State self-control and food choices:
How can healthy food choices
be stimulated?
A
State self-control and food choices: How can healthy food
choices be stimulated?
State Of Self-Control And The Effect Of Influence Techniques On Healthy Food Choices
University of Groningen Faculty of Economics and Business
Master Thesis Business Administration – Marketing Management August 2013
Yorick van Asselt Postweg 38 8181 VJ Heerde +31653550174
Yorick.van.asselt@gmail.com S1705229
B
Management Summary
Objective: Due to the fact that people are struggling with healthier food choice, this study is aimed at investigating the influence of self-control on food choice. Namely, under a low state of self-control people tend to make their food choice more impulsively. Therefore this research investigates how people can make more healthy food choices with the help of influence techniques on product packaging. This is analyzed by the examination of food choices with no influence techniques, health claims or heuristics under a state of high- and low self-control.
Methods: One hundred thirty six student, with an average of (n) years participated in the experiment. This study used a 2 (high vs. low self-control) x 3 (no influence technique vs. health claims vs. heuristics) design as within subjects factor. The dependent variable in this research is the number of healthy food choices in a food choice task.
Results: This research has found that heuristics positively influence a healthy food choice when an individual does not have the goal to eat healthy. Moreover no evidence is found for the effect of state self-control on food choice. Besides no significant effect of health claims on food choice is found.
C
Preface
With the finalization of my master thesis I have finished my master in Business
Administration – Marketing Management. During my master I have done an internship at one of the biggest Fast Mover Consumer Goods (FMCG) companies in the Netherlands, FrieslandCampina. As an intern in the Flavored Drinks marketing team I have
researched claims of dairy products. I got interested in the effect of those claims and this started me thinking about this kind of subject for my thesis. My mentor Stefanie Salmon told me about her paper and new questions arose in my mind. This resulted in the current research question of this paper.
In this preface I would like to take the opportunity to thank certain people who
supported me during the months of writing my thesis. First I would like to give a special thank to my colleagues Geert Frank, Suzanne van den Eshof and Shirley Lau. They made it possible to combine my thesis writing with an internship and motivated me during the whole process. Furthermore I would like to thank Redmar Roelofs, who helped me as a sparring partner. Moreover my supervisor, Stefanie Salmon, was really important to me for successfully finalizing me thesis. From the begin till the end of the process she gave me useful feedback and support. I would also like to thank Jenny van Doorn for her helpful comments on my work. Finally I would like to give a special thank to my friends and family for their support and relaxation. This is in my opinion essential to keep performing on a high level.
With finalizing my thesis a new episode in my life will start. Due to successful combining my study and internship I got the opportunity to stay longer at my current employer, FrieslandCampina. As a consequence I have to say farewell to a great period in my life, namely the student life. I would like to end with a quote which I picked up during my exchange period in Finland. It helped me with struggles during my study and will
probably still helps me during my career: “All you have to do is know where you're going.
The answers will come to you of their own accord.” – Earl Nightingale.
D
Content
1 Introduction ... 1
1.1 Problem Statement & Research Questions ... 3
1.1.1 Problem Statement ... 3
1.1.2 Research Questions ... 3
1.1.3 Empirical Research Questions... 4
1.2 Academic and managerial relevance ... 4
1.3 Structure of the thesis ... 5
E
4 Results ... 24
4.1 Sample characteristics ... 24
4.1.1 Demographics total sample ... 24
4.1.2 Demographics per condition ... 25
4.2 Manipulation checks ... 26
4.3 Reliability ... 27
4.4 Hypotheses tests ... 28
4.5 Additional Results ... 35
4.5.1 Control variables ... 35
4.5.2 Ego depletion and Trait self-control ... 36
5 Conclusions and recommendations ... 37
5.1 Summary and conclusions ... 37
5.2 Managerial implications and recommendations ... 39
5.3 Limitations and future research ... 40
References 42 Appendices 50 Appendix A: Questionnaire ... 50
Appendix B: Packaging Types ... 56
1
1 Introduction
Suppose you have a busy day at work. A lot of deadlines, your computer is not working well and a lot of miscommunication with your colleagues. At the end of this stressful day you go home by train or car. On the go you decide to buy some food, so you can appease your hunger till dinner. When you enter a petrol store or a on-the-go store at the train station, you stand in front of the shelf. Is it likely that you buy something tasty, even when you normally have the intention to eat healthy? And how do you make this trade-off decision? People are able and willing to make trade-trade-offs among any conflicting attributes. For instance the choice between a higher quality and a lower priced item (Payne et al, 1993). Do you think impulsive or can you afford, after a stressful day, to make a thoughtful food choice?
2 (Fox and Duggan, 2012). So, it can be concluded that society is increasingly occupied
with healthy food consumption.
However, nearby 43% of the Dutch consumers consider eating more healthy as difficult to achieve (Paulussen & Temminghoff, 2010). Research supports this by stating that especially in situations where individuals have low self-control, consumers are unable to make well-considered, balanced decisions when they are standing in front of the shelf (Bargh, 2002; Wansink & Sobal, 2007). Namely, under conditions of low self-control, consumers are more likely to use impulsive decision making strategies, such as salient cues or heuristics (Fennis et al., 2009). Those cues are important tools for marketers in the communication on package. This affects judgments of consumers about the product’s attributes such as quality and healthiness. Those cues can be applied in terms of color, package size, visuals and text (Underwood et al, 2002; Shiv et al., 2005). In obesogenic food environments people are less able to resists temptations due to those cues and this leads to unhealthy food choices(Baumeister et al., 1998). For instance, consumer behavior at on-the-go stores at train- and petrol stations is driven by impulsive and thoughtless decision making. In these situations people are not able to exert self-control. Therefore no well-considered decisions are made and people are more likely to buy tasty products. Hence it is interesting to research if it is possible to influence consumers’ thoughtless food choices towards a more healthy direction.
Therefore this research examines the possibility to make use of low state self-control, by using those cues for healthy products. Those cues can be used as influence technique in order to stimulate healthy food choices. In this research two influence techniques are researched under a state of high and low self-control.
3 on package, which offers the consumer information about the product (Gigerenzer et al.,
1999). For instance this information could be based on the scarcity principle where opportunities seem more valuable to us when they are less available or might be lost. People tend to think that things that are difficult to get are typically better than those that are easy to get. (Cialdini, 2001).
Due to the fact that people are struggling with a lack of self-control in eating more healthy, it is interesting to gain more knowledge about the effect of those influence techniques in a state of high and low self-control. Therefore this research examines more specifically how in a trade-off situation an influence technique is able to let individuals make a more healthy food choice. In those situations people have to value conflicting attributes in order to make a decision. For instance tastiness and healthiness of a product. Resolving those sort of conflicts are a primary source of depletions and leads to low state self-control (Payne et al., 1993, Wang et al., 2010). In those situations it is interesting to know if a sophisticated heuristic comply to stimulate a healthy food choice or is it prudent to use a health claim? This study compares the effects of those influence techniques on food choice of individuals in conditions of high state self-control versus low state self-self-control.
1.1 Problem Statement & Research Questions
1.1.1 Problem Statement
The aim of this research is to study how influence techniques are effective in influencing food choice under high and low self-control. In order to research how individuals can be influenced to make healthy food choices, the following problem statement is formulated:
What is the effect of self-control on food choice, and what is the effect of influence techniques on food choice under conditions of low versus high self-control?
1.1.2 Research Questions
In order to answer the problem statement, several research questions are formulated. These questions are related to the different aspects mentioned in the problem statement, so it can be translated more specifically. The following questions are examined:
4
2) What is a health claim? 3) What is a heuristic?
1.1.3 Empirical Research Questions
4) What is the effect of self-control on food choice?
5) What is the effect of health claims and heuristics on food choice under conditions of high- and low self-control?
1.2 Academic and managerial relevance
A consumers’ state of self-control varies over time (Baumeister et al., 1998). Some studies promote the exertion of self-control in order to make healthier choices, while others argue that this may not be the best option and claim that impulsive and automatic choices could stimulate healthier choices (Herman & Polivy, 2011; Salmon et al., 2012). Therefore this research contributes to the need of more insight of the role of self-control with regard to healthier food choices.
Besides a study of Hill (2009) showed that only 1 in 20 dieters are successfully in maintaining weight loss. This indicates that a strict and well-balanced regulation of the self is not a winning strategy in order to eat healthier. Therefore it may be easier to change a consumers’ environment instead of changing their mind (Wansink, 2006). One of these cues in a consumers’ environment are packaging cues which can be used as an influence technique. In this research two of those influence techniques, health claims and heuristics, are examined under high and low state self-control.
5 making process and that health claims could positively influence a healthy food choice
for individuals with high state self-control.
Besides this study examines another influence technique which is used on product packaging, namely heuristics. A heuristic is a simple decision rule which is applicable in situations where individuals make their choices thoughtless. In literature it is stated that heuristics especially have effect when people make their decisions impulsively (Gingerenzer & Gaissmaier, 2011; Shah & Oppenheimer, 2008). Research states that under conditions of low self-control, people tend to employ strategies such as heuristics. This research complements to the existing literature in the need for more information about mechanisms involved in food choice. (Fennis et al., 2009; Steptoe et al., 1995).
Furthermore this research is managerial relevant with regard to the aim of organizations to serve the consumer as best as they can. Communication can be more effective, when a company has more knowledge about the consumer. As a consequence, the company may obtain a competitive advantage and increase profitability (Vynke, 2002; Wedel and Kamakura, 2003). This research contributes to a better understanding of the marketing communications for food products. Companies will be able to improve their nutrition communication and improve their health positioning.
Moreover this research can be valuable for society with regard to the state of health. It gives a better insight in the food choice process. Besides the outcomes can make it possible to influence consumers in a healthy food choice. Authorities and companies have a better insight in the effect of influencing techniques under different circumstances. In the end this information constitutes to more healthy food choices.
1.3 Structure of the thesis
6
2 Theoretical Framework
This chapter gives an overview of the variables used for this research and explains those in more detail. This is done by a literature review which is the base of the research. Firstly, the independent variable self-control is explained and its relationship with the dependent variable of this research, food choice. Secondly influence techniques as a moderating variable, is elaborated. Therefore three levels of this variable are explained, namely the presence of health claims and heuristics. Based on the literature review in this chapter, hypotheses and a conceptual framework are drawn.
2.1 Self-control
7 foods. People get more vulnerable to various substance-abuse and eating disorders
(Tangney et al., 2004). Hence, literature states that under a state of high self-control, individuals make well-balanced, thoughtful decisions and this is related to healthier outcomes. This is in contrary of situations where people with a state of low self-control has to make food choices. In those conditions a individual tends to make thoughtless and impulsive food choices, which is associated with unhealthy behaviors.
Capacity for self-control varies dispositional among individuals. Those who are high in trait self-control are good at controlling and overriding impulses, while people who are low in trait self-control are more likely to act on impulse as a result of their shortage to suppress or override predominant action tendencies (Baumeister & Heatherton, 1996; Vohs et al, 2007; Tagney et al., 2004). A study of Friese & Hofmann (2008) investigated this statement, where participants completed the self-control scale (Tagney et al., 2004). In a second session, The Single Category Implicit Association Test (Karpinski & Steinman, 2006) had to be fulfilled. This tests measures automatic reactions towards, in this case, potato chips. As a result participants with low self-control where more likely to consume potato chips, than participants who are high in trait self-control.
8 Based on the existing literature this research argue that individuals with high state
self-control are more likely to make a healthy food choice than individuals with low state self-control. Food choices of people with a low level of self-control depends more on external cues in the decision environment. These cues frequently lead to unhealthy choices. Those cues are preferred above long-term health benefits, which are more likely to taken into account in food choices of people with high state self-control. Therefore this research states the following hypothesis:
2.1.1.1
Hypothesis I:
9
2.2 Influence techniques
Food packaging has become a powerful communication tool. It can draw attention and creates unique association which contribute to differentiating from competition and creating additional value in the consumer’s mind. Besides it reaches consumers at the time of purchase (Young, 2004; Chandon, 2012). Except from mandatory components such as size information and nutrition information, the elements and information on package are chosen by marketers (Kiesel et al., 2011). Two of those influence techniques are examined in this research, namely health claims and heuristics. The effect of those influence techniques is researched in high- and low state self-control. Hereby more insight will be obtained about how individuals could be influenced to make a more healthy food choice.
2.2.1 Health claims
Recently, products that claim to improve consumers’ health have become very popular (Urala, Schutz, & Spinks, 2011). Since the Nutrition Labeling and Education Act (NLEA) by the Food and Drug Administration (FDA) was passed in the United States in 1990 there is a growing health trend in marketing (Andrews et al., 1998). Health claims can be seen as a legitimate educational tool, which may affect consumer preference. Besides, communicating health claims could influence consumer behavior and support well-informed food choices. (Patch et al., 2004; Parker, 2003). Namely, nutrient and health claims convey information about food characteristics and health-related food benefits, that might otherwise remain unknown to consumers (Leathwood et al., 2007; Van Trijp et al., 2007). The use of health claims about food characteristics concerns a wide variety. It includes all health claims concerning the elimination and reduction of ingredients, but also added ingredients. Examples of divergent claims on package are ‘100% natural’, ‘les sugar’, ‘low in calories’, ‘extra calcium’ and 0% fat (Keller et al, 1997; Wiliams, 2005; van Kleef et al., 2005; van Osselaer et al., 2001).
10 nutrient may lead to a halo effect. This is also called the overgeneralization effect, where
consumers are tended to generalize the positive perception to other nutrient levels, which are not included in the claim. In addition, the so-called “magic bullet effect” makes consumers ascribe inappropriate health benefits to the product (Roe et al., 1999). Furthermore, several studies have showed that a product with a health claim increase consumers’ expectations about the healthiness of a product. Through a health claim consumers tend to produce more positive attitudes toward the nutritional value of a product in comparison with an option without a health claim (Garretson & Burton, 2000; Tuorila & Cardello, 2002; Kozup et al., 2003; van Trijp & van der Lans, 2007). This can result in a ‘halo’ effect, affecting attitude about nutritional attributes unrelated to the health claim (Ford et al., 1996).
11 benefits, which needs more effort to process in a depleted condition. People in depleted
condition are less able to resists temptations and short-term benefits. As a consequence people with low state self-control are not able to make well-balanced decisions. In contrast to individuals in a depleted conditions, people with high state self-control the reflective system is more active. Thereby long-term goals are more accessible in the decision making process. Those differences results in the phenomenon that the effectiveness of health claims is likely to change in different conditions of state self-control.
To examine the effect of a health claim under conditions of high versus low self-control, it is important to take in to consideration how individuals process information. Literature has stated that a two-system model explains social behavior as a joint function of reflective and impulsive processes. The reflective system produces behavioral decisions, which are based on knowledge about facts and values and probability of potential consequences. Besides the impulsive system of an individual’s provokes behavior which is based on associative links and motivational orientations (Strack and Deutsch, 2004). Both systems operate in parallel. However, there is an asymmetry such that the impulsive system is a continuous process, while the reflective system may be disengaged. The reflective system requires a high amount of cognitive capacity, while the impulsive system requires little cognitive capacity (Strack and Deutsch, 2004).
12 in to consideration when making a well-balanced decision. As mentioned before
literature has shown that these claims have a positive effect on purchase intention. Therefore it is likely that when individuals in a state of high self-control are more likely to make a healthy food choice than when there is no influence technique. Based on these findings the following hypothesis is composed:
2.2.1.1
Hypothesis II:
Individuals with high state self-control are more likely to make a healthy food choice in the presence of a health claim compared to when there is no influence technique.
Besides food choices of people with high state of self-control, the following section explains the presence of health claims in food choices of individuals with a low state of self-control. An important question is how health claims have an effect during fast and frugal decisions. Do consumers take a health claim into account while they make a decision? The lexicographic decision offers an answer to this question. This principle explains how the presence of a health claim plays a role in food choice in individuals with low state self-control.
13 mechanism is mainly used in situations where information is costly, time is pressing or
the consequences of making mistakes are not heavy (Gigerenzer et al., 1999).
This statement of Gigerenzer et al. ( 1999) contributes to the possibility that health claims could have an effect during fast and frugal food decisions. Namely, if individuals have the intention to eat healthy, it is likely that health is one of the most important determinant in decision making. In those situations a health claim can serve as a cue, which makes the consumer aware about the health attribute of a specific product. As a consequence the consumer can be triggered by this health claim. Hence the health claim is processed in the impulsive model where judgments are based on various simplified procedures. In these situations the health claim works as a heuristic and therefore this research hypothesizes that a health claim can work as a heuristic, when people have the intention to eat healthy.
2.2.1.2
Hypothesis III:
14
2.2.2 Heuristics
In contrast to health claims, heuristics influence individuals in a more subtle manner. This influencing technique focuses on impulsive and thoughtless decision making of consumers. Before analyzing the effect of heuristics, the definition of the concept heuristic is given.
Research defines heuristics as efficient cognitive processes that ignore part of the information, with the goal of making decisions more quickly and frugally based on only a few important pieces of information (Gigerenzer & Gaissmaier, 2011l Scheibehenne et al., 2007). Shah and Oppenheimer (2008) listed five aspects which explains why heuristics rely on effort reduction, namely: examining fewer cues, reducing effort of retrieving cue values, simplifying the weighting of cues, integrating less information, and examining fewer alternatives. Basically, heuristic is the umbrella name of the construct of various heuristics, which are based on three common building blocks, namely:
1. Search rules specify in what direction the search extends in the search space 2. Stopping rules specify when enough information has been found so that search
can be stopped
3. Decision rules specify how the information found should be processed to make a final decision (Gigerenzer et al, 1999).
People make decisions using fast en frugal heuristics that are tailored to particular decision environments. These heuristics are fast, because they do not involve much computation. Besides they are frugal because people only search for limited information and one or a few cues in the decision making environment (Gigerenzer et al., 1999). In order to make a decision it is stated that individuals first examine the structure of the information available in the specific environment. This is followed by the choice for a simple decision heuristic that will process the limited information to come up with a decision for action (Gigerenzer et al., 1999).
15 Humans are able to make good decisions by using computationally realistic shortcuts, or
heuristics, that are well matched to the structure of information in particular task environments. In modern society the variety in supply of food is still rising, while people make nearby 200 food decisions a day (Schwartz, 2004; Wansink, 2006). This is a lot of information people have to process and therefore heuristics offer a satisfying solution. Namely, literature stated that food choice decisions are often based on heuristics that facilitate in making a satisfactory choice with minimum effort (Bröder, 2003; Gingerenzer and Goldstein; Roering et al., 1986). However, Scheibehenne et al. (2007) argue that theories about food choices should not be limited to a single model. There are several heuristics mentioned in literature, which could be relevant for the food decision making process.
For instance, there is the principle of reciprocity, which is a heuristic where a consumer feels the need to return a favor. Secondly, the principle of scarcity contains the need to acquire what is less available. Moreover, the authority heuristic implies the tendency of people to follow the recommendation of an expert. People think that if an expert said so, it must be true (Cialdini, 2001). For instance this is the case in advertisements where an expert is the endorser. Such advertisement could be in a television commercial, but also on the packaging of a product. Furthermore, a heuristic which is applicable for an individual in the consideration set formation is the recognition heuristic which quickly reduces the choice set of a consumer. This heuristic increases the probability that a product will be a considered for purchase by priming a familiar brand (Marewski et al., 2010; Coates et al., 2004). Another heuristic is the principle of social proof is examined, where consumers have the tendency to adopt the option which is preferred by others (Cialdini, 2009; Lun et al., 2007) This heuristic is based on the influence of one’s social context. For instance, people will be influenced by the information of what majority has chosen.
16 Based on the literature this research argues that individuals with high state self-control
have more capacity to make a well-balanced decision. Hereby they can compute more information in the environment. Those cues are less effective, because consumers do not use an impulsive decision making strategy and also take into account other information in the decision making environment . In contrast to high state self-control people with low state self-control are more likely to base their decisions in an impulsive manner. Heuristic cues offer the possibility to make decisions with less information, a minimum cognitive effort and time. Therefore it is hypothesized that individuals under low self-control can be influenced by heuristics in order to make a healthy food choice.
In contrast to individuals with low state self-control, the reflective system of people with high state self-control is more active. They are able to take long-term benefits into account and make more well-balanced food choices. Those individuals are less likely to make decisions in an impulsive manner. Therefore it is hypothesized that the presence of a heuristic related to a healthy option, does not have a significant effect on the food choice of individuals with high state self-control.
2.2.2.1
Hypothesis IV:
Individuals with a low state of self-control are more likely to make a healthy food choice when a heuristic is related to a healthy option, compared to when there is no influence technique.
2.2.2.2
Hypothesis V:
17
2.2.3 Conceptual Model
Based on the theoretical framework of this research a conceptual model is drawn. In this model it is shown that the main effect of this research is the effect of individuals’ state of self-control on food choice. Besides there is looked for an significant effects of influence techniques on the relationship of high- and low self-control and food choice.
18
3 Methodology
This chapter describes the method used for examining the hypotheses of this research. First, the study type, acquiring participants and the research design is explained. Furthermore the way in which the variables are measured is specified. In the latter two section the procedure of participating and the plan for analyzing the results is described.
3.1 Study type
This research is defined as a formal study, which is based on a research question and hypothesis. The aim of this research is to test the hypothesis and answer the research question. Based on the hypothesis a causal relationship among variables is explained, namely among independent variables, dependent variables and moderating variables (Cooper & Schindler, 2006).
3.2 Participants
Demographic variables such as age, gender and educational level are not expected to determine if someone do their groceries. In society every adult person buys groceries. Therefore there is not a specific target group set for this survey. In order to get a representative sample size per condition there are at least 20 respondents required per condition. So for this research at least 120 respondents are required.
3.3 Design
For this research, a 2 (high self-control – low self-control) by 3 (no influence technique – health claim - heuristic) design is used. The matrix below shows the different conditions:
No influence technique Health claim Heuristic
High self-control Condition 1 Condition 2 Condition 3
Low self-control Condition 4 Condition 4 Condition 6
Table 1
19
3.4 Variables
This section describes the manipulation and measurement of the variables in this research. The independent variable (self-control) and dependent variable (food choice) are explained. Besides the moderating variables, health claims and heuristics, are described.
3.4.1 Independent variable
In this research the condition of self-control is used as independent variable. Two levels are defined: high self-control and low self-control. The level of self-control will be manipulated by a ego depleting task adopted from Baumeister et al. (1998). Due to this task it is possible to determine the effect of influencing techniques on food choice under state of high self-control and a state of low self-control. Participants are asked to complete a regulatory-depletion task. This task is based on studies of Baumeister et al (1998) and Hagger et al. (2010), which showed to be an established and validated task to influence self-control.
There are two versions of the task, where in both versions task each participant is requested to retype a text. In order to manipulate a state of high self-control the only thing a respondent has to do is retyping the text. This is presumed as a relatively easy task. However the task to manipulate the state of self-control of an respondent to be low, is way more difficult. The respondent has to retype the text, but have to stick to two rules, namely:
Don’t use the blank button
Don’t retype any the letter ‘e’
20
3.4.2 Dependent variable
The dependent variable in this research is food choice, which is measured by a trade-off decision between two products. One product is assumed to be healthy and the other one as unhealthy. In this research it is hypothesized that the independent variable, self-control, has an influence on food choice. So it is researched if there is an significant difference between respondents with a high state of control and a low state of self-control. In the next sections the manipulation of product packaging is described. This hypotheses are also measured by a significant difference in food choice.
3.4.3 Moderating variables
In this research two influence techniques are examined as moderating variable in the relationship between self-control and food choice. The effect of health claims and heuristics is examined. In order to test this effect, consumers are shown a trade-off situation where they have to indicate which product they would chose at that moment. The two products which are shown are comparable, but one of them is assumed to be more healthy. The following table shows the six trade-offs:
Overview product pairs:
Option 1: Option 2: Healthy option
Chocolate bar Cereal biscuit
Crisps Rice crackers with peanuts Ribble crisps Mixed and raisins
Cola Orange juice
Chocolate drink Drink yoghurt
Brownie pudding Yoghurt
Table 2
21
3.4.3.1
Health claims
The first moderating variable in this research is the effect of a health claim on product packages. The manipulation is performed by adding a health claim on product packages in two conditions
A manipulation check is performed by questions about the attractiveness, tastiness and healthiness of the products. The answers about products with health claim are checked with the packages without health claim. By showing a significant difference between those product packages the validity of the manipulation could be checked. The formulation of the health claim is based on the definition of van Trijp et al. (2007), where health claims inform consumers about the food characteristics and health benefits of the product. In the following table the health claims per product are shown:
Overview health claims
Product Health claim
Cereal biscuit With important dietary fiber Rice crackers with peanuts Contains iron en omega 3 Mixed nuts and raisins >50% less fat
Orange juice Rich of vitamine C
Drink yoghurt Contains important vitamins for your immune system
Yoghurt 0% fat
Table 3
3.4.3.2
Heuristics
22 which are already used for products which you can buy in the supermarket. Therefore
the aim of this research will not be obvious for the respondents. In table 4 the different heuristics are shown which are used in this research.
Overview heuristics
Product Heuristic
Cereal biscuit Snack of 2012
Rice crackers with peanuts Most sold among students Mixed nuts and raisins Recommended by Jamie Oliver Orange juice Economy package
Drink yoghurt Value package
Yoghurt Seasonal variation
Temporary available Table 4
3.4.4 Control variables
In order to check if there are any differences in which state and condition of participants complete the questionnaire, control variables are measured. These control variables are hunger, affect and healthy eating goal.
Affect is measured with one item: ‘How do you feel at the moment?’. This is measured on a 7 point Likert scale ranging from 1 (very negative) to 7 (very positive)
Hunger is measured with one item, namely: ‘How hungry do you feel at this moment?’ Healthy eating goal is measured with one item: ‘To what extent do you have the goal to eat healthy?’. The latter two items are measured on a 7 point Likert scale, ranging from 1 (not at all) to 7 (very much).
3.4.5 Other variables
23 are for ego depletion. By asking this it is possible to check if there are significant
differences among the respondents and conditions. Besides questions are asked about the trait self-control among differences. Namely, this can also have an effect on the outcomes of this research.
3.5 Procedure
The questionnaire is composed by the help of the online application Qualtrics. Participants got an online invitation by e-mail or via an hyperlink. The distribution of the questionnaire was randomized. So participants are randomly assigned to a certain condition and therefore different version of the retype task and packaging for the trade-off choices.
3.6 Plan of analysis
24
4 Results
In this chapter several analyses are executed in order to describe the main results. First, the demographics of the total sample size and per condition are summarized. Furthermore, the manipulations and the reliability of the questions are checked. After conducting those analyses, the stated hypotheses of this research are tested. Those results are the base for answering the research question of this paper.
4.1 Sample characteristics
This section describes the demographics of the total sample and also per condition. With the help of the online application, Qualtrics.com, data is gathered of 136 respondents. The participants are obtained by the snowball-technique. This implies that respondents are first acquired from the researcher’s close family and friends, who where then asked to forward the questionnaire to their family and friends. This resulted in originally 139 respondents who completed the questionnaire. However 3 respondents did not complete the ‘self-control manipulation’ task or all food choices. As a consequence those data is deleted from the dataset. So this resulted in the total number of 136 (n=136) respondents.
4.1.1 Demographics total sample
25
Graph 1: Age distribution Graph 2: Education distribution
4.1.2 Demographics per condition
In order to check if the respondents are equally randomized among the different conditions, a randomization check is done. This contains a check if there are no significant differences in demographic characteristics of participants per condition. The confidence interval for these analysis are set at 95%, which implies that the p-value has to be higher than .05 to conclude that the variable is randomly assigned.
The dependent variable ‘gender’ with ‘condition’ as independent variable, are tested through a chi-squared test. The outcome is not significant: (F(5, 136) = 4,440, p = .488) which implies that the conditions are equally randomized with regard to gender. Moreover, the variables ‘age’ and ‘education level’ are tested by an one-way ANOVA test. The result of the ANOVA for Age (F(5,135) = .212, p = .957) and Education Level (F(5,135) = ,358, p = .876) shows that the respondents are equally randomized in terms of Age and Education Level.
26
4.2 Manipulation checks
A manipulation check is performed in order to check if product pairs are actually a trade-off choice between one unhealthy, tempting food product and one product which is significantly more healthy, but less palatable. In order to measure if the product pairs are considered as trade-offs, the respondents evaluated the products in terms of attractiveness, tastiness and healthiness. A paired-samples T-test is performed to check if the products differ significantly on the variables ‘attractive’, ‘tasty’ and ‘healthy’. The results of this test are shown in the table 5.
Manipulation check: means product pairs
Product pair Attractive Tasty Healthy Trade-off pair
Chocolate bar Cereal cookie 5.265 4.191* 5.552 4.493* 1.757 5.022* Yes Crisps
Rice crackers with peanuts
5.132 3.103* 5.250 3.743* 2.103 4.596* Yes Crisps
Mixed nuts and raisins
4.000 4.728* 4.919 4.191* 1.941 3.221* Yes Cola Orange Juice 4.824 4.669 4.728 5.103* 1.824 5.132* No Chocolate drink Drink yoghurt 4.596 5.103* 4.654 5.419* 2.706 5.081* No Pudding Yoghurt 4.904 5.059 4.713 5.022 1.934 5.015* No Table 5
* Significant difference in means
27 deviations and t-test scores of the perceived attractiveness, tastiness and healthiness are
shown.
The other product pairs in table 5 differed significantly in terms of healthiness. However the product pairs are not considered as a trade-off choice because of the outcomes on attractiveness and tastiness. Namely, yoghurt drink, was evaluated significantly more attractive and tasty than the chocolate drink. This is in contrast to the expectations, because the chocolate drink is positioned more as an indulgence product than the yoghurt drink. Also Orange Juice is evaluated as more tasty than cola, while there was no significant difference for attractiveness. Moreover for the pudding and yoghurt pair no significant difference is shown in terms of attractiveness and tastiness. Though the trade off pairs are used for testing the hypotheses of this research.
4.3 Reliability
In the questionnaire, a construct of several questions is used to measure the variables ‘depletion sensitivity’ and ‘trait self-control’. In order to check if the respondents have given the same answers, Cronbach’s alpha is calculated. The summated scales of this measure has to be at least 0.60 in order to be reliable (Hair et al., 2010).
Cronbach’s Alpha variables
Construct Questions Cronbach’s Alpha
Depletion sensitivity 11 .855
Trait self-control 13 .752
Table 6
The Cronbach’s alpha of 11 questions about ‘Depletion sensitivity’ in the questionnaire scored .855. After using the ‘Recode into same variable’ function in SPSS, the Cronbach’s Alpha of trait self-control had a score of .752 for the 13 items of the trait self-control construct.
28 depletion sensitivity is (F(5, 135) = .847, p= .519) and trait-self-control shows (F(5,135)
= .623 p = .683).
4.4 Hypotheses tests
This section tests the hypotheses of this research and helps answering the research question. A between-subjects ANOVA is performed to check if the hypotheses in this research can be accepted or rejected.
In table 7 the means and standard deviation are shown of both conditions. The average amount of healthy food choices for trade off product pairs was 1.022 (SD=0.91) on a scale ranging from one to three healthy choices. The table shows that the means of people in high state self-control is lower than those with low state self-control. Besides it becomes clear that the means for individuals with high state self-control and no influence technique are higher than those with low state self-control. Furthermore the standard deviations and sample size per condition are shown for the tradeoff pairs in this research. These numbers give a first indication of the results for the trade off pairs in this research.
Descriptives: trade off pairs
29 Table 7 shows that, for individuals with high self-control, the means of products with an
influence technique are higher. This is consistent with the hypotheses stated in this research. Furthermore the low mean of products with a health claim for people with low self-control is not in line with the stated hypotheses.
In the between-subject analysis food choice is selected as dependent variable. Self-control and influence technique are included as between subject factors. The confidence interval is set at 95%, which implies that p-value < .05 has to be fulfilled, in order to be significant. Self-control consists of 2 levels, namely: ‘1 = low self-control’ and ‘2 = high self-control’. Furthermore influence technique consists of 3 levels: ‘0= no influence technique’, ‘1 = health claim’, and ‘2 = heuristic’. Hunger is added as a covariate, because this variable differs between conditions. Besides the goal to eat healthy is added as a covariate, because this variable is included in hypothesis 3. The variable food choice for trade off product pairs consists of the sum of the healthy food choices in a range of three trade-off decisions. The following table shows the outcomes of the between-subject ANOVA analysis for the trade-off pairs:
Between-Subjects ANOVA : Trade-off pairs
Variable Hypothesis B t P-value
Constant 3.245 4.605 .000
Hunger .682 .967 .327
Goal to eat healthy 17.352 24.618 .000
Self-control H1 .502 .712 .400
Influence technique 1.422 2.018 .137
Self –Control * Influence technique
H2, H3, H4 .381 .540 .584
Table 8
30 outcomes when ‘goal to eat healthy’ is excluded as covariate. This is also the case for the
between-subject analysis with all product pairs.
Between-Subject ANOVA trade-off pairs – without goal to eat healthy as covariate
Variable Hypothesis B F P-value
Constant 43.165 51.764 .000
Hunger 2.045 2.440 .121
Self-control H1 .416 .499 .481
Influence technique 2.006 1.203 .304
Self –Control * Influence technique
H2, H3, H4 .414 .497 .610
Table 9
Furthermore a split analysis for self-control is performed. Hereby the difference between high versus low self-control with regard to the effect of influence techniques is analyzed. Table 14 shows the relevant outcomes, which reveals no significant effects.
Between-Subjects ANOVA with split file analysis for self-control
Variable B F P-value
High self-control Influence technique .385 .523 .596 Low self-control Influence technique 1.432 1.575 .215 Table 10
31 Respondents divided by healthy eating goal
Variable
condition
N
No healthy eating goal Self-control High Low 42 40 Influence technique No influence technique Health claim Heuristic 25 31 26 Healthy eating goal Self-control High Low 23 31 Influence technique No influence technique Health claim Heuristic 19 17 18 Table 11Table 11 shows that almost every condition has enough respondents to find significant effects which are valid. Only the levels of influence techniques for people with a healthy eating goal do not have 20 respondents or more. Therefore the validity of any results with less than 20 respondents per condition is taken into account in the result section.
Between-Subject ANOVA with Median Split – healthy eating goal
Variable B F P-value No Healthy eating Goal Constant .822 1.467 .201 Hunger .642 1.145 .288 Self-control .309 .551 .460 Influence technique* 1.778 3.172 .048*
Self –Control * Influence technique .050 .089 .915
Healthy Eating Goal Constant .842 .978 .451 Hunger .749 .870 .356 Self-control .120 .140 .710 Influence technique .063 .073 .930
Self –Control * Influence technique 2.045 2.376 .104 Table 12
32 Due to the significant result of the effect of influence technique on food choice for people
with no healthy eating goal, a post-hoc analysis is performed. Moreover, a Tukey HSD analysis is chosen to determine significant differences between the different levels of influence technique.
Post-hoc analysis for individuals with no healthy eating goal
Levels of influence technique Std. Dev. Sig.
No influence technique vs. health claim .201 .213 No influence technique vs. heuristic .210 .025*
Health claim vs. heuristic .200 .516
Table 13
* = significant
33
Hypothesis 1: Influence of self-control on food choice
.In hypothesis 1 it is stated that individuals with high state self-control are more likely to make a healthy food choice than individuals under low state of self-control. The ANOVA analysis for all product pairs shows no main effect of state of self-control on food choice. This implies that high self-control (M=3.169) does not lead to more healthy food choices in comparison with low self-control (M=3.070), (F(1, 135)= .055, p = .816). Besides the tradeoff product pairs do also not show a significant difference between a high and low state of self-control, F(1, 135)= .499, p = ,481). Based on this study it can be concluded that there is no significant relationship between healthy food choice and an individual’s state of self-control. So:
Hypotheses 1 has to be rejected.
4.4.1.1
Hypothesis 2
The second hypothesis of this research states that individuals with a high state of self-control are more likely to make a healthy food choice with the presence of a health claim compared to when there is no influence technique. In the ANOVA analysis there is no significant main effect in healthy food choices among individuals in a high or low state of self-control and with the presence or absence of influence techniques, (F(1, 135)= .502, p = .606). Also the tradeoff product pairs does not show a significant main effect(F(2, 135= .497, p = .610).
Moreover, a split analysis is performed to test the effect of influence technique for individuals with high state self-control and in conditions where people have low state self-control. Both conditions (high versus low self-control) do not show a significant effect. Namely the outcomes of high state self-control show (F(2, 135)= .523, p = .596), while the results for low state self-control show (F(2, 135)= .1575, p = .215).
Hypothesis 2 has to be rejected.
4.4.1.2
Hypothesis 3
34 choice with the presence of a health claim related to the healthy option, compared to
when there is no influence technique.
The ANOVA analysis for all product pairs shows that there is a significant influence of the healthy eating goal as a covariate in the model, (F(1, 135)= 34.287, p = .000). Besides, this is also the case for the tradeoff pairs: (F(1, 135)= 24,618, P = .000). Due to the fact that p < .05 it can be concluded that the goal to eat healthy significantly influences a health food choice. Namely, if an individual’s goal to eat healthy is higher, it is more likely that he or she will make a more healthy food choice. However there is no significant relationship in interaction effect of self-control and influence technique on food choice. All product pairs show (F(2, 135)= .463, p = .630), while also in the ANOVA analysis for tradeoff pairs shows no significant effect (F(2, 135)= .540, p = .584).
The ANOVA analysis in this research showed a significant effect of healthy eating goal as covariate. Therefore a media split of ‘goal to eat healthy’ is performed to check the differences between people who have the goal to eat healthy and individuals who do not have this goal. For individuals who have the goal to eat healthy no significant interaction of state self-control and influence technique is found (F(2, 54)= .2.054, p = .139).
Hypothesis 3 has to be rejected.
4.4.1.3
Hypothesis 4 and 5
35 Namely, based on the results there is no significant outcome possible for the post-hoc
analysis. This leads to the following conclusion:
Hypothesis 4 has to be rejected. Hypothesis 5 is accepted.
Moreover a split analysis is performed where a difference is made between people with the goal to eat healthy and people for who this goal is less important. This between-subjects ANOVA shows that influence technique has a significant effect on individuals food choice who do not have the goal to eat healthy (F(2, 135)= 3.172, p = .048). Due to this result a post-hoc analysis is performed to find significant differences between the three levels of influence technique. This analysis shows a significant difference between the condition where no influence technique is present and the condition where a heuristic is shown, because p-value = .025. Due to the fact that p-value < .05 is can be concluded that heuristics are effective in influencing individuals food choice when a person do not have the goal to eat healthy.
4.5 Additional Results
In the additional results of this research the influence of the covariates ‘Goal to eat healthy’ and ‘Trait self-control’ are described. Due to the fact that there is no main effect of state of self-control on a healthy food choice, additional analysis with regard to trait self-control is performed. Namely, it is likely that this variable could influence individuals’ food choice.
4.5.1 Control variables
As mentioned before, ‘goal to eat healthy’ positively influences a healthy food choice, which contains that if someone has the goal to eat healthy, this person is significantly more likely to make a healthy food choice.
36 (F(1, 135)= .967, p = .327). Because it did not show a significant effect, there is no
significant difference in outcomes among the condition due to the differences in hunger.
Furthermore the influence the control variables ‘depletion sensitivity’ and ‘trait self-control’ are examined. These variables are included in the Between-Subjects ANOVA analysis as covariates.
The depletion sensitivity did not show a significant influence on a health food choice for all the product pairs (F(1,135)= .074, p = .786). Moreover, there is also no significant influence for the tradeoff pairs (F(1,135)= .954, p = .331). Hence it can be concluded that when an individual is more sensitive for ego depletion, the chance does not become bigger that he or she makes a more unhealthy food choice.
The effect of trait self-control as a covariate in the model showed no significant result for all product pairs (F(1,135)= 2.095, p = .150). However a significant effect for the tradeoff product pairs is showed (F(1, 135)= 5.719, p = .018). Because p-value < .05, it implies that individuals with a higher trait self-control are more likely to make a healthy food choice for tradeoff pairs.
4.5.2 Ego depletion and Trait self-control
Due to the significant influence of trait self-control, as a covariate, it is interesting to know if there is a direct relationship between trait self-control and a healthy food choice. Therefore an ANOVA is performed, with food choice as dependent variable and trait self-control as independent variable.
The results show no significant relationship between an individual’s trait self-control and a healthy food choice for all product pairs, namely: (31, 135)=1.331, p = 0.144). Besides there is also no significant influence of trait self-control on food choice for the tradeoff pairs (F(31, 135)= 1.484, p = .073). It can be concluded that with a higher trait self-control it is not more likely that an individual makes a healthy food choice.
37 It is also interesting to know if depletion sensitivity has an interaction effect with a state
of self-control. A Between-Subjects ANOVA shows that for all tradeoff pairs there is no significant effect between those variables (F(25, 135)= .800, p = .727). When all product pairs are taken into account for food choice, there is also not an interaction effect (F(25, 135)= .603, p = .919).
5 Conclusions and recommendations
In this final chapter of this research an answer is given to the research question wherefore the hypotheses are stated. Based on the results of this research, conclusion are drawn. Moreover, managerial implications and recommendations are given. Furthermore the limitations of this research and directions for further research are discussed.
5.1 Summary and conclusions
The last years there is a growing trend of people who want to eat healthier, but research indicates that consumers do not succeed. This study is constructed to research the effectiveness of influence techniques under high versus low state self-control. The findings of this study can help finding a solution to let individuals make a more healthy food choice. By distributing a survey, including an ego depleting task and manipulated tradeoff product pairs, possible relationships between variables are analyzed.
38 The second hypothesis of this research is also not supported, which stated that
individuals with a high state of self-control are more likely to make a healthy food choice with the presence of a health claim compared to when there is no influence technique. Also for the third hypothesis in this research no significant relationship is found.
Similar to the outcomes of these hypotheses, a study of Garretson and Burton (2000) concluded that there is no such influence between health claims and food choice. Furthermore the findings of this research are in line with studies which showed insignificant effects of health claims on nutrition- or brand attitudes (Ford et al., 1996; Keller et al. 1997; Mitra et al, 1999). An explanation could be that individuals do not rely on health claims in making nutrition- or brand attitudes. Therefore it could be wise to offer more information in the Nutrition Facts panel, which gives a more truthful and non-misleading view.
The fourth hypothesis stated that individuals with a low state of self-control are more likely to make a healthy food choice when a heuristic is related to a healthy option, compared to when there is no influence technique. In despite of the difference in means, there was no significant relationship found. Besides the fifth hypothesis was confirmed, stating that individuals with high state self-control are not more likely to make a healthy food choice when a heuristic is related to a healthy option, compared to when there is no influence technique. These hypotheses are in line with a study of Scheibehenne et al. (2007) which stated that theories about food choice should not be limited to a single model. They showed that there is no favored model to individual decision making, such as heuristic decision making. In contrast to this statement, the finding of the effect of heuristics on food choice for individuals is in line with studies of Cialdini (2001), Coates et al. (2004) and Marewski et al. (2010). These studies state that heuristics positively influence an individual’s purchase intention. Nevertheless these studies did not make a clear distinction between goal to eat healthy. This study states that heuristics are effective when an individual does not have the goal to eat healthy.
39 those questions. This gives an indication that the health claims and heuristics were not
used too obvious or emphatic.
Furthermore this research found that people, who do not have the goal to eat healthy, were prone to heuristics which are not related to eating healthy. It is likely that they are more sensitive to short-term gratification than people with a healthy eating goal, which is more relevant for the long-term. Though literature did not give an indication about the role of the goal to eat healthy and therefore it is not hypothesized in this research. However, enough respondents belong to this condition. Therefore this is tested as additional result and the significant relationship between heuristics and a healthy food choice for people with no healthy eating goal is found. Moreover, no significant effect was found for people with a healthy eating goal. Besides this condition did not have 20 respondents, but only 18 persons. However the high p-value for this 18 persons indicates that probably no significant effect would be found with 2 more respondents for this conditions. The difference
The difference in effectiveness, based on healthy eating goal, could be complementary to the LEX principle, where information is costly, time is pressing or consequences of making mistakes are not heavy (Bettman, 1979; Gigerenzer et al., 1999). It is likely that people who have the goal to eat healthy, consider the healthiness of a product as most important attribute. Therefore it could be that those persons base their decision on the perceived healthiness of a product and take no heuristics into account in the decision making process when these are not related to health. It could be the case that they already have chosen the information, which is applicable in the decision making process and as a consequence heuristics are not effective. Besides it could be the case that health claims are effective. However in this research healthy versus unhealthy products are tested. In those situations an individual, who already has the goal to eat healthy, could also notice the difference in healthiness without any claims.
5.2 Managerial implications and recommendations
40 the effect of health claims on food choice. Therefore firms should be aware that
contradicting findings concerning the effectiveness of health claims.
This research showed that goal to eat healthy has a significant effect as covariate. Besides the results of the effect of heuristics on food choice are different for individuals with- or without the goal to eat healthy. Therefore this research gives an indication for further research that it is important to take individuals goal to eat healthy into account.
Moreover, the findings about the effect of heuristics are interesting to marketers. Marketers can use communication on package to affect judgments of consumers about their products (Shiv et al., 2005). This research indicates that people, with no goal to eat healthy, are affected by heuristics on package. Therefore marketers can consider to use certain heuristics in the communication on package (Underwood, 2002). Besides those heuristics can use a tool to let people make more healthy food choices, in mindlessly manner. It is even more interesting because these individuals do not have the goal to eat healthy.
5.3 Limitations and future research
A first limitation of this research is the fact that the survey is only spread among Dutch. The hypotheses are based on research in also other countries. However the outcomes of the research could be different in another country. For instance the understanding of a specific claim could be interpreted differently between countries. Namely, research showed differences in interpretation of health claims between countries (Van Trijp & Van der Lans, 2007). Besides there is a difference in legislation for labeling, presentation and advertising on food packages between countries (Leathwood et al., 2007). Therefore future research is needed for different geographical areas.
41 In despite of the fact that there are no indications that the results of this study would be
different, a broader sample would be better.
Another limitation of this research is that the determining factors of food choice could differ between each food category and nutrition-health relationship. A tailored approach is needed when communicating health claims (Ascheman-Witzel and Hamm, 2010). It could be the case that the effect of heuristics and health claims differ per product category. Besides there could be a difference for products which are used on daily base, and products with a more luxury or indulgence proposition. For instance, a health claim could be more powerful for products which are consumed on a daily base. Namely, the long-term impact of daily consumption is greater than an indulgence product which a person consumes once a week. With regard to heuristics it could also be the case that heuristic types differ per product category. An interesting question would be if the recommendation of an expert will have the same effect on dairy drinks as it has for the snacks such as crisps. Future research could examine which type of health claims and heuristic is the most effective for certain product categories.
A fourth limitation of this research is the fact that respondents only had to choose between two products, while in the supermarket a decision is made when a person is standing in front of a shelf. Namely, it could be the case that the effect of a heuristic will reduce, because people tend to make decisions in the supermarket on the basis of affective product features (Bruyneel et al., 2006). However research has shown that subtle heuristics are still capable in influencing consumers choice, both within and beyond the health domain (Fennis et al., 2009). This is an interesting issue for future research to sort out.
42
References
Aschemann-Witzel, J., Hamm, U., (2010. Do consumers prefer Foods with nutrition and health claims?, Results of a purchase simulation, Journal of Marketing Communications, Vol. Vol. 16, p. 47-48.
Baumeister, R. F., & Heatherton, T. F., (1996). Self-regulation failure: An overview,
Psychological Inquiry, Vol. 7, p. 1–15.
Baumeister, R. F., Heatherton, T. F., & Tice, D. M., (1994). Losing control: How and why people fail at self-regulation. San Diego, CA: Academic.
Baumeister, R.F., Bratslavsky, E., Muraven, M., & Tice, D.M., (1998). Ego depletion: Is the active self a limited resource?, Journal of Personality and Social Psychology, Vol. 74, p. 1252- 1265.
Baumeister, R.F., Vohs, K.D., & Tice, D.M., (2007), The strength model of self-control, Current directions in psychological science, Vol. 16(6), p. 351-355.
Bech-Larsen, T., and K.G. Grunert., (2003), The perceived healthiness of functional foods: A conjoint study of Danish, Finnish and American consumers’ perception of functional foods, Appetite, Vol. 40, No. 1, p. 9–14.
Bargh, J. A., (2002). Losing consciousness: Automatic influences on consumer judgement, behavior, and motivation, Journal of Consumer Research, Vol. 29, p. 280-285.
Bettman, J., (1979). An information processing theory of consumer choice”. Reading, MA: Addison-Wesley.
Brown, C. L., & Carpenter, G. S. (2000). Why is the trivial important? A reasons-based account for the effects of trivial attributes on choice. Journal of Consumer Research, 26(4), 372-385.
Burton, S., Andrews J.C., Netemeyer R.G, (2000), Nutrition ad claims and disclosures: Interaction and mediation effects for consumer evaluations of the brand and the ad,
Marketing Letters, Vol. 11, no. 3, p. 235–47.
Bröder, A., (2003), Decision making with the ‘‘adaptive toolbox’’: Influence of environmental structure, intelligence, and working memory load, Journal of
Experimental Psychology: Learning, Memory,& Cognition, Vol. 29, p. 611–625.
Caswell, J.A., Ning, Y., Liu, Y. & Mojduszka, E.M., (2003), The impact of new labelling regulations on the use of voluntary nutrient-content and health claims by food manufacturers, J Public Pol Mark, Vol. 22, 147–158.
