attitudes toward the product or brand (Grunert et al., 2010). In processing and interpreting the nutrition information on the label, several factors such as message characteristics, but also the receivers’ motivation,
opportunity and ability to process the message information can influence the level of this
processing and therefore also processing outcomes. In the next section we will take a further look at the actual message characteristics that distinguish the monochrome and color-coded GDA from one another. Additionally, by using the MOA theory of MacInnis, Moorman and Jaworski (1991) I will set out how these different elements are presumably processed by consumers.
Motivation – I want to interpret nutrition information
The MOA theory is a useful tool in explaining how communication effectiveness (in the current study that is label understanding and liking) can be improved by (increasing)
8 individuals’ levels of Motivation, Opportunity and Ability to process this information. As described by the MOA theory (MacInnis et al., 1991), motivation is a force that pushes people towards certain goals. Grunert and Wills (2007) found that not only nutrition knowledge but also interest in nutrition information (the motivation to look for and interpret nutrition information) can ease label comprehension and use. It is not expected that motivation has an effect on the relation between label type and label comprehension: since both highly or lowly motivated consumers will find the numerical nutrition information they are looking for on both labels. However, in line with Grunert and Wills (2007) it is assumed that motivation can be related with nutrition knowledge, through more active search for nutrition information and mere expose effects. This implies that consumers with high motivation are more likely to have high nutrition knowledge because in the past they engaged in active search for nutrition information.
Opportunity – I am in the opportunity to process nutrition information
The main element in which the monochrome and color-coded GDA differ, is the manner in which they present informational references. Reference point information (such as
percentages of daily values on the GDA format) has proven to be essential in processing numerical nutrition information and consumers tend to experience difficulties with interpreting nutrition information that lacks reference information (Visschers & Siegrist, 2009). It is therefore no surprise that standard nutrition tables which only show absolute nutrient amounts, have found to be difficult to understand, and not preferred (Viswanathan & Hastak, 2002; Van Herpen et al., 2014). This can be explained by the fact that “numerical information lacks meaning by itself and has to be compared with other information to be interpreted meaningfully” (Viswanathan, 1994, p. 49). Both the GDA and the color-coded GDA label resolve this lack of reference by presenting a numerical (relative) guideline or directive of an adults’daily amount (monochrome GDA), and a combination of this numerical information and color (color-coded GDA).
However, with its additional colors, it seems that the color-coded GDA exceeds the monochrome label in providing comprehendible referential information. This is not a surprise when we know that research on consumer impressions of nutrition labels has shown that consumers not only experience to be overly cognitively challenged when having to process numbers, but more importantly, they still experience high cognitive load when having to interpret percentages as on the monochrome GDA label (Barone, Rose, Manning, & Miniard,
9 1996). It seems plausible that the more negative evaluation of the monochrome GDA that Malam et al. found (2009), is an effect of this cognitive hassle. Evidently, in grocery shopping situations the opportunity and time to process nutrition information is generally relatively low (Feunekes, 2008). Facilitating the opportunity to process can be done by reducing the time needed to evaluate the presented information (MacInnis et al., 1991). While interpreting absolute values and percentages on the monochrome GDA label might involve more time and cognitive effort, traffic lights have been found to facilitate consumers in making quick
comparisons and decisions, with relatively low cognitive effort (Malam et al., 2009). This can be explained by the grounded theory that explains that to humans, colors contain symbolic elements and meaning, and automatically activate associations (Elliot, Maier, Moller, Friedman, & Meinhardt, 2007).
The ability – I can interpret the nutrition information
Based on MOA theory, ability is the extent to which consumers have the necessary resources to make an outcome happen (MacInnis et al., 1991), which is in this case interpreting nutrition information. Anderson and Reder (1979) suggest that deeper levels of comprehension are achieved when various mental concepts and relevant knowledge structures are activated. In the case of interpreting food labels, it has been found that label understanding and ultimately label use are determined by such a knowledge structure: nutrition knowledge (Grunert et al., 2010). The resource- matching perspective (Hu, Huhmann and Hyman, 2007) suggests that information processing is optimal when available resources (e.g. nutrition information) and required resources (e.g. nutrition knowledge) are matched. Thus, when consumers can rely on required resources to solve the cognitive task, information processing is optimal. The resource-matching perspective also assumes that when a cognitive task (e.g. interpreting a nutrition label) requires unavailable resources, or if the available resources (e.g. nutrition information on the GDA) exceed required resources, then the performance of the cognitive task (e.g. interpreting the label) is negatively impacted and cannot be completed successfully (Anand & Sternthal, 1990; Meyers-Levy & Peracchio, 1995).Hence, it is expected that consumers with high nutrition knowledge can rely more on established knowledge structures than consumers with low nutrition knowledge. It is expected that:
H2. Label comprehension will be higher when nutrition knowledge is high, than when nutrition knowledge is low.
10 Subsequently, when consumers with low nutrition knowledge judge a cognitive task as overly challenging because they do not possess the required nutrition knowledge structure, avoidance of the effortful cognitive task can occur: consumers may feel confused and / or they may adopt heuristic processing (Malhotra, 1984), because they search for a way to induce
cognitive strain (Wright, 1975). When this occurs, it is expected that colors can be applied as simple heuristic cues to interpret the levels of nutrients: hence the symbolic nature of colors and the established associations they posses, the processing of these colors only asks for low cognitive effort. It is therefore expected that consumers with low nutrition knowledge (perceive to) comprehend the color-coded GDA better than the monochrome GDA:
H3. When nutrition knowledge is low, label comprehension will be higher for the color coded-GDA than for the monochrome GDA.
Hence the fact that the color-coded GDA label only differentiates from the monochrome GDA in additional coloring, and consumers with high nutrition knowledge are expected to not need to fall back on colors, it is expected that there is no significant difference in label
comprehension when nutrition knowledge is high.
The effect of established levels of comprehension
Now we get to the point where consumers have reached some level of label comprehension, with our without the help of available nutrition knowledge. Previous research on the
comprehension of nutrition labels has mainly focused on label effectiveness as the ability to communicate product healthfulness and to induce actual understanding (Feunekes et al., 2008; Van Herpen & Trijp, 2011). Thus, these studies mainly focused on objective comprehension: the extent to which respondents interpret the information on the label
correctly. However, in the current study I would like to break down the concept of
comprehension into two components: objective (or correct) comprehension of the message, as intended by the sender, and subjective (or perceived) comprehension of the message (e.g. ‘I believe to comprehend the nutrition label’). Subjective label comprehension will be
examined, simply because of the fact that whether consumers draw correct or incorrect conclusions from a nutrition label, the belief in one’s capability to interpret the label is believed to have a profound effect on label liking, as will be explained in the next section.
11 In understanding the positive effects that perceived label comprehension can have on the message (e.g. the label) and ultimately on the message source (e.g. the brand), several studies on subjective comprehension can be cited. First, a study on the effects of subjective comprehension in advertising on ad perceptions by Mick (1992), shows that deeper levels of subjective comprehension are positively related to attitudes toward the source message. This is based on the idea that when it comes to the impact of perceived ability on self-regulatory mechanisms and decision making, subjective comprehension often enhances “the belief in one’s own capabilities to mobilize cognitive resources and courses of action needed to meet given situational demands” (Wood & Bandura, 1989, p. 408). In other words, if consumers
feel as if they understand the label, this can result in the belief that one is self capable of
resolving a cognitive task as interpreting the nutrition label. In turn, when the receiver can gain a personal advantage from the source or if the source allows the receiver to fulfill a certain task (e.g. interpreting nutrition information) this can raise the receivers’ feeling of competence (Knowles & Linn, 2004). Raising the receivers’ feeling of competence has found to be a helpful way in inducing resistance and in increasing the likelihood of adapting
persuasive messages (Ratneshwar & Chaiken, 1991; Hu, Huhmann and Hyman, 2007). These ideas and the fact that food labels can be conceptualized as source messages, lead to the hypothesis that:
H4. (Subjective) label comprehension will have a positive effect on label liking.
Although previously assumed that in order to evaluate the label, consumers have to
understand it to some degree (Grunert et al., 2010), I wish to highlight that comprehension of
the label does not in every case need to be a prerequisite or requirement for label liking (Grunert & Wills, 2007). Consumers may like the label because they find it easy to interpret or useful, but it is expected that they can also like the appearance of the label because of the colors used, because consumers generally seem to like seeing colors on food labels (Grunert & Wills, 2007). As specified in hypothesis 1, it is expected that label comprehension will be highest for the color-coded GDA, and that the same accounts for label liking.
The effects of label liking
Via or around label comprehension, we now come to a point where (through evaluation) the label is liked or not, and to a critical point in this study: does this have any effect on the
12 consumers’ evaluation of the brand showing the label? As mentioned in the introduction, it is expected that liking of the label can have radiating effects on the variables brand attitude, brand credibility and purchase intention (Kozup et al., 2003; Andrews et al., 2011). It is expected that for the color-coded GDA, liking of the label can lead to a more positive evaluation of the product or brand, through peripheral processing of the colors (Petty and Cacioppo, 1986), and as we have read, even when the label is not understood correctly (Grunert & Wills, 2007). In semiotics it is proposed that positive connotative meanings that are generated during interpretation (e.g. ‘What a helpful label this is’ or ‘I like the colors’), activate positive evaluations about the message source through overall aesthetic enjoyment (Mick,1992). Similarly, radiating effect is supported by the article of MacInnis and Jaworski (1989) which describes how positive thoughts (e.g. ‘I like this label’) or negative thoughts (‘this label is hard to interpret’) generated by a stimulus cause development of strong beliefs about the object, here being the brand. Thus, when consumers generate positive or negative feelings, this is believed to spill over to general attitudes toward the brand. According to Mick (1992), the more positive these overall brand evaluations are, the bigger the chance the brand will be chosen over other brands in consumer decision making contexts.
Furthermore, as for the effect of label liking on brand credibility, research has found that when brands present clear information about a product, this leads to more positive feelings about the message source, which can lead to positive effects on the credibility of the brand (Baek, Kim, & Yu, 2010). Hence clarity is here conceptualized as ‘the lack of
ambiguity of product information’ (e.g. straightforward and clear nutrition information) (Erdem, Swait, & Louviere, 2002) nutrition labels can be treated as tools which eliminate (nutrition) information ambiguity. In turn, it is therefore proposed that because of its clear information the label is liked more, which ultimately leads to a more positive brand
credibility. In summary, these findings on the potential positive effects of label liking result in the following hypothesis:
H5. Label liking will have a positive effect on the attitude towards the brand, purchase intention and brand credibility.
In turn, because of the overall expectation that label liking will be higher for the color-coded GDA (hypothesis 1) and the above described presumptions, it is also expected that these positive effects will result in a higher brand attitude, brand credibility and purchase intention for the color-coded GDA condition:
13
H6. Brand attitude, brand credibility and purchase intention will be higher for the color-coded GDA than for the monochrome GDA.
The previously outlined theoretical framework resulted in the development of the conceptual model as shown in Figure 3.
Figure 3. Conceptual model.
METHOD
Design and participants
A self-administered online survey (created with Qualtrics Online Survey Software) was distributed via Facebook and email. In total, 124 complete responses were obtained which consisted of 37 male participants and 87 female participants, in the age range of 18-73 (M= 30, SD=13.40). Due to the relatively high amount of students, participants are slightly higher educated than the national population (Centraal Bureau voor de Statistiek [CBS], 2013): 58.1% (n=72) achieved a university degree, 22.6% (n=28) had a high education degree (hbo), 12.9% (n=16) completed senior secondary education and 6.5% (n=8) completed a form of secondary vocational education. Participants were randomly assigned to one of 2 conditions (type of label: monochrome GDA, n= 59; or color-coded GDA n= 65).
14 Materials
Label formats. As described, the two nutrition labels examined in this study are the
monochrome GDA and the color-coded GDA label (see Figure 4). These formats show details on five nutrients: calories, sugar, fat, saturated fat and salt. Per nutrient, the formats present the absolute amount of the nutrient, as well as a percentage relative to the guideline of the daily amount per 100 gram. The color-coded GDA label combines this numerical nutrition information with the colors green, amber and red. All the label examples in this study are created with Photoshop graphics editor and are presented with actual nutrient amounts and with appropriate numerical information and colors, based on front of pack criteria for traffic light labeling formulated by the FDA (2012), and FSA (2007). Dutch phrasing on the labels is consistent with the actual GDA information used on food labels in the Netherlands.
Monochrome GDA Color-coded GDA
Figure 4. Examples of the nutrition labels used in the current study
Product. Pre-packaged soups were used in this study because of two reasons. First, research
has shown that interest in nutrition information differs between product categories: nutrition information is generally less relevant for fresh products (e.g. fruit, vegetables) but more relevant in processed products with a lower degree of transparency, like ready meals (Directorate General for Health and Consumer Protection, 2005). Secondly, the nutrition knowledge scale used in this study (see Appendix A for the specific items that were used and
Measures) has been found to correlate relatively high with vegetable consumption, compared
to other foods and drinks (Dickson-Spillmann, Siegrist, & Keller, 2011).
In each condition three Unox soup variants where shown: asparagus, tomato and curry soup. Depending on the condition, these packages showed one of the label formats, combined with an enlarged view of that specific label (see Appendix B for all used visuals). These soups almost all differed in the amounts of calories, sugar, fat, saturated fat and salt. Only two of the soups (asparagus and tomato soup ) contained the same levels of salt. As the referential
15 information in this study was based on actual FSA guidelines (2007), the labels used in the coded condition only included green and amber elements. The GDA label and the color-coded GDA in Figure 4 contain a red element, but these were only used as examples in the introduction text that participants read (see Procedure).
Procedure
Once participants had indicated they agreed on participating in this study, they first had to answer some questions on demographics. Then, participants had to answer randomly ordered questions to determine their level of nutrition knowledge. Also, they were presented with questions regarding their motivation to search for and understand nutrition information in grocery shopping situations. After completing these questions, the participants were randomly assigned to one of the two conditions: the monochrome GDA or the color-coded GDA
condition.
Participants first were presented with a picture of one of the labels and introduction text of the condition specific nutrition label, see Appendix C for the specific examples. The text consisted of information on how the label and numbers and/or colors should be
interpreted. Participants were instructed to read the text carefully, because they would be presented with some questions about the label. The monochrome GDA introduction text was based on GDA information provided by the Dutch Nutrition Center (Voedingscentrum, n.d.). Additional information on the colors used in the color-coded GDA text was based on
information from the FSA on how traffic light colors should be interpreted by consumers (2013). Participants had to indicate on a 5-point Likert scale how familiar they were with the presented label from 1 (unfamiliar) to 5 (familiar).
Subsequently, participants were exposed to the three soup packages with a
monochrome or color-coded GDA label on them (Appendix B), and had to answer questions regarding actual (objective) understanding, perceived healthfulness of the soups and which elements of the label participants based their conclusions on, ending with questions regarding their perceived understanding (subjective comprehension) of the labels shown. Finally, participants had to answer questions regarding their attitude towards the label, their attitude towards the brand, the brands’ credibility and their purchase intention.
16 Measures
Unless indicated otherwise, items were presented in a random order. See Appendix D for specific scales and items that were used.
Demographics. Participants were asked about their age, gender, education level and
their field of occupation or education. The latter variable has been included because it is expected that participants with educational or professional backgrounds in medicine, health or nutrition generally have a higher levels of nutrition knowledge (Dickson- Spillmann et al., 2011).
Nutrition knowledge. Nutrition knowledge was measured with the 20 item Consumer
Nutrition Knowledge Scale (CoNKS) which has been developed and validated by Dickson-Spillmann et al. ( 2011). The20 questions were translated applying a back-translation procedure. First, the questions were translated from English to Dutch by a master student in English literature. These translations were reviewed by a registered nutritionist and edited where necessary. This meant for instance that ‘gruyère cheese’ was substituted by ‘belegen kaas’, which is probably a more familiar cheese amongst Dutch consumers, and also contains similar nutrient quantities. Then, the Dutch translations were translated back to English by a second degree English teacher. For an overview of the translated version of the CoNKS, see Appendix A. To avoid the risk of an ‘easy’ answer choice when respondents would
experience difficulty in identifying the best option (don’t know), or forced choice (false/true), nutrition knowledge was measured on a 7-point Likert scale from 1 (completely disagree) to 7 (completely agree). Correct responses were scored as 1, while incorrect responses and
‘neutral’ responses (4) were scored as 0. The score for nutrition knowledge was calculated as the percentage of correct answers. Based on their scores, participants were divided into a low and a high nutrition knowledge group, by using a median split (Mdn =68.42). A reliability analysis was used to test reliability of the 20 items of the Dutch CoNKS version, which showed to be reliable (α=.69, M=72.6, SD=15.41). However, item 6 (‘Oily fish contains healthier fats than red meat’) was deleted. This because the item was answered correctly by 84% of the participants and it is believed that this high rate of correct responses could be caused by recent media attention on omega-3 fatty acids and fish oil, and that this has increased awareness and knowledge of the benefits fish and fish oil hold. Deleting this item caused a slight increase of the Cronbach’s Alpha to α =.71 (M = 67.86, SD = 15.33).
Motivation. As motivation is treated as the interest to search for and interpret nutrition information (Feunekes et al., 2008), respondents were asked to indicate on a 5-point Likert scale if they are never (1) or always (5) motivated to ‘read’ and ‘understand’ nutrition
17 information (about calories, sugar, fat, saturated fat and salt) when doing grocery shopping. A principal component analysis showed that the two items belong to one component (Eigen value 1.69), which explains 84.96% of the variance. Both items positively correlate with the first component (>.7) and seem to measure the motivation to interpret nutrition information: a higher score on these items seems to reflect a higher motivation. A reliability analysis showed that this scale has good reliability: α =.82.
Objective label comprehension. To simulate grocery shopping situations in which
consumers are presented with multiple products, and to measure objective comprehension, participants were asked which one of the three presented soups was highest or lowest in calories, in saturated fat, in sugar, fat, saturated fat and salt. This measurement is comparable to tasks applied in previous studies by Grunert et al. (2010) and the FSA (2005). Also, to measure perceived healthfulness of the products, respondents had to range the three soups from healthy (1) to least healthy (3). A ‘right’ order for soups’ healthfulness was established by a nutritionist who scrutinized the soups on healthfulness. The objective of including this test was not primarily to determine perceived product healthfulness, but again to simulate grocery shopping conditions and corresponding decision processes. Because traffic light colors have shown to be possible influencers of label interpretation and perceived
healthfulness (Aschemann-Witzel et al., 2013; Schuldt, 2013), participants were also asked which of the label elements (calories, sugar, fat, saturated fat, salt and for color-coded GDA: the colors) they mainly used in determining product healthfulness, ranging from 1 (most important) to 5 or 6 (least important). By asking this, it could be determined whether the use of color (scored as 1) or the other elements (scored as 0) as ‘healthfulness indicators’, affected scores on label comprehension.
Subjective label comprehension. In line with the research and measurements of
Grunert et al. (2010), subjective comprehension was measured by asking respondents to indicate whether they understood the presented label on a 7-point Likert scale (1: completely disagree, 7: completely agree). Also, participants were asked to indicate if the label enabled them to interpret nutrition information and helped them in comparing products, and if they thought the label was easy to use (1: completely disagree to 7: completely agree).
Label liking. To my knowledge, previous researchers have not yet constructed a
specific tool to measure label liking. The interpretation of the attitude toward the label is that it should reflect subjects' overall evaluations of the stimulus. To measure attitude toward the label, a construct of seven, 7-point semantic differentials has been used. These items were unappealing/appealing, bad/good, unpleasant/pleasant, unfavorable/favorable,
18 unlikable/likable and are similar to the scale applied by Spears and Singh (2004).
‘Interesting/boring’ and ‘uninformative/informative’ were added because the labels could be treated as informational stimuli (Biehal, Stephens & Curio, 1992) and because of their perceived relevance. A principal component was conducted on the seven items with orthogonal rotation (Varimax). This analysis showed that two components can be
distinguished, with an Eigen value of 4.41 for the first component and 1.21 for the second component. All seven items were retained because they not only are believed to belong to the entire construct of label liking, but reliability was higher when all items were included
(α=.90) than when two separate components were distinguished (α =.88, α =.86). In combination they explained 80.45% of the variance. The Kaiser-Meyer-Olkin measure
verified the sampling adequacy for the analysis, KMO=.86. All items positively correlate with component one (>.7), with the highest consistency for ‘good’ (.86), and seem to measure overall label liking: a higher score on the items seems to reflect higher label liking. Finally, to determine the overall label score, participants had rate the label on a scale from 0 to 10.
Brand attitude. Participants first had to indicate on a 5-point Likert scale how familiar
they were with the brand (1: unfamiliar, 5: familiar). To measure brand attitude, a construct of five 7-point semantic differentials has been used, similar to Spears and Singh’s measurements (2004). Participants were asked to indicate whether they found the brand to be
unappealing/appealing, bad/good, unpleasant/pleasant, unfavorable/favorable, unlikable/ likable. A principal component analysis showed that the six items belong to one component (Eigen value 4.41), which explains 74.02% of the variance. All items positively correlate with the first component (>.7) (with the highest consistency for ‘pleasant’ and a component value of .90) and seem to measure brand attitude: a higher score seems to reflect a more positive brand attitude. A reliability analysis showed that this scale has good reliability: α=.93. To determine how participants evaluated the fact that the brand presented the label, participants were also asked to indicate on a 7-point Likert scale if they ‘appreciated’, and found it ‘useful’ and ‘good’ that the brand showed the label (1: completely disagree, 7: completely agree).
Brand credibility. Six 7-point Likert items were used to measure brand credibility (1:
completely disagree, 7: completely agree), similar to measurements used by Sweeney and Swait (2008). A principal component analysis showed that the six items explain 81.18% of the variance, and belong to an one-dimensional scale with an Eigen value of 4.87. All items positively correlate with the first component (>.7), with a highest consistency for ‘this brand is dedicated to deliver promises it made’ and a component value of . 94. This construct seems
19 to measure brand credibility: a higher score on items seems to reflect higher perceived brand credibility. A reliability analysis showed that this scale has good reliability: α=.95.
Purchase intention. In line with Baek, Kim and Yu (2010) purchase intention was
measured by asking participants to indicate on a 7-point Likert scale how likely it was that they would buy soups from the brand they saw and how likely it was that they would choose soups from this brand over soups from other brands (1: very unlikely, 7 very likely). A principal component analysis showed that the two items explain 84.73% of the variance, and belong to a one-dimensional scale (Eigen value is 1.69). Both items positively correlate with the first component (>.7) ,with a highest consistency ‘would purchase’ (component value of .92) and seem to measure purchase intention: a higher score on items seems to reflect a higher intention to purchase the brand. A reliability analysis showed that this scale has good
reliability: α=.81.
RESULTS
Sample descriptives
The average age in the monochrome condition was 29.44 years (SD=12.79) and in the color coded condition the average age was 30.61 years (SD=14.02). With respect to gender, 32.2% of the participants in the monochrome condition were male (n=19) and 67.8% were female (n=40). In the color-coded condition, 27.7% was male (n=18) and 72.3% was female (n=47). Statistics showed age did not significantly differ between conditions t(122) = -0.48, p=.632, and that the two conditions did not differ with respect to gender X²(1)= .30, p=.695 and education level U =1911, p=.973. This implies that differences in the groups regarding dependent variables cannot be caused by differences in these background variables.
Nutrition knowledge. Overall, scores on nutrition knowledge ranged from 2.5 to 9.5
and were not extremely high (M=6.79, SD= 1.53) which is consistent with participants’ reactions in the comments section of the experiment: they found the questions not that easy to answer. Several bivariate analyses were conducted to determine whether background
variables were related to nutrition knowledge. These results showed there was a significant relationship between gender and nutrition knowledge (rS= .23, p=.009): women scored
slightly higher on nutrition knowledge (M= 69.93 SD=14.51) than men (M= 63.02,
20 and subjective comprehension (p>.05), which implies that objective and subjective
comprehension were ultimately not influenced by gender. All other associations such as age, level of education and educational/professional background but also motivation to search for and interpret nutrition information were not statistically significant (p>.05). Therefore, the assumption that consumers with high motivation would have higher nutrition knowledge than consumers with low motivation is rejected.
Condition effects on label related dependent variables
The first hypothesis proposed that label comprehension and label liking would be significantly higher for the color-coded GDA condition than the monochrome GDA condition. Tests of mean differences on these dependent variables showed the following results.
Objective label comprehension. In general, participants scored relatively high on label
comprehension questions (M=92.47, SD=15.56) and scores did not significantly differ between conditions, t(122)=-0.52, p= .607, 95% CI [-7.00,-4.11]. This implies that
participants in both conditions were equally well able to interpret the labels correctly. Two other components that were measured as possible indicators of label comprehension (the score on perceived healthfulness and the label element that played the most crucial role in
determining product healthfulness) did also not correlate with objective comprehension. This implies two things: first, whether participants were good or bad at choosing the healthiest and least healthiest soup, this did not relate to label comprehension. For actual scores on product healthfulness per condition, see Appendix E. Second, when consumers applied the colors as main product healthfulness indicator, instead of basing product healthfulness on other elements such as calories or fat, this did not affect actual label comprehension.
Subjective label comprehension. Perceived label comprehension showed to be
significantly higher for the color-coded GDA condition (M=5.48, SD=1.08) than for the monochrome GDA condition (M=5.07, SD=1.59), t(122)=-1.69, p=.047, 95% CI [-0.89, 0.69]: participants who saw the color-coded GDA label felt to understand the label better than participants who saw the monochrome version. The three questions on ease of use,
helpfulness for interpretation and facilitation (which were added to enucleate possible differences in comprehension) showed no significant difference between the two label conditions (p>.05). Bivariate analyses also showed that there was no significant association
21 with variables such as age, gender, education, familiarity with the label and
educational/professional background and objective and subjective comprehension (p>.05).
Label liking was significantly higher for the color-coded GDA (M=5.23, SD=0.93)
than for the monochrome GDA (M=4.63, SD=1.13), t(122)=3.22, p=.001, 95% CI [0.96, -0.23]. In addition, the participants in the color-coded GDA condition gave the label a significantly higher score from one to ten (M=7.30, SD= 0.99), than participants in the monochrome GDA condition (M=6.81, SD=1.43), t(122)= 2.22, p= .028, 95% CI [0.92, -0.05]. Based on these results, the first hypothesis is partially supported: although participants are not better at correctly interpreting the color-coded GDA, results confirm that participants’
feeling of comprehension is higher for the color-coded GDA and they like the color-coded
label more. Bivariate analyses showed no significant associations with background variables (p>.05).
Condition effects on brand related dependent variables
Hypothesis 6 proposed that brand attitude, brand credibility and purchase intention would be significantly higher for the color-coded GDA condition. To test these effects, three
independent t-test were performed on both conditions.
Brand attitude. The attitude participants had towards Unox did not significantly differ
between the color-coded (M=4.94, SD=1.07) and the monochrome GDA condition (M=4.79,
SD= 0.93), t(122)= -1.31, p= .097, 95% CI [-0.62, -0.13]. However, it must be mentioned that
when participants were particularly asked about the fact the brand showed the respective label (instead of asking them about their general attitudes toward the brand) reactions were
significantly more positive for the color-coded GDA label: participants in the color-coded GDA condition rated the label significantly higher on the elements: ‘appreciate’, ‘good’ and ‘useful’. Data on these variables for both groups are presented in Table 1.
Brand credibility. Results showed that there was no significant difference for brand
credibility between the color-coded GDA (M=5.03, SD=0.89) and the monochrome GDA condition (M=4.76, SD= 1.22), t(122)=1.39, p= .084, 95% CI [-0.64, -0.11].
Purchase intention. The independent t-test showed that participants in the color-coded
GDA condition scored significantly higher on the intention to purchase Unox soups (M=5.02,
SD=1.47) than participants in the monochrome GDA condition (M=4.48, SD=1.62), t(122)=
-0.20, p=.027, 95% CI [-1.09, 0.01]. These results support the purchase intention component of hypothesis 6: participants who saw the color-coded GDA label have a significantly higher intention to buy Unox soups in the future. However, when it comes to brand attitude and the
22 brand credibility, this part of the hypothesis is not supported: there was no significant
difference between conditions for these particular variables.
Table 1
Mean Scores, Standard Deviations and Independent t-test by Label Format
Label format Monochrome GDA (n=59) Color-coded GDA (n=65) M SD M SD t (p)
The fact that the brand shows the label I,
.. appreciate 5.41 1.28 5.82 0.93 -2.02(*)
.. find good 5.54 1.13 5.86 0.83 -1.78(*)
.. find useful 5.34 1.32 5.87 0.97 -2.49(**)
Note. *p <.05, ** p <.01.
Model testing
To test the conceptual model which has been proposed in the theoretical framework, the following analyses were conducted.
Moderating effect of nutrition knowledge
To test the moderating effect of nutrition knowledge on the relationship between label type and label comprehension (hypothesis 3), a two way ANCOVA was performed and showed that there was no significant interaction effect between nutrition knowledge and label type on objective comprehension F(1, 119)=0 .92, p=.338 and subjective comprehension (F(1, 119) = 0.41, p= .524). Therefore, hypothesis 3 is rejected. However, there was a significant direct effect of nutrition knowledge on objective comprehension (F(1, 119)=5.26, p= .024): participants’ scores on objective label comprehension were significantly higher when
nutrition knowledge was high (M=95.79, SD=11.56) than when nutrition knowledge was low (M=89.56, SD= 17.96) which is in line with hypothesis 2. No direct effect of nutrition
knowledge on subjective comprehension was found (p >.05), which implies that whether participants had high or low nutrition knowledge, participants perceived to understand the labels equally well.
23 The effect of label comprehension on label liking
Hypothesis 4 proposed that label comprehension has a positive effect on label liking. In line with the expectation that the component of subjective comprehension would be of greatest importance in determining establishing label liking, subjective comprehension of the label (‘I understand this label’) relates directly positive and moderate to label liking, r=.46, p<.001. Moreover, subjective label comprehension did not directly relate to brand attitude, brand credibility and purchase intention (p>.05), which implies that possible effects of subjective comprehension on brand attitude, brand credibility and purchase intention run through label liking. Regarding the effect of objective label comprehension on label liking, results showed no significant relationship. Additionally, objective comprehension was not related to brand attitude, brand credibility and purchase intention (p>.05).
To test the exact effects of subjective comprehension on label liking per condition, two separate simple regression models were conducted. For the monochrome condition, the
regression model with label liking as a dependent variable and subjective comprehension as an independent variable showed to be significant , F(1, 57)=14.43, p<.001. The model can be applied to predict label liking for participants who saw the monochrome label, where
subjective comprehension accounts for 20.2% of the variation in label liking (R²= .202). When perceived comprehension of the label increases with one, label liking will increase with .32 (b*= .32, t=3.79, p< .001, 95% CI [0.15, 0.49]). Also, for the color-coded GDA condition the regression model showed to be significant F(1, 63)=13.49, p<.001. The model can
therefore be applied to predict label liking for participants who saw the color-coded GDA label, where subjective comprehension accounts for 17.6% of the variation in label liking (R²= .176). When perceived comprehension of the label increases with one, label liking will
increase with .36 (b*=.36, t=3.67, p< .001, 95% CI [0.17, 0.59]).
Effects of label liking on brand related dependent variables
Hypothesis 5 proposed that label liking has a positive effect on brand attitude, brand credibility and purchase intention. To test the applicability of the proposed model for both conditions, regression models per condition and per dependent variable were conducted.
Brand attitude. For the monochrome condition, a linear regression model with brand
attitude as a dependent variable and label liking as an independent variable showed to be significant, F(1, 57)=14.90, p<.001, and can therefore be used to predict brand attitude. Label liking accounts for 20.7% of the variation in brand attitude (R²= .207). When label liking
24 increases with one, brand attitude will increase with .46 (b*=.46, t=3.86, p< .001, 95% CI [0.22, 0.70]). The same regression model was applied for the color-coded condition and showed to be significant F(1, 63)=12.81, p=.030, and can therefore also be used to predict brand attitude. Label liking accounts for 17.6% of the variation in brand attitude (R²= .176). When label liking increases with one, brand attitude will increase with .29 (b*= .29, t= 3.54,
p= .030, 95% CI [0.20, 0.68]). 1
Brand credibility. For the monochrome condition, a linear regression model with
brand credibility as a dependent variable and label liking as an independent variable showed to be significant, F(1, 57)=13.02, p<.001, and can therefore be used to predict brand
credibility. Label liking accounts for 18.6% of the variation in brand credibility (R²= .186). When label liking increases with one, brand credibility will increase with .46 (b*=.46, t=3.61,
p< .001, 95% CI [0.21, 0.72]). The linear regression model for the color-coded condition also
showed to be significant, F(1, 63)=12.76, p=.002, and can therefore be applied to predict brand credibility. Label liking accounts for 16.6% of the variation in brand credibility
(R²=.166). When label liking increases with one, brand credibility increases with .31 (b*=.31,
t=3.42, p=.002, 95% CI [0.20, 0.69]).2
Purchase intention. A last regression model was used to test the influence of label
liking on purchase intention, which showed to be significant for the monochrome condition (F(1, 57)=13.70, p< .001). Results show that 19.4% of the variation in purchase intention can be explained by label liking (R²=.194) and when label liking increases with one, the intention to purchase Unox soups increases with .63 (b*= .63, t=3.70, p<.001, 95% CI [0.29, 0.97]). The model showed also to be significant for the color-coded condition (F(1, 63)=12.12,
p=.004). Results show that 16.6% of the variation in purchase intention can be explained by
label liking (R²= .166) and when label liking increases with one, the intention to purchase Unox soups increases with .56 (b*=.56, t=3.48, p=.004, 95% CI [0.28, 0.96]). 3
1 Age (r
S= -.23, p= .010) and brand familiarity (X² (92) = 167.67, p< .001) have shown to correlate with brand attitude. Taken together with label liking, these variables account for 34.3% of the variation in brand attitude (R²=.343) in the monochrome condition and for 24.4% in the color-coded condition (R²=.244).
2
Bivariate analysis showed that brand credibility was weakly negative associated with age (rS= -0.24, p =.008). Taken together with label liking, the two variables account for 24.7% of the variation in brand attitude (R²=.247) in the monochrome condition, and for 26.7% in the color-coded condtition ( R²=.267).
3
Bivariate analysis showed that purchase intention was positively associated with participants’ familiarity with Unox X²(1)= 61.00, p=.006. Familiarity with the brand and label liking combined, account for 40.1% of the variation in intention to purchase Unox (R²=.401) in the monochrome condition, and for 38.8% in the color coded condition (R²=.388).
25 In summary, results show that the proposed model is largely applicable for both
conditions. It can be concluded that (subjective) label comprehension has a positive effect on label liking, and in turn, label liking increases brand attitude, brand credibility and the
26
Figure 5. Final conceptual model.
Note. Conditions are separated by ‘/’: monochrome condition (left) / color-coded condition (right). Beta values depicted in italic, explained
variance depicted in bold (R²). *Effect of nutrition knowledge only significant on objective comprehension. **Only a significant effect for subjective comprehension on label liking.
27
DISCUSSION AND LIMITATIONS
In a nutshell, this study was set out to address the differences between the monochrome and color-coded GDA on comprehension, liking and the ultimate effects on brand attitude, brand credibility and purchase intentions. Several interesting conclusions can be drawn from this study. Starting off with the first hypothesis, overall we can infer that the color-coded label version is evaluated significantly more positive than the monochrome GDA label, which is in line with previous studies (Grunert & Wills, 2007; Malam et al., 2009). Not only did results show that the color-coded GDA is perceived as more easy to understand (subjective label comprehension), it is also better liked and rated than the monochrome GDA label.
Participants indicated that they were more appreciative of the color-coded label and found it more helpful and good than the monochrome version. Above all, purchase intentions of Unox soups were significantly higher when participants saw the color-coded version on the soup package. This implies that if brands want consumers to choose their product over products from other brands, they are more likely to achieve this by showing the color-coded GDA label on their products. An important remark that has to be made is that we do not yet know
whether this is also the case when the label consists of more red an amber elements. As we have read in the introduction, when labels consist of more red and amber elements this does not inevitably have to result in fewer sales (Sacks et al., 2009), but there are several studies that are less supportive of this thought. Due to the signaling effect the colors amber and red are believed to attract attention (Hieke & Wilczynski, 2012; Van Herpen et al., 2014), but it is also found that the color red could be associated with danger (Pravossoudovitch, Cury,
Young, & Elliot, 2014). Thus, the current study tells us what effect green and amber
assembled labels have on the dependent variables, but the specific question if consumers will tend to avoid products (or brands) with amber and red label elements (Euractiv, 2010), has not yet been answered. Additionally, in the current study a general evaluation of three products was measured. Therefore there are no insights provided yet on how individual products (with different color assembled labels) are evaluated, and more importantly, how this affects brand evaluations and purchase intentions.
As predicted, and in line with previous studies (Grunert & Wills, 2007; Malam et al., 2010) participants found the color-coded GDA more easy to comprehend than the
28 interpretability of nutrition information and facilitating product comparison, tend to be in favor of the color-coded GDA label. This absence of a significant effect between the
conditions could be caused by more deep processing of the objective comprehension task due to the specific experimental setting. Hence the fact that participants were probably well in the opportunity to actually comprehend the presented labels, it is expected that the ability to compare nutrient levels and therefore the perceived interpretability of nutrition information were relatively high. It could be that because participants had sufficient opportunity to process, and therefore were more capable to determine which product contained the least or most of a certain nutrient, scores on interpretability of the nutrition information and product comparison are higher than they would be in natural settings, especially for the monochrome GDA label. It is expected that in actual grocery shopping situations, where the opportunity and capability to process and obtain actual label comprehension is lower, the color-coded GDA label will score significantly higher on these variables, because the label is believed to relieve cognitive strain through the color elements.
Where previous research has shown that nutrition knowledge is a prerequisite for label comprehension (Grunert et al. 2010), no such results have been found in the current study: whether nutrition knowledge was low or high, participants scored relatively high on objective label comprehension, and both labels were equally well comprehended. In interpreting these results, there are two observations that can be made. First, the manner in which actual label comprehension was tested was based more on correctly reading off nutrient quantities, than it was about inferences or implications that could be drawn from the nutrition information. It could be that nutrition knowledge as it was measured in the current study, did not perfectly match the comprehension test participants had to fulfill. The used measurement for nutrition knowledge might be more linked to understanding in terms of inferences that consumers draw from the nutrient information. However, the particular measurement and conceptualization of objective label comprehension was chosen because it corresponds to the objective set out by the European Commission (2007), that Front of pack (FOP) nutrition labeling should enable consumers to compare products based on their nutrient quantities. Second, in this
experimental set up, participants may be somewhat forced to pay relatively high attention to the label because they had to answer questions about the label. This setting differs from grocery shopping situations in which consumers are less capable and not as much in the opportunity to process nutrition information (Feunekes et al., 2008). The higher opportunity and cognitive attention to process the nutrition information in the experiment, may also have resulted in relatively high scores on objective comprehension: once participants properly
29 focused on the labels, they were well able to determine which product contained the highest or lowest amounts of for instance salt or calories.
As hypothesized, not actual label comprehension, but perceived (subjective) label comprehension is the main predictor component of comprehension on label liking: when consumers perceive the label as more easy to understand, this results in higher appreciation of the label. Furthermore, in comparing both labels, the color coded-GDA was significantly better liked and was not influenced by background variables such as age, gender and education level. This implies that all consumers, whether young/old, male/female and with low or high education, equally liked the label. Moreover, even low or high nutrition
knowledge did ultimately not lead to higher label liking. This because nutrition knowledge only had an effect on objective comprehension of the label, but objective comprehension did have no significant effect on label liking. This implies that it does not matter whether
consumers have low or high nutrition knowledge in order for them to like the label, or in the end, to prefer the color-coded GDA label. Furthermore, although the amounts of explained variance are relatively small, label liking showed to have a positive effect on brand attitude, brand credibility and purchase intention. These positive effects of label liking did not result in different levels of brand attitude and brand credibility between both conditions, but did result in a significant main effect on purchase intention: participants who saw the color-coded GDA label were significantly more likely to buy, and choose Unox soups over comparable soups of other brands in the future. However, an observation which can be made, is that since
participants were not asked about their particular shopping behavior regarding (prepackaged) soups, it is not possible to determine how results of purchase intentions relate to actual
(previous) purchase behavior of participants. This remark emerged in the comments section of the experiment: two participants indicated they never buy prepackaged soups and therefore rated their purchase intention relatively low. However, this does not take away the fact that purchase intention was significantly higher for participants in the color-coded GDA condition.
The current study found no significant effects of label type on brand credibility and brand attitude. There are three possible explanations for this. First, to determine brand
credibility, participants were not specifically asked about the trustworthiness/ believability of the fact that the brand showed the specific nutrition information. The conceptualization of brand credibility was namely based on two elements: trustworthiness (the notion that the brand is willing to deliver what it promises) and expertise (the notion that the brand is capable to deliver what it promises) (Sweeney & Swait, 2008). According to Baek et al. (2010) and in line with signaling theory, brands generally function as signals for (unobservable) quality and
30 consumers believe that brands are motivated to meet high expectations, and be truthful about their products and the claims they make (Sweeney & Swait, 2008). This could mean that, whether presented with a monochrome or a color-coded GDA label, consumers perceive the brand Unox as trustworthy and the nutrition information as just, which makes there is no difference found between conditions. Second, brand credibility (and also brand attitude) is a result of over time brand to customer communication, and represents a summary of the brand-consumer relationship (Sweeney & Swait, 2008). It seems plausible to assume that this relationship is not easily affected. Finally, three participants commented at the end of the experiment that they felt a bit wary of the fact that the brand Unox was represented in the experiment, and that that they were curious about the motives or influence that Unox had in the study. Although only mentioned by three participants, it could be that more participants experienced some level of skepticism. This may have impacted or distorted results on brand attitude and brand credibility. Nonetheless, it is not expected that this skepticism toward the brand has had an effect on the evaluation of the labels, because in the introductory text participants read it was mentioned that brands have to rely on European guidelines when creating these labels. Even when participants did not read this particular part of the text, it is not likely that participants experienced this skepticism. This because several studies have shown that consumers are relatively confident about Front of Pack information that is presented in the form of ‘fact panels’, because consumers generally seem to believe that this information is regulated by the government (Levy, 1995; Feunekes et al., 2008).
Future research
Although this study provides valuable insights into understanding the effects of showing a monochrome or color-coded GDA label on brand evaluations, I wish to emphasize the introductory nature of this study. Several recommendations for future research will be mentioned.
In the current study, product specific nutrient amounts, percentages and additional colors were shown. Future research should focus on different label assemblies that vary in their nutrient amounts, so that the effects of different color combinations (green, amber and red) can be tested separately, per product.
Although it seems a justified choice to use prepackaged soups as stimuli, examining products from one product category (prepackaged soups) is rather specific. Future research should therefore focus on different product categories. In doing so, these studies could additionally examine if the effects differ between well known and less or unknown brands, to
31 test whether this results in more profound results or effects on brand attitude and brand
credibility.
As mentioned in the discussion, another point of improvement is the simulation of grocery shopping situations, in which the opportunity to process nutrition information is significantly lower, than it was in this experimental setting. Future research should strive to provide experimental settings which better reflect natural grocery shopping situations and/or simulate corresponding cognitive/mental states.
Finally, the sample may not be the best representative of the larger population. Not only was the sample not that large, the average education level of the sample was relatively high and consisted of mostly female participants. It is therefore suggested that future research should strive for a more diverse sample, which for instance reflects different age groups, nationalities and education levels better.
REFERENCES
Ajzen, I. (1991). The theory of planned behavior. Organizational behavior and human
decision processes, 50(2), 179-211.
Anand, P.,& Sternthal, B. (1990). Ease of message processing as a moderator of
repetition effects in advertising. Journal of Marketing Research, 27, 345–353.
Anderson, J. R., & Reder, L. M. (1979). An elaborative processing explanation of depth of
processing. In S. Cermak and F.I.M. Craik (Ed.), Levels of processing in human
memory (pp. 385-403). Hillsdale, NJ: Lawrence Erlbaum Associates.
Andrews, J. C., Burton, S., & Kees, J. (2011). Is simpler always better? Consumer evaluations
of front-of-package nutrition symbols. Journal of Public Policy & Marketing, 30(2),
175-190.
Aschemann-Witzel, J., Grunert, K. G., Trijp, H. V., Bialkova, S., Raats, M. M., Hodgkins, C.,
Koenigstorfer, J. (2013). Effects of nutrition label format and product assortment on
