The Effect of a Multiple Traffic Light Label and the Moderating Effect of Type of Food on the Purchase Intentions of the Healthier Food Option

The Effect of a Multiple Traffic Light Label and the

Moderating Effect of Type of Food on the Purchase

Intentions of the Healthier Food Option

Effectively Promoting Healthy Food Choices

By

Julian Mikolajczyk

University of Groningen Faculty of Economics and Business

2

The Effect of a Multiple Traffic Light Label and the

Moderating Effect of Type of Food on the Purchase

Intentions of the Healthier Food Option

Effectively Promoting Healthy Food Choices

By

Julian Mikolajczyk j.mikolajczyk@student.rug.nl

S3032213

First supervisor: Martine van der Heide, MSc. Second supervisor: prof. dr. ir. Koert van Ittersum

Master Thesis University of Groningen Faculty of Economics and Business

3

Abstract

There is quite some disagreement on the effects of MTL labels. Some researchers say that MTL labels do not influence behavior and are purely intended to inform consumers about the healthiness of food. Conversely, other studies show that MTL labels do have such influence on consumer behavior and choices. This research aimed at filling this knowledge gap by examining the effectiveness of a MTL label in increasing the purchase intentions of the healthier option. Besides a moderating and main effect of the type of food were included. An online study was conducted in The Netherlands. Participants were randomly allocated to the control condition where a standard FOP label is used, or the MTL condition where a MTL label is used. Participants had to indicate their purchase intentions of twelve packaged food products regularly bought by Dutch consumers. The purchase intentions of the healthier option were tested using virtue, neutral and vice food categories. The use of a MTL label did not positively influence shoppers’ purchase intentions of the healthier option. However, when the moderating effect of type of food is added and the level of education is controlled for, the use of a MTL label did influence shoppers’ purchase intentions of the healthier option. The MTL label did not have a stronger effect for neutral foods (vs. vice vs. virtue) in selecting the healthier option. However, the type of food (virtue, neutral and vice) did influence the purchase intentions of the healthier option. The highest purchase intentions were found in the virtue category and the lowest in the vice category.

4

Contents

2.1 How do FOP labels work? ...8

2.2 What is the optimal design of a MTL label? ...9

2.3 Do FOP labels influence purchase intentions? ... 10

2.4 Does the type of food have an effect on purchase intentions? ... 12

2.5 Conceptual model ... 14

3. Methodology ... 15

3.1 Participants ... 15

3.2 Design ... 15

3.3 Nutrition label format ... 15

3.4 Procedure ... 16

3.5 Measurement ... 16

3.6 Products ... 17

4. Results ... 18

4.1 Health consciousness... 18

4.2 Purchase intentions of the healthier option... 19

4.3 Purchase intentions controlled for gender, age and level of education ... 22

4.4 Perceived healthiness ... 23

4.5 Influence of perceived healthiness on purchase intentions ... 26

5. Discussion ... 27

5.1 Conclusion ... 29

References ... 31

Appendix 1. Output Factor Analysis ... 35

Appendix 2. Output Reliability Analysis... 37

Appendix 3. Output Repeated Measures ANOVA (purchase intentions) ... 39

Appendix 4. Output Repeated Measures ANOVA (perceived healthiness) ... 41

Appendix 5. Output Repeated Measures ANCOVA (gender, age, level of education) ... 43

5

1. Introduction

Whereas Back-Of-Pack (BOP) nutrition information is mandatory on most prepacked foods, Front-Of-Pack (FOP) nutrition information is still limited and often voluntary in most countries. However, BOP nutrition information seems to be ineffective against the increasing obesity rates (Elshiewy and Boztug, 2018). Therefore, the political and societal acceptability of FOP nutrition labels is growing. This can be seen by the increasing FOP implementations and interventions by governments all over the world. Feunekes et al. (2009, p. 58) suggests that ‘’a FOP label would facilitate making healthier choices by incorporating benchmark information that enables consumers to interpret the information and/or by providing an advice that includes an interpretation of the nutritional information.’’ There are different FOP nutrition labels which vary in shape, color and size. Also the messages of the labels differ and can be proscriptive, prescriptive or both. Some labels focus on both positive and negative nutrients, others focus just on the critical nutrients. According to Kanter et al. (2018) the most common critical nutrients that have been included in FOP nutrition labelling schemes are fats, saturates, sugars and sodium. Sometimes FOP labels also include the amount of proteins, fibers, fruits and/or vegetables.

There are four main FOP nutrition labels that can be distinguished: Health labels (e.g. the Keyhole label, the Choices Programme label), Traffic light labels (e.g. single traffic light, multiple traffic lights), Warning labels (e.g. high in sugar, high in fat) and Summary indicator labels (e.g. Health Star Ratings, Nutri-score). According to Kanter et al. (2018, p.1401) these four main FOP nutrition labels can be categorized into four systems. 1) Interpretive Nutrition Rating System (INRS): provides nutrition information as guidance rather than specific facts. 2) Reductive system: shows information only, with no specific judgement, opinion or recommendation. 3) Evaluative/summary indicator system: combines several criteria to establish one indication of the healthiness of a product and shows judgement, opinion or recommendation with no specific information. 4) Nutrient-specific system: provides nutrition information for a set of nutrients.

6 proves to be most effective (Hawley et al., 2013; Hersey et al. 2013; Maubach and Mather, 2014; Cecchini and Warin, 2016). According to Feunekes et al. (2009, p. 58) a MTL label ‘’shows the amount of the five key nutrients energy, total fat, saturated fatty acids, sugar and salt in each serving. The nutrients can score green, amber or red, respectively, indicating ‘‘Go’’, ‘‘Ok’’ and ‘‘Think before you eat too much of this although a little bit will never hurt.’’ Currently there are three countries that implemented traffic light labels. South Korea was the first country to implement a voluntary traffic light label on children’s food products in 2011 for total sugars, fat, saturated fat and sodium. The United Kingdom followed in 2013 by implementing a voluntary traffic light label for energy, fat, saturated fat, sugar and salt. In the same year Ecuador implemented a mandatory traffic light label for sugar, fat and sodium.

In The Netherlands the Consumentenbond (Consumer Association) lobbies to introduce a traffic light system. Currently there are no official accepted FOP labels that indicate the healthiness of a food product in The Netherlands. Research shows that the old FOP label (the Choices Programme) was not clear to consumers (Consumentenbond, 2016). After a successful campaign against the label the minister of Public health, Welfare and Sport announced that it will disappear in 2018 (Voedingscentrum, 2016). Conversely, another research shows that 71% of the Dutch consumers are positive about having a FOP label. In the research three FOP labels are compared and they found that consumers were most positive about the MTL label. What they liked about the MTL label was that it is easy to understand, complete, logical, the colors attract attention and make the label easily readable (Consumentenbond, 2018).

8

2. Literature review

2.1 How do FOP labels work?

According to Kanter et al. (2018) the variety of FOP nutrition labels have two common goals. The first goal is to inform, guide and shape food choices and behaviors. This is done by communicating complex information to consumers in an easily understood and standardized format. Though, the variety of FOP nutrition labels differ in the extent to which they assist consumers to make healthfulness judgements. Evidence on the effectiveness of FOP nutrition labels on consumer understanding and perception of the healthiness of food is quite established. However, evidence on the effect they have on actual behavior is still limited. The second goal is to stimulate industry reformulation. ‘’FOP labels have a positive stimulating effect on product innovation in that it encourages food manufacturers to alter the nutritional composition of their foods in beneficial ways’’ (Kleef and Dagevos, 2015, p. 300). Thus, food manufacturers that make their products healthier (e.g. less sugar, less fat) in order, for example in the case of a MTL label, to get a ‘greener’ traffic light (Baltas, 2001).

‘’The inclusion of more attention-grabbing and easily interpretable front-of-pack (FOP) nutrition information is one of the public policies that can be implemented to empower consumers to identify unhealthful food products and to make more informed food choices’’ (Machín, Aschemann-Witzel et al., 2017, p.55). So attention and the ease of information processing are important concepts when it comes to effective FOP labels. According to Kleef and Dagevos (2015, p.300) the different FOP nutrition labels signal the healthiness of a food product in different ways. They also state that when shopping for groceries, due to time pressure and distractions, people have limited motivation and opportunity to extensively look for information. Another factor influencing the level of attention and processing of information is the goal of the shopping trip. When consumers have a specific health goal they are more likely to process the information on the FOP label, especially when these health goals are focused on specific nutrients (van Herpen and van Trijp, 2011). However, ‘’it has been argued that the contribution of FOP labels to dietary change is minor at best, and that their ability to modify consumers' food choices strongly depends on consumer interest in and motivation for eating more healthfully’’ (Machín, Aschemann-Witzel et al., 2017, p.56).

9 et al., 2006). According to Elshiewy and Boztug (2018, p.56) avoiding unhealthy nutrients like e.g. sugars and fats is more important to consumers, than consuming healthy nutrients. Nutrition labels are capable in disclosing such information, increase attention to it, and reduce the processing costs necessary to evaluate product healthiness. They also provide a framework to define consumers who make healthier purchases due to nutrition information. ‘’First, consumers need to recognize the label. Then, they need to be motivated and able to process the information. The disclosed information must increase nutrition knowledge, prompt a desire for dietary change, and enable the consumer to use the information for a change in behavior. Importantly, any trade-off considerations must ultimately turn in favor of health.’’

2.2 What is the optimal design of a MTL label?

A traffic-light system should include information about the content of key nutrients (energy, fats, saturates, sugars and salt), portion sizes that are meaningful and recognizable (e.g. 1 burger), percentage of Reference Intake (RI) (per portion), color code (red/amber/green) and could use text descriptors to classify the colors (low/medium/high) (Department of Health, 2016). When FOP nutrition labels are limited to unhealthy nutrition content consumers’ motivation is expected to increase, because these nutrients are more relevant to consumers (Elshiewy and Boztug, 2018). However, the interpretation of a MTL label can still be a difficult task since, ‘’this system still requires consumers to arrive at their own overall healthful judgement based on the simultaneous evaluation of the content of several nutrients’’ (Machín, Aschemann-Witzel et al., 2017, p.56).

10 only in comparison with other information numerical nutrition content can be interpreted meaningfully (Elshiewy and Boztug, 2018).

A MTL label could have some potential problems (Hawley et al., 2013, p.435, 436). During a study of the UK FSA (Food Standards Agency) some people did not realize that the colors red/amber/green had meaning. Some people thought the colors were used to make the labels more noticeable. In addition, others thought the colors were linked to specific nutrients (e.g. fats are always viewed in red). The study of the UK FSA found that this problem could be solved by adding the text high/medium/low to the colors and in this way consumers could better estimate the correct levels of nutrients in food products (Malam et al., 2009). The research concluded that when using a traffic light label the best liked and most comprehensive label is a combination of using the colors red, amber and green, a text that indicates high, medium or low, and a percentage of Guideline Daily Amount (GDA). Because of new European rules the new term Reference Intakes is used instead of GDA.

Research from Roberto et al. (2012) investigated how the Facts Up Front labeling system could be improved. They found that adding a color-coded traffic light system including high/med/low text performed the best on measures of nutrition knowledge and label perceptions. Other research by Hersey et al. (2013, p.1) compared multiple papers that studied the ability of consumers to select healthier products using nutrient‐specific systems that incorporate text and color codes with multiple‐level summary icons. They found that ‘’studies indicate that consumers can more easily interpret and select healthier products with nutrient‐ specific FOP nutrition labels that incorporate text and symbolic color to indicate nutrient levels rather than nutrient‐specific labels that only emphasize numeric information, such as Guideline Daily Amounts expressed as percentages and/or grams.’’

2.3 Do FOP labels influence purchase intentions?

11 FOP labelling systems’ abilities to influence consumer behavior is needed. Van Herpen et al. (2014) found that the MTL label was most effective in communicating the healthfulness of products. This holds for products evaluated in isolation, in comparison with another product and also across and within product categories. Further, they state that the ability of nutrition labels to inform consumers about the healthfulness of a product is not necessarily sufficient for consumers to change their existing product preferences, and that more research is needed on this. Consumers can more easily identify healthier foods using nutrient-specific schemes labels (e.g. Traffic Light labels) compared to summary systems (e.g. Health Star Ratings). ‘’Summary systems may influence consumers to purchase healthier products.’’ However, when it comes to nutrients-specific systems more research is needed to assess the influence of such a label on consumers’ purchases (Hersey et al., 2013, p.1).

12 Even though at first studies did not find effects of FOP labels on purchase intentions or consumption some studies did expect that these effects may appear, but more research was needed to support this. As can be concluded from the previous some papers found that FOP labels have an effect on purchase intentions and consumption, however evidence for this effect is still limited. This can for example be seen in the results of the research of The National Institute for Health and Environment in The Netherlands. They concluded that a FOP label can fit in with the goal of the food policy to inform consumers about healthy food, but that no convincing evidence was found on the effects of FOP labels on consumer behavior or product innovation (Rijksinstituut voor Volksgezondheid en Milieu, 2016). Also Machín, Aschemann-Witzel et al. (2017) emphasize that evidence of FOP labeling schemes performance in the market situation is still limited. Nutritional label use affects purchasing behavior mainly because consumers want to avoid the negative nutrients in food products. These effects can be greater when the labeling is combined with an information campaign to educate consumers. Purchasing behavior is affected by the nutritional information, because it influences valuations and perceptions of the product (Drichoutis et al., 2006). Formally, this study proposes:

H1: The use of a Multiple Traffic Light label will positively influence shoppers’ purchase intentions of the healthier option.

2.4 Does the type of food have an effect on purchase intentions?

13 generally more easily classified as either ‘unhealthy’ (e.g. confectionery and sugar) or ‘healthy’ (e.g. fish, fruit and vegetables, and eggs).’’

In this study the type of food is categorized into virtues (healthy foods), vices (unhealthy foods) and neutral foods. According to van Doorn and Verhoef (2011, p.168) ‘’vice and virtue foods are typically conceptualized in relation to each other as relative virtues and relative vices. Relative vices (also known as “wants”) are products that provide immediate pleasurable experience (such as the good taste of chocolate cake), but contribute to negative long-term outcomes (such as future weight gain and related health problems). Relative virtues (also known as “shoulds”) are less gratifying and appealing in the short term but have less negative long-term consequences than vices and therefore are a more prudent choice.’’

When it comes to FOP labels the effect the type of food has on behavior is limited. Hence, the focus of this paper is to investigate if the moderating effect of type of food also holds for actual behavior. Thus, if there exists a moderating effect of the type of food (virtue vs. neutral vs. vice) on the purchase intention of the healthier option when using a MTL label. Formally, this study proposes:

H2: The use of a Multiple Traffic Light label will have a stronger effect on the purchase intentions of the healthier option for neutral foods (vs. vice vs. virtue).

Furthermore, this study predicts that the moderator type of food will also have a main effect on the purchase intentions of the healthier option, such that this effect will be stronger within the virtue category. This prediction is based on research by Nikolova and Inman (2015) which states that consumers are more likely to use nutrition information in healthier categories. Besides they found that consumers are more likely to ignore nutrition information of vice foods. Consumers tend to do this to prevent the negative emotions that could arise from considering how unhealthy the food is. ‘’In contrast, consumers are more willing to base their decisions on the nutrition content in healthier categories that meet their health-related goals’’ (Nikolova and Inman, 2015, p.820). Formally, this study proposes:

14

2.5 Conceptual model

The conceptual model in figure 1 summarizes the hypotheses and the central theme of the paper.

+ +

+

Multiple Traffic Light label Purchase intentions of the healthier option

Type of food: neutral foods (vs. vice foods vs. virtue foods

15

3. Methodology

3.1 Participants

The data was collected through an online survey between November 2018 and December 2018 in The Netherlands. The sample was drawn randomly from the population. The online survey was distributed under the personal network of the author via social media and email. The target group consisted of people of all ages between 16-75 living in The Netherlands.

3.2 Design

This study used a mixed factorial design. The independent variable (MTL label) is included as a between subjects factor with two levels (control condition vs. MTL label condition). The moderator variable (type of food) is tested within subjects on three levels (virtue vs. neutral vs. vice). This design means that participants were randomly assigned to one of the two experimental conditions. (I) A control condition with FOP nutrition information using a standard label, including virtue, neutral and vice foods. (II) A MTL condition with FOP nutrition information using a MTL label, including virtue, neutral and vice foods. Randomization was automatically generated by the online software used.

3.3 Nutrition label format

A MTL label was used that includes information about calories, fats, saturates, sugars and salt content per portion. Also, the percentage of Reference Intake of the nutrients, color code (red/amber/green) and text descriptors were used to classify nutrient content as low, medium and high, except for calorie content which is presented in white. See figure 2 for a standard label and figure 3 for a MTL label.

16

3.4 Procedure

Participants were exposed to pre-packaged products from two virtue categories, two neutral categories and two vice categories. Within these categories two options were presented: a healthier variant and a less healthy variant. A standard FOP label in the control condition (see figure 2) and a MTL label in the MTL label condition (see figure 3) were shown next to the products. In the MTL label condition information about the MTL label was given, where was explained how the label works and how the nutritional information should be interpreted. The order of presentation of the product categories was randomized over participants. Also the presentation of the products was randomized (showing the healthier variant first vs. showing the less healthy variant first).

Initially, participants were asked to indicate their purchase intentions of the two presented options per category. Then, participants had to rate the healthiness of the twelve products presented individually (also in randomized order). First asking the purchase intentions ensures that the participants are not influenced by the perceived healthiness question, and thus are not put into a ‘healthiness mindset’ which could influence the purchase intentions. The perceived healthiness of the products was added as a manipulation check to see if the participants could classify the virtue products as healthy, the neutral category as healthy nor unhealthy, and the vice category as unhealthy. After rating the products the participants were asked to complete questions on demographics and their health consciousness.

3.5 Measurement

The purchase intentions and perceived healthiness were measured using the scales of Feunekes et al. (2009, p. 60). All questions employed 5-point Likert-type rating scales, unless otherwise indicated. Purchase intentions was measured by the question, ‘How likely is it that you will buy this product?’ (1=extremely unlikely, 5=extremely likely). The perceived healthiness was measured by the question ‘How healthy is this product to you?’ (1=not healthy at all, 5=very healthy). Also some demographic question were included that asked about their age, gender, education level and dietary restrictions.

17 and I do not worry much about the healthiness of food. 4) It is important to me that my diet is low in fat. 5) I always follow a healthy and balanced diet. 6) It is important for me that my daily diet contains a lot of vitamins and minerals. 7) The healthiness of snacks makes no difference to me. 8) I do not avoid foods, even if they may raise my cholesterol. These questions were measured on a 7-point Likert scale (1=strongly disagree, 7=strongly agree).

3.6 Products

18

4. Results

In total 103 respondents completed the survey. Of these respondents 54.4% is male and 45.6% is female. The sample is quite young with 75.7% of the respondents in the age range between 18 and 34 years old. Further, the sample is highly educated with 45.6% of the respondents having a Bachelor’s degree and 28.2% a Master’s degree. All data analyses were performed in SPSS software.

4.1 Health consciousness

The health consciousness of the respondents was analyzed with a factor analysis using the method of Maximum Likelihood with Varimax rotation, and the internal reliability of each factor was tested using Cronbach’s alpha. The results of these analyses can be found in Appendix 1 and 2. First tested was if the factor analysis was appropriate. The KMO measure of sampling adequacy showed a score of p=.714 which is good. Bartlett’s test of Sphericity shows a highly significant result of p=.000, so we reject the null hypothesis that the factors are uncorrelated. The communalities of all items are >.4 which is good. Looking at the Rotated Component Matrix we can see Factor 1 consists of: item 2, 4, 5 and 6. Factor 2 consists of item 1,3 and 8. And factor 3 consists of item 7. Since only item 7 loads on factor 3 this factor will be discarded. Factor 1 is labeled as Healthy Diet and Factor 2 is labeled as Healthiness of food choices. Item 1, 3, 7 and 8 were recoded into new variables since they were negatively stated. The internal consistency of the marker items was tested with a reliability analysis. Factor 1 shows a Cronbach’s Alpha of .690 which is good and factor 2 shows a Cronbach’s Alpha of .651 which also above the threshold of α=.6. These two factors are able to explain 50.89% of the variance which is not ideal, but still acceptable. Factor 1 has a mean of 4.32 and factor 2 a mean of 4.49. This shows that the respondents scored quite neutral on the health consciousness question. See table 1 for an overview of the results of the factor and reliability analysis.

Table 1: Health consciousness

Description of the subscales to determine the health consciousness

Statement Mean SD Factor

loading

Communality Item-total correlation

Healthy Diet

2. I am very particular about the healthiness of

19 food I eat.

4. It is important for me that my diet is low in fat

3.93 1.6 .718 .530 .643

5. I always follow a healthy and balanced diet.

3.87 1.6 .678 .601 .585

6. It is important for me that my daily diet contains a lot of vitamins and minerals.

5.15 1.4 .778 .631 .653

Eigenvalue= 2.58 Variance%= 32.29 Cronbach’s a= .690

Healthiness of food choices

1. The healthiness of food has little impact

4.69 1.6 .758 .622 .440

3. I eat what I like and I do not worry much about the healthiness of food.

4.77 1.7 .763 .672 .526

8. I do not avoid foods, even if they may raise my cholesterol.

4.00 1.7 .705 .543 .422

Eigenvalue= 1.49 Variance%= 18.60 Cronbach’s a= .651

4.2 Purchase intentions of the healthier option

20 On average the highest purchase intentions of the healthier option were found in the virtue category (mean= 3.61), however this was the only category where the purchase intentions decreased in the MTL condition. The neutral category (mean= 3.26) has the second highest purchase intentions of the healthier option. This category showed the highest increase in purchase intentions in the MTL condition. The lowest purchase intentions of the healthier option were found in the vice category (mean=3.10) and these purchase intention increased slightly in the MTL condition. A summary of these results can be found in table 2.

Table 2: Mean purchase intentions control and MTL condition

Purchase intentions of the healthier option Food type Control=1 MTL=2 Mean Std. Deviation N Virtue 1,00 3.75 1.0 46 2,00 3.47 1.0 44 Total 3.61 1.0 90 Neutral 1,00 3.15 1.0 46 2,00 3.36 0.9 44 Total 3.26 0.9 90 Vice 1,00 3.05 1.1 46 2,00 3.14 1.1 44 Total 3.10 1.1 90

21 the healthier option. So we reject H1: The use of a Multiple Traffic Light label will positively influence shoppers’ purchase intentions of the healthier option.

There was a significant main effect of type of food being rated F(2,176)=8.031, p=.000. This effect indicates that the ratings of the three food categories significantly differed. This means that the type of food has a main effect on the purchase intentions of the healthier option, whereas the highest purchase intentions were found in the category virtue foods. The Bonferroni corrected post hoc tests showed that the ratings of the food categories 1 virtue and 2 vice significantly differ (p=.002). The category virtue and neutral also significantly differ (p=.025). However, the category vice and neutral did not significantly differ (p=.534). Figure 4 shows that the ratings between the virtue and vice category are relatively big (mean virtue=3.61, mean vice=3.10). The figure also shows that the difference between the neutral and vice category are relatively small (mean neutral=3.26, mean vice=3.10). So we accept H3: The purchase intentions of the healthier option will be higher for virtue foods (vs. vice vs virtue).

Figure 4: Mean purchase intentions

22 intentions of the healthier option for neutral foods (vs. vice vs. virtue). The small differences between the ratings from the control and MTL condition in the different food categories can be seen in figure 5.

Figure 5: Mean purchase intentions control and MTL condition

4.3 Purchase intentions controlled for gender, age and level of education

Multiple repeated measures analyses of covariance (ANCOVA) considering a 95% confidence level are used to analyze if gender, age and level of education influence the purchase intentions of the healthier option. All covariates were recoded into categorical variables. Gender was already coded correctly in male and female, age was divided in a under 34 years and a above 34 years group and level of education was divided in low educated (lower than a Bachelor’s degree) and highly educated (Bachelor’s degree and higher). Since these variables are categorical they were added as Fixed Factors (independent factors) in the ANCOVA.

23 The covariate age shows that the under 34 age group (mean=3.38) had slightly higher purchase intentions of the healthier option than the above 34 age group (mean=3.14). The analysis did not show significant results for the main effect of age F(1, 86)=1.941, p=.167. So the ratings between the age groups are similar. Controlling for age did not lead to significant results in the main effect of Control versus MTL condition F(1, 86)=.112, p=.739 and did also not lead to significant results in the interaction effect of Type of Food x Control versus MTL condition F(2, 172)=1.933, p=.148.

Highly educated respondents (mean=3.41) have higher purchase intentions of the healthier option than low educated respondents (mean=3.03). The analysis shows significant results for the main effect of level of education F(1, 86)=4.342, p=.040. So the ratings between highly educated and low educated respondents significantly differ. Controlling for level of education did not lead to significant results in the main effect of Control versus MTL condition F(1, 86)=.115, p=.735. However, level of education did lead to significant results in the interaction effect of Type of Food x Control versus MTL condition F(2, 172)=3.347, p=.037.

4.4 Perceived healthiness

The perceived healthiness of the products was added as a manipulation check to see if the participants could classify the virtue products as healthy, the neutral category as healthy nor unhealthy, and the vice category as unhealthy. In order to analyze this a repeated measures analysis of variance (ANOVA) considering a 95% confidence level is used. The results of this analysis can be found in Appendix 4.

24

Table 3: Mean perceived healthiness control and MTL condition

Perceived Healthiness Food type Control=1 MTL=2 Mean Std. Deviation N Virtue 1,00 3.58 0.5 54 2,00 3.50 0.6 51 Total 3.54 0.5 105 Neutral 1,00 3.24 0.8 54 2,00 3.13 0.7 51 Total 3.19 0.8 105 Vice 1,00 1.44 0.6 54 2,00 1.57 0.6 51 Total 1.50 0.6 105

Mauchly’s Test of Sphericity shows a significant value of p=.031, which is less than p=.05, so we cannot reject the hypothesis that the variances of the differences between levels were significantly different. In other words the assumption of sphericity has not been met. So we use Greenhouse-Geisser corrected degrees of freedom to assess the significance of the corresponding F values. Looking at Levene’s Test of Equality of Error Variances we see that all three food categories are above the significance value of p=.05. This indicates that variances are homogeneous for all levels of the repeated measures variables. There was a non-significant main effect of control condition versus MTL condition F(1, 103)=.099, p=.754. This effect indicates that the control condition (mean=2.76) and MTL condition (mean=2.73) gave similar ratings. So the perceived healthiness of the products do not differ when comparing a MTL label to a standard FOP label.

25 figure also shows that the difference between the virtue and neutral category relatively small (mean virtue=3.54, mean neutral=3.19).

Figure 6: Mean perceived healthiness

The interaction effect Type of Food x Control versus MTL condition was insignificant, F(1.877, 193.306)=1.312, p=.271. So there is no interaction effect between the use of a MTL label and the type of food when it comes to the perceived healthiness. This indicates that the respondents gave similar ratings in the control condition and MTL condition when looking at the different food categories. The small differences between the ratings from the control and MTL condition can be seen in figure 7.

26

4.5 Influence of perceived healthiness on purchase intentions

27

5. Discussion

There is quite some disagreement on the effects of MTL labels. Some researchers say that MTL labels do not influence behavior and are purely intended to inform consumers about healthy food (Hawley et al., 2013; Borgmeier and Westenhoefer, 2009; van Herpen et al., 2014). Conversely, other studies show that MTL labels do have such influence on consumer behavior and choices (Cecchini and Warin, 2016; Machín, Aschemann-Witzel et al., 2017; Mhurchu et al., 2018). This research aimed at filling this knowledge gap by examining the effectiveness of a MTL label in increasing the purchase intentions of the healthier option. Besides a moderating and main effect of the type of food were included. Some papers found that FOP labels mainly influence products that have a positive and healthy image. However, these studies found an effect on health perceptions, not on actual behavior. The focus of this paper was to investigate if the moderating and main effect of type of food also holds for actual behavior. An online study was conducted with a control condition and a MTL label condition. The purchase intentions of the healthier option were tested using virtue, neutral and vice food categories.

Not in line with the hypothesis, the MTL condition did not significantly differ compared to the control condition (where a standard FOP label is used). So the use of a MTL label did not positively influence shoppers’ purchase intentions of the healthier option. As expected, the biggest increase in purchase intentions was found in the neutral category. However, the results showed no significant differences compared to the control condition. Since this effect is not significant we reject the hypothesis that the use of a MTL label will have a stronger effect for neutral foods (vs. vice vs. virtue) in selecting the healthier option. However, the type of food (virtue, neutral and vice) did influence the purchase intentions of the healthier option. The highest purchase intentions were found in the virtue category and the lowest in the vice category. This is in line with the hypothesis, which stated that the purchase intentions of the healthier option will be higher for virtue foods (neutral vs. vice).

28 difference in purchase intentions when looking at the interaction effect between the use of a MTL label and the type of food. So the use of a MTL label does influence shoppers’ purchase intentions of the healthier option when the moderating effect of type of food is added and the level of education is taken into account.

Also the perceived healthiness of the products was included to check if the respondents could determine the healthiness of the products when using a MTL label. The control condition showed similar scores as the MTL condition. In both conditions the respondents were able to classify the virtue products as (moderately) healthy, neutral products as healthy nor unhealthy and the vice products as unhealthy. So the perceived healthiness of the products was assessed correctly by the respondents and did not differ between the MTL label and the standard FOP label. Results of a regression analysis showed that the perceived healthiness of the products did not influence the purchase intentions. Besides the health consciousness of the respondents was tested. The sample was neutral to moderately health conscious. The healthiness of food choices was perceived slightly more important than following a healthy diet.

It should be noted that evidence on the ability of MTL labels to influence the purchase intentions is limited. The paper of Machín, Aschemann-Witzel et al. (2017) is one of the few papers that found that a MTL label did increase the purchase intentions of the healthier option. However, these effects only occurred when a health goal was added to the shopping trip. A previous study with the same study design and context, but without a health goal, showed no significant results. In this study no health goal was induced. The ability of FOP labels to influence food choices depends strongly on consumer interest in and motivation for eating more healthfully (Machín, Aschemann-Witzel et al., 2017, p.56). The sample in this study scored quite neutral on the health consciousness measure, which could imply that the sample did not had the motivation to eat more healthfully. This can explain why no significant results were found.

29 MTL label and their influence on purchase intentions is limited. It could be that adding these features to a label have a small impact on the purchase intentions.

After controlling for level of education significant results were found. However, the group of low educated people was quite small. This influences the robustness of this finding. Also, the experimental setting of this study might give different results compared to a real life situation, since the processing of information is diverse. When people shop for groceries they have limited time and are faced with distractions. This could in turn lead to limited motivation and opportunity to extensively look at the FOP information (Kleef and Dagevos, 2015). So FOP labels might be less effective in a shopping environment. Furthermore, the findings are based on a non-representative sample. The sample was quite young and highly educated. So the findings cannot be generalized to the population of The Netherlands. Also the findings are based on a hypothetical choice of products.

Future research could have a focus on measuring actual behavior in a more real-life experiment. This can be for example be done by creating a virtual shopping task where respondents have more freedom in choosing their own groceries (instead of having to rate fixed products). Also more in-depth questions could be asked about the health goals and motivations of the respondents to see how much impact this has on their purchase intentions. This could be interesting since health goals and motivations influence the level of attention and processing of information. Besides, more FOP labels could be compared to the MTL label. The labels in this study were quite similar. Adding different FOP labels might lead to significant differences in purchase intentions. Controlling for level of education did show significant results. However, in this study no theory was devoted to this and the sample size of low educated respondents was small. Future research could give a bigger role to the level of education when investigating the effects of MTL labels on the purchase intentions of the healthier option.

5.1 Conclusion

31

References

Baltas, G. (2001). Nutrition labelling: issues and policies. European journal of marketing, 35(5/6), 708-721.

Cecchini, M., Warin, L. (2016). Impact of food labelling systems on food choices and eating behaviours: a systematic review and meta‐analysis of randomized studies. Obesity reviews, 17(3), 201-210.

Consumentenbond (2016), Resultaten Panelonderzoek Het Vinkje. Retrieved from:

https://www.consumentenbond.nl/binaries/content/assets/cbhippowebsite/actie-voeren/vinkjes/opmaak_resultaten_onderzoek_het_vinkje-v4b.pdf

Consumentenbond (2018), Consumentenonderzoek Voedselkeuzelogo’s. Retrieved from: https://www.consumentenbond.nl/binaries/content/assets/cbhippowebsite/landingspaginas/acti es/weet-wat-je-eet/consumentenonderzoek-voedselkeuzelogos-nl.pdf

Department of Health. (2016). Guide to creating a front of pack (FoP) nutrition label for pre-packed products sold through retail outlets. London: Department of Health.

Drichoutis, A. C., Lazaridis, P., & Nayga Jr, R. M. (2006). Consumers' use of nutritional labels: a review of research studies and issues. Academy of marketing science review, 2006, 1.

Elshiewy, O., & Boztug, Y. (2018). When Back of Pack Meets Front of Pack: How Salient and Simplified Nutrition Labels Affect Food Sales in Supermarkets. Journal of Public Policy & Marketing, 37(1), 55-67.

Feunekes, G. I., Gortemaker, I. A., Willems, A. A., Lion, R., & Van Den Kommer, M. (2008). Front-of-pack nutrition labelling: testing effectiveness of different nutrition labelling formats front-of-pack in four European countries. Appetite, 50(1), 57-70.

32 Hersey, J. C., Wohlgenant, K. C., Arsenault, J. E., Kosa, K. M., & Muth, M. K. (2013). Effects of front‐of‐package and shelf nutrition labeling systems on consumers. Nutrition Reviews, 71(1), 1-14.

Hui, S. K., Bradlow, E. T., & Fader, P. S. (2009). Testing behavioral hypotheses using an integrated model of grocery store shopping path and purchase behavior. Journal of consumer research, 36(3), 478-493.

Kanter, R., Vanderlee, L., & Vandevijvere, S. (2018). Front-of-package nutrition labelling policy: global progress and future directions. Public health nutrition, 21(8), 1399-1408.

Kleef, E. V., & Dagevos, H. (2015). The growing role of front-of-pack nutrition profile labeling: a consumer perspective on key issues and controversies. Critical reviews in food science and nutrition, 55(3), 291-303.

Machín, L., Aschemann-Witzel, J., Curutchet, M. R., Giménez, A., & Ares, G. (2017). Does front-of-pack nutrition information improve consumer ability to make healthful choices? Performance of warnings and the traffic light system in a simulated shopping experiment. Appetite, 121, 55-62.

Machín, L., Arrúa, A., Giménez, A., Curutchet, M. R., Martínez, J., & Ares, G. (2017). Can nutritional information modify purchase of ultra-processed products? Results from a simulated online shopping experiment. Public Health Nutrition, 21(1), 49-57.

Machín, L., Cabrera, M., Curutchet, M. R., Martínez, J., Gimenez, A., & Ares, G. (2017). Consumer perception of the healthfulness of ultra-processed products featuring different front-of-pack nutrition labeling schemes. Journal of Nutrition Education and Behavior, 49, 330-338.

Malam, S., Clegg, S., Kirwan, S., McGinigal, S., Raats, M., Shepherd, R., ... & Dean, M. (2009). Comprehension and use of UK nutrition signpost labelling schemes. London: Food Standards Agency.

33 Mhurchu, C. N., Eyles, H., Jiang, Y., & Blakely, T. (2018). Do nutrition labels influence healthier food choices? Analysis of label viewing behaviour and subsequent food purchases in a labelling intervention trial. Appetite, 121, 360-365.

Nikolova, H. D., & Inman, J. J. (2015). Healthy choice: the effect of simplified point-of-sale nutritional information on consumer food choice behavior. Journal of Marketing Research, 52(6), 817-835.

Overheid.nl (2017). Voedingsbeleid: Brief van de minister van Volksgezondheid, Welzijn en Sport. Retrieved from: https://zoek.officielebekendmakingen.nl/kst-31532-184.html

Rijksinstituut voor Volksgezondheid en Milieu (2016). De waarde van een voedselkeuzelogo

voor het voedingsbeleid. Retrieved from:

https://www.rijksoverheid.nl/documenten/rapporten/2016/12/09/de-waarde-van-een-voedselkeuzelogo-voor-het-voedingsbeleid

Roberto, C. A., Bragg, M. A., Schwartz, M. B., Seamans, M. J., Musicus, A., Novak, N., & Brownell, K. D. (2012). Facts up front versus traffic light food labels: a randomized controlled trial. American journal of preventive medicine, 43(2), 134-141.

Roininen, K., Lähteenmäki, L., & Tuorila, H. (1999). Quantification of consumer attitudes to health and hedonic characteristics of foods. Appetite, 33(1), 71-88.

Van Doorn, J., & Verhoef, P. C. (2011). Willingness to pay for organic products: Differences between virtue and vice foods. International Journal of Research in Marketing, 28(3), 167-180.

Van Herpen, E., Hieke, S., & van Trijp, H. C. (2014). Inferring product healthfulness from nutrition labelling. The influence of reference points. Appetite, 72, 138-149.

34

Voedingscentrum (2016) Vinkje. Retrieved from:

https://www.voedingscentrum.nl/encyclopedie/Vinkje.aspx

Watson, W. L., Kelly, B., Hector, D., Hughes, C., King, L., Crawford, J., & Chapman, K. (2014). Can front-of-pack labelling schemes guide healthier food choices? Australian shoppers’ responses to seven labelling formats. Appetite, 72, 90-97.

37

Appendix 2. Output Reliability Analysis

43

Appendix 5. Output Repeated Measures ANCOVA (gender, age, level of

education)

The Effect of a Multiple Traffic Light Label and the

Moderating Effect of Type of Food on the Purchase

Intentions of the Healthier Food Option

Julian Mikolajczyk

S3032213

Introduction

•

Front-of-Pack (FOP) labels

•

Increasing implementations and interventions

•

Disagreement on the effects of MTL labels

•

Inform about the healthiness of food

(Borgmeier and Westenhoefer, 2009; van Herpen

et al., 2014)

•

Influence behavior and choices

(Machín, Aschemann-Witzel et al., 2017; Mhurchu et al.,

2018)

‘’More research on the impact FOP labelling systems have on purchasing and

Theoretical framework

How do FOP labels work?

(Kanter et al., 2018)

•

Inform, guide and shape food choices and behaviors

•

Stimulate industry reformulation

Does the type of food have an effect on purchase intentions?

•

Largest effect of FOP labels found for ambiguous foods (

Mhurchu et al., 2018)

•

Consumers are more likely to use nutrition information in healthier

categories

(Nikolova and Inman, 2015)

What is the optimal design of

a MTL label?

(Malam et al., 2009;

Conceptual model

•

H1: The MTL label will positively influence shoppers’ purchase intentions of the

healthier option.

•

H2: The MTL label will have a stronger effect on the purchase intentions for

neutral foods (vs. vice vs. virtue).

•

H3: The purchase intentions of the healthier option will be higher for virtue

Methodology

Control condition

Results

H1 rejected: No significant main effect of a MTL label (p=.984)

•

(mean control=3.32 and mean MTL=3.32)

H2 rejected: No significant moderating effect of type of food (p=.148)

•

Neutral category biggest increase

H3 accepted: Significant main effect of type of food (p=.000)

Results

Purchase intentions of the healthier option controlled for gender, age and level

of education

•

No significant main effect found for gender (p=.537) and age (p=.897).

•

Significant main effect found for level of education (p=.040).

•

Highly educated people (mean=3.41) have higher purchase intentions than low

educated people (mean=3.03).

Controlling for level of education and adding the moderating effect of type of

food, the MTL label did positively influence shoppers’ purchase intentions

Discussion

‘’The ability of FOP labels to influence food choices depends strongly on

consumer interest in and motivation for eating more healthfully’’

(Machín,

Aschemann-Witzel et al., 2017, p.56)

•

Effective MTL labels

•

Health goals

•

Educational campaigns

•

Motivate consumers to eat more healthfully

•

Increase their health consciousness

References

• Hawley, K. L., Roberto, C. A., Bragg, M. A., Liu, P. J., Schwartz, M. B., & Brownell, K. D. (2013). The science

on front-of-package food labels. Public Health Nutrition, 16, 430-439.

• Kanter, R., Vanderlee, L., & Vandevijvere, S. (2018). Front-of-package nutrition labelling policy: global

progress and future directions. Public health nutrition, 21(8), 1399-1408.

• Machín, L., Aschemann-Witzel, J., Curutchet, M. R., Giménez, A., & Ares, G. (2017). Does front-of-pack

nutrition information improve consumer ability to make healthful choices? Performance of warnings and the traffic light system in a simulated shopping experiment. Appetite, 121, 55-62.

• Malam, S., Clegg, S., Kirwan, S., McGinigal, S., Raats, M., Shepherd, R., ... & Dean, M. (2009).

Comprehension and use of UK nutrition signpost labelling schemes. London: Food Standards Agency.

• Mhurchu, C. N., Eyles, H., Jiang, Y., & Blakely, T. (2018). Do nutrition labels influence healthier food

choices? Analysis of label viewing behaviour and subsequent food purchases in a labelling intervention trial. Appetite, 121, 360-365.

• Nikolova, H. D., & Inman, J. J. (2015). Healthy choice: the effect of simplified point-of-sale nutritional

information on consumer food choice behavior. Journal of Marketing Research, 52(6), 817-835.

• Roberto, C. A., Bragg, M. A., Schwartz, M. B., Seamans, M. J., Musicus, A., Novak, N., & Brownell, K. D.

(2012). Facts up front versus traffic light food labels: a randomized controlled trial. American journal of preventive medicine, 43(2), 134-141.

• Van Herpen, E., Hieke, S., & van Trijp, H. C. (2014). Inferring product healthfulness from nutrition

labelling. The influence of reference points. Appetite, 72, 138-149.

• Westenhoefer, J., & Borgmeier, I. (2009). Impact of different food label formats on healthiness evaluation