“Persuasion to the limit: the effectiveness of the persuasion strategy social proof on the sales of condoms”

“Persuasion to the limit: the effectiveness of the

persuasion strategy social proof on the sales of

condoms”

Empirical analysis of the effect of social proof and self-esteem on the sales

of embarrassing products

By

Esmée Siebring

“Persuasion to the limit: the effectiveness of the

persuasion strategy social proof on the sales of

condoms”

Empirical analysis of the effect of social proof and self-esteem on the sales of embarrassing products Master Thesis By Esmée Siebring | S2182300 Trompsingel 15-2 | 9724 CX Groningen | +31(0)657537083 | E.Siebring@student.rug.nl University of Groningen Faculty of Economics and Business

MSc Marketing Management

Supervisor: Dr. M. Keizer External supervisor: Dr. J. Wan

PREFACE

Dear reader,

In front of you lies my MSc Marketing Management thesis, the final piece to finalize the master program at the Rijksuniversiteit Groningen. The five months it took to write this thesis have been challenging and interesting. A relative short period of time in which I have learned a lot. Moreover, the topic “persuasion to the limit: the effectiveness of the persuasion strategy social proof on the sales of condoms” provided me the opportunity to combine my interest in both marketing and psychology. The possibility to choose a consumer psychology topic, a topic in my field of interest, truly made me enjoy the process of writing this thesis.

Now, I would like to take the opportunity to thank my first supervisor Dr. M. Keizer for sharing his vision, providing me with such helpful feedback and taking the time to reply to my questions. Without his guidance this thesis would not be what it is now. Then, I would also like to thank my second supervisor Dr. J. Wan for evaluating my master thesis. Next, I would like to thank my family and friends who have always supported me during the process of writing a master thesis.

As a final remark, I wish you lots of joy and fun while reading my master thesis.

Esmée Siebring

ABSTRACT

This study is carried out because of the lack of empirical evidence regarding the influence of persuasive tactics on the sales of embarrassing products. In particular the effect of the persuasion strategies by Cialdini (2001) on the sales of embarrassing products is lacking. This current study attempted to find new ways to increase the sales of embarrassing products and focused on the influence of the persuasion strategy social proof on the sales of embarrassing products. First, this paper has discussed relevant literature about the constructs sales of embarrassing products, social proof and self-esteem. Subsequently, the methodology section has discussed the 2 x 2 between-subjects design and the experiment which has been conducted amongst one hundred twenty-three participants. The experiment has tested whether the social proof condition (as opposed to the control condition) significantly influenced the sales of embarrassing products. Also, by means of a moderation analysis it has been tested whether one’s self-esteem moderated the aforementioned relationship, and thereby, it was tested whether self-esteem had a significant effect on the relationship between social proof and the sales of embarrassing products or not. Next, the results section has discussed the results of the conducted experiment. They have partly aligned with the hypotheses, however, the two-way ANOVA and two-way ANCOVA analyses did not show any pronounced effects, which meant that no interesting conclusion could be drawn from these results. Besides, the reliability of the outcomes with regard to the measurement of self-esteem are questionable due to extremely high mean and median scores. Lastly, the author has discussed these results, their theoretical and managerial implications, and their research limitations in the discussion section of this paper.

TABLE OF CONTENTS

1. INTRODUCTION ... 6

2. LITERATURE REVIEW ... 8

2.1 Sales Embarrassing Products ... 8

2.2 Social Proof ... 9

2.3 Self-Esteem ... 14

2.4 Conceptual Model ... 17

3. METHODOLOGY ... 18

3.1 Participants and Study Design ... 18

3.2 Sample Description ... 19

3.3 Pretest ... 20

3.4 Procedure and Measurements ... 21

3.4.1. Manipulation of the principle of social proof. ... 21

3.4.2. Self-esteem measurement. ... 23

3.4.3. Measurement of the purchase intention towards the embarrassing product ... 25

3.5 Data Analysis ... 26 4. RESULTS ... 26 4.1 Manipulation Check ... 26 4.2 Hypotheses Testing ... 28 4.3 Control Variables ... 31 5. DISCUSSION... 34 5.1 Summary ... 34 5.2 Theoretical Implications... 35 5.3 Managerial Implications ... 35

5.4 Limitations and Future Research ... 37

5.5 Conclusion ... 40

6. REFERENCE LIST ... 41

1. INTRODUCTION

Once a year my high school hosted a talent show for all young musicians of the school. I have always aspired to be the host of the night ever since I started high school. It was the year 2010, my junior year of high school, a big night was ahead of me. I was finally going to host the talent show in front of a big crowd. However, the nerves took complete control of me, which resulted in an embarrassing event. Instead of introducing myself, I introduced myself as my co-host and vice versa. When realizing what just happened, I decided to correct my introduction, which led to a very awkward situation. I have never felt so embarrassed in my life before as I did in that particular moment.

What made me feel so embarrassed on that stage? Mistakes are human and should not be something we are embarrassed about. Still, this knowledge did not keep me from feeling extremely embarrassed. In general, the social evaluations of present others (i.e. present others who are judging one’s behavior in a particular situation or moment) may cause feelings of embarrassment. Especially when these social evaluations are negative. Therefore, people try to avoid these negative evaluations and uphold a steady positive public image, so that they cannot be negatively evaluated. They do so by conforming and pleasing others (Dahl, Manchanda and Argo, 2001; Edelmann and McCusker, 1986). Finally, embarrassment can be defined as “an aversive and awkward emotional state following events that increase the threat of unwanted evaluations from a real or imagined social audience” (Dahl et al., 2001: 474). The embarrassing situation described in the beginning of this chapter is related to the definition by Dahl et al. (2001: 474).

However, a gap can be found in current research on embarrassment. Embarrassment in purchase situations have not been studied much in previous research. These purchase situations can vary greatly: from different physical environments and time constraints to differences in one’s emotional state. Subsequently, the different contexts or situations might influence one’s purchase decision. Therefore, the aim of this paper is to gain insights into the effect of different purchase situations on purchase decisions with respect to embarrassing products. Next, these purchase decisions may have an influence on sales, or for the purpose of this study, sales of embarrassing products.

purchase decision in a particular purchase situation. Remarkably, previous literature mainly focused on the influence of such persuasive tactics on the sales of regular products or other consequences that such persuasive tactics might have on regular products. However, persuasive tactics might also have an influence on the sales of embarrassing products. More specifically, it might be that by changing these persuasive tactics, sales of embarrassing products might go up. An example of such persuasive tactics which may influence the sales of embarrassing products are the persuasion strategies by Cialdini (2001).

Again, previous literature solely questioned the influence of the persuasion strategies by Cialdini (2001) on regular products. Not much research can be found that questions the influence of these persuasion strategies on the sales of embarrassing products, and moreover, how these can be deployed in order to increase sales. Especially the principle of social proof might be of great importance because consumers are less embarrassed when they see that others are purchasing the embarrassing product. Moreover, this is a cue for them that this is the right thing to do as these consumers believe that others might have superior information that oneself is lacking. Next, following the majority is desired as people do not want to be different because they are afraid that they will not be socially accepted by the group. Therefore, by behaving in the same manner as the others, people will gain social approval (Cialdini and Goldstein, 2004; Mollen, Rimal, Ruiter and Kok, 2013).

Consequently, it will be tested in this research whether the persuasion strategy social proof by Cialdini (2001) can be manipulated in such a way that the sales of embarrassing products will increase. By presenting people with an advertisement including the social norm which shows the consumer behavior of others, participants will believe that this consumer behavior is the right thing to do, and thus, they will most likely follow the consumer behavior of others. As a result, participants will feel less embarrassed when purchasing the embarrassing product, and on top of that, they will feel socially accepted by the group. Therefore, it is expected that social proof positively influences the sales of embarrassing products.

self-esteem are more sensitive to social norms in purchase situations because they are often looking for social approval and ways to minimize threats to their social identity. As a consequence, it is expected that consumers with low self-esteem are most likely to be influenced by the effect of social proof as opposed to people with high self-esteem. Also, by following the consumer behavior of others, threats to their social identity will be minimized. Thus, an advertisement that includes a social norm stating the appropriate consumer behavior in a particular purchase situation has a bigger effect, meaning higher purchase intentions towards the embarrassing product, on consumers with low self-esteem as opposed to consumers with high self-esteem. Therefore, examining the effect of self-esteem on the relation between social proof and sales of embarrassing products might be of great relevance to this research as well.

In conclusion, this paper extends existing literature by providing new ways to stimulate the sales of embarrassing products, by focusing on the influence of persuasive tactics such as the persuasion strategy social proof (Cialdini, 2001) on the sales of embarrassing products and whether this possible effect is moderated by one’s level of self-esteem.

2. LITERATURE REVIEW

As mentioned in the introduction, this research focuses on the influence of social proof on the sales of embarrassing products moderated by one’s level of self-esteem. In this paragraph the variables and the possible relationships will be discussed in depth.

2.1 Sales Embarrassing Products

Experiencing embarrassment when purchasing embarrassing products in a retail environment may have consequences for the sales of embarrassing products. Embarrassment is an emotion which is negatively evaluated, and people try to avoid this emotion at all times by means of developing coping strategies or by avoiding this embarrassing purchase situation completely, meaning that no embarrassing products will be purchased in the end (Modigliani, 1968; Nichols, Raska and Flint, 2014; Blair and Roese, 2013). Therefore, in order to increase the sales of embarrassing products, marketing managers should come up with new ways to limit these feelings of embarrassment, so that it will not affect the sales of embarrassing products. But first the concept embarrassment will be discussed into detail.

tremor (Modigliani, 1968). It is experienced as a negative emotion, which negatively affects one’s social identity. In order to avoid embarrassment, people avoid situations that threaten one’s social identity (Modigliani, 1968). Thus, people aim to uphold a positive self-concept and avoid situations that threaten their social identity.

It should be noted that the degree to which one feels that their social identity is threatened, and embarrassment is experienced, differs per person or personality. According to Modigliani (1968) some people are more susceptible to embarrassment than others. This depends on (1) one’s susceptibility towards evaluations of others, and (2) one’s immediate assumption that those evaluations are more negative than they actually are.

But how can embarrassment be defined? Many definitions can be found in literature. According to VandenBos (as cited in Krishna, Herd, Aydinoglu, 2015: 474) embarrassment can be defined as “a self-conscious emotion in which a person feels awkward or flustered in other people’s company or because of the attention of others, as, for example, when being observed engaging in actions that are subject to mild disapproval from others”. Edelmann and McCusker (1985: 133) define embarrassment as “a self-presentational difficulty resulting from a concern with our observable behavior and a desire to conform and please others”. Tangeny (as cited in Tsao and Chan, 2011: 504) elaborates on this definition and states that “embarrassment is an awkward feeling aroused as a result of an individual’s behavior that fails to gain social acceptance, but it is not morally condemned”. Last but not least, Dahl et al. (2001: 474) define embarrassment as “an aversive and awkward emotional state following events that increase the threat of unwanted evaluation from a real or imagined social audience”.

These definitions and concepts explained above all describe the concept of embarrassment in its broadest sense, whereas, in my opinion, the definition by Dahl et al. (2001: 474) describes general embarrassment in the most complete manner as it emphasizes the causes of embarrassment. More specifically, the definition encloses feelings of embarrassibility as a result of negative evaluations of others and one’s desire to gain social approval, which are the center pieces in experiencing embarrassment. Therefore, this definition will be adopted in this paper.

have interesting implications for marketers who want to increase the sales of embarrassing products. Subsequently, embarrassing products can then be defined as “products people need and seek out, but do not discuss openly because it evokes aversive feelings and feelings of awkwardness resulting from purchasing such goods, which are subject to disapproval from others” (Katsanis, as cited in Londono Roldan, 2013; VandenBos, as cited in Krishna, Herd, Aydinoglu, 2015: 474). This definition is most complete as it takes into account the essential parts that describe general embarrassment while defining products that makes one feel embarrassed.

An example that may clarify this definition of embarrassing products is the example by Nichols et al. (2014). Their study showed that college students felt embarrassed when purchasing condoms in a retail environment as this threatened their social identity and indicated that one is promiscuous. Being promiscuous is negatively evaluated and socially disapproved by others, hence, college students want to avoid this image of being promiscuous.

In conclusion, in order to increase the sales of embarrassing products, marketing managers should develop coping strategies or find other ways to limit threats to one’s social identity. In case these threats are limited, embarrassment will not be experienced, and thus, barriers to purchase an embarrassing product are eliminated.

2.2 Social Proof

Using social proof might be a very effective way to increase the sales of embarrassing products. Here, social proof is defined as “a heuristic by which behavior is viewed as correct in a given situation to the degree to which we see others performing it” (Rao, Greve and Davis, 2001: 504). Thus, if consumers are seeing that other consumers are purchasing embarrassing products, it is a cue for them that this is the right thing to do as consumers do not want to be different. More specifically, others will not disapprove of this consumer behavior (Mollen, Rimal, Ruiter and Kok, 2013). Thus, consumers do not have to feel embarrassed about purchasing this embarrassing product as they now know that other consumers have bought the embarrassing product before them. Besides, consumers will achieve the goal of accuracy as well, as they behaved in the correct way according to the social norm in the given purchase situation, and moreover, gained social approval because of behaving in that particular way (Cialdini and Goldstein, 2004; Mollen et al., 2013). This feeling is desired as people want to be part of the group and gain social approval of this group.

Let us now dive in the concept of social proof by studying the following studies. These studies discuss examples of social proof experiments with regards to the sales of regular products. However, these studies might still have implications for my research as well. These studies illustrate that social proof cues have an effect on consumer’s willingness to purchase regular products. For this current paper it might mean that once consumers learn (i.e. by means of an advertisement that shows the consumer behavior of others) that others bought the embarrassing products, consumers should be more motivated to buy the embarrassing products themselves as well.

Another study by Mollen et al. (2013) examined the effects of social proof on unhealthy and healthy food choices at an on-campus food court. Here, a healthy food choice (i.e. a salad) and an unhealthy food choice (i.e. a hamburger) were offered at the same price, so that the price did not intervene in the results of the experiment. To test whether social proof had an influence on a particular food choice, an advertising message including a descriptive social norm (i.e. a message showing what most other consumers do) was presented to a consumer. It was hypothesized that a social proof advertising message including a descriptive social norm promoting the healthy food choice would have a positive influence on the sales of the healthy food choice. This means that the social proof advertising message stated that most people chose the salad as opposed to the hamburger. The results aligned with this hypothesis, meaning that consumers who were exposed to an advertisement stating the descriptive social norm purchased the healthier food choice more often as opposed to consumers who were not exposed to the social proof advertisement.

Note that social proof has already been widely examined in prior research. Cialdini (2007) states that consumers are looking at the opinions, decisions and behaviors of others to find out what is appropriate behavior in a certain situation. Cialdini, Wosinka, Barret, Butner and Gornik-Durose (2012: 1243) agree with this statement by Cialdini (2007) and define social proof as “a validation of the correctness of one’s (purchase) decisions and behavior in comparison to others”. Therefore, consumers are often following the behavior of, for example, friends or family in purchase situations (Cialdini et al., 2012). In this paper I adopt the definition of social proof by Rao et al. (2001: 504) who define social proof as “a heuristic by which behavior is viewed as correct in a given situation to the degree to which we see others performing it”.

Subsequently, making a decision based on a heuristic, in this case, the principle of social proof, reduces search costs but may also lead to erroneous judgements, overvaluation on one’s product choice and post-purchase disappointment (Rao et al., 2001). Namely, decisions are often made because one believes that others made a decision based on some superior information that oneself is missing (Rao et al., 2001). However, decision making of a group is often based on reactance rather than on superior information, meaning that a situation evolves in which everyone believes that decisions are based on some superior information others had, while the truth is that decisions are only made by simply following the behavior of others, without knowing how one came to that particular decision. Especially, consumers that feel insecure may rely on these heuristics as a shortcut for decision making (Rao et al., 2001).

If insecure consumers use social proof as a heuristic, they basically rely on a social norm, which can be defined as “a given guideline or behavioral expectation within a group or society that shape what is deemed normal or desirable” (Frederiks, Stenner and Hobman, 2015: 1387). Furthermore, previous literature by Cialdini and Goldstein (2004) already stated that consumers are eager to conform to the consumer behavior of others in order to enhance, protect or repair their self-esteem. Subsequently, marketers often use these social norms to evoke certain consumer behavior. For example, by presenting consumers with a positive social norm (e.g. desirable behavior in a purchase situation) consumers will shift their behavior towards this positive social norm because that behavior is deemed desirable in that particular situation (Frederiks et al., 2015). An example of such a positive social norm has been illustrated by an advertisement of a low-fat cheese in a study by Salmon et al. (2015). Here, marketing managers placed banners at the point-of-sale in a supermarket stating that the low-fat cheese was the most sold cheese in the assortment. By placing the banners at the point-of-sale, the marketing managers wanted to shift its consumers towards a certain behavior (i.e. choose this type of low-fat cheese).

provoke certain consumer behavior. Hence, by showing a descriptive social norm, marketers want consumers to conform to the desirable behavior (i.e. the behavior illustrated by the positive descriptive social norm) (Frederiks et al., 2015).

In conclusion, marketers can influence consumer behavior by using the principle of social proof in their advertising messages to achieve a certain goal (Cialdini, 2007). Marketers may persuade consumers by using a descriptive social norm to which one wants to conform (Cialdini and Goldstein, 2004). Previous research already stated that consumers are eager to conform to such a descriptive social norm in order to enhance, protect or repair their self-esteem (Cialdini and Goldstein, 2004). Therefore, gaining insights in the principal of social proof might have implications for marketing managers. By deploying the principle of social proof in their advertising messages and letting consumers rely on heuristics, marketing managers might be able to stimulate and achieve a particular goal. For this research, this means realizing an increase in the sales of embarrassing products. Therefore, it is expected that once consumers have learned that others bought the same embarrassing product, they are more motivated to buy the embarrassing product as well. Stated formally:

H1: The social proof condition (advertising message informing consumers about the consumer

behavior of others) as opposed to the control condition has a positive influence on the sales of embarrassing products.

2.3 Self-Esteem

Therefore, high self-esteem enhances advertisement effectiveness, and this will ultimately increase sales (Fennis and Stroebe, 2016). But what is self-esteem exactly?

Self-esteem is a widely discussed topic, which is not an unambiguous concept. According to Blascovich and Tomaka, Rosenberg, Schooler, Schoenbach and Rosenberg (as cited in Von Soest, Hansen, Wagner and Gerstorf, 2018: 945) self-esteem can be defined as “an individual’s general attitude toward or evaluation of the self and reflects people’s beliefs about how worthy they are as a person”. On the other hand, Rosenberg (as cited in Akhtar and Shaukat, 2016: 277) defines self-esteem as “an individual’s beliefs, thoughts, and feelings about oneself, one’s importance, and one’s self-worth. The latter definition as described by Rosenberg (as cited in Akthar and Shaukat, 2016: 277) is considered to be the most complete definition of self-esteem and will therefore be adopted in this paper.

Besides differences in levels of self-esteem (i.e. some people possess low self-esteem, whereas others have high self-esteem), Modigliani (1968: 315) acknowledged different sorts of self-esteem, which can be defined as follows: (1) “the individual’s own conception of his situational self-esteem, called situational-self-esteem (SSE)”, (2) “the conception which others present have of his situational self-esteem, called situational-public-esteem (SPE)” and (3) “his own perception of his situational-public-self-esteem, called situational-subjective-public-esteem (SSPE)”. Modigliani (1968) states that, if one experiences differences in the SPE as compared to the SSPE, feelings of embarrassment arise. Hence, a loss in SSPE leads to a loss in the SSE, which ultimately results in feelings of embarrassment (Modigliani, 1968). Tangney, Miller, Flicker and Barlow (1996) agree with the rather complicated explanation by Modigliani (1968) and state that a temporary loss of self-esteem is caused by experiencing deficiencies in one’s presented self, resulting in feelings of embarrassment. This occurs following several stages. Firstly, one should be violated by the occurrence of unpleasant events. Subsequently, the social presence of others during the occurrence of these unpleasant events should lead to a loss in SSPE. Finally, a loss in SSPE leads to a loss in the SSE, which ultimately results in feelings of embarrassment (Modigliani, 1968).

Subsequently, according to Edelmann and Modigliani (in Keltner and Buswell, 1997) these people as described by Modigliani (1968) who on top of their low self-esteem also experience an acute disappointment with the self or fail to act in accordance with personal standards are more likely to experience embarrassment as opposed to people with high self-esteem who experience the same disappointment with the self or fail to act in accordance with personal standards. This matches with the different levels of situational-self-esteem (i.e. high vs low) resulting in different beliefs, attitudes or consumer behavior.

Subsequently, high self-esteem might then have implications for marketers. A research by Hong and Zinkhan (1995) has shown that a positive self-concept influences one’s purchasing behavior, meaning that advertising or the message a business wants to convey should be congruent with one’s self-concept in order to establish effective advertising (i.e. establishing brand memory, creating brand attitude and increasing purchase intentions in the pre-purchase stage). Here, self-concept is defined as “the totality of the individual’s thoughts and feelings having reference to himself as an object (Rosenberg, as cited in Hong and Zinkhan, 1995: 248)”. This definition of self-concept is rather similar to the definition of self-esteem by Rosenberg (as cited in Akthar and Shaukat, 2016: 277), which is adopted in this paper. Then, the positive effect of self-concept on advertising effectiveness appears to be similar for a positive self-validation, or more broadly spoken, high self-esteem. This means that high self-esteem reflects the ability of a consumer to be able to evaluate a persuasive advertising message. Therefore, high self-esteem enhances advertisement effectiveness and ultimately increases sales (Fennis and Stroebe, 2016).

Drolet, 2008). On the other hand, consumers who have high self-esteem do not have such a desire to conform to the behavior of others as compared to the desire consumers with low self-esteem have (Arndt et al., in Cialdini and Goldstein, 2004). Furthermore, it is expected that consumers with high self-esteem are less susceptible to the evaluations of others, and thus, less susceptible to embarrassment (Modigliani, 1968). Thus, as they acknowledge less threats to their social identity, people with high self-esteem are more likely to participate in this risky consumer behavior and buy the embarrassing product.

Hence, consumers with high self-esteem as opposed to consumers with low self-esteem are more likely to purchase embarrassing products. Stated formally:

H2: High self-esteem as opposed to low self-esteem has a positive influence on the sales of

embarrassing products.

Additionally, feelings of insecurities and uncertainty are more likely to result in behavior consistent with the social proof principle, meaning that people with low self-esteem are more likely to follow or conform to the behavior of others (Cialdini, 2007). By conforming to the behavior of others, people hope to gain social approval of others and, thereby, reduce the threat to their social identity. Thus, conforming to the behavior of others, ultimately, positively affects one’s self-esteem. People go at great length to conform to the behavior of others to restore their self-esteem and social identity, even if one is not personally, directly or publicly rejected by others (Cialdini and Goldstein, 2004). Cox and Bauer (in Kropp, Lavack and Silvera, 2005) agree with this statement and posit that consumers with low self-esteem are more likely to be influenced by the decisions and opinions of others. Thus, in order to be able to make the right purchase decision, consumers with low self-esteem are more likely to be persuaded by the behavior of other consumers. Applied to the current study, this implies that:

H3: An advertising message informing a consumer about the consumer behavior of others has the

biggest effect on consumers with low self-esteem, which ultimately results in an increase in sales of embarrassing products.

The literature review discussed the theoretical concept behind the variables and the hypotheses showed why the expected relations among the dependent variable, independent variable and the moderator may exist. Figure 1 is a graphical representation of the relations among these variables. It shows that an advertising message in the social proof condition, which can be defined as “a validation of the correctness of one’s (purchase) decisions and behavior in comparison to others” (Cialdini et al., 2012), ultimately leads to an increase in sales of embarrassing products, which can be defined as “ products people need and seek out but do not discuss openly because it evokes aversive feelings and feelings of awkwardness resulting from purchasing such goods, which are subject to disapproval from others” (Katsanis, as cited in Londono Roldan, 2013; VandenBos, as cited in Krishna, Herd, Aydinoglu, 2015: 474). The moderator in this relationship, that is one’s self-esteem, affects the relationship between the principle of social proof and the sales of embarrassing products directly (Baron and Kenny, 1986).

3. METHODOLOGY

3.1 Participants and Study Design

The aim of this study was to assess the influence of both the independent variable (i.e. social proof) and the moderator (i.e. one’s self-esteem) on the dependent variable (i.e. sales of embarrassing products). The present study had a 2 (social proof: an advertising message showing the behavior of others vs an advertising message not showing the behavior of others) x 2 (self-esteem: high vs low) between-subjects design, in which participants were randomly assigned to the social proof condition or the control condition. Furthermore, the purchase intention towards the embarrassing product and one’s level of self-esteem were measured. Here, self-esteem was

measured by means of a self-esteem scale which indicated a participant’s level of self-esteem. Afterwards, participants were divided into two conditions (a high self-esteem condition vs a low self-esteem condition) based on the median score of the participants altogether. The study was designed with the survey platform Qualtrics.

Consumers in the Netherlands were asked to participate in this study. Participants were mainly recruited via social media (e.g. Facebook and LinkedIn) and WhatsApp (see appendix A). All reside in the Netherlands, participated on voluntarily basis and met the age requirement. Here, participants were required a minimum age of eighteen years, because for the purpose of this study it is assumed that all people above eighteen earn their own money and are responsible for making their own purchase decisions in a retail environment.

3.2 Sample Description

Two hundred two people participated in this study. However, forty-five participants (22% of the total sample) were excluded from the survey because they either did not complete the survey (n = 44) or did not meet the age requirement (n = 1). One hundred fifty-seven participants remained (70 men, 87 women; Mage = 32.9, SDage = 14.2) and were included in further analyses. Nevertheless,

in the end, another thirty-four participants were excluded from subsequent data analysis due to failing the manipulation check in the Qualtrics survey, meaning that it is assumed that they did not recognize the consumer behavior of previous others (n = 30) or indicated a percentage below the threshold of 80% (n = 4). The failure of the manipulation check will be further explained in the results section.

In the end, one hundred twenty-three participants (51 men, 72 women; Mage = 31.3, SDage =

12.5) were included in subsequent analyses from which forty-five participants were randomly assigned to the social proof condition and seventy-eight participants were randomly assigned to the control condition. Subsequently, the median of the Rosenberg Self-Esteem scale served as a decision point and assigned respondents right after their participation to either a low or high self-esteem condition. Thus, sixty-five participants were assigned to the low self-self-esteem condition, whereas fifty-eight participants were assigned to the high self-esteem condition.

3.3 Pretest

Firstly, a pretest was conducted in order to come up with empirical evidence that the used product during the main study would be perceived as embarrassing. Eight products were chosen for this pretest based on previous literature in which it was stated that these products were perceived as embarrassing. Hence, the products used during the pretest were the following: (1) condoms, (2) adult diapers, (3) adult magazines, (4) laxatives, (5) douche, (6) lice shampoo, (7) undergarment and (8) hemorrhoid medication (Lau-Gesk and Drolet, 2008; Jones et al., 2017; Dahl et al., 2001; Moore et al., 2008).

Subsequently, consumers - acquired via my own social network (i.e. WhatsApp) - were asked to imagine themselves to be in a purchase situation (i.e. their favorite drugstore) having a strong purchase intention towards an embarrassing product. Next, they had to indicate which embarrassing product would make them feel most embarrassed by ranking the aforementioned eight embarrassing products (1 = “most embarrassing product” to 8 = “least embarrassing product”). Twenty-one people participated in this pretest. From the results it can be concluded that purchasing hemorrhoid medication is perceived as most embarrassing (i.e. seven participants indicated that hemorrhoid medication was the most embarrassing product).

product mentioned in the pretest, due to the necessity of the medication. Hence, the results might be misleading. Consequently, it is decided to use the product that is perceived as the second most embarrassing product, which are condoms. Six participants indicated that condoms were the most embarrassing product.

3.4 Procedure and Measurements

The study began with a brief introduction about the purpose of this study. Here, it is explained that the study consists of two independent research parts, each belonging to different research topics for different graduation projects. The first research topic was part of the graduation project of the fictitious person Els Janssen with questions regarding demographic information (e.g. gender, age, education, occupation and annual income before taxes) and one’s self-esteem. The second research topic was part of my own graduation project for which I used my own name. This part covered questions regarding one’s purchase intention towards the embarrassing product shown in an advertisement. By splitting the research into two different parts the participants were manipulated in such a way that the dependent variable (i.e. sales of condoms) is not dependent on one’s level of self-esteem. In this manner the risk of contamination will be limited.

Then, measuring one’s self-esteem was done by means of using the Rosenberg Self-Esteem (RSE) scale, which consisted of ten different items regarding one’s self-esteem. Subsequently, in the second part of the survey, participants were randomly assigned to one of the two social proof conditions each showing different advertisements (i.e. an advertisement showing the consumer behavior of others as opposed to a neutral advertisement not showing the consumer behavior of others). In this manner, participants were manipulated to be able to infer whether participants would comply to the principle of social proof or not. Then, participants were asked to indicate the likelihood of purchasing the embarrassing product by measuring their purchase intention right after being confronted with the advertisement. Lastly, a manipulation check was included to measure their awareness of and the extent to which social proof influenced their purchase decision. At the end of the questionnaire participants were thanked for participating in this study (see appendix B and C).

advertisement in the social proof condition should elicit the same behavior as has been illustrated in the advertisement, meaning that, showing the consumer behavior of other people (i.e. buying the embarrassing product) should increase the purchasing likelihood of the participant in the social proof condition. The following advertisements have been used:

2. Control condition:

In order to prepare the collected data for any statistical analyses, a dummy variable for the independent variable social proof was created. The dummy variable “social proof” indicated the condition the participant was in (1 = “social proof condition”; 2 = “control condition”). By distinguishing whether a participant was in the social proof or control condition, the data was now fit to perform statistical analyses, so that the effect of social proof on any variable can be examined. Thus, the variable “social proof” is being used in further analyses as the independent variable.

3.4.2. esteem measurement. To measure one’s self-esteem, the Rosenberg

students to adults). The RSE scale consists of ten items measured on a 4-point Likert-scale (α = 0.85, M = 3.64, SD = 0.43). Sample items of the RSE scale are: “on the whole, I am satisfied with myself”, “at times I think I am no good at all” and “I feel that I have a number of good qualities”. In order to maximize the response rate and, on top of that, force participants to have an opinion about all items, it is decided to keep the measurement scale rather simple and not present a neutral option. Hence, the items are measured on a 4-point Likert-scale (1 = “strongly disagree” to 4 = “strongly agree”), which should also clearly distinct low self-esteem (e.g. strongly disagree and disagree) from high self-esteem (e.g. strongly agree and agree). In order to increase the internal consistency of the RSE scale, five items were recoded into positively formulated items instead of negatively formulated items. As the Cronbach’s Alpha shows above, this had led to an internally consistent self-esteem measurement scale. The recoded items are the following: “at times I think I am no good at all”, “I feel I do not have much to be proud of”, “I certainly feel useless at times”, “I wish I could have more respect for myself” and “all in all, I am inclined to feel that I am a failure” (see appendix D).

Nevertheless, to be able to say something about the effect of different levels of self-esteem on the purchase intention towards embarrassing products, different groups should be created. Consequently, the median was used to assign participants to groups with either low self-esteem (n = 65; Mdn < 3.6) or high self-esteem (n = 58; Mdn > 3.6). Here, low self-esteem means lower than the median rather than participants truly having low self-esteem. Note that the impact of these relatively high self-esteem scores will be discussed in the discussion section.

3.4.3. Measurement of the purchase intention towards the embarrassing product. As the

experiment is conducted in an online survey it is not possible to actually sell embarrassing products, and thus, it is hard to measure actual sales. Measuring one’s purchase intention towards embarrassing products is an indicator of one’s likelihood to actually purchase and is therefore the most relevant measure to measure potential sales. In order to measure one’s purchase intention towards the embarrassing product, participants in the social proof condition and control condition were asked on a 5-point Likert-scale to indicate their likelihood of purchasing the embarrassing product. Items of the purchase intention scale were: “the advertisement convinced me to purchase condoms”, “I am planning to purchase condoms in the near future” and “I am sure to purchase this type of condoms”. It is decided to keep the measurement scale as simple as possible. Hence, the items were measured on a 5-point Likert-scale (1 = “extremely unlikely” to 5 = “extremely likely”), which should also minimize a participant’s frustration level. Next, the three items were combined and recoded into different variables, which illustrated the mean scores of the purchase intention per participant per condition and were called “Socialproof_condition_DV” (α= 0.83, M = 2.42, SD = 0.85) and “Control_condition_DV” (α = 0.90, M = 2.41, SD = 0.91), referring to the purchase intention of participants in either the social proof condition or control condition.

3.5 Data Analysis

The present study had a 2 (social proof: an advertising message showing the behaviors of others vs an advertising message not showing the behavior of others) x 2 (self-esteem: high vs low) between-subjects design. Three hypotheses are being tested in the context of this full model (see literature review, Figure 1). The effect of the independent variable social proof on the dependent variable sales has been tested by means of a one-way ANOVA analysis to test whether the group means significantly differ from each other. On the other hand, the effect of the moderator self-esteem and the interaction effect between the independent variable and the moderator has been tested by means of a two-way ANOVA analysis to test if there is a significant difference between people with low or high self-esteem in how they respond to the principle of social proof in terms of purchase intention with regard to embarrassing products.

Next, two covariates were included in a two-way ANCOVA analysis to increase the accuracy of the results. It might be that one’s annual income before taxes over 2017 and the manipulation check, which takes the influence of the consumer behavior of others into account, influence the dependent variable sales of embarrassing products. By including these covariates in a two-way ANCOVA these possible effects are being controlled for, and thus, led to more reliable results.

4. RESULTS 4.1 Manipulation Check

However, as mentioned in the beginning of the methodology chapter, it was decided to exclude thirty-four participants in the social proof condition from subsequent analyses. From this number, thirty participants indicated the option “unknown” when asked for the percentage of people that purchased this type of condoms before them, whilst the advertisement showed a percentage of 83%. Hence, these participants did not remember the percentage of people that purchased this type of condoms before them. Besides, four participants indicated a percentage below the threshold of 80% (i.e. as the correct percentage was 83%, it was decided that all indicated percentages below 80% were considered as too different), meaning that the indicated percentage was too deviant. This could be due to several reasons. For example, these people did not see the social norm at all in the advertisement or they simply could not remember the exact percentage and chose to tick the “unknown option” or may have felt pressure to indicate a random percentage.

Next, it was assumed that these thirty-four participants did not pay close attention to the presented social norm, and thus, were not able to recall the exact percentage of people that purchased this type of condoms before them. This led to the decision to exclude these participants from any subsequent analyses because the objective was to examine whether social proof influenced one’s purchase intention. Hence, a participant should be aware of the presented social norm in a given purchase situation in the first place. As these participants could not recall the exact percentage of people who purchased this type of condoms before them, it was assumed that they did not paid close attention to the advertisement, which resulted in not being able to recall the exact percentage. Therefore, it can be questioned whether they have been influenced by the social norm in the advertisement at all, and thus, to limit the risks of including unreliable results, these people have been excluded from subsequent analyses. Ultimately, one hundred twenty-three participants (51 men, 72 women; Mage = 31.30, SDage = 12.50) were included in further analyses.

extent to which a participant believes that the consumer behavior of others influenced their own purchase decision.

4.2 Hypotheses Testing

By means of a 2 (social proof: an advertising message showing the behavior of others vs an advertising message not showing the behavior of others) x 2 (self-esteem: high vs low) ANOVA, it was tested whether the consumer behavior of others, or more specifically, the principle of social proof, had an influence on the sales of embarrassing products and whether this relationship is moderated by one’s self-esteem. The results of the statistical tests will now be discussed per hypothesis.

H1: The social proof condition (advertising message informing consumers about the consumer

behavior of others) as opposed to the control condition has a positive influence on the sales of embarrassing products.

Beforehand, it was hypothesized that once consumers learned that others bought the same embarrassing product, they were more motivated to buy the embarrassing product as well. Therefore, the effect of the independent variable social proof on the dependent variable sales of embarrassing products was examined by means of a one-way ANOVA analysis to test whether people who were assigned to the social proof condition and were exposed to an advertisement showing the consumer behavior of others significantly differed from people in the control condition who were exposed to a neutral advertisement which did not show the consumer behavior of others.

However, this test did not show a statistically significant main effect between the two different experimental conditions (F(1, 121) = 0.00, p = .94), and thus, did not align with previous findings in literature. Moreover, the means of both groups are more or less similar and neither align with previous findings in literature or the above stated hypothesis (Msocialproof = 2.42, SDsocialproof =

0.85; Mcontrol = 2.41, SDcontrol = 0.91). Thus, there is not enough evidence to infer that people who

H2: High self-esteem as opposed to low self-esteem has a positive influence on the sales of

embarrassing products.

Prior research has shown that people with high self-esteem as compared to people with low self-esteem do not have such a desire to conform to and please others (Arndt et al., in Cialdini and Goldstein, 2004). Moreover, it is expected that they are less susceptible to negative evaluations of others, and ultimately, embarrassment (Modigliani, 1968). Therefore, they are more likely to participate in risky consumer behavior, such as purchasing an embarrassing product. Also, prior literature stated that high self-esteem positively influences one’s purchasing behavior (Hong and Zinkhan, 1995; Fennis and Stroebe, 2016). Hence, it was hypothesized that people with high self-esteem are more likely to purchase embarrassing products. Subsequently, the Levene’s Test for Equality of Variances shows that the variances between both groups are equal (F(3, 119) = 1.17, p = .33). Consequently, a two-way ANOVA, which premises the homogeneity of the variances, was performed.

However, the results do not show a significant main effect for the effect of different levels of self-esteem on the sales of embarrassing products (F(1, 119) = 0.19, p = .67). Nevertheless, the means of both groups show a slight difference in the expected direction, meaning that people with high self-esteem (M = 2.48, SD = 0.86) are more likely to purchase embarrassing products, as opposed to people with low self-esteem (M = 2.35, SD = 0.91). However, the difference between the group means do not significantly differ from each other. Hence, there is not enough evidence to infer that high self-esteem results in a greater purchase intention towards embarrassing products, and thus, I cannot reject the null-hypothesis stating that there is no difference in sales based on one’s self-esteem.

H3: An advertising message informing a consumer about the consumer behavior of others has the

Not only have the main effects been tested, as described by the first two above stated hypotheses, but also the interaction effect by means of a moderation analysis (Figure 2). The results do not align with the expectation that

social proof has the biggest effect on people with low self-esteem as the results show that social proof actually has the biggest effect on people with high self-esteem who have not been exposed to an advertisement showing the social norm (M = 2.57, SD = 0.15). This partly aligns with previous literature, which stated that high self-esteem positively influences one’s purchasing behavior and that people with high self-esteem are less susceptible to

the negative evaluations of others, embarrassment and are also less willing to conform to and please others (Arndt et al., in Cialdini and Goldstein, 2004; Hong and Zinkhan, 1995; Fennis and Stroebe, 2016; Modigliani, 1968). However, it can be also said that the results of this thesis partly align with the aforementioned hypothesis as people with low self-esteem, who have been exposed to an advertisement showing the social norm (M = 2.49, SD = 0.19), have higher purchase intentions towards embarrassing products as opposed to people with high self-esteem who have also seen an advertisement with the social norm (M = 2.35 SD = 0.19), which carefully suggest an interaction effect.

Nevertheless, there is not enough evidence to infer that there exists an interaction effect between the principle of social proof and one’s self-esteem, as the effect is not significant (F(1, 119) = 1.67, p = .20). This implies that there is not a significant difference between people with low or high self-esteem in how they respond to the principle of social proof in terms of purchase intention with regard to embarrassing products. The means of the four different groups do not significantly differ from each other, and thus, it is not possible to draw interesting conclusions from these results.

4.3 Control Variables

The results from the two-way ANOVA do not show any significant effects, meaning that the results cannot be interpreted in a meaningful manner. In this case it might be that other factors exists that may influence the relationship between social proof and the sales of embarrassing products. These factors, which are called covariates, are not particularly interesting for the purpose of this research but might somehow disturb this assumed relationship. Consequently, these factors should be controlled for by means of performing a two-way ANCOVA analysis. The two-way ANCOVA analysis increases precision of the test results, removes sources of variance in the dependent variable (i.e. the sales of embarrassing products) due to uncontrolled factors and account for systematic differences across treatment groups (i.e. participants who have been exposed to the advertisement with the social norm and participants who have been exposed to the advertisement without the social norm) that were not controlled for in the experimental design.

For this research, uncontrolled factors that may serve as covariates are one’s annual income before taxes over 2017 and the manipulation check. Hence, the two-way ANCOVA analysis compares the mean values of a response variable between groups (i.e. participants who have seen the advertisement with the social norm and participants who have seen the advertisement without the social norm) when the response variable covaries with other continuous variables (i.e. one’s annual income before taxes over 2017 and the manipulation check).

Firstly, participants were asked for their annual income before taxes over 2017 by means of a multiple-choice question, in which the participant had to choose one of the six different income groups, ranging from €0 – €10,000 per year to over €90,000 per year. Feelings of awkwardness are limited by providing the participant with several income groups rather than asking them to write down their exact income per year. In this manner, one feels more comfortable to answer, meaning that one is more likely to answer a question about one’s income in the form of a multiple-choice question rather than an open question as it remains rather vague how much one earns over a year. Secondly, a manipulation check has been performed by means of a 5-point Likert-scale (1 = “strongly disagree” to 5 = “strongly agree”; M = 2.37, SD = 0.93) asking participants to what extent they believed that the consumer behavior of others influenced their own purchase decision.

are equal have been complied with (F(3, 117) = 0.733, p = .54), which is a necessary condition in case covariates are being added to the model. This means that there is not enough evidence to infer that the variances are unequal, and thus, the null-hypothesis stating that the variances of both groups are equal cannot be rejected.

Next, the results of the two-way ANCOVA may be interpreted as the assumptions of the Levene’s Test of Equality of Error Variances have been complied. At the confidence level of 95%, the covariate “income” shows a nearly significant positive effect on the sales of embarrassing products (F(1, 115) = 3.67, p = .058), meaning that a higher income results in a higher purchase intention towards embarrassing products across the two different treatment groups equally. This might occur because people with a higher income simply have more money to spend as opposed to people with a lower income. The other covariate, manipulation check, shows a significant positive effect on one’s purchase intention towards embarrassing products across the different treatment groups equally (F(1, 115) = 10.66, p = .00). This means that if one is more aware of the influence of the consumer behavior of others on one’s own consumer behavior, one will indicate a higher purchase intention towards embarrassing products. This is rather odd, as participants in the control condition should not necessarily indicate a high reliance on the consumer behavior of others. It could be that participants in this survey felt that it was socially desirable to state that one is highly influenced by the consumer behavior of others when making consumer behavior decisions for their own. Additionally, it might be that participants already had the knowledge that most people are influenced by the consumer behavior of others when making consumer behavior decisions for their own and therefore indicated a higher score. Besides, it might also be that people who already believe that the consumer behavior of others influences one’s own consuming behavior are instantly looking for social proof cues in their environment and, therefore, indicate higher purchase intentions towards embarrassing products.

result on testing the main and interaction effect of the model. Now, the results of the main and interaction effect will be discussed into detail.

The insignificance of the first main effect means that the first hypothesis cannot be accepted. Being informed by the social norm in a given purchase situation (M = 2.38, SD = 0.85) does not have a significant different effect on one’s purchase intention towards embarrassing products as opposed to people who are not being informed about the social norm in a given purchase situation (M = 2.41, SD = 0.91). As with the results of the one-way ANOVA analysis, the means of both groups are not following the expected direction, meaning that the means of both groups are rather similar. Moreover, the difference in means of both groups do not significantly differ (F(1, 115) = 0.85, p = .36), and thus, these results cannot be interpreted in a meaningful manner.

Next, the second hypothesis cannot be accepted either. People with high self-esteem (M = 2.46, SD = 0.86) do not have a significant greater purchase intention with regards to embarrassing products as opposed to people with low self-esteem (M = 2.34, SD = 0.91). Thus, the group means of both groups do not significantly differ from each other (F (1, 115) = 0.02, p = .90) and thus, these results cannot be interpreted in a meaningful manner.

Finally, there is not a significant difference between people with low or high self-esteem in how they respond to the principle of social proof in terms of purchase intention with regard to embarrassing products (Figure 3). This means that the interaction effect between the variables remains insignificant as well (F(1, 115) = 2.68, p = .10). The results of the two-way ANCOVA analysis neither align with the

expectation that social proof has the biggest effect on people with low self-esteem since the results show that social proof has the biggest effect on people with high self-esteem in the control condition (M = 2.60, SD = 0.14), and thus, people who have not been exposed to an advertisement with the social norm. This aligns with

people with high self-esteem are less likely or susceptible to conform to and please others, social approval and ultimately embarrassment (Arndt et al., in Cialdini and Goldstein, 2004; Modigliani, 1968). Moreover, high self-esteem positively influences one’s purchasing behavior (Hong and Zinkhan, 1995; Fennis and Stroebe, 2016). Besides, the results of this current research show that people with low self-esteem ultimately have higher purchase intentions towards embarrassing products (M = 2.43, SD = 0.18) as compared to people with high self-esteem (M = 2.18, SD = 0.19). Moreover, these means show greater differences between the different groups than the mean differences of the two-way ANOVA analysis, which means that the suggestion for an interaction effect is stronger in this analysis due to the inclusion of the two covariates in the two-way ANCOVA analysis. Nevertheless, no interesting conclusions can be drawn from these results as these results do not significantly differ from each other.

In conclusion, the results of both the two-way ANOVA as well as the ANCOVA analysis appear not to be significant, meaning that all drawn conclusions cannot be interpreted in a meaningful manner.

5. DISCUSSION 5.1 Summary

The purpose of this research was to find out whether the principle of social proof had a significant influence on sales, and in particular, sales of embarrassing products. Prior literature already stated that consumers are often following the consumer behavior of others, as they believe that these consumers indicate the appropriate behavior in a particular purchase situation or have superior information that oneself is lacking. Especially consumers with low self-esteem are sensitive to these social norms in purchase situations since these people are often looking for social approval and ways to minimize threats to their social identity. By following the consumer behavior of others, people with low self-esteem are minimizing these threats to their social identity as they believe that this is the most appropriate way to behave in a given purchase situation. Consequently, it was hypothesized that an advertisement showing the social norm leads to higher purchase intentions towards embarrassing products, and thus, has the biggest effect on people with low self-esteem as opposed to people with high self-self-esteem or people who have not been exposed to the social norm in that purchase situation.

difference in the expected direction, whereas other mean scores are rather similar. Likewise, the results do not completely align with prior literature and the hypotheses of this research. It appears that people with low self-esteem who have been exposed to the consumer behavior of others in a particular purchase situation have higher purchase intentions towards embarrassing products as opposed to people with high self-esteem who have also been exposed to the consumer behavior of others, as was hypothesized initially. Nevertheless, the highest purchase intentions towards embarrassing products is coming from the people with high self-esteem who have not been presented with the consumer behavior of others at all, which is rather contradicting to the third hypothesis of this research. However, previous studies align with this result and acknowledge the strong influence of high self-esteem on sales. They showed that people with high self-esteem are less susceptible to the opinions and evaluations of others, feelings of embarrassment and that people with high self-esteem are less likely to conform to and please others (Arndt et al., in Cialdini and Goldstein, 2004; Modigliani, 1968). Moreover, previous literature stated that high self-esteem positively influences one’s purchasing behavior, which means most likely for this paper, an increase in the sales of embarrassing products Hong and Zinkhan, 1995; Fennis and Stroebe, 2016).

Nevertheless, in the end, it should be noted that no interesting conclusion can be drawn from these results as all results appear not to be significant. Therefore, the reliability of these results should be questioned.

5.2 Theoretical Implications

This research is contributing to scientific literature as it attempted to find new solutions to increase the sales of embarrassing products. Up till now, previous studies mainly focused on the effects of plain marketing tactics (i.e. shelf space or product packaging) on the sales of embarrassing products (Dahl et al., 2001; Jones, Barney and Farmer, 2017). Currently, there is not much research to be found that looks at persuasive tactics and how these persuasive tactics can be influenced in such a way that the sales of embarrassing products will go up.

norm stating that a high percentage of consumers purchased the embarrassing product just before a particular single consumer did, is in some way depicted in the advertisement for embarrassing products, and will result in an increase in sales, as consumers are always looking at the consumer behavior of others to validate the appropriate behavior in a purchase situation (Cialdini, 2007). In addition to that it was examined whether the influence of social proof was different for people with low or high self-esteem, in which self-esteem served as a moderator. A moderator which has not been studied much before either.

The results partly aligned with the hypotheses mentioned in the literature review and showed that people with low self-esteem ultimately have higher purchase intentions towards embarrassing products as compared to people with high self-esteem. However, this study did not find any compelling evidence, and thus, no interesting conclusions can be drawn from these results. Nevertheless, from the previous literature by Cialdini and Goldstein (2004) it can be suggested that there is a difference between people with low or high self-esteem in how they respond to the principle of social proof in terms of general consumer behavior. Elaborating on that point of view, there might still be an effect of social proof on the sales of embarrassing products which is moderated by one’s self-esteem. Especially since embarrassment is an emotion that is negatively evaluated and is trying to be avoided by all means (Modigliani, 1968). Subsequently, social proof in particular might offer the possibility to minimize threats to one’s social identity or social disapproval (Cialdini and Goldstein, 2004; Modigliani, 1968). Moreover, this current research has severe limitations – which will be discussed later – meaning that these results might be different when this research is replicated under different and better research conditions.

5.3 Managerial Implications

Then, replicating this study may have relevant implications and opportunities for businesses in the retail environment. Replicating this research might gain relevant insights for marketing managers who are thinking about a way to advertise embarrassing products, so that the sales of embarrassing products will go up. The replicated study might help marketing managers to increase the sales of embarrassing products by means of creating advertisements for embarrassing products. Moreover, marketing managers may learn new information about the content of the advertisement that is most desired, meaning that it may help marketing managers to include a social norm in the advertisement in the most effective way (i.e. spoken messages in a television advertisements, slogans or textual messages in a print advertisement).

Another relevant implication for marketing managers is that high self-esteem is positively related to one’s purchasing behavior (Hong and Zinkhan, 1995; Fennis and Stroebe, 2016). Moreover, people with high self-esteem are less susceptible to the opinions and evaluations of others, feelings of embarrassment and they are less likely to conform to and please others (Arndt et al., in Cialdini and Goldstein, 2004; Modigliani, 1968). Therefore, marketing managers should find ways to make consumers feel good about themselves, already in the pre-purchase stage. For example, by simple greeting consumers in a store, having some small talk or give a little compliment to the consumer. Subsequently, when consumers feel good, or more specifically, have high self-esteem, they are more likely to purchase. Hence, the sales of embarrassing products will ultimately increase.

5.4 Limitations and Future Research

the differences between the social proof condition and the control condition for different levels of self-esteem.

Moreover, thirty-four participants out of the seventy-nine participants were excluded from further data analysis due to the assumption that they failed the manipulation check. These participants were solely excluded from the social proof condition in which participants were presented with an advertisement showing the consumer behavior of others. This resulted in an unbalanced study design, in which the number of participants in the control condition (n = 78) exceed the number of participants in the social proof condition (n = 45). Unbalanced study designs have a reduced statistical power as opposed to balanced designs, which are preferred for a two-way ANOVA analysis. The problem with unbalanced study designs is, basically, the treatment contrasts become correlated when unequal numbers of participants are present in the various groups, which results in the estimates being partly dependent on each other. Ultimately, the estimates of the main effects and the interaction effect need correction before performing the two-way ANOVA analysis, because the treatment contrasts will otherwise result in variances that are either too small or too large (Landsheer and van den Witteboer, 2015: 2). As it is least preferred for the treatment contrasts to be correlated, more relevant insights will be gained in case the study design is balanced. Therefore, an interesting issue worth exploring in the future might be the effects of an advertisement in which the social norm is presented in a more visible manner. By means of another font type, size or another text color, the visibility of the social norm in the advertisement should increase and result in participants noticing the social norm more often. Another improvement might be to present the participant with more than one advertisement. For example, for this research three advertisements of the same brand of condoms could be used. Each advertisement should have its own lay-out but contain the same descriptive social norm. The advertisement will be presented to the participants right after one another rather than presenting them at the same time. By means of repetition and enhanced visibility of the descriptive social norm it will be easier for participants to recall the percentage of people who purchased this type of condoms before them. Thus, less participants will be excluded, resulting in a more balanced study design, which is the most preferred study design.