Master Thesis MSc in Marketing Management “A choice-based conjoint analysis: Can we trust the traffic light labeling system?”

(1)

Master Thesis

MSc in Marketing Management

“A choice-based conjoint analysis: Can we trust the traffic

light labeling system?”

Philipp Leirich

(2)

“A choice-based conjoint analysis: Can we trust the traffic

light labeling system?”

Master Thesis

Faculty of Economics and Business

MSc Marketing Management

Date: 14-06-2016 Supervisors: prof. dr. ir. K. van Ittersum

(3)

1

Management Summary

The goal of this research is to provide highly needed insights of the supporting influence of a colored traffic light labeling system on the healthiness of choice. By introducing additionally a promotion and prevention focus through salience of color, this research investigates if the expected positive influence can be even enhanced. Moreover, differences between weak and strong dietary restraints were studied.

Based on the results, it can be assumed that the use of a colored traffic light labeling system (vs. numerical) has a positive impact on the healthiness of choice. The average means indicate, that the use of a colored traffic light label (vs. numerical) is up to five times healthier. Interestingly, the positive effect can be enhanced by using a promotion focus through salience of color, which contains an expansion in size of low level and therefore healthy nutrients. This holds only true for the condition of promotion and is not applicable for a prevention focus of high level (unhealthy) nutrients. Further, this research does not provide any evidence that there are any differentiating effects between participants who were weak or strong in their dietary restraints.

Following these findings, some managerial recommendations can be given. Health organization as well as the government can positively influence public health on voluntary basis by introducing a mandatory traffic light labeling for food and drink packages. On basis of a promotion focus, the use of salience of color for low level nutrients can enhance the positive effect in order to reduce nutrition-related diseases. A differentiation between in terms of dietary restraints is not necessary and will not positively nor negatively influence the outcome.

Keywords: traffic light label, salience of color, healthiness of choice, food packages, FOP

(4)

2

Preface

With this research paper, my time as a master student at faculty of economics and business, ends. Becoming a student in the field of marketing management at the RUG was a great pleasure, a challenge and a not describable experience which passed by too quickly. Coming to the end of my academic career, I can say that focusing my studies on marketing is the right choice and will guide me through my future business career. I appreciate what the whole staff of the RUG did for me and other students during the past year and I leave the Netherlands with a lot of new knowledge, experiences and even some new friends.

Special thanks go to my supervisor professor Koert Van Ittersum who guided and supported me during the past 6 month with a lot of passion and valuable insights. Additionally, I would like to thank Dr. Ellen Van Kleef from the Wageningen University, who was highly interested in my research topic and also gave me important constructive feedback during the whole process of writing my master thesis.

Philipp Leirich

(5)

3

Table of Content

1 Introduction ... 4

2 Theoretical Framework ... 6

2.1 Literature Review ... 6

2.1.1 Food Labeling Systems ... 6

2.1.1.1 Back-of-Pack Nutrition Facts ... 6

2.1.1.2 Guideline Daily Amount ... 7

2.1.1.3 Traffic Light Labeling System ... 8

2.1.2 Dietary Restraints ... 9

2.1.3 Information Processing and Salience of Color ... 11

2.2 Conceptual Model and Hypothesis ... 13

3 Research design ... 14

3.1 Method: Choice-Based Conjoint Analysis... 14

3.2 Attributes and Levels ... 15

3.3 Questionnaire ... 15

3.4 Choice Design ... 17

3.5 Preparing for Analysis ... 18

4 Results ... 20

4.1 Sample ... 20

4.2 Testing the Hypothesis ... 22

5 Conclusion and discussion ... 28

5.1 Managerial implications and recommendations ... 30

5.2 Limitations and Directions for Future Research ... 31

6 References ... 33

7 Appendices ... I

7.1 Appendix A: Qualtrics Questionnaire... I

7.2 Appendix B: SPSS Output ... X

(6)

4

1 Introduction

(7)

5

an important factor of the traffic light label, the provision of information processing cues can influence the customer’s decision making process with a “stop” and “go” logic behind the colors green and red (Trudel et al., 2015). Connecting this with the regulatory focus theory, it is possible that consumers associate the color green with a promotion focus, which means they have the enduring or acute goal to focus on the achievement of gains and acquire positive outcomes. Or, in contrast, consumers with a prevention focus are not very motivated by achieving gains, but more by preventing losses and negative outcomes, illustrated by the color red (Higgins, 1997). To combine the colored traffic light labeling system with the focus regulatory theory, salience of color plays an immanent role and is part of further investigation. Hence, a traffic light label needs to catch the customers’ visual attention through the use of color (red, yellow, green) and size (extension), to activate a promotion or prevention focus. Based on this background, the main research questions are as follows:

Does the implementation of a traffic light (i.e. color-coding, text, percentage of GDA) as a front-of-package (FOP) nutrition label influence the healthiness of consumer choice?

Is the impact of a colored TL on Consumer Choice predominantly influenced by dietary restraints?

Is the salience of the colors red and green triggering a promotion or prevention focus which next increases acceptance of healthy food or increases rejection of unhealthy food?

This study contributes to academic theories by supporting existing literature which verifies the influence of a colored traffic light label on the healthiness of choice. Additionally, possible influences of dietary restraints will be investigated and connected to theory. Furthermore, this research offers managerial implications by testing if a promotion or prevention focus, indicated by salience of color, can enhance the healthiness of choice.

(8)

6

2 Theoretical Framework

2.1 Literature Review

With a review of the current literature and theory, this section will offer an overview of concepts and theories regarding the topic to provide a better theoretical understanding. Additionally, it is substantial to validate the proposed research model as well as its hypothesis and research questions, which will be presented in chapter 2.2.

First, the emergence of most important labeling systems, especially Front-of-Pack (FOP) traffic light label as an independent variable, will be elaborated. Next, possible moderators will be examined, which are namely dietary restraints and salience of color, ending with a conceptual model.

2.1.1 Food Labeling Systems

Food labels can be a tool to assist consumers making their diet choices and to address a strategy to lower the intake of unhealthy food and the risk of obesity by empowering them with information (Campos, Doxey & Hammond, 2011). An increase of studies evaluating the understanding and acceptance of food labels can be noticed since 2002 (Grunert & Wills, 2007). According to Grunert & Willis (2007), who created a review on the European research on consumer responses to nutrition information on food labels, a simplified FOP information is preferred by the consumer. Nevertheless, they differ in terms of liking and evaluation between various labels, e.g. Nutrition Facts, Guideline Daily Amount (GDA), Traffic Lights (TL) or health logos/ratings. With a summary of former studies, Campos et al. (2011) concluded that nutrition labels are perceived as highly credible and that there is a positive link between the use of nutrition labels in general and healthier diets, since about two-thirds of shoppers read the nutrition labels before buying a diet product (Campos et al., 2011; Cecchini & Sassi, 2015; Cecchini et al., 2010). Consumers appear to be able to comprehend and apply key information of labeling systems, but with an increasing complexity they become more and more confusing, which will be a central topic in the following. Cowburn & Stockley (2005) point out that interpretational aid may support customers during the comprehension process, e.g. traffic light.

2.1.1.1 Back-of-Pack Nutrition Facts

(9)

7

(Grunert & Wills, 2007). The most prominent label, which is mandatory for any food package in Europe, is on the backside of food packages (see Figure 1). The attempt to provide useful information, by giving an overview of the ingredients, calories and other components to the consumer, has been overwhelming. On the one hand, studies show that providing unaided information about various macro and micro nutrients does not help during the decision process of the average customer, instead, the complexity of back-of-pack nutrition labels are difficult to assess and therefore confuse the customer (Wansink, Sonka & Hasler, 2004).

Figure 1, Back-Of-Pack Nutrition Facts

On the other hand, calories alone are not enough information to make healthy choices for customers, which results in a fast evaluation of nutrition labels nowadays (Jones & Richardson, 2007). Unfortunately Jones & Richardson (2007) don’t show which information is actually needed to make better and healthier choices, but as mentioned earlier, an interpretational aid might improve the comprehension process.

2.1.1.2 Guideline Daily Amount

(10)

8

recommended percentage daily intake for adults (see Figure 2) (Corsetto, Muller & Ruffieux, 2015).

Figure 2, Guideline Daily Amount

But the GDA label has been criticized because the overall references are not based on science, rather on an average adult intake and thus not appropriate to groups with diverse prerequisites (Hawley et al., 2013). The authors Hawley et al. (2013) contributed highly to the understanding of how consumers understand, perceive and use GDA labels. They concluded that the processing of information is difficult for the average user and that a percentage of daily intake has no real impact on the decision making process, since a fair amount of consumers do not understand the labels (see Jones & Richardson, 2007). Especially consumers over 65 years think that percentage GDA information is difficult to comprehend (Hawley et al., 2013). Helfer & Shultz (2014) support this suggestion by exploring the processing time of a GDA label, concluding that it needs the most time of all food labels and does not add value to the nutrition choices.

Summarizing the findings from the presented literature, it can be suggested that percentages as a method of providing dietary information alone should be avoided and connected with a visual aid to process the complex information more easily and intuitively.

2.1.1.3 Traffic Light Labeling System

(11)

9

As Drichoutis et al. (2012, p. 524) states, the TL can be defined as follows: “The “traffic light” label places colors next to each nutrient of a product, similar to traffic lights, which indicate low, medium, and high assessments of the nutrient. Usually foods are labeled with a panel of colored spots relating to the amount of sugar, salt, fat, and saturated fat.” The definition provided indicates that a TL summarizes and extents previously mentioned labeling systems (e.g. BOP Nutrition Facts and GDA) by using color as a reference for high/medium/low level intake (see Figure 3).

Figure 3, Traffic Light Labeling System including GDA

Thus, supporting the process of evaluation by guiding consumers to important nutrients, which possibly leads to a selection of healthier food products (Aschemann-Witzel et al., 2013; Bollinger, Leslie & Sorensen, 2010; Borgmeier & Westenhoefer, 2009; Hawley et al., 2013; Lobstein & Davies, 2009; Trudel, Murray, Kim & Chen, 2015; Trudel & Murray, 2011). For that reason, the authors Drichoutis, Lazaridis & Nayga (2005) summarize: a traffic light label is the preferred option between all available FOP labels.

To support prior research, the following hypothesis will be investigated:

H1: A colored traffic light labeling system (vs. numerical) has a positive effect on the healthiness of consumer choice.

2.1.2 Dietary Restraints

(12)

10 self-discipline.

Consumers rely heavily on external cues during the process of exerting self-regulation, for example food labels, to regulate their eating behavior (Trudel & Murray, 2011). The authors show that dieters who have the goal to limit the consumption of food tend to spend more time on cost-related attributes (e.g. counting fat and calories) than on pleasure attributes (e.g. enjoy taste). As a result, dieters are better in controlling their eating behavior than non-dieters. Dieters can be considered as health conscious consumers, which means they are concerned about their wellbeing and willing to improve or maintain their health by undertaking health actions, e.g. gathering information, consume organic food (Becker et al., 1977; Michaelidou & Hassan, 2008). Based on Gould (1988), Naylor et al. (2009, p. 223) define health consciousness as “individual difference variable that assesses the degree to which a person plays an active role in maintaining his or her health.” Thus, health consciousness can be seen as a motivational component. With a focus on cost-related attributes, dieters are more motivated than non-dieters to increase their self-regulation strength (Trudel & Murray, 2013). Especially with available nutritional information dieters focus on the costs of consumption with the result that they have greater willpower to resist the temptation of unhealthy food (e.g. high sugar, high sutured fat, high calories), which would else conflict with dieters’ long-term goals (Trudel et al., 2015). As Trudel et al., (2015) point out, when investigating the processing of information, here of TL-labels, it should be differentiated by the consumers goals to understand the true effects. On the one hand the authors show that color-coded labels provide non-dieters with a visual information processing cue, which has direct influence on the perception of foods’ health quality in terms of a “stop and go” logic behind the colors green and red. On the other hand dieters’ food health perception is not directly affected by the colors green and red, merely the evaluation and information recall in comparison to the traditional BOP nutrition facts.

(13)

11 Thus, the following hypothesis will be investigated:

H2: The influence of a colored traffic light labeling system on healthiness of consumer choice is moderated by dietary restraints.

Further, the information processing will be elaborated with a focus on salience of color to investigate how color-coded traffic light might improve the choice-process.

2.1.3 Information Processing and Salience of Color

When consumers are confronted with complex food choices in terms of selection, they are less able to make conscious choices (Balcombe et al., 2010). The result is that consumers use heuristics to make their decisions. The authors Mhurchu & Gorton (2007) showed that consumers do not understand by nature how to balance the consumption of different nutrition’s, rather base the choice only on the fat content, regardless of any other nutrition levels. Connecting this with the notion of Grunert & Wills (2007) that calories and fat are taken as the central interest of consumers, it is understandable that the consumer lost the overview and can hardly make precise decisions because the focus is wrong. Particularly relevant for the comprehension of food labels is how consumers process the given information, since they have a restricted ability to process information with the outcome that their attention is selective (Trudel & Murray, 2011). As a result, consumers with a goal (e.g. become healthier) are more likely to regulate their behavior. Due to the limited amount of mental resources, there is an increased probability of self-regulatory failure if information is not easily accessible (Polivy et al., 1986; Vohs & Faber, 2007). Color signals and salient sizes of labels might help to support the process to evaluate the information; therefore salience of color and size will be further elaborated.

(14)

12

focus on the achievement of gains and acquire positive outcomes. In contrast, consumers with a prevention focus are not very motivated by achieving gains, but more by preventing losses and negative outcomes, illustrated by the color red (Higgins, 1997).

Stimuli that transmit a positive message can produce approach responses, while stimuli with a negative message produce avoidance responses (Elliot & Convington, 2001). According to Elliot and Maier (2007), this process takes usually place in a very automatic way without consciousness reasoning. The visual attention is very important here, because attention can be reached by salient stimuli which “pop out” from their environment, for example colors, shape or size (Connor, Egeth & Yantis, 2004). This process depends on two distinct types of attention. The Bottom-up mechanism which runs passively on simple sensory input and reacts automatically to salient visual cues of latent importance (Connor et al., 2004; Theeuwes, 2010). And the counterpart, top-down mechanism, which operates actively on voluntary long-term strategies through feedback processing, controlling the bottom-up mechanism on expectancy and goals (Connor et al., 2004; Theeuwes, 2010). This means for food labels, and especially for the colored traffic light label, that both mechanism need to be activated and engaged simultaneously. The first step is to catch the customers’ visual attention through the use of color (red, yellow, green) and size (enlarged). Second, to bring the transferred information in line with his long-term strategies: For example reducing bodyweight.

Thus, the following hypothesis will be investigated:

(15)

13

2.2 Conceptual Model and Hypothesis

Figure 4 provides a visual representation of relations and hypothesis, which will be explored in this master thesis.

Figure 4, Conceptual Framework

The main purpose of this research is 1) to support previous research results by confirming the positive impact of a colored traffic light labeling system on healthiness of consumer choice, 2) to investigate if consumers’ choice is moderated by dietary restraints and 3) to investigate if salience of color moderates the impact of a colored traffic light label on healthiness of choice. Therefore, the following research questions (RQ) have been formulated:

RQ1: Does the implementation of a traffic light (i.e. color-coding, text, percentage of GDA) as a front-of-package (FOP) nutrition label influence the healthiness of consumer choice? RQ2: Is the impact of a colored TL on Consumer Choice predominantly influenced by dietary restraints?

(16)

14

3 Research design

3.1 Method: Choice-Based Conjoint Analysis

(17)

15

3.2 Attributes and Levels

In this research, five different attributes will be the examined, namely: Calories, fat, saturated fat, sugar and salt. Each attribute consists of three levels, since all of them refer to a low, medium and high level of nutrient. The condition established by the number-of-levels effect is fulfilled, which says that every attribute should have more or less the same number of levels. Otherwise attributes with more levels will achieve much higher average importance than other attributes and therefore bias the outcome (Wittink, Huber, Zandan & Johnson, 1992).

Attribute Definition Level

Calories Calories are units of food energy. 175cal; 225cal; 275cal

Fat Fat is one of three macro-nutrition. 3g; 13g; 23g

Saturated Fat

Saturated fat is an altered form in which the fatty acids all have single bonds.

1g; 5g; 9g

Sugar Sugar is short-chain, soluble carbohydrates. 5g; 16g; 27g

Salt Salt is a mineral composed primarily of sodium chloride (NaCl).

0.3g; 1.4g; 2.3g

Table 1, Overview Attributes and Levels

The chosen attributes and its levels (see Table 1) base on guidelines of the UK government (2013), who define the criteria for 100g of food with low, medium and high content level of each nutrient. High and low levels are chosen slightly above or below the limit and medium levels are equally spaced between these boundaries.

3.3 Questionnaire

This study consists out of a CBC-part and a questionnaire which than focuses on possible moderators and necessary information to gain insights concerning the mentioned research questions on basis of the conceptual model. The choice sets as well as the questionnaire were created with support of the program “Sawtooth” and uploaded via “Qualtrics”.

(18)

16

To receive qualitative responses, which means the results are high in validity and reliability, the survey combines random tasks and fixed tasks. First of all, 12 random tasks collect data of different combinations of attributes, followed by two manipulated fixed tasks, which serve as the dependent variable in this thesis to answer the research questions and to verify the hypothesis. For the purpose of this research, the 12 random tasks help the participant to become familiar with the choice decision process and with the condition they are assigned to. Additionally, the hope is that the random tasks resemble the responses of the fixed tasks to confirm the results. If they do not confirm the results, there might be something wrong with the data collection, data coding or the respondents answered very inconsistently (Orme, 2015). This might be important for future research of the data, when the calculation includes the estimated part-worth.

Each choice set consists out of one healthy, neutral and unhealthy choice-option, which is defined by the dominance of high, medium or low nutrient levels. The special feature here is that depending on the displayed condition, the participants are expected to shift their choices, because very healthy or very unhealthy nutrients become more visible. The goal is that participants do not make their choices only on basis of calories or fat.

(19)

17

3.4 Choice Design

Participants are randomly assigned to one of six conditions (see Table 2), which results in a study design of 2x3. This ensures that differences in background or personality are spread equally across all conditions in order to minimize the intervention of extraneous variables which might bias the results (Aronson, Wilson & Brewer, 1998). The attribute “calories” is stable throughout all conditions without manipulating the color (non-color) or size (promotion, prevention focus), because the amount of calories depends on other attributes (fat, carbohydrates and protein) and is not under investigation. All other attributes are part of the manipulations.

Condition one contains a regular size of one centimeter expansion and a manipulation of color (no color), depending on the nutrient level. The second condition “promotion” contains a manipulation of size and color (no color), when the nutrient level is low and marked as “green”. The expansion is manipulated through a variation of size of from one centimeter to three centimeters. All other attributes are stable during this condition. The third condition “prevention” manipulates size and color (no color), when nutrient level is high and marked as “red”, without manipulating any other attribute. The expansion is manipulated through a variation of size from one centimeter to three centimeters.

No Color Color 1. Condition: Regular Size 2. Condition: Promotion 3. Condition: Prevention

Table 2, Overview conditions with the example attribute “fat”

(20)

18

attributes with three levels, resulting in 3x3x3x3x3 = 243 possible combinations of stimuli. Each of the three products, here sandwiches, is presented on a profile card, which can be selected by the participant, depending on his/her best choice. Additionally, in a discrete choice, participants are given the opportunity to select the alternative “no choice” option to make the decision more realistic.

The reason, on the one hand, why the full profile method is used is that it is close to reality and imitates a real marketplace decision, because the fictive object has to be evaluated in one piece. On the other hand, the decision is very complex and depends only on a selection of attributes in a given order, which means that in reality there could be other possible attributes which might influence the decision process. Moreover, the order of attributes might bias the results as well. This has to be kept in mind during the evaluation of the data.

Figure 5, Example Screenshot of CBC Question (Condition: Color x Regular Size)

3.5 Preparing for Analysis

Moderator Dietary Restraint

(21)

19

the median of two are assigned to the condition “Weak Restraint”. Participants with a stronger restraint behavior above the median of two are assigned to condition “Strong Restraint”.

Construct Cronbach’s Alpha

Dietary Restraint 0.75

Table 3, Reliability Analysis Dietary Restraint

Dependent Variable: Healthiness of Choice; Choice Sets

The two relevant fixed choice sets, which function as a dependent variable, were separately recoded, by defining the healthiness of each profile on basis of their nutrient level dominance. This means that the higher the dominance of unhealthy nutrient levels, the unhealthier is the choice in total. Each choice set consists out of one “unhealthy”, “neutral” and “healthy” choice, which is the result of an individual manipulation. The recoded variable has the values -1 (= unhealthy), 0 (= neutral) and 1 (= healthy), which means that a score below zero is unhealthy and a score above zero is healthy.

Since this research looks for one overall measure for the choice of healthiness, these two separate recoded dependent variables were combined into one variable with the values -2 (= very unhealthy), -1 (= unhealthy), 0 = (neutral), 1 (= healthy) and 2 (= very healthy). Hence, the same condition holds true: a score below zero is unhealthy and a score above zero is healthy. Participants who decided to choose option “no choice” are considered as if they selected the healthy choice option, because it can be suggested that none of the possible options is healthy enough to make a choice, but would respectively relate to the intention to buy healthy food.

Independent Variables: Conditions

(22)

20

4 Results

4.1 Sample

Overall and after cleaning the data, the sample consists of a total of 601 respondents who participated in the “Qualtrics” online survey, which was conducted in the United States by a research agency. Through a link, which distributes all participants equally over all six conditions, the collection process took one day and resulted in approximately 100 participants per condition. Because of this, a normal distribution can be expected and is supported by a significant Kolmogorov-Smirnov-Test (see Appendix B). Cleaning the data means that all participants who did not finish the survey were deleted. Also participants who selected “No Option” for both dependent choice sets were excluded because they are not relevant to answer the research questions.

Women and men are almost equally distributed through the whole sample, with 48.4% females and 51.6% males. The age is not as equal distributed as the gender, where 66.7% of the participants are between 21 and 40 years old (see Appendix B). Table 4 represents an overview of the descriptive statistics concerning the respondents.

Age Frequency Percentage Gender Percentage

18 to 20 6 1.0 Male 48.4 21 to 30 199 33.1 Female 51.6 31 to 40 202 33.6 41 to 50 91 15.1 51 to 60 76 12.6 61 to 70 25 4,2 71 and older 2 0.3 Total 601 100 100

Table 4, Descriptive Statistics Age and Gender

(23)

21

high. For this reason the BMI is object of further investigation as a possible covariate. Possibly, a difference in gender and age might influence future results as well, for which reason gender and age are used as covariates later on as well.

Figure 6, Descriptive Statistics BMI-Class of Dieting Participants

Previously to the analysis, a brief investigation of the descriptive statistics, at this point the means of the dependent variable, helps to get an overview over the data and its possible results. As Table 5 shows, the average healthiness of choice increases with the introduction of color over all different size conditions. This could indicate that there is a significant positive influence of color on the dependent variable. Furthermore, the mean of color in combination with a promotion focus has the highest mean in total, which could possibly mean that this condition has the strongest effect of all. To investigate and create evidence for the significance of the hypothesis, an ANOVA will be further conducted.

Condition N Mean

No Color x Normal Size 103 0.09

Color x Normal Size 101 0.50

No Color x Promotion 95 0.26

Color x Promotion 102 0.72

No Color x Prevention 99 0.14

Color x Prevention 101 0.45

Table 5, Means of Dependent Variable sorted by Condition (see Appendix B) 2%

25%

41% 32%

(24)

22

4.2 Testing the Hypothesis

With the goal to accept or reject the hypothesis (see Chapter 2), a multifactorial ANOVA is performed. The ANOVA uses the differences in means between groups, here the six conditions, to draw conclusions of their influence.

The created independent variables, which consist out of the color and size conditions, as well as dietary restrain (see Chapter 3.5) act as fixed factors with possible influence on the dependent variable. Table 6 summarizes the Between-Subjects Effects results.

Source Type III Sum of Squares Df Mean Square F Sig. Corrected Model 38.53a 11 3.503 4.013 .00*** Intercept 71.20 1 71.20 81.57 .00*** IV1_Color_NoColor_All 25.07 1 25.07 28.73 .00*** IV2_Prevention_RegularSize_Promo tion 5.33 2 2.67 3.05 .05** Restrain_Median_Split 3.68 1 3.68 4.22 .04** IV1_Color_NoColor_All* IV2_Prevention_RegularSize_Promo tion .53 2 .27 .30 .74 IV1_Color_NoColor_All* Restrain_Median_Split 2.54 1 2.54 2.91 .09† IV2_Prevention_RegularSize_Promo tion* Restrain_Median_Split .51 2 .26 .29 .75 IV1_Color_NoColor_All* IV2_Prevention_RegularSize_Promo tion * Restrain_Median_Split 2.65 2 1.32 1.52 .22 Error 514.12 589 .87 Total 631.00 601 Corrected Total 552.65 600

a. R-Squared = .07 (Adjusted R-Squared = .05)1

Table 6. Between-Subjects Effects ANOVA

1_{†p < .10}

(25)

23

The overall model of the performed ANOVA and the intercept are significant (Corrected Model Sign. = .00; Intercept Sign = .00). Adjusted R-Squared (=.05) shows that the model explains 5% of the variance of “healthiness of choice”. Further, the variables concerning each hypothesis will be elaborated.

Effect of Traffic Light Label on Healthiness of Choice

The results of Table 6 show that there is a main effect of the independent variable “Color” [F(1; 514.12) = 28.73, p = .00] on healthiness of choice. This effect is positive according to the baseline. Descriptive statistics show that the mean of condition “No Color x Regular Size” vs. “Color x Regular Size” increases with the introduction of color (see Appendix B). A higher score indicates a healthier choice due to the coding of the variable.

By creating a figure out of the total means the positive effect of color with a regular size becomes visible (see Figure 7). This means, that a colored traffic light leads to an increase of healthiness of choice (vs. numerical). Therefore, the results confirm H1: A colored traffic

light labeling system (vs. numerical) has a positive effect on the healthiness of consumer choice.

(26)

24

Moderating Effect of Salience of Color

The results of Table 6 show that there is a main effect of the moderator “Salience of Color” [F(2; 514.121) = 3.05, p = .05] on healthiness of choice. This effect is positive according to the baseline. Descriptive statistics show that the mean of condition “No Color x Regular Size” vs. “No Color x Promotion”, “Color x Promotion”, “No Color x Promotion” and “Color x Promotion” increases with the introduction of salience of color (see Appendix B) and is reinforced by the use of regular colors (see Figure 8).

Figure 8, Estimated Marginal Means (Salience) of Healthiness of Choice

With the implementation of Figure 9, the positive effect of color in combination with salience becomes clearer. Even though, the interaction between color and salience is not significant for the performed ANOVA (Sign. = .74), it could be argued that a promotion focus (low nutrient level, large green) seems to be more effective than a prevention focus (high nutrient level, large red). In order to verify whether or not the healthiness of choice differs between a promotion and prevention focus, an independent t-test with healthiness of choice and salience is performed additionally to the ANOVA. To identify possible dissimilarities between the color and non-color condition, the results are split.

The independent t-test for a colored traffic light label was significant, t(197.13) = -2.20, p = 0.03, and shows that the promotion focus (M = .72; SD = .94) and prevention focus (M = .45; SD = .81) differ from the average healthiness of choice (see Appendix B). Further, the average mean indicates that a promotion focus is more effective than a prevention focus. The

0,14 0,09 0,26 0,45 0,5 0,72 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8

Prevention Regular Promotion

(27)

25

independent t-test for a non-colored traffic light label was not significant, t(192) = -.93, p = .36, and shows that the promotion focus (M = .26; SD = 1.03) and prevention focus (M = .14; SD = .80) does not differ from the average healthiness of choice (see Appendix B).

Figure 9, Estimated Marginal Means (Color) of Healthiness of Choice

Exploring Figure 9, it can be seen that the average mean of regular size with color (M = .50) is higher than a prevention focus (M = .45). Hence, a manipulation of salience in combination with color is only effective for a promotion focus and not for prevention focus, which might be caused due to another moderating effect by strong or weak dietary restraints, which will be elaborated in the following. Therefore, the results confirm H3: Salience of color moderates

the impact of a colored traffic light label on healthiness of choice.

Moderating Effect of Dietary Restraint

The results of Table 6 show that there is a main effect of the moderator “Dietary Restraint” [F(1; 514.12) = 4.22, p = .04] on healthiness of choice.

The role of the moderator “Dietary Restraint” is not quite clear from the beginning, because the main effect is significant which means that dietary restraints influence the healthiness of choice, but the interaction terms closely fail to support its influence on the independent variables and is therefore only marginally significant (Sign. = .09).

With the aim to verify whether or not the healthiness of choice differs between strong or weak

(28)

26

dietary restraint participants, an independent t-test with healthiness of choice and dietary restraint is performed with split results for color and non-color. The independent t-test for a colored traffic light label was not significant, t(301.06) = .29, p = .79, and shows that strong dietary restraints (M = .54; SD = .92) and weak dietary restraints (M = .57; SD = .94) do not differ from the average healthiness of choice (see Appendix B). The independent t-test for a non-colored traffic light label is significant, t(266.31) = 2.64, p = .01, and shows that strong dietary restraints (M = -.01; SD = .95) and weak dietary restraints (M = .28; SD = .94) differ in the average healthiness of choice (see Appendix B). Further, the average mean indicates that weak dietary restraint participants are stronger effected by a non-colored traffic light label than strong dietary restraint participants.

The recent results explain why the interaction term closely fail to support the influence, which is due to the significant influence of the non-color condition and the not significant influence of the color condition. Since further interpretation focuses on the influence on a colored traffic light label, the results reject H2: The influence of a colored traffic light labeling system on

healthiness of consumer choice is moderated by dietary restraints.

Covariates

As indicated previously, it can be suggested that possible covariates bias the recent results. Therefore, an ANCOVA is conducted with gender, BMI and age as covariates. This analysis could not support these suggestions due to insignificant main effects of the covariates (see Appendix B).

In summary, the results suggest that a colored traffic light label can increase the healthiness of choice and is moderated by salience of color (H1 & H3). Moreover, there is no moderating effect of dietary restraints on the influence of a colored traffic light label (H2).

(29)

27

Hypothesis Result Comment

H1 A colored traffic light labeling system

(vs. numerical) has a positive effect on the healthiness of consumer choice.

H2 The influence of a colored traffic light

labeling system on healthiness of consumer choice is moderated by dietary restraints.

Significant main effect of “Dietary Restraint” on healthiness of choice but not significant interaction term. The independent t-test revealed that the color condition has no sign. differences in means. H3 Salience of color moderates the impact

of a colored traffic light label on healthiness of choice.

(30)

28

5 Conclusion and discussion

The main goal of this research is to support recent literature and its results by confirming the positive effect of a colored traffic light label on the healthiness of choice, using choice based conjoint data, which is very similar to actual market place choices (Rao, 2014). Additionally, possible moderators which might influence the effect were used to explore if the impact of a traffic light label is pre-dominantly influenced by strong or weak dietary restraints and if salience of color could enhance the effect of a traffic light label by using a promotion or prevention focus.

Based on an performed ANOVA, which is used to analyze differences among group means, it has been investigated if there are differences concerning the influence of the traffic light labeling system, under certain conditions, on healthiness of choice.

The main findings of this research are:

 There are large differences in means between the influence of a non-colored and a colored traffic light labeling system on healthiness of choice.

 Furthermore, the inspection of means revealed that the influence of a traffic light labeling system on healthiness of choice is positive and increases its effect with the introduction of color, as it was expected.

 The results indicate that consumers’ choice is more than five times healthier with the support of a colored traffic light label than without color, due to a higher average score.

 The analysis of the possible moderators, dietary restraints and salience of color, showed that there are large differences in means between their moderating impacts on the influence of a traffic light labeling system on the healthiness of choice.

 But an independent t-test revealed that there is no moderating effect of dietary restraints for the use of a colored traffic light label. Therefore it can be concluded that the impact of a colored traffic light label is not pre-dominantly influenced by dietary restraints under the conditions of this research.

(31)

29

 The results of the covariance, to test for a possible influence of BMI, age or gender did not lead to any significant results.

These results support the findings and theories of recent publications of researchers who gained insights in the possible influence of a colored traffic light labeling system on the healthiness of choice, for example Aschemann-Witzel et al. (2013), Hawley et al. (2013) and Trudel & Murray (2011). Backed up with these results, it can be suggested that color supports the evaluation process of food labels and guides consumers to important nutrients, which next can increase the healthiness of choice. In line with the regulatory focus theory, individuals with a promotion focus regulate their behavior towards positive outcomes (Aaker & Lee, 2001). The results indicate that consumers with a promotion focus tend to make healthier choices than consumers with a prevention focus, which could be explained with the association of a “stop” and “go” logic behind the colors green and red. Additionally, this could mean that the majority of participants focus on achieving gains instead of preventing losses. The consumers’ goals could possibly be to eat healthier instead of eat food which does not contain high fat or high calories.

Thus, the results support the idea, in order to guide consumers to healthier choices, that the use of a promotion focus through salience of color can be more efficient than the use of a prevention focus. Furthermore, the responses to the colored traffic light labeling system were not moderated by dietary restraints, as suggested by the authors Trudel et al. (2015), and can be interpreted equally across these different groups. Since only 23% of the participants are dieting during the participation of the survey, a higher percentage of dieters might lead to a different result.

(32)

30

5.1 Managerial implications and recommendations

Beside the contribution to current research literature, this thesis includes some managerial implications which could be used by the World Health Organization, in cooperation with the government to, lead consumers to healthier choices and there reduce overweight, obesity and other diseases which might be associated with unhealthy eating behavior.

Firstly, the introduction of a colored traffic light labeling system does have significant impact on healthiness of choice. The results show that color supports the evaluation process of food labels and guides consumers to the important nutrition facts, which might have been possibly not recognized due to high complexity. Especially old people, who tend to have problems to process and understand nutrition information could profit from this aid. Respectively, a recommendation would be the use of colored traffic light labels additionally to BOP nutrition facts to guide customers along the important nutrients.

Secondly, personal dietary restraints do not influence the positive effect of a colored traffic light label on healthiness of choice. Therefore, it is not necessary to differentiate between these groups.

(33)

31

5.2 Limitations and Directions for Future Research

Although this research paper gives some important insights to the topic of colored traffic light labeling systems, there are limitations which need to be mentioned.

Concerning its generalizability, this study is limited to only one product: Sandwiches. Therefore, these results should not be used to draw any conclusions to other food products or product categories, but should be part of future research. For instance, it might be interesting to compare results of products between or within different product categories. In addition, it can be suggested that traffic light labels interact with familiar brands differently because they have a certain image, which might also be interesting for future research.

Furthermore, this study was conducted via an online survey on Qualtrics. With support of choice sets the marketplace situation is imitated as realistically as possible but due to the online survey situation, the participant is still in confronted with an artificial situation, thus, future research could include a lab experiment or a field study concerning the topic of interest. Since the data collection took place in the U.S., the results should not be used to draw conclusions for other countries, because the eating and buying behavior might be different compared to other countries or parts of the world, for instance Europe. Therefore, it might be interesting to compare the results between different countries.

The descriptive statistics showed that actually 52.2% of the participants suffer from overweight or obesity. This circumstance might also bias the results and should be object of further investigation.

Different covariates which could possibly influence the results have been investigated in this research. But there might be other influencing variables which have not been considered. For instance, depending on nutrition knowledge, the influence of a traffic light label could differ in its strength due to a different evaluation process. Also the age structure could have an impact on the results, as it has been indicated previously.

In this research, the choice sets were manipulated by using an unhealthy, neutral and healthy option with the goal to see differences in the participants’ choice with the introduction of color and salience. Since selection of nutrient level for each nutrient was an individual decision without statistical background, different nutrient levels within each option are possible and might lead to new results.

(34)

32

(35)

33

6 References

Aaker, J. L., & Lee, A. Y. (2001). “I” seek pleasures and “we” avoid pains: The role of self-regulatory goals in information processing and persuaion. Journal. Of. Consumer.

Research., 28(1).

Aronson, E., Wilson, T. D., & Brewer, M. B. (1998). Experimentation in social psychology. In The Handbook of social psychology (Vol. 1, pp. 99–142).

Aschemann-Witzel, J., Grunert, K. G., van Trijp, H., Bialkova, S., Raats, M. M., Hodgkins, C., … Koenigstorfer, J. (2013). Effects of nutrition label format and product assortment on the healthfulness of food choice. Appetite, 71(February 2016), 63–74.

Asioli, D., Naes, T., Ovrum, A., & Almli, V. L. (2016). Comparison of rating-based and choice-based conjoint analysis models. A case study based on preferences for iced coffee in Norway. Food Quality and Preference, 48, 174–184.

Balcombe, K., Fraser, I., & Falco, S. Di. (2010). Traffic lights and food choice: A choice experiment examining the relationship between nutritional food labels and price. Food

Policy, 35(3), 211–220.

Becker, M. H., Maiman, L. A., Kirscht, J. P., Haefner, D. P., & Drachman, R. H. (1977). The Health Belief Model and Prediction of Dietary Compliance : A Field Experiment.

Journal of Health and Social Behavior, 18(4), 348–366.

Bollinger, B., Leslie, P., & Sorensen, A. (2010). Book Calorie Posting in Chain Restaurants. Borgmeier, I., & Westenhoefer, J. (2009). Impact of different food label formats on

healthiness evaluation and food choice of consumers: a randomized-controlled study.

BMC Public Health, 9, 184.

Burton, S., & Kees, J. (2011). Flies in the Ointment? Addressing Potential Impediments to Population-Based Health Benefits of Restaurant Menu Labeling Initiatives. Journal of

Public Policy & Marketing, 31(2), 232–239.

Campos, S., Doxey, J., & Hammond, D. (2011). Nutrition labels on pre-packaged foods: a systematic review. Public Health Nutrition, 14(8), 1496–1506.

(36)

34

Cecchini, M., Sassi, F., Lauer, J. A., Lee, Y. Y., Guajardo-Barron, V., & Chisholm, D. (2010). Tackling of unhealthy diets, physical inactivity, and obesity: Health effects and cost-effectiveness. The Lancet, 376(9754), 1775–1784.

Connor, C. E., Egeth, H. E., & Yantis, S. (2004). Visual attention: Bottom-up versus top-down. Current Biology, 14(19), 850–852.

Corsetto, P., Muller, L., & Ruffieux, B. (2015). Helping Consumers With a Front-of-Pack Label: Numbers or Colours? Experimental. International Journal of Advance

Technology & Engineering Research (IJATER), 1(1), 30–35.

Cowburn, G., & Stockley, L. (2005). Consumer understanding and use of nutrition labeling: a systematic review. Public Health Nutr, 8(1), 21–28.

Department of Health, U. (2013). Guide to creating a front of pack (FoP) nutrition label for pre-packed products sold through retail outlets.

Diener, C. (2011). Using choice modeling to supercharge your business : Know what your

customer wants, know what you can do about it : the essential non-technical guide to choice modeling, its benefits and applications. Ithaca, NY: Paramount Market

Publishing.

Drichoutis, A. C., Lazaridis, P., & Nayga, R. M. (2005). Nutrition knowledge and consumer use of nutritional food labels. European Review of Agricultural Economics, 32(1), 93– 118.

Drichoutis, A. C., Nayga, R. M., & Lazaridis, P. (2012). Nutritional Labeling. The Oxford

Handbook of the Economics of Food Consumption and Policy.

Eggers, F., & Sattler, H. (2011). CBC H I T - CBC HILCA Preference Measurement with Conjoint Analysis OVERVIEW OF STATE - OF - THE - ART APPROACHES AND RECENT DEVELOPMENTS, 3(1), 36–47.

Elliot, A. J., & Maier, M. A. (2007). Color and psychological functioning. Current Directions

in Psychological Science, 16(5), 250–254.

(37)

35

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(3), 430–439.

Helfer, P., & Shultz, T. R. (2014). The effects of nutrition labeling on consumer food choice: A psychological experiment and computational model. Annals of the New York Academy

of Sciences, 1331(1), 174–185.

Higgins. (1997). Beyond pleasure and pain. American Psychologist, 52(12), 1280–300. Hoch, S. J., & Loewenstein, G. F. (1991). and Consumer Self-Control, l(March), 492–508. Jones, G., & Richardson, M. (2007). An objective examination of consumer perception of

nutrition information based on healthiness ratings and eye movements. Public Health

Nutrition, 10(3), 238–244.

Kleef, E. Van, & 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.

Kozup, J. C., Creyer, E. H., & Burton, S. (2003). Making Healthful Food Choices: The Influence of Health Claims and Nutrition Information on Consumers’ Evaluations of Packaged Food Products and Restaurant Menu Items. Journal of Marketing, 67(2), 19– 34.

Lobstein, T., & Davies, S. (2009). Defining and labelling “healthy” and “unhealthy” food.

Public Health Nutrition, 12(3), 331–340.

Louviere, J. J., Hensher, D. A., & Swait, J. D. (2000). Stated Choice Methods: Analysis and Application. Journal of Applied Econometrics, 420.

Malhotra, N. L. (2010). Marketing Research (6th Editio). New Jersey: Prentice Hill.

Meyers-Levy, J., & Peracchio, L. A. (1995). Understanding the Effects of Color : How the Correspondence between Available and Required Resources Affects Attitudes. Journal

of Consumer Research, 22(2), 121–138.

(38)

36

International Journal of Consumer Studies, 32(2), 163–170.

Naylor, R. W., Droms, C. M., & Haws, K. L. (2009). Eating with a Purpose: Consumer Response to Functional Food Health Claims in Conflicting Versus Complementary Information Environments. Journal of Public Policy & Marketing, 28(2), 221–233. Ni Mhurchu, C., & Gorton, D. (2007). Nutrition labels and claims in New Zealand and

Australia: A review of use and understanding. Australian and New Zealand Journal of

Public Health, 31(2), 105–112.

Orme, B. (2015). Including Holdout Choice Tasks in Conjoint Studies. Sawtooth Software

Research Paper Series, 1–7.

Polivy, J., Herman, C. P., Hackett, R., & Kuleshnyk, I. (1986). The effects of self-attention and public attention on eating in restrained and unrestrained subjects. Journal of

Personality and Social Psychology, 50(6), 1253–60.

Rao, V. R. (2014). Applied conjoint analysis. Applied Conjoint Analysis (Vol. 9783540877). Theeuwes, J. (2010). Top-down and bottom-up control of visual selection. Acta Psychologica,

135(2), 77–99.

Trudel, R., & Murray, K. B. (2011). Why Didn’t I Think of That? Self-Regulation Through Selective Information Processing. Journal of Marketing Research, 48(August), 701–712. Trudel, R., & Murray, K. B. (2013). Self-regulatory strength amplification through selective

information processing. Journal of Consumer Psychology, 23(1), 61–73.

Trudel, R., Murray, K. B., Kim, S., & Chen, S. (2015). The Impact of Traffic Light Color-Coding on Food Health Perceptions and Choice, 21(3), 255–275.

Vohs, K. D., & Faber, R. J. (2007). Spent resources: Self‐regulatory resource availability affects impulse buying. Journal of Consumer Research, 33(4), 537–547.

Wansink, B., Sonka, S. T., & Hasler, C. M. (2004). Front-label health claims: When less is more. Food Policy, 29(6), 659–667.

Wittink, D., Huber, J., Zandan, P., & Johnson, R. M. (1992). The Number of Levels Effect in Conjoint: Where Does It Come From and Can It Be Eliminated? Sawtooth Software

(39)

I

7 Appendices

7.1 Appendix A: Qualtrics Questionnaire

Introduction:

Purchasing a Sandwich

You are about to make 14 purchase decisions. Specifically, we ask you to make 14 sandwich purchases. You are about to be presented with 14 choice sets, each consisting of three sandwiches that you may consider buying. The sandwiches you are going to look at are described based on five characteristics:

Calories 175 cal 225 cal 275 cal

Fat 3 g 13 g 23 g

Saturated Fat 1 g 5 g 9 g

Sugar 5 g 16 g 27 g

Salt 0.3 g 1.4 g 2.3 g

For each choice set, we ask you to select the sandwich that you would purchase.

(40)

II

Dependent Choice sets

(41)

III

(42)

IV

(43)

V

(44)

VI

(45)

VII

(46)

VIII

(47)

IX

Basic information and demographics

(48)

X

7.2 Appendix B: SPSS Output

Kolmogorov-Smirnov-Test:

One-Sample Kolmogorov-Smirnov Test

CBCDV_Comp ute_new

N 601

Normal Parametersa,b Mean ,36 Std. Deviation ,960 Most Extreme Differences

Absolute ,233 Positive ,161 Negative -,233 Kolmogorov-Smirnov Z 5,714 Asymp. Sig. (2-tailed) ,000 a. Test distribution is Normal.

b. Calculated from data.

Age:

Age

Frequency Percent Valid Percent Cumulative Percent Valid 18 to 20 6 1,0 1,0 1,0 21 to 30 199 33,1 33,1 34,1 31 to 40 202 33,6 33,6 67,7 41 to 50 91 15,1 15,1 82,9 51 to 60 76 12,6 12,6 95,5 61 to 70 25 4,2 4,2 99,7 71 and older 2 ,3 ,3 100,0 Total 601 100,0 100,0 Gender: Gender

Frequency Percent Valid Percent Cumulative Percent Valid

(49)

XI

BMI:

BMI_Classification

Frequency Percent Valid Percent Cumulative Percent Valid Underweight 16 2,7 2,7 2,7 Normal 271 45,1 45,2 47,8 Overweight 191 31,8 31,8 79,7 Obesity 122 20,3 20,3 100,0 Total 600 99,8 100,0 Missing System 1 ,2 Total 601 100,0 Dieting:

Are you currently dieting

Frequency Percent Valid Percent Cumulative Percent Valid

Yes 138 23,0 23,0 23,0 No 463 77,0 77,0 100,0 Total 601 100,0 100,0

Frequencies, Dependent Variable:

(50)

XII Mean ,45 ColorNoSize N Valid 101 Missing 0 Mean ,50 CBCDV_Compute_new

(51)

XIII

ANOVA: Effect of salience, color and dietary restraints on healthiness of choice Descriptive Statistics

Dependent Variable: CBCDV_Compute_new IV1_Color_

NoColor_All

IV2_LargeGreen_Ne utral_LargeRed

Restrain_Median_Split Mean Std. Deviation N

(52)

XIV Between-Subjects Factors Value Label N IV1_Color_NoColor_All 0 No Color 297 1 Color 304 IV2_LargeGreen_Neutral_L argeRed -1 Prevention 200 0 No Size 204 1 Promotion 197 Restrain_Median_Split ,00 Weak Restraint 330 1,00 Strong Restraint 271

Tests of Between-Subjects Effects

Dependent Variable: CBCDV_Compute_new Source Type III Sum of

Squares

(53)

XV

Independent t-test: Healthiness of choice, grouped by salience, split by color

Group Statistics IV1_Color_NoColor_All IV2_LargeGreen_Neutral_La rgeRed N Mean No Color CBCDV_Compute_new Prevention 99 ,14 Promotion 95 ,26 Color CBCDV_Compute_new Prevention 101 ,45 Promotion 102 ,72 Group Statistics IV1_Color_NoColor_All IV2_LargeGreen_Neutral_Lar geRed

Std. Deviation Std. Error Mean

No Color CBCDV_Compute_new Prevention ,796 ,080 Promotion 1,023 ,105 Color CBCDV_Compute_new Prevention ,806 ,080 Promotion ,937 ,093

Independent Samples Test

IV1_Color_NoColor_All Levene's Test for Equality of Variances

F Sig.

No Color CBCDV_Compute_new

Equal variances assumed 16,694 ,000 Equal variances not

(54)

XVI

Color CBCDV_Compute_new

Equal variances assumed 1,717 ,192 Equal variances not

assumed

IV1_Color_NoColor_All t-test for Equality of Means t df

Equal variances assumed -,927 192 Equal variances not assumed -,923 177,416 Color CBCDV_Compute_new

Equal variances assumed -2,201 201 Equal variances not assumed -2,202 197,124

IV1_Color_NoColor_All t-test for Equality of Means Sig. (2-tailed) Mean Difference

Equal variances assumed ,355 -,122 Equal variances not assumed ,357 -,122 Color CBCDV_Compute_new

Equal variances assumed ,029 -,270 Equal variances not assumed ,029 -,270

(55)

XVII Std. Error Difference 95% Confidence Interval of the Difference Lower No Color CBCDV_Compute_new

Equal variances assumed ,131 -,381 Equal variances not assumed ,132 -,382 Color CBCDV_Compute_new

Equal variances assumed ,123 -,512 Equal variances not assumed ,123 -,512

IV1_Color_NoColor_All t-test for Equality of Means 95% Confidence Interval of the Difference Upper No Color CBCDV_Compute_new

Equal variances assumed ,137 Equal variances not assumed ,139 Color CBCDV_Compute_new

Equal variances assumed -,028 Equal variances not assumed -,028

Independent t-test: Healthiness of choice, grouped by dietary restraints, split by color

Group Statistics

IV1_Color_NoColor_All Restrain_Median_Split N Mean Std. Deviation No Color CBCDV_Compute_new

(56)

XVIII

Color CBCDV_Compute_new

Weak Restraint 158 ,57 ,940 Strong Restraint 146 ,54 ,918

Group Statistics

IV1_Color_NoColor_All Restrain_Median_Split Std. Error Mean No Color CBCDV_Compute_new Weak Restraint ,072 Strong Restraint ,085 Color CBCDV_Compute_new Weak Restraint ,075 Strong Restraint ,076

IV1_Color_NoColor_All Levene's Test for Equality of Variances

F Sig.

Equal variances assumed ,137 ,711 Equal variances not

assumed Color CBCDV_Compute_new

Equal variances assumed ,007 ,933 Equal variances not

assumed

(57)

XIX

Equal variances assumed 2,644 295 Equal variances not assumed 2,641 266,308 Color CBCDV_Compute_new

Equal variances assumed ,267 302 Equal variances not assumed ,268 301,063

IV1_Color_NoColor_All t-test for Equality of Means Sig. (2-tailed) Mean Difference

Equal variances assumed ,009 ,293 Equal variances not assumed ,009 ,293 Color CBCDV_Compute_new

Equal variances assumed ,789 ,029 Equal variances not assumed ,789 ,029

IV1_Color_NoColor_All t-test for Equality of Means Std. Error Difference 95% Confidence Interval of the Difference Lower No Color CBCDV_Compute_new

Equal variances assumed ,111 ,075 Equal variances not assumed ,111 ,075 Color CBCDV_Compute_new

(58)

XX

IV1_Color_NoColor_All t-test for Equality of Means 95% Confidence Interval of the Difference Upper No Color CBCDV_Compute_new

Equal variances assumed ,511 Equal variances not assumed ,511 Color CBCDV_Compute_new

Equal variances assumed ,239 Equal variances not assumed ,238

ANCOVA: Effect of salience, color, dietary restraints and covariate gender on healthiness of choice

Squares

(59)

XXI IV1_Color_NoColor_All * IV2_LargeGreen_Neutral_L argeRed * Restrain_Median_Split 2,647 2 1,324 1,519 ,220 Error 512,370 588 ,871 Total 631,000 601 Corrected Total 552,649 600 a. R Squared = ,073 (Adjusted R Squared = ,054)

ANCOVA: Effect of salience, color, dietary restraints and covariate BMI on healthiness of choice

Squares

(60)

XXII

ANCOVA: Effect of salience, color, dietary restraints and covariate age on healthiness of choice

Squares

(61)

XXIII

Descriptive statistics: Realistic Information provision & to make better choices:

The nutrition information was very realistic

Frequency Percent Valid Percent Cumulative Percent Valid 1 5 ,8 ,8 ,8 2 9 1,5 1,5 2,3 3 13 2,2 2,2 4,5 4 34 5,7 5,7 10,1 5 73 12,1 12,1 22,3 6 200 33,3 33,3 55,6 7 267 44,4 44,4 100,0 Total 601 100,0 100,0

The nutrition information allowed me to make better choices

(62)

XXIV

7.3 Appendix C: Scale

Dietary Restraints by Polivy et al. (1986)

How often are you dieting?

 Never; rarely, sometimes, often, always (Scored 0–4)

Weight fluctuation

What is the maximum amount of weight (in pounds) you have ever lost within 1 month?

 (0–4, 5–9, 10–14, 15–19, 20+ lbs (Scored 0–4))

What is your maximum weight gain (in pounds) within a week?

 (0–1, 1.1–2, 2.1–3, 3.1–5, 5.1+ lbs (Scored 0–4)) In a typical week, how much does your weight fluctuate?

 (0–1, 1.1–2, 2.1–3, 3.1–5, 5.1+ lbs (Scored 0–4))

How many pounds over your desired weight were you at your maximum weight?

 (0–1, 2–5, 6–10, 11–20, 21+ lbs (Scored 0–4))

Concern of dieting

Would a weight fluctuation of 5 pounds affect the way you live your life?

 Not at all; slightly, moderately; very much (Scored 0–3) Do you eat sensibly in front of others and splurge alone?

 Never; rarely, often, always (Scored 0–3) Do you give too much time and thought to food?

 Never, rarely, often; always (Scored 0–3). Do you have feelings of guilt after overeating?

 Never, rarely, often, always (Scored 0–3). How conscious are you of what you are eating?