How subliminal priming influences the Pay What You Want mechanism

(1)

1

How subliminal priming influences the Pay What You

Want mechanism

Anouk Teunisse

11145404

22-6-2016

University of Amsterdam Master Business Administration

(2)

2

Statement of originality

This document is written by Student Anouk Teunisse who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Preface

Before you lies the thesis “How subliminal priming influence the Pay What You Want (PWYW) mechanism”. It has been written to fulfill the graduation requirements of the Business Administration Master at the University of Amsterdam (UvA). I was engaged in researching and writing this thesis from January to June 2016.

I really enjoyed writing this thesis. The topic was very interesting, especially because this topic was a big gap in the literature. Based on existing literature I could formulate a few assumptions and during the whole process I was very curious to find out if these assumptions would be confirmed.

I would like to thank my supervisor, Tina Dudenhöffer for her excellent guidance and support during this process. Next to that I would like to thank my parents, my brother and my

boyfriend for their unconditional support. Your wise counsel and supporting words have, as always, helped me a lot. And finally I wish to thank all of the respondents, without whose cooperation I would not have been able to conduct this analysis.

I hope you enjoy your reading.

Anouk Teunisse

(4)

4

Abstract

The Pay What You Want mechanism is an increasingly popular concept for companies to differentiate themselves from each other. But there is one big disadvantage of this concept, namely that customers can exploit their control. The goal of this research is to find out if subliminal priming can influence the amount paid with the Pay What You Want mechanism and to see if there are any differences in high vs. low primes and number words vs. Arabic digits. The following research question was drafted: Can subliminal priming influence the

amount paid with the Pay What You Want mechanism and is there a difference between high vs. low primes and Arabic digits vs. number words?

To answer this research question, an online survey and a field experiment has been conducted. 254 respondents filled in the online survey and 66 additional respondents participated in the field experiment. All the respondents filled in a survey, during this survey a video was shown. Respondents were randomly assigned to one of five different groups. Four groups had a high or a low price anchor with an Arabic digit or a number word. The fifth group was the control group. The respondents were exposed to different versions of the same video. While the content of the video did not change, the subliminal primes changed for every group. In the video of the control group, there was no prime shown. After seeing the video, the respondent had to (indicate what to) pay for the cup of coffee. These outcomes were compared and there was no significant differences between the different priming groups. This was the case in both the online survey as the field experiment.Which means that in this work no indication is found that subliminal priming can help with manipulating the payment process with the Pay What You Want mechanism and also the value and the different presentations of the anchor have no influence on the amount paid.

(5)

5

1. Introduction

1957 was the year that people heard for the first time about subliminal priming (Weir, 1984). The idea of being manipulated without one’s awareness was so abhorrent to people that subliminal advertising has been legally banned in Australia, Britain, and the United States. But people were also trying to get some benefits out of subliminal priming. American consumers appeared to spend more than 50 million dollar annually on self-help audiotapes that contain subliminal messages (Pratkanis & Aronson, 1992). The industry flourished, even though scientific testing of such tapes in areas of self-esteem (Greenwald, Spangenberg, Pratkanis & Eskenazi, 1991), memory improvement (Audley, Mellett & Williams, 1991; Greenwald et al., 1991), and weight loss (Merikle, & Skanes, 1992) failed to find any evidence for the effectiveness of this phenomenon. Another body of research has confirmed the existence of subliminal priming, however it only works under certain circumstances (Strahan, Spencer & Zanna, 2002).

Another well-known phenomenon in the marketing field, is the Pay What You Want mechanism. The idea of this mechanism is that people can pay whatever they want for a product (Fernandez & Nahata, 2009). Both the consumer and the company can benefit from this. The most important benefit for the company is that it can differentiate itself from its competitors. The advantage for the consumers is that they can decide for themselves what the product is worth. This can lead to lower payments, which is a disadvantage for the company. The question in this research is if subliminal priming can subconsciously influence the response of the consumer on the amount paid with the Pay What You Want mechanism. .

(7)

7 Furthermore, this research aims to fill up the gap in literature regarding the combination of the Pay What You Want mechanism directly with subliminal priming. In the existing literature, researchers combined both the Pay What You Want mechanism and subliminal priming with anchoring (Mussweiler & Englich, 2005; Gneezy, Gneezy, Riener & Nelson, 2012). Results show that subliminal primes can serve as an anchor and that anchoring can influence the amount paid with the Pay What You Want mechanism. But no existing research has combined the Pay What You Want mechanism directly with subliminal priming. This research will explore both the concepts and the relationship between the Pay What You Want mechanism and subliminal priming. Next to that, different subliminal primes will be tested. Both the value of the anchor and different forms of presentation, such as Arabic digits vs. number words, will be controlled. An online survey and a field experiment will be conducted to see if there are differences within these groups and if subliminal priming shows any effect within the Pay What You Want mechanism. This leads to the following research question: Can

subliminal priming influence the amount paid with the Pay What You Want mechanism and is there a difference between high vs. low primes and Arabic digits vs. number words?

1.1 Structure thesis

This thesis will have the following structure. After the introduction, chapter two will expand on the existing literature about subliminal priming and the Pay What You Want mechanism. Also the different types of anchor, mainly focussing on the value and the presentation, will be discussed. Finally the hypotheses will be presented.

The research methods, the data collection, the experiment and the online survey will be discussed in chapter three. All the information which would be necessary to replicate the experiment, will be given.

(8)

8 The fourth chapter will outline the results. The hypotheses will be discussed to see if they are confirmed or not. Also the control variables will be discussed to see if these control variables have a significant influence on the amount paid with the Pay What You Want mechanism.

The final chapter, chapter five, will cover the conclusion and discussion. Conclusions will be discussed and important elements which might have influenced the research will be outlined.

2. Literature review

This research covers two topics: the Pay What You Want mechanism and subliminal priming. In this literature review both topics are discussed separately. Also different types of anchoring will be explained, connections will be made and the hypotheses will be drafted.

2.1 Subliminal priming

In 1957, James Vicary, an advertising executive from New Jersey claimed to have

substantially increased Coke sales by 18 percent and popcorn sales by over 50 percent, by secretly flashing the words “EAT POPCORN” and “DRINK COKE” onto the movie screen at a local theatre. People were outraged and alarmed (Weir, 1984). According to Pratkanis (1992) people were mostly frightened because this new technique was so devilish that it could bypass their conscious intellect and beam subliminal commands directly to their

subconscious. The Federal Communications Commission immediately investigated the Vicary study and ruled that the use of subliminal messages could result in the loss of a broadcast license (Pratkanis, 1992). In reality Vicary was lying about the increased sales. He had never flashed anything on the movie screen; it was just a hoax to save his floundering advertising company (Weir, 1984). But despite this false experiment, people do believe that subliminal

(9)

9 priming has an influence on people’s decisions. Subliminal priming is the stimulation of the unconscious and it increases the probability of later occurrence as well as the related cognitive tasks. It is commonly used within the advertisement industry where a certain images activate certain desires (Psychology dictionary, 2016). Researchers concluded that subliminal priming can have an effect on people, but only when certain conditions are met. Strahan et al. (2002) found out that both the priming of goal-relevant cognitions and the motive to pursue the goal were necessary for ads to be more persuasive. Subliminal priming per se had no effect. It was only effective when it was applied in combination with a relevant motivational state which influences the pursuit of the goal. Higgins (1996) explains that for primes to influence

behaviour they must be both accessible and applicable. Also Karremans et al. (2006) confirm that subliminal priming alone does not have any effect. It only influences the behaviour of people when they already have a motivation to act. In other words, priming alone does not always determine behaviour. This thought is shared by Zanna et al. (2002), who found out that when people are primed with a sad face and they expect to interact with another person (i.e., they are motivated to restore their mood) they will be more persuaded by an

advertisement for a mood-restoring product. In contrast, if people are primed with a neutral stimuli, even if they expect to interact with another person, there is little reason to expect that sadness would be activated and, hence, that mood restoration would play any role in their evaluation of the product. This means that the prime will activate certain thoughts or feelings, but only when there is a motivational state. When these thoughts are activated, this can influence the person’s opinion (Murphy & Zajonc, 1993), purchase behaviour (Karremans, Stroebe, & Claus, 2006) and the decision making process (North, Hargreaves, & McKendrick, 1999).

(10)

10 Shanks et al. (2013) explain that instead of, or in addition to, the normal route from conscious intentions to behaviour, individuals can be induced to act socially or unsocially, walk faster or slower, behave more or less intelligently, or perceive accurately or inaccurately as a result of the activation of subliminal primes. According to Murphy and Zajonc (1993) subliminal priming can be used to influence people’s affective reactions to an unfamiliar object. In a series of experiments, Murphy and Zajonc (1993) found that participants liked Chinese ideographs that were preceded by a subliminally presented smiling face better than the same ideographs preceded by a subliminally presented scowling face.

Subliminal priming can also influence the behaviour of people indirectly by functioning as an anchor (Mussweiler & Englich, 2005). Anchoring is a decision making bias where judgments are unduly influenced by numeric information that serves as a starting value for estimations (Slovic & Lichenstein, 1971). Also Reitsma-van Rooijen and Daamen (2006) confirm that there is a subliminal anchor effect. There is one important aspect of this anchor, namely that it is short-lived (Mussweiler & Strack, 2001). This idea is based on subliminal priming studies which showed that the duration of the effects of subliminally primes are restricted presumably because the activation level of the prime decreases over time (e.g., Forster & Davis, 1984). Reitsma- van Rooijen and Daamen (2006) explain that an anchor effect can only work in combination with time pressure for the respondent. The first reason is that in the conditions with time pressure, the time between the last subliminal presentation of the anchor value and the absolute estimate is shorter than in the conditions without time pressure. Because the accessibility level of the subliminally presented anchor value will decrease within time, the chances that the anchor value will still be activated is higher in the conditions with time pressure than in the conditions without time pressure. The second reason is that time pressure

(11)

11 interferes with extensive recall processes and increases reliance on the first thought that comes to mind.

2.2 Pay What You Want

The Pay What You Want mechanism is an unusual strategy where consumers can pay what they want, including nothing, for a product the company sells. The company cannot refuse the price paid by the consumer (Fernandez & Nahata, 2009). This sounds negative for the seller, because the consumers can pay nothing for the goods. But when the valuation of the good is positive, consumers will never free ride in equilibrium (Fernandez & Nahata, 2009). This was also the case by the release of the band Radiohead’s album “In Rainbows” in 2007. Fans were invited to download the album from the band’s website for any price they chose, including nothing. Hundreds of thousands of fans chose to pay for something they could have received for free, and Radiohead collected hundreds of thousands of dollars from its album sales. Other artists (e.g., Girl Talk) and video game companies (e.g., World of Goo) also have had some level of success in using Pay What You Want (Gneezy et al., 2012). More companies decided to use the Pay What You Want mechanism, for example for-profit organizations (as opposed to, e.g., museums or charities) (Gneezy et al., 2012) and restaurants (the Guardian, 2016). The Pay What You Want mechanism can even lead to an increase in revenues (Kim, Natter & Spann, 2009). This is because the consumer behaviour with the Pay What You Want

mechanism is largely driven by the individuals identity and self-image concerns; individuals feel bad when they pay less than the appropriate price. This can even lead them to pass on the opportunity to purchase the product (Gneezy et al., 2012).

(12)

12 Consumers react differently when they already know what the product is worth (Gneezy et al., 2012). This change in behaviour is caused by the fact that they already have a price anchor. They already have an external reference price which they compare to their own internal price (Kim et al., 2009). The external price anchors yield two types of effects. First, they can provide a standard relative to which the price of other products is evaluated. Second, they can stimulate thoughts about the features of products that might be sold at these prices. Both of these factors can influence the price that consumers are willing to pay for a product (Adaval & Wyer, 2011).

2.3 Different kind of anchors

As explained above, price anchors can influence the Pay What You Want mechanism. The question arises whether different types of price anchors, such as the value and the presentation of the anchor, have different influences on the amount paid with the Pay What You Want mechanism. In paragraph 2.3.1. there will be discussed if there is a difference between high and low anchors and in paragraph 2.3.2. there will be discussed if there is a difference between Arabic digits and number words.

2.3.1 Value of the anchor

Anchoring is a popular topic in the literature. A few studies about anchoring are done in combination with an auction. Ku, Galinsky, & Murnighan (2006) conducted six studies using a diverse set of methodologies with an auction element in it. The aim of this research is to test whether low anchor values in auctions can actually produce higher final prices than high anchor values. One of the conclusions was that lower starting prices result in higher final prices. But this was mainly because the low starting price generates a lot of traffic. Looking at an intrapsychic level, starting prices had an assimilative effect on value judgments. Low

(13)

13 starting prices generate lower value estimates, which is consistent with traditional anchoring findings. Which means that if the generation of traffic will be excluded, low starting prices will lead to lower value estimates.

Adaval and Wyer (2011) also conducted research about anchoring, but this research was not done in combination with an auction. Participants were subliminally primed with a high or low price anchor. Afterwards they were asked to estimate the price they were willing to pay for a DVD player and a pair of shoes. As expected, estimates increased with the value of the subliminally primed numbers to which participants were exposed. That is, participants estimated that they were willing to pay more for a product when the primed price was high instead of low.

The study of Adaval and Wyer (2011) was combining anchoring with the willingness to pay (WTP). But what is the influence of the value of the anchor on the Pay What You Want mechanism? Jang & Chu (2012) explain that there is a difference between the WTP and the price people pay with the Pay What You Want mechanism. WTP stands for the maximum price at or below which a consumer will definitely buy one unit of the product (Varian, 1992). This is not always the same amount that people pay with the Pay What You Want mechanism, because there are other influences. One of these influences with the Pay What You Want mechanism is selfishness (Jang & Chu, 2012). When people can pay whatever they want to pay for a product or service, selfishness can lead to low payments. But Jang and Chu (2012) also explain that people are not always selfish under this pricing scheme. They have a duality about being selfish or being fair. Because of this duality, people are not always paying the maximum amount they are willing to pay with the Pay What You Want mechanism. Another influence is the social environment of people. The authors (Jang & Chu, 2012) show that

(14)

14 consumers can be influenced into taking fair actions by providing cues about ‘‘socially

correct’’ actions others are taking. The reason for this is that many consumers act unfairly not because of their inherent propensity, but because they believe others are doing the same. Next to that, self-image is also very important for people. Greenberg (1983) argues that allocators’ adherence to distributive justice is due to their concern for the impressions they make on themselves (self-image) and on others (impression management). This leads to the conclusion that Pay What You Want and WTP are different mechanisms which can lead to different amounts paid.

Johnson and Cui (2013) found some contradicting results that not using an external reference prices may be the most beneficial strategy for the firm. Both the minimum and maximum prices will have a negative influence on consumers' chosen prices in comparison to not offering an external reference price. However, a suggested price strategy appears to be an effective means of maximizing the firm's yield while giving consumers the freedom to choose their own price, especially when the suggested price is close to the consumer's internal

reference price. This means that it depends what kind of external reference price it is. A minimum or maximum price will negatively influence the prices paid, but the suggested price will result in a high chosen price, even higher than when participants are given no external reference price. A suggested price is likely the most effective external reference price tool for managing the firm's yield and minimizing chosen prices that are too low for the firm. By suggesting a price, firms can influence consumers to choose prices close to the suggested price without setting either an upper or lower limit on consumers' chosen prices. In addition, this strategy enables firms to offer Pay What You Want pricing while communicating their product's value (Johnson & Cui, 2013). This means that it is possible to influence people through an external reference price, but only when people have the freedom to choose the

(15)

15 price themselves. This is comparable with subliminal priming influencing the Pay What You Want mechanism, because the Pay What You Want mechanism still gives the consumers the option to choose their own prices without an upper and lower bound. The only difference with subliminal priming is that they do not know they are being primed.

As explained above, subliminal priming might influence the Pay What You Want mechanism. Looking at earlier studies in combination with the WTP mechanism, high anchors lead to a high amount paid and the other way around. This leads to the following hypotheses regarding the Pay What You Want mechanism:

Hypothesis 1: A high subliminal primed price, in contrast to a low subliminal primed price, will positively influence the amount paid with the Pay What You Want mechanism

Hypothesis 2: A low subliminal primed price, in contrast to a high subliminal primed price, will negatively influence the amount paid with the Pay What You Want mechanism

2.3.2 Presentation of the anchor

As stated before, the value of the anchor might be influencing the outcomes. The question arises whether the way the anchor is presented (Arabic digit vs. number words) influences the results and if this can influence the amount paid with the Pay What You Want mechanism. There is already some research about this topic, but there is not one main conclusion. This research will first examine if there is a difference between processing Arabic digits and number words and then test if there is a relationship between subliminal priming and the presentation of a number.

(16)

16 Fias et al. (2001) conducted a research about how people process Arabic digits. They

questioned whether Arabic numbers were processed more like pictures or like words. There is a big difference between these processing types and it is known that picture naming is

interfered by an incongruent word, but word naming is hardly hindered by the presence of an incongruent picture. In the research of Fias et al. (2001), they concluded that the Arabic digit naming is interfered by the presence of an incongruent word numeral, while naming the word number is not influenced by the presence of an incongruent Arabic digit. This will lead to the conclusion that Arabic numbers are processed like picture, while word numbers aren’t. This concludes that there is a difference between the processing of Arabic digits and word numbers. Fias et al. (2001) also conclude that Arabic digits are always processed with semantic mediation, whereas written number words can either access semantics, or their phonological codes directly. The results suggest that number words, much like words in general, are read based on non-semantic processing pathways (Fias et al., 2001).

The idea that Arabic digits are always processed with semantic mediation is not shared by all researchers. Cipolotti & Butterworth (1995) explain that there is a double route from Arabic input till spoken verbal output, one semantic and one non-semantic route. Cipolotti (1995) suggested the existence of an asemantic route for reading aloud Arabic numerals in which Arabic numerals are translated into spoken number names without computing a semantic representation. Also Deloche and Seron (1987) and Cohen and Dehaene (1991) suggested the existence of an asemantic route for processing numbers though they have not formally

(17)

17 Van den Bussche et al. (2009) claim that the context will direct the subject’s attention to a certain processing level and will determine whether the stimuli will be semantically or asemantically processed. A small target set will stimulate an asemantic processing strategy. Paying attention to certain lower features, like letters or bigrams, is sufficient to fulfill the task. Once a semantic or asemantic strategy has been selected to perform the task on the targets, a similar kind of strategy (semantic or asemantic) will also be adopted for the processing of the primes. Thus, for a small target set the primes will only be processed asemantically. A large target set will encourage subjects to use a semantic strategy to process the stimuli since paying attention to lower feature levels will no longer sufﬁce to classify the targets. In the current research Arabic digits will be primed, which will be a small target and processing will be asemantic according to Van den Bussche et al. (2009). Number words are being primed as well, posing a larger target and according to Van den Bussche et al. (2009) the words will be processed semantically.

It is clear that there is a difference in processing Arabic digits and number words. But the question arises how this difference influence the amount paid with the Pay What You Want mechanism. Damian (2014) was also interested in the differences between Arabic digits and number words. In the first experiment of his research, he established that number words were named faster than corresponding digits, but only if the two notation formats were presented in separate experimental blocks. In the second experiment he compared the naming of number words and Arabic digits to a manual magnitude judgement task. Digits were named slower, but categorized faster than number words. This suggest that processing of the two notation formats is asymmetric, with digits gaining rapid access to numerical magnitude

representations, but slower retrieval of lexical codes, and the reverse for number words (Damian, 2014).

(18)

18 As stated before, there is a difference in processing Arabic digits and number words. But as explained earlier there is not one main conclusion regarding this subject. Different authors have different conclusion about the route Arabic digits and number words follow. But as Damian (2014) explains, the latencies of Arabic digits are a lot faster and digits are gaining more rapid access to numerical magnitude representations than number words. There is no contradicting research about this statement. This is crucial in this current research, where the subliminal primes are shown for 23 MS. If the latencies of Arabic digits are a lot faster, there is a big chance that people remember these primes better than number word primes. This leads us to the following hypotheses.

Hypothesis 3: A high subliminal primed Arabic digit, will have a more positively influence on the amount paid with the Pay What You Want mechanism compared to a high subliminal primed word number.

Hypothesis 4: A low subliminal primed Arabic digit, will have a more negatively influence on the amount paid with the Pay What You Want mechanism compared to a low subliminal primed word number.

The theoretical model of this thesis is as following:

Value of the anchor Presentation of the anchor

Subliminal priming Amount paid with the PWYW mechanism

(19)

19

3. Methods & Analysis

This chapter is divided in five paragraphs. In the first paragraph there will be explained why the product coffee is chosen for this study. The second paragraph will describe how the pre-test has been conducted. The third paragraph will discuss the online survey and the field experiment will be discussed in the fourth paragraph. Paragraph five will explain how the variables are reflecting the question of this thesis.

3.1 The product

The research design of this study is a combination of an online survey and a field experiment, with coffee as the product. The focus is set on coffee, because it is a low cost, regular bought, product. To make sure the product is an independent variable, a new coffee brand, AT coffee, has been created. Because this is a new coffee brand, people do not know the ‘regular’ price for a cup of AT coffee. This will help to exclude habit behaviour, which means that customer already have a common value (Nunes & Boatwright, 2004) which can influence the amount paid. One side note should be made, because coffee is a regular bought product people already have a mental range in mind what they are willing to pay for a cup of coffee.

3.2 Pre-test

The prices of two big coffee companies in the Netherlands, Starbucks and the Coffee

Company, were compared. For a normal size black coffee, the mean was 3 euro. A high and low standard were created by altering it with a difference of two euro. This means that the primes were set on five and one. Then a pre-test was conducted, to see if one and five euro are not commonly used prices to pay. Fifty respondents filled in the survey. The mean age was 27 years old and 19 respondents were men. The respondents were asked to answer 4 questions: 1. What is the price you will pay for a cup of black coffee to go? 2. What is the price you will

(20)

20 pay for a cup of black coffee to go on a street vendor? 3. What is your age? 4. Are you a man or woman? Results show that people were willing to pay less for a cup of black coffee on a street vendor than in other settings. The mean of the answers on question one was €2,08. The mean of the answer on question two was €1,64, see table 1 and 2.

N Minimum Maximum Mean Std.

Deviation Price for a

cup of coffee

50 1.00 3.00 2.08 0.46912

Table 1. Descriptives question one, pre-test

N Minimum Maximum Mean Std.

Deviation Price for a

cup of coffee

50 0.50 3.00 1.64 0.51431

Table 2. Descriptives question two, pre-test

For the online survey, the mean of question one is the most important outcome. This might be different for the field experiment, because this will be done in a pop-up coffee store at the canteen of the University of Amsterdam. Because this is not a ‘real’ coffee shop, the prices of a street vendor might be more important. These outcomes show that the low and high prices are not the same as the chosen primes, which means that the chosen primes can be set at one and five euro.

(21)

21

3.3 Online survey

254 respondents participated in the online survey. Most of the respondents were contacted through Facebook. It included a short survey which took around three minutes. All the data was collected in April 2016 and is strictly confidential and anonymous. The results are collected in the program Qualtrics with permission of the University of Amsterdam. Respondents were first asked about their age, gender, nationality, income and education. Next, the respondents saw a one minute during video about the coffee growing process. The respondents were told that they had to see the video to show how the beans were made. This information will distract the respondent from the real purpose of the video, namely the priming effect. The method of this experiment is a quantitative 2x2 factorial design, see table 3. There are five different groups. In four of the five groups there are subliminal messages hidden in the video. The other group is the control group.

Subliminal priming High Anchor Low Anchor

Number word Five One

Arabic digit 5 1

Table 3. 2x2 experimental factorial design.

The respondents were randomly assigned to one of the groups, to make sure that every respondent had the same chance to participate in a certain condition. After the respondents saw the video, they continued with the survey. A fictitious situation was presented which participants were asked to envision. The respondents were asked to imagine that they entered a coffee shop, walked towards the counter and then they had to decide themselves what they wanted to pay for a cup of black AT coffee. A picture of a cup AT Coffee was added to make sure there will be no differences in the amount paid because of the size of the cup. Next there

(22)

22 was a question about the quality of the beans and finally they were asked if there was

something striking about the video. This last question is a control question to see if people are aware of the priming or not. If people were aware of the prime, they will be deleted from the dataset.

As explained earlier, subliminal priming only works in combination with time pressure (Reitsma- van Rooijen and Daamen, 2006), which means that there cannot be much time between the video and the questions about the amount people are willing to pay. Directly after the respondents saw the video, the respondents were asked to fill in the price they would pay for a cup AT coffee. Another important aspect of the survey is that the respondents were asked, before they saw the video, to count the amount of women in the video. This question was included to increase their attention and to keep them focused on the video.

As explained there are five different respondent groups. See table 4 for the group distribution.

Group 1 This group will be subliminally primed. They will be primed with the Arabic number 5. Group 2 This group will be subliminally primed. They

will be primed with the verbal number five. Group 3 This group will be subliminally primed. They

will be primed with the Arabic number 1. Group 4 This group will be subliminally primed. They

will be primed with the verbal number one. Group 5 This will be the control group. This group

will not be primed. Table 4. Distribution of the priming groups (online survey)

(23)

23 The video which was shown to all the respondents, is a very neutral video. The video was made with the program Windows Movie Maker. The video shows how the coffee is made. The primes were presented for 23 MS, including one frame. The image of the prime was a black background with a white front. It was exported on Youtube. The duration of the prime, 23 MS, was based on the research of Fishbach et al. (2003), where was shown that subliminal priming was possible till 23 MS. People are being primed three times, because the video will last one minute, one time priming might be not enough.

3.4 Field experiment

The field experiment was done in a pop-up coffee shop. This was done at the University of Amsterdam. In the canteen of the University of Amsterdam, a small pop-up coffee store arose. 66 respondents participated in the field experiment. The data for this field experiment was collected in April 2016 and is strictly confidential and anonymous. The results are collected in the program Qualtrics with permission of the University of Amsterdam. The respondents had to fill in the same survey as the respondent from the online survey, the only difference is that instead of asking what they would pay for a cup of AT coffee, they are actually paying for the cup of coffee. The purpose of this field experiment is to see if there are differences between the intention to pay and the amount paid in a real situation. Because the field experiment was done with 66 respondents, only the high and low Arabic number primes were shown. There were three groups in the field experiment, see table 5.

(24)

24 Group 1 This will be the control group. This group

will not be primed.

Group 2 This group will be subliminally primed. They will be primed with the Arabic digit 5.

Group 3 This group will be subliminally primed. They will be primed with the Arabic digit 1.

Table 5. Groups distribution priming field experiment.

The videos of these groups are the same as the videos of the online survey. But now the respondents will first order a cup of coffee and after this they will fill in the first five control questions and they will see the video. As soon as the video stops, respondents are asked to pay a self-chosen amount for the coffee. After paying, the cup of coffee was given to them. This was intentionally because else the taste of the beans could influence the amount paid. Also the coffee machines were shielded, to avoid that the Senseo logo could influence the amount paid. It is important to note that there were no regular prices stated, to prevent other anchors. After the respondents paid for their coffee, they had to answer three more questions about the number of women, the quality of the beans and if they saw something striking about the video.

3.5 Explanation of the variables

This work contains both independent as dependent variables. The independent variable, subliminal priming, is processed in the video the respondents see. The dependent variable, the amount people pay at the Pay What You Want mechanism, is tested by asking respondents how much they would pay for a cup of AT coffee. In the field experiment it was tested by letting people actually pay. The first moderator, the value of the anchor, is measured by

(25)

25 putting high and low anchors in the video. The second moderator, the presentation of the anchor, is measured by priming number words and Arabic digits in the video. The outcomes of the different groups will show if there is indeed a relationship between these variables.

4. Results

4.1 Online survey

The online experiment has been conducted among 254 respondents. On basis of the question if the respondents saw something striking about the video, 40 respondents were excluded from the analysis. These respondents had indicated that they saw the prime. Because this research is about subliminal priming, these respondents needed to be excluded. Next to that two outliers were deleted. The final dataset of the online survey included 212 respondents. A G-Power test has been conducted to determine the sample size for this specific research. The results of the G-Power test show a minimum of 200 respondents, so after excluding the respondents the sample size is still big enough. Which means that the probability of detecting a “true” effect is high. This is of course only possible if there is any effect.

The first five questions of the online survey were control questions to see if there were any differences between and within the different priming groups. In the distribution overall there were some noteworthy differences in the spreading of income, nationality and how they perceived the quality of the beans. But also within the spreading of the other variables were some differences. Looking at gender, 92 respondents (43,4%) were male and 114 respondents (53,8%) were female. There were 6 missings (2,8%). Looking at gross annual income, 169 respondents (79,7%) earned between 0-20.000 euro gross a year, 19 respondents (9%) earned between the 20.000-35.000 euro gross a year, 18 respondents (8,5%) earned between

(26)

35.000-26 65.000 a year and 4 respondents (1,9%) earned 65.000 euro or higher gross a year. There were 2 missings (0,9%). Looking at nationality, 207 respondents (97,6 %) were Dutch, 3 (1,4%) respondents had another nationality. There were 2 missings (0,9%). Looking at the question about the quality of the beans 70 respondents (33%) thought the quality of the beans was medium, 136 (64,2%) respondents thought the quality was high and 5 respondents (2,4%) thought the quality was very high. There was 1 missing (0,5%). And finally looking at the highest degree of the respondents, 2 respondents (0,9%) only finished elementary school, 69 respondents (32,5%) only finished high school, 30 respondents (14,2%) achieved a bachelor degree on college, 70 respondents (33%) completed a bachelor degree at the university, 34 respondents (16%) finished an master, 4 (1,9%) respondents had a PHD degree and there were 3 missings (1,4%). Next to that, important to mention; 52 respondents were assigned to the control group, 40 respondents were assigned to the priming group with the Arabic digit 5, 45 respondents were assigned to the priming group with the Arabic digit 1, 42 respondents were assigned to the priming group with the word number one and 33 respondents were assigned to the priming group with the word number five. Not all the groups have the same amount of respondents because of the 42 deleted respondents. By the variable amount paid, there were 2 missings. And finally the variable age has also been looked at to see if there were any big differences, but there were not.

The demographic data per different priming group was also looked at. There were only some differences of distribution within the variable gender and quality of the beans. Table 6 shows the differences within the gender variable and table 7 shows the differences within the quality of the beans perception.

(27)

27

Male Male % Female Female % Missings Missings %

Control group 26 50 24 46,2 2 3,8

Priming 5 11 27,5 27 67,5 2 5

Priming 1 24 53,3 21 46,7 0 0

Priming one 14 33,3 27 64,3 1 2,4

Priming five 17 51,5 15 45,5 1 3

Table 6. Differences in the distribution of the variable gender within the online survey.

Medium Medium% High High% Extr. high1 Extr. high% Missings Missings% Control group 15 28,8 37 71,2 0 0 0 0 Priming 5 9 22,5 29 72,5 2 5 0 0 Priming 1 19 42,2 22 48,9 3 6,7 1 2,2 Priming one 18 42,9 24 57,1 0 0 0 0 Priming five 9 27,3 24 72,7 0 0 0 0

Table 7. Differences in the perception of the quality of the beans, within the online survey.

As shown in table 6, there are some differences in the distribution of the respondents within the variable gender. Expected is that the difference in gender will have no influence on the amount paid, especially because the reasoning behind this question was that men are often getting paid more than women. But looking at income, there are no differences between the groups, that´s why there is expected that there will be no influence. For the quality of the beans there was not such a theoretical explanation. That’s why a chi-test has been conducted, see table 8. There was a statistically significant association between the perceived quality of

1

(28)

28 the beans and the amount paid, χ2(1) = 15.270, p = .047. Moreover, based on the ϕ statistic the association seems to be of medium size (ϕ= .27).

Different priming groups Quality beans Control Priming group 5 Priming group 1 Priming group one Priming group five χ 2 Φ Medium quality 15 (-.5) 9 (-1.2) 19 (1.2) 18 (1.1) 9 (-.6) 15.270 0.269 High quality 37 (.6) 29 (.6) 22 (-1.2) 24 (-.6) 24 (.6) Very high quality 0 (-1.1) 2 (1.1) 3 (1.9) 0 (-1.0) 0 (-.9)

Table 8. Chi-test influence quality of the beans on the different priming groups

The next step was to conduct a One-Way ANOVA test. The one-way analysis of variance (ANOVA) is used to determine whether there are any significant differences between the means of two or more independent groups. And that is exactly what this research is about, looking at the differences in amount paid between the different priming groups. On basis of Skewness and the Kurtosis, histograms and boxplots there is concluded that all the groups were normally distributed except group 2, with the prime Five. The outliers, respondent number 181 and 199 were deleted to obtain a normally distributed dataset. Afterwards a One-Way ANOVA, Bonferroni test has been conducted. The output shows that the Test of

Homogeneity of Variances is not significant (p= .731), which means that the One-Way ANOVA, Bonferroni is a good test to conduct. Looking at the significant of this test, it is not

(29)

29 significant, F(4, 205) = .241, p > 0.05. Which means that the video has no influence on the amount paid, see table 9.

SS DF MS F Sig.

Priming group .557 4 .139 .241 .915

Error 118.444 205 .578

Total 119.002 209

Table 9. One-Way ANOVA online survey

Table 10 shows the mean, standard deviation and number of respondents of the different priming groups of the online survey.

Mean Std. Deviation N Control _1,95 _,76 ₅₁ Priming 5 _1,97 _,71 ₄₀ Priming 1 _2,05 _,85 ₄₅ Priming One _1,96 _,705 ₄₁ Priming Five _1,89 _,745 ₃₃ Total _1,97 _,755 ₂₁₀

Table 10. Mean and standard deviation of the online survey

Next to that, a One-Way ANOVA was conducted with the option contrasts, see table 11. Planned contrasts revealed that the control group did not significantly increased the amount paid compared to the other priming groups, t(205) = .137, p = .892. Planned contrasts also revealed that the group primed with 5, did not significantly increased the amount paid compared to the other priming groups, t(205) = -.046, p = .963. Also the priming group with the digit 1, did not significantly increased the amount paid compared to priming group with the number five and one, t(205) = -.897, p = .371. Finally the priming group with the number

(30)

30 one, did not significantly increased the amount paid compared to the priming group with the number five, t(205) = -.413, p = .680.

Amount paid Value of Contrast Std. Error T df Sig. Contrast 1 .0335 .24502 .137 205 .892 Contrast 2 -.0096 .20883 -.046 205 .963 Contrast 3 -.1292 .14401 -.897 205 .371 Contrast 4 -.0734 .17777 -.413 205 .680

Table 11. Contrasts online survey

To check if the control variables have any influence on the amount paid, different One-way ANOVA’s has been conducted. The variable gender has a non-significant Homogeneity of Variances (p = .24) and looking at the significance of the One-Way ANOVA test, it is

significant, F(1, 203) = 20.782, p < 0.05. Which means that gender has a significant influence on the amount paid (see Appendix A, table 12). The variable age has a non-significant

Homogeneity of Variances (p = .104) and looking at the significance of the One-Way ANOVA, it is significant, F(23, 183) = 2.747, p < 0.05. Which means that age has a

significant influence on the amount paid (see Appendix A, table 13). The variable income has a non-significant Homogeneity of Variances (p = .546) and looking at the significance of the One-Way ANOVA, it is significant, F(3, 204) = 4.441, p < 0.05. Which means that income has a significant influence on the amount paid (see Appendix A, table 14). The variable degree has a non-significant Homogeneity of Variances (p = .116) and looking at the significance of the One-Way ANOVA, it is significant, F(5, 201) = 3.184, p < 0.05. Which means that degree has a significant influence on the amount paid (see Appendix A, table 15). The variable nationality has a non-significant Homogeneity of Variances (p = .761) and

(31)

31 looking at the significance of the One-Way ANOVA, it is not significant, F(1, 206) = 1.485, p > 0.05. Which means that nationality has no significant influence on the amount paid (see Appendix A, table 16). The variable quality of the beans has a non-significant Homogeneity of Variances (p = .260) and looking at the significance of the One-Way ANOVA, it is

significant, F(2, 206) = 5.032, p < 0.05. Which means that the perception of the quality of the beans has a significant influence on the amount paid (see Appendix A, table 17).

4.2 Field experiment

The field experiment has been conducted among 66 respondents. On basis of the question if the respondents saw something striking about the video, 6 respondents were excluded from the analysis. No outliers were excluded. Because the field experiment is just a confirmation of the online experiment, no G-Power test has been conducted.

For the field experiment the same analysis as the online survey has been done. Looking at gender, 29 respondents (48,3%) were male and 31 respondents (51,7%) were female. Looking at gross annual income, 55 respondents (91,7%) earned between 0-20.000 euro gross a year, 3 respondents (5%) earned between the 20.000-35.000 euro gross a year and 2 respondents (3,3%) earned between 35.000-65.000 euro gross a year. Looking at nationality, 45

respondents (75%) were Dutch, 15 (25%) respondents had another nationality. Looking at the question about the quality of the beans 2 respondents (3,3%) thought the quality of the beans was low, 23 respondents (38,3%) thought the quality was medium, 34 respondents (56,7%) thought the quality was high and 1 respondent (1,7%) found the quality very high. And finally looking at the highest degree of the respondents, 17 respondents (28,3%) only finished high school, 9 respondents (15%) achieved a bachelor degree on college, 19 respondents (31,7%) completed a bachelor degree at the university, 12 respondents (20%) finished an master and 3

(32)

32 respondents (5%) had a PHD degree. Looking at the distribution of the different groups, 22 respondents were assigned to the control group, 17 respondents to the prime group 5 and 21 respondent were assigned to the prime group 1. Also in the field experiment there were no big differences in the distribution of age within the different priming groups. Finally we looked at the demographic data per different priming group of the field experiment. There were no big differences in distribution of the respondents, which means that we can continue with

analysing the data.

The next step was to conduct a One-Way ANOVA. On basis of Skewness and the Kurtosis, histograms and boxplots there is concluded that all the groups were normally distributed. A One-Way ANOVA, Bonferroni test has been conducted. The output shows that the Test of Homogeneity of Variances is not significant (p = .162), which means that the One-Way ANOVA, Bonferroni is a good test to conduct. Looking at the significance of this test, it is not significant, F(2, 57) = .664, p > 0.05. Which means that the priming in the video of the field experiment has no influence on the amount paid (see table 18).

SS Df MS F Sig.

Priming group .558 2 .279 .664 .519

Error 23.960 57 .420

Total 24.518 59

Table 18. One-Way ANOVA field experiment

Next to that, a One-Way ANOVA was conducted with the option contrasts, see table 19. Planned contrasts revealed that the control group did not significantly increased the amount paid compared to the other priming groups, t(57) = -1.110, p = .272. Planned contrasts also

(33)

33 revealed that the priming group with the digit 5, did not significantly increased the amount paid compared to priming group with the digit 1, t(57) = .380, p = .705.

Amount paid Value of Contrast

Std. Error T df Sig.

Contrast 1 -.1932 .17405 -1.110 57 .272

Contrast 2 .0804 .21153 .380 57 .705

Table 19. Contrasts field experiment

Table 20 shows the mean and std. deviation of the different priming groups of the field experiment. Mean Std. Deviation N Control _1,39 _,77 ₂₂ Priming 5 _1,15 _,62 ₁₇ Priming 1 _1,23 _,51 ₂₁ Total _1,27 _,64 ₆₀

Table 20. Mean and standard deviation of the field experiment

To check if the control variables have any influence on the amount paid, different One-way ANOVA’s has been conducted. The variable gender has a non-significant Homogeneity of Variances (p = .204) and looking at the significance of the One-Way ANOVA, it is not significant, F(1, 58) = 2.687, p > 0.05. Which means that gender has no significant influence on the amount paid (see Appendix B, table 21). Age has a non-significant Homogeneity of Variances (p = .471) and looking at the significance of the One-Way ANOVA, it is not significant, F(16, 43) = 1.304, p > 0.05. Which means that age has no significant influence on the amount paid (see Appendix B, table 22). The variable income has a non-significant Homogeneity of Variances (p = .446) and looking at the significance of the One-Way ANOVA, it is not significant, F(2, 57) = .837, p > 0.05. Which means that income has no

(34)

34 significant influence on the amount paid (see Appendix B, table 23). The variable degree has a non-significant Homogeneity of Variances (p = .477) and looking at the significance of the One-Way ANOVA, it is not significant, F(4, 55) = 2.384, p > 0.05. Which means that degree has no significant influence on the amount paid (see Appendix B, table 24). Important to notice is that it is close to significant, which is called marginally significant. So there is no significant influence but there is a marginally significant influence. The variable nationality has a non-significant Homogeneity of Variances (P = .146) and looking at the significance of the One-Way ANOVA, it is not significant, F(1, 58) = 1.697, p > 0.05. Which means that nationality has no significant influence on the amount paid (see Appendix B, table 25). The variable quality of the beans has a non-significant Homogeneity of Variances (p = .085) and looking at the significance of the One-Way ANOVA, it is not significant, F(3, 56) = .848, p > 0.05. Which means that the perception of the quality of the beans has no significant influence on the amount paid (see Appendix B, table 26).

4.3 Reflection of the hypotheses

The results from both the field experiment as the online survey show that the video has no influence on the amount paid. Which will lead to the following conclusion of the hypotheses:

 Hypothesis 1: A high subliminal primed price, in contrast to a low subliminal primed

price, will positively influence the amount paid with the Pay What You Want mechanism

This hypothesis will be rejected. There is no significant influence of a high subliminal prime on the amount paid with the Pay What You Want mechanism.

(35)

35  Hypothesis 2: A low subliminal primed price, in contrast to a high subliminal primed

price, will negatively influence the amount paid with the Pay What You Want mechanism

This hypothesis will be rejected. There is no significant influence of a low subliminal prime on the amount paid with the Pay What You Want mechanism.

 Hypothesis 3: A high subliminal primed Arabic digit, will have a more positively

influence on the amount paid with the Pay What You Want mechanism compared to a high subliminal primed word number.

This hypothesis will be rejected. There is no significant influence of a high subliminal primed Arabic price on the amount paid with the Pay What You Want mechanism.

 Hypothesis 4: A low subliminal primed Arabic digit, will have a more negatively

influence on the amount paid with the Pay What You Want mechanism compared to a low subliminal primed word number.

This hypothesis will be rejected. There is no significant influence of a low subliminal primed Arabic price on the amount paid with the Pay What You Want mechanism.

5. Conclusion and discussion

5.1 Conclusion

This research questioned if it is possible to unconsciously manipulate the payment process of the Pay What You Want mechanism through subliminal priming, resulting in the question:

(36)

36

Can subliminal priming influence the amount paid with the Pay What You Want mechanism and is there a difference between high vs. low primes and Arabic digits vs. number words?

Hereby filling up the gap in literature regarding the relationship of the Pay What You Want mechanism with subliminal priming. Next to that, this research investigated the subliminal priming variables and has tested different subliminal primes. Both the value of the anchor and the different presentations, such as Arabic digits vs. number words, were being controlled. An online survey and a field experiment has been conducted and conclusions are drawn.

The online survey has been conducted under 254 respondents. 40 respondents indicated that they saw the prime, which means that they were excluded from the research. Next to that, 2 outliers were deleted. The field experiment was conducted to see if the intention to pay, which was measured in the online survey, was the same as people really had to pay. 66 respondents anticipated in the field experiment. Six respondents indicated that they saw the prime, which led them be excluded from the research. The online survey was conducted with Qualtrics, but most of the respondents were drawn through Facebook. The field experiment was conducted at the canteen of the University of Amsterdam. The respondents were randomly assigned to one of the five videos. In four of the five videos were primes processed. High vs. low primes and number words vs. Arabic digits were shown. There was also one control group. After seeing the video, the respondents had to indicate what they would pay for a cup of AT Coffee. In the field experiment there was no indication question, but people really had to pay. The results show that there were no significant differences between the different priming groups, also not between the primed groups and the control group. This led to the conclusion that there is no significant relationship found between subliminal priming and the Pay What You Want mechanism. The only significant relationship was found in the online survey between

(37)

37 all the control variables, except nationality, and the amount paid. These variables do have a direct influence on the amount paid with the Pay What You Want mechanism.

5.2 Discussion

The Pay What You Want mechanism has a few elements which can influence the amount paid. As explained before, selfishness can have a big influence on the amount paid. If the selfishness is high, it can lead to low payments (Jang & Chu, 2012). Other influences which can have a big influence on the amount paid are self-image and social pressure (Greenberg, 1983). A possible explanation for the non-significant results is that these influences were very strong in this research and they were overruling the influence of the subliminal prime on the amount paid. But there is no existing research about this topic. Future research should be conducted to see if influences can overrule each other and under what circumstances this occurs. In this current research, the external influences might be strengthened by the

experimental set-up. During the field experiment the respondents were asked to pay after they saw the video. They paid directly at the researcher, which means that the payment was not anonymous. There were often more people in line for the coffee, who were able to see how much the respondent paid for the cup of coffee. This might have strengthened the social pressure aspect. Jang and Chu (2012) explain that social pressure is mainly from the social environment of people. The authors show that consumers can be influenced into taking fair actions by providing cues about ‘‘socially correct’’ actions others are taking. The reason for this is that many consumers act unfairly, not because of their inherent propensity but because they believe others are doing the same (Jang & Chu, 2012). This can be strengthened because of the public payment. Respondents could already have an idea what to pay based on the payment of the respondents in the line before them. To act “socially correct” the respondents had to pay an amount which would be in line with the payment of the other respondents.

(38)

38 During the online survey it was anonymous, which means that social pressure did not

influence these results.

There is another possible explanation for the non-significant results. Researchers concluded that subliminal priming works, but only when the motive to pursue the goal was present (Strahan et al., 2002). Subliminal priming per se had no effect. This important element was processed in the online survey by making people imagine that they were in the specific situation. They had to image they entered a coffee shop and were going to buy a cup of coffee. If the respondents really placed themselves in that situation, the motivation state should be achieved. But if the respondents were not able to image themselves in that specific situation, than the motivational state might have not been achieved. Which could influence the results of the online survey. This explanation is only applicable for the online survey. During the field experiment, the respondents were only participating if they had the intention to buy a cup of coffee. This means that in the field experiment this motivational state was achieved. But there is one important side note. The field experiment was conducted with the help of 66 respondents. The sample of the field experiment is on the small end to draw strong

conclusions. This could also be the reason why all the tests are not significant. Important to notice; the goal of the field experiment was to confirm that the amount intended to pay is equal to the actual amount paid, in a real-life setting. This is the reason why 66 respondents were enough for this research.

Looking at the total mean score of the online survey and the field experiment, there was a big difference. The total mean score of the online survey was €1,97 and the total mean score of the field experiment was €1,27. As explained above, the field experiment was on the small end to draw strong conclusions. Further research should be conducted to see if these

(39)

39 differences are smaller when the sample size of the field experiment is bigger. And also to see if the results of the field experiment are still not significant. The big differences between these two means can also be occurred because of the set-up of the online field experiment. As explained with the pre-test, people are willing to pay less for a cup coffee on a street vendor than in a coffee shop. The pop-up coffee shop at the UvA can be compared to a street vendor, because it wasn’t a real coffee shop. So this can partly explain why the average paid price is a lot lower at the field experiment.

Also the duration of the subliminal prime, 23 MS, could have had an influence on the results. Fishbach et al. (2003) explain primes can be shown subliminal till 23 MS. If the prime will be shown longer than 23 MS, people will notice the prime. Looking at our research, especially the online survey, a serious number of respondents saw the prime. This is not in line with the research of Fishbach et al. (2003). The respondents which indicated they have seen the prime, were deleted out of the data set. The other respondents did not indicate that they saw the prime, but it is of course possible that more respondents saw the prime. If respondents saw the prime, it was not subliminal anymore which might have influenced the results. Looking at the number of respondents which saw the prime, there was a difference between the field

experiment and the online survey. 40 respondents out of 254 saw the priming at the online survey, which is almost 16%. This percentage was a lot lower at the field experiment, namely 9%. The cause of this difference can be the experimental setting. The field experiment was conducted in a busy room and there was a lot of noise and distraction. The online survey was mostly completed at home or in a more quiet room. Which means that the respondents were probably less distracted at the online survey than the field experiment. The more distracted the respondents are, the smaller the change that they will process the prime.

(40)

40 Another influence might be the frequency of the priming. In this experiment the respondents were primed three times during the video. It is not clear what the influence is of how often the prime has been shown, but it might have influenced the results. Till now there is no existing research about this topic which can function as a frequency guideline. Future research should be conducted to see if results differ based on how often a prime is shown. If there is proof that the frequency of the prime shown can influence the results of a research, it is also important to find out what frequency is most influencing.

There is also another reason why the location of the field experiment can be a discussion point. As explained before, consumers react differently when they already know what the product is worth (Gneezy et al., 2012). This change in behaviour is caused by the fact that they already have a price anchor. The field experiment was done at the canteen of the

University of Amsterdam. In this specific canteen, respondents were not able to buy a cup of coffee. But in the same hallway of the canteen, there is also a cafe. In this cafe it is possible to buy a cup of coffee for €1,25. Next to that, there are also coffee machines in the hallway. These coffee machines sell a cup of coffee for €0,50. These prices might have functioned as an external reference price, which could have influenced the amount paid by the respondents. But of course to exclude all the external reference prices is very hard in every experimental set-up.

As stated before, the AT Coffee brand was created to make sure nobody would have a price anchor in their mind. But for next researches it can also be an option to look at a totally unfamiliar product, which makes it even harder for people to come up with a reference price. The external reference prices as explained above with the other coffee prices, would not be there then.

(41)

41 There was one more striking element during the field experiment. Some respondents said before paying: let’s see what I have in my wallet. Conclusions can be made that people were deciding what to pay on base of the amount of money they carried with them. If it was almost nothing, they apologized most of the time for it. Maybe people were willing to pay another price, but they did not have that amount of money in their wallet. This could have influenced the amount paid at the field experiment. Important to notice is that the location of the field experiment could have had some influence on this. At the University of Amsterdam, students can only pay with their student card or debit card. Which means that fewer students carry cash money with them. In this field experiment, student had to pay with cash. This could have influenced the results.

Looking at the distribution of the respondents, in both the online survey and the field

experiment, most of the respondents had an income between 0-20.000 euro. Income can be a big influence on the amount paid. Further research should investigate the influence of income on the relationship between subliminal priming with the Pay What You Want mechanism. Finally, there was also one proven significant difference within the different priming groups. A Chi-test has confirmed that the perception of the quality of the beans was different within the different priming groups, at the online survey. This could have influenced the results, saying that people who think the quality of the beans are high, would pay more for a cup of coffee than people who think the quality of the beans are low. Further research should be conducted to see if this have influenced the amount paid with the Pay What You Want mechanism.

Overall, subliminal priming is a complex topic that involves many variables, factors and situation cues. This research was the first research which directly combined subliminal

(42)

42 priming with the Pay What You Want mechanism. The outcomes suggested that there was no significant relationship, but further research should be conducted to see if these discussion points have influence on the outcomes.

(43)

43

Appendix

A. SS Df MS F Sig. Gender 10.917 1 10.917 20.782 .000 Error 106.635 203 .525 Total 117.552 204

Table 12. One-Way ANOVA, influence of gender on the amount paid (online survey)

SS Df MS F Sig.

Age 30.412 23 1.322 2.747 .000

Error 88.097 183 .481

Total 118.509 206

Table 13. One-Way ANOVA, influence of age on the amount paid (online survey)

SS Df MS F Sig.

Income 7.279 3 2.426 4.441 .005

Error 111.454 204 .546

Total 118.733 207

Table 14. One-Way ANOVA, influence of income on the amount paid (online survey)

SS Df MS F Sig.

Degree 15.918 5 3.184 6.241 .000

Error 102.534 201 .510

Total 118.452 206

(44)

44

SS df MS F Sig.

Nationality .850 1 .850 1.485 .224

Error 117.883 206 .572

Total 118.733 207

Table 16. One-Way ANOVA, influence nationality on the amount paid (online survey)

SS Df MS F Sig.

Quality beans 5.499 2 2.750 5.032 .007

Error 112.560 206 .546

Total 118.060 208

Table 19. One-Way ANOVA, influence of the quality of the beans on the amount paid (online survey) B. SS df MS F Sig. Gender 1.086 1 1.086 2.687 .107 Error 23.433 58 .404 Total 24.518 59

Table 21. One-Way ANOVA, influence of gender on the amount paid (field experiment)

SS df MS F Sig.

Age 8.010 16 .501 1.304 .239

Error 16.508 43 .384

Total 24.518 59

(45)

45

SS Df MS F Sig.

Income .699 2 .350 .837 .438

Error 23.819 57 .418

Total 24.518 59

Table 23. One-Way ANOVA, influence of income on the amount paid (field experiment)

SS Df MS F Sig.

Degree 3.623 4 .906 2.384 .062

Error 20.896 55 .380

Total 24.518 59

Table 24. One-Way ANOVA, influence of degree on the amount paid (field experiment)

SS Df MS F Sig.

Nationality .697 1 .697 1.697 .198

Error 23.821 58 .411

Total 24.518 59

Table 25. One-Way ANOVA, influence of nationality on the amount paid (field experiment)

SS Df MS F Sig.

Quality of the beans 1.066 3 .355 .848 .473

Error 23.452 56 .419

Total 24.518 59

Table 26. One-Way ANOVA, influence of the quality of the beans on the amount paid (field experiment)

(46)

46 C. Brand AT Coffee

(47)

47 E. Questions online survey

(48)

(49)

49 F. Primes shown in the video

(50)

How subliminal priming influences the Pay What You Want mechanism