At the core: Metaphors and Preference Consistency

Table Of Contents

Abstract

List Of Tables

List Of Figures

1. Introduction

2. Theoretical Basis
2.1 Emotional Versus Cognitive Decision Making
2.2 The Measure Of Inconsistency
2.3 Priming.
2.4 The Experimental Setup Of Amir, Ariely And Lee
2.4.1 Experiment 1A by Amir, Ariely and Lee
2.4.2 Experiment 1B by Amir, Ariely and Lee
2.4.3 Experiment 2 by Amir, Ariely and Lee
2.4.4 Experiment 3 by Amir, Ariely and Lee
2.4.5 Experiment 4 by Amir, Ariely and Lee
2.5 Metaphors And Their Influence On Human Behavior
2.6 The Unpaired Sample T-Test
2.7 The Wilcoxon-Mann-Whitney-Test (Also Called Mann-Whitney-U-Test)

3. Research Hypothesis

4. Our Experimental Set Up

5. Proband Recruitment And Proband Pool For Marketing Chair

6. Introduction Into Collected Experimental Data

7. Testing For Differences Between Groups
7.1 Test For Difference Between Full And Sliced Group
7.1.1 Test for difference between full and sliced group without outliers
7.1.2 Test for difference between full and sliced group without super consistent probands 33
7.1.3 Test for difference between full and sliced group without outliers and without super consistent probands
7.1.4 Test for difference between full and sliced group considering complete data sets only
7.2 Test For Difference Between Full And Sliced Group Within Interquartile Ranges
7.2.1 Test for difference between full and sliced group within interquartile range of errors
7.2.2 Test for difference between full and sliced group within interquartile range of reading time
7.2.3 Test for difference between full and sliced group within interquartile range of age
7.3 Test For Difference Between Full And Sliced Group Concerning Age
7.3.1 Test for difference between full and sliced group concerning people in the typical age of students
7.3.2 Test for difference between full and sliced group concerning people older than students
7.4 Test For Difference Between Full And Sliced Group Regarding Genders
7.4.1 Test for difference between full and sliced group regarding women
7.4.2 Test for difference between full and sliced group regarding women without super consisten probands
7.4.3 Test for difference between full and sliced group regarding men
7.4.4 Test for difference between full and sliced group regarding men without super consistent probands
7.5 Test For Difference Between Full And Sliced Group Regarding The Daytime Of Data Recording
7.5.1 Test for difference between full and sliced group between nine a.m. and twelve a.m.
7.5.2 Test for difference between full and sliced group between one p.m. and five p.m

8. Modified Experiment
8.1 Set Up Of Modified Experiment
8.2 Descriptives Of Modified Experiment
8.3 Hypothesis Testing

9. Critique
9.1 Critique Regarding The Experimental Set Up Of Amir, Ariely And Lee
9.2 Critique Regarding The Experimental Set Up Of Lee And Schwarz
9.3 Critique Regarding The Evaluation Methodology Of Amir, Ariely And Lee

10. Conclusion

References

Appendix A

Appendix B

Appendix C

Appendix D

Appendix E

Abstract

In 2009 the three academics On Amir, Dan Ariely and Leonard Lee published the article "In Search of Homo Economicus" in the Journal of Consumer Research (JCR). After having conducted five experiments the conclusion of this article was that people who put more trust in their feelings show more consistent preferences than people who base their decisions on analytics and deliberation. Supported was that resulting conclusion of the psychological fundamental that emotions are nothing more than efficient programmes, designed by evolution to ensure species preserving reactions on external influences. The author of the present paper tested the aforementioned tenor critically by setting up two experiments using the previous work by Amir, Ariely and Lee as a basis. In the first experiment 131 probands were tested, in a second slightly modified experiment 26 persons participated in the experiment. After several tests of the collected data material the conclusion of Amir, Ariely and Lee cannot be supported. In fact, data of persons who are assumed to be deliberate thinkers showed lower numbers of errors

Keywords: preference consistency, emotions, decision making

List Of Tables

Table 1: Output Of 30 Paired Comparisons

Table 2: Descriptives Of Proband Pool

Table 3: Descriptives Of All Collected Data.

Table 4: Analysis Of Relation Between Errors Of Full-Fruits Group And Sliced-Fruits Group

Table 5: SPSS Output Of Test Between Groups

Table 6: SPSS Output Of Test Between Groups Without Outliers

Table 7: SPSS Output Of Test Between Groups Without Super Consistent Probands

Table 8: Short Descriptives Of Sliced Group Without Outliers And Without Super Consistent Probands

Table 9: SPSS Output Of Test Between Groups Without Outliers And Without Super Consistent Test Persons

Table 10: SPSS Output Of Test Between Groups With Complete Data Sets Only

Table 11: SPSS Output Of Test Between Groups Within Interquartile Range Of Reading Time

Table 12: SPSS Output Test Between Groups Within Interquartile Range Of Reading Time 36

Table 13: SPSS Output Of Test Between Groups Within Interquartile Range Of Reading

Time And With Complete Data Sets Only

Table 14: SPSS Output Of Test Between Groups With Interquartile Range Of Age

Table 15: SPSS Output Of Test Between Groups Concerning Age Of Students

Table 16: SPSS Output Of Test Between Groups Concerning Age Of Students Without Outliers.

Table 17: SPSS Output Of Test Between Groups Concerning Age Of People Older Than 29 Years

Table 18: SPSS Output For Women Of Full And Sliced Group

Table 19: SPSS Output For Women Of Full And Sliced Group Without Super Consistent Probands

Table 20: SPSS Output Of Full And Sliced Group Regarding Men

Table 21: SPSS Output Of Full And Sliced Group Regarding Men Without Super Consistent Probands

Table 22: SPSS Output For Probands Of Full And Sliced Group Tested Between Nine A.M. And Twelve A.M.

Table 23: SPSS Output For Students Of Full And Sliced Group Tested Between Ten A.M. And One P.M

Table 24: SPSS Output For Probands Of Full And Sliced Group Tested Between One P.M. And Five P.M

Table 25: Descriptives Of Modified Experiment

Table 26: SPSS Output Of Test Between Groups Within Modified Experiment...

Table 27: Errors Of Full-Fruits Group And Sliced-Fruits Group In Modified Experiment

Table 28: List Of Gadgets Considered Best And Worst

Table 29: Calculation Of Sum Of Ranks.

List Of Figures

Figure 1: Geometrical Representation Of Preferences

Figure 2: Used Gadgets To Be Presented To Test Persons

Figure 3: Experimental Process Of One Exemplar Sequence Of Test Run

Figure 4: Process Of One Exemplar Sequence Of Experiment

Figure 5: Age Classes Of Persons Interested In Further Experiments

Figure 6: Boxplot For Ages In Proband Pool

Figure 7: Errors Of Full-Fruits Group And Sliced-Fruits Group

Figure 8: Errors Of The Whole Experiment Converted Into Standard Normal Distribution

Figure 9: Errors Converted Into Exponential Distribution

Figure 10: Sample Evaluation

Figure 11: Errors Of Full-Fruits Group And Sliced-Fruits Group Converted Into Standard Normal Distribution

Figure 12: Histogramm Of Frequency Of Errors Within Full-Fruits Group

Figure 13: Histogramm Of Frequency Of Errors Within Sliced-Fruits Group

Figure 14: Physiological Performance Curve - Relation Between Daytime And Performance Ability

Figure 15: Modified Experiment

Figure 16: Histogramm Of Modified Experiment

Figure 17: Experimental Process Of One Exemplar Sequence Of Test Run (Slides Translated Into English Language)

1. Introduction

The interaction between different fields of research becomes more and more important. The complexity of the world we live in does not allow a restricted view on problems anymore. More holistic approaches tend to be a future way of solving tasks. That is why neuro-science and marketing have begun to merge in recent years. In today's high competitive markets it is not enough to know that people buy goods. In order to achieve at least small advantages towards competitors on the market and consequently reach higher sales it is much more relevant to know why people decide for certain goods and how to influence them in their decision making. But one can also apply a vice ve]rsa view and find out how to help consumers to decide in consistent ways concerning their preferences as well as making consumers less susceptible for advertising tricks and influences on their decision behavior. In this paper a further link between a psychological approach and marketing tools is conducted. This interdisciplinary view on people's decision making will deliver one more element for understanding consumer behavior. Based on the previous work of On Amir, Dan Ariely and Leonard Lee another experimental set up was conducted to explore preference consistency under different primings of experiment participants.

On Amir owns an associate professorship in marketing at University of California in San Diego, California. Amir opened several papers on decision making and consumer behavior to scientific discussion. His latest publication "Can uncertainty improve promotions" appeared in the Journal of Marketing Research in 2010 (Amir 2011). Dan Ariely is James B. Duke professor of behavioral economics at Duke University in Durham, North Carolina. He shows a long record of publications in the fields of individual behavior and decision making (Ariely 2011). The last of the triumvirate of experts is Leonard Lee who is assistant professor in the marketing division at Columbia Business School in New York. After his Ph.D. in marketing in 2006, he co-worked on a number of papers on preferences and publicized two articles each in the Journal of Consumer Research (JCR) and Psychological Science (Lee 2011).

The article "In Search of Homo Economicus: Cognitive Noise and the Role of Emotion in Preference Consistency" by Amir, Ariely and Lee, published in JCR in August 2009, represents the fundamental for this paper. This triad of researchers used four experimental set ups to test their research hypothesis "... that a higher degree of reliance on emotional responses ... will generate a higher level of preference consistency" (Amir, Ariely, Lee 2009).

In order to find out if similar results can be generated and if the conclusion of Amir, Ariely and Lee is reproductable more than 130 test persons were asked to decide between products which were presented in paired comparisons. The paper presented to you will provide statistical findings, contributing to the interdisciplinary field of neuro-marketing and test the hypothesis by Amir, Ariely, and Lee in a critical way and therefore disprove their conclusion that emotions lead to more consistent preferences.

2. Theoretical Basis

In the article "In Search of Homo Economicus ..." Amir, Ariely and Lee measure preference consistency by applying the methodology of series of choices between two goods. In every round of the choice series the tested person has to choose one out of the two goods. During this choosing process mistakes in the preferences of the proband may occur. These mistakes are inconsistencies concerning the preferences of the individual. Mistakes in preferences can appear because of different reasons. Maybe the person is not sufficiently concentrated during the choosing process, maybe some information was transferred incompletely or the individual just does not understand the transmitted information. In the past, such inconsistencies where treated as so called noise and did not find great consideration in scientific evaluations. But Amir, Ariely and Lee took a closer look at the noise that appeared when testing persons for preference consistency. They assume that this noise contains "information about the ability of decision makers to perform good decisions ... " (Amir, Ariely, Lee 2009). The three academics explored, whether emotional decision making might create a higher level of preference consistency and consequently better decisions.

2.1 Emotional Versus Cognitive Decision Making

There is a huge number of psychological models which claim that human decisions are influenced by two subparts of thinking: emotion and cognition. Just to name a few models from psychology, there are e.g. the emotion-cognition dual process-model by Loewenstein and O'Donoghue from 2004, the associative system versus rule-based system by Sloman from 1996 and nonverbal processes versus verbal processes by Paivio from 1986. All these models have in common that they base on the distinction of emotion and cognition. In this context one can state that more emotional decisions are "holistic, affective, concretive and passive" (Amir, Ariely, Lee 2009). The cognitive approach to decisions is characterized by analysis, logic, abstract thinking and activity. Consequently, Amir and his colleagues evaluated if decisions, which are made more on an emotional basis, are more consistent than those decisions that are made cognitively. These emotional decisions are considered to reveal true preferences of the individual, due to evolutionary development of human thinking (Amir, Ariely, Lee 2009, page (p.) 175). The task of emotions is to decide in a fast and correct way, as evolutionary psychology suggests. In the early days of mankind, when people did not have cognitive abilities like nowadays, one had to decide as quick as possible to ensure surviving when facing dangers like predators or enemy tribes, for example. Emotions ensured to escape from threatening situations by evoking fear and hence making one escape from danger, for instance. The same goes for multiplying of people. Love is the most popular emotion people know and this is nothing more than a kind of program that is running in a person's head, caused by evolution to make sure human race produces descendants. Thinking of this argumentation has to lead to the fact that emotional decision making produces more consistent results regarding preferences of individuals than decision making which is based on deliberation. Additionally, Amir et al. (based on Epstein 2003, p. 160) consider emotional decisions to be much more stable over time because they focus on the most relevant aspects of the decision problem while deliberative thinking might change. Cognitive thoughts may change because of new information the individual received, because of a change of thinking, because another decision model was applied or because information causes costs. Subsequently, these changes may lead to inconsistent decision making. As a conclusion from this opinion towards emotions in decision making Amir and his colleagues rise the hypothesis " ... that a greater reliance on emotional responses ... will generate ... higher ... preference consistency." (Amir, Ariely, Lee 2009, p. 175).

2.2 The Measure Of Inconsistency

Based on fundamental economic theory Amir, Ariely and Lee oriented in transitivity. Transitivity says that, if an individual is confronted with a set of goods named A, B and C, then this person must have a structure in his or her preferences when choosing between paired comparisons of these goods. This means that, if A > B and B > C, then A > C. This methodology was introduced by B. Babington Smith and M. G. Kendall in their article "ON THE METHOD OF PAIRED COMPARISONS" (Babington Smith, Kendall 1940). Presumed that this structure holds, if the individual chooses the goods, then there is no mistake in the person's preferences and so there is no inconsistency. But if the choosing person does not follow this restriction, then there is inconsistency in the decision behavior. These mistakings or inconsistencies are counted and summed up for each individual and the result is the so called inconsistency measure.

In their paper Babington Smith and Kendall apply two methods to find and count the number of consistency mistakes:

They use a tabular form to show the output of one person that was asked to choose between six goods which were presented in paired comparisons (Babington Smith, Kendall 1940, p. 325 and 26). Generally, a total number of 2 pairs results from a paired comparison. For six 6 goods A, B, C, D, E and F, as applied in their paper, a number of [illustration not visible in this excerpt] = 30 "preferences of a single observer" (Babington Smith, Kendall 1940, p. 325) has to be checked for inconsistencies. See table one for their used tabulation to show the output of a tested person who was asked to choose between the 30 pairs out of six products.

Table 1: Output Of 30 Paired Comparisons

illustration not visible in this excerpt

(Out of "ON THE METHOD OF PAIRED

COMPARISONS", Babington Smith, Kendall 1940, p. 326)

In this first table one can see the six goods A to F. As Babington Smith and Kendall explained, every good in one of the rows is compared to one good in the column. Good A in the second row (counted from upper row to lower row) is compared to good B in column three (counted from left to right). The number "1" means that A is preferred to B, for example. Hence, when comparing good B in row three, one can find a number "0" in the second column, what means that B is not preferred to A. Another example is the comparison of good E to D: one can see there is a number "1" written in the table and so good E is preferred to good D. In contrast, when comparing D to E there is "0"¸this means that D is not preferred to E. The same procedure works for the entire table. In order to find inconsistencies the experimenter now has to check transitivity orders in the following way: A > B and B > C and logically A > C, resulting in the triad A > B > C. If there appears a mistake in this logic, one mistake can be counted. Just to show one example for a working triad: As one can learn from the table A > B, indicated by the number "1". Then B > C, indicated by "1". Now, one needs to think logically, because to make this triad work, A must be preferred to C and this means in the table there must be written a number "1" in the comparison of A to C. If one now takes a look into the corresponding cell of table 1, one can see there is in fact a number "1" and thus the triad is consistent. For another example for an inconsistent triad, take a look at the comparison between goods B, E and F. B > E and E > F. Consequently, B should be preferred to F and number "1" should be written in the corresponding cell. But there is a "0"! This means that the tested person indicated that he or she prefers F to B. This is an inconsistent choice and means that one can count one mistake in the preference consistency of the tested individual.

Another possibility to illustrate the preferences of a tested individual is to present the results geometrically, as one can learn from figure one. If A is preferred to B (A > B), Babington Smith and Kendall wrote A → B in the figure or vice versa B ← A. The arrow just indicates the direction of the individual's preference regarding the paired comparison between two products.

Figure 1: Geometrical Representation Of Preferences

illustration not visible in this excerpt

(Out of "ON THE METHOD OF PAIRED

COMPARISONS", Babington Smith, Kendall 1940, p.

Figure one illustrates the same preferences indicated by the tested person that were also shown in table one. In a whole there are five mistakes in the preferences of this proband. In figure one a preference inconsistency is indicated if there is a circular relationship between three goods. The five inconsistencies are: ABD, ACD, ADF, AED and BEF. It is up to the person who evaluates inconsistencies which method is considered easier to find inconsistencies. Actually, the arrow graph of figure one needs some more time to be established but with a short glance one can find circular relations very quickly. However, when working with an output like table one the evaluating person just needs some practice and afterwards the entire evaluation process of finding mistakes in preference consistency is going to work quickly, too. The number of mistakes a tested person can commit in the experiment is described in the work of Kendall and Babington Smith, too. The maximum number of mistakes is

illustration not visible in this excerpt

for even numbers of goods that are to be compared to each other. For odd numbers of goods one has to apply the following formula:

illustration not visible in this excerpt

In the same work they also prove that the minimum number of mistakes is zero (Babington Smith, Kendall 1940, p. 327).

2.3 Priming

The triumvirate of researchers used different primings in their experiments in order to test their hypotheses. Concerning Bargh and Chartrand priming "... (deals) with the ways which internal mental states mediate, in a passive and hidden manner, the effects of the social environment on psychological processes and responses." (Bargh, Chartrand 2000). In other, much more simplified, words priming is needed to explore how people behave if they are put under certain influences which are determined by the experimenter. Furthermore, priming consists of two sub definitions: supraliminal and subliminal priming. For this paper supraliminal priming is the most important sub group of priming which is needed to understand the experiments by Lee et al. and the following experiment that was conducted in order to provide further research results for exploring preference consistency. That is why this paper focuses on supraliminal priming. Bargh and Chartrand (2000) also use the words "conscious priming" instead of supraliminal. The reason for this is that the participant perceives the priming while solving a "conscious" task. During the choice phase of the Amir, Ariely and Lee experiments the individuals are exposed to the choice stimuli (priming stimuli) and they are "fully aware of the ... stimuli ..." (Bargh, Chartrand 2000, p. 7) themselves. However, the participants do not know what the priming stimuli are all about. They do not have any knowledge about the underlying purpose of the priming. In contrast, subliminal priming are stimuli that are perceived nonconsciously.

The primings of Lee, Amir and Ariely are that they showed only names or only pictures to the both groups of experiment 1A. In experiment 1B they showed names-only, pictures-only and pictures and names to the participants. Experimental set up number two contained primings of either black and white pictures of the products or color pictures. In the third experiment participants were primed through a first step in which one group had to indicate two decisions in which they trusted their feelings and the second group had to indicate ten such decisions. In the last experimental set up the priming was done through the introduction of digit-codes. At the beginning of the experiment one group had to keep a three-digit code in mind in order to participate in a lottery to win one of the gadgets they had seen in the choosing stage. Group number two had to recall a ten-digit code. This so called cognitive load should ensure that the participants who had to keep ten numbers in mind trusted in their feelings much more than the low-loaded group.

2.4 The Experimental Setup Of Amir, Ariely And Lee

The products were shown to the probands in randomized paired comparisons in which the tested persons always had to choose one of the two presented products. Lee et al. calculated the number of product pairs that is shown to the testing persons like this: (P-1) x P/2 = number of product pairs (Amir, Ariely, Lee 2009, p. 176). P is a variable representing the number of goods that are to be presented to the probands. For their experimental series (five in total) Lee and his colleagues used ten goods (in experiments 1A, 2 and 3). From this a number of 45 = [(10-1) x 10/2] pairs is resulting. In experiments 1B and 4 they used eight products with a number of 28 pairs. The products that were used were small and cheap electronic gadgets, like a radio pen with ear phones, a voice recording key chain with LED or a multi tool with an LED, too. Pictures of the used products are listed in figure two on the next page.

In a first step they presented the names, the pictures and descriptions of the products to the participants of the experiments. The participants had as much time as they wanted to study the products and their descriptions. Afterwards, the choosing process began. The participants were shown the product pairs and they had to evaluate which one they preferred. The five experiments in this experimental series were conducted to find out how much the probands depended on their emotional approach when choosing goods.

illustration not visible in this excerpt

Figure 2: Used Gadgets To Be Presented To Test Persons

2.4.1 Experiment 1A by Amir, Ariely and Lee

In this first experiment visual stimuli were presented differently to the testing persons. One group saw only the names of the products and then had to choose. The other group saw pictures of the products and then had to decide. Using the unpaired sample t-test the triad of researchers found that participants who saw the pictures showed a significant lower level of preference inconsistency than those participants of the names group (mean value (M) of inconsistencies in pictures group: 2.7, M of inconsistencies in names group: 4.6, t (532) = 4.08, p < 0.001). Due to the fact that pictures evoke emotions and the names primarily touch the cognitive system of humans, Lee, Ariely and Amir figured out "preference consistency is associated with greater reliance on emotional responses" (Amir, Ariely, Lee 2009, p. 178).

2.4.2 Experiment 1B by Amir, Ariely and Lee

In the second, slightly changed experiment, the scientists added a third condition. Furthermore, the first condition (pictures are shown only) and the second condition (names are shown only) stayed the same as in experiment 1A. The third condition also contained names and pictures during the choice period. The results were that the pictures-only group made significantly fewer mistakes (M of pictures-only group: .4) than those participants who saw only the names (M of names-only group: 1.4), so the experiment 1B supports the findings of 1A. The errors in the names-and-pictures group were quite similar to those of the names- only group (M of names-and-pictures group: 1.2) (Amir, Ariely, Lee 2009, p. 178 to 79). Due to the fact that the names activate cognitive thinking it appears logically that there were much more mistakes in the names-and-pictures group than in the pictures-only group.

2.4.3 Experiment 2 by Amir, Ariely and Lee

In this set up the participants were presented black-and-white pictures of the goods to be chosen and color pictures. Both groups could also see the names of the products in the choice phase. Background for this priming was that color photos were expected to cause greater degree of emotionality with the participants than the black-and-white pictures. After the experiment the probands had to rate how good the pictures they had seen before represented the goods. A scale from 1 (not at all) to 7 (very well) was used. The outcome was that the black-and-white group showed twice as many mistakes than the color group (M black-and- white group: 2.2, M color group: 1.1). Additionally, participants' answers on ratings of the pictures did not reveal any significant difference on how they perceived the photographs.

2.4.4 Experiment 3 by Amir, Ariely and Lee

In experiment part number 3 Amir et al. influenced situational trust of participants in order to figure out "... if reliance on emotional reactions ... generates greater preference consistency." (Amir, Ariely, Lee 2009, p. 180). Stimuli were kept constantly. In the first step of this set up participants were asked to name two decisions in the past in which they trusted their feelings and then it turned out to be the right decision. The second group had to name ten such decisions. This approach should make people more confident in their decisions (group with two indicated cases) or less confident (group with ten decisions to be described). In part two of this experiment probands were randomly assigned to the either names-only condition or pictures-only condition. Consequently, low trust participants (those who had to name ten past decisions) committed fewer errors (M: 2.52) than the high-trust group (M: 6.28).

2.4.5 Experiment 4 by Amir, Ariely and Lee

In this set up choice stimuli were still the same. The first step of experiment 3 was left out. But now participants were set under a cognitive load. In order to take part in a lottery in that they could win one of the products they had chosen, participants had to keep in mind either a three-digit code (low cognitive load) or a ten-digit code (high load). The high-loaded group was expected to rely more on its emotions due to the fact that they had to concentrate on recalling a ten-digit code. Finally, also this presumption turned out be true on t-test base. The M of errors in the high-loaded group was .7 and in the low-loaded group 1.9.

The purpose of this experimental series was to check participants preference consistency under different influences and so to create reliable and valid results.

2.5 Metaphors And Their Influence On Human Behavior

Norbert Schwarz and Spike W. S. Lee from University of Michigan examined if metaphors used in english language show effects on the behavior of experiment participants. German Norbert Schwarz, professor of psychology at University of Michigan and member of German National Academy of Science Leopoldina, and Spike W. S. Lee, a PhD from University of Michigan, who publicized several articles on social and psycological issues in e.g. psychological science, show long records of experience in exploring human behavior. They concentrated on the link between metaphors concerning moral purity and physical cleanliness (Lee, Schwarz 2010). The exact metaphors Spike Lee and Norbert Schwarz turned their attention to were "dirty hands" and "dirty mouth", described in their article published in Psychological Science in 2010. Their research hypothesis to be tested was " If ... people prefer purification of the "dirty" body part ..." (Lee, Schwarz 2010, p. 1), which was involved in committing a misdeed, "... over the purification of other body parts." If so, they found evidence for the fact that metaphors used in everyday life are related to feelings and behavior of human beings. The experiment designed to test the hypothesis contained a scenario in which the tested persons had to imagine that they were working in a law-firm. In their firm they were competing for promotion with a coworker. This coworker had lost an important document which would boost the coworkers career and ruin the career of the tested persons.

Now, the participants had found the document and were told they should leave either a voice mail (using their mouths for speaking) or an e-mail (using their hands to type the text message). The message they had to transmit contained the name of the certain participant and the information that the participant either found the document (ethical behavior) or that the participant could not find the document (unethical behavior because the experiment participant was told that he or she had found the document). In a further step the participants were asked to rate the degree of how much they desired certain goods, by applying a scale of 1 (completely undesirable) and seven (completely desirable). Furthermore, they should indicate their willingnesses to pay for the products they had rated. Under these products there were also hand sanitizer and mouthwash. Finally, the data were analysed and Schwarz and Lee figured out that participants rated mouthwash higher, if they lied in their voice mail (M of voice mail lying group: .21). The group of participants lying in the e-mail message rated hand sanitizer higher than mouthwash (M: .31). As a result of their analysis Schwarz and Lee concluded "... that the embodiment of moral purity is specific to the motor modality involved in a ... transgression ..." (Lee, Schwarz 2010, p. 2) and that the individuals who were participating in the experiment showed much more desire to clean the part of their body which was used to commit the transgression.

2.6 The Unpaired Sample T-Test

The unpaired sample t-test is a decision rule on mathematical basis. T-test (or student's t-test) delivers analysis for the difference between the empirically evaluated mean values of two sample sizes. It is used to test hypotheses concerning expected values of one or two samples from normally distributed populations with an unknown standard deviation or unknown variance (Bortz 2005). Characteristics of the standard normal distribution are that it is symmetric and its skewness measure is zero, that its mean is equal to its mode and its median. The mean can be any numerical value (negative, zero or positive) (Anderson et al. 2007). T- test belongs to the group of parametrical tests and it needs interval scaled data. T-test works with population parameters of spread and arithmetic mean, which are estimated by use of the sample size. This test offers a help to decide whether a difference in the means of two sample groups did just appear by accident or whether there are really significant differences between the two evaluated groups. In other words, t-test assesses, if two examined groups differ systematically in their means or not (Friese et al. 2004). T-test assumes that recorded data are normally distributed. In case of a violation of this assumption the non-parametrical "Wilcoxon-Mann-Whitney-Test" (from now on WMW) has to be applied to explore differences between two sample groups of recorded data. WMW-Test also needs to be applied in case of scarcity of values. Generally, literature opinion is to use WMW-Test, if a sample size of less than 20 is available. The formula to calculate the test statistic of t-test is denoted like this:

illustration not visible in this excerpt

After the calculation of the test statistic the result is compared to a standardized t distribution table. If the calculated t of the test statistic is higher than the number of the t distribution table, a significant difference between the two compared samples is identified and the null hypothesis (H₀) representing the fact that there is no difference within data material can be rejected. P-value approach is another method to find out whether to reject the H₀ and to accept the alternative hypothesis (H₁). The H1 states that there is a difference between the compared groups. If the value of the test statistic is higher than the value out of the table, then p-value is smaller than the used confidence level and the H₀ is rejected and vice versa (Anderson et al. 2007). Programmes like SPSS statistics software and Microsoft Excel ease the mathematical procedure which was described before and help focusing on research contents.

2.7 The Wilcoxon-Mann-Whitney-Test (Also Called Mann-Whitney-U-Test)

This statistical test is a non-parametrical method. It serves as a tool to find out, if two independent populations differ from each other. In contrast to the t-test, WMW-methodology does not test for the difference of the means, but it tests "... if two populations are identical." (Anderson 2007, chapter 19, p. 28). So, H0 is: The two populations are identical, and H1 is: The two populations are not identical. This test does not need the assumption of normally distributed values. It only needs a scale which is at least ordinal and two independent samples (Anderson 2007). In order to apply this method, at first one has to combine both groups into one group and then the task is to rank the data from the lowest to the highest value (ascending order), if two values are equal one has to assign the average of the tied ranking (e.g. imagine there are two equal values in a combined group, the first equal value is on rank 15 and the second equal value is on rank 16, then the average rank is (15+16)/2=15.5, so every value of the both equal ones receives rank 15.5). Then the sum of ranks of the values per group are calculated by the sum of the ranks. This sum of ranks is called T. Then the values for the mean and the standard deviation have to be computed.

The M is calculated like this:

illustration not visible in this excerpt

Computation of SD:

illustration not visible in this excerpt

As a next step the function as to be transformed into a standard normal distribution by

applying this formula:

illustration not visible in this excerpt

This z-value is going to be transformed by looking up the corresponding value in a standard normal distribution table, considering that this is a two sided approach and one has to double the value taken from the standard normal distribution table. This value is called p-value and is compared to the level of confidence applied for the hypothesis. Is the confidence level lower than the value taken from standard normal distribution table, then H0 is rejected. Also for this test several computer programmes calculate the p-value and therefore ease the process of analysing data material. In the present paper SPSS software will be used for calculation of p- values.

3. Research Hypothesis

In utility theory preferences are an expression for relative and subjective evaluation of the bundles of goods, named A and B. These goods or bundles of goods are evaluated under consideration of their expected satisfaction of the individual's needs (Piekenbrock, 2011). Due to inherent degree of need satisfaction through bundles of goods in the individual one can conclude that the individual's inner evaluation processing is stimulated when being confronted with sliced fruits which show their inner core. So, the research hypothesis for this paper is that probands who are shown sliced fruits reveal a higher level of preference consistency than the group of test persons who were confronted with unsliced, full fruits.

[...]