The unhealthy = tasty intuition, a construal level perspective

(1)

The Unhealthy = Tasty Intuition,

a Construal Level Perspective

Master Thesis

Name: Cecile Krijnen

Student Number: 10874100 Submission Date: 24-06-2016

Study: MSc Business Administration – Marketing Track

Institution: University of Amsterdam Thesis Supervisor: dr. A. Zerres

(2)

2

Statement of Originality

This document is written by Student Cecile Krijnen who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

This study takes a Construal Level Theory approach to the phenomenon of the Unhealthy = Tasty Intuition. An offline experiment (N=112) was conducted to investigate the moderating effect of an abstract or concrete mindset on the relationship between healthy and unhealthy portrayed food and taste. This moderating effect of mindsets was tested on taste perceptions and purchase intentions. The results reject an Unhealthy = Tasty Intuition in the Netherlands, but suggest a Healthy = Tasty Intuition. Healthy information about a cookie resulted in more favorable taste evaluations than unhealthy information. A moderating effect of mindset was found on both the relationship between portrayed healthiness and taste and on the relationship between portrayed healthiness and purchase intentions. Compared to people under a concrete mindset, people under an abstract mindset rated the taste of a healthy portrayed cookie more favorably than of an unhealthy portrayed cookie. However, people under a concrete mindset did not assess the taste or purchase intention higher of unhealthy portrayed cookies, compared to people under an abstract mindset. This complies with the unexpected finding of the Healthy = Tasty Intuition.

(5)

5

Introduction

Consumers often indicate to care about their health and mention it to be an important motive in their choices for food. Interestingly, their subsequent behavior does not always correspond to reach this goal (Ronteltap, Sijtsema, Dagevos and de Winter, 2012). There is a conflict and gap between what people know they should do and what they actually do (Ronteltap et al., 2012; Bisogni, Jastran, Seligson and Thompson, 2012). The facts about the global and still continuing rise of overweight and obesity are clear, it is a major health challenge (Ng et al., 2014). One of the reasons for the growing rate of overweight is the inclination of consumers to overconsume unhealthy food because they presume this to taste better (Raghunathan, Naylor and Hoyer, 2006). According to this presumption, healthy food is associated with a minor taste quality while unhealthy food enjoys more favorable associations such as better taste and enjoyment. This Unhealthy = Tasty Intuition (UTI) leads to suboptimal food choices and contributes to the view that healthiness and tastiness are inversely connected

(Raghunathan et al., 2006).

Claims of healthiness are often used in food marketing. These claims lead to favorable consumer impressions regarding caloric intake. But health associations are not always

favorable, they can also lead to bad impressions in the form of minor taste expectations (Schuldt and Hannahan, 2013). This complicates the tradeoff people face regarding food choices; people express to care for their health, but at the same time prefer tasty food over presumed less tasty but healthier options. Long-term health ambitions are compromised for the short term hedonic goal of higher perceived taste.

This research takes a Construal Level Theory (CLT) perspective. This theory holds that actions and events are construed on different levels. Higher construal levels comprise more abstract representations of a specific event or action, while low construal levels

(6)

6 of abstractness. Fujita and Han (2009) investigated how a concrete or abstract mindset

influences subsequent food choices. Respondents primed to an abstract mindset choose healthier food options over unhealthy food options. Similarly, respondents with a concrete mindset prefer the unhealthy option to the healthy one. This illustrates that food choices are not based on what was being thought about but rather in which manner was thought. This construal level approach illustrates how specific mindsets steer food choices.

In the present research, an offline experiment is conducted to test the potential moderating effect of different construal levels on the UTI. Following this intuition, I am testing the idea that an abstract mindset directs people to more healthful choices and a concrete mindset to less healthful choices. This is the first research that tests the interaction between these two mechanisms. If food choices can be directed by influencing construal levels, would that then moderate the relationship between healthiness and taste? This will also be tested for purchase intentions.

The theoretical framework describes the current literature in the field of the Unhealthy = Tasty Intuition and Construal Level Theory. The hypotheses are formulated based on this framework and the research gap. Thereafter, the research method will clarify the research design and process. Lastly, the results of the hypotheses are discussed, implications for practice are given and directions for future research are suggested.

(7)

7

Theoretical Framework

Unhealthy = Tasty Intuition

Unhealthy food and tastiness are inversely related (Raghunathan et al., 2006). Consumers have the tendency to follow this UTI and associate unhealthy food with a better taste and healthy food with an inferior taste. This means that a negative correlation exists between healthiness and tastiness (Raghunathan et al., 2006; Mai and Hoffman, 2015). This intuition might originate from times during human evolution where high fat and caloric foods were essential sources of energy needed for survival. A preference for energy-dense foods was functional and thus an evolutionary advantage (Mai and Hoffman, 2015).

This UTI is based on two types of sources; the internal and external source. The internal source posits that an inversed relationship exists between healthy things and

enjoyable things (Raghunathan et al., 2006). Consumers generally believe wholesome food is less enjoyable and expect to receive more enjoyment from unhealthy food. Westcombe and Wardle (1997) demonstrated how lower fat labeled food was presumed to be less enjoyable than food containing a normal fat content. Healthy food can on one hand result in positive expectations regarding caloric intake, but they also promote negative expectancies of inferior taste quality (Tuorila, Cardello and Lesher, 1994; Schuldt and Hannahan, 2013). This

inversed relationship between healthy and enjoyable things presumably originates from the Protestant work ethic. This work ethic established greater attachment to utilities than to enjoyment.Activities or stimuli were assigned to either the serious utility category or the fun-related enjoyment category. The moral obligation to prioritize necessities over luxuries resulted in people working hard to earn the right to indulge. This might have strengthened the belief that healthy food is less tasty than unhealthy food and unwholesome food is compatible with a fun intuition (Raghuanathan et al., 2006; Mai and Hoffman, 2015).

(8)

8 The external sources such as mass media and personal communication stimulate the UTI. Examples of personal communication can be found in the way parents encourage their children to eat healthy vegetables and discourage them from overconsuming harmful unhealthy food such as candy. Another illustration is mass media that exposes the public to this relationship. For example, through television or magazines, the inversed relationship between healthy and enjoyable things is emphasized (Raghunathan et al., 2006).

Does this intuition reside in nurture or is it something that resides in nature? Werle, Trendel and Ardito (2012) relate this intuition to cultural differences. Their research shows that the UTI that was discovered in the United States, does not hold in France. In France, unhealthy food is linked with bad taste and healthy food is associated with tastiness. If indeed cultural differences are the cause of this intuition, then the UTI is an intuition that is and can be taught.

Construal Levels in Food Choices

CLT explains how psychological distance influences peoples’ thoughts and behavior. Abstraction broadens peoples’ mental horizons which has implications for decisions and

behaviors. Because it shifts time perspectives and permits people to consider more

psychologically distant things (Trope, Liberman and Wakslak, 2007; Burgoon, Henderson and Markman, 2013). In other words, people construe actions and events on different levels. Higher construal levels indicate a more abstract and general understanding of the action and are more focused on the reasons behind the action and its effects, while lower construal levels of actions comprise details or specifics of an action and illustrate more concretely how the action is done (Vallacher and Wegner, 1987; Trope and Liberman, 2003; Trope et al., 2007). Lower construal levels are linked to more impulsiveness and lower self-motivations compared

(9)

9 to people in higher construal levels who are more persistent in their behavior and self-control (Vallacher and Wegner, 1987; Vallacher and Wegner, 1989; Fujita and Han, 2009).

Fujita & Han (2009) showed how choices were not based on what was being thought about but rather in which manner was being thought. Based on the CLT, they investigated how food choices are influenced by levels of abstraction. The results showed that participants chose healthy options over unhealthy options in the high level construal condition compared to the low level construal condition. This reveals how consumers can be primed in a mindset that influences their subsequent food choice. Hence, healthy eating can be interpreted at different representation levels. The research of Ronteltap et al. (2012), showed how

individuals who think about healthy eating at higher representation levels were more health minded than people who think about healthy eating at lower representation levels. Health, dieting and lifestyles are examples of the broader themes that are represented at these higher construal levels. In contrast, food products are mainly associated with concrete, low construal levels.

In general, consumers indicate to view health as an important outcome. Yet the value of these outcomes is often discounted when it takes place in a more distant future (Trope and Liberman, 2000). This illustrates the implicit tradeoff consumers face regarding health aspirations and hedonic experiences in food choices (Schuldt and Hannahan 2013).

Consumers constantly engage in a cost-benefit analysis and the dilemma of self-control when they face food choices. They often contradict and fail to live up to their indicated aspirations of foregoing healthy food choices by pursuing tasty food. Long-term nutritional ambitions are compromised for the short-term hedonic goal of perceived superior taste (Mai and Hoffman, 2015; Ramanathan and Menon, 2006).

(10)

10

Research Gap

A concrete or abstract mindset thus influences the way information is perceived and how decisions are made. Combining these theories could mean that an individual’s construal level can affect this intuition. The UTI posits that unhealthy portrayed food is perceived as tastier than healthy portrayed food. However, according to the CLT, higher levels of abstraction are able to direct people to healthier options versus lower level construals. So, if food choices can be steered by influencing construal levels, would this moderate the relationship between healthiness and taste? No research has yet investigated the moderating effect of primed construal levels on the relationship between portrayed healthiness and taste and portrayed healthiness and purchase intentions. In this study, these two theories will be tested to discover if a primed abstract or concrete mindset moderates the UTI. If food choices can be directed by influencing construal levels, would that then moderate the relationship between healthiness and taste? And what is the effect on the relationship between healthiness and purchase intention?

Theory for Hypotheses

Healthy labeled food products can promote the positive assumption that it is a healthier choice in for example caloric intake or fat content. However it also provokes the negative assumption that such foods have minor taste quality (Tuorila et al., 1994; Raghunathan et al., 2006). Consumers assign higher taste expectations to full fat products than fat free products. A composition of lower fat or sugar content reduces taste evaluations (Mai and Hoffman, 2015). Raghunathan et al. (2006) demonstrated in an experiment how enjoyment ratings were higher in the condition who received unhealthy information compared to the healthy information condition. Based on this theory it can be proposed that an UTI exists. Therefore I expect to replicate this UTI effect and predict:

(11)

11

H1: The taste of an unhealthy portrayed product is evaluated more favorably than the taste of a healthy portrayed product.

Even though people are unable to taste differences between low fat or sugar products and full fat or sugar products, healthier portrayed products also reduce purchase intentions. This means that the UTI does not just affect taste judgments but also actual decisions (Mai and Hoffman, 2015; Raghunathan et al., 2006). This is in accordance with the finding of Mai and Hoffman (2015) who demonstrated that purchase intentions were lower for healthy portrayed food items. Therefore I predict that the same UTI does not only affect taste but also purchase intention.

H2: The purchase intention of an unhealthy portrayed product is higher than the purchase intention of a healthy portrayed product.

Consumers prefer food that, according to the nutrition label, contains lower fat and sugars because they perceive it as healthier. However, after sensory judgments, food containing higher fat and sugar levels is favored (Hoppert, Mai, Zahn, Hoffmann and Rohm 2012). This suggests that the UTI is influentially adequate to also affect the hedonic quality of taste experiences. Westcombe and Wardle (1997) demonstrate how health-minded consumers rated the higher fat labeled food as less pleasant compared to less health-minded consumers. Also, the results of Fujita and Han (2009) show that the number of people who chose healthy options over unhealthy options was greater in the high level construal condition than in the low level construal condition. Product evaluations were not based on what was being thought about, but rather in which manner was being thought. Therefore, the mindset matters. Based on these theories, I expect that a primed abstract (vs. concrete) mindset steers consumers to think more in broader themes such as health and lifestyles. In accordance with an abstract

(12)

12 mindset I predict consumers to favor the tastiness of a healthy portrayed stimuli in

comparison with an unhealthy portrayed stimuli.

H3a: Under an abstract mindset, the taste of a healthy portrayed product is evaluated more favorably than the taste of an unhealthy portrayed product.

Consumers indicate health as an important outcome, yet the value of these outcomes is often discounted when it takes place in a more distant future. Meaning that consumers often know the most optimal choice for their health but decide to forego a suboptimal choice in the here and now. The closer an event or choice takes place in time, the more will be relied on affective (vs. cognitive) factors which in this case leads to food choices based on tastiness (Trope and Liberman, 2000). Lower construal levels are also linked to higher impulsive behaviors compared to higher construal levels that is associated with higher self-control (Vallacher and Wegner, 1987; Vallacher and Wegner, 1989; Fujita and Han, 2009). Based on these theories I expect consumers in a concrete mindset to prefer the unhealthy stimuli compared to the healthy portrayed stimuli.

H3b: Under a concrete mindset, the taste of an unhealthy portrayed product is evaluated more favorably than the taste of a healthy portrayed product.

In comparison to health as abstract motive, concrete motives like price and sensory appeal may overrule this abstract value at moments of purchase (Ronteltap et al., 2012). This means that the short term and concrete benefits are deemed more important at that moment in time. Fujita and Han (2009) showed that levels of abstraction influence subsequent food choices. People primed inhigher construal levels made more healthful choices compared to people who were primed in lower construal levels. Based on these theories, I expect that people who are primed with a concrete mindset have a higher likelihood of purchasing unhealthy

(13)

13

H3

H1 H4

H1

portrayed food compared to abstractly primed people. Therefore, I predict that the same mechanism does not only affect taste but also purchase intention. This results in the following hypotheses.

H4a: Under an abstract mindset, the purchase intention of a healthy portrayed product is higher than of an unhealthy portrayed product.

H4b: Under a concrete mindset, the purchase intention of an unhealthy portrayed product is higher than of a healthy portrayed product.

Conceptual Model

This conceptual framework provides an overview of all examined relationships in the study.

Portrayed Healthiness (Healthy or Unhealthy product information) Taste Concrete or Abstract Mindset Purchase Intention H1 H2 H1

(14)

14

Method

Design

This study has an experimental design of 2 (portrayed healthiness: healthy/unhealthy) by 2 (construal level: abstract/concrete) between-subjects design with as dependent variables: evaluated taste and purchase intentions. In total 112 people (57 female, Mage = 39.8 years, Age

Range: 16-79 years) voluntarily participated in the offline experiment for which the data was gathered in the form of a survey (see Appendix 1). In total 110 people were of Dutch origin and the remaining two were German. Participants were recruited in the city center of Utrecht and were randomly assigned to one of the four experimental groups. They were primed with either an abstract or concrete mindset and received either healthy or unhealthy information concerning a cookie, as can be seen in Table 1. Participants remained unaware of the other conditions.

Table 1: 2x2 Experimental Design

Healthy information Unhealthy information

Abstract mindset AH AU

Concrete mindset CH CU

A private label supermarket cookie was selected in the experiment that is not well known and not widely available, in order to control for recognition of the cookie. The private label brand of Hoogvliet was chosen, because the supermarket only holds 64 stores in the Netherlands. Deliberately a simple cookie was chosen in order to prevent a ceiling effect. The cookie is a plain grain biscuit that can be interpreted in two ways; it is a grain biscuit and therefore healthy or it is a cookie, which is normally considered unhealthy. Please find an image of the cookie in Appendix 2 and the nutritional information in Appendix 3.

(15)

15

Manipulation of healthiness

I manipulated the level of healthiness by changing the information about the cookie. Two versions of information provision were tested, both having two treatments. All participants were shown the same factual nutritional value of the cookie before the tasting (Appendix 3). The cookie and nutritional information was the same in all conditions, only the information regarding healthiness differed, as can be seen in Table 2. Participants in the first version received qualitative information about the cookie and were told it was a classical cookie, which is the unhealthy condition. In the second treatment, participants evaluated the same cookie but received the information that it was a classical cookie made with a specific interest in health. Participants in the second version of the pretest had to read a text about good and bad fats before they were told how many grams of fat their cookie contained, please find this text in Appendix 4 (Raghunathan et al., 2006). Participants in the first treatment were told that their cookie consisted of mainly good fats (healthy condition). The second treatment received the information that their cookie contained mainly bad fats (unhealthy condition).

Table 2: Healthiness Manipulation

Version 1 Healthy condition This is a classical cookie that has been made with particular interest on healthiness as it contains less sugars and less saturated fats than comparable cookies. So it is a healthier choice.

Unhealthy condition This is a classical cookie that contains a normal level of sugar and normal distribution of good and bad fats just like other cookies.

Version 2 Healthy condition This cookie contains 2,5 grams of fat: 2 grams of good (unsaturated) fats and 0,5 grams of bad (saturated) fats.

Unhealthy condition This cookie contains 2,5 grams of fat: 2 grams of bad (saturated) fats and 0,5 grams of good (unsaturated) fats.

A pretest was conducted in which 54 people (30 female, Mage = 31.8 years, Age Range: 15-68

years) voluntarily tasted a cookie. They assessed the flavor of this stimulus and rated it on a ten point grading scale. This test was used to see if the manipulations of the qualitative information regarding healthiness and nutritional information gave enough variance for further research. Based on the results of the pretest it was decided to continue with version one of the healthiness manipulation. An independent sample t-test supported the effectiveness

(16)

16 of the healthiness manipulation of version one (Mhealthy = 6.08, SDhealthy = 1.61; Munhealthy =

4.46, SDunhealthy = 2.26, t(26) = -2.10, p < 0.05). The effectiveness of version two was not

supported (Mhealthy = 5.36, SDhealthy = 1.45; Munhealthy = 6.36, SDunhealthy = 1.78, t(28) = 1.63, p >

0.05). Therefore the healthiness manipulation of version one is integrated in the main study.

Manipulation of construal level

The effectiveness of the construal level manipulations was verified through a second pretest with a different sample. Participants were primed either to an abstract or concrete mindset. Participants in the abstract condition were asked to think about cookies on a higher construal level (see Table 3). Abstract self-thought was elicited by asking participants to focus on an idea, attitude or opinion they hold regarding their lives. By probing questions and digging deeper into the reason behind their answers, an abstract response and mindset is elicited. Concrete self-thought was provoked by letting participants focus on concrete experiences or specific events in their lives (Updegraff and Suh, 2007; Freitas, Gollwitzer and Trope, 2004; Fujita and Han, 2009). Participants in the concrete condition were asked to think about the last cookie purchase they did in the supermarket of their preference. Specific questions force participants to remember exactly how they performed their actions. So the questions were very specific, for participants to enter a more concrete mindset.

(17)

17

Table 3: Construal Level Manipulation

Abstract Before we start, I want you to briefly think about cookies.

1. Why do you buy cookies?

And why is that important to you? 2. How do cookies fit in your lifestyle?

And why is that important to you?

3. How important is the packaging of a cookie to you? And why is that the case?

Concrete Before we start, I want you to briefly think about the last purchase of cookies you did in your preferred supermarket.

1. How do you usually go to your supermarket?

2. Can you remember exactly where in your supermarket you find the cookies? Please describe where this is.

3. Where are they placed on the shelfs?

More on the top or more on the bottom? More on the left or more on the right?

4. Please describe how you usually open up the packaging of a cookie.

Thereafter, the Behavior Identification Form of Vallacher and Wegner (1989) was used to measure the effectiveness of the manipulation. In their technique eating can for example be seen as getting nutrition, which is a higher level construal, or putting food in my mouth, indicating a lower level construal. Participants indicated which description best described the five actions visible in Table 4. One option is a more abstract option and the other is more concrete. The abstract and concrete answering options are deliberately shown in a random order.

Table 4: Construal Level Statements

Greeting someone Saying ‘hello’ Showing friendliness

Taking a test Showing my knowledge Answering questions

Eating Getting nutrition Putting food in my mouth

Making a grocery list Writing things down Getting organized

Getting groceries Putting it in my basket Caring for myself/my family

These five statements at the end of the questions indicate the effectiveness of the construal level manipulations. The answers were coded 0 for the concrete option and 1 for the abstract option. Five statements indicating a concrete or abstract mindset have been taken together to form one variable on a scale of 0 to 1. If someone answers the five statements with all five abstract options then the outcome of this participants is 0, compared to 1 if all five concrete

(18)

18 answering options were chosen. A pretest was conducted in which 26 participants (13 female, Mage = 44.5 years, Age Range: 21-81 years) voluntarily participated.

An independent sample t-test supported the effectiveness of the construal level

manipulation (Mabstract = .68, SDabstract = .21; Mconcrete = .43, SDconcrete = 0.18, t(28) = -3.222, p =

0.004 < 0.01), a mean of 1 indicating complete abstractness and a mean of 0 a completely concrete mindset. These results support the effectiveness of the manipulation and show how the abstractly primed condition is significantly different from the concretely primed group.

Experiment description

As a cover story, participants were told that the research is focused on the degree to which packaging influences taste perception. I explained to hypothesize that if a cookie is wrapped in an attractive packaging, the perceived tastiness is higher compared to a neutral packaging. In this experiment, the cookie was presented in a neutral transparent packaging (see Appendix 2). Participants were told they were in the control condition in order for me to compare them to other groups who get to see a different packaging. The tasting had to feel real for the participants, avoiding that they notice they are actually taking part in an experiment

concerning healthiness and mindset manipulations. For this reason, several packaging related questions are included to the research to make the cover story believable. Three questions were used from the Product Aesthetics Centrality Scale of Bloch, Brunel and Arnold (2003). Actually all participants are in the ‘control condition’ and no one got to see a different

packaging. Only at the end of the experiment, I showed participants the original packaging of the cookie they tasted. Both pretests holding the same cover story did not generate any suspicion. All measures and scales have been designed in English and have been back

translated from Dutch to English by two independent people to prevent any loss of meaning in translation.

(19)

19 The manipulated construal-level was evoked at the start of the experiment, before they see and taste the cookie (see Appendix 1). The participants need to be in this mindset from the moment they receive the information, express their expectations and rate the cookie. Because it is unclear when the effect will happen. According to Lee, Frederick and Ariely (2006)no cognitive correction takes place anymore after the actual tasting of the cookie. The mindset manipulation happened exactly the same way as the second pretest. Participants were primed either abstractly or concretely through the method of abstract or concrete self-thought

(Updegraff and Suh, 2007).

After having moved the participants in a concrete or abstract mindset, the participants got exposed to either healthy or unhealthy nutrition information about the cookie. All

participants had to read the nutritional information before tasting the cookie. This information contains the real nutritional values of the cookie, only the (un)healthy information that is being told differs. Before tasting, participants answered their level of agreement to three statements concerning packaging on a ten-point grading scale. Thereafter the participants expressed their expectations of the cookie. These expectations were measured by three questions regarding tastiness, healthiness and level of enjoyment on ten-point grading scales: “not tasty at all - very tasty”, “not healthy at all - very healthy” and “not enjoyable at all - very enjoyable”. After having tasted the cookie, all participants indicated the level of tastiness, the

level of sweetness and were asked how likely it is that they would buy the cookie. All this happened based on a ten point grading scale: “not tasty at all -very tasty”, “not sweet at all - very sweet” and “not likely at all - very likely”.

Three questions that indicate taste have been combined to form one scale. These were the questions regarding taste expectation, expected level of enjoyment and actual tastiness. The Cronbachs Alpha of these three items is .82. Similar to the pretest, the five statements

(20)

20 indicating a concrete or abstract mindset have been taken together to form one variable on a scale of 0 to 1.

After having measured their construal level, the experiment was finalized with the gathering of demographic information of the participants. As a control variable, participants were also asked the last time they ate something. This might influence their taste evaluation of the cookie. If it has been a long time since one’s last meal, then the taste of a product is most likely higher because of the desire to eat something (Moskowitz, Kumraiah, Sharma, Jacobs and Sharma, 1979). As a second control variable, participants were asked with which traditional culinary culture they grew up, as this could also affect taste perceptions.

(21)

21

Results

The means, standard deviations and correlations of the main variables of this study are reported in Table 5. The table clarifies the correlations and strength between variables. For this analyses the independent variables (portrayed healthiness and mindset), dependent variables (taste and purchase intention) and control variables (gender, age, education, last eating time 2 hours ago, last eating time 3 or more hours ago and Dutch kitchen) are included.

Table 5: Means, Standard Deviations, Correlations

Variables M SD 1 2 3 4 5 6 7 8 9 10

1. Gender 1.51 .50

2. Age 39.79 17.67 .04

3. Education 2.71 1.28 -.01 .09

4. Last Eating time 2 hours agoa

.28 .45 .00 .06 -.00 5. Last Eating time 3 or more hours agoa .24 .43 -.03 -.13 -.04 -.35** 6. Dutch Kitchenb .78 .42 .07 .19* .08 -.05 .15 7. Portrayed Healthinessc 0.50 .50 -.02 .22* -.11 -.10 .15 .24* 8. Mindsetd _0.50 _.50 _-.09 _.09 _-.06 _-.06 _-.02 _-.02 _.00 9. Taste 6.32 1.52 .02 .04 -.20* -.15 .28** .02 .24** .22* (α .82) 10. Purchase Intention 4.54 2.69 .10 .09 -.25** -.08 .23* .22* .28** -.06 .70**

**Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed). a_{Reference group is in the last hour}

b _{Reference group is other kitchens.}

c_{Portrayed Healthiness (Unhealthy = 0, Healthy = 1)} d_{Mindset (Concrete = 0, Abstract = 1)}

There are five significant correlations between control variables and variables from the model. First, portrayed healthiness is significantly correlated with age (r = .22). Also Dutch kitchen versus other kitchens shows a significant positive correlation with age (r = .19). Third, taste is negatively correlated with education (r = -.20) significant at the .05 level. This means that the higher the age of the participants, the lower they rate tastiness. Fourth, purchase intention is negatively correlated with education level (r = .25) significant at the .01 level.

(22)

22 This correlation shows a similar direction as education level and evaluated taste. Not

surprisingly then is the significant highly positive relation between taste and purchase intention (r = .70). This is as expected, if participants enjoy the taste of the cookie, the

likelihood of buying is also higher. There was also a positive relation between people who ate three hours ago (versus in the last hour) and taste (r = .28). This was intuitively expected, because people who have not eaten for a long time are more likely to desire a cookie. Portrayed healthinessis significantly positively correlated with taste (r = .24) and purchase intention (r = .28). This is unexpected as this does not align with the UTI. Mindset shows a positive correlation with taste (r = .22), indicating a relationship between an abstract mindset and taste evaluation. Mindset is not significantly correlated with purchase intention.

Manipulation check – portrayed healthiness

I have used an independent sample t-test to check if the informed healthiness of the cookie was significantly differently perceived across the healthy informed groups (Abstract-Healthy and Healthy) and unhealthy informed groups (Abstract-Unhealthy and Concrete-Unhealthy). The healthy informed groups rated the healthiness higher on a scale of 1 to 10 (M = 6.91, SD = 1.18) than the unhealthy informed group (M = 5.50, SD = 1.18), this difference was significant (t (110) = -4.83, p < .01). Thus, the manipulation of healthiness was

successful, although both groups score above five on a ten-point scale.

Manipulation check – abstract and concrete mindset

An independent sample t-test was executed to check if the primed abstract or concrete

mindset was significantly different across the abstractly primed groups (Abstract-Healthy and Abstract-Concrete) and concretely primed groups (Healthy and

(23)

23 0.25, SDabstract = .18; Mconcrete = 0.59, SDconcrete = .27). The difference between the means is

significant (t (110) = 7.854, p < .01), with a mean of 0 being completely abstract and a mean of 1 completely concrete. This significant difference indicates that people in the concrete group answered more than half of the options to the concrete pole and the people in the abstract group answered less than half to the concrete pole.

Hypotheses Testing

To test whether the effect of the UTI is still present, an independent sample t-test was conducted with a healthy portrayed cookie and unhealthy portrayed cookie as grouping variables and evaluated taste as dependent variable. There is a significant difference between the unhealthy portrayed cookie (M = 5.95, SD = 1.42) and healthy portrayed cookie (M = 6.69, SD = 1.55, t(110) = -2.63, p < 0.05) in taste evaluation. The results show a significant relationship, though the direction was reversed to what was expected in the hypothesis. Participants in the condition exposed to the healthy portrayed information rated the taste of the cookie higher than participants in the unhealthy condition. Therefore, Hypothesis 1 is not supported. In fact, a positive significant effect was found, while a negative effect was

expected. This suggests a Healthy = Tasty Intuition and means that receiving healthy information regarding the cookie results in more favorable taste evaluations compared to receiving unhealthy information.

An independent sample t-test was performed to test Hypothesis 2. The results show a significant effect opposed to what was expected. Participants in the healthy portrayed

condition indicated a significantly higher likelihood of purchasing the cookie (M = 5.29, SD = 2.83) compared to participants exposed to unhealthy information (M = 3.79, SD = 2.35, t(110) = -3.05, p < .01). This finding rejects Hypothesis 2, because the likelihood of buying

(24)

24 the finding of Hypothesis 1. A higher evaluated taste would intuitively result in a higher likelihood of purchasing. This is confirmed by a positive correlation (r = .70), between taste and purchase intention as can be found in Table 5. This correlation is regarded high as it is based above .5, significant at the .01 level. The results indicate that portraying healthy product information compared to unhealthy product information results in a higher likelihood of purchasing the cookies.

A two-way analysis of variance was conducted to test Hypothesis 3a and 3b. This analysis compared the main effects of portrayed healthiness and the mindset and tested the interaction effect between these two variables on taste evaluation. Portrayed healthiness included two levels (unhealthy portrayed information and healthy portrayed information) and mindset consisted of two levels (concrete and abstract). The assumption of homogeneity of within-group variances is met (F = .319, p > .05). The total variance explained by the model is 15.6%.

Table 6: Factorial ANOVA (N=112), Dependent Variable: Taste Evaluation

SS DF MS F η2 Sig. Portrayed Healthiness Mindset Portrayed Healthiness*Mindset Error Total 15.25 12.44 12.44 217.18 4732.89 1 1 1 108 112 15.25 12.44 12.44 2.01 7.59 6.19 6.19 .07 .05 .05 .007 .014 .014

(25)

25 There was a significant main effect of portrayed healthiness on taste perception, as can be seen in Table 6 (F (3, 108) = 7.59, p < .01, η2 = .07). This indicates a significant difference between unhealthy portrayed information (M = 5.95, SD = 1.42) and healthy portrayed information (M = 6.69, SD = 1.55). The result reconfirms that the healthy portrayed cookie is evaluated more favorably than the unhealthy portrayed cookie, similar to the finding of Hypothesis 1. The main effect of mindset on taste perception was also significant (F (3, 108) = 6.19, p < .05, η2 = .05). This shows a significant difference between the concrete (M = 5.99, SD = 1.41) and abstract (M = 6.65, SD = 1.57) mindset. There was a significant interaction effect between healthiness and mindset on the evaluated taste (F (3, 108) = 6.19, p < .05, η2 = .05), see the visualization in Figure 1. The moderation is driven by the contrasts involving the Abstract-Healthy condition, in which participants were abstractly primed and received healthy information regarding the cookie (M = 7.36, SD= 1.40). This condition was significantly different from Abstract-Unhealthy (M = 5.95, SD = 1.44; t(54) = 3.71, p < .01), Concrete-Healthy (M= 6.02, SD = 1.42; t(54) = 3.55, p < .01) and Concrete-Unhealthy (M = 5.95, SD = 1.42; t(54) = 3.73, p < .01). This supports Hypothesis 3a; under an abstract mindset, the taste of a healthy portrayed cookie is evaluated more favorably than the taste of an unhealthy portrayed cookie. There were no other significant differences in the model. The difference Abstract-Unhealthy (M = 5.95, SD = 1.44) and Concrete Healthy (M = 6.02, SD = 1.42) was insignificant with a p-value of .85 (t(54) = -.19). There was no difference between the Abstract-Unhealthy condition and Concrete Unhealthy condition (M= 5.95, SD = 1.42; t(54) = .00, p = 1.00). In other words, there were no differences in taste evaluations between participants under concrete or abstract mindsets who received unhealthy information. This means that when unhealthy product information is portrayed, an abstract or concrete mindset yields the same taste evaluations.

(26)

26 There were also no significant differences between the Concrete-Healthy condition and Concrete-Unhealthy condition (t(54) = .19, p > .05). This implicates that concretely primed participants rated an unhealthy and healthy portrayed cookie almost equally with a mean difference of .07. Therefore Hypothesis 3b is not supported, participants did not rate the taste of an unhealthy portrayed cookie more favorably than the taste of a healthy portrayed cookie, but only marginally insignificant the opposite way. For a concrete mindset there is no difference between healthy-unhealthy product information, but there is a significant difference for abstract mindset. For abstract mindset, it matters which information regarding healthiness of the cookie is portrayed. The difference between the healthy and unhealthy information only has an effect for an abstract mindset. For the concrete condition, healthy information does not matter, but for the abstract condition it matters and has a positive effect on taste. Under an abstract mindset the cookie is perceived as more tasty when participants received healthy product information. So you need an abstract mindset for the information about healthiness to affect the taste evaluation. Under an abstract mindset, healthy product information results in an increased taste perception. Unhealthy product information about the cookie resulted in significantly lower taste ratings compared to healthy information. Adding healthy information only reinforces taste evaluations for people under an abstract mindset. But unhealthy

information does not weaken the taste evaluation for an abstract mindset opposed to a concrete one. In total 7% of the effect is driven by healthiness and 5% by the mindset, meaning that healthiness has a higher contribution in the explanatory power of the model.

A two-way analysis of variance was conducted to test the influence of portrayed (un)healthiness of the stimuli and mindset on purchase intention. Portrayed healthiness included two levels (unhealthy portrayed information and healthy portrayed information) and mindset consisted of two levels (concrete and abstract). The assumption of homogeneity of

(27)

27 within-group variances is met (F = .66, p > .05). The total variance explained by the model is 15.6%.

Table 7: Factorial ANOVA (N=112), Dependent Variable: Purchase Intention

SS DF MS F η2 Sig. Portrayed Healthiness Mindset Portrayed Healthiness*Mindset Error Total 63.00 18.89 43.75 680.21 3110.00 1 1 1 108 112 63.00 18.89 43.75 6.95 10.00 3.00 6.95 .09 .03 .06 .002 .086 .010

Figure 2: Interaction of Portrayed Healthiness and Mindset on Purchase Intention

All effects were statistically significant at the .05 level except for the main effect of mindset. The main effect for portrayed healthiness produced an F ratio of F (3, 108) = 10.00,

significant at the .01 level, η2 = .09, as can be found in Table 7. This indicates a significant difference between unhealthy portrayed information (M = 3.79, SD = 2.35) and healthy portrayed information (M = 5.29 , SD = 2.83). The main effect for mindset yielded an F ratio of F (3, 108) = 3.00, p > .05, η2 = .03, indicating that there was a non-significant difference between a concrete (M = 4.13 , SD = 2.50) and abstract mindset (M = 4.95, SD = 2.84). The interaction effect was significant (F (3, 108) = 6.95, p < .05, η2 = .06).

Looking at Figure 2, the effects seem to be driven by the Abstract-Healthy condition (M = 6.32, SD = 2.60). This is confirmed by the several t-tests that were performed to test for differences in specific contrasts. This Abstract-Healthy condition was significantly different from the Abstract-Unhealthy condition (M =3.57, SD = 2.40; t (54) = 4.12, p < .01) It was

(28)

28 also significantly different from both concrete conditions; Concrete-Healthy (M = 4.25, SD = 2.70, t(54) = 2.92, p < .01) and Concrete-Unhealthy (M = 4.00, SD = 2.33, t(54) = 3.52, p < .01). These findings support Hypothesis 4a. Under an abstract mindset, the likelihood of buying a healthy portrayed cookie was significantly higher than for an unhealthy portrayed cookie. All other contrasts were insignificant. There was no significant difference (p > .05) between Abstract-Unhealthy (M = 3.57, SD = 2.40) and Concrete-Healthy (M = 4.25, SD = 2.70, t(54) = -.99, p > .05). The same holds for Abstract-Unhealthy and Concrete-Unhealthy (M = 4.00, SD = 2.33, t (54) = -68, p > .05). This means that Hypothesis 4b is not supported. People under a concrete mindset do not have a higher likelihood of purchasing unhealthy portrayed products than healthy portrayed products. There were no significant differences between these groups. Having an abstract or concrete mindset does not significantly yield different purchase intentions when unhealthy information is portrayed. Only people under an abstract mindset who are exposed to healthy product information indicate a higher likelihood of purchasing. The relation between portrayed healthiness and purchase intention is

moderated by mindset. The results show a significant interaction effect. In an abstract mindset, healthy information has a positive effect on purchase intention. This means that it matters if you are in a concrete or abstract mindset for the purchase intention of a healthy portrayed cookie. For an unhealthy portrayed cookie there was no significant difference between the abstract and concrete condition. Again the interaction effect was mainly driven by portrayed healthiness, which explained 9% of the effect. Mindset only has a moderating effect, there was no direct effect.

When visually comparing Figure 1 and Figure 2, it is noticeable that for purchase intention, healthiness has a higher influence under an abstract mindset than for taste. The interaction for purchase intention shows a steeper line than the line for taste. It is a difference of approximately 1.5 point on the rating scale. Also a slight trend seems to be going on in

(29)

29 which unhealthy information decreases purchase intention under an abstract mindset.

However this was not proven significant and can therefore be coincidental. Healthy

information has a stronger influence on purchase intention than it has on taste. Taste is being influenced somewhat less than purchase intention when healthy information is given under an abstract mindset.

Regression Analyses

The two-way ANOVA already accepts the moderating effect of mindset. As an additional test of reliability, I also added different control variables in a regression analysis. The assumption of homogeneity of variances and the assumption that residuals are normally distributed are both approximately met (see Appendix 5). There is no multicollinearity (VIF is structurally below 1.5 < 4).

The variables portrayed healthiness and mindset have a direct significant effect, the moderation is also significant (see Table 8). This confirms the findings of the ANOVA. After controlling for age, education, last eating time 2 hours ago, last eating time 3 or more hours ago and Dutch kitchen, the effect is still significant. The total variance explained by this model is 26% (F (9, 102) = 4.07; p < .01). The main effects remain and the moderation is significant. In comparison with the ANOVA, the addition of control variables and interactions explained an additional amount of variance (from 15.6% to 31.4%). The additions to the model add that the control variable last eating time 3 or more hours ago has a significant positive effect on taste. Also the moderation between portrayed healthiness and education was significant and confirms the results similar to those found in the ANOVA.

(30)

30

Table 8: Regression (N=112), Dependent Variable: Taste

R R2 _R2_Change _B _SE _β _t

Step 1 Gender Age Education

Last eating time 2 hours ago

Last eating time 3 or more hours ago Dutch Kitchen .34 .12* .04 .01 -.52 -.18 .97 -.14 .28 .01 .28 .33 .36 .35 .01 .08 -.17 -.05 .27** -.04 .15 .862 -1.85 -.53 2.74 -.40 Step 2 Gender Age Education

Last eating time 3 or more hours ago Dutch Kitchen Portrayed Healthiness Mindset .45 .20** .08** .13 .00 -.43 -.08 .91 -.24 .60 .67 .27 .00 .27 .32 .34 .34 .29 .27 .04 .02 -.14 -.02 .26** -.07 .20* .22* .49 .20 -1.59 -.26 2.65 -.71 2.10 2.49 Step 3 Gender Age Education

Last eating time 3 or more hours ago Dutch Kitchen

Portrayed Healthiness Mindset

Moderation Portrayed Healthiness*Mindset Moderation Portrayed Healthiness *Age Moderation Mindset*Age

Moderation Portrayed Healthiness *Education Moderation Mindset*Education .56 .31*** .12** .28 -.01 .62 .20 1.20 -.34 -.07 -.12 1.37 .01 .01 -1.12 -.71 .26 .02 .47 .32 .34 .33 .79 .75 .55 .02 .02 .55 .54 .08 -.13 .20 .06 .34* -.09 -.02 -.04 .31* .21 .17 -.31* -.20 .86 -.74 1.31 .61 3.47 -1.04 -.09 -.16 2.50 .87 .69 -2.05 -1.32 *** Correlation is significant at the 0.00 level (2-tailed).

An additional moderating effect of education was found on the relationship between portrayed healthiness and taste. For low education there is practically no effect while for high education a negative effect was found (Figure 3). In other words, the higher the portrayed healthiness of the cookie, the lower is the taste evaluation for higher educated people. This interesting finding points to an UTI for higher educated people.

(31)

31

Figure 3: Interaction of Portrayed Healthiness and Education on Taste Evaluation

The two-way ANOVA already supports the moderating effect of mindset on purchase intention. An additional test of reliability is also conducted for purchase intention. The same control variables as in the previous regression analysis are added. All assumptions of running a regression are (approximately) met: homogeneity of variances, normal distribution of residuals and no multicollinearity (VIF is structurally below 1.5 < 4), please see Appendix 4.

After controlling for gender, age, education, last time eaten 2 hours ago, last time eaten 3 hours or more ago and Dutch kitchen,the main effect of health is significant. This is similar to the findings of the ANOVA (see Table 9). For mindset there is also no significant main effect. The moderation between health and mindset is significant. All in all, the model confirms the ANOVA. The only difference in comparison to the ANOVA is the additional amount of variances. The addition of control variables and interactions explained an additional amount of variance (from 15.6% to 30%). The additional explained variance is explained by the direct effects of last eating times 3 or more hours ago and education. For purchase intention, education has a direct negative effect while for taste this was an indirect effect.

(32)

32

Table 9: Regression (N=112), Dependent Variable: Purchase Intention

R R2 _R2_Change _B _SE _β _t

Step 1 Gender Age Education

Last eating time 3 or more hours ago Dutch Kitchen .39 .15** .38 .01 -1.21 .06 1.34 1.06 .49 .01 .49 .58 .62 .60 .07 .08 -.23* .01 .21* 1.76 .78 .85 -2.50 .10 2.19 1.76 Step 2 Gender Age Education

Last eating time 3 or more hours ago Dutch Kitchen Portrayed Healthiness Mindset .46 .21** .06* .51 .00 -1.07 .20 1.23 .87 1.04 .89 .48 .01 .48 .56 .60 .60 .50 .48 .10 .02 -.20* .03 .20* .14 .19* .17 1.07 .25 -2.24 .35 2.03 1.46 2.07 1.86 Step 3 Gender Age Education

Last eating time 3 or more hours ago Dutch Kitchen

Portrayed Healthiness Mindset

Moderation Portrayed Healthiness*Mindset Moderation Portrayed Healthiness*Age Moderation Mindset*Age

Moderation Portrayed Healthiness*Education Moderation Mindset*Education .55 .30*** .09* .59 -.00 -.24 .42 1.57 .75 .86 -1.45 2.64 -.01 .02 -1.66 .38 .47 .03 .84 .57 .62 .59 1.41 1.33 .98 .03 .03 .98 .97 .11 -.01 -.05 .07 .25* .11 .16 -.27 .43* -.05 .16 -.26 .06 1.25 -.08 -.29 .73 2.55 1.29 .61 -1.09 2.70 -.19 .64 -1.70 .39 *** Correlation is significant at the 0.00 level (2-tailed).

Both regression analyses confirm the strong similarities between taste and purchase intention like in the ANOVAS. Not only are they strongly correlated, there are similarities in the way they can be explained. The ANOVAS and regression analyses have an equal share of the variances that can be explained in taste and purchase intention. Therefore it seems that purchase intention and taste are driven by the same mechanism.

(33)

33

Discussion

The aim of this study was to test the moderating effect of construal levels on the relation between healthy and unhealthy portrayed food and taste and on purchase intentions.

Results of Hypothesis 1 show that participants did not evaluated taste according to the UTI. In contrast, taste evaluations of healthy portrayed food were higher than for unhealthy portrayed food. This means that participants do not act according to the UTI. Instead, they expect healthy food to have a better taste compared to unhealthy food. Therefore Hypothesis 1 is rejected. The results indicate the existence of an opposed Healthy = Tasty Intuition (HTI) instead. A possible explanation is found in the theory of Werle et al. (2012) who relate this to cultural differences. Their research proved that in France, unhealthy food is linked with bad taste and healthy food is associated with tastiness. This means that food is perceived

differently among different cultures. The research of the UTI has been conducted in the United Sates (US). The US is among the countries with the highest numbers and largest increases of obesity. This is the result of different, generally more unhealthy, eating habits and lifestyles (Ng et al., 2014). In the US about 36.5% of the population was obese in 2014, in France this number is roughly one third: 12.5% of the French population was measured obese (OECD, 2016). Both considered Western-European countries, the Dutch population is more alike the French compared to the US. The obesity percentage in 2012 was 12.7% in the Netherlands (Nationaal Kompas Volksgezondheid, 2014). The Dutch number is very similar to the number of France. This research was conducted in the Netherlands and considered mainly Dutch participants. Therefore it is likely to assume a belief in the HTI in the Netherlands and partly assign the opposed intuition to cultural differences.

(34)

34 Another factor that might play a role in the result of the HTI is the education level regarding food consumptions. Results from a study of Kidwell, Hasford and Hardesty (2015) show that consumers have become more mindful of food consumptions and choices. They show that by emotional ability consumers gain control over their food choices. More specifically, they demonstrate how education reduces consumer beliefs in the UTI. In the Netherlands, the Voedingscentrum is one of the main educational initiatives focused on nutrition. This institute educates and advises people in making healthful food choices, what can be marked effective looking at the relatively low obesity rate compared to for example the US. Consumers get more health-conscious and self-motivated in pursuing healthy food

choices because of education. These health-conscious consumers check nutrition labels and generally search for healthier alternatives (Mai and Hoffman, 2012). In other words, health-minded consumers also pursue healthier choices in their purchase behavior. Another study demonstrates that dieters rate food as less tasty when it is considered as less healthy. Healthy food instead is associated with a more favorable taste (Irmak, Vallen and Robinson, 2011). All this indicates that the belief in the UTI thus differs per country and also strongly depends on the educational initiatives and motivations of consumers. This explains that this intuition does not reside in nature but is something that is and can be taught.

Hypothesis 2 predicted that the UTI also had its implications in purchase intentions in a way that unhealthy product information results in a higher likelihood of purchasing. The findings show the opposed result, in the similar direction of the finding of Hypothesis 1. Healthy portrayed cookies are more likely to be bought than unhealthy portrayed cookies. The purchase intention is largely explained by the fact that taste and purchase intention are related and driven by the same mechanism and people prefer the taste of the healthy portrayed cookie. Another explanation can be that healthy labels upgrade the value of the product. Healthy product information provides a motive to buy and comes along with a higher

(35)

35 willingness to pay (Bower, Saadat and Whitten, 2002). A study of Kozup, Creyer and Burton (2003) confirms this and declares that favorable nutritional information leads to more positive attitudes towards the product and purchase intentions. A final factor that could have led to these results can be found in the study of Cranage, Conklin and Lambert (2004). They showed that when nutritional information was present, people have a higher purchase intention than when no nutritional information was present. People made more healthful food choices when they were informed about the nutritional information. This suggests that showing nutrition information already guides people in the direction of thinking through their actions.

Hypotheses 3a and 3b tested the moderating effect of mindset on the relation between portrayed healthiness and taste. Results showed that this relationship is moderated in such a way that compared to people under a concrete mindset, people under an abstract mindset rate a healthy portrayed cookie more favorably than an unhealthy portrayed cookie. A positive effect was found for healthiness: the more abstractly people are primed, the higher they rate the taste. This was conform Hypothesis 3a and is a contribution to the literature. The results show that it indeed matters in which manner is being thought, rather than what was being thought about (Trope and Liberman, 2000). This elaborates on the findings of Fujita and Han (2009) and shows that not only food choice but also taste is affected by the manipulated mindset. Adding healthy information only reinforced taste perceptions for people under an abstract mindset. People under a concrete mindset did not rate the unhealthy portrayed cookie more favorably than the healthy portrayed cookie. They rated the tastiness rather equally of healthy and unhealthy portrayed food and therefore seem to be indifferent. This means that under an abstract mindset, the taste evaluation is perceived as higher only when healthy product information is portrayed. People who were primed abstractly but received unhealthy information about the cookie rated the taste significantly lower. This is different from what was proposed, so hypotheses 3b was not supported. People under a concrete mindset were

(36)

36 expected to be led by affective factors in the here and now (Trope and Liberman, 2000; Fujita & Han, 2009), but they assessed the flavor of the unhealthy and healthy cookie rather equally. It is possible that no effect was found because the UTI was not supported. Participants did not act according to the UTI. In contrast, people under an abstract mindset acted according to a HTI and perceive healthy food as tastier. Another possible explanation of the unsupported hypothesis can be that an even stronger concrete mindset was needed to support Hypothesis 3b. Even though the manipulation checks showed a significant difference in mindsets and in portrayed healthiness.

Similar results were found for purchase intention. The relation between portrayed healthiness and purchase intention was, as predicted, moderated by the mindset. This happened in the same direction as with taste. Under an abstract mindset, the likelihood of purchasing a healthy portrayed cookie is higher than for an unhealthy portrayed cookie. Hypothesis 4a is supported, however Hypothesis 4b is not supported. It does not work the other way around; people under a concrete mindset do not have a higher likelihood of purchasing unhealthy portrayed food compared to healthy portrayed food. This suggests that people who think at lower construal levels do not necessarily have the impulse of buying unhealthy portrayed food. People primed in higher construal levels are more inclined to buy healthy portrayed food compared to people in lower construal levels. Abstractness results in more conscious deliberation and choosing the healthy option (Fujita & Han, 2009). The effect of purchase intention shows similar patterns as with taste. Therefore, the explanations for taste are also relevant for purchase intention in explaining this difference between the abstract and concrete mindset.

An interesting additional finding is that purchase intention is also educationally based. The higher the level of education, the lower is the likelihood of purchasing under an abstract mindset. A reason can be that information is processed in a different way under this mindset

(37)

37 and education serves as a buffer for purchase behavior. The results showed that higher

portrayed healthiness decreases taste perception for these people. In this study, higher educated people tend to be more critical and are less easily persuaded by the healthiness claims on products. They seem to be more thoughtful and show self-control in purchase situations. Lower educated people associate healthiness with a better taste. It can be that they have a higher belief in the socially desirable intuition that healthy food is good. This is the opposite of the UTI and is partly driven by people’s knowledge about healthy food.

This research contributed to the literature by disconfirming the existence of an UTI in the Netherlands, as an opposite HTI is found. Besides, the research has deepened the CLT more specifically in the food context. The research supports and reconfirms the effectiveness of mindset manipulations. More specifically, how abstraction as a tool can be used in steering people’s food choices and even actual taste experiences. Mindset manipulations were

adequately capable of moderating the relationship between portrayed healthiness and taste and between portrayed healthiness and purchase intentions.

Practical Implications

The contributions for practice are two-fold. Firstly, these findings are relevant in the battle against overweight and obesity. People have already learned that healthy portrayed food is not less tasty than unhealthy portrayed food, which is a positive development. But still, the global battle against obesity is not yet won. Abstraction as a tool might help policymakers in health campaigns by effectively stimulating people to forego healthy food choices and thus pursue healthy lifestyles.

Secondly, for food companies and more specifically marketers, it provides insights in the functioning and priming of consumers’ mindsets that influences their food choices when

(38)

38 taking into account the differences in healthy portrayed information. Despite the fact that companies already make use of healthy labels, the study provides evidence for the importance of the combination of abstract consumer mindsets together with healthy labeling. If

institutions and marketers can prime consumers to higher levels of abstraction, this positively affects taste evaluations of healthy products. Purchase intentions have shown to follow from a positive taste experience. This proposes new opportunities for food marketers and provides more insights in ways they can persuade consumers. Advertising techniques could be adjusted accordingly.

Limitations and Future Research

Like every study, this study has some limitations. Firstly, the research was conducted in the Netherlands and mainly consisted of Dutch participants. A broader cultural diversity of participants could have yielded different results. The UTI was not found in the Netherlands for which one of the reasons is the cultural difference in comparison to the US. It would be interesting to test the UTI in different countries to see which countries and culture support this intuition.

Secondly, the offline experiment was conducted among 112 people. This study only yielded differences in the abstract condition for people who received healthy information. It would also be interesting to test this mechanism on a larger sample with a stronger concrete mindset manipulation. A stronger manipulation might affect taste and purchase intention for the concrete mindset for healthy and unhealthy portrayed food.

Thirdly, the study covered the subject of purchase intention, which was asked after they tasted the cookie. The results could have been influenced by the fact that a positive answer was elicited because they received a free cookie. Also, the purchase intention was not

(39)

39 tested separately from taste but followed directly after the tasting. It might be that the declared purchase intention was the consequence of the tasting. In normal purchasing situations, there is no opportunity to pre-taste the cookie and people have to base their decision on the

packaging and information available. It should also be noted that this purchase intention was a hypothetical scenario and no actual purchase followed from it. Future research is needed to test the construal level mechanism in actual purchasing situations.

Lastly, a deeper understanding of the moderating effect of education levels on the relationship between portrayed healthiness and taste would be interesting to investigate.

The unhealthy = tasty intuition, a construal level perspective