The impact of nutrition labeling and health literacy on the healthiness of food choices in restaurants

(1)

The impact of nutrition labeling and health

literacy on the healthiness of food choices in

restaurants

By

Frank Hofman

s2327287

University of Groningen

Faculty of Economics and Business

MSc Marketing

Marketing Management

January 2019

Supervisor: Martine van der Heide Co-assessor: Jenny van Doorn

(2)

2

Abstract

(3)

1. Introduction

Over the last decades, the spread of non-communicable diseases such as cardiovascular disease, diabetes type-2 and obesity, have increased dramatically (Schmidhuber and Traill, 2009; Waterlander, Steenhuis, de Boer, Schuit and Seidell, 2012; WHO, 2009). Overconsumption and the immoderate ingestion of unhealthy (saturated) fats and sugars play a huge role in this development. Additionally, it is becoming more common for people to eat out at a bar or at a restaurant, especially for millenials who on average consume 2-3 meals a month at a restaurant or bar in comparison to gen X’ers who on average eat out only once a month (Kuhns and Saksena, 2017). Frequently eating out is associated with excess weight gain over time (McCrory, Fuss, Hays, Vinken, Greenberg and Roberts, 1999). Moreover, food that is consumed away from home has lower nutritional quality compared to food that is cooked at home (Finkelstein, French, Variyam and Haines, 2004). The combination of the change in eating out behavior and the healthiness and overconsumption when eating out results in a higher consumption of unhealthy, high calorie meals. This leads to restaurants being seen as a potential target for interventions to improve diet and therefore reduce diet related diseases (Goffe, Penn, Adams, Araujo-Soares, Summerbell, Abraham, White, Adamson and Lake, 2018). Policy makers have to choose from a range of possible intervention methods when trying to improve the diet and healthiness of the population. An important characteristic of these interventions is the covertness of the interventions (Goffe et al., 2018). Covert interventions are not noticeable by consumers, whereas overt interventions are those that are obvious to the consumer, such as menu labeling.

In a new legislation, as a part of the affordable care act issued by former United States president Obama, restaurant and fast food chains are now required to put nutrition labeling on their menus (Affordable care act, 2010). This specific legislation went into effect in 2018. Nutrition labeling in restaurants can influence consumers in two different ways. First, there is a direct effect of the consumer eating less because of the increased awareness on the nutritional values. Second, there is an indirect effect of restaurant industry changing the composition of their menus to make it more appealing (Block and Roberto, 2015).

In a study done by Burton, Creyer, Kees and Huggins (2006), consumers were asked to estimate the nutritional content of typical restaurant meals. These consumers highly underestimated the amount of fats and calories in them. The amount of fat and saturated fat in those meals was two times higher than the consumers’ estimation. This shows a lack of nutritional knowledge. A possible solution for this lack of knowledge in regard to calories consumed is providing nutritional labeling in restaurants.

(5)

population has sufficient knowledge about these labels and knows how to interpret it. These studies do not cover what the most efficient methods of providing nutritional information in restaurants are. To make menu labelling in restaurants more impactful we need to know which methods are the most effective.

While some authors argue that nutrition labeling in restaurants is not an effective intervention strategy or did not have the intended effects (Elbel, Kersh, Brescoll and Dixon, 2009; Sinclair, Cooper and Mansfield, 2014), others reported a significant change in the food choices made when nutrition labeling is introduced (Pulos and Leng, 2010; Burton, Howlett and Tangari, 2009). The general consensus from these studies is that the average consumer has no idea how much calories they consume from restaurant meals and that only health conscious customers are able to benefit from the nutritional information (Burton et al, 2009).

Furthermore, consumers may not use or understand the nutritional information provided. Krukowski, Harvey-Berino, Kolodinsky, Narsana and DeSisto (2006) found that only 64%-73% of community and college students were able to accurately report knowledge on daily caloric needs. This shows that the generation that eats out most often has a low health literacy. In this paper health literacy is defined as an individual's’ capability to obtain and understand health information to make appropriate health-related decisions (Huizinga, Carlisle, Cavanaugh, Davis, Gregory, Schlundt and Rothman, 2009). Furthermore, Krukowski et al. (2006) pitch a possible solution for the problem of low health literacy among restaurant clients. Namely, categorizing food in low, medium and high caloric groups on the menu instead of providing all the nutritional information. They suggest that this would result in a better effectiveness of the intervention. However, this intervention technique has not been studied and is provided merely as a suggestion by these authors. Therefore, I will study two different types of menu labelling to see if there are different effects on low and high health literate consumers.

Concluding, I will study the effect providing nutritional information in restaurants on the healthiness of the food choices to provide current research with more clarity on this topic. Second, this paper will contribute to the current literature by studying the impact of health literacy on the effectiveness of nutrition labeling in restaurants as an intervention method. Thirdly, I will test if different methods of providing nutritional information in restaurants have a different effect on consumer with high health literacy or low health literacy. Therefore, I study the effect of health literacy on this relationship and to test if different nutritional labeling techniques have different effects on the lower health literate group compared to no nutrition labeling. This results in the following research questions:

What is the impact of nutrition labeling in restaurants on the healthiness of the food choice?

(6)

6

restaurants on the healthiness of the food choice?

(7)

2. Literature review

2.1 Nutrition labeling and the healthiness of food choices

Cross-sectional and longitudinal studies have shown that frequently eating out at restaurants is associated with a higher BMI (body mass index) and weight gain (Bezerra, Curioni and Sichieri, 2012; Duffey, Gordon-Larsen, Jacobs, Williams and Popkin, 2007). This is due to the increased portion size, higher amounts of less favorable nutrients (saturated fats, sugars and salt) and lower amounts of favorable nutrients (calcium and vitamins). These three components can result in a positive energy balance and therefore lead to weight gain and obesity.

Consumers unknowingly consume 900 additional calories a week from eating in a restaurant (Burton et al. 2009). They have trouble determining which foods are high in calories and contain negative nutrients (Wansinck and Chandon, 2006). This underestimation of calories consumed can cause significant weight gain in the long run. Furthermore, consumers also lack knowledge on the amount of energy they need to consume daily (Elbel et al., 2009). The average consumer has limited knowledge on the total caloric and nutritional contents of restaurant meals. Making this information available for the consumer could result in a better management of caloric intake. Furthermore, Burton et al. (2009) argue that providing consumers with nutritional information can improve goal activation in health-conscious consumers. This in turn can result in increased goal-directed behaviors.

According to Burton and Creyer (2004), providing nutritional labeling is an effective strategy for reducing calorie consumption when there is a greater discrepancy between the perceived content and the actual content. In fine dining restaurants, providing clients with nutrition labeling on menus resulted in fewer calories consumed (Fotouhinia-Yepes, 2013). The inclusion of nutrition labeling was highly valued by women and older clients. This could result in increased traffic and loyalty to restaurants providing nutrition labeling.

Sinclair et al. (2014) argue that labeling of menus with calories alone has no effect on total calories consumed. This claim is supported by Swartz, Braxton and Viera (2011), they found that calorie labeling does not have the intended effect on food selection. However, there are some side notes to their research. When the impact of sex was tested in this setting, women used menu labeling to select and consume less calories whereas men did not. Furthermore, the effect of menu labeling on food consumption seems to be contextual and require further research to study different settings (VanEpps, Roberto, Park, Economos and Bleich, 2016).

(8)

8 types of effects and impacts on the consumer’s food choice. Therefore, three types of menus will be tested. First, a regular menu without any nutrition labeling (control menu). Second, a menu with specific nutritional information such as calories and macronutrients, and third, a menu with easier to understand nutritional categories (Krukowski et al.,2006) (See appendix 1 for the different menus).

Drichoutis and Lazaridis (2005) describe the mechanisms behind when consumers choose to use nutritional labels. First, the use of nutritional labeling is an act of information search. Therefore, a consumer will use the label as long as the additional costs spend using the label do not outweigh the additional benefits. In this case additional costs being the time spend reading the label and the benefits can be a healthier food choice. Second, if a consumer perceives that their future health is at risk they are more likely to use nutritional labeling. Perceptions of risk motivate a people to increase their information search. The authors conclude with the effectiveness and use of labels is mainly influenced by socio-economic, income and education. They argue that nutritional labeling is only beneficial for a part of the population which falls into different categories.

With previous literature divided on the matter of effectiveness of nutrition labeling in restaurants, the first objective of this paper is to clarify this relationship. This results in the first hypothesis:

H1: Nutrition labeling in restaurants increases the healthiness of the food choice.

2.2 Health literacy and the healthiness of food choices

The concept of health literacy builds upon the relationship between food knowledge and food choices, but more generally the relationship between knowledge and behavior (Vaitkeviciute et al., 2015). Food literacy is not limited to nutrition knowledge, but also includes the ability to obtain, process and understand basic nutritional information as well as the proficiency to use this information to make improved health decisions (Kolasa, Peery, Harris and Shovelin, 2001).

According to social learning theory, knowledge creates a precondition for change. This includes self-regulation of health-related habits (Rosenstock, Stretcher and Becker, 1988). However, knowledge on its own is not sufficient for a change of individual behavior (Clifford, Anderson, Auld and Champ, 2009). Therefore, it is needed to move beyond knowledge to more comprehensive concepts such as literacy to effect change in diet.

(9)

levels of SPFL (Poelman, Dijkstra, Sponselee, Kamphuis, Battjes-Fries, Gillebaart and Seidell, 2018). According to previous research, a low health literacy and numeracy act as an obstacle to consumer understanding and the interpretation of food related information (Rothman, Housam, Weiss, Davis, Gregory, Grebretsadik and Elasy, 2006). The ability to read and interpret numbers and includes basic computation, logic and understand measurements, plays an important role in health literacy and when this is low consumers will lack the basic processes to make the proper interpretations of the nutrition label. Based on the mechanisms described in the studies mentioned above, I argue for a direct positive relationship between health literacy and the healthiness of the food choice in restaurants. The second hypothesis is constructed from these arguments:

H2: Health literacy has a positive influence on the healthiness of the food choice.

2.3 The moderating effect of health literacy

Consumers have trouble interpreting and understanding nutrition labels. Consumers that have a higher health literacy are able to better understand nutrition labels which results in a higher consumption of vegetables and a reduction in the consumption of sugar packed sodas (Persoskie, Hennessy and Nelson, 2017).

In today’s society there is an overload of nutritional information which leaves consumers confused and unsure about which nutritional advice to follow. According to Kickbusch (2008), increasing health literacy among the society will result in a better way to process health and food related information and allows for better decision making.

Consumers with a lower health literacy and numeracy (ability to apply numerical concepts) make worse judgements in regard to food compared to higher literate consumers (Malloy-Weir and Cooper, 2017). Furthermore, consumers who have a higher nutritional knowledge and numeracy were shown to use food labels more often (Hess, Visschers and Siegrist, 2012). Furthermore, lower health literate consumers showed less use of nutrition labels since they might not understand the numbers on it. However, both of these studies do not cover the effect of health literacy on nutrition labeling in restaurants. Merely the effect of health literacy on the effect of nutrition labeling in general.

One of the possible explanations of the lack of effect of nutrition labeling on healthiness of the food choice in some studies done in fast food restaurants is the clientele of these establishments. They are most often frequented by minorities and lower income groups (Block, Scribner and DeSalvo, 2004). Groups with a lower socioeconomic status (SES) tend to make less use of nutrition labeling because of worse health literacy and numeracy compared to groups with a higher SES (Malloy-Weir and Cooper, 2017; Noguira, Thai, Nelson and Oh, 2016; Feng and Fox, 2018).

(10)

10 following:

H3: A higher health literacy positively influences the relationship between nutrition labeling and the healthiness of the food choice.

2.4 Conceptual Model

(11)

3. Methods

The data needed for this study was collected via a survey among 138 participants. 35 were not included in the final sample due to incomplete data and dietary restrictions. The final sample for this study consisted out of 103 adults of which 46 were male and 57 were female. The three different menus were evenly distributed among the participants. The average age was 29 and they ate out on average 2.8 times a month. 24.3% of the participants eats out once a month, 29.1% eats out twice a month and 73.8% eats out 3 or more times a month.

3.1 Study Design

Participants of the survey (n=103) were randomly assigned to one of three menu labeling conditions. (1) a menu without any calorie labels (no calorie labels), (2) a menu with calorie labeling that contains kcal, carbohydrates, proteins and fats (calorie labels), (3) a menu with a three-item categorization of the number of calories by low, medium and high (Roberto, Larsen, Agnew, Baik and Brownell, 2010) (shown in appendix 1).

All menus contain the same 15 items, 5 starters, 5 main courses and 5 desserts. The items on the menu were selected so when together they formed a menu of a typical Dutch restaurant. (starter: tomato soup, carpaccio, vegan bitterballen, bread with spreads and a shrimp cocktail. Main courses: entrecote, grilled salmon, mixed grill, vegetarian risotto and a Caesar salad. Desserts: crème brûlée, dame blanche, cheese platter, pure chocolate mousse and mixed fruits with Greek yoghurt).

For groups 2 and 3, the calorie labels were presented in a column after the menu item and before the price. The calorie values of the meals were gathered from caloriecontrol.org.

3.2 Key measures

3.3.1. Dependent variable - Healthiness of the food choice

This paper relies on total calories consumed during the stay at the restaurant to calculate the healthiness of the food choice. This is in line with previous literature that covers the healthiness of eating out (Roberto et al., 2010). The total calories consumed are calculated by adding up the calories of the starter, main and dessert.

3.3.2. Independent variable - Nutrition labeling

(12)

12 The second menu will be the same as the first menu but with added nutritional information (Kcal, carbohydrates, protein and fats). According to Krukowski et al., (2006), consumer with a lower health literacy may have trouble with such a. in-depth description of the healthiness of the meals. Therefore, I have added a third category to test whether this holds. This menu will be similar to the first menu but with a tiered description of the number of calories in the meal. Low, medium and high. This could result in a better comprehension of the healthiness of the meal by the lower literate consumers.

For the starters, the low label will apply for meals below 200 calories, the medium category for meals between 200 and 300 calories and the high calorie label will be for meals of over 300 calories. For the main course, the low label we be for meals of below 600 calories, the medium label for meals between 600 and 800 calories and the high label will be for meals of 800 calories or more. Finally, for the desserts the low label will apply for desserts that have 200 or less calories, the medium label for desserts that have 200-300 calories and high for desserts that have 300 calories or more. The cut-off points for these labels have been decided with the average number of calories consumed when eating out in mind. (Urban, Weber, Heyman, Schichtl, Verstraete, Lowery and Masters, 2016). On average restaurant goers consume 1205 calories when eating out. When participants in this study choose only low-calorie options, they at most consume 837 calories which is in the range of a healthy dinner. The medium calorie options lead up to about 1200 calories consumed which is in line with the average restaurant goer. Choosing high caloric options results in consuming 1679 calories minimum, which is much higher than the average restaurant goer.

3.3.3. Moderating variable - Food literacy

(13)

Table 1. Health literacy 12-item questionnaire

Item Min-Max

1. When I have questions on healthy nutrition, I know where I can find information on this issue?

Disagree strongly = 1 to Agree strongly = 4; I do not have experience with these issues = 0

2. In general, how well do you understand the following types of nutritional information?

(A) Nutrition information leaflets (B) Food label information

(C) TV or radio program on nutrition

(D) Oral recommendations regarding nutrition from professionals.

(E) Nutrition advice from family members or friends

Very bad = 1 to Very good = 5; I do not make use of this kind of information = 0

3. How familiar are you with the Food Pyramid? Very bad = 1 to Very good = 5

4. I know the official Dutch recommendations about fruit and vegetable consumption.

Disagree strongly = 1 to Agree strongly = 4

5. I know the official Dutch recommendations about salt intake. Disagree strongly = 1 to Agree strongly = 4

6. Think about a usual day: how easy or difficult is it for you to compose a balanced meal at home?

Very hard = 1 to very easy = 4; not applicable = 0

7. In the past, how often were you able to help your family members or a friend if they had questions concerning nutritional issues?

1 = Never to always = 5; there have never been any questions = 0

8. There is a lot of information available on healthy nutrition today. How well do you manage to choose the information relevant to you?

Very bad = 1 to very good = 5; I have not been interested in these issues = 0

9. How easy is it for you to judge if media information on nutritional issues can be trusted?

(14)

14 10. Commercials often relate foods with health. How easy is it for

you to judge if the presented associations are appropriate or not?

Very hard = 1 to very easy = 4

11. How easy is it for you to evaluate if a specific food is relevant for a healthy diet?

12. How easy is it for you to evaluate the longer-term impact of your dietary habits on your health?

3.3.4. Control variables

To solidify the results of this paper several control variables are included. I have included 7 control variables which could influence the number of calories consumed during the dinner. First, to control for the hunger of the participant which can influence which meal they choose, I included hunger as a control variable. Hunger was measured on a 7-point Likert scale ranging from extremely hungry to not hungry at all. Second, the meal price can influence the choice of which dish is consumed therefore price is included as a control variable. Third, according to Swartz, Braxton and Viera (2011), women use nutrition labeling to a greater extent when compared to men. This leads to gender being included as a control variable. Fourth, following the SPFL survey, I ask the educational level of the respondent to test if there is an influence of educational level on food literacy and healthiness of the food choice. Fifth, the body mass index of the respondents will be calculated to study if a respondents BMI has a relationship with health literacy and participants with an unhealthy BMI (>25) are more or less affected by nutrition labeling compared to participant with a healthy BMI (<25). Sixth, the respondents age will be included as a control variable. Seventh, how often participants eat out will be included as the final control variable. How often a participant eats out might influence restaurant behavior and which dishes are chosen.

3.3 Analysis strategy

The independent variable (menu category) is a categorical variable and the dependent variable (total calories) is an interval variable. Since, the data is collected once per participant and the observations are independent, a multiple regression analysis was performed to test the hypothesis. A multiple regression analysis is used to learn more about the relationship between several independent variables (the IV, MOD and control variables) and a dependent variable.

(15)

significant negative relationship between the total calories consumed during dinner and the menu category. It shows that when menu category 1 (no caloric information) was not given less calories were consumed. There is also a strong positive correlation between price and total calories consumed. This correlation can be easily explained. The most expensive meals contain the most calories and the cheapest meals contain the least number of calories. Furthermore, age negatively correlates with the BMI of participants, which is in line with previous studies done on that topic (Meeuwsen, Horgan and Elia, 2010). Lastly, there appears to be no statistically significant relationship between the SPFL and the total calories consumed during the dinner.

Before testing the three hypotheses, I tested for multicollinearity. Table 3 shows the results for this test. All the variables were included in this test except for the difference between actual calories consumed and perceived calories consumed since it is derived from total calories consumed and the perceived calories consumed. Including this variable will result in a very high VIF. This means that is a strong inter-association between those IV’s. However, this is not an issue since when removing this variable all the VIF values return to acceptable levels. All the tolerance values are above 0,2 and the VIF values are below 10 which means that multicollinearity is not an issue.

(16)

(17)

(18)

18

4. Results

4.1 Hypothesis testing

Before testing the three hypotheses, I performed an linear regression analysis to test whether the two models are significant and if the variance explained in model 2 is significantly more than in model 1. Model 1 (without the interaction term) is significant at F (2,100) = 3.347, p < 0.05. Model 2 (including the interaction term) is significant at F (3,99) = 4.476, p < 0.01. Model 2 with the interaction term between the menu category and SPFL accounted for significantly more variance than just menu category and SPFL by themselves. R^2 change = 0.057, p = 0.013. This indicates that there is potentially significant moderation between the menu category and SPFL on the total calories consumed.

Table 3 reports the results of the multiple regression analysis that includes for the first model the menu category, SPFL and both interaction terms. The second model includes the control variables: hunger, price, gender, education, BMI, age, eating out and perceived calories consumed. The control variable Difference actual and perceived calories consumed was dropped for this analysis since it correlated very highly with calories consumed. For testing the first hypothesis (nutrition labeling in

restaurants increases the healthiness of the food choice), we have to look at the variables menu 2 and

menu 3. The second model shows that these variables are significant (menu 2: p=.0064, menu 3:

p=.0183). This means that the coefficients are statistically different from zero. Since the p-value of

(19)

Figure 1: Differences between the menu types by calories consumed.

To further study the difference between menu 2 (with nutritional labeling) and menu 3 (with nutritional labeling and categories), I performed a one-way anova in which I compared the means of both menus. The results from this analysis show that there is no statistically significant difference between the two groups (F (1,66) = 0.669, p =.416).

For testing the second hypothesis (health literacy has a positive influence on the healthiness

of the food choice), we have to look at SPFL in model 2. Hypothesis 2 tests the direct effect of SPFL

on the number of calories consumed during dinner. SPFL in model 2 is not statistically significant (p=.1602). Therefore, the coefficients are not statistically different from zero. Since the p-value of SPFL is not statistically significant there is no evidence for a relationship between SPFL and the number of calories consumed during dinner. Therefore, we have to reject the second hypothesis.

For testing the third and final hypothesis (A higher health literacy positively influences the

relationship between nutrition labeling and the healthiness of the food choice), we have to look at

(20)

20 the pattern difference in the number of calories consumed between the three menus is similar to that of figure 1. Concluding, price is the only significant control variable. This can be explained by the cheapest meals containing the least number of calories and the more expensive meals containing more calories.

(21)

Table 4: Multiple regression analysis results

4.2 Follow-up analysis

(22)

22 To follow up on this analysis, I wanted to study if participants were able to more accurately predict their total number of calories consumed when nutritional information was shown on the menus. To study this, I subtracted the predicted number of calories from the actual number of calories the participants consumed. Since participants over- and underestimated the predicted number of calories consumed, I created the absolute variable of the difference. Similar to the previous analysis, I performed an multiple regression analysis. However, the absolute difference between the calories consumed and perceived calories consumed is used as the dependent variable. The outcome of this regression was not significant (p-values: menu 2 p=.0577, menu 3 p=.8659). This means that in this sample there is no statistically significant relationship between the different types of menu and how accurate the participant predicted the number of calories consumed. However, the P-value of the first menu is marginally significant, it suggests that participants which were shown menu 2 were able to more accurately predict their calories consumed.

(23)

5. Discussion

The aim of this study was to clarify the relationship between nutritional labeling in restaurants and the healthiness of the food choice and to study the moderating effect of SPFL on this relationship. The underlying mechanisms for the relationship have been set out and elaborated upon. While some studies argue that nutrition labeling in restaurants is not an effective intervention strategy (Elbel, Kersh, Brescoll and Dixon, 2009; Sinclair, Cooper and Mansfield, 2014), other found that it indeed can be an effective to lower the number of calories consumed during the stay at the restaurant (Pulos and Leng, 2010; Burton, Howlett and Tangari, 2009). This study aimed to clarify this disagreement. The healthiness of the food choices in restaurants appears to be influenced by whether or not nutritional information is displayed on the menu. Participants, on average, consumed 126 calories less when they were shown menu 2 and 112 calories less when shown menu 3, when compared to menu 1. Menu 3, with the added labels of low, medium and high, does not have a stronger effect on the number of calories consumed compared to menu 2. A possible explanation for this can be that the participants in our sample mostly have a basic understanding of nutritional information and that the added labels do not provide more information to them when making a food choice. This is confirmed by the average SPFL of 36 which means that more than half of the sample has adequate to excellent health literacy. Even participants who scored lower and have inadequate or problematic health literacy it might be possible to understand the concept of basic caloric needs.

Furthermore, I studied the direct and moderating effect of SPFL on the number of calories consumed and the main relationship. According to the analysis there is no statistically significant effect of SPFL in my model. This means for the direct effect that the number of calories consumed by the participants is not influenced by their food literacy and that participants with a lower food literacy were just as able to make an informed decision as participants with a higher food literacy. Furthermore, SPFL does not have a moderating effect on the relationship between the different menus and the total amount of calories consumed.

The findings above allow me to answer the research questions. First, adding nutritional information on menus in restaurant decreases the total amount of calories consumed. Second, in this setting there is no influence of food literacy on the relationship between nutrition labeling in restaurants on the healthiness of the food choice.

(24)

24 The second and third hypothesis pose the theory that a higher SPFL has a negative effect on the total amount of calories consumed and on the relationship between menu labeling and the total amount of calories consumed. I argued that this was the case due to that consumers that have higher food literacy are able to better understand and interpret nutrition labeling (Persoskie et al., 2017). The analysis has shown no support for either of these relationships. This means that there is, in this case, no influence of SPFL on total calories consumed and on the relationship between menu labeling and the total amount of calories consumed. A possible explanation for this could be a high level of basic understanding of calories in my sample which is not reflected in the SPFL. Furthermore, 90,3% of the sample has completed a MBO education or higher and 66% of the participants completed a HBO or WO education. Therefore, it can be assumed that participants have a basic understanding of numeracy and how it relates to calories and food consumption. Apparently, the participants, even those with a lower SPFL, are capable of making healthy decisions in regard to food choices when provided with nutritional information and that including additional labels (low, medium and high) does not have added benefits.

From the follow-up analysis, I can conclude that providing participants with nutritional information does not aid them with accurately estimating the number of calories they consume. It seems to be that estimating the number of calories consumed during the dinner is quite difficult with or without nutritional labeling. A possible explanation for this could be that participants simply do not remember the calories of the meals they have chosen, they only use the information when deciding and forget it afterwards. Furthermore, I tested for the effect of subsequent food choices on the healthiness of the dinner. It seems to be that for the most part food choices are not influenced by the healthiness of the previous course. However, this was not the case for all menu choices.

5.1 Theoretical implications

This paper adds incrementally to the literature by examining the relationship between menu labelling and the healthiness of the food choice and the influence of SPFL on this relationship by a survey under 103 participants. By proving that the main relationship is significant but that SPFL does not affect the relationship between menu category and calories consumed, this paper contributes to theory on several areas.

(25)

when compared to the menu with only nutritional information. Furthermore, this paper compares 3 different conditions and 2 different menu labels to provide clarity on the subject of the effectiveness of menu labeling.

Second, SPFL was found to not have any statistical influence on the relationship of nutritional labeling and healthiness of the food choice and as a direct effect. This paper has shown that SPFL does not play a role in the ability of participants to make a healthy decision in this setting. There are several possible explanations for this lack of effect. First, as previously discussed, the level of basic nutritional education in my sample appears to be sufficiently enough for participants to make healthy food choices when presented with nutritional labels. This can be due to how common nutritional labels are in the everyday lives of inhabitants of the Netherlands. Second, SPFL is not overreaching enough as a construct to predict the decision capabilities in regard to food choice. It does not account for numeracy and understanding of macronutrients.

Lastly, I performed two follow-up analysis to answer questions that rose during the study. Those showed that providing nutritional information on menus does not influence the predictability of the number of calories consumed during dinner by the participants. This means that nutritional labeling results in participants deciding to consume less calories but are not helping them with retaining the number of calories consumed. Additionally, I tested for the predictability of healthy food choices by the healthiness of the previous food choice. The results showed that, for the most part, a healthy starter does not necessarily result in a healthy main dish. Furthermore, it means that consumers do not consider the healthiness of the previous course when deciding their next course. The aim of these follow-up analysis is to provide future research with a base to build further upon.

5.2 Practical implications

In addition to theoretical implications, there are several lessons to be drawn from this research for manager and policy makers, which can aid in their decision making. First, this study has shown that adding nutrition labelling in restaurants reduces the total amount of calories consumed during the dinner. This proves the worth of the intervention technique of adding caloric information on menus. This technique can be added to the arsenal of intervention strategies which policy makers use to try to influence the healthiness of the general population for the better. Restaurants are one of the places which most people regularly visit and where they overeat. Reducing the amount of calories they consume during these visits can help in reducing non-communicable diseases. This is especially important for the millenials, since they eat out more often.

(26)

26 adequate healthy decision. In the Netherlands, all products that are meant for consumption in supermarkets are labelled with the number of calories, macronutrients, the percentage of the daily caloric needs and the advised daily caloric needs. This possibly adds to a more general understanding of caloric information and is not necessarily reflected by SPFL.

5.3 Limitations and directions for further research

There are several limitations to this study which can inspire and direct future research. First, since this study was done via a survey, its findings are only a reflection of the real world. Future research could follow up on this study by performing it in a restaurant setting and where participants beforehand are not informed that they are partaking in a study. This will result in findings that are more generalizable across the population. Second, the setting of this study was a fine-dining restaurant. Future research could test if the results still hold for other types of eating establishments such as self-service or fast food joints.

In terms of measurement techniques, the SPFL metric focuses on individual skills and abilities that are necessary for healthy food choices. However, the instrument does not measure all potential aspects which are relevant for the broad concept of food literacy. Cultural, political and societal aspects of food literacy are not reflected in the SPFL but can be studied by future research. Furthermore, SPFL is a self-reported form of literacy. People might over-estimate their own health literacy, or might be inclined to give the ‘appropriate’ answer. Future studies could develop a more objective method for measuring food or health literacy.

The composition of this study's participants reduces its generalizability across the general populations. First, for the most part the participants consisted of people aged below 30. While some participants were aged between 50-70 they were severely outnumbered. Second, the number of participants to this study also hinders it generalizability.

(27)

6. Conclusion

First, the aim of this study was clarifying the effect of nutritional labeling in restaurants on the amount of calories consumed during the visit. Three different menu compositions were developed to test the different relationships. The findings show that providing menus with nutritional labeling decreases the number of calories consumed during the visit. Furthermore, adding categorical calorie distinctions (high, medium and low) on the menus with nutritional labeling does not further decrease the amount of calories consumed.

(28)

28

7. References

Bezerra, I. N., Curioni, C., & Sichieri, R. (2012). Association between eating out of home and body weight. Nutrition reviews, 70(2), 65-79.

Block, J. P., & Roberto, C. A. (2014). Potential benefits of calorie labeling in restaurants.

Jama, 312(9), 887-888.

Block, J. P., Scribner, R. A., & DeSalvo, K. B. (2004). Fast food, race/ethnicity, and income: a geographic analysis. American journal of preventive medicine, 27(3), 211-217.

Burton, S., & Creyer, E. H. (2004). What consumers don't know can hurt them: consumer evaluations and disease risk perceptions of restaurant menu items. Journal of Consumer Affairs,

38(1), 121-145.

Burton, S., Creyer, E. H., Kees, J., & Huggins, K. (2006). Attacking the obesity epidemic: the potential health benefits of providing nutrition information in restaurants. American journal of public

health, 96(9), 1669-1675.

Burton, S., Howlett, E., & Tangari, A. H. (2009). Food for thought: how will the nutrition labeling of quick service restaurant menu items influence consumers’ product evaluations, purchase intentions, and choices?. Journal of Retailing, 85(3), 258-273.

Clifford, D., Anderson, J., Auld, G., & Champ, J. (2009). Good Grubbin': impact of a TV cooking show for college students living off campus. Journal of nutrition education and behavior,

41(3), 194-200.

Cohen, D., & Farley, T. A. (2008). Peer reviewed: eating as an automatic behavior.

Preventing chronic disease, 5(1).

Colatruglio, S., & Slater, J. (2016). Challenges to acquiring and utilizing food literacy: Perceptions of young Canadian adults. Canadian Food Studies/La Revue canadienne des études sur

l'alimentation, 3(1), 96-118.

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.

Duffey, K. J., Gordon-Larsen, P., Jacobs Jr, D. R., Williams, O. D., & Popkin, B. M. (2007). Differential associations of fast food and restaurant food consumption with 3-y change in body mass index: the Coronary Artery Risk Development in Young Adults Study–. The American journal of

(29)

Elbel, B., Kersh, R., Brescoll, V. L., & Dixon, L. B. (2009). Calorie labeling and food choices: a first look at the effects on low-income people in New York City. Health affairs, 28(6), w1110-w1121.

Feng, W., & Fox, A. (2018). Menu labels, for better, and worse? Exploring socio-economic and racial-ethnic differences in menu label use in a national sample. Appetite.

Fotouhinia-Yepes, M. (2013). Menu Calorie Labelling in a Fine Dining Restaurant: Will it Make a Difference?. Journal of Quality Assurance in Hospitality & Tourism, 14(3), 281-293.

Goffe, L., Penn, L., Adams, J., Araujo-Soares, V., Summerbell, C. D., Abraham, C., ... & Lake, A. A. (2018). The challenges of interventions to promote healthier food in independent takeaways in England: qualitative study of intervention deliverers’ views. BMC public health, 18(1), 184.

Hess, R., Visschers, V. H., & Siegrist, M. (2012). The role of health-related, motivational and sociodemographic aspects in predicting food label use: a comprehensive study. Public health

nutrition, 15(3), 407-414.

Huizinga, M. M., Carlisle, A. J., Cavanaugh, K. L., Davis, D. L., Gregory, R. P., Schlundt, D. G., & Rothman, R. L. (2009). Literacy, numeracy, and portion-size estimation skills. American

journal of preventive medicine, 36(4), 324-328.

Kickbusch, I. (2008). Health literacy: an essential skill for the twenty-first century. Health

Education, 108(2), 101-104.

Kolasa, K. M., Peery, A., Harris, N. G., & Shovelin, K. (2001). Food literacy partners program: a strategy to increase community food literacy. Topics in Clinical Nutrition, 16(4), 1-10.

Krukowski, R. A., Harvey-Berino, J., Kolodinsky, J., Narsana, R. T., & DeSisto, T. P. (2006). Consumers may not use or understand calorie labeling in restaurants. Journal of the American Dietetic

Association, 106(6), 917-920.

Kuhns, A., & Saksena, M. (2017). Food Purchase Decisions of Millennial Households Compared to Other Generations.

Malloy‐Weir, L., & Cooper, M. (2017). Health literacy, literacy, numeracy and nutrition label understanding and use: a scoping review of the literature. Journal of Human Nutrition and Dietetics,

30(3), 309-325.

(30)

30 Meeuwsen, S., Horgan, G. W., & Elia, M. (2010). The relationship between BMI and percent body fat, measured by bioelectrical impedance, in a large adult sample is curvilinear and influenced by age and sex. Clinical nutrition, 29(5), 560-566.

Nogueira, L. M., Thai, C. L., Nelson, W., & Oh, A. (2016). Nutrition label numeracy: disparities and association with health behaviors. American journal of health behavior, 40(4), 427-436.

Patient Protection and Affordable Care Act, 42 U.S.C. § 18031 (2010).

Persoskie, A., Hennessy, E., & Nelson, W. L. (2017). Peer Reviewed: US Consumers’ Understanding of Nutrition Labels in 2013: The Importance of Health Literacy. Preventing chronic

disease, 14.

Poelman, M. P., Dijkstra, S. C., Sponselee, H., Kamphuis, C. B., Battjes-Fries, M. C., Gillebaart, M., & Seidell, J. C. (2018). Towards the measurement of food literacy with respect to healthy eating: the development and validation of the self perceived food literacy scale among an adult sample in the Netherlands. International Journal of Behavioral Nutrition and Physical Activity,

15(1), 54.

Pulos, E., & Leng, K. (2010). Evaluation of a voluntary menu-labeling program in full-service restaurants. American Journal of Public Health, 100(6), 1035-1039.

Roberto, C. A., Larsen, P. D., Agnew, H., Baik, J., & Brownell, K. D. (2010). Evaluating the impact of menu labeling on food choices and intake. American journal of public health, 100(2), 312-318.

Rosenstock, I. M., Strecher, V. J., & Becker, M. H. (1988). Social learning theory and the health belief model. Health education quarterly, 15(2), 175-183.

Rothman, R. L., Housam, R., Weiss, H., Davis, D., Gregory, R., Gebretsadik, T., ... & Elasy, T. A. (2006). Patient understanding of food labels: the role of literacy and numeracy. American

journal of preventive medicine, 31(5), 391-398.

Schmidhuber, J., & Traill, W. B. (2006). The changing structure of diets in the European Union in relation to healthy eating guidelines. Public Health Nutrition, 9(5), 584-595.

Sinclair, S. E., Cooper, M., & Mansfield, E. D. (2014). The influence of menu labeling on calories selected or consumed: a systematic review and meta-analysis. Journal of the Academy of Nutrition and Dietetics, 114(9), 1375-1388.

Swartz, J. J., Braxton, D., & Viera, A. J. (2011). Calorie menu labeling on quick-service restaurant menus: an updated systematic review of the literature. International Journal of Behavioral

Nutrition and Physical Activity, 8(1), 135.

(31)

human energy requirements and US Department of Agriculture database information: a multisite randomized study. Journal of the Academy of Nutrition and Dietetics, 116(4), 590-598.

Vaitkeviciute, R., Ball, L. E., & Harris, N. (2015). The relationship between food literacy and dietary intake in adolescents: a systematic review. Public health nutrition, 18(4), 649-658.

VanEpps, E. M., Roberto, C. A., Park, S., Economos, C. D., & Bleich, S. N. (2016).

Restaurant menu labeling policy: Review of evidence and controversies. Current obesity reports, 5(1), 72-80.

Wansink, B., & Chandon, P. (2006). Meal size, not body size, explains errors in estimating the calorie content of meals. Annals of internal medicine, 145(5), 326-332.

Waterlander, W. E., Steenhuis, I. H., de Boer, M. R., Schuit, A. J., & Seidell, J. C. (2012). Introducing taxes, subsidies or both: the effects of various food pricing strategies in a web-based supermarket randomized trial. Preventive medicine, 54(5), 323-330.

(32)

32

8. Appendix

(33)

Menu 1: starters without nutritional labeling.

(34)

(35)

(36)

(37)

(38)

(39)

(40)

(41)

8.2 Appendix 2: Binary regression analysis Healthiness subsequent food

choice

8.3 Appendix 3: Survey

Note: 1 of 3 menus shown in appendix 1 were given to a participant to choose from, menus were randomly divided between participants.

Thank you for agreeing to take part in this study about restaurant menu choices.

In this test you will be asked to look at a restaurant’s menu and you will be asked to choose a dish from each course.

Completing the survey will take no more than 5-10 minutes. The survey is anonymous and the answers that you will provide will be kept in the strictest confidentiality.

Menu questions

Imagine that you are having a 3-course dinner at a restaurant with a friend. The waiter comes by to show you the menu. You will be asked to choose a dish for every course.

Please read the menu carefully and choose your starter dish

o

Tomato soup (1)

o

Carpaccio (2)

o

Vegan bitterballen (3)

o

Bread with spreads (4)

o

Shrimp cocktail (5)

(42)

42 Please read the menu carefully and choose your main dish

o

Entrecote (1)

o

Grilled salmon (2)

o

Mixed grill (3)

o

Vegetarian risotto (4)

o

Ceasar salad (5)

Please read the menu carefully and choose your dessert

o

Crème brûlée (1)

o

Dame blanche (2)

o

Cheese platter (3)

o

Pure chocolate mousse (4)

o

Mixed fruits with greek yoghurt (5)

SPFL

In the following section you will be asked several questions about nutritional information

When I have questions on healthy nutrition, I know where I can find information on this issue.

o

Strongly disagree (1)

o

Somewhat disagree (2)

o

Somewhat agree (3)

o

Strongly agree (4)

(43)

In general, how well do you understand the following types of nutritional information?

Very bad (1) Slightly bad (2) I do not make use of this kind of information (3) Slightly well (4) Very good (5) Nutrition information leaflets (1)

o

Food label information (2)

o

TV or radio program on nutrition (3)

o

Oral recommendations regarding nutrition from professionals (4)

o

Nutrition advice from family members or friends (5)

o

How familiar are you with the Dutch 'schijf van 5'?

o

Extremely familiar (1)

o

Very familiar (2)

o

Moderately familiar (3)

o

Slightly familiar (4)

(44)

44

To what extent do you agree with the following statement: 'I know the official Dutch recommendations about fruit and vegetable consumption'.

o

Somewhat agree (3)

o

Strongly agree (4)

To what extent do you agree with the following statement: 'I know the official Dutch recommendations about salt intake'.

o

Somewhat agree (3)

o

Strongly agree (4)

Think about a usual day: how easy or difficult is it for you to compose a balanced meal at home?

(45)

In the past, how often were you able to help your family members or a friend if they had questions concerning nutritional issues?

o

There have never been any issues (1)

o

Never (2)

o

Sometimes (3)

o

About half the time (4)

o

Most of the time (5)

o

Always (6)

There is a lot of information available on healthy nutrition today. How well do you manage to choose the information relevant to you?

o

I have not been interested in these issues (1)

o

Extremely bad (2)

o

Somewhat bad (3)

o

Neither good nor bad (4)

o

Somewhat good (5)

o

Extremely good (6)

(46)

46 Very hard (1) Somewhat hard

(2)

Somewhat easy (3)

Very easy (4)

How easy is it for you to judge if media information on nutritional issues can be trusted? (1)

o

Commercials often relate foods with health. How easy is it for you to judge if

the presented associations are appropriate or not?

(2)

o

How easy is it for you to evaluate if a specific food is

relevant for a healthy diet? (3)

o

How easy is it for you to evaluate the longer-term impact of your dietary habits on your health? (4)

o

Questions regarding the menu

Did you notice the nutrition labeling on the menu?

o

Yes (1)

o

No (2)

Did the nutrition labeling on the menu influence you to eat more/less calories?

o

To eat more calories (1)

o

Did not influence me (2)

(47)

How many calories do you think you have consumed?

________________________________________________________________

Did the pricing of the dishes influence your decisions?

o

Yes (1)

o

No (2)

General questions

In this final section you will be asked several personal questions

What is your age?

________________________________________________________________

What is your gender?

o

Male (1)

o

Female (2)

What is your highest level of completed education?

o

Highschool (1)

o

MBO (2)

o

HBO (3)

o

WO (4)

What is your height? (in cm)

________________________________________________________________

What is your weight? (in kg)

________________________________________________________________

(48)

48 When filling in this survey I was hungry

o

Strongly agree (22)

o

Agree (23)

o

Somewhat agree (24)

o

Neither agree nor disagree (25)

o

Disagree (27)

o

Do you have any dietary restrictions?

o

None (5)

o

Vegan (1)

o

Vegetarian (2)

o

Gluten (3)

o

Other: (4) ________________________________________________

How many times a month do you eat out?

________________________________________________________________

(49)

(50)

(51)

9. Results

9.1 Hypothesis testing

Before testing the three hypotheses, I performed an linear regression analysis to test whether the two models are significant and if the variance explained in model 2 is significantly more than in model 1. Model 1 (without the interaction term) is significant at F (2,100) = 3.347, p < 0.05. Model 2 (including the interaction term) is significant at F (3,99) = 4.476, p < 0.01. Model 2 with the interaction term between the menu category and SPFL accounted for significantly more variance than just menu category and SPFL by themselves. R^2 change = 0.057, p = 0.013. This indicates that there is potentially significant moderation between the menu category and SPFL on the total calories consumed.

Table 3 reports the results of the multiple regression analysis that includes for the first model the menu category, SPFL and both interaction terms. The second model includes the control variables: hunger, price, gender, education, BMI, age, eating out and perceived calories consumed. The control variable Difference actual and perceived calories consumed was dropped for this analysis since it correlated very highly with calories consumed. For testing the first hypothesis (nutrition labeling in

restaurants increases the healthiness of the food choice), we have to look at the variables menu 2 and

menu 3. The second model shows that these variables are significant (menu 2: p=.0064, menu 3:

p=.0183). This means that the coefficients are statistically different from zero. Since the p-value of

(52)

52 Figure 1: Differences between the menu types by calories consumed.

To further study the difference between menu 2 (with nutritional labeling) and menu 3 (with nutritional labeling and categories), I performed a one-way anova in which I compared the means of both menus. The results from this analysis show that there is no statistically significant difference between the two groups (F (1,66) = 0.669, p =.416).

For testing the second hypothesis (health literacy has a positive influence on the healthiness

of the food choice), we have to look at SPFL in model 2. Hypothesis 2 tests the direct effect of SPFL

on the number of calories consumed during dinner. SPFL in model 2 is not statistically significant (p=.1602). Therefore, the coefficients are not statistically different from zero. Since the p-value of SPFL is not statistically significant there is no evidence for a relationship between SPFL and the number of calories consumed during dinner. Therefore, we have to reject the second hypothesis.

For testing the third and final hypothesis (A higher health literacy positively influences the

relationship between nutrition labeling and the healthiness of the food choice), we have to look at

(53)

the pattern difference in the number of calories consumed between the three menus is similar to that of figure 1. Concluding, price is the only significant control variable. This can be explained by the cheapest meals containing the least number of calories and the more expensive meals containing more calories.

(54)

54 Table 4: Multiple regression analysis results

9.2 Follow-up analysis

(55)

To follow up on this analysis, I wanted to study if participants were able to more accurately predict their total number of calories consumed when nutritional information was shown on the menus. To study this, I subtracted the predicted number of calories from the actual number of calories the participants consumed. Since participants over- and underestimated the predicted number of calories consumed, I created the absolute variable of the difference. Similar to the previous analysis, I performed an multiple regression analysis. However, the absolute difference between the calories consumed and perceived calories consumed is used as the dependent variable. The outcome of this regression was not significant (p-values: menu 2 p=.0577, menu 3 p=.8659). This means that in this sample there is no statistically significant relationship between the different types of menu and how accurate the participant predicted the number of calories consumed. However, the P-value of the first menu is marginally significant, it suggests that participants which were shown menu 2 were able to more accurately predict their calories consumed.

(56)

56

10. Discussion

The aim of this study was to clarify the relationship between nutritional labeling in restaurants and the healthiness of the food choice and to study the moderating effect of SPFL on this relationship. The underlying mechanisms for the relationship have been set out and elaborated upon. While some studies argue that nutrition labeling in restaurants is not an effective intervention strategy (Elbel, Kersh, Brescoll and Dixon, 2009; Sinclair, Cooper and Mansfield, 2014), other found that it indeed can be an effective to lower the number of calories consumed during the stay at the restaurant (Pulos and Leng, 2010; Burton, Howlett and Tangari, 2009). This study aimed to clarify this disagreement. The healthiness of the food choices in restaurants appears to be influenced by whether or not nutritional information is displayed on the menu. Participants, on average, consumed 126 calories less when they were shown menu 2 and 112 calories less when shown menu 3, when compared to menu 1. Menu 3, with the added labels of low, medium and high, does not have a stronger effect on the number of calories consumed compared to menu 2. A possible explanation for this can be that the participants in our sample mostly have a basic understanding of nutritional information and that the added labels do not provide more information to them when making a food choice. This is confirmed by the average SPFL of 36 which means that more than half of the sample has adequate to excellent health literacy. Even participants who scored lower and have inadequate or problematic health literacy it might be possible to understand the concept of basic caloric needs.

Furthermore, I studied the direct and moderating effect of SPFL on the number of calories consumed and the main relationship. According to the analysis there is no statistically significant effect of SPFL in my model. This means for the direct effect that the number of calories consumed by the participants is not influenced by their food literacy and that participants with a lower food literacy were just as able to make an informed decision as participants with a higher food literacy. Furthermore, SPFL does not have a moderating effect on the relationship between the different menus and the total amount of calories consumed.

The findings above allow me to answer the research questions. First, adding nutritional information on menus in restaurant decreases the total amount of calories consumed. Second, in this setting there is no influence of food literacy on the relationship between nutrition labeling in restaurants on the healthiness of the food choice.

According to Cohen and Farley, eating is a behavior that is carried out with little regard to intention or self-regulation (Cohen and Farley, 2008). Studying a restaurant’s menu is possibly the only the occasion when cognition plays a role and when rational food decisions can be made (Pulos and Leng, 2010). Providing participants with menus that contain nutritional labeling provided them with more information about their food choice and thus were able to make a more informed and healthy decision.

(57)

the total amount of calories consumed and on the relationship between menu labeling and the total amount of calories consumed. I argued that this was the case due to that consumers that have higher food literacy are able to better understand and interpret nutrition labeling (Persoskie et al., 2017). The analysis has shown no support for either of these relationships. This means that there is, in this case, no influence of SPFL on total calories consumed and on the relationship between menu labeling and the total amount of calories consumed. A possible explanation for this could be a high level of basic understanding of calories in my sample which is not reflected in the SPFL. Furthermore, 90,3% of the sample has completed a MBO education or higher and 66% of the participants completed a HBO or WO education. Therefore, it can be assumed that participants have a basic understanding of numeracy and how it relates to calories and food consumption. Apparently, the participants, even those with a lower SPFL, are capable of making healthy decisions in regard to food choices when provided with nutritional information and that including additional labels (low, medium and high) does not have added benefits.

From the follow-up analysis, I can conclude that providing participants with nutritional information does not aid them with accurately estimating the number of calories they consume. It seems to be that estimating the number of calories consumed during the dinner is quite difficult with or without nutritional labeling. A possible explanation for this could be that participants simply do not remember the calories of the meals they have chosen, they only use the information when deciding and forget it afterwards. Furthermore, I tested for the effect of subsequent food choices on the healthiness of the dinner. It seems to be that for the most part food choices are not influenced by the healthiness of the previous course. However, this was not the case for all menu choices.

10.1 Theoretical implications

This paper adds incrementally to the literature by examining the relationship between menu labelling and the healthiness of the food choice and the influence of SPFL on this relationship by a survey under 103 participants. By proving that the main relationship is significant but that SPFL does not affect the relationship between menu category and calories consumed, this paper contributes to theory on several areas.

The impact of nutrition labeling and health literacy on the healthiness of food choices in restaurants