Unconscious enhancement of self-esteem : subliminally priming compliments to enhance self-esteem

(1)

Unconscious enhancement of self-esteem:

Subliminally priming compliments to enhance self-esteem

Bachelor thesis Economics and Business, Business Studies

Author: N. A. Scheijbeler (10658165)

Supervisor: T. Dudenhöffer, MSc Topic: Subliminal Priming University of Amsterdam 29 June 2016

(2)

Statement of Originality

This document is written by Student Nina Scheijbeler 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)

Abstract

In this research the influence of a subliminally primed compliment on a person’s self-esteem was investigated. Self-esteem plays a crucial role in a person’s customer behaviour. To measure the effects on esteem an evaluative conditioning method was used. The level of upfront self-esteem, self-perception and the implicit self-theory a person adheres to, were also studied; these variables could influence the effect. A survey-based experiment, in which a subliminal message was hidden in a video, was used to test the effect on self-esteem. The complete dataset (n = 109) did not confirm a significant effect of a subliminally primed compliment on self-esteem. However, the effect of the video on the self-esteem for participants with high, average or low upfront self-esteem was investigated. Significant differences were found between the control group and the experimental group in the high upfront self-esteem category (n = 57) and the low upfront self-esteem category (n = 35). Self-esteem was enhanced significantly in the experimental condition. This suggests that subliminally primed compliments do affect the level of self-esteem. The enhancement of esteem is larger for the high esteem group compared to the low self-esteem group. This could be caused by the fact that people with a high level of self-self-esteem are better at internalizing positive feedback, like compliments. Subgroups of participants that had a high level of self-esteem combined with either an average score on the traits feminine or good-looking, or adherence to the incremental theory, also showed a significant enhancement of their self-esteem. No further effects could be confirmed for the other variables investigated.

(4)

4.3 Main effect 17 4.4 Moderating effects 20 4.4.1 Upfront Self-esteem 21 4.4.2 Implicit Self-theory 25 4.5 Additional analysis 26 4.5.1 Self-perception 27 4.5.2 Individuals Interests 29 4.5.3 Subgroups 33 5. Discussion 36 5.1 Contributions 37 5.2 Limitations 38 5.3 Application 39 6. Conclusion 40 References 42 Appendices 45

Appendix A: Online survey questions 45

Appendix B: Comparison of self-esteem measurements 55

(5)

1. Introduction

Sigmund Freud provided important insight into the concept of the unconscious. According to Freud (1915) the human mind is only partially accessible to the human conscious, while the importance of unconscious intuitive cognitive processes cannot be underestimated. He suggested that human behaviour is mainly based on the unconscious mind. Freud’s theory reinstated the human unconscious mind as a factor of importance in psychological research.

Research conducted within the field of subliminal priming shows that there are ways to influence the unconscious mind (Strahan, Spencer & Zanna, 2002; Karremans, Stroebe & Claus, 2005; Bustin, Jones, Hansenne & Quoidbach, 2015). By priming individuals subliminally, certain associations or knowledge in people's memory can be activated unconsciously in a person’s brain. The activation of certain parts of the human brain can influence their behaviour as well as their feelings and values (Dijksterhuis, Aarts & Smit, 2004).

Self-esteem is a critical part of customer behaviour, people’s purchase decisions are based on the image they have of themselves (Onkvisit & Shaw, 1987). This topic is especially interesting for retail, such as (online) clothing stores. Whether customers buy a product is influenced by their self-esteem. This involves acceptance and rejection of certain symbolic goods and people’s effort to maintain or enhance their self-esteem (Banister & Hogg, 2003). People do not regularly buy clothing in which they think they look bad, they buy clothes that maintain or enhance their self-esteem. Retail stores could try to increase customer’s self-esteem. This would make customers more inclined to buy a product. An increase in self-esteem could form a person’s attitude towards a product more positively, people could feel they need a certain product to maintain or enhance their positive self-image (Sirgy, 1985).

Dijksterhuis (2004) showed that evaluative conditioning, when used subliminally, also works for making a person's attitude more positive towards him- or herself, by pairing the word ‘I’ with a positive trait terms. To influence self-esteem, a change in interaction with other individuals has to be made. Feedback from other individuals is an example of this (Paul, 2001). Compliments can increase someone’s self-esteem (Leary, Tambor, Terdal, & Downs, 1995). Whether using evaluative conditioning to subliminally prime compliments also enhances self-esteem, has not yet been investigated. If this does have an effect on self-self-esteem, this could be applied to online clothing stores. By subliminally priming customers that are looking to buy a fashion item online, with a compliment, the self-esteem of these customers could grow. This could have a positive result on their customer behaviour. This is a situational application; subliminal priming can have an effect on the purchase decision of customers, for instance when

(6)

trying on a clothing item, but only if it influences the customer at the exact moment of hesitation of buying or not buying the product (Strahan et al., 2002). Whether it is ethically acceptable to use this kind of method to try to change the way a person feels about themselves or the decisions he or she makes without people being aware of this is questionable. Subliminal priming infringes the autonomy of a person. People should be protected against harm or exploitation by subliminal priming (Murphy, Illes, & Reiner, 2008). Discussions are going on about the use of subliminal priming with the goal to improve a person’s life. It is much harder to influence somebody unconsciously in comparison to manipulating somebody’s attitude or perception towards objects with conscious changes in for example colour, another neuro-marketing method (Murphy et al., 2008).

Not everybody reacts to compliments in the same way. For instance, people with low esteem cannot easily accept compliments due to their low beliefs. People with high self-esteem are more accepting to compliments and could thus show a higher effect of compliments on self-esteem (Josephs, Bosson, & Jacobs, 2003). The level of self-esteem of a person is not easy to determine just by looking at them, but research indicated that people with a high level of self-esteem usually have certain characteristics. They are emotionally stable, extraverted, and conscientious (Robins, Tracy, Trzesniewski, Potter, & Gosling, 2001). These characteristics could help identify people with high self-esteem and could explain why they show certain customer behaviour. Whether compliments affect a person’s self-esteem also depends on a persons believes on the malleability of personality. According to the implicit self-theory a person is either an incremental theorist or an entity theorist. Somebody believes that a person’s personality can change by somebody’s own effort, or not (Park, & John, 2010). So whether a compliment influences somebody, depends on the way somebody thinks about this malleability.

The question that was studied is whether compliments have an effect on self-esteem if primed subliminally. Therefore the research question investigated in this research is: Is it possible to enhance self-esteem through subliminally primed compliments? Is the effect of these compliments on self-esteem moderated by the level of self-esteem a person has upfront? And does the implicit self-theory a person endorses, moderate the effect of a subliminally primed compliment on somebodies esteem? The effect of subliminally primed compliments on self-esteem is investigated, using the evaluative conditioning method.

This is investigated by making use of an online survey based experiment in which a compliment is masked in a fashion video. A minimum of 100 respondents answers questions regarding their self-perception and self-esteem before and after watching the video. The idea of respondents on the malleability of a person's personality is also measured. These data are

(7)

analysed to answer to the research questions stated above. Investigating this topic will further our understanding of the importance of the unconscious, and to what extent people can be influenced by subliminal messages. The workings of these cognitive processes have certain implications for customers as well as businesses.

In the following parts of this paper, a theoretical framework will illustrate the previous research that has been done within these topics, the research field will be specified, the conceptual model will be developed and the research questions will be discussed. Thereafter the research design, experimental procedure, sample, data collection, and data analysis will be illustrated and explained. Thereafter the results will be analysed and discussed. In the last part of the paper a concluding paragraph will answer the research questions, and give concluding thoughts on the research performed and found results.

2. Theoretical framework

In this paragraph the existing literature on subliminal priming, self-esteem, implicit self-theories and evaluative conditioning will be presented; including previous research about unconscious alteration of behaviour through subliminal priming. These theories form the basis of the conceptual model and the hypothesis for this research paper.

2.1 Subliminal priming

The idea that people could be influenced through their unconscious mind started to grow in the 1950’s. This was caused mainly by the claims made by James Vicary in 1957 (Pratkanis, 1992). He claimed that he could encourage people to eat more popcorn and drink more Coca Cola in the cinema, by stimulating them subliminally with the phrase “Eat popcorn, Drink Coke” on the movie screen and therefore influencing them through their unconscious mind without them knowing. His claims were ungrounded, but caught the attention of a lot of people (Pratkanis, 1992).

Further research on subliminally influencing behaviour brought no clear view of its effect on behaviour (Pratkanis & Aronson, 1992). Meanwhile findings on subliminal priming did show that there is a way to influence the unconscious mind of people (Strahan et al., 2002). Research in the neurological field also identified an effect of subliminal priming. Findings showed that a different area within the brain was active when people got primed subliminally, in comparison to situations in which the brain consciously perceived messages. This indicated that the unconscious

(8)

mind is indeed affected by subliminally primed messages (Libet, Alberts, Wright, & Feinstein, 1967).

Numerous previous studies suggest that subliminal priming can affect the behaviour of people, but only with weak persuasion effects (Trappey, 1996). More research was done on the effect of subliminal priming, and more recent researchers found ways in which the effect of subliminal priming on behaviour was visible in a more meaningful way. The effect of subliminal priming is more meaningful, when certain conditions are met (Bustin et al., 2015). For instance, the effect of priming people with a product or a brand will be larger when the product is relevant to a person's individual goals and needs (Karremans et al., 2006). In addition, the effect is also more meaningful if the people being primed are motivated to pursue their goal or needs (Strahan et al., 2002). A larger effect of subliminal priming was also found in personality traits (Bustin et al., 2015). When the ‘personality’ of the primed product or brand matches the personality of the person being primed, the subliminal prime has a larger effect on persuasion. So situational factors are of importance. Furthermore, not all individuals are as sensitive towards (the same) messages below threshold (Bustin et al., 2015).

Goals and the willingness to pursue these goals can, however, also arise unconsciously. Different experiments have shown that the willingness to pursue a goal is influenced and controlled by social features that have become associated with goals. These features could be certain social norms, or communication with others (Custers, & Aarts, 2010). An example of this is that the scent of citrus-scented all purpose cleaning agent caused people to keep their direct environment cleaner, for instance their desk at work. While not being aware of the scent, it influenced their thinking and doing. The scent of lemon was associated with cleaning, and the norm was to keep your desk at work clean (Holland, Hendriks, & Aarts, 2005). Conscious behaviour can thus be activated by conscious and unconscious goals and motivation. Subliminal priming in the form of audio, video images or scents can under certain circumstances alter an individual’s conscious behaviour (Karremans et al., 2006).

2.2 Evaluative conditioning

Research has shown that attitudes of people can also be manipulated through subliminal priming (Arnold & Boggs, 2013). A method to form attitudes of people towards an object is ‘evaluative conditioning’. In this area of research an object (conditioned stimulus) is paired with a positive or negative stimulus (unconditioned stimulus) (Dijksterhuis, 2004). As shown in earlier research the effect of evaluative conditioning is also visible when the conditions are presented subliminally

(9)

(De Houwer, Hendrickx, & Baeyens, 1997; Krosnick, Betz, Jussim, & Lynn, 1992; Niedenthal, 1990). Research has shown that people with a certain need or goal can be influenced subliminally by certain words or images. Subliminal conditioning can actually motivate individuals to pursue certain needs (Bustin et al., 2015). Veltkamp, Custers & Aarts (2011) showed that subliminally pairing customer behaviour with positive trait terms can make people show certain customer behaviour. Dijksterhuis (2004) showed that this method, when used subliminally, can also be used to ameliorate a person's attitude towards him- or herself, by pairing the word I with positive trait terms.

2.3 Self-esteem

Unconscious processes play a role in all psychological processes (Bargh, 1984). Research on these processes has shown that implicit self-esteem can be enhanced, using evaluative conditioning, by subliminally priming positive trait terms paired with the word ‘I’ in trials (Dijksterhuis, 2004). This is a directive way to influence self-esteem that does not require conscious intervention.

Self-esteem is of importance for people's happiness and wellbeing (Arnold & Boggs, 2013), as well as meaningful to customer behaviour (Onkvisit, Shaw, 1987). Self-esteem is the value a person puts on his/her personal self worth in relation to others and the environment (Arnold & Boggs, 2013). Following Greenwald & Banaji (1995, p.11), implicit self-esteem is defined as; “the introspectively unidentified (or inaccurately identified) effect of the self-attitude on evaluation of self-associated and self-dissociated objects”. It is a positive evaluation of self associated stimuli. An increase in implicit self-esteem increases total self-esteem (Dijksterhuis, 2004). This differentiation between implicit and explicit self-esteem, shows that part of a person’s self-esteem is formed by unconscious stimuli. This means discrepant self-esteem can occur when somebody, for instance, has low implicit self-esteem but high explicit self-esteem. Implicit esteem can also be viewed as a buffer of esteem, for when the explicit self-esteem is compromised (Arndt, Greenberg, Pyszczynski, & Solomon, 1997). Implicit self-self-esteem is therefore of great importance to somebody’s self-esteem and self worth.

To influence self-esteem, a change in interaction with other individuals or him- or herself has to be made. Social comparison and identification with others can also influence somebody’s self-esteem (Paul, 2001). High self-esteem is something everybody needs and wants, to feel good about themselves and to feel happy (Arnold & Boggs, 2013; Dijksterhuis, 2004). Compliments can increase someone’s self-esteem (Leary, Tambor, Terdal & Downs, 1995). A moderating variable, which could influence the effect of compliments on someone’s self-esteem, is the level

(10)

of self-esteem a person has upfront. For instance, people with low self-esteem evaluate compliments in a different way than people with high self-esteem do (Marigold, Holmes, & Ross, 2007). People with low self-esteem have difficulties internalizing positive feedback, because it is contradictory to their self-beliefs (Josephs, Bosson, & Jacobs, 2003). Generally people evaluate themselves positively and do not have a low level of self-esteem. People with a higher level of self-esteem think better about themselves and their abilities, this happens automatically and unconsciously (Farnham, Greenwald, & Banaij, 1999).

To measure self-esteem, multiple measurements have been developed. There are measures to assess explicit as well as implicit self-esteem. These measures capture how people like themselves, feel about themselves, and believe that they are worth something or at least the same as others. Scales that have been used a lot are for instance the 10 item Rosenberg scale (Rosenberg, 1965), and the Janis and Field (Heatherton, Wyland, & Lopez, 2003). These scales measure self-esteem according to multiple statements with multiple-choice answers, about how people feel in certain situations and interact with other people, among other things. It is, however, questionable whether these self-esteem measurement scales give a good view of somebody’s complete self-esteem. The scales only correlate weakly with the implicit self-esteem measurement methods like the Implicit Association Test (IAT), and the Name-letter task (Bosson, Swann, & Pennebaker, 2000). These scales are indirect measures of cognitive processes and check whether people feel more positive towards certain letters that are part of their name, or associations with certain objects. These scales have, however, been used in multiple studies and helped investigate the topic of self-esteem.

2.4 Implicit Self-theory

Whether feedbacks from others, like compliments, really influence somebody’s self-esteem depends on the amount of self-esteem somebody already has (Marigold, Holmes, & Ross, 2007; Josephs, Bosson, & Jacobs, 2003). The implicit self-theory also plays a part in the acceptance of compliments and the way this makes somebody feel. It is an important aspect in understanding customer behaviour (Park, & John, 2010).

Implicit self-theories are beliefs about the malleability of our personality; a distinction is made between entity theorists and incremental theorists. Incremental theorists belief that one’s personality can be improved with effort, while entity theorists belief that personality is fixed and cannot be directed (Park, & John, 2010). This changes the way people think about self-enhancement (Dweck, 2000). Incremental theorists seek to improve themselves, while entity theorists seek opportunities to feel better about them selves (Park & John, 2010). These theories

(11)

are applicable to all sorts of traits. To test whether someone is an entity or incremental theorist, the Implicit Persons Theory Measure developed by Levy, Stroessner, & Dweck (1998) can be used. This measurement consists of eight statements that test whether a person thinks personality is malleable.

The implicit self–theories also affect the way customers feel about a brand. For entity theorists for example, a brand can be seen as an opportunity to think more positively about themselves. Incremental theorists on the other hand will not use a brand and its personality traits, to feel more positive about him or her self. They are less affected by brand images and experiences and their self-perception does not change (Park & John, 2010).

2.5 Conceptualization

This overview of the existing literature on the topics of this paper offers clear insight into what is known and unknown about the subject. The area of research for the current study combines the research on subliminal priming, the independent variable, with the technique of evaluative conditioning, to investigate its effect on self-esteem, the dependent variable. The conclusion that can be drawn from this literature is that self-esteem is of importance for people to feel good about themselves (Arnold & Boggs, 2013; Dijksterhuis, 2004). In addition to this, it is known that self-esteem can be influenced by communication with other individuals, by getting feedback (Paul, 2001). Compliments are a form of feedback. In this paper only this positive form of feedback, compliments, will be investigated. Trying to lower a person's self-esteem is considered unethical because of the importance of self-esteem for somebody's well being (Arnold & Boggs, 2013; Dijksterhuis, 2004).

Research also pointed out that evaluative conditioning can be used, to change a person’s attitude towards an object (Levey & Martin, 1975). As shown in earlier research, the effect of evaluative conditioning is also visible when the conditions are presented subliminally (De Houwer et al., 1997; Krosnick, Betz, Jussim, & Lynn, 1992; Niedenthal, 1990). Subliminal priming can be used to influence people; their values, attitudes and behaviour (Dijksterhuis et al., 2004). Dijksterhuis (2004) showed that evaluative conditioning in subliminal priming can enhance someone’s attitude towards him- or herself, when using positive trait terms as stimuli. Lastly, research showed that the level of self-esteem people have, influences their behaviour and characteristics, as well as their sensitivity to feedback (Josephs, Bosson, & Jacobs, 2003). The level of self-esteem can possibly moderate the effect of a subliminally primed compliment on someone’s self-esteem. Another variable that can have a moderating effect on the influence of a subliminally primed compliment on somebody’s self-esteem is the implicit self-theory that a

(12)

person believes in. Based on these earlier investigations, the following conceptual model was created (see Figure1).

Figure 1

Conceptual model for enhancing self-esteem through subliminal priming

Based on this acquired knowledge the questions investigated in this study are: Is it possible to enhance self-esteem through subliminal primed compliments? Does the upfront level of self-esteem influence this effect? And does the implicit self-theory a person believes in influence this effect? These questions were addressed through an evaluative conditioning survey-based experiment.

In line with the discussed theory the following hypotheses were developed: Hypothesis 1: Subliminally primed compliments have a positive effect on self-esteem Hypothesis 2: The effect of the subliminally primed compliments on self-esteem is larger

for people with a high level of self-esteem upfront.

Hypothesis 3: The effect of the subliminally primed compliment on self-esteem depends on the implicit self-theory somebody adheres to; the effect is larger for incremental theorists.

3. Methodology

To investigate whether a subliminally primed compliment has a positive effect on a person’s self-esteem, the technique of evaluative conditioning was used. The research was based on the theories that are available on this topic. Based on this existing knowledge an experiment-based survey was performed, and primary data was collected from human subjects. This quantitative research strategy with a deductive research approach should bring reliable results. By using a

Subliminal

primed

compliment

Self-‐esteem

Implicit self-‐

theory

Level of self-‐

(13)

cross-sectional method the effect of the subliminal primed compliments at a certain moment in time could be measured. In this paragraph it is described how this effect was investigated.

3.1 Sample

To investigate the effect of the independent variable, the subliminal primed compliment, on the dependent variable, self-esteem, an online experiment in Dutch was conducted through Qualtrics. The respondents (N = 126, Mage = 40) in this experiment were Dutch-speaking women. The sampling method that was used was non-probability convenience sampling; the survey was distributed through email and social media to Dutch-speaking females that were willing and able to participate. Some participated in the survey on their computer, others on their smartphone. The data were collected in The Netherlands, mainly in Amsterdam. All men were omitted from the sample in order to find results that are easily applicable to the fashion and clothing industry. Six incomplete surveys were neglected, and obvious outliers were identified and removed from the dataset. One suitable application for subliminally primed compliments in real-life would be in the clothing industry. The focus on female participants for this research, made it easier to find a sample group that, in general, is interested in fashion and clothing. Since not all women might enjoy fashion in the same way, or have the same opinion about fashion, additional questions were added to cover this aspect.

Participants were divided into two groups; an experimental group (n = 63) and a control group (n = 63). The mean age of the participants was equally high in the experimental group as in the control group (Mage = 40). Participants were not informed whether they were in the control or experimental group. They were informed the told the research was about personality, and instructed to answer the questions truthfully. They were informed about the voluntary nature of their participation and the anonymous and protective way their data was collected, stored and processed for analysis. No information could be traced back to specific individuals. The respondents were also aware that the data was not used for any other purpose than this study. Lastly they also had the right to ask questions, and quit or pause the survey-based experiment at any time.

3.2 Experimental procedure

In the first questions of the survey-based experiment general information about the respondents was gathered. Their age, mother tongue and whether they wear glasses or lenses. These control variables gave additional insight into the effect of a subliminally primed compliment. In the second part the 10-item esteem scale of Rosenberg was used to measure the level of

(14)

self-esteem of the respondents, using a Dutch translation of Franck, De Raedt, Barbez, & Rosseel (2008). The respondents indicated their view on 10 statements on a four-point scale – from strongly agree, to strongly disagree. Statements such as; ‘on the whole, I am satisfied with myself’.

In the next set of eight questions it was tested in which implicit self-theory the respondents believed. The first four questions represented the Entity theorists, and the last four the incremental theorists. This was done according to the eight-item Implicit Persons Theory measure, with a response format of a seven-point scale – from strongly agree to strongly disagree. On for example a statement as; ‘everyone is a certain kind of person, and there is not much that they can do to really change that’ (Levy et al., 1998). This measures shows whether participants think the personality of a person is malleable.

To measure whether their personality would be influenced by certain personality traits a number of questions was integrated to test whether participants perceived themselves to be glamorous, good looking and feminine (Park & John, 2010). These personality traits are according to earlier research, traits that women would like to enhance themselves on (Park & John, 2010). The traits also match the video that was shown in the survey. The way a participant perceives herself could influence the way she responds to, or perceives, a compliment.

In the next part of the survey a video was shown. In the video that the experimental group watched the compliment and the object of association, the person herself, were subliminally primed to the respondents. This procedure is known as subliminal forward conditioning, the conditioned stimulus is followed by the unconditioned stimulus, both below threshold. The same video was shown to the control group, but without any masked messages or form of evaluative conditioning. The video that was shown was a video that was supposed to be experienced as positive by most females. It showed a model posing in white clothing in front of a clear blue ocean (Emily DiDonato Italia, 2014). A fashion video was chosen, because of the application of subliminally primed compliments to the clothing sector. Self-esteem is known to be critical in the purchase decision in buying clothing (Onkvisit & Shaw, 1987; Banister & Hogg, 2003; Sirgy, 1985). The video had a duration of 55 seconds. The masked message that was shown to the experimental group was the word ‘ik’ (Dutch word for ‘I’), followed directly by the phrase ‘je bent mooi’ (Dutch for ‘you are beautiful’). These words were presented in the middle of the screen in colours that fitted the colours of the video (blue background and white letters) the total image covering almost the whole screen, starting at 33 seconds into the video. The masked messages were both shown for 10 milliseconds (Dijksterhuis, 2004). The video was created using Adobe Premiere CC 2015. Participants were told to pay close attention while watching the video.

(15)

After watching the video the survey was continued, starting with questions regarding the video the participants had just seen. These questions were added to see whether the respondents really felt positive about the video, had paid close attention, and whether the video made them perceive themselves differently. In order to get reliable results, self-esteem was measured again after the video, using a different measurement scale to prevent participants from repeating their answers. A selection of questions of a revised Janis and Field Scale, as descripted by Heatherton et al. (2003) was used. Eleven items measured self-esteem on a five-point scale – from practically never, to all the time. An example of a statement of the scale is: ‘how often do you feel inferior to most of the people you know’. According to Heatherton & Polivy (1991) self-esteem can fluctuate and manipulations can have immediate temporary effect on self-esteem. Comparing two measures of self-esteem, within a time frame of, on average, five minutes, will therefore be able to show an effect of the manipulation of self-esteem.

After this second assessment of self-esteem, the participants were asked to complete a couple more debriefing questions, asking whether they had seen anything unusual while watching the video, and they were thanked for their participation. The exact questions used in the survey are shown in Appendix A. Any additional information and access to all data is available on request.

3.3 Statistical analysis

The independent variable, the subliminally primed compliment, was presented to the participants in the experimental group hidden in a video as part of the survey. The object of association; the person herself, was also subliminally primed to affect self-esteem. All participants in the experimental group were shown the same subliminal messages. ‘Je bent mooi’ and ‘ik’. The dependent variable, self-esteem, was measured twice using two different measurement scales. By comparing these, the effect of the subliminally primed compliment was measured. This was done by calculating the level of self-esteem according to the Rosenberg scale before the participants watched the video (Rosenberg, 1965), and the level of self-esteem after watching the video according to the scale of Janis and Field (Heatherton et al., 2003).

To calculate the self-esteem according to Rosenberg (Rosenberg, 1965), the reversed scale answers were adjusted, and the scores for the answers given were aggregated. The same was done for the Janis and Field Scale (Heatherton et al., 2003). To compare the two scales with each other the Janis and Field scale was corrected (by multiplying the scores of the respondents with 4/5, because of the different response formats used). A possible increase in self-esteem could then be

(16)

calculated. The reliability of the scales was then tested using Cronbach’s Alpha. This was also usedto see whether the two scales measure the same aspect.

The first analysis was done on the descriptive statistics that were brought forward through SPSS. A paired t-test showed whether there was a difference in self-esteem before and after watching the video. A Mann-Whitney test could then be performed to see if the difference was larger in the experimental group, compared to the control group.

The initial measure of self-esteem was also used to look into the level of self-esteem the participant have upfront, and its moderating effect. Participant with a level of self-esteem above average (>29), were categorised as people with a high level of self-esteem. People with a level of self-esteem below average (<29) were categorised as participants with low self-esteem. A two-way ANOVA was then performed to test whether one of these groups’ self-esteem changed significantly after watching the video, and whether the upfront self-esteem influences the effect of the subliminal primed messages. Next Mann-Whitney tests were performed for each separate category.

The other moderating variable, the implicit self-theory a person believes in, was measured by the Implicit Persons Theory measure (Levy et al., 1998). A higher score on the scale indicated a stronger belief in the incremental theory. The respondents were categorised as entity theorist or incremental theorist, based on whether they scored higher on the first four, or the last four questions of the Implicit Persons Theory measure. Most of the participants were either incremental theorists or entity theorists, but there were also a few respondents that did not clearly agree with one of the two theories. The reliability of this Implicit Persons Theory scale was also assessed; to be sure the scale measured the right variable. With this data a two-way ANOVA analysis was performed to check for differences in self-esteem after watching the video. This was done for the group of entity theorists, the group of incremental theorists and the participants that were neither one nor the other.

Hereafter the effect was tested for people that perceived themselves to be good-looking, glamorous, or feminine. The participants were categorised in three groups for the three traits. People that scored above average (>4) were categorised as having a high-self perception, below average (<4) was categorised as low self-perception. The others participants were categorised as having average self-perception (=4). A two-way ANOVA analysis showed whether perceiving yourself to be a certain way influences the level of self-esteem of a participant after having seen the video.

Whether they liked or disliked the style of the dress, and for the different ways that the respondents perceived the fashion style and the model in the video, were taken into account. It

(17)

was tested whether this changes the way somebody’s self-esteem changes after having seen the video. Participants were again categorised into three groups, based on the degree in which they liked the model, style and fashion used in the video. High (>4), low (<4) and average (=4).

Multiple independent-samples t-test analyses were then conducted to see if any subgroup’s, participants with a certain combination of scores on variables, self-esteem was affected by the subliminal primed compliment.

4. Results

In this paragraph the results from the analysis are given and examined in detail. At first the pre-tests are discussed, and next the main hypothesised effect is tested. Followed by the possibly moderating effects and additional effects.

4.1 Pre-tests

In the pre-tests of the survey-based experiment, the masked messages in the video were presented for 20 milliseconds. Analysing the first 16 responses to the survey showed that the messages, at least the phrase, were not subliminal. All respondents from the experimental group reported having seen the primed phrase ‘je bent mooi’ (‘you are beautiful’) in the debriefing questions. However, they did not report seeing the word ‘ik’ (‘I’). After this pre-test the length of the primes was altered to masked messages of 10 milliseconds. While analysing the responses on this new version of the survey, two out of the 10 respondents (20%) from the experimental group reported having seen the primed phrase. Again none of them indicated seeing the word ‘ik’ (‘I’). Therefor it was decided that the messages of 10 milliseconds were indeed subliminal, and the investigation and data collection was continued with this version of the survey and primed messages.

4.2 Descriptive statistics sample

The participants of the survey were divided into two groups, the experimental group and the control group. Within both groups, the average age of the participants was found to be 40 years old. Of these Dutch-speaking females, 91% indicated having Dutch as their mother tongue. Of the 109 participants 14% usually wears glasses or lenses, but did not while filling in the survey. These participants were not taken out of the sample, because they were able to fill in all the questions, and reported their answers in a logical way. To be sure that people paid close attention to the video, a question about video was posed just after the video. This was a simple question regarding an animal that they had seen while watching the video. 27% was not able to give the

(18)

right answer to this question, and therefore most probably have not paid close attention while watching the video. Another interesting result is that of the 109, 21 participant (19%), took more than 15 minutes time to fill in the survey, while most participants were able to fill in the survey within 10 minutes. They will therefor probably have paused the survey to continue it later, in one case even continuing the next day. In this case the effect of the subliminal primed compliment could be different for these participants. These descriptive statistics have consequences for the validity of the results.

4.3 Main effect

In the survey-based experiment, for each participant the level of self-esteem before and after watching the video was calculated according to the measurement scale of Rosenberg and of Janis and Field (Rosenberg, 1965; Heatherton et al., 2003). In Table 1, the levels of self-esteem of the participants in the experimental (EG) and control group (CG), before and after watching the video, are shown. For the experimental group, the mean of self-esteem after watching the video was only slightly higher than before seeing the video. For the control group the mean after watching the video was found to be only slightly lower compared to before watching the video. Alongside this, the maximum as well as the minimum level of self-esteem after the participants watched the video, were higher for the experimental group, compared to the control group. The standard deviations are high, large differences in self-esteem levels were visible between participants. Another notable result is the minimum level of self-esteem for the control group after having watched the video. This minimum (11.20) is lower than the other minimum values measured. Also the mean of self-esteem for the control group after watching the video was relatively lower compared to the mean of the experimental group.

Table 1

Descriptive statistics levels of self-esteem before and after watching the video

N Minimum Maximum M SD Before EG* 46 20.00 39.00 30.15 4.65 After EG 46 21.60 39.20 30.61 5.06 Before CG** 63 20.00 39.00 30.38 4.05 After CG 63 11.20 37.60 29.92 5.76 * EG = Experimental group ** CG = Control group

(19)

In the survey-based experiment, the video with the subliminal message was shown to the experimental group. Of these participants 27% detected the phrase ‘je bent mooi’, 9% indicated having seen the word ‘mooi’ (‘beautiful’) and 6% reported to have seen an incorrect variation on the phrase ‘je bent mooi’. It was decided to omit the 27% of the experimental group that saw the subliminally primed phrase when looking into the effect of the compliment on self-esteem (n EG = 46, M age = 42, n CG = 63, M age = 40).

Analysing the difference between the levels of self-esteem of the participants before and after watching the video resulted in the statistics shown in Table 2. The mean in the control group was negative. This means that on average the self-esteem of the participants in the control group had decreased after watching the video. The mean of the experimental group was positive, the difference between the two measures of self-esteem is positive. This indicates that the self-esteem of the participants in the experimental group had grown. When looking at the self-esteem results table (Appendix B) we see that 50% of the respondents in the experimental group and 48% in the control group have an increased self-esteem. The increase in self-esteem for participants in the control group went up to a maximum increase of 13.00. In the experimental group the largest increase was 14.40. The largest decrease in self-esteem in the control group was 22.80 while in the experimental group this was only 7.20 (see Table 2). Overall the self-esteem in the experimental group has grown more.

Table 2:

Descriptive statistics differences in self-esteem between before and after watching the video

N Minimum Maximum M SD

Difference EG* 46 -7.20 14.40 .46 3.65

Difference CG** 63 -22.80 13.00 -.46 6.71

* EG = Experimental group ** CG = Control group

The scales used to measure self-esteem were tested for reliability. The Reliability of the Rosenberg scale was high, with Cronbach’s Alpha = .86 (N = 10). All items of the scale have good correlation with the total score of the scale. Also the Jenis and Field scale showed a high reliability, with Cronbach’s Alpha = .83 (N = 11). Both scales have a high Cronbach’s Alpha (both > .80), therefore the two scales are both reliable to measure self-esteem and results can be compared for self-esteem calculations.

(20)

Next a Paired sample t-test was performed (N > 30) to check for significant differences between the levels of self-esteem before and after the participants had seen the video. Normality of the data was assessed using the Shapiro-Wilk and Kolmogorov-Smirnov tests, because of the small sample (p > .05) (Appendix C, Table 1). The results are shown in Table 3 (for additional output see Appendix C, Table 2 and 3). On average the participants in the control group did report a higher level of self-esteem before watching the video (M = 30.38, SE = .51) compared to after watching the video (M = 29.92, SE = .73). This difference, - .46, BCa 95% CI [-1.23, 2.15], was not significant t (62) = .55, p = .56. Participants in the experimental group reported a lower level of self-esteem before watching the video (M = 30.15, SE = .69) compared to after having watched the video (M = 30.61, SE = .75). The difference of .46, BCa 95% CI [-1.54, .63], was not significant either t (45) = -.85, p = .40. The self-esteem level of the participants has not changed significantly neither for the experimental group nor the control group.

Table 3:

Output Paired sample t-test comparing self-esteem before and after the video

* EG = Experimental group **CG = Control group

To check whether the difference in self-esteem before and after watching the video was larger for the experimental group that saw the video with the subliminal primed message, compared to the control group that did not see the subliminal message, a Mann-Whitney test was performed. The datasets were not distributed normally (p < .05) according to the Shaprio-Wilk test (Appendix C, Table 4). The difference between the control group and the experimental group was tested. See Table 4 for the output of difference in self-esteem between the control and experimental group after and before watching the video (grouping variables; 0 = control group and 1 = experimental group) (for additional output see Appendix C, Table 5). The experimental

M SD Std. Error Mean 95% Confidence Interval of the difference t df Sig. (2-tailed) Lower Upper

Pair 1: Before EG* After EG

.46 6.71 .85 -1.23 2.15 .55 62 .59

Pair 2: Before CG** After CG

(21)

group (Mrank = 53.20) did not have a statistically higher or lower amount of self-esteem compared to the control group (Mrank = 56.32) before watching the video, U = 1366.00, z = -.51, p = .61, r = -.05. Also the self-esteem after watching the video did not differ significantly between the experimental group (Mrank = 55.92) and the control group (Mrank = 54.33), U = 1406.50, z = -.26, p = .79, r = -.03.

Both groups reported comparable levels of self-esteem before and after watching the video, no significant difference between the two treatments was found. Hypothesis 1 could therefore not be confirmed, the results show no significant effect of the subliminal primed messages on the self-esteem of the participants.

Table 4

Output Mann-Whitney test comparing self-esteem Control group and Experimental group

Before After

Mann-Whitney U 1366.00 1406.50

Wilcoxon W 2447.00 3422.50

z -.51 -.26

r ( = !_! ) -.05 -.03

Asymp. Sig. (2-tailed) .61 .79

Note. Grouping variable: Treatment (0 = control group, 1 = experimental).

A Pearson correlation analysis showed a positive correlation, r = .72, p = .00, between the levels of self-esteem before and after watching the video in the experimental condition (Appendix C, Table 6). This indicates a high relation between the two. A higher level of self-esteem before watching the video results in a higher level of self-esteem after watching the video for the experimental group. This effect of the high self-esteem group is described in more detail in the following paragraph.

4.4 Moderating effects

To see whether any effect can be found, multiple possible moderators were tested and results are expected to point out whether any effect is present when the participants have a certain level of upfront self-esteem, adhere a certain implicit self-theory, perceive themselves a certain way, liked or disliked the video or a combination of these variables.

(22)

4.4.1 Upfront Self-esteem

First the dataset was divided into three categories: high upfront self-esteem (> 29) average upfront self-esteem (= 29) and low upfront self-esteem (< 29). See Frequencies in Table 5. Most participants, more than 50%, showed a high level of self-esteem upfront.

Table 5

Frequencies for categories of self-esteem

Frequency Percent

High self-esteem 57 52.30

Average self-esteem 17 15.60

Low self-esteem 35 32.10

Total 109 100.00

According to the Shapiro-Wilk test the data within these groups was not normally distributed (Appendix C, Table 7). However, equality of variances for the three categories in self-esteem is assumed considering the results of the Levene’s test, F (2, 106) = 1.17, p = .32 (Appendix C, Table 8). Because of the robustness of ANOVA against the violation of the normal distribution assumption, an ANOVA analysis was performed to compare the main effect of the primed compliment on the level of self-esteem, with the interaction effect of the level of upfront esteem. The test compared the level of upfront esteem in relation to the level of self-esteem after having seen the video in the experiment, to see if a statistically significant effect on the dependent variable could be found. The outputs are shown in Table 6 and 7 (treatment is not included in this ANOVA analysis because of violation of the equal variances assumption)

(23)

Table 6:

Output two-way ANOVA upfront Self-esteem

Type III Sum of Squares df Mean Squares F Sig

Corrected Model 220.81 2 110.41 3.90 .02 Intercept 7591.25 1 75691.25 2675.98 .00 Self-esteem 220.81 2 110.41 3.90 .02 Error 2998.26 106 28.29 Total 102691.84 109 Corrected Total 3219.07 108

Note. Dependent variable: Self-esteem after watching the video. R Squared = .06 (Adjusted R Squared = .051).

Table 7:

Post hoc output: Bonferroni Self-esteem (I) Self-esteem (J) Mean Difference Std. Error Sig. 95% CI Lower Bound Higher Bound Bonferroni Low Average High Average High Low High Low Average -.32 -2.95 .32 -2.62 2.95 2.62 1.57 1.14 1.57 1.47 1.14 1.47 1.00 .03 1.00 .23 .03 .23 -4.15 -5.73 -3.50 -6.20 .17 -.95 3.50 -.17 4.15 .95 5.73 6.20

Note. Dependent variable: Self-esteem after watching the video. The error term is Mean Square (Error) = 28.285

The results in Table 6 and 7 show a significant effect of the upfront level of self-esteem on the self-esteem after having watched the video, F (2, 106) = 3.90, p = .02. The Bonferroni post hoc tests revealed that self-esteem after watching the video was higher for the participants with a high level of upfront self-esteem, than for participants with a low level of upfront self-esteem, p = .03. The difference in esteem after watching the video, between the group with high self-esteem and average self-self-esteem upfront, however, was not significant, p = .23. In addition there was no significant difference in the level of self-esteem after watching the video between the low

(24)

upfront esteem and the average upfront esteem category, p = 1.00. The mean of self-esteem after the video was higher for respondents with high upfront self-self-esteem (M=31.57, SD = 5.97) compared to low upfront esteem (M = 28.62, SD = 4.27). A variance of 6% in self-esteem scores of the participants after they watched the video, could be explained by the different levels of upfront self-esteem (𝑅!_{= .06) (see below Table 6).}

To further investigate the difference between the high esteem group, and the low self-esteem group is, two spate tests were performed. Normality of distribution was assessed with the Shaprio-Wilk test, the dataset of the high upfront self-esteem group (n = 57) was not distributed normally, but the data of the low upfront self-esteem group (n = 35) was distributed normally (Appendix C, Table 10). Table 8 shows the result of the Mann-Whitney test for the high upfront level of self-esteem group, comparing self-esteem levels after watching the video between the control group (n = 35) and the experimental group (n = 22).

Table 8

Output Mann-Whitney test comparing Control group and Experimental group for the participants with high level of upfront self-esteem

After

Mann-Whitney U 208.00

Wilcoxon W 838.00

Z -2.91

r ( = !_! ) -.39

Asymp. Sig. (2-tailed) .00

Note. Grouping variable: Treatment 0 = control group, 1 = Experimental group).

Looking at the results of this Mann-Whitney test (grouping variables; 0 = control group, 1= experimental group), it can be concluded that the difference in the self-esteem between the control group (Mrank = 23.94) and the experimental group (Mrank = 37.05) after watching the video, is statistically significant for the participants with a high level of upfront self-esteem, U = 208.00, z = -2.91, p = .00, r = -.39). The effect size is negative, participants with a high level of self-esteem score lower on self-esteem after watching the video in the control group, compared to the experimental group (additional output see Appendix C, Table 11).

In Table 9 the output of the independent t-test for the participants with low upfront self-esteem can be found. These results also suggest a significant difference in self-self-esteem, between the control group and the experimental group after watching the video for the category with low

(25)

level of self-esteem. Within the category of participants with a low level of upfront self-esteem, the participants in the experimental group (M = 26.67, SE = .86) reported a higher level of self-esteem after watching the video compared to the control group (M = 30.68, SE = .96). This difference, -1.45, BCa 95% CI [1.40, 6.64], was significant t (33) = 3.12, p = .00 (equal variances not assumed, p = .50).

Table 9

Output Independent t-test comparing the experimental group and control group for the participants with an average level of upfront self-esteem

Levene's Test for Equality of Variances

t-test for Equality of Means

F Sig. t df Sig. (2-tailed) Mean Differe nce Std. Error Difference 95% Confidence Interval of the Difference Lower Upper After Equal variances assumed .46 .50 3.12 33 .00 4.02 1.29 1.40 6.64 Equal variances not assumed 3.11 32.35 .00 4.02 1.29 1.39 6.65

Based on these last two tests it is clear that within the group of participants with a high level of upfront self-esteem, as well as the group of participants with low upfront self-esteem, the self-esteem after watching the video was significantly different between the control group and the experimental group. An effect of the subliminal primed compliment was not found for all

participants, the average upfront self-esteem category did not show any significant enhancement of self-esteem. This means that also hypothesis 2 had to be rejected. If the main effect is not supported, no moderating effect can be found either. Meanwhile, a significant effect was found for participants with a high level of upfront self-esteem and participants with a low level of upfront self-esteem. Only for these participants an effect in the experimental group was found. This could indicate an effect of the primed message that was masked in the video for these groups of participants. However, the effect of in the experimental group is larger for the high self-esteem

(26)

group, compared to the low self-esteem group (see Appendix C, Table 9). 4.4.2. Implicit self-theory

The same was done for the other moderating variable, the implicit self-theory. In Table 10 the distribution of the incremental and entity theorists is given. In the sample used for this investigation, more than half of the respondents were incremental theorists. In five cases a clear adherence into one of the two theories was not found. When investigating the whole population an even distribution of the theories would be expected (Levy et al., 1998).This distribution was found using the Implicit Persons Theory measurement scale (Levy et al., 1998). The reliability of this test is high, Crohnbach’s Alpha = .90. All items have a good correlation with the total score of the scale (> .30).

Table 10

Frequencies of the Implicit self-theory

Entity 46 42.20

Incremental 58 53.20

None 5 4.60

Total 109 100.00

For the incremental theorists as well as the entity theorists, another two-way ANOVA was conducted. Again, the dataset was not distributed normally, but equality of variances was assumed, F (2, 106) = .20, p = .82 (see Appendix C, Table 12 and 13). The output is given in Table 11.

These outputs indicate that no difference in the levels of self-esteem after watching the video could be linked to the implicit self-theory a participant adheres to, F (2, 103) = 1.52, p = .22.

(27)

Table 11

Output two-way ANOVA Implicit self-theory Type III Sum

of Squares

df Mean Squares F Sig

Corrected Model 159.71 5 31.94 1.08 .38 Intercept 24200.08 1 24200.08 814.75 .00 Treatment .05 1 .05 .00 .97 Selftheory 90.13 2 45.07 1.52 .22 Treatment*Selftheory 84.48 2 42.24 1.42 .25 Error 3059.36 103 29.70 Total 102691.84 109 Corrected total 3219.07 108

Note. Dependent variable: Self-esteem after watching the video. R Squared = .050 . (Adjusted R Squared =.003).

The outputs one’s again show that the effect of the treatment (video) of the participants is not significant, F (1, 103) = .000, p = .97. Also the combination of the treatment and the self-theory a person adheres to, does not bring a significant difference between the three categories, F (2, 103) = . 1.42, p = .25. The Implicit self-theory a person adheres to, does not influence the level of esteem a person had after watching the video. No significant difference in self-esteem was found between the categories.

No main effect was found for the subliminal primed compliment on the self-esteem of the participants. Therefor the hypothesised moderating effect of the implicit self-theory a person adheres to, could also not be confirmed. No support was found for hypothesis 3. Besides this, the results showed no effect of the implicit self-theory on the self-esteem of the participants.

4.5. Additional analysis

Next to the moderating effects, there are some other variables that could influence the way the subliminal message is perceived by the participants. The video in which the primed messages were masked is of importance for the way the compliment gets across. Whether the video influences the participants depends on the way they perceive the video and their selves, and whether they like of dislike the model and fashion style in the video.

(28)

4.5.1 Self-perception

The way participants perceive themselves can influence the way they perceive the video in which the prime is masked. Self-perception was measured on the basis of three traits: good looking, glamorous and feminine. The survey showed that 58% of the respondents considered themselves to be good looking. However, 60% did not describe them selves as glamorous. Furthermore, 79% of the respondents, all female, described themselves as feminine. These three factors correlate with each other, see results Table 12. All three correlate positively with each other, r = .49 r = .35, r = .51 (p ≤ .01). These correlations show a tendency to relation between the traits. The traits ‘feminine’ and ‘glamorous’ show a higher relation. Whether the level that the participants scored on the three traits, matches the characteristics of the video, could influence in which way the participants were influenced by the video.

Table 12

Correlations between traits

Good-looking Glamorous Feminine

Good-looking r 1 109 .49 .00 109 .35 .00 109 Sig. (2-tailed)* N Glamorous r .49 .00 109 1 109 .51 .00 109 Sig. (2-tailed) N Feminine r .35 .00 109 .51 .00 109 1 109 Sig. (2-tailed) N *Significant at 0.01 (2-tailed).

On the basis of the 7-item scale for personality traits used for these three traits, all scores above four were denoted as high scores, all scores below four were seen as low scores on the traits. Scores of four were seen as average. The frequencies of the scores on the three traits are shown in Table 13.

(29)

Table 13

Frequencies of high, average and low score on traits (Good-looking, Glamorous & Feminine)

Frequency Percentage Good-looking High 63 35 11 57.80 32.10 10.10 Average Low Glamorous High 25 18 66 22.90 16.50 60.60 Average Low Feminine High 86 16 7 78.90 14.70 6.40 Average Low

Whether the level of self-perception influences the effect of the subliminally primed messages was assessed using a two-way ANOVA analysis. The data was again not normally distributed, but equality of variances was assumed for all traits, p > .05 (Appendix C, Table 14 and 15). A two-way ANOVA was performed for good-looking, glamorous and feminine. See results in Table 14 for the output.

These results show that the effect of perceiving yourself to be good-looking on the level of self-esteem after watching the video was not significant, F (2, 83) = 1.08, p = .35. The same was true for perceiving yourself to be glamorous, F (2, 83) = 1.83, p = .17, and perceiving yourself to be feminine, F (2, 83) = 1.56, p = .22. No significant main effect for one of the traits, or a combination of the traits on the self-esteem of the participants after watching the video was found all p-values in Table 13 are > .05. Self-perception did not influence self-esteem after watching the video. Self-perception on the basis of these three traits did not influence the self-esteem of the participants after having seen the video.

(30)

Table 14

Output two-way ANOVA Self-perception

Type III Sum of Squares df Mean Square F Sig. Corrected Model 974.51 25 38.98 1.44 .11 Intercept 21284.19 1 21284.19 787.05 .00 Treatment _3.49 ₁ _3.49 _{.13 .72} Good-looking 58.37 2 29.18 1.08 .35 Glamorous _99.18 ₂ _49.59 _{1.83 .17} Feminine 84.58 2 42.29 1.56 .22 Treatment * good-looking 21.81 2 10.90 .40 .67 Treatment * glamorous 27.80 2 13.90 .51 .60 Treatment * feminine 14.24 2 7.12 .26 .77 Good-looking * glamorous 42.39 3 14.13 .52 .67 Good-looking * feminine _27.11 ₄ _6.78 _{.25 .91} Glamorous * feminine 37.56 1 37.56 1.39 .24

Treatment * good-looking * glamorous 8.44 1 8.44 .31 .58

Treatment * good-looing * feminine 49.07 2 24.54 .91 .41

Error 2244.56 83 27.04

Total 102691.84 109

Corrected Total 3219.07 108

4.5.2. Individuals interests

After the video, self-perception based on the three traits discussed, was assessed with questions regarding the woman and clothing in the video. Whether people perceived themselves to be more good-looking, glamorous and feminine in the style of clothing and the model in the video, was then taken into account. As well as how much they liked or disliked the style and the model in the video. This was analysed to see any effect of the subliminally primed compliment on self-esteem could be found. When the video did not interest the respondents, the influence of the masked messages could be smaller. Four questions measured whether or not the participants liked the

(31)

video, the model and the style of fashion. On the basis of the average score of participants on these questions three categories were created. Participants that scored below the average score of four were put in the category ‘low’. Participants that scored above four were categorised in the category ‘high’. Table 15 shows the distribution of the participants for these three categories.

Table 15

Frequencies of low, average and high scores on interest in video, model and fashion style

Low 102 93.60

Average 6 5.50

High 1 .90

Total 109 100.00

A two-way ANOVA analysis was then performed to check if a certain category scores higher on self-esteem after watching the video. This data was non-normal distributed, but no violation of the equal variances assumption was assumed, F (1, 106) = .95, p = .33 (Appendix C, Table 16 and 17). The output of the ANOVA analysis is shown in Table 16.

Table 16

Output two-way ANOVA for Interest in video Type III Sum

of Squares

df Mean Square F Sig.

Corrected Model 154.88 4 38.72 1.31 .27 Intercept 10448.12 1 10448.12 354.61 .00 Interest 100.93 2 50.46 1.71 .19 Treatment 9.62 1 9.62 .33 .57 Interest* Treatment 5.28 1 5.28 .18 .67 Error 3064.19 104 29.46 Total 102691.84 109 Corrected Total 3219.07 108

(32)

This test showed no statistical significant effect of the level of interest in the video, model and style of clothing on the level of self-esteem after having watched the video, F (2, 1) = 1.71, p = .19). Also no statistical significance was found when looked at treatment and the level of interest in the video together, F (1, 1) = .18, p = .67.

Whether participants would feel more good-looking, glamorous or feminine after watching the video was also tested with a two-way ANOVA. The normality assumption was violated, but again variance was equal between the categories, F (1, 107) = .59, p = .44 (see Appendix C, Table 18 and 19). Table 17 shows the frequencies of participants scoring high, average or low on these three traits. In Table 18 the results of the analysis are given.

Table 17

Frequency of Low, average and high Self-Perception scores on good-looking, glamorous and feminine, after watching the video

Frequency Percent Low 102 93.60 Average 4 3.70 High 3 2.80 Total 109 100.00 Table 18

Output two-way ANOVA. Interest in video

Type III Sum of Squares

df Mean Square F Sig.

Corrected Model 148.94 5 29.89 1.00 .42 Intercept 12741.84 1 12741.37 427.48 .00 Treatment 20.25 1 20.52 .68 .41 TraitAfter 84.07 2 42.04 1.41 .25 Treatment * TraitAfter _25.17 ₂ _12.58 _.42 _.66 Error 3070.13 103 29.81 Total 102691.84 109 Corrected Total 3219.07 108

(33)

These results indicated no significant effect of the perception after watching the video. The perception did not affect self-esteem after watching the video, F (2, 103) = 1.41, p = .25. Also no significant interaction was found for the treatment and the level in which the video changed the participant’s perception on the basis of the three traits discussed, F (2, 103) = .42, p = .66.

Right after the participants watched the video, they were asked whether they liked the video they just watched. The frequencies of the answers given are shown in Table 19. Of the participants, 42% had a positive opinion (fun or beautiful) about the video, and 58% had a negative opinion (boring or stupid) about the video. Most people found the video boring.

Table 19 Opinion on video Frequency Percent Leuk (‘fun’) 3 2.75 Saai (‘boring’) 48 44.04 Mooi (‘beautiful’) 43 39.45 Stom (‘stupid’) Total 15 109 13.76 100.00 To see whether a positive or negative opinion about the video had an effect on the self-esteem after having seen the video a Mann-Whitney test was performed, because of the normality assumption being violated (Appendix C, Table 20). The output of this test is shown in Table 20 (grouping variables; 0 = positive opinion, 1 = negative opinion). The difference in self-esteem after watching the video between the positive opinion participants (Mrank = 52.75) and the

negative opinion participants (Mrank = 56.64) is not statistically significant, U = 1345.50, z = -.64, p = .53, r = -.06. The opinion towards the video does not influence the level of self-esteem or effect of the subliminally primed compliment.

Unconscious enhancement of self-esteem : subliminally priming compliments to enhance self-esteem