Marketers foregoing the beauty bias : examining personality factors that influence the evaluation of advertising depicting an imperfect spokesperson or situation

(1)

Marketers foregoing the beauty bias: Examining personality factors

that influence the evaluation of advertising depicting an imperfect

spokesperson or situation.

Lilla Magony Master’s Thesis Student number: 11417463

University of Amsterdam Faculty for Economics and Business

Economics Master: Behavioural Economics and Game Theory track 15 ECT

Supervisor: dhr. Dr. Jan B. Engelmann 15. December 2017

Abstract

The emerging trend in advertising is put to a test to appraise possible factors that might influence the positive evaluation of the stimulus depicting imperfect spokesperson or situation. Age, Social Status, Social Comparison Attitudes, Assimilation scores, and Satisfaction with self are tested as possible influencing factors. While it seems the respondents preferred the imperfect ad on the emotional scale in a single preference questions, and it was shown that the factors mentioned

above influence the score of reported emotions about the stimuli, the difference between the two conditions were not significant in every case.

(2)

Statement of Originality

This document is written by Lilla Magony who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are 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.

I. Introduction and Motivation

“Everything has beauty, but not everyone sees it.” Confucius

“Glamorous, air-brushed images of celebrities do have the power to make people feel bad about themselves. 80% of teen girls compare themselves to images they see of celebrities; among who compare themselves to these images, almost half say

it makes them feel dissatisfied with their own appearance.” Today-AOL: Ideal to Real Body Image survey, 2014

Preference for beauty and perfection is coded in humanity by evolution and exploited by marketers who depict the desired beauty and aspirational target audience in the advertisements. This image is not real. What is seen on the screen is an artificial image, and the consumers were not aware up until now of the amount of time, effort and money put into the process. The overdose of these extremely idealized images distorted the perception of the average beauty standards, and some experts claim it contributed to the rise of eating disorders and high dissatisfaction with people's own appearance.

In the past years, with the spread of internet and accessibility of information, the tricks used in the media to achieve the perfect look are more well-known. One of the first pioneers in changing the landscape was Dove in 2004 with the Campaign for real beauty, where the focus of the campaign was a more realistic and positive body image portrayal. While the project was highly successful, there were no similar attempts until last year, up until now. In the past months, the number of examples increased dramatically. There seems to be an emerging trend, that raises the question of the real strength of these ads with imperfect situation or spokesperson: Is there a difference in processing these ads, and are the evoked emotions different? Most probably it is not a silver bullet that works for everyone, but how can we

(4)

Once we understand them better, the advertising materials can improve on

effectiveness, hopefully can be more realistic and less frustrating for the consumers. However, while there are numerous studies focusing on the effects of beauty and attractiveness, the literature is not yet mature on the topic of realistic protagonists, or the imperfections of the spokesperson - most probably because it was not widely used so far. With this research, we attempt to take one step further and examine the effects of a more realistic media portrayal and the possible benefits of abandoning the well-established habits of presenting the perfect beauty. Furthermore we try to identify personality factors that might make consumers open to this way of

communication.

Therefore, in my thesis I am going to review and summarize the available literature, which is going to provide the foundation of the chapter afterwards, where possible theories are presented and the fit with current practical examples are discussed. The theory is tested with an online research, which is described in detail in Chapter IV, and the results are available in Chapter V.

(5)

II. Literature review and theoretical background

1. The Beauty Bias

The preference for attractive faces is inherited from ancestors of humankind and built-in into the everyday heuristic-toolkit people use to navigate through life.

Evolutionary psychology already proved that from a very early age, infants as young as 6 months old prefer attractive faces over less attractive ones (Saad, 2004), and this preference persists in later stages of life. Social psychology refers to the

phenomenon as “what is beautiful is good” effect, but also defined as a transference effect or a halo effect, defined as a “heuristic to attribute competence, kindness, other characteristics to attractive people” (Eagly et al, 1991). From now on, we refer to this heuristic as “Beauty bias”, even though it does not only include beauty. Studies in evolutionary psychology demonstrated there is a consistent pattern in gender difference, as men value beauty and health more than women, meanwhile women value social status more than men (Saad, 2004), but there is a mental shortcut when judging other people. According to a study lead by D’Alessandro, ‘attractive women are perceived to be selected more often as work colleagues, for hiring, and as dating partner’. (D’Alessandro & Chitty, 2011). Bloch and Richins added based on their results, that physical attractiveness is positively related to certain benefits, like social power, an increased level of self-esteem and the advantages of receiving positive responses from others. (Bloch & Richins, 1992)

2. Beauty in advertising

Since the spread of visual and audiovisual channels in advertising the chance to use visual cues was open for marketers. Soon the principle of the ‘Picture Superiority Effect’ was known, meaning that based on evolutionary inherent reasons images attract more attention, comprehension and liking than copy only, and it also draws more attention to use a model to demonstrate the product than display the product only. (Bjerke & Polegato, 2006)

(6)

It is also known that people are ‘drawn towards persons and things they like,

evaluate highly and prefer’ (Bjerke & Polegato, 2006), a phenomenon often referred as ‘Immediacy principle’, that explains the importance of likeable persons in

communication and as brand spokesperson.

Saad quotes an article from M. J. Baker and Churchill in the early years of

advertising, in 1977, suggesting that ‘Physically attractive endorsers positively affect an ad’s evaluation’ (Saad, 2004), although it is not entirely clear what are his

measures for success, but much further research based on this claim takes it

forward and specifies their KPIs. Buunk and Dijkstra proved that an attractive source leads to a more positive attitude toward the product it advertises, therefore it leads to a stronger purchase intention. In the same paper the authors cite the work of Zajonc from 2011, about the advantages of using attractive models in advertising and the mechanism is explained as follows: in general, higher familiarity leads to higher liking, therefore displaying attractive people in advertising materials results in more likes and may elicit a “warm glow” in the viewers, compared to least attractive models. (Buunk & Dijkstra, 2011)

Exploring the mechanisms of advertising, we need to differentiate between factors that influence the effectiveness directly or indirectly. The direct influence is ‘through the elicitation of product argument from the picture’ (Bower & Landrecht, 2001), suggesting idea that it is supposed to influence the customers on the rational level, while the indirect effectiveness can be measured by the credibility of the

spokesperson, the product evaluation by the consumers or the reported purchase intentions. Source credibility is ‘the degree to which the source of the message is perceived as being credible by the target market’. (D’Alessandro & Chitty, 2011) The higher the perceived level of credibility, the advertisement is thought to rank better on the scale of persuasiveness. Source credibility is difficult to measure objectively, it might be influenced by emotional factors. It is also impacted by identification of the customers with the endorsers, as it is shown to reach higher level if the

spokesperson is from similar cultural background (D’Alessandro & Chitty, 2011). Many studies examined the effect of credibility on advertisements, with Bower and her team focused on the level of attractiveness of the spokesperson. They found that a NAM (Normally Attractive Model) might be more credible in a specific product category, the one that focuses on solving a problem of the customer (Bower & Landrecht, 2001). Furthermore, they suggested that NAMs are recognized more

(7)

trustworthy than HAMs (Highly Attractive Models) because of the perceived similarity between the NAM and the consumer, but sufficient evidence was not found to

confirm this theory. Despite it is not entirely proved yet, it poses two points that we need to examine: the importance of the advertised product type and the effect of relatable spokesperson or similarity (which will be explored in the next chapter). As mentioned, the problem-solving products achieved better source credibility if they display a NAM, because the beauty bias might suggest to people that the highly attractive people are problem-free and have better lives than normal people (this effect might be amplified if the person in the ad is a celebrity and details about their lifestyle are known or implied). It is important to mention though that ‘certain product classes generate greater involvement as a whole than other classes’ (Feiereisen et al, 2009). Another study Baker and Churchill (1977) also concluded that the product category matters, and it has different effects depending on the gender. If the product is related to romance, like a perfume, male subjects reported a higher purchase intent if the depicted female model was highly attractive. However, when the product category was not related to romance (for example a coffee advertisement), men in the experiment indicated higher purchase intent if the model was normally attractive compared to highly attractive model. (Bower & Landrecht, 2001). We can conclude therefore that product category is a significant factor when deciding to use a highly attractive or normally attractive spokesperson. In this thesis, this will be taken into consideration, but testing the effects of it is outside of the scope of this work. The influencing factors of product category presented another phenomenon, the selectivity hypothesis, the gender difference in processing advertising stimuli. Men mostly follow schema-based strategies, while women show more detailed processing strategies and ‘engage in more comprehensive, elaborate and subjective processing’ (Feiereisen et al, 2009). The paper of Gulas and McKeage also refers to differences depending on gender. When women targeted with the implied message of ‘You can look like this’ with the product it advertises, it is often interpreted by women as ‘You also should look like this’. At the same time advertisers might imply the message towards men suggesting ‘You could have her’, but it is often interpreted as ‘You should be drawing attention of women like this’ with the usage of the given product (Gulas & McKeage, 2000). The difference in gender will be examined in this study, but the misled interpretation of the advertising messages (either if it is intended or otherwise) is becoming a big problem for society.

(8)

Bloch and Richins claim the ‘standards set by fashion models, rather than being attainable by typical consumers, is unattainable, highly unrealistic, distant from the average person, and any comparison with these images can lead to dissatisfaction with the self’. (Bloch & Richins, 1992). They are not alone with this suggestion. Many other papers propose a similar claim, such as D’Alessandro and Chitty. According to them, despite the portrayed ideal, the average weight of real women has increased in the past decade, widening the gap ‘between the cultural norm and the biological reality’ (D’Alessandro & Chitty, 2011), and with the constant portrayal of thin models, young and vulnerable women in his experiment drew the conclusion that a thin body shape is the ideal shape. Richins came to the same conclusion in 1991, saying that ‘exposure to HAMs leads to a lowered satisfaction with one’s own attractiveness among young female adults, because this ‘repeated exposure to the extremely

attractive models in advertising and other media influences consumers’ perception of what constitutes an acceptable physical appearance’ (Bloch & Richins, 1992). This phenomenon was called the ‘Beauty Myth’ by Stephens, Hill and Hanson (1994), claiming the distorted image portrayed by the media leads to poor self-esteem, that might result in excessive dieting and eating disorders. (Saad, 2004). Gulas and McKeage concluded that consumers realize the images the mass media shows are stylized and idealized, but ‘still draw meaning from it’ (Gulas & McKeage, 2000). Given the above information it sounds counterintuitive that ‘consumers may like idealized advertising imagery just like escaping reality with TV. Realism leads to less effective advertising’. (Gulas & McKeage, 2000). To understand the underlying mechanism, we need to look into the Social Comparison Theory in the next chapter.

3. Social Comparison Theory

The foundation of Social Comparison Theory was laid down by Festinger in 1954. It refers to the principle that people have a fundamental desire to evaluate aspects of themselves, such as their opinions, achievements and abilities - as the first

proposition of the theory states. Wood added another factor to the theory, the case of personal traits such as physical attractiveness and circumstances also part of Social Comparison (Wood 1989) The further propositions of the theory are:

(9)

- In absence of objective basis for comparison, people’s need can be satisfied by a social comparison with other people

- Such social comparison will, when possible, be made with similar others. (Martin & Kennedy, 1994)

The aim is accurate self-evaluation, ‘judgement of value, worth or appropriateness of one’s abilities, opinions and personal traits.’

Social comparison can occur both consciously and subconsciously (Bjerke &

Polegato, 2006), and there are individual differences in self-monitoring and Attention To Social Comparison (ATSCI) (Gulas & McKeage, 2000) . These differences are also expressed with the concept of Social Comparison Orientation (SCO), the ‘personality disposition of individuals who are strongly oriented to Social

Comparison, who are particularly interested in their own standing relative to others’. These people are labelled as having high SCO level, that means they tend to

engage in more comparison. (Buunk & Dijkstra, 2011). The same authors in this paper demonstrated a remarkable finding about the connection of high SCO values and the attractiveness of the model in the advertising when the participants are primed on being the only female person in the group. There was no difference in case of low SC (Social Comparison) subjects, but the high SC figures correlated with higher liking of the product when the priming did not take place. However, when the high SCO participants were primed on being the only female in the group, they showed a less positive attitude when the model highly attractive compared to a normally attractive model (Buunk & Dijkstra, 2011), although it need to be stated that the paper does not present sufficient and plausible explanation for this finding. As we can see, the Social Comparison Orientation values might have an influential factor on the evaluation of advertisement depending on the attractiveness of the model. To understand the components of the Social Comparison Orientation, we need to also examine the components of the Self Concept. Bjerke and Polegato, referring to Prince defined Self Concept as ‘the system of thoughts and feelings about the self which is a driving force for much of the human behaviour’. Based on Onkvisit and Shaw (1987), they claim that it is multidimensional and it has four components or four separate “selves”:

1: The real self, also referred to as the actual or objective self. - ‘The way the person actually is.’

(10)

3: The ideal-self (the potential self-actualization) - ‘The way she or he would like to be’

4: Looking-glass-self, also known as social self - The way she or he thinks others regards her or him’

‘When a female model is an advertiser’s presenter, she should represent the ideal self-image of the targeted female consumer’. Advertising usually addresses the third one, the ideal self, to achieve identification with the customers. (Bjerke & Polegato, 2006). Self-concept is believed to be stable by late adolescence, at least on

important dimensions (Richins, 1991). ‘Once the self-concept is structured, new information is selectively interpreted’ based on the Cognitive consistency

principle. This claims based on the work of Osgood and Tannenbaum (1955), that ‘people value harmony among their thoughts, feelings and behaviour, and willing to maintain consistency between these elements’ (Gulas & McKeage, 2000). If the consistency between the different elements is not maintained, Self Discrepancy might occur. According to Self Discrepancy Theory, ‘dissatisfaction occurs when there is a discrepancy between the ideal level of an attribute and the actual level’ (Richins, 1991; D’Alessandro & Chitty, 2011)

Social Comparison can affect not only concept, but also feelings or self-perception. As the definition states, ‘feelings about the self vary across social contexts and for some people fluctuate around some baseline level of self-concept, particularly in adolescence’. (Richins, 1991).

Social Comparison also can influence people’s subjective well-being. (Richins, 1991) Comparing oneself with others might result in different valence. If the comparison happens with a person who is superior in a specific domain, assimilation might

enhance self-image to result in hope or admiration. Assimilation is when ‘perceiving oneself as similar to the comparison target under normal conditions. Exposure to an attractive model will induce an assimilation process in women in which one considers oneself similar to the target’ (Buunk & Dijkstra, 2011). When comparing oneself with a superior comparison target the self image might suffer and the contrast might result frustration. (Buunk & Dijkstra, 2011)

Besides Self-evaluation, Martin and Kennedy defined two further categories of Social comparison. Self enhancement is an ‘individual’s biased attempts to maintain

positive views of him/herself to protect or enhance self-esteem’. It occurs ‘when one assumes similarity on surrounding dimensions (those involved in comparison but not

(11)

the focal dimension under evaluation)’. The other category is Self Improvement, when an individual attempts to learn how to improve or to be inspired how to improve a particular attribute’. It occurs ‘when one learns from a superior other’ (Martin & Kennedy, 1994).

Based on the Social Comparison theory, in an ideal case ‘people seek to satisfy their need for self-evaluation by comparing themselves with people who share the same characteristics. However, in absence of similar people, models in ads can act as a benchmark’ (D’Alessandro & Chitty, 2011). There is numerous research about the effect of this comparison. ‘Consumers see these idealized images and (consciously or unconsciously) compare their more mediocre selves and lives with the idealized images’. Consumers are surrounded with ‘haunting images of perfection and wealth and the increasingly desperate realization that they will never achieve the idealized state depicted in advertising’ . ‘American consumers are constantly aware of the discrepancy between life as it is lived and life as it is pictured in advertising’ (Richins, 1991, Spitzer 1962). She goes on and based on the work of Schudson (1984) claims ‘advertising does not claim to picture reality as it is but reality as it should be - life and lives worth emulating’ (Richins, 1991).

Previously it was mentioned this might negatively affect young women’s self esteem, and Gulas and McKeage proved that it is not exclusive to females, physical

appearance and financial success portrayed in the ads also negatively affected the self-esteem of young males (Gulas & McKeage, 2000). The majority of the studies focus though on female customers, because they perceive more sharply the gap between the self and the idealized image created by the advertising and media, and Feiereisen and her coauthors claim women are therefore less satisfied with their bodies, and they tend to have higher physical concerns than men. (Feiereisen et al, 2009) This might derive from the fact that evolutionarily women connect success to attractiveness, and men to social status. Some women might experience negative feelings by the comparison to the idealized images and the Highly Attractive Models, creating a negative impact on advertising effectiveness. (Feiereisen et al, 2009) Another possible influencing factor during the social comparison is the different levels of attractiveness between the real world and the idealized image depicted in the media. According to the Adaptation Level Theory, ‘people compare a new stimulus with a subjective average or neutral point (the adaptation level)’. Exposition to a different than average level of attractiveness might temporarily alter one’s

(12)

comparison standards, even when their self-concept is stable. Comparison standard is calculated taking all the ‘modal or average value of all outcomes known to the person, weighted by its salience’ (Richins, 1991). It was also suggested that the measures and standards used to determine self-evaluation have many components and can be altered by social setting can make ‘one particular standard more

cognitively accessible than others’.The best demonstration of the contrast effect is presented by Kenrick and Gutierres (1980), who showed that after watching an episode of Charlie’s Angels, male respondents rated average attractive females lower that those who haven’t watched the same footage beforehand. Based on the integration theory we can conclude that when it comes to judging, ‘in relation to the implicit or perceived distribution within that class’, the stimuli is categorized as

belonging to a particular class and judged within that class. (Richins, 1991)

As mentioned before during the literature overview, the advertisers target mostly the ideal self of customers to achieve identification, hoping that a match or at least a congruence ‘between consumers’ self-image and the brand image contributes to positive attitudes toward a brand’(Bjerke & Polegato, 2006). Feiereisen and her team confirms this, adding that media portrayal being congruent with self-schemes of the customers should generate more positive attitudes than the incongruent ones (Feiereisen et al, 2009).

The question arises, what is the reason of congruity? There are many theoretical ideas about how the consumer identifies themselves with the Highly Attractive

Models despite knowing their lives are not real, but there is a little research aiming to figure out if portraying a more realistic and normally looking model might be more appropriate. According to Bower, Social Comparison Jealousy can cause the derogation of the Highly attractive Models, and harm the reputation of her person, her expertise, her knowledge or might undermine the product argument. Interesting proposition, although this effect was not present in every study, either because it depends on the product category or maybe the negative effect needs to overcome a specific threshold to be noticeable. (Bower, 2001). Nevertheless, there are further studies examining the effects of displaying HAMs. In an experiment of D’Alessandro and Chitty, a Chinese clothing brand displayed the products on thin or somewhat fatter models. The results showed that consumers felt the fatter model made them feel more comfortable about their body shape, and more predisposed toward the brand. (D’Alessandro & Chitty, 2011). The mechanism is not entirely clear in this

(13)

case, but it seems emotions have an important role to play. We might assume that if a brand makes respondents feel better about themselves, this is reflected in a better brand evaluation. However, the study goes on and states that in general there were no differences in source credibility and attractiveness, but at the end displaying a thin model resulted a higher score in brand attractiveness. The explanation to this is unclear. In the paper of Feiereisen we can find a possible theory, stating that the high level of physical concern caused by the exposure to HAM might lead to a lowered satisfaction with one’s own body image (Feiereisen et al, 2009). Therefore the subject might long for ideals of beauty in the advertisement, and aspire to the notion that the advertised product might solve this discrepancy - this theory might only be plausible if the participant has high SC values.

To summarize the latest paragraphs, we can state that advertising with HAMs may result in lower satisfaction with self, but the respondents found the stimuli appealing, and the product likeable. Richins goes further to claim that ‘HAMs are more effective than those with less attractive models or without models’ (Richins, 1991), although this is the statement that will be examined later to see if it really holds under every circumstance.

4. Imperfection and Reality: Imperfection is reality

On the contrary, we can see successful examples when the advertisers portrayed imperfect spokespeople. The most well-known example is the campaign prepared for the brand Dove by its umbrella company, Unilever in 2004. The ‘Campaign for real beauty’ was claimed to be a great success; the proof being that it still runs 13 years later, and takes an important role in Dove’s global marketing. It was not only

overachieving in terms of commercial success and sales, but also generated huge volume of media sensation, created PR-value for the brand, and has been endorsed by celebrities, gender scholars, and professional associations alike. (Johnston &Taylor, 2008). There were also attempts to set healthy standards in the world of media, many fashion shows have banned models under a certain BMI index, and the Ban Anorexy in Marketing movement had similar goals for advertising materials (Gulas & McKeage, 2000).

(14)

D’Alessandro and Chitty examined the portrayal of realism in advertising from more perspective, and concluded that based on research and real life example ‘suggests that marketers can use more obtainable, realistic and socially responsible depiction of feminine body shape’. He also adds that the results showed the portrayal of more realistic human being showed this effect for older women more. He goes on to

suggest based on further studies (Martin; Veer et al, 2004) that ‘women identify more with media representation of an attainable body shape’. The conclusion of the paper based on the book of Webb that the brand spokesperson do not necessarily need to come through as ‘Beautiful’, the key is they have to connect emotionally with the audience.

A recent study also suggests that not everyone is satisfied with the current state of art in advertising. A report published by market research and strategic consultancy, LHBS suggests that 69% of women do not feel represented in advertising, the question remains if how can we identify and characterize this consumer group (LHBS, 2017).

(15)

III. Theory and Hypotheses

The overview above shows there are numerous studies on the connection between attractiveness and the advertising effectiveness. We have seen above the

advantages of using a spokesperson in the advertising, who is beautiful, healthy and in a desirable situation. Many theories and research projects underline the idea to portray attractive spokespersons in communication, that it became the ‘Cult of unrealizable beauty’ as Lakoff and Scherr phrased it in 1984. Since that time, the situation became worse with the widespread of image manipulation toolkits and programs, and the image we see today in marketing materials is the result of long castings, involvement of hair and make-up professionals, and many hours of retouching. In one of their campaigns, the brand Dove also tried to point out the manipulation of media and present the point that the perception of average beauty is distorted. It is a fair assumption to think that nowadays with the spread of the internet and easy access to information, manipulation and image editing softwares, the

consumers are more aware of unreality of the image from the media than ever before.

Maybe as a response to this, an interesting trend emerged in the past months in advertising, where some of the bigger brands aired advertising that depicts an

imperfect situation or not highly attractive spokesperson. The examples from the last months include a TV advertisement from IKEA titled “Every other week” that portrays a divorced couple trying to cope with the changed parental situation of their child with the help of the furniture shop. This commercial is part of their biggest umbrella claim: “Where life happens”, presumably referring to the reality of life and the challenges their customers face in everydays. In the same series IKEA explored further situations that are not common in the advertising world, where the usual family consists a happily married parents and two well-dressed and well-behaved children. Contrary to that, IKEA portrayed a first encounter between a couple adopting a child, and a middle-aged father who is trying to come to terms with his short-tempered teenage daughter. In both situations the IKEA products were reliable partners to the customers that helped them in this difficult situation.

Two brands from the automotive sector also joined this trend and prepared online marketing material displaying bittersweet stories. Ford created short films for their

(16)

online channels in Denmark where they portray the difficulties of a family when the parents decide to divorce.

Mercedes-Benz launched worldwide a series of online films with the campaign claim “Grow up.” featuring various stories of relationships and families where the main characters argue with each other, get into a fight or insult each other.

The clothing brand Desigual also had a move against the mainstream beauty ideals by contracting a model, Winnie Harlow, who has a special skin condition, called vitiligo.

As we can see, more and more brands join the idea of going against the usual portrayal of beauty and aspirational target audience. This is refreshing, and might be a good direction to decrease the social damage of highly unrealistic media image, but it is also interesting to try to dig deeper into the mechanism to try understand how these commercials work. There are possible explanations to the application of the imperfect advertising:

1. Marketers have a never-ending struggle to get the attention of the customers, break out from the cluster, and be more memorable with stepping away from the crowd.

2. The portrayal of the imperfect, but real situations is still a novel approach in the advertising language that can generate extra PR value to the company. 3. For a specific type of consumers who are open-minded, higher educated, and

have lower SCO values, the initial, negative emotional response evoked by the real situation is cognitively adjusted, and pushes it into a better

perception/opinion about the brand.

While the first two option might work only for the marketers who act first, the third one might mean a long-term change in advertising.

To prove this, we will try to narrow down the characteristics of the specific target audience who are open to this way of advertising, but at first we will explore how well the examples fit with the literature.

During the literature review it seemed crucial to consider some factors that determine if the less attractive or imperfect spokesperson can be effective. These were the product category of the brand, the gender of the intended target audience, the achieved source credibility, whether the brand ambassador can successfully evoke assimilation/congruence, and the implied gap between the self-image and ideal self. We are going to review this through four examples.

(17)

There are two campaigns from the fashion industry, the first one, Dove shower gel and firming lotion, might be considered as a problem-solving product, therefore the success of the campaign underlines the theoretical background. It is almost

exclusively aimed for female customers, and the assumption is that it scored high on the assimilation attribute, especially with the narrator claiming that it is not a

challenge to show the product effectiveness on models who are perfect already. This also communicated confidence from the brand that the product can live up to

promises in real life conditions beyond the idealized world of advertising, and this might have increased the perceived credibility. Furthermore, it does not pose an unrealistic and unattainable example for the ideal self of the consumers, narrowing the gap between the self image and ideal self, therefore the respondents might feel better about themselves.

However, in case of the clothing brand Desigual, the products might be considered beauty-enhancing function, and in this case we have only assumptions about the mechanism, because for this type of product the literature suggests using a HAM with immaculate features. Also, a rare skin condition is difficult to prompt assimilation with the target audience. Looking into the brand manifesto of the Spanish company, we learn they consider uniqueness as a foundation of their brand: “Everyone has something unique and different about them, and that's what inspires us.” With this embrace for differences, we can assume that the portrayal of the model is the

manifestation of the brand philosophy. This kind of congruity might sound credible for the female audience, and it also eases the expectations on the ideal self, as they are not trying to exclusively prescribe what is the ideal beauty.

IKEA and Mercedes-Benz are both less focussed on a female audience, rather targeting female and male customers, but we need to bear in mind for some people a car might be the predictor of social status, and Mercedes-Benz, as a premium brand might influence more the men who are seeking comparison in social status rather than in attractiveness. The situations in the web-series are realistic, full of tension and portraying the problems with human relationships, but some of the circumstances might be a bit above the average financial situation (quitting from a leading position at the company, family quarrel during a nice trip). Assimilation of the target audience might be easy, as it depicts human relationships even though the social status is different, people still can relate to a ‘father’ or a ‘spouse’, but the circumstances might widen the gap between self image and ideal self.

(18)

In case of the IKEA ad, assimilation of the target audience might be quite easy, resonates well with the consumers and creates empathy. There is no reason to suggest it creates unrealizable standards. Actually, the majority of people do not find it ideal to belong to the group of divorced people, but the endorsers in the

commercial are still desirable and respectable for their exemplary behaviour in this difficult situation. Therefore we can suggest this ad does not widen the gap between the self image and ideal self and remains credible for the target audience.

What we also can observe with the selection of marketers who opted for this kind of communication, is that none of them is at the lower end of the market, none of the advertised products are the cheapest possible option. We need to make a remark though that IKEA has an affordable selection of products, but in that segment they can win customers with their prices. Where IKEA needs to compete for customers with other furniture retailers is the middle and upper priced products.

This suggestion also links into the conclusion of the literature review, that the less attractive advertising might be effective for consumers after late adolescence. This will lead to our first hypothesis:

H1a: The age of the respondents positively correlates with the reported

positive evaluation about the advertisement that depicts negative situation or spokesperson.

We expect that respondents who evaluated the treatment video more positively, will be from the older age group.

As in most cases, the age is also positively correlated with the social status, so we can also try to show a similar relationship with the social status:

H1b: The social status of the respondents positively correlates with the reported positive evaluation about the advertisement that depicts negative situation or spokesperson.

Just as in case of the age, we expect that respondents with higher Social Status measured by income metrics, age, education, living circumstances and satisfaction scores, evaluate the treatment advertisement more positively, if we assume that higher social status means less social comparison.

(19)

The literature suggests, the reception of this kind of communication also depends on the Social Comparison Orientation of the participants. We will explore this claim to try to create a profile or identify characteristics of customers, who might be the ideal subject for advertising depicting imperfect spokesperson.

We might evaluate the attention to the social comparison, to what extent does the person engage in comparison.

H2a: Subjects with lower social comparison values have a more positive evaluation of the real, imperfect adverts.

An inverse relationship is expected between the social comparison values and the positive evaluation: the more likely it is that the person often engages in Social Comparison, the less likely that the evaluation of the treatment ad is positive.

Furthermore, we can also try to find a connection between their reported satisfaction and their self-image. Suggesting that respondents who are more satisfied with their current situation are less likely to be bothered by the reality and imperfection of the treatment video, we presume that they are going to evaluate it more positively. H2b: Subjects with smaller gap between the self-image and ideal self show more positive evaluation of the real, imperfect adverts.

And last but not least, we can try to find evidence for the higher assimilation of the target audience:

H3: There is a positive correlation between the assimilation and the positive evaluation of the real, imperfect adverts.

We expect that respondents will report higher scores on the Treatment ad in terms of it being more relatable and they empathise more with the characters in it, and this assimilation is going to increase the positive evaluation of the treatment video.

(20)

IV. Research Design and Data

1. Research design

The research was conducted as an online questionnaire in a randomized, controlled survey. Half of the participants were asked to evaluate in detail a 60 second long commercial from IKEA, titled ‘Every other week’. The other participants focused on another IKEA 60 second ad based on normal advertising standards, with desirable spokespersons and situation, which is referred to as the Control ad. (The links to the videos are attached in the Appendix). The latter advert is not entirely realistic; it uses a common method of advertising, the ‘dramatization’ of the problem the product is able to solve. The control ad operates with the anticipated usual tone of voice. It is upbeat, dynamic, funny, while the portrayed family is ideal and aspirational. The situation is idealized, but there is less focus on each character’s attributes. Their look is self-explanatory and less in the center of the story, than in case of a

beauty-enhancing product.

Displaying two ads from the same brand should eliminate the different brand perceptions between the treatment and baseline groups, as there will not be

significant differences caused by the displaying of two separate brands, and there is no expected strong negative feelings towards the brand or the advertised product. The questionnaire contains three main parts, and is attached in the Appendix. The first part poses a simple choice question to all of the participants after both ads are presented in random order to the respondents. A time component is added to force the respondents to reply as soon as possible - with this time pressure, we try to measure their initial overall preference of the two options, hopefully with the least possible rational reasoning.

In the second part, the participants are randomly presented again with one of the videos, and asked to evaluate it in detail from different aspects on a 7 point Likert scale: the basic, positive and negative emotions, purchase intent, assessment of manipulation, credibility, the novelty of the idea and the level respondents identified themselves with the main characters.

The third part contained basic demographic data questions, statements to measure current life satisfaction and Social Comparison Orientation of the respondents

(21)

(based on Scale for Social Comparison (INCOM, Iowa-Netherlands Comparison Orientation Scale), measured with responses on the Likert scale.

2. Data in general

The responses were collected during a three-week period in November 2017. The 153 respondents were recruited through Social Media, the participation was

voluntary, and the sampling is non-representative. After screening and filtering, the analysis was conducted with 102 valid observations using StataSE 15.0.

43% of the respondents were male, the mean age is 32.18 years with a standard deviation of 9.51. The youngest respondent was 18, the oldest was 65, and on average 36.2 minutes were spent on the survey.

When presented with the single choice between the two ads, 62.75% preferred the treatment ad over the control one (37.25%).

3. Reported emotional scores

The mean intensity and standard deviation of the emotions about one of the randomly presented video are summarized in Table 4.1. To conduct the tests on the emotions, the responses were

combined into new variables to merge the positive and negative emotions. The main attributes of these new, merged variables are presented in Table 4.2. and the results of the distribution test implies these are approximately normally

distributed and therefore can be used in our regressions. (Figure 4.1)

When comparing these average and individual emotion values in the treatment and control conditions with ANOVA analysis in the two conditions, a new variable is also included to

account for the fact if the respondent is answering Figure 4.1: Distribution of the new

(22)

questions about their preferred ad, or the one they did not like (variable ‘Liked’). The Mean Positive Emotions do not show any significant difference (Pr=0.306) between the 2 groups, while the same statistics with the Mean Negative Emotions depending on the treatment results Pr=0.1605. The averages are shown on Figure 4.2. In case of the individual emotions,

significant difference has been found with the following items: The treatment video reportedly made the respondents feel sad (Pr = 0.000), The treatment video

reportedly made the respondents more hopeful (Pr=0.0034), the video was found more credible (Pr=0.0004), respondents claimed it is more likely they are going to remember it in one week (Pr=0.011), had better scores on the question of

Empathising with the main characters (Pr: 0.000), and provided the perception that the video is displaying real life (Pr=0.000).

Interesting to note, once we account for the fact of the respondents are evaluating the video they preferred, the significance level of the ‘Made me feel happy’

agreement increases (Pr=0.0090) regardless of the treatment.

There seems to be a slight difference in some cases in the frequency of the selected Likert items, as the distribution of the responses in the control option is more dense on the middle values, while the treatment video seems to induce higher valence to either direction, as shown on an example presented on Figure 4.3. However, with the ‘Strongly disagree’ option selected the most often in both cases, the difference is not statistically

significant.

Figure 4.3: Frequency of selected Likert scale items in

Control and Treatment conditions to express to what extent did the video make the respondents feel Uplifted.

(23)

Furthermore, some of the variables show interesting variation between the 2 conditions. The mean of some emotions depending on the group are depicted in Figure 4.4.

Table 4.1.: Main attributes of emotions

N Mean Std. Dev. Variance

Angry 95 1.705 1.465 2.146 Sad 97 2.402 1.840 3.388 Happy 99 3.070 2.046 4.188 Disgusted 95 1.389 1.160 1.346 Nervous 95 1.957 1.662 2.764 Indifferent 95 2.789 1.961 3.848 Inspired 98 3.091 1.850 3.424 Uplifted 98 3.214 2.011 4.046 Encouraged 97 2.783 1.696 2.879 Challanged 96 2.281 1.677 2.814 Belonging 94 2.542 1.751 3.068 Satisfied 97 2.608 1.811 3.282 Hopeful 99 2.939 1.883 3.547 Ashamed 94 1.351 0.958 0.918 Inferior 93 1.451 1.088 1.185 Outsider 96 2.060 1.485 2.206 Unattractive 91 1.351 1.058 1.119 Incapable 96 1.604 1.277 1.631 Frustrated 97 2.226 1.890 3.573

Figure 4.4: Mean values

(24)

Table 4.2: Main attributes of the Mean and Total variables

N Mean Std. Dev Variance Min max

Positive Emotions Mean 100 2.803 1.458 2.126 1 6.125 Positive Emotions Total 102 21.509 12.083 146.014 0 49 Negative Emotions Mean 99 1.848 1.021 1.044 1 6 Negative Emotions Total 102 16.333 9.149 83.709 0 52

4. Satisfaction, Novelty and SCO scores

The question referring to the level of satisfaction of the respondents in general with their life circumstances provided a mean of 5.284 with standard deviation of 1.205. The distribution in the 4 age groups is depicted on the Figure 4.5

It is notable that the satisfaction seems to raise with age, the frequency of higher Likert items raises throughout age groups. However, in the last age group it drops, most probably because of the low amount of respondents. As Figure 4.6. Depicts, the 95% Confidence interval (marked with grey) increases dramatically in the older age group, assumedly because of the uncertainty caused by the few observations.

Figure 4.5: Frequency of General Satisfaction

scores divided by Age Groups

Figure 4.6: Prediction for General Satisfaction

(25)

In the regressions the Mean of Satisfaction scores will be used, that is a mean of the all statements regarding satisfaction with appearance, social status or general

satisfaction, the agreement with negative statements are used in this variable after being recoded. The mean of this score is 4.516 with standard deviation of 1.155. The mean of reported openness to novelty is 5.25 with standard deviation of 1.052, and is not significantly different for the treatment group.

The Social Comparison Orientation statements were also combined into 1 variable from 11 statements. The negative statements were recoded just as with the satisfaction scores, so the higher the figure, it is more likely that the respondent is going to engage in social comparison. The mean value for the SCO 3.835 with the standard deviation of 1.049, the density is shown on Figure 4.7, we conclude its distribution allows us to use in our analysis.

Figure 4.7: Distribution of the Social Comparison

(26)

V. Results

To evaluate the Hypotheses, we describe below the methods and models were used.

H1a: The age of the respondents positively correlates with the reported

positive evaluation about the advertisement that depicts negative situation or spokesperson.

H1b: The social status of the respondents positively correlates with the reported positive evaluation about the advertisement that depicts negative situation or spokesperson.

The first hypothesis assumes positive correlation between age and the positive evaluations about the treatment video. To measure this, the responses about

positive emotions were combined into one variable using their mean and generating a quasi-continuous dependent variable.

In the second part of the first hypothesis, there is a correlation assumed between the reported positive emotions and the social status. Social status is to be indicated by age, the level of education, money spent on the rent and the status satisfaction scores reported by the respondents. The latter is an inference from the satisfaction scores to the real social status. These factors might be correlated so at first we will have a look at the independent variables. Age seems to be a good predictor if someone lives in a village or smaller entity and the occupation variable, but not the education, rent or the score on satisfaction with social status.

In the first step we start with examining the effect of age in separation, mainly with using the following model in regressions (1) - (5):

𝐸𝑚𝑜𝑡𝑖𝑜𝑛 = 𝐴𝑔𝑒 + 𝐴𝑔𝑒 ∗ 𝐴𝑔𝑒 + 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 + 𝐿𝑖𝑘𝑒𝑑 + 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 ∗ 𝐴𝑔𝑒 + + 𝐿𝑖𝑘𝑒𝑑 ∗ 𝐴𝑔𝑒

As the variable ‘Liked’ (if the subject is evaluating the video they liked) contains information already about the treatment, we do not use an interaction term for treatment and age.

(27)

As the preference question can be regarded a positive evaluation of the video, another way is to check if there is any significant relation between the simple choice outcome and age. In this case the former is used as a dependent variable and regressed on age using logit regression in the last column of Table 5.1.

Table 5.1.: Linear regression results: Relationship between Positive Emotions and Age

(1) Challanged (2) Hopeful (3) Product desirability (4) Positive Emotions, Mean (5) Negative Emotions, Mean (6) Preference (Logit) Age 0.143** (0.064) 0.170** (0.064) -0.142** (0.062) -0.078 (0.057) 0.004 (0.045) 0.003 (0.098) Age * Age -0.002** (0.000 -0.002** (0.000) -0.002** (0.000) -0.001** (0.000) 0.000 (0.000) -0.001 (0.001) Treatment -0.904 (1.197) 0.218) (1.229) -0.591 (1.214) -1.009 (1.036) 0.631 (0.848) Treatment * Age 0.030 (0.38) 0.238 (0.036) -0.019 (0.037) 0.028 (0.031) -0.012 (0.027) Liked -0.148 (0.878) 1.294 (1.005) -2.048** (0.886) 0.928 (0.799) -0.375 (0.620) Liked * Age 0.291 (0.025) 0.002 (0.025) 0.095*** (0.023) 0.011 (0.020) -0.011 (0.017) Constant -0.105 (1.203) -0.534 (1.235) 1.475 (1.250) 1.407 (1.171) 1.703* (0.993) 0.617 (1.798) Robust standard errors in parentheses.

*** p<0.01, ** p<0.05, * p<0.1

Age: Age of the respondents expressed in years

Treatment: If the individual has seen the treatment or control video (binary, Treatment=1, Control=0)

Liked: If the individual is asked to evaluate the video they preferred (binary, Evaluating the preferred one=1, Evaluating the ad that was not preferred=0)

Challenged: Level of agreement with the statement that the video made the respondents feel Challenged

Hopeful: Level of agreement with the statement that the video made the respondents feel hopeful Product Desirability: Level of agreement with the statement that the video displayed the product in a desirable way

Positive Emotions: Mean of all reported positive Emotions Negative Emotions: Mean of all reported negative Emotions

Preference: Simple choice preference between the 2 videos. (Binary, Treatment=1, Control=0)

Feeling Challenged, Feeling Hopeful and Desirable Product presentation have a statistically significant coefficient, although the effect seems rather small. This effect

(28)

disappears when the other emotions are combined into a general score of positive emotions, is not significant for the negative emotions, and the Logit model also failed to show the influence of the Age variable. Unfortunately it is difficult to provide proof for this hypothesis, but theory suggests other confounding variables can help, so let’s examine what happens if we revise our model and include further variables for the next part of the Hypothesis.

In the upcoming regressions, we will include other variables to account for the Social Status, and use the following model:

𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝐸𝑚𝑜𝑡𝑖𝑜𝑛𝑠

= 𝐴𝑔𝑒 + 𝐴𝑔𝑒 ∗ 𝐴𝑔𝑒 + 𝑆𝑡𝑎𝑡𝑢𝑠 + 𝑆𝑡𝑎𝑡𝑢𝑠 ∗ 𝐴𝑔𝑒 + 𝑅𝑒𝑛𝑡

+ 𝐶𝑖𝑡𝑦 𝑜𝑓 𝑟𝑒𝑠𝑖𝑑𝑒𝑛𝑐𝑒 + + 𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛 + 𝑂𝑐𝑐𝑢𝑝𝑎𝑡𝑖𝑜𝑛 + 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 + 𝐿𝑖𝑘𝑒𝑑 + 𝐿𝑖𝑘𝑒𝑑 ∗ 𝐴𝑔𝑒

Table 5.2.: Linear regression on age and social status variables

(1) Mean of positive emotions (2) Mean of positive Emotions (3) Mean of Positive Emotions Age 0.151** (0.059) 0.140** (0.059) 0.163** (0.77) Age * Age -0.002** (0.000) -0.002** (0.000) -0.002** (0.001)

Satisfaction with Status 0.131

(0.079) 0.145* (0.082) 0.251 (0.302)

Satisfaction with Status *

Age -0.004 (0.009) Rent 0.033 (0.074) 0.028 (0.074) 0.016 (0.071) City of residence 0.195) (0.156) Education -0.285 (0.138) -0.288** (0.138) -0.259* (0.146) Occupation -0.002 (0.093) Treatment -0.132 (0.286) -0.108 (0.289) -0.059 0.2999 Liked 1.247*** (0.272) 0.474 (0.857) 0.772 (0.834) Liked *Age 0.236 (0.022) 0.013 (0.022)

(29)

Constant 0.014 (1.109) 0.352 (1.110) -0.605 (1.546) N 100 100 100

Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Age: Age of the respondent measured in years

Satisfaction with status: Level of agreement to the statement that the respondent is satisfied with their social status

Rent: Average monthly spend on rent, split into 5 groups City: City of residence split into 4 groups based on size Education: Level of highest education split into 6 groups Occupation: Occupation of the respondent (categorical)

Treatment: If the individual has seen the treatment or control video (binary, Treatment=1, Control=0) Liked: If the individual is asked to evaluate the video they preferred (binary, Evaluating the preferred one=1, Evaluating the ad that was not preferred=0)

As it is presented in the table, including the confounding parameters, age becomes a significant determinator of the positive emotions. This implies that the positive

evaluation of the treatment advert slightly increases with age as hypothesised. As per the second part of the hypothesis regarding the social status, we get mixed results. Education seems to be a significant determinant of the positive evaluation in most cases, and in regression (2) the Satisfaction with Social Status variable also becomes significant on the 90% level.

H2a: Subjects with lower social comparison values have a more positive evaluation of the real, imperfect adverts.

The next hypothesis assumes that the SCO value predicts the positive evaluation of the treatment ad.

The higher the SCO value, the more often a person compares themselves to others, and SCO values are partially determined by age and correlated with the general satisfaction measures, both the opposite direction, meaning that the older a

respondent is, or the more satisfied they are with their lives, it is less likely that they are going to engage in social comparison. This simple regression is in Table 5.3.:

Table 5.3.: Linear regression of confounding parameters of SCO scores

(1)

SCO scores

Age -0.037***

(30)

Gender 0.186 (0.197) General Satisfaction -0.153* (.0075) Constant 5.757*** (0.518) Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Age: Age of the respondent measured in years

Gender: Gender of the respondent (1=Male, 2=Female)

General Satisfaction: Question to measure the general life satisfaction scores of the respondent

SCO: Social Comparison Orientation scores, variable created from various statements

Based on these results Age and Gender are excluded from the next regression, which is presented in Table 5.4. With the most important results from the regression that uses the following model as a basis:

Positive/Negative Emotions = SCO + Treatment + Liked + Treatment*SCO

Table 5.4.: Regression results to determine the effect of SCO variable

(1)

Mean of positive emotions (2) Mean of Negative Emotions

SCO 0.292* (0.149) -0.240* (0.141) Treatment 0.962 (0.985) -0.579 (0.776) Liked 1.265** (0.286) -0.662** (0.225) Treatment * SCO -0.320 (0.254) 0.220 (0.179) Constant 1.101** (0.532) 2.997*** (0.595)

SCO: Social Comparison Orientation measures

Treatment: wether the the treatment or control video was seen (binary, Treatment=1, Control=0) Liked: If the individual is asked to evaluate the video they preferred (binary, Evaluating the preferred one=1, Evaluating the ad that was not preferred=0)

(31)

The table above shows the Social Comparison Orientation might be a determinant of the evaluations and consumer choices, although it is significant only on 90% level. Keeping everything else constant, 1 point increase in the SCO scores increases the reported positive emotions by 0.29 point, while decreases the Negative Emotion, which is the opposite of the hypothesised effect.

H2b: Subjects with smaller gap between the self-image and ideal self show more positive evaluation of the real, imperfect adverts.

We attempt to show a relationship between the respondents’ gap between self image and ideal self, using the satisfaction scores.

Some questions in the survey were aimed to measure the current satisfaction of the respondents, and further questions inquired about to what extent would they change their life, to imply the underlying dissatisfaction. Based on the theory about the different layers or self-concept, the difference in these responses might be used to measure the gap between the self-image and ideal self. Although these require more in-depth research to examine with more sophisticated questions, we attempt to provide a topline view on this Hypothesis.

The variable to measure the size of the gap between self-image and the ideal self, is created as the difference between the reported generic satisfaction and the

agreement that changes needed in the respondent’s life were taken. The latter was recoded, and the bigger the difference between the two figures in the variable created, the higher is the assumed gap. If the first number is smaller than the second, it means the person would want to change a lot, and the measure is negative. If the first score, the original satisfaction score, is higher than the answer about the changes to reach the ideal self, the variable to measure the gap is positive. It was measured with general satisfaction statements (variable called ‘Gap General’), and specific questions about the appearance (variable referred to as ‘Gap

Appearance’)

Once the new variables were defined this way, different regressions were tested, summarized in Table 5.5 based on the following model:

(32)

Table 5.5.: Linear regression results to identify the effect of the gap between self-image and ideal self

(1)

Encourage (2) Hopeful (3) Standing out (4) Assimilation (5) Mean of positive emotions Gap General 0.274** (0.125) 0.236* (0.104) 0.284** (0.132) 0.152** (0.076) 0.117* (0.070) Liked 1.549*** (0.416) 1.626*** (0.400) 1.045** (0.457) 1.279*** (0.281) 1.342*** 0.263 Treatment -0.349 (0.337) 0.668* (0.364) 0.060 (0.346) -0.306 (0.290) -0.286 (0.272) Liked * Gap -0.143 (0.175) -0.094 (0.151) -0.005 (0.177) Constant 1.705*** (0.258) 1.355*** (0.236) 3.366*** (0.322) 1.741*** (0.199) 1.972*** (0.190) N 96 98 101 99 99 R2 0.157 0.249 0.166 0.180 0.209

Gap General: Gap between self-image and ideal self, General

Treatment If the individual has seen the treatment or control video (binary, Treatment=1, Control=0) Liked: If the individual is asked to evaluate the video they preferred (binary, Evaluating the preferred one=1, Evaluating the ad that was not preferred=0)

Encourage: Level of agreement with the statement that the video made the respondents feel encouraged

Hopeful: Level of agreement with the statement that the video made the respondents feel hopeful Standing out: Estimated likelihood that the video stands out from the majority of advertisements Assimilation: Combined variable to account for the identification with the spokesperson in

advertising and responses to the question of to what extent did the respondents feel addressed with the video

As we can see, the gap between the ideal self and self image influences the reported positive emotions, in most cases for both the treatment and control ads: the higher the gap, the evaluation is more positive. Note that if the person would like to change many things in their lives and appearance, then the value of the gap is negative. In our regressions we showed the change of the gap scores by 1 unit to the positive direction would increase the dependent variables, the positive emotions also to positive direction with the coefficients reported in the table.

(33)

H3: There is a positive correlation between the assimilation and the positive evaluation of the real, imperfect adverts.

The last hypothesis focuses on testing the assimilation assumption from the

literature. In our survey, it was measured by statements about to what extent did the respondent feel addressed by the advert and the level of identification they

experienced with the main characters. These two variables were combined into the ‘Assimilation’ variable for our analysis.

Anova tests showed that the difference in reported assimilation scores was not statistically significant between the Treatment and Control groups.

However, running a very similar regression than in the previous hypothesis on the Mean of Positive Emotions variable to see to what extent it is determined by the assimilation scores, we find the suggestion from previous studies confirmed, the assimilation scores partially determine the evaluation of the ads. The notable result is that in the first regression the Assimilation has the opposite effect than expected, as it decreases the expected mean of positive emotions.

Empathy also can be considered as a weak form of assimilation, therefore we can attempt to replace the ‘Assimilation’ variable in our regression to the empathy

scores. Further details can be found in Table 5.6., the base model used is almost the same as in H2b:

𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑜𝑟 𝑁𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝐸𝑚𝑜𝑡𝑖𝑜𝑛

= 𝐴𝑠𝑠𝑖𝑚𝑖𝑙𝑎𝑡𝑖𝑜𝑛 𝑠𝑐𝑜𝑟𝑒𝑠 + 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 + 𝐿𝑖𝑘𝑒𝑑 + 𝐿𝑖𝑘𝑒𝑑 ∗ 𝐴𝑠𝑠𝑖𝑚𝑖𝑙𝑎𝑡𝑖𝑜𝑛 + 𝐺𝑒𝑛𝑒𝑟𝑎𝑙 𝑆𝑎𝑡𝑖𝑠𝑓𝑎𝑐𝑡𝑖𝑜𝑛

(34)

Table 5.6.: Linear regression results to examine the relationship between assimilation and the evoked positive or negative emotions

(1) Mean of Positive emotions (2) Mean of Negative Emotions (3) Mean of Positive Emotions (4) Mean of Negative Emotions (5) Mean of Positive Emotions Assimilation -0.615*** (.102) -0.215** (0.108) 0.624*** (0.138) Treatment -0.182 (0.226) 0.177 (0.192) -0.707** (.271) 0.439** (0.213) -0.176 (0.229) Liked 1.224** (0.414) -0.907** (0.382) 1.60** (0.554) -0.912 (0.462) 1.220** (0.445) Empathy 0.527 (0.094) -0.198* (0.114) Liked * Assimilation -0.187 (0.147) 0.156 (0.116) -0.199 (0.165) Liked * Empathy -0.263* (0.150) 0.112 (0.125) Satisfaction 0.153* (0.091) Constant 0.909*** (0.231) 2.527*** (0.346) 1.017*** (0.231) 2.564*** (0.382) 0.094 (0.585) N 99 98 99 97 99

Treatment If the individual has seen the treatment or control video (binary, Treatment=1, Control=0) Liked: If the individual is asked to evaluate the video they preferred (binary, Evaluating the preferred one=1, Evaluating the ad that was not preferred=0)

Assimilation: Combined variable to account for the identification with the spokesperson in advertising and responses to the question of to what extent did the respondents feel addressed with the video Empathy: To what extent did the respondents empathise with the main characters in the video Satisfaction: Level of agreement with the statement that the respondent is satisfied with their lives

With these results we can confirm Assimilation also proved to be significant determinant of the positive emotions, the relationship between the variables were shown. The effects surprisingly seemed to be to the opposite direction: higher assimilation scores imply a decrease in the reported positive emotions.

The reason why it might be negative direction if people do not like to feel assimilation if they are dissatisfied with their current life and situation. Therefore, the General life satisfaction score was also included in the last regression (5) and it appeared to be justified with significant coefficients, with the trend reversed, in the direction of the coefficient matches the hypothesised values.

(35)

VI. Discussion and implications

1. Evaluation of the results

The study presented us with various insights and mixed results.

Roughly two thirds of the participants preferred the Treatment ad when it came to simple choice between the videos, and only one third picked the Control ad as their favourite choice. Despite this major win for the Treatment ad, the attempt to

disentangle the underlying mechanisms of the choice, the reasons are still not entirely clear. There were few categories where the difference between the two videos were significant (for example Making the respondents sad, Empathizing with the main character and the perception of displayed reality and so on). Based on the literature and theory, these significant variables cannot only be responsible for the overwhelming win of the treatment video.

While these results are credible, do not entirely explain why the respondents preferred the Treatment video better, the difference between the treatment and control videos are smaller than expected. Obviously not every factor could be surveyed in this questionnaire. It might simply be that the creative execution was liked better, or the Treatment ad was preferred only in relation with the control ad, and the control ad was not liked. Some respondents mentioned it was too busy, and made them dizzy.

It is difficult to provide a compelling argument about the hypothesised influence of Social Status, as the coefficient was found to be significant only on the 90% level. One possible reason is the collected data was not precise enough to describe Social Status, or there is not enough predictive power in this variable. Even if this is the case they are useful confounding variables to examine the effects of age. The literature suggested stabilizing self-evaluation after late adolescence, which was found when the confounding variables were added, and the respondents were split into age group - we found that the mean of reported positive emotions rose when entering the second age group, even if it was not significantly different for the treatment and control versions.

What was also clear from the research is that the Social Comparison Orientation partly drives the evaluations, and that there is a clear relationship between the two

(36)

variables. The higher the Social Comparison Orientation, that is someone more likely to engage in comparing themselves to others, the higher the reported positive

emotions - for both videos. One possible explanation is to this effect that is opposite as expected, that higher SCO values might imply higher attentiveness to media and television, which leads to higher acceptance of the ads, the tools and the tone of voice they use. This however was not measured in the current study therefore we cannot test it.

Evidence was found though for the assimilation assumptions, that higher assimilation scores induce higher liking - this confirmed the results from the literature review. We have also seen that the current satisfaction level of the respondents is crucial in this case. If they were feel better about their own situation, would that also mean that they like the ads better? This, as many other factors could be determined in a follow-up study that can take on with the recent research but with polished methods.

2. Implications

In a follow-up study some shortcomings of this research can be taken into consideration.

Self-reported measures which were the main input for this study are not always reliable, and does not provide insight on the underlying mechanisms in the consumers’ brain. The outcome, a mark on a Likert scale cannot represent the process that takes place in the respondents’ head. Likert scale measures are often criticised for causing distortions either because the respondents try to guess the desirable response and try to comply with the expectations (social desirability bias), or because the tendency to avoid the extreme ends of the scale (central tendency bias), or because the respondents seem to agree with the statement. To account for this latter bias the questionnaire contained statements in positive and negative framing, but in a more extensive study the self-reported measures could be combined with other methods, like biometric techniques. Unfortunately applying these methods were out of the means this master’s thesis.

It is also worth mentioning another weakness of the self-reported measures, that it relies on the learned concepts of the emotions that might differ for each person. Based on the work of Lisa Feldman Barett, we might assume there are individual

Marketers foregoing the beauty bias : examining personality factors that influence the evaluation of advertising depicting an imperfect spokesperson or situation