The Effect of Fractal Smartphone Wallpapers on Aesthetic Liking and Touching
Master Thesis MSc Marketing University of Groningen Faculty of Economics and Business
1st Supervisor
Dr. Yannick Joye
2nd Supervisor
Dr. Carmen Donato
By
Table of Content
Abstract 4
1. Introduction 5
2. Literature Review 8
2.1 Importance of Smartphone Design & Wallpapers 8 2.2 Fractal Geometry of Nature and Aesthetic Liking 9
2.2.1 Complexity 10
2.2.2 Symmetry 11
2.2.3 Angularity 11
2.3 Psychological Frameworks 12
2.3.1 Processing Fluency Theory 12
2.3.2 Arousal Theory 13
2.4 Fractals in Nature 13
2.5 Fractal Dimension 15
2.6 Fractal Aesthetics 16
2.7 Neurophysiological Response to Fractals 18
2.8 Touching 19
2.9 Smartphone Brand Image 20
2.10 Construal Level 22
2.11 Conceptual Model 23
3. Methodology 25
3.1 Participants and Design 25
3.2 Materials 26
3.2.1 Smartphone Brand 26
3.2.2 Wallpapers 27
3.3. Measurements, Correlation and Reliability Analysis 27
3.3.1 Aesthetic Liking 27
3.3.2 Touching 28
3.3.4 Behaviour Identification 28
3.4 Procedure 29
4. Results 30
4.1 Descriptive Statistics 30
4.2 Hypotheses Testing 30
4.2.1 ANOVA : Aesthetic Liking 30
4.2.2 ANOVA : Touching 33
4.2.3 ANOVA : Willingness to Buy 35
4.3 Correlation Analysis on Touching and WTB 37
4.4. Mediation Analysis on Touching 37
4.4.1 Apple Brand (Low FD, Mid FD, High FD) 37 4.4.2 OnePlus Brand (Low FD, Mid FD, High FD) 38 4.5 Mediation Analysis on WTB
4.5.1 Apple Brand (Low FD, Mid FD, High FD) 38 4.5.2 OnePlus Brand (Low FD, Mid FD, High FD) 39
4.6 Construal Levels 40
4.6.1 ANOVA : Construal Level & Aesthetic Liking 41 4.6.2. ANOVA : Construal Level & Touching 42 4.6.3 ANOVA : Construal Level & WTB 43
5. Conclusion 45
5.1 General Discussion 45
5.2 Limitations & Further Research 47
References 48
Abstract
Many customers today are judging smartphone’s attractiveness by its design. However, many smartphone’s design look similar and has reached to saturation. The use of wallpapers can help smartphone brand to differentiate themselves from competitors. Smartphone wallpapers often display natural content or forms that have fractal-like properties. This visual characteristics of the nature is perceived to be aesthetically pleasing by our brain. Past research has studied how visual characteristics of nature can affect customer liking. However, the study of fractal dimension and its effect on customer behaviour is limited.
This research will explore on how fractal wallpaper with different fractal value (low, intermediate, high) is linked to aesthetic liking. Moreover, we want to investigate whether fractal wallpaper can also affect other customer behaviour, such as touching and willingness to buy. Additionally, we want to know whether smartphone brand also plays an important role.
1. Introduction
Aside from the functional benefits offered by smartphones, many customers nowadays are judging the smartphone’s attractiveness by its design. Maybe one particular phone has such strong hardware components, such as 4GB of RAM, waterproof case, and large batteries. But with its unappealing design, customers end up choosing the other smartphone that has a better design, albeit with inferior hardware components. With strong emphasis on smartphone design preference, many customers see the design as an important criterion for the purchase decision. The smartphone’s design serves more than just how comfortable or lightweight the phone is, but it is also a form of personal expression (Castells, 2006). The smartphone’s design itself can be judged in terms of its colour, finishes, and build materials.
Past studies have investigated how the colour of the smartphones can affect customer’s aesthetic liking (Nanda et al, 2008), or how the phone’s materials such as the shape, and size influence customer’s aesthetic liking (Chuang et al., 2001; Yun et al., 2003). However, nowadays the smartphone design has reached saturation, and you can see that most smartphones’ design from every brands look somewhat similar. Interestingly, the one thing that can also differentiates one brand to other brands in regard to smartphone design is the wallpapers used.
Often, smartphone wallpapers display natural content or forms that have fractal-like properties. This visual characteristics of the nature can be perceived to be aesthetically pleasing by our brain (Redies, 2007). Past research has studied how visual characteristics of nature, such as complexity, symmetry, angularity, can affect customer liking or other customer behaviour. However, the study of fractal dimension and its effect on customer behaviour is limited. In addition, past studies tend to focus on how these visual characteristics is only linked to aesthetic liking, while the connection to any marketing behaviour is not widely discussed. This thesis will explore on how can we use the right wallpaper to increase more aesthetic liking by focusing our research in one of the most important visual characteristics of nature, fractal dimension, and whether it can influence the customers to touch the smartphone. In other words, we are going to focus on the underlying visual grammar of nature and how it could influence aesthetic liking and enhancing the probability to touch the product. Moreover, we are going to stay away from exploring the meaningful content of nature (plants, trees, etc.).
Touching is an interesting customer behaviour because this habit can also measure other customer behaviours such as willingness to buy. In 2003, Illinois state attorney general’s office warned shoppers to be cautious with retailers who encourage them to hold and touch the products when shopping. It is argued that physically holding the product may lead to unplanned purchases (Peck & Shu, 2009).
The research will be tested by using wallpapers with low, intermediate, and high fractal characteristics, and observe whether certain fractal values (low, intermediate, high) are perceived to be more aesthetically pleasing by the participants. In addition, we will also investigate whether the smartphone brands (high-end and low end) moderates the effect of the wallpapers’ attractiveness toward aesthetic liking.
To conclude, the author propose the following research question along with the sub questions:
Does using smartphone’s wallpapers with intermediate values of fractal dimension can increase the likelihood to touch the smartphone?
Is the effect of fractal dimensions in smartphone wallpapers towards aesthetic liking moderated by the smartphone brand?
2. Literature Review
The literature review is meant to introduce the theoretical background of this research to the readers. On this section, we will start by exploring what kind of smartphone designs customers find attractive and how wallpapers can affect smartphone attractiveness. This will be linked to our main discussion, which is the fractal geometry study, a study on the visual characteristics of nature. This research will put an emphasis on fractal dimensions. Furthermore, we will see how past fractal dimension study is limited compared to other visual characteristics’ studies that are linked to customer behaviour. In addition, we will explore on how fractal dimension is connected to aesthetic liking, and eventually lead to touching the product. Afterwards, we will also discuss how touching can also influence willingness to buy, and how smartphone brand images moderate the effect of wallpapers toward aesthetic liking.
2.1 Importance of Smartphone Design & Wallpapers
Past research has investigated on how customers judge the smartphone designs. The design is often judged in terms of its colour, finishes, and build materials. For example, Nanda and colleagues (2008) investigated how smartphone’s colour can affect aesthetic liking, where male participants would prefer darker smartphone colours compared to females. In addition, Chuang et al (2001) and Yun et al (2003) focused on how phone’s materials such as shape, and size influence the customer’s aesthetic liking. They found that the smaller and the lighter the phone was better. The result of these studies were mainly applicable in 2000s, where each smartphone brands had distinct smartphone designs in regard to colour, shape, and size. Nowadays the innovation in smartphone design has reached saturation. All smartphones tend to look the same in terms of their design as they are mostly focusing in creating even thinner phones. Therefore, focusing on design is not enough to differentiate themselves.
Interestingly, wallpapers used on smartphones are very unique to each smartphone brands. The wallpapers are more than just displaying beautiful images. It serves a purpose. As visual stimuli, it can provide a quality perception, creating strong associations with the brand and might make it easier to attract attention from the customers (Henderson et al., 2003). The wallpapers are found to provide enhanced pleasure for customers, and the use of the right wallpaper can increase the aesthetic liking of the smartphone. Heimler argued that the wallpapers of smartphone can complement the smartphone design because it creates a visual stimulus that customers find interesting and can give more positive judgments towards the smartphone design (2002). It is interesting that wallpapers can have positive effect towards customers. We argue that smartphone design should not only about the physical features, but paying more attention on the wallpapers can enhance the liking of smartphone design.
We understand that smartphone design is an important criterion for customers. Due to similarities between each smartphones’ design, the use of wallpapers can help smartphone brands to differentiate themselves from other competitors, enhance the aesthetic liking of the design, and finally attract attention from the customers. However, how can we know which wallpapers or images are aesthetically more pleasing? This will be discussed in the following section.
As we have discussed before, the use of smartphone wallpapers can enhance the aesthetic liking of smartphones. Before we move along further, it is necessary to briefly discuss about aesthetic experience. Aesthetics is tightly linked to our perception (Zeki, 1999) and the perception itself is mediated by the sensory organs and the brain. Similary, Redies (2007), on his research, he argued that aesthetic perception is affected by many different channels and regions of visual system in our brain, and it is part of human nature.
The study of aesthetic perception started from the 19th century. At this time, laws of human vision was defined (Wurtz & Kandel, 2000) and researchers began to understand how our brain is linked to the aesthetic perception. More importantly, their studies focused on how humans’ aesthetic perception is developed from looking at visual arts such as artworks. Particularly, they are exploring how the underlying elements of the arts can influence aesthetic liking. We can also consider wallpapers from smartphones, computers, and TVs as visual arts because these wallpapers can be aesthetically pleasing too. Furthermore, what makes certain wallpapers to be more aesthetically pleasing can be examined by the visual elements hidden in the wallpapers.
Past research in aesthetics study did investigate how the underlying characteristics of images can affect aesthetic liking. There are several elements of visual characteristics related to aesthetic liking that are widely discussed: complexity, symmetry, and angularity. The last one, fractals, is also an important visual element, but the number of studies and its implications are limited compared to other elements of visual characteristics we mentioned briefly before. Before we go further in fractal geometry study, we will discuss the first three visual elements first.
2.2.1 Complexity
that participants perceived both the abstract and representational images that have intermediate level of complexity to be more aesthetically pleasing. The past studies on complexity and aesthetic liking also use other visual stimuli besides artwork. Menzel and colleagues’ (2015) focus on how complexity in face photographs is linked to beauty; Braun and colleagues’ (2013) research investigated the complexity of print advertisements and architectural images.
2.2.2 Symmetry
Osborne (1986) argued that images are symmetrical if they repeat either their main forms or a mirror image of their main forms on either side of a medial axis. In other words, symmetry can exist along one or several axes, and it is related with the amount of redundancy in a stimulus. Consequently, symmetry can reduce the processing demand required to identify images (Leder, 2013). Osborne further pointed out that symmetry is related to aesthetic liking. The repeating pattern provides a very elementary aesthetic stimulus and may serve to arouse attention. Scientist and mathematicians often linked symmetry with beauty (Mcmanus, 2005). However, philosophers and art historian believe that symmetry is often perceived as rigid. Asymmetry appears to be more dynamic and less predictable beauty (Osborne, 1986). Symmetrical images are everywhere, especially in nature: reflection of a mountain in a lake, starfish, flowers, etc.
Symmetry have been used in numerous experiments linked to aesthetic liking. As an example, this visual element is used to determine the aesthetic liking of facial photographs in Menzel and colleagues’ research (2015). In addition, Jacobsen and colleagues’ research (2006), regarding how brain is correlated to aesthetic judgment of beauty, employed symmetry because aesthetic judgments are known to be often guided by criteria of symmetry.
2.2.3 Angularity
Barona’s research (2009) investigated the angularity effect in aesthetic preference. Their conclusion also supported Bar and Neta’s findings (2009), where participants found that angular hexagons are less aesthetically pleasing than round circles. It is argued that sharp contoured objects are perceived to inflict physical harm to the individual. Bar, Neta (2007), and Larson et al (2009) supported this hypothesis, they found that there was an increase of amygdala activity in response to the presentation of abstract shapes representing facial expressions of threat. However, the literature on preference on rounded shapes compared over angular shapes has its own inconsistencies. Westerman and colleagues (2012) insisted that even though angular shapes are associated with threat, and more aggressive negative states, the valence attributed to curved shapes is less clear and may be determined by many factors, such as orientation, texture, and familiarity.
2.3 Psychological Frameworks
Now that we have discussed symmetry, complexity, angularity, we are going to discuss the psychological frameworks that explain why these elements of visual characteristics can affect aesthetic liking.
2.3.1 Processing Fluency Theory
opportunity for learning and more positive affective response. However, based on processing fluency theory, more complex images have more information or elements that need to be processed. That is why people would prefer images with intermediate level of complexity. Some empirical studies, however, showed contrasting result. For example, Landwehr and colleagues (2011) research found that there is a positive effect of complexity such that prototypical but complex car designs are most successful in terms of sales.
2.3.2 Arousal Theory
Berlyne (1970) postulated that human aesthetic preferences are a function of a stimulus’ “arousal potential” (which was largely determined by visual complexity and aesthetic preference) in an inverted-U shaped pattern, in which, medium level of complexity was often found to be preferred. Berlyne (1970) suggested that when respondents were presented with complex stimuli, they feel uncertainty and arousal. However, further increase in complexity would reduce arousal. In other words, visual stimuli with low complexity leads to low preference and arousal. People will seek to maintain a level of arousal that is constant with their preferred level of stimulation. Highly aroused individuals will seek out certainty, while people with low arousal will seek out more stimulation, more complex, less certain, visual environments (Forsythe et al, 2011). According to Gestalt psychology, humans tend to find patterns in stimuli, even the ones that are randomly generated (Hochberg, 1968). Each repetition of the stimulus presents more opportunities for the randomness to be simplified or encoded by the respondent. In regard to symmetry, symmetrical patterns are predictable, but it provides a very elementary aesthetic stimulus and may serve to arouse attention.
2.4 Fractals in Nature
visual characteristic of environment that can affect aesthetic liking. The use of fractal studies are appropriate in this study because fractals are often found in natural environment. In addition, to my knowledge, none of these aesthetic literature had attempted to make any links to marketing research. These studies’ results solely discussed aesthetic liking or preference, but it is unclear whether their findings are applicable to real marketing situation. In this research, we are going to apply our findings in fractals to real marketing situation.
Interestingly, we found that most smartphone brands often use wallpapers with nature forms. From the high-end brand such as Apple to lower-end brand such as Huawei, they all use natural sceneries wallpapers (Figure 2.1 and 2.2). These nature forms are unique and possess certain visual characteristics that our brain finds pleasing (Redies, 2007). Natural objects have high degree of fractal content (Gouyet 1996) and humans have preference for such environments (Kaplan & Kaplan, 1989). To determine which kind of wallpapers are more preferred by customers, we can explore this more by focusing our research in fractals.
The study of nature’s fractal characteristics started in the 1970s. Mandelbrot (1982) was the famous mathematician that coined the term “fractals” for the very first time and in his book, Fractal Geometry of Nature, he argued that fractals consist of patterns that recur on finer and finer scales, building scale-invariant shapes of immense complexity. Objects that are made by humans tend to have smooth characteristic with no rough edges. In contrast, the boundaries of natural forms are often best characterized by irregularity and roughness.
An example of natural fractal object is tree. Trees, when we zoom in or zoom out our view towards the branches, the patterns of the branches observed have the same statistical qualities (Taylor, 2011). This is often called as self-similarity. High-self similarity implies that an image as a whole has an appearance similar to its parts. The self-similarity characteristic are mostly found in natural sceneries such as the trees, etc. For example, in Figure 2.3, a photograph taken by Taylor (2011) shows how the trees’ branches have similar patterns when we magnify the picture, and the branches represent the high-self similarity. In other words, fractals have been described as the “fingerprints of nature” (Taylor, 2003) because its repeating patterns can be found in mountain ranges, coast lines, clouds, rivers, Figure 2.3 Tree Branches
trees, plants, etc (Gouyet, 1996). Thus, fractals geometry is an important visual characteristic of nature that can help us to investigate which kind of wallpapers are aesthetically more pleasing.
The term ‘biophilic’ fractal is used to classify fractals that possess positive aesthetic qualities that are deeply rooted in their natural appearance. It is defined by their resulting organic visual aesthetics. Taylor and Sprott (2008) asserted that fractals represent an amazing visual element for aesthetic investigations: artificial patterns that can capture important visual qualities of our natural environment.
2.5 Fractal Dimension
The link between fractals and aesthetic perception started from Clint Sprott’s work in 1993. He proposed the use of fractal dimension to measure the complexity of fractal images, and to investigate the relationship between fractals and aesthetic perception. Fractal dimension (D) reflects the density of edges in binarized images and is closely related to complexity (Mureika, & Taylor, 2013). Fractal dimension value usually ranging between 1 and 2. For example, a smooth line contains no fractal structure at all has a D value of 1, while completely filled area such as a square has a value of 2 because it is able to cover the whole surface. By repeating the patterns of fractal line, it will make the line to begin to occupy space but not entirely filling the whole surface. These patterns of fractal lines create a D value between 1 and 2. Taylor & Sprott (2008) suggested that by increasing the amount of the fine structure in the fractal’s repeating patterns, the D value moves closer to 2 but it will never be exactly at 2.
Each natural objects appear to have its own unique fractal dimension. Numerous past studies, for example, have found that in average, plants and trees have D value of 1.28-1.90 (Morse et al., 1985); snowflakes with D value of 1.7 (Nittmann & Stanley, 1987); waves with D value of 1.3 (Werner, 1999); and clouds with D value of 1.30-1.33 (Lovejoy, 1982). Figure 2.4, Figure 2.4 Clouds and Tree Branches
taken by Taylor and colleagues (2011), illustrates the natural environment along with its fractal dimension value. The clouds and the forests have D value of 1.3 and 1.89 respectively. For images with low fractal value, the small content of the fine structure builds a smoother, sparse shape. In other hand, images with fractal value close to 2, the larger content of fine structure builds a shape full of intricate detailed structure (Taylor & Sprott, 2008). We can see this from the tree branches image. With its higher fractal dimension value, it appears to have more rough edges, and more complex patterns compared to the clouds. Forsythe and colleagues (2011) conducted a study on preference of natural scenery images. Their research found that pictures of natural environments were rated more beautiful and more fractal than other images (abstract images, figurative images) by participants and fractal dimension was accounted for 30% of the variance in judgments of the beauty of natural scenes.
2.6 Fractal Aesthetics
Interestingly, fractal images are also found on computers as screensavers. Past studies have investigated how these computer-generated fractal images can enhance aesthetic liking and produce positive feelings (Draves et al, 2007; Taylor & Sprott, 2008). Electric sheep is a good example, it is regarded to be sophisticated in terms of both the inventive methods used to create the images and the appeal of the visual appearance (Taylor & Sprott, 2008). Electric Sheep was first developed by artist Scott Draves. Initially the images were called as fractal flames. As the algorithm to make the fractal images expanded, the images were later called as Electric Sheep. Figure 3 shows the examples of Electric Sheep images taken from Taylor and Sprott’s (2008) research. Interestingly, the fractal pattern of these screensavers are very similar to the fractal patterns found in the natural environment. Participants were asked to rate the aesthetic appeal of these Electric Sheep images, and they concluded that participants perceived the images with intermediate fractal values (1.3-1.5) to be more aesthetically pleasing. In addition, they also found that images with mid-D reduce their physiological stress, and they feel more relaxed.
meshes with increasingly small square sizes. Their study concluded that fractals generated by nature, mathematics and art that have intermediate fractal dimension value (1.3-1.5) have the highest aesthetic liking.
2.7 Neurophysiological Response to Fractals
We have concluded that many past research suggested images with intermediate fractal dimension value to have the highest aesthetic liking. We will try to explain why this is the case.
It is argued that images with mid-D can reduce stress, increase feeling of relaxation and enhance aesthetic liking (Taylor, 2006). Taylor and Sprott’s (2008) experiment investigated on respondents’ brain activity when they are exposed to images with intermediate D. It is concluded that this specific value range activate distinctly different visual areas of the brain than high-D fractals (Watts & Taylor, 2007). Likewise, EEG experiments found that alpha waves (a state of being wakefully relaxed) are optimum for intermediate D (Hagerhall et al., 2008). In addition, in 2011, Taylor and colleagues conducted another research on neurophysiological responses toward fractal images. This is done by monitoring participant’s quantitative EEG response while viewing fractals images. They concluded that images with intermediate D generated maximal alpha response in the frontal region of our brain, and therefore it is more relaxing and aesthetically more pleasing. This does make sense because the aesthetic experience that we feel is a product of brain function (Zeki, 1999). Another theory, which is the Savannah theory, suggested that humans prefer lower D values simply because it mimics the complexity of the African savannah scenery, where our ancestors spent considerable amount of time in this landscape, and the low visual complexity facilitates detection of predators in the surrounding vegetation (Wise, & Leigh-Hazzard, 2000). Consequently, this theory believed that we perceived images with high D to be too intricate and complicated. Therefore, my first hypothesis is:
H1: Wallpaper with intermediate fractal dimension will lead to higher aesthetic liking in comparison to wallpaper with lower or higher fractal dimensions.
how aesthetic liking of the wallpapers can influence touching, and also willingness to pay. Moreover, the smartphone brand image will also be discussed.
2.8 Touching
Most past literatures (Peck, & Childer, 2003; Peck, & Wiggins, 2006; Peck, & Shu, 2009; Hulten, 2013) only focused on conducting studies that show the benefits of touching products, or usually they would argue that shoppers touch the products simply because they want to obtain information of the product and make evaluation (McCabe & Nowlis, 2003). However, it is interesting to see that these past research did not explore on how the aesthetic side of the products itself might have an influence on the shopper’s attention and willingness to touch the products. Although Hulten (2013) did mention that shoppers generally want to touch products because they are attracted to it, the reason why was not elaborately explained. Therefore, this research is trying to link how the visual stimuli from the smartphone can affect aesthetic liking and touching the product.
In retail atmosphere, shoppers experience brands and products through vision, sound, smell, touch, and taste (Hulten, 2013) which highlights the significance of sensory cues and stimuli. Touching a product can have a persuasive influence on customer’s attitudes and behaviour. Peck & Childers (2003) found that touching can increase purchasing intention towards the product and increase the confidence in the evaluation of the products. People touch to gather information about a product and to help them make judgments. Previous marketing studies suggested that some product categories encourage touching more than others (Grohmann, Spangenberg, & Sprott, 2007). Smartphones falls into this category, because sense of touch allows customer to judge the texture, hardness, and weight (Klatzky, & Lederman, 1992). Touching appears to have a powerful effect towards purchasing decision. In fact, even if there is no information regarding the product, touching still can influence persuasion significantly (Peck & Wiggins, 2006).
(Schiffman, 2001). It is evident that electronics gadgets such as smartphones rely on tactile senses to determine its product features such as hardness, roughness, weight (Boyd, 2011). The use of smartphone wallpaper can create a visual stimulus that can attract our attention, because shoppers are attracted to images that are aesthetically pleasing for them. The use of visual stimuli might enhance the probability to touch the product (Kahn, & Deng, 2010). The aesthetics of the wallpaper is the first step to attract the customer, because we will rely on our vision to observe the visual stimuli. Our visual impression will further tell us to try get closer to the product and touch it (Ludden et al., 2006). Gallace & Spence (2011), on their research on aesthetics and touching, suggested that if customers perceived a certain product to be aesthetically pleasing, they will more likely to touch the product. Once a consumer is paying attention to a product based on its tactile properties, it is less likely that they will shift their attention to a competing product or brand. If the smartphones can capture their audiences based on the tactile input they will have a clear advantage over their competitors, which is important in a cluttered market places.
In addition, we also expect that effect of the wallpapers’ aesthetics towards touching can also lead to more willingness to buy. Aesthetic liking plays an important role for purchase decisions (Faerber, et al 2010). As we have discussed before, customers rely on tactile input to make product evaluations, specifically the design. The wallpapers can attract customers because it is aesthetically pleasing, and chances are they will try to approach the product and try. The more probability to touch the product, the more chances the customers would buy it. Consequently, I propose:
H2 : Wallpaper with intermediate fractal dimension will directly influence the probability of the customers to touch the smartphone in comparison to wallpapers with low or high fractal dimension.
H3 : Wallpapers with higher aesthetic liking will increase the probability for the customers to touch the smartphone
H4 : The more customers touch the smartphone, the more they are willing to pay.
2.9 Smartphone Brand Image
associations perceived by the consumer including the personality projected by the brand (Batra, & Homer, 2004). Brand associations can help motivate consumers to use their brand because of the social and psychological nature of the brand and the feelings and attitude arise from using the brand (Sirgy 1982). Knapman (2014) argued that consumers are strongly influenced by brand when it comes to buying smartphones. Remedios & Nathwani (2014) research suggested that the participants’ preference for smartphone depends heavily on the brand. They concluded that participants are not willing change their smartphone brand even if the other brands have the same functionality as their current smartphone brand. Participants associated Apple’s & Samsung’s phone to be in a high quality due to its premium pricing, great design and the brand logo itself simply make their phones to be perceived in a more positive brand image. Emerging brands from China such as Xiaomi, Huawei, OnePlus also offer similar premium design and the same functionality with significantly lower price than Apple and Samsung. However, their smartphones are seen to be inferior due to their poor brand image. Consumers tend to perceive these emerging brands as copycats because their phone design are very similar to the high-end brands. We argue that smartphone brand image may influence the effect of wallpapers’ aesthetics. High-end smartphone brands like Apple and Samsung already have the brand image of high quality and design, or in other words, shoppers already see these brands as aesthetically pleasing.
With the addition of using wallpapers with intermediate fractal dimension, customers should find these high-end smartphone brands to be even more aesthetically pleasing due to their brand association of high quality, beautiful design, In other words, the wallpapers can complement the smartphone design’s aesthetics. Although emerging brands such as Xiaomi, Huawei, OnePlus can also enjoy the benefit of using wallpapers with similar value of fractal dimension, the effect might not be as significant as the higher-end brands due to their inferior brand image on their design. I propose:
H5: Smartphone brand will directly influence the aesthetic liking of the smartphone design.
2.10 Construal Level
Construal level theory (CLT) is one of the leading contemporary theory of mental construal, with rich implications and applications in consumer science. Trope & Liberman’s (2003) construal level approach has become a notable topic for social psychology in general, and for its research on judgment and decision making in particular. CLT can be used to explain and predict the consumer behaviour in terms of consumer’s preference, purchasing decisions and intentions, brand representation, risk taking, and other consumer behaviour (Fiedler, 2007). Consumer preference is part of psychological-distance dimension, a central variable of CLT. For example, customers often have to choose between two products, one product has more “desirable” attributes (design, image of the of the product), and one product that has more “feasible” attributes (cheap, ease of use). CLT suggested that customers would prefer to chose products that have more “desirable” attributes than the “feasible” counters. This can be applied to our situation, where customers would often choose smartphone brands that have more desirable qualities such as the premium design, and luxurious brand. In addition, Dhar & Kim (2007) argued that consumer mindset is at a greater psychological distance from the moment of purchase, where abstract attributes of the products are very important (design, quality). Conversely, in the final purchasing decision, the low-level features (price) might be emphasized. Thus, in our case, smartphone design is an important attribute because customers pay a lot of their attention on the abstract attributes first when it comes to looking for a new smartphone. The fractal wallpapers can enhance the “desirable” attribute of the smartphone.
individuals who can see their action in terms of causal effects, social meanings, and self-descriptive implications. To put it into our situation, individuals with low construal levels will look and buy smartphones simply because they need one. They buy it because the smartphone offer functional benefits such as to communicate with other people, access the internet, and to take photos. In other hand, people with high construal level will buy smartphones because of its abstract or desirable attributes, such as the premium design, the quality that the brand conveys, and the feeling you get from buying premium smartphone. In our research, we are going to see whether construal level can affect any of the consumer behaviour.
2.11 Conceptual Model
3. Methodology
We analyzed previously collect data in order to investigate the effect of wallpapers with three different levels of fractal dimension values (low, intermediate, high) towards aesthetic liking, how the liking on the wallpapers can affect touching, and willingness to buy. Additionally, we wanted to investigate the moderating effect of smartphone brand image (higher end, lower end) on the relationship between the wallpapers and aesthetic liking. In this data collection section we will give a thorough information on the data and its measurements.
We expect to have higher aesthetic liking in wallpapers with intermediate fractal dimension value, compared to wallpapers with low or high fractal dimension value. As a consequence, we argued that the higher aesthetic liking in wallpapers with intermediate value will lead to more probability to touch the smartphone and buy it. In addition, we expect that the effect of wallpapers towards aesthetic liking will be higher in higher-end smartphone compared to lower-end smartphone.
3.1 Participants and Design
240 participants participated in this study, with 30 participants in each condition. The participants were mainly from Indonesia and The Netherlands. The study was 2 (smartphone brand: Apple, OnePlus) x 4 (Wallpaper type: control, low dimension, intermediate dimension, and high dimension) between-subjects design. Therefore, the survey has 8 version and each respondent only took part in one of the condition. In order to receive large number of respondents in such limited amount of time, all the participant were approached via internet, particularly social media applications (Facebook, Line, Whatsapp). We asked 240 people via personal message on these social media applications to participate our survey.
3.2 Materials
3.2.1 Smartphone Brand
We used Apple to represent the high-end smartphone brand, and OnePlus for lower-end smartphone brand. These two brands are widely popular in their respective consumer segment, and therefore they are the perfect choice. We chose the latest phones released by the these two brands in 2015: iPhone 6s and OnePlus 2. We expect that the effect of wallpapers towards aesthetic liking is significantly higher in Apple compared to OnePlus brand.
3.2.2 Wallpapers
As we can see from Figure 3.2,there are four types of wallpapers used. To make sure that the fractal wallpapers does have an effect on consumer behaviour, we used the white wallpaper as the control condition to compare it with the main conditions later in the analysis. In control condition, respondents are more likely to judge the smartphone simply by its design and the brand. While in main conditions, we expect that the fractal images will affect the way the respondents perceive the smartphone design. The fractal images were created from www.sciencevsmagic.net/fractal. This site lets us simply create fractal images through five different control in regard to the the base, segments, mirror, depth, and the angle of the fractals. We created one set of similar fractal images to maintain the consistency of the result. We ended up producing three similar fractal images, and the difference between them is only the number of iterations used. For example, to produce fractal with low FD we used 2 iterations, mid FD with 3 iterations, and high FD with 4 iterations.
3.3 Measurements, Correlation, and Reliability Analysis
We wanted to investigate to what extent smartphone fractal wallpaper and smartphone brand image can affect aesthetic liking, touching, and willingness to buy. In this section, we will explain how we measure each of the variables using the appropriate scales. Three constructs in the model (aesthetic liking, touching, and willingness to buy) are measured with multiple questions or items. Before proceeding with the analysis, we are going to check whether each items correlate with each other, and thus constitute a reliable scale. If they do, we are going to make a new sum variable that depicts the whole construct up in one score..
3.3.1 Aesthetic Liking
showed that the internal consistency of the scale including the items was satisfactory (Cronbach’s α = 0.94). Thus, we can sum the six items that measure aesthetic liking into one variable (SUMAestheticLiking).
3.3.2 Touching
In terms of touching, we asked respondents to what extent they “want to approach where the smartphone is displayed”, whether they “want to take the smartphone in their hand,” whether they want to “touch the smartphone” and whether they want to “try the smartphone” using 7-points Likert scale from strongly disagree to strongly agree. These scales are adapted from Hulten’s (2013) research. We noticed that studies on touching behaviour are often conducted through experimental design and observational method. Consequently they do not hand out surveys to the participants. Therefore, we tried to adapt the questions from their observation sheet into series of statements. For example in Hulten’s (2013) observation sheet, there was an observation on “Approach behaviour of shoppers in terms of getting close to the product(yes/no)” and we changed it into “I would like to come closer to the product” to better fit our survey. These four items significantly correlate with each other (Appendix C2), and reliability analysis showed that the four items measuring touching have a Cronbach’s α = 0.94. Therefore, we can sum the four items into one variable (SUMTouching).
3.3.3 Willingness to Buy
To measure respondents’ willingness to buy, we asked respondents to what extent they agree with the following statements using 7-point Likert scale from strongly disagree to strongly agree: “I would like to reach out and grab the smartphone”, “the smartphone’s design really “speaks” to me and I feel that I must buy it”, and “the smartphone has a really great design and I feel a strong urge to buy it”. The scales are based on Bloch and colleagues (2013) in regard to visual product aesthetics and willingness to buy. In our survey, the statements are adapted specifically for the smartphone design. The three items significantly correlate with each other (Appendix C3), and reliability analysis showed that the three items have Cronbach’s α = 0.87. Thus, we can sum the three items into one variable (SUMBuying).
3.3.4 Behaviour Identification
to measure individual differences in level of personal agency. Each item on the Behaviour Identification Form represents an act identity followed by two alternative identities : one lower and one higher in level of construal. In our survey, we used fourteen behavioural statements. High construal identification is valued as 1, while low construal identification is valued as 0. Therefore, to determine whether participants have high construal level, we simply calculate how much high construal identifications that they choose. Participants who have a total score of 7 and higher are considered to have high construal level. The reliability analysis showed that the fourteen statements have Cronbach’s α = 0.66. Thus, we can sum all the fourteen BIF statements into one variable (SUMBIF). Correlation analysis showed that participants’ construal level is highly correlated with aesthetic liking (r = -0.172, p < 0.001) and willingness to buy (r = -0.127, p < 0.001). However, construal level is not correlated with touching behaviour (p = 0.389).
3.4 Procedure
To collect data from participants, we designed a questionnaire (Appendix A) and distributed it to the internet. The questionnaire started with a brief introduction on the purpose of this research, putting emphasis on how participants’ involvement are crucial. Demographic questions, age and gender, were asked in the following section. There were eight versions of the survey, and each participants only took part in one condition. To make sure that the survey was distributed evenly and randomly, we used the randomize logic from Qualtrics.
In the questionnaire, we asked our respondent to imagine a situation where they are in phone store looking for a new phone. Respondents were then presented with a picture of the smartphone using a wallpaper, and were asked about their opinion in regard to the aesthetics of the smartphone design (six items), their willingness to touch (four items) and buy the phone (three items). In the last section of the survey, we asked respondents to identify fourteen behavioural statements.
4. Results
4.1 Descriptive Statistics
To get an overview, we provide descriptive statistics of the sample (Appendix B). Table 4.1. shows the distribution of respondents’ gender and age. In total, two hundred and forty respondents participated in our questionnaire, and within each condition 30 respondents participated. The sample consists of 105 (44%) men and 135 (56%) women, with an average age of 28 (SD = 9.33). The age ranged from 17 to 70 years old.
4.2 Hypotheses Testing
4.2.1 2-Way ANOVA : Wallpapers and Smartphone Brand on Aesthetic Liking
In order to analyse the influence of fractal wallpapers and smartphone brand image on aesthetic liking, we performed a 4 (wallpapers: control vs low FD vs mid FD vs high FD) x 2 (smartphone brand: iPhone vs OnePlus) between subjects ANOVA on aesthetic liking.
For the main effects, high end smartphones are liked significantly more than their low-end counterparts, F(1, 232)= 19.72, p < .001. Second, the different fractal wallpapers had a significantly different effect on aesthetic liking, F(1, 232)= 34.34, p < .001. This suggest that changes in aesthetic liking is caused by the fractal wallpapers. In terms of interaction effect, there was no significant wallpaper by smartphone brand interaction for aesthetic liking, F(1, 232)= 0.67, p = 0.57.
As we can see from Figure 4.2.1.A and Figure 4.2.1.C, wallpapers with low FD, mid FD, and high FD) scored higher aesthetic liking than using white wallpaper. This supports our literature review where we argued that fractal wallpapers can increase aesthetic liking. Based on the pairwise comparison test (Table 4.2.1.B), for iPhone, changes in aesthetic liking from white wallpaper to : low FD, mid FD, and high FD are all significant. Changes in aesthetic liking from low FD wallpaper to : mid FD, and high FD are all significant. Lastly, changes in aesthetic liking from mid FD to high FD wallpaper is not significant. In regard to OnePlus, changes in aesthetic liking from white wallpaper to : Low FD, Mid FD, and High FD are all significant. Changes in aesthetic liking from Low FD to Mid FD, and High FD are not significant. Similarly, changes in aesthetic liking from Mid FD to High FD is also not significant
Table 4.2.1.C Means of Fractal Wallpapers’ Aesthetic Liking Between Smartphone Brands
There appears to be a peak on aesthetic liking in wallpaper with mid fractals. In Figure 4.2.1.A, we can see that the mean difference between mid FD wallpaper and high FD wallpaper in aesthetic liking is not much in each smartphone brands. The pairwise comparison test showed that difference between mid FD wallpaper and high FD wallpaper is not significant in both iPhone and OnePlus. This means that the increase of aesthetic liking from mid FD wallpaper to high FD wallpaper is not significant. Lastly, Figure 4.2.1.A showed us that participants prefer the high-end smartphone more than the lower end. From white wallpaper to high fractal wallpaper, the high-end brand has more aesthetic liking than the lower-end.
33 4.2.2 2-Way ANOVA : Wallpapers and Smartphone Brand on Touching
In order to analyse the influence of fractal wallpapers and smartphone brand image on touching, we performed a 4 (wallpapers: control vs low FD vs mid FD vs high FD) x 2 (smartphone brand: iPhone VS OnePlus) between subjects ANOVA on touching.
For the main effects, high end smartphones were touched significantly more than their low-end counters, F(1, 232)= 8.78, p < .001. Second, the different fractal wallpaper had a significantly different effect on touching F(1, 232)= 31.14, p < .001 meaning that changes in willingness to touch is caused by fractal wallpapers. In terms of interaction effect, there was no significant wallpaper by smartphone interaction for touching, F(1, 232)= 1.52, p = 0.21.
Figure 4.2.2.A Figure 4.2.1.A Means of Willingness to Touch Between Fractal Wallpapers and Smartphone Brands . Error Bars 95% CI.
As we can see from Figure 4.2.2.C, wallpapers with low FD, mid FD, and high FD scored higher touching than using white wallpaper. Based on the pairwise comparison test (Table 4.2.2.B), For iPhone, changes in touching from white wallpaper to: Low FD, Mid FD, High FD are all significant. Changes in touching from Low FD wallpaper to Mid FD is significant, while High FD is not significant. Lastly, changes in touching from Mid FD to High FD is also not significant. In regard to OnePlus, changes in touching from white wallpaper to: Low FD, Mid FD, and High FD are all significant. Changes in touching from Low FD to Mid FD and High FD are not significant. Similarly, changes in touching from Mid FD to High FD is not significant.
Similar to aesthetic liking, there appears to be a peak on touching in wallpaper with mid fractals. In Figure 4.2.2.A, we can see that the mean difference between mid FD wallpaper and high FD wallpaper in touching is not much in OnePlus brand. Interestingly, iPhone with high FD wallpaper results in decreasing willingness to touch. Lastly, Figure 4.2.2.A showed us that participants prefer to touch the high-end smartphone more than the lower end. From white wallpaper to high fractal wallpaper, the high-end brand has more likeness of touching than the lower-end.
We can conclude that fractal wallpaper with higher FD will lead to higher probability of touching, specifically using wallpaper with mid FD. The use of high fractal wallpaper does not seem to affect touching significantly. Thus, H2 is partially supported. Wallpaper with intermediate dimension value will significantly lead to higher probability of touching in comparison to wallpaper with lower or higher fractal dimensions, specifically in Apple brand. Fractal wallpapers in high-end brand can enhance more probability of touching than the lower-end brand. Regardless how low or how high the wallpaper’s fractal value is, consumers will prefer to touch more of the high-end smartphone brand than the lower end.
4.2.3 2-Way ANOVA : Wallpapers and Smartphone Brand on Willingness to Buy
In order to analyse the influence of fractal wallpapers and smartphone brand image on willingness to buy, we performed a 4 (wallpapers: control vs low FD vs mid FD vs high FD) x 2 (smartphone brand: iPhone VS OnePlus) between subjects ANOVA on willingness to buy.
For the main effects, high end smartphones are significantly have more willingness to buy than their low-end counters (F(1, 232)= 18.02, p < .001. Second, the different fractal wallpaper had a significantly different effect on willingness to buy F(1, 232)= 22.47, p <
.001, meaning that changes in willingness to buy is caused by the fractal wallpapers. In terms of interaction effect, there was no significant wallpaper by smartphone interaction for willingness to buy, F(1, 232)= 1.57, p = 0.19.
As we can see from Figure 4.2.3.A, wallpapers with low FD mid FD, and high FD scored higher willingness to buy than using white. Based on the pairwise comparison test (Table 4.2.3.B), for iPhone, changes in WTB from white wallpaper to: Low FD, Mid FD, and High FD are all significant. Changes in WTB from Low FD to Mid FD is significant, while High FD is not significant. Lastly, changes in WTB from Mid FD to High FD is significant. In regard to OnePlus, changes in WTB from white wallpaper to: Low FD, Mid FD, and High FD are all significant. Changes in WTB from Low FD to Mid FD and High FD are all not significant. Lastly, changes in WTV from Mid FD to High FD is also not significant.
Similar to aesthetic liking and touching, there appears to be a peak on willingness to buy in wallpaper with mid fractals. In Figure 4.2.3.A, we can see that the marginal difference between mid FD wallpaper and high FD wallpaper in willingness to buy is quite high specifically in iPhone. High fractal wallpaper in iPhone leads to significant drop in willingness to buy and it is statistically significant, p < .001. Lastly, Figure 4.2.3.A showed us that participants are more willing to buy the high-end smartphone brand than the lower end. From white wallpaper to high fractal wallpaper, the high-end brand have more willingness to buy than the lower-end.
4.3 Correlation Analysis on Touching and Willingness to Buy
We wanted to check whether touching can influence willingness to buy. To test these two variables, we conducted correlation analysis. We found that touching and willingness to buy is highly correlated and is significant (r= 0.770, p < 0.001). This suggests that the more customers touch the smartphone, the more they willingness to buy. Therefore, H4 is supported.
4.4 Mediation Analysis on Touching
To see whether the effect of fractal wallpapers on touching is mediated by aesthetic liking, we used the macro Process by Hayes. We performed mediation analysis by comparing each two fractal wallpapers (low vs mid, mid vs high, low vs high).
4.4.1 Mediation on Apple Brand (Low FD, Mid FD, High FD)
We performed the first mediation analysis on Apple brand. We used touching as the dependent variable, fractal wallpaper ( low FD, mid FD) as the independent variable, and aesthetic liking as the mediator. The analysis showed that the bias-corrected 95% confidence interval with 1000 bootstrap samples for the indirect effect of low FD and mid FD wallpaper on touching through aesthetic liking did not include zero (0.09 to 1.44). This suggest that the increasing probability of touching behaviour from low fractal wallpaper to mid fractal wallpaper is mediated by aesthetic liking.
For the second mediation analysis, with only using wallpapers with mid FD and high FD, the analysis showed that the bias-corrected 95% confidence interval with 1000 bootstrap samples for the indirect effect of mid FD and high FD wallpaper on touching through aesthetic liking did include zero (-0.17 to 0.39). This means that the increasing probability of touching behaviour from mid fractal wallpaper to high fractal wallpaper is not mediated by aesthetic liking.
4.4.2 Mediation on OnePlus Brand (Low FD, Mid FD, High FD)
We performed mediation analysis on OnePlus brand. We used touching as the dependent variable, fractal wallpaper (low FD to mid FD) as the independent variable, and aesthetic liking as the mediator. The analysis showed that the bias-corrected 95% confidence interval with 10000 bootstrap samples for the indirect effect of low FD and mid FD wallpaper on touching through aesthetic liking did include zero (-0.07 to 1.01).
Similarly, For mid FD and high FD wallpaper, the analysis showed that the bias-corrected 95% confidence interval with 10000 bootstrap samples for the indirect effect of mid FD and high FD wallpaper on touching through aesthetic liking did include zero as well (-0.29 to 0.55).
Lastly, for low FD and high FD wallpaper, the analysis showed that the bias-corrected 95% confidence interval with 10000 bootstrap samples for the indirect effect of low FD and high FD wallpaper on touching through aesthetic liking also included zero (-0.03 to 0.40). This means that the increasing probability of touching behaviour from all fractal wallpaper in OnePlus brand is not mediated by aesthetic liking.
To conclude the mediation analysis, we can say that H3 is partially supported. Aesthetic liking mediates the effect of fractal wallpaper on touching, particularly in Apple brand with mid fractal wallpaper. There is no mediating effect of aesthetic liking on fractal wallpapers with OnePlus brand. Fractal wallpaper with higher aesthetic liking will result in more probability to touch the smartphone.
4.5 Mediation Analysis on Willingness to Buy
To see whether the effect of fractal wallpapers on willingness to buy is mediated by aesthetic liking, we used the macro Process by Hayes. We performed mediation analysis by comparing each two fractal wallpapers (low vs mid, mid vs high, low vs high).
4.5.1 Mediation on Apple Brand (Low FD, Mid FD, High FD)
FD wallpaper on WTB through aesthetic liking did not include zero (0.13 to 1.54)..
Similarly, between low FD and high FD wallpaper, the analysis showed that the bias-corrected 95% confidence interval with 10000 bootstrap samples for indirect effect of low FD and high FD wallpaper on WTB through aesthetic did not include zero (0.17 to 0.73). Thus, the increasing probability of willingness to buy from low FD to mid FD wallpaper, and from low FD to high FD wallpaper is mediated by aesthetic liking.
In contrast, with mid FD and high FD wallpaper, the analysis showed that the bias-corrected 95% confidence interval with 10000 bootstrap samples for indirect effect of mid FD and high FD wallpaper on WTB through aesthetic liking did include zero (-0.39 to 0.60). Aesthetic liking does not mediate the increasing probability of willingness to buy from mid FD to high FD wallpaper.
4.5.2 Mediation on OnePlus Brand (Low FD, Mid FD, High FD)
For low FD and mid FD wallpaper, the analysis showed that the bias-corrected 95% confidence interval with 1000 bootstrap samples for the indirect effect of low FD and mid FD wallpaper on WTB through aesthetic liking did include zero (-0.07 to 0.91).
Similarly, Between mid FD and high FD wallpaper, the analysis showed that the bias-corrected 95% confidence interval with 10000 bootstrap samples for the indirect effect of mid FD and high FD wallpaper on WTB through aesthetic did include zero as well (-0.34 to 0.65).
Lastly, for low FD and high FD wallpaper, the analysis showed that the bias-corrected 95% confidence interval with 10000 bootstrap samples for the indirect effect of low FD and high FD wallpaper on WTB through aesthetic did include zero too (-0.05 to 0.40). This suggest that aesthetic liking does not mediate the effect of fractal wallpaper towards willingness to buy on OnePlus.
wallpaper with higher aesthetic liking will result in more willingness to buy.
4.6 Construal Levels
As we can see from Table 4.6.1, most of the participants answered all fourteen statements with highly construal answers. In each statements, more than half of the participants chose the highly construal identification. In average, participants chose ten highly-construal identifications from fourteen statements. We can conclude that most of our participants are high-level agents, they are the individuals who can see their action in terms of causal effects, social meanings, and self-descriptive implications. This suggests that these individuals would look and buy smartphone simply not because of its functional benefits, but more of the desirable attributes that the smartphone can offer (design, brand image).
4.6.1 Two-Way ANOVA : Wallpaper and Construal Level on Aesthetic Liking
In order to analyse the influence of fractal wallpapers and construal level on aesthetic liking, we performed a 4 (wallpapers: control vs low FD vs mid FD vs High FD) x 2 (construal level: Low vs High) between subjects ANOVA on aesthetic liking. For the main effect, construal level does not affect aesthetic liking significantly, F(1,232)= 1.42, p = 0.23, meaning that changes in aesthetic liking is not caused by construal level. In terms of interaction effect, there was no significant wallpaper by construal level interaction for aesthetic liking, F(1,232)= 1.28, p = 0.28.
In both high and low construal level, wallpapers with low FD, mid FD, high FD scored higher aesthetic liking than using white wallpaper. Based on Table 4.6.1.B, for low construal level, changes in aesthetic liking from white wallpaper to low FD is not significant, while mid FD and high FD are all significant. Changes in aesthetic liking from low FD to mid FD is significant, while high FD is not significant. Lastly, changes in aesthetic liking from mid FD to high FD is also not significant. In regard to high construal level, changes in aesthetic liking from white wallpaper to: low FD, mid FD, and high FD are all significant. Changes in Figure 4.6.1.A Means of Fractal Wallpapers’ Aesthetic
Liking Between Construal Levels. Error Bars 95% CI.
aesthetic liking from low FD to mid FD is significant, while to high FD is not significant. Finally, changes in aesthetic liking from mid FD to high FD is not significant. It is interesting to see that compared to low construal, the highly construal participants rated the white wallpaper and low FD wallpaper more aesthetic liking. While for Mid FD and high FD, low construal participants rated more aesthetic liking than the highly construal. For low construal participants, there is a peak on aesthetic liking in wallpaper with mid FD.
4.6.2 Two-Way ANOVA : Wallpaper and Construal Level on Touching
In order to analyse the influence of fractal wallpapers and construal level on touching, we performed a 4 (wallpapers: control vs low FD vs mid FD vs High FD) x 2 (construal level: Low vs High) between subjects ANOVA on touching. For the main effect, construal level affect touching significantly F(1,232)= 6.06, p < 0.001, meaning that changes in willingness to touch is caused by construal level. In terms of interaction effect, there was significant wallpaper by construal level interaction for touching, F(1,232)= 3.65, p < 0.001.
Figure 4.6.2.A Means of Willingness to Touch Between Fractal Wallpapers and Construal Levels. Error Bars 95% CI.
As we can see from Figure 4.6.2.A, wallpapers with low FD, mid FD, and high FD scored higher touching than using white wallpaper. According to Table 4.6.2.B, for low construal level, changes in touching from white wallpaper to low FD and high FD are not significant, while to mid FD is significant. Changes in touching from low FD to mid FD is significant, while to high FD is not significant. Lastly, changes in touching from mid FD to high FD is significant. In regard to high construal level, changes in touching from white wallpaper to low FD, mid FD, and high FD are all significant. Changes in touching from low FD to mid and high FD are not significant. Similarly, changes in mid FD to high FD is not significant. Compared to low construal, the highly construal participants rated more touching in white wallpaper and low FD wallpaper. While for low construal participants, they rated more touching in mid FD and high FD wallpapers. For low construal participants, there is a peak of willingness to touch in wallpaper with mid FD.
4.6.3 Two-Way ANOVA : Wallpaper and Construal Level on Willingness to Buy
In order to analyse the influence of fractal wallpapers and construal level on WTB, we performed a 4 (wallpapers: control vs low FD vs mid FD vs High FD) x 2 (construal level: Low vs High) between subjects ANOVA on WTB. For the main effect, construal level does Figure 4.6.3.A Means of Willingness to Touch Between Fractal
Wallpapers and Construal Levels. Error Bars 95% CI.
not affect WTB significantly, F(1,232)= 0.26, p = 0.60. In regard to interaction effect, there was no significant wallpaper by construal level interaction for WTB, F(1,232)= 1.09, p = 0.352.
In both high and low construal level, wallpapers with low FD, mid FD, high FD scored higher WTB than using white wallpaper. Based on Table 4.6.3.B, for low construal level, changes in WTB from white wallpaper to: low FD, mid FD, high FD are all significant. Changes in WTB from low FD to mid FD is significant, while to high FD is not significant. Lastly, changes in WTB from mid FD to high FD is not significant as well. In regard to high construal, changes in white wallpaper to: low FD, mid FD, high FD are all significant. Changes in WTB from low FD to mid FD is significant, while to high FD is not significant. Lastly, changes in WTB from mid FD to high FD is not significant. Interestingly, highly construal participants scored more WTB in white wallpaper and mid FD wallpaper. In contrast, low construal participants scored more WTB in mid FD and high FD. For both high and low construal level, there is a peak of WTB in wallpaper with mid fractal.
5. Conclusion
5.1 General Discussion
Past literature on fractals and its effect on customer behaviour is limited. Previous studies tend to focus on how visual characteristics is only linked to aesthetic liking, while the connection to any marketing behaviour is not widely discussed. We found interesting results that are consistent with our literature review.
This paper contributes to the literature by using smartphone fractal wallpaper to measure aesthetic liking, touching, willingness to buy, and whether smartphone brand also play an important part. By testing the first hypothesis, we found that wallpaper with intermediate fractal dimension value will significantly lead to higher aesthetic liking in comparison to wallpaper with lower or higher fractal dimensions, specifically in Apple brand. This is consistent with past studies, where images with mid fractal value is more preferred. Thus, H1 is partially supported. In regard to fractal wallpaper and smartphone brands, we found both high-end and low-end smartphone brand to have more aesthetic liking with fractal wallpaper. However, the effect is higher in high-end smartphone brand. Consistent with our argument in literature review, high-end smartphone brand has better brand image when it comes to smartphone design. Thus, the use of fractal wallpaper enhances their design’s aesthetic even more. Hence, we can accept H5. However, H6 is rejected, as we found that smartphone brand does not moderate the effect of fractal wallpapers on aesthetic liking. It appears that smartphone brand only have a direct influence on aesthetic liking.
mediation effect with OnePlus brand. This is consistent with our findings from literature, where customers are more willing to touch products that they found aesthetically pleasing. The use of fractal wallpaper can attract them to try and touch the smartphone. Therefore, H3 is partially supported.
Lastly, we included willingness to buy because it is significantly correlated with touching. Past literature have found that touching products can increase willingness to buy. Similar to touching, we found that fractal wallpaper with mid dimension value will lead to more willingness to buy. We also found that aesthetic liking mediates the effect of fractal wallpaper on willingness to buy, particularly in Apple brand with mid fractal wallpaper. Similar to mediation with touching, we found no mediating effect with OnePlus brand. In addition, we also found that most of the participants have high-construal levels. In each statements in behaviour identification form, more than 50% participant chose the highly-construal identification. We found that 210 participants have high highly-construal level. This would mean that these participants would look and buy smartphone simply not because of its functional benefits, but more of the desirable attributes that the smartphone can offer (design, brand image). Construal level affect touching significantly, but it did not affect aesthetic liking and WTB significantly.
It is true that fractal wallpaper with intermediate fractal dimension is indeed more preferable in regard to aesthetic liking and interestingly to marketing behaviours. We believe that fractal wallpaper with mid value have the right amount of complexity, and number of iterations. Linking back to processing fluency theory, fractal wallpaper with mid value often have symmetrical patterns as well, so it has enough information or elements that our brain can process. The higher the fractal value, the more elements that our brain need to process. In regard to arousal theory, participants should perceive fractal wallpaper with mid value to have enough complexity and creates arousal. Further increase in complexity, or fractal value, would reduce arousal.
