The influence of digital signage on consumers: how
consumer characteristics and emotions evoke the most
favourable consumer behavior.
By
Tessa Koomen
The influence of digital signage on consumers: how
consumer characteristics and emotions evoke the most
favourable consumer behavior.
By Tessa Koomen
University of Groningen Faculty of Economics and Business
Master’s Thesis Msc. Marketing Management Completion date: February 12, 2020
Professor Rankestraat 14a 9713GE Groningen t.koomen@student.rug.nl
+316 15 59 07 24 Student number: 2777290
Supervisor: dr. J. Berger
Abstract
This research investigates the influence of digital signage content on consumer behavior. We conducted an experiment to examine 1) which digital signage content (affective, cognitive, or mixed) evokes the most favourable consumer behavior; 2) what the role of emotions (positive and negative) is in this relationship; and 3) if consumer characteristics have an influence on the consumer behavior evoked by digital signage content. An experiment tests the theoretically derived hypotheses and indicated that the relationship between digital signage and consumer behavior is not statistically significant proven. We found that positive and negative emotions caused by digital signage increase word-of-mouth intentions. Furthermore, the findings suggest that the consumer trait of extraversion shows the strongest moderating influence of all traits on affective digital signage content on word-of-mouth. Conscientious consumers have the strongest direct negative influence on word-of-mouth.
Keywords: digital signage, emotions, cognitive, word-of-mouth, willingness-to-pay, Big Five
Management summary
Companies spend a large share of their budget on in-store marketing, hoping to grab consumers at the point of purchase to increase sales. Digital signage is one of the newest in-store marketing stimuli and is used for many purposes, such as advertising and the provision of news. However, in the existing digital signage literature, a gap is highlighted by Garaus et al. (2017) in the effectiveness of the drivers and different behavioural responses. This study sets out to address a part of this research gap. Therefore, we have examined the effect of different types of digital signage content (affective, cognitive or mixed) on consumer behavior (willingness-to-pay and word-of-mouth) via emotions (positive and negative) with the moderating effect of consumer characteristics (Big Five Traits). This study will give marketers a better understanding of the effectiveness of digital signage and consumer behavior on this stimulus. With a better understanding of the available marketing tools and consumer behavior, marketers are able to use the tools in such a way that it will alternate consumer behavior in a positive way.
We have conducted an online experiment to determine three outcomes: 1) which digital signage content (affective, cognitive, or mixed) evokes the most favourable consumer behavior; 2) what the role of emotions (positive and negative) is in this relationship; and 3) do consumer characteristics have an influence on the consumer behavior evoked by digital signage content. We expected that the use of digital signage would increase willingness-to-pay and word-of-mouth intentions, but that the influence of affective content was stronger than cognitive or mixed content. Furthermore, positive emotions would have a positive mediating influence on the relationship between digital signage and consumer behavior, while negative emotions would have a negative mediating influence. Finally, we expected that the consumer traits of extraversion, neuroticism and openness would react stronger to affective digital signage content, while conscientiousness and agreeableness would react stronger to cognitive content.
We have tested the theoretically derived hypothesis by executing an online experiment. The participants were randomly exposed to one of the three experimental conditions of the experiment, either affective, cognitive or mixed content. The three experimental conditions were identical, except the content of the digital signage. We have created two different stimuli for the each of the experimental conditions, which are based on previous research and the study by Garaus et al. (2017). To analyse the results of our experiment we have used SPSS software. Regression analysis techniques have been used to examine the effects between digital signage, willingness-to-pay, word-of-mouth and emotions and consumer characteristics.
Table of contents
I. Introduction 1
II. Literature review 4
a. Theoretical background 4
b. Research model and hypotheses development 8
III. Methodology 18
a. Participants and research design 18
b. Stimulus generation 18
c. Measures and procedure 19
d. Plan of data analysis 22
IV. Results 24
a. Descriptives 24
b. Convergent validity 24
c. Analysis 25
V. Discussion 31
a. Future research directions and limitations 33
b. Theoretical implications 33
c. Managerial implications 34
VI. References 36
VII. Appendixes 44
A. Digital signage content 44
B. Questionnaire scales and validity 45
C. Experiment questionnaire English 46
D. Experiment questionnaire Dutch 48
E. Hayes (2017) PROCESS model 5 on willingness-to-pay 51 F. Hayes (2017) PROCESS model 15 on willingness-to-pay 52 G. Hayes (2017) PROCESS model 15 on word-of-mouth 53
1 I. Introduction
Instead of deciding upfront what to purchase, more than 76% of the consumers decides in-store what they will purchase (The Point of Purchase Advertising International, 2012). This means that companies can have a large influence on this decision, because they have till the point of purchase to trigger consumers. Companies are able to create more touch points with the consumer when consumers decide what to purchase in-store compared to when they decide on forehand. Therefore, a lot of companies spend a large share of their budget on in-store marketing. With this in-in-store stimuli, companies hope to “grab consumers” at the point of purchase with their merchandising to increase sales (Abratt and Goodey, 1990). Kollat and Willet (1969) suggest that in-store stimuli act as reminders of shopping needs and therefore assist in making purchase decisions and offering consumers new ways of satisfying needs. In-store stimuli can be defined as techniques such as in-In-store siting, on-shelf positions, price-off promotions, sampling, point-of-purchase displays, coupons and in-store demonstrations (Abratt and Goodey, 1990).
But are these companies spending their budget in the right way? Is in-store marketing really that effective? One of the newest in-store stimuli is digital signage, which is according to Dennis, Michon, Brakus, Newman, and Alamanos (2012) mostly a flat LCD or plasma screen linked with digital content. The aim of digital signage is to communicate with shoppers while they are in the process of buying. Digital signage is used for many purposes, such as advertising, the provision of news and community information, and the enhancement of image. Where traditional point-of-sale (POS) media use predefined content to an unspecific crowd, digital signage provides strategies for presenting promotions, such as showing specific information on a specific date and location (Turov, Shilov and Teslya, 2019). The flexibility of the medium is also mentioned by Müller et al. (2009) as a benefit. Where traditional POS-media is in-store during a predefined time period, digital signage content can be changed every second. This would mean that during daytime, the digital signage could be showing diapers advertisements, while in the evening hours beer advertisements are shown. With digital signage, retailers are able to show more relevant content to the consumer compared to traditional POS-media. Besides these benefits, different studies have found positive effects of digital signage (e.g. Dennis et al., 2012; Garaus, Wagner and Manzinger, 2017). The use of digital signage systems instead of static displays in public places is increasing (Harrison and Andrusiewicz, 2004), and also urban areas and retailers start to recognize the potential of an extra attention moment (Bauer, Dohmen and Strauss, 2011). Because the increase in use and its high potential, we will focus on the effects of digital signage to help marketers, manufacturers and retailers in understanding the effectiveness of digital signage.
2 these resources and to give them a possibility to differentiate from their competitors (Abratt and Goodey, 1990).
Digital signage is thus an important tool for retailers. Even though a lot of research has been done on general in-store marketing, Garaus et al. (2017) highlighted the need for further research in the area of digital signage. They did a field experiment where they studied what the effect of digital signage content was on behavioural responses such as impulsive purchases and store loyalty, through the mediating effect of emotions and cognitive evaluations, like store image and perceived merchandise quality. The results of this study indicated that affective digital signage content evokes more positive emotions than cognitive or mixed content and that affective digital signage content increases impulsive purchases and store loyalty. Furthermore, positive emotions mediate the relationship of digital signage on impulse purchases, while emotions (both positive and negative) and perceived merchandise quality mediate the relationship between digital signage and store loyalty. After their experiment they highlighted a gap in the existing literature in the effectiveness of drivers of digital signage, such as consumer characteristics, situational variables, and habituation effect. This study sets out to address a part of this research gap with the focus on consumer characteristics, measured by the Big Five Traits. Because of time constraints it is not possible to study all drivers mentioned by Garaus et al. (2017). Hu and Jasper (2006) concluded in their study that stores should be more personable, which could be achieved by introducing social cues that are relevant to consumers’ lifestyles and values. However, at this moment personalization of digital signage is still challenging (Turov et al., 2019). The ideal digital signage solution would be providing individualized content for the audience at any time (Wibotzki, Sandkuhl, Smirnov, Kashevnik, and Shilov, 2017). Therefore, by focussing on consumer characteristics in this study, we will be able to recognize different consumers and create specific content for them. More knowledge on consumer characteristics in combination with digital signage will simplify personalization of digital signage content. Another research gap pointed out by Garaus et al. (2017) is the use of different behavioural responses, namely word-of-mouth and willingness-to-pay. This research will also address this research gap and study the relationship between digital signage and word-of-mouth and willingness-to-pay. Therefore, this study extends the research by Garaus et al. (2017) by studying different dependent variables (word-of-mouth and willingness-to-pay) and add the moderating effect of consumer characteristics.
3 characteristics (the Big Five Traits) have an influence on the consumer behavior evoked by digital signage content.
4 II. Literature review
We will start the literature review with discussing the different concepts that will be used in this study. The aim of discussing these concepts is to set clear definitions, which we can use during the rest of this study. Afterwards, section b will introduce the conceptual model and hypotheses will be developed based on previous research.
a. Theoretical background Digital signage
The main concept of this research is one of the newest in-store marketing tools, digital signage. The existing literature has various definitions for this term. Digital signage is described as a flat LCD or plasma screen linked with digital content by Dennis et al. (2012). Garaus and Wagner (2019) describe a digital signage system as a private screen network that displays varied content. The definition of Wibotzki et al. (2017) is even more precise, they define digital signage as “the provision of content (advertisements, news, assistance) on
electronic scoreboards or large displays in places where many people are present or passing by”. Different researchers have already showed the positive effects of digital signage on
consumer behavior. Garaus and Wagner (2019) found a positive relationship between digital signage mounted at the checkout area and overall store satisfaction through reducing the perceived waiting time and creating favourable waiting experiences. According to Roggeveen, Nordfält and Grewal (2016) the long-term sales of hypermarkets could increase by 3% due to digital signage. Dennis et al. (2012) showed that digital signage adds positive perceptions of the mall environment, emotions, and approach behavior, such as spending. This study will focus on the different types of advertisement content that a digital signage can present. According to Dennis et al. (2012) digital signage content typically includes advertisements, community information, entertainment, and news. Burke (2009) found that messages that relate to the task at hand and the shoppers’ current need state get the most response of shoppers. In the advertising literature, a differentiation is made between emotional (affective) and informational (cognitive) advertisement. This study will follow the content types used by Garaus et al. (2017) and will differentiate between three types of digital signage content: affective, cognitive, and mixed content.
5 Garaus et al. (2017) are not the only ones who found differences between the three content types. Where affective advertisements focus on feelings, cognitive advertisements focus on thinking and according to Zajonc (1980), feelings and thinking are two independent evaluation systems. More recent research by Dennis, Newman, Michon, Brakus and Wright (2010) has further developed this finding. They found that emotional content is processed via the peripheral route, while informative content is processed via the central route, which is consistent with the Elaboration Likelihood Model by Petty and Cacioppo (1986). The ELM is a model that describes how attitudes are created and changed. The model suggest that there are two routes to persuasion: the central route and the peripheral route. The central route has a high elaboration likelihood and is based on the careful and thoughtful consideration of arguments central to the issue. The behavioural outcome will be determined by the consumers’ cognitive response (Dennis et al., 2010). On the other hand, the peripheral route has a low elaboration likelihood and is based on the affective associations or simple inferences tied to peripheral cues (Petty and Cacioppo, 1986). The behavioural outcome is not determined by cognitive processes, but by external factors such as emotional appeal (Dennis et al., 2010). Shiv and Fedorikhin (1999) support this model by arguing that the cognitive processing required for cognitive stimuli is more than for emotional stimuli, since emotional stimuli are processed automatically and cognitive stimuli require more elaborate processing. The first type of digital signage content this study will use is affective content. Based on the ELM, affective content will be processed through the peripheral route, since the content will lead to affective associations. Matzler, Bidmon and Grabner-Kräuter (2006) describe affect as “valenced feeling states and emotions”. According to Edell and Burke (1987) affective advertisements give rise to feelings by creating a particular experience during product use. Yi (1990) found that emotional reactions are triggered by affective context and that affective context influences brand evaluations. For the development of affective advertisements, faces are commonly used because these evoke automatic responses (Nenkov and Scott, 2014). Hu and Jasper (2006) agree that affective content is often manipulated by faces. Furthermore, they found that two out of three affective response dimensions (pleasure and arousal) are positively affected by in-store graphics manipulated with faces.
6 important to the potential consumer and should present data which the consumer accepts as being verifiable (Puto and Wells, 1984).
In this research, we will use a third type for digital signage content, namely mixed content. This content type will combine affective and cognitive content. We have added this third content type to make sure the differences in results we might get are assigned to either one of the two conditions, and not just coincidence.
Willingness-to-pay
In this research we will focus on consumer behavior by utilising the following two variables: willingness-to-pay and word-of-mouth, since these variables require further research according to Garaus et al. (2017) in the field of digital signage.
In marketing research, price is an important variable since it contributes to sales volumes, margins and product positioning (Le Gall-Ely, 2009) and it is the only element of the marketing mix that generates income (Breidert, 2005). The number one problem many marketing executives face is setting the right price for their product or service (Breidert, 2005). One of the most important price variables is willingness-to-pay. Willingness-to-pay can be defined as the maximum price a buyer accepts to pay for a given quantity of goods and services (Kalish and Nelson, 1991; Kohli and Mahajan, 1991; Wertenbroch and Skiera, 2002; Le Gall-Ely, 2009). Willingness-to-pay is an interesting price variable to study, since the variable enables economists to estimate the demand curve according to price and design optimal price schedules (Gafni, 1998; Wertenbroch and Skiera, 2002).
Breidert (2005) makes a distinction between the reservation price and the maximum price. The distinction between these prices is that the reservation price does not depend on a reference product, while the maximum price does. Consumers’ willingness-to-pay depends on the perceived economic value and on the utility of the good (Breidert, 2005). If the consumer believes that there is no alternative offering, the willingness-to-pay is equal to the utility of the good, which is the reservation price. If the consumer believes that there is an alternative offering, but with an economic value below utility, the willingness-to-pay is equal to the economic value of the product, which is the maximum price.
Word-of-mouth
The other variable, word-of-mouth, is an important consumer behavior variable, since positive word-of-mouth reduces the need for marketing expenditures and it might increase revenue when attracting new consumers (Reichheld and Sasser, 1990). Word-of-mouth is defined by Westbrook (1987) as “informal communications directed at other consumers about the
ownership, usage, or characteristics of particular goods and services and/or their sellers”.
Hennig-Thurau, Gwinner, Walsh and Gremler (2004) describe word-of-mouth as “any
comments (positive or negative) received or spread by the actual, former or potential customer about any product or service”. Brown, Barry, Dacin and Gunst (2005) adopt in their
7 one does business with a company or store and making positive recommendations to others about a company or store. The power of word-of-mouth is that it is more credible than commercial sources of information, that it is two-way communication and it is considered a risk reliever by providing potential consumers with a description of what the experience would be (Derbaix and Vanhamme, 2003).
Word-of-mouth can be used for shaping the impressions other people and themselves have, the regulation of emotions, connecting with others, information acquisition, or convincing others to a certain action or opinion (Berger, 2014). Word-of-mouth is the result of consumer involvement, in which three states are identified: product-involvement, self-involvement, and other-involvement (Westbrook, 1987). Product-involvement is when the consumer wishes to talk about the purchase and the gratifications the purchase provides. When the consumer wants to tell others about the purchase to gain attention, recognition, or status, the consumer is situated in the self-involvement state. When the consumer wants to help other consumers by sharing knowledge or experience, we speak of the other-involvement state (Berger, 2014).
Emotions
Previous research has already shown that the relation of a stimulus and consumer behavior is influenced by emotions created by that specific stimulus. This study will investigate what the mediating influence of emotions is on the relationship between digital signage and consumer behavior.
James (1884) has described emotions as (sets of) sensations that resulted from a physiological response to a stimulus. However, recently disagreement arose about the definition of emotions and the boundaries of it (Adolphs, Mlodinow and Feldman Barret, 2019). Most emotion scientists view emotions as a complex response pattern with experiential, physiological, cognitive and intentional elements (Omdahl, 1995; Scherer, 2005). Schreuder, Van Erp, Toet and Kallen (2016) define emotion as “a short-term state that is directly related to the
environmental stimuli”. They argue that this state could be observed consciously or processed
unconsciously. The most complete definition of emotion is given by Yuan and Dennis (2016): “a subjective feeling related to personal needs, goals, or concerns towards the self or others
and is typically triggered by events or objects in one’s environment rather than internal factors”. Chaudhuri (2006) has identified three categories of emotions: emotion characterized
8
Consumer characteristics
Besides the mediating effect of emotions, we will also study the moderating effect of consumer characteristics on the relationship between digital signage and consumer behavior. For the description of personality traits, the Big Five Traits is regarded as the most widely accepted and most prominent model (Rammstedt, Goldberg and Borg 2010). The Big Five Traits is a model to describe personality traits, and interpret and predict human behavior. Abood (2019) describes a trait as something what makes and distinguishes an individual, the inner desire of the individual that affects his behavior frequently. The Big Five Traits model consists of the following five traits: extraversion (surgency), neuroticism (emotional stability), conscientiousness (dependability), agreeableness, and culture, intellect or openness (Goldberg, 1992). It is necessary to mention that the traits are not opposites, but orthogonal dimensions (Matzler, Faullant, Renzl and Leiter, 2005). Srivastava, Chandra and Shirish (2015) have described the personalities in different traits. They relate the traits of being social, active, outgoing, and experiencing positive emotions related to extraversion. Neuroticism refers to the traits of being anxious, depressed, angry, embarrassed, emotional, worried, and insecure. Being careful, thorough, responsible, organized, hardworking, achievement-oriented, and persevering are the traits of conscientiousness. Agreeableness is referred to as being kind, considerate, likable, helpful, cooperative, courteous, flexible, trusting, good-natured, forgiving, soft-hearted, and tolerant. Openness refers to the traits of being imaginative, cultured, curious, original, broad-minded, intelligent, and artistically sensitive, and to the willingness to try new and different things.
b. Research model and hypotheses development
9 Figure 1: Conceptual model
At the end of this research we want to have three outcomes determined: 1) which digital signage content (affective, cognitive, or mixed) evokes the most favourable consumer behavior, 2) what is the role of emotions (positive and negative) in this relationship, and 3) do consumer characteristics have an influence on the consumer behavior evoked by digital signage content.
As explained before, the S-O-R-theory (Mehrabian and Russel, 1974) suggests that a stimulus triggers a response mediated by the internal evaluation of the organism. We can apply this theory to our study with digital signage as stimulus and consumer behavior (willingness-to-pay and word-of-mouth) as the response. Based on the S-O-R-theory, we can expect a difference in consumers’ response when showing the digital signage to consumers. This is supported by various studies (Dennis et al., 2010; Dennis et al., 2012; Dennis, Brakus and Alamanos, 2013). These studies found that digital signage content contains cues which evoke specific experiences in customers, which in turn, positively affect consumers’ approach behavior.
10 consumers are willing to spend more. When consumers spend more, this could be due to buying more items, but it could also indicate a higher willingness-to-pay. Roggeveen et al. (2016) found that in hypermarkets digital signage result in greater sales receipts, more items purchased and more time spent. A greater sales receipt and more items purchased imply that consumers are willing to spend more, which could also indicate that people are willing to pay more for one product. Based on previous literature, we expect the relationship between digital signage and consumers’ willingness-to-pay to be positive.
Hypothesis 1a: Digital signage has a positive influence on consumers’ willingness-to-pay.
Different studies have already shown that consumers’ purchase behavior can be significantly influenced by word-of-mouth (Arndt, 1967; Brown and Reingen, 1987). Keller (2007) found that positive word-of-mouth mediates the relationship between advertising content and consumer responses. Hogan, Lenon and Libai (2004) found that word-of-mouth can be increased by increasing advertising. Since digital signage shows advertisements, this would imply that there is a positive relationship between digital signage and word-of-mouth. According to Oliver (1997) the pleasurable fulfilment of needs and desires leads to satisfaction and according to Brown et al. (2005) leads satisfaction to positive word-of-mouth. Various studies (Willems, Smolders, Brengmana, Luyten, and Schöning, 2017; Garaus and Wagner, 2019) have found the positive influence of digital signage on store satisfaction. This implies that digital signage increases satisfaction, which in turn leads to positive word-of-mouth intentions. Besides this, digital signage is an extra attention moment in-store, which creates an extra experience for the consumer, thereby increasing their involvement. According to Westbrook (1987) and Berger (2014), word-of-mouth is the result of consumer involvement. We expect that by creating an extra experience, in the form of digital signage, consumers are more inclined to talk with their relatives about the experience. Therefore, the effect of digital signage on consumers’ word-of-mouth intentions will be positive.
Hypothesis 1b: Digital signage has a positive influence on consumers’ word-of-mouth intentions.
The Elaboration Likelihood Model describes how attitudes are created and changed. The model suggests that affective content is processed through the peripheral route, while cognitive content is processed through the central route. This implies that the digital signage content types are processed differently through consumers and the consumer response for each content type will be different. The affect referral hypothesis (Wright, 1975) suggests that in the decision-making process of choosing brands, consumers choose the brand for which the retrieved affect is the most positive instead of using specific attribute information. Based on this theory, consumers would prefer affective digital signage content instead of cognitive or mixed content.
11 (affective vs. cognitive vs. mixed) on behavioural responses (impulse purchases and store loyalty) with the mediating effect of emotional processes (positive and negative emotions) and cognitive processes (store image and merchandise quality perceptions). The results of their study confirmed that the most positive emotions and most favourable store and merchandise evaluations are evoked by showing affective digital signage content rather than cognitive or mixed content. These results support the previously mentioned theoretical reasoning that affective content result in more favourable responses than mixed or cognitive content. The study by Garaus et al. (2017) has partly the same subject as present study, however we will study willingness-to-pay and word-of-mouth as dependent variables, and additionally investigate the moderating effect of consumer characteristics. Garaus et al. (2017) showed that the most favourable behavioural responses are evoked by showing affective digital signage content. Based on Garaus et al. (2017) we would expect that affective digital signage content has a higher influence on consumer behavior (willingness-to-pay and word-of-mouth) than cognitive digital signage content.
But not only Garaus et al. (2017) have found positive influences of affective content. Sherman, Mathur and Smith (1997) have found that pleasure, one of the dimensions of affective response, positively affect money spent and liking of the store. Furthermore, they found that one of the other dimensions of affective response, arousal, has as well a positive influence on money spent, but also on time spent and the number of items purchased. This implies that affect increases the willingness to spend, and with that the willingness-to-pay. Bigné and Andreu (2004) found that consumers who experience greater pleasure and arousal showed an increased level of satisfaction and more favourable behavioural intentions, such as willingness-to-pay more. Park, MacInnis, Priester, Eisingerich and Iacobucci (2010) showed that consumers who feel emotionally attached to a brand are willing to spend more social and financial resources. By showing affective (emotional) content, consumers feel more emotionally attached to a brand (Dennis et al., 2013) which might result in higher word-of-mouth intentions and higher willingness-to-pay. On the other hand, cognitive content shows informative content, which does not make consumers more emotionally attached.
Mixed digital signage content is a combination of affective and cognitive content. We expect that the cognitive components will negatively influence the effectiveness of the mixed digital signage content, while affective components will positively influence the effectiveness. For that reason, we expect affective digital signage content to have a higher influence on consumer behavior (willingness-to-pay and word-of-mouth) than mixed digital signage content.
Hypothesis 2a: Compared to cognitive and mixed digital signage content, affective content has the strongest positive influence on consumers’ willingness-to-pay.
12 Since mixed content will consist of components of affective and cognitive content, we expect that the influence of both components will determine the effectiveness of mixed content. We have hypothesized that affective digital signage content will have the strongest positive influence on our consumer behavior variables. Therefore, we expect that this positive influence will be reflected in the mixed content, while the influence of cognitive content will negatively reflected. Therefore, the expectation is that the stronger influence of affective content in mixed content will make the influence of mixed content stronger than the influence of cognitive content. Although little previous research is available, this expectation is supported by findings of Dennis et al. (2014) and Garaus et al. (2017).
Hypothesis 2c: Compared to cognitive digital signage content, mixed content has the strongest positive influence on consumers’ willingness-to-pay.
Hypothesis 2d: Compared to cognitive digital signage content, mixed content has the strongest positive influence on consumers’ word-of-mouth intentions.
Following Garaus et al. (2017) we will make a distinction between positive and negative emotions. The Construal Level Theory (CLT) argues that an individuals’ willingness to think about abstract and future goals could be increased by positive emotions, while negative emotions increase consumers’ focus on immediate and proximal concerns (Yuan and Dennis, 2016). Edell and Burke (1987) have shown that positive and negative feelings can be generated by advertisements and that both feelings can make unique contributions to a shopper’s attitude towards the advertisement. Combining these findings with the study of Yoo, Park and Macinnis (1998) who found that store characteristics induce shoppers’ in-store emotions, supports our mediating effect of emotions on the relationship between digital signage content and consumer behavior.
Dennis et al. (2012) have studied the effect of emotions within the digital signage field. They evaluated the impact of digital signage in a shopping mall on shoppers’ perception of the retail environment, positive affect, and approach behavior. The results indicated that digital signage enhances shoppers’ evaluation of the retail environment. Positive emotions are triggered by the shoppers’ assessment of their environment, which influence shoppers’ approach behaviors, such as additional spending. This finding is consistent with the principle of cognitive mediation (Lazarus, 1991) which describes that a specific stimulus, such as digital signage, influences the perception of other attributes of the store environment. This perception will influence emotions and in turn will influence behavior.
13 distracted from negative mood states by a high working memory demand. A number of other studies have supported this theory (DeFraine, 2016; Kron, Schul, Cohen and Hassin, 2010; Davidson and Irwin, 1999). Applying this theory to digital signage indicates that digital signage will create an extra attention moment in-store and thus demanding working memory. Therefore, we expect that consumers’ negative emotions are negatively influenced by digital signage.
Hypothesis 3a: Digital signage has a negative influence on consumers’ negative emotions.
Edell and Burke (1987) have already shown that positive and negative feelings can be generated by advertisements. They also found that affective advertisements give rise to feelings (positive and negative) by creating a particular experience during product use. Hu and Jasper (2016) found that affective content for in-store graphics positively affects pleasure and arousal, which are two dimensions of affective response. This finding is confirmed by Dennis, Brakus, Gupta and Alamanos (2014), who found that affective experiences are created by affective digital signage content. Therefore, we expect that affective content will have a positive influence on positive emotions. This expectation is also confirmed by the study of Garaus et al. (2017) who found that positive emotions are evoked by affective digital signage content.
Hypothesis 3b: Affective digital signage content has a positive influence on consumers’ positive emotions.
self-14 evaluation due to positive emotions, which induces an increase in the perceived value of object in the environment. Our expectation therefore is that consumers with a high score on positive emotions would be willing to pay more for the product than they would normally, while negative emotions would decrease willingness-to-pay. This is also supported by a study of Peters, Slovici and Gregory (2003), who found that buyers with a stronger positive feeling were willing to pay more.
Hypothesis 4a: Positive emotions caused by the use of digital signage have a positive influence on consumers’ willingness-to-pay.
Hypothesis 4b: Negative emotions caused by the use of digital signage have a negative influence on consumers’ willingness-to-pay.
The variance in word-of-mouth intentions can be explained for a large part by emotional responses (Maute and Dube, 1999). Berger (2014) says that sharing with others (word-of-mouth) facilitates emotion, which also indicates that there is a relationship between emotions and word-of-mouth. Westbrook (1987) found that feelings directly relate to word-of-mouth in the post-purchase period. White (2010) proposed that over time positive emotions will have a positive impact on word-of-mouth, while negative emotions will have a negative impact on word-of-mouth. However, his research only found a significant impact of positive emotions, but not a significant impact of negative emotions. Furthermore, he found that positive emotions lead to higher satisfaction, whereas negative emotions do not seem to influence satisfaction levels. According to Brown et al. (2005) a higher customer satisfaction leads to a spread of positive word-of-mouth. Based on these studies we expect that positive emotions will have a positive influence on positive word-of-mouth intentions. On the other hand, consumers with negative emotions will have a lower customer satisfaction and therefore spread less positive or even negative word-of-mouth (Brown et al., 2005). Since our measurement for word-of-mouth is positively framed, we expect that the relationship between negative emotions and word-of-mouth intentions will be negative. Consumers who are experiencing negative feelings are not likely to encourage, recommend or say positive things about a product.
Hypothesis 5a: Positive emotions caused by the use of digital signage have a positive influence on consumers’ word-of-mouth intentions.
Hypothesis 5b: Negative emotions caused by the use of digital signage have a negative influence on consumers’ word-of-mouth intentions.
The moderating effect of consumer characteristics will be tested by utilizing the Big Five Traits. Previous literature has already shown that the human decision making process is influenced by personality (Adamopoulus, Ghose and Todria, 2018).
15 conscientiousness and agreeableness are more likely to repost negative information on micro blogs, because of their responsibility and care to spread and publicize (negative) information. According to Devaraj, Easley and Crant (2008) consumers with a high score on conscientiousness tend to take actions immediately when necessary and consumers with a high score on agreeableness are enthusiastic and enjoy to cooperate with others. Matzler et al. (2008) found that consumers with a high score on openness have an open and positive attitude to exchange information with others. This is also found by Yin et al. (2020), who found that a high score on openness is related with a high likelihood to share (negative) information with others. The description of these traits support this finding. Being careful and responsible are used for describing conscientiousness, while being kind and helpful are used for describing agreeableness. Therefore, we expect that consumers with a high score on conscientiousness or agreeableness want to share their experiences to help others to make a good choice. Consumers who score high on openness are willing to try new and different things. We expect that these consumers are innovators or early adapters in the Diffusion of Innovation theory by Rogers (1962). This would imply that there is not much (experience) information available. Therefore, we think that consumers with a high openness score are more willing to share information to spread their knowledge. Furthermore, Oreg and Sverdlik (2014) state that consumers with a high score on openness are likely to address their concerns and help targets to see their own perspectives. Based on previous findings, we expect that consumers with a high score on conscientiousness, agreeableness or openness show a stronger positive influence on word-of-mouth intentions.
Hypothesis 6: Consumers with a high score on conscientiousness show a stronger positive influence on word-of-mouth than the other traits.
Hypothesis 7: Consumers with a high score on agreeableness show a stronger positive influence on word-of-mouth than the other traits.
Hypothesis 8: Consumers with a high score on openness show a stronger positive influence on word-of-mouth than the other traits.
16 moderating effects on the relationship between affective digital signage content and consumer behavior (willingness-to-pay and word-of-mouth). Openness is only mentioned in the study by Matzler et al. (2006), but we expect this trait also to have a stronger moderating influence on affective digital signage content on consumer behavior. Openness is described as being imaginative, which we think fits more with emotions than information.
Hypothesis 9a: Consumers with a high score on extraversion show a stronger moderating influence on affective digital signage content on willingness-to-pay than cognitive or mixed digital signage content.
Hypothesis 9b: Consumers with a high score on extraversion show a stronger moderating influence on affective digital signage content on word-of-mouth than cognitive or mixed digital signage content.
Hypothesis 10a: Consumers with a high score on neuroticism show a stronger moderating influence on affective digital signage content on willingness-to-pay than cognitive or mixed digital signage content.
Hypothesis 10b: Consumers with a high score on neuroticism show a stronger moderating influence on affective digital signage content on word-of-mouth than cognitive or mixed digital signage content.
Hypothesis 11a: Consumers with a high score on openness show a stronger moderating influence on affective digital signage content on willingness-to-pay than cognitive or mixed digital signage content.
Hypothesis 11b: Consumers with a high score on openness show a stronger moderating influence on affective digital signage content on word-of-mouth than cognitive or mixed digital signage content.
17 that these traits have a stronger moderating effect on the relationship between cognitive digital signage content and consumer behavior.
Hypothesis 12a: Consumers with a high score on conscientiousness show a stronger moderating influence on cognitive digital signage content on willingness-to-pay than affective or mixed digital signage content.
Hypothesis 12b: Consumers with a high score on conscientiousness show a stronger moderating influence on cognitive digital signage content on word-of-mouth than affective or mixed digital signage content.
Hypothesis 13a: Consumers with a high score on agreeableness show a stronger moderating influence on cognitive digital signage content on willingness-to-pay than affective or mixed digital signage content.
18 III. Methodology
The Methodology chapter will explain how empirical evidence is collected to test the derived hypotheses. The chapter starts with a description of the participants and the research design, followed by an explanation about the generation of the stimulus used in the experiment. Furthermore, we will provide an explanation about the development of the measurement tools and the full procedure. The chapter will be finalized by providing a plan of data analysis, in which is presented how the collected data is going to be analysed.
a. Participants and research design
In order to test the derived hypotheses, an online experiment was created. Participation of this study was voluntary. Participants could engaged in the experiment via online self-selection, which means that a link was shared via social media channels and participants could choose if they wanted to participate. In this online experiment 171 participants participated. However, there were some participants who did not complete the experiment, so these were deleted from the sample, resulting in a sample size of 136.
The study was an online experiment and used a single (content: affective vs. cognitive vs. mixed) between-subjects factorial design. During the experiment participants were asked to evaluate a product without any information on what they needed to focus on. The participants were exposed to one of the three experimental conditions of the experiment: affective vs. cognitive vs. mixed digital signage content. The three experimental conditions were identical, except for the content of the digital signage. The allocation of the content to participants is randomized (44 affective content, 46 cognitive content, 46 mixed content).
Although subjects were assigned at random to the three experimental conditions, an analysis was conducted to determine whether the groups differed on the basis of gender, age and education.
b. Stimulus generation
A digital signage system can show different types of content, such as advertisements, news, and assistance (Wibotzki et al., 2017). The advantage of digital signage is that it is flexible in changing the content it shows (Müller et al., 2009). We have based the design of our digital signage content on the study by Garaus et al. (2017). However, the difference will be that we have used static content rather than dynamic content. Because of time and money constraints it is not feasible to create dynamic content for this study. Ervasti, Häikiö, Isomursu, Isomursu and Liuska (2015) found that there is no significant difference in the effectiveness of digital signage content when comparing still content with video or animation content. Therefore, we expect that static content generates the same results as dynamic content.
19 cognitive advertisement should present factual, relevant information about the brand or product (Puto and Wells, 1984). The cognitive digital signage content therefore displayed information about apples communicated through “Did you know”-questions. Two advertisements have been created, the first advertisement showed an apple splashing in water while informing that an apple consists of water for 85%. The second advertisement informs that apples can decrease the risk of strokes and heart diseases. The mixed digital signage content showed one of the cognitive (scene: an apple consists for 85% of water) or affective advertisements (scene: child eating apple).
We decided to minimize the influence of price communication on willingness-to-pay by leaving any communication about prices. This is supported by Dennis et al. (2015) who found that an advertisement displaying the price of a product influences how much consumers bid. All digital signage contents can be found in Appendix A.
c. Measures and procedure
The experiment was a voluntary experiment, which is shared via different types of social media. When a participant decided to participate in the study, they were asked to select their language (Dutch or English). The study is set up in English and Dutch since we expect a lot of the respondents will be Dutch and we do not want the participants’ English proficiency to influence the test results. The experiment is checked with backward translation. The experiment was completed by 128 Dutch speaking participants and 8 English speaking participants.
After selecting the language, the participants were directed to a welcome page. This welcome page tells the participants that the study takes approximately five to ten minutes of their time and thanks the participants for participating in the study. When continuing to the next page, the participants started with the experiment. Participants were asked to fill in demographical information, such as age, gender, and educational level. We have used this information to test if there are any differences between consumer groups. After completing these questions, the participant was directed to a new page, containing the Big Five Traits questionnaire.
The original Big Five Traits questionnaire was designed by Goldberg (1992). However, we wanted to keep the questionnaire as short as possible due to the voluntary nature of the study. Therefore, we used the shortened version by Srivastava et al. (2015) which is based on the questionnaires by Saucier (1994), Gosling, Rentfrow and Swann (2003) and Lang, John, Lüdtke, Schupp and Wagner (2011).
20 The scale for extraversion (α = 0.745) consisted of the following three items: extraverted, enthusiastic, talkative. Moody, easily upset and anxious are the items that described the neuroticism trait (α = 0.593) . Conscientiousness (α = 0.791) was measured by the items dependable, self-disciplined, and organized. Srivastava et al. (2015) have deleted the item ‘dependable’ after a factor analysis, however, since this item is mentioned in other researches as well (Saucier, 1994; Gosling et al., 2003), we will still ask for this item in the experiment. A reliability analysis before starting the data analysis will decide whether this item was a good measure or not. Agreeableness (α = 0.796) can be described by the items sympathetic, warm, and kind. Creative, imaginative, and unconventional described the trait of openness (α = 0.732). Appendix B shows the item wording and reliability statistics of these scales.
After the Big Five Traits questionnaire, some additional information about digital signage was given to the participants.
“Digital signage is one of the newest in-store marketing tools. We can describe digital signage as the provision of content (advertisements, news, assistance) on electronic scoreboards or large displays in stores. On the next page you will see an advertisement which will be presented on a digital signage in your hometown's supermarket.”
When the participant wanted to continue the study, the digital signage content was shown. The content of the digital signage is randomized and could be affective, cognitive or mixed content. 44 participants have been exposed to affective content, 46 participants to cognitive content and 46 participants have seen mixed content. We have transformed this data into a seven-point Likert scale for analysis. The higher this score, the more affective the content is that is shown to the participant (1 = cognitive, 4 = mixed, 7 = affective). The different content types can be found in Appendix A.
After the participant has seen the digital signage, they will be asked about their current emotional state. Following Garaus et al. (2017), we have used the eight-item scale developed by Babin and Darden (1996) to measure positive and negative emotions. The scale consisted of four positive emotions (happy, pleased, satisfied, content; α = 0.847) and four negative emotions (unhappy, despair, unsatisfied, annoyed; α = 0.815). These emotions were presented in a matrix with a six point Likert scale ranging from I do not feel at all to I feel very much. The scale for positive emotions was calculated by taking the average of the four items. The higher this average score is, the more positive the participant’s emotions were. The calculation for the negative emotions scale was the same, by taking the average of the four items. The higher this average score is, the more negative the participant’s emotions are. The item wording and reliability statistics of these scales can be found in Appendix B.
21
To what extent is it likely that you say positive things about the product to others?
To what extent is it likely that you encourage friends and relatives to buy the product?
To what extent is it likely that you recommend the product to others?
The questions were scaled with respect to the Likert scale of seven points. The scale was calculated by taking the average of the three items. The higher this average score is, the higher the consumers’ word-of-mouth intentions. Appendix B shows the item wording of the word-of-mouth scale and reliability statistics.
When the participant proceeds with the study he will be asked about his willingness-to-pay for the product. Le Gall-Ely (2009) describes in her paper different ways to measure willingness-to-pay including their disadvantages and benefits. She divides the measures in three different categories: methods based on sales data, methods based on survey data, and purchase offers. This study does not have access to sales data so we will perform an experiment through a survey. According to Breidert, Hahsler and Reutterer (2015) there is a difference between direct and indirect surveys. They argue that with direct surveys, such as expert judgment and customer surveys, participants are asked to state how much they would be willing to pay for some product, while in indirect surveys, such as a conjoint analysis and discrete choice analysis, a rating or ranking procedure is applied in order to estimate a preference structure to derive the willingness-to-pay. According to Le Gall-Ely (2009) there are three measures of willingness-to-pay based on survey data: a conjoint analysis, contingent valuation and psychological prices, and simulated purchase tests. The advantage of all these methods is that they are simple, direct measurements of willingness-to-pay usable at the point-of-sale for all types of products. However, a conjoint analysis can suffer from a hypothetical and informational bias. Therefore, there might be a difference between the participant’s willingness-to-pay and the willingness-to-pay in a real-time situation. Contingent valuation and psychological prices might suffer from the same biases (hypothetical, informational) but might also suffer from a strategic bias, since participants tend to over- or underestimate their willingness-to-pay. Because of monetary and time constraints we have decided to do a direct survey. We are aware of the limitations of this choice, and would recommend to improve these for future research. Participants were asked the price that they would be willing to pay for the product that the digital signage has showed them, in euros per kilogram. The higher this average score, the higher the average willingness-to-pay is.
22
d. Plan of data analysis
To analyse the results of our experiment we have used SPSS software. We started with cleaning the gathered dataset before analysing. The data was checked for missing variables and outliers, to remove these from the sample. After the data cleaning, we have transformed our digital signage content variables into one variable, a seven-point Likert scale (1= cognitive, 4 = mixed, 7= affective). Subsequently, we have computed the scales for the Big Five Traits, emotions and word-of-mouth.
After we had cleaned the dataset, regression analysis techniques were used to examine the direct effects between the variables of digital signage content, willingness-to-pay, and word-of-mouth, but also for the indirect effects of positive and negative emotions and consumer characteristics.
The direct relationship between digital signage content and willingness-to-pay and word-of-mouth (hypotheses 1 and 2) will be tested using a linear regression and a one-way ANOVA. We will perform two linear regressions, one for both dependent variable. Furthermore, we will check these results and test hypothesis 2 with a one-way ANOVA. In this way it is also possible to compare the means of the different content types.
To test for the mediating effects of emotions and moderating effects of consumer characteristics (hypothesis 3 to 13), we will use the Hayes (2017) PROCESS macro for SPSS. PROCESS macro is a statistical tool for the analysis of mediation and moderation. According to Hayes (2013) this tool can be used to “describe and understand the conditional nature of
the mechanism by which a variable transmits its effect on another”. Since our conceptual
model has a mediation and moderating effect included, we will use this macro to simplify the analysis. The Hayes (2017) PROCESS macro has 92 models, which all represent a different conceptual model.
For our analysis of the mediating effect of emotions (hypotheses 3 to 5)we will use the Hayes (2017) PROCESS model 4. Model 4 describes the direct effect of X on Y, even as the indirect effect of X on Y through M. In our case this would mean that we analyse the direct effect of digital signage content on willingness-to-pay and word-of-mouth. The indirect effect we will describe is digital signage content on willingness-to-pay and word-of-mouth through emotions. The conceptual diagram of model 4 can be found in Figure 2.
23 describes the conditional direct effect of X on Y. In our conceptual model this would mean that we will study the conditional direct effect of consumer characteristics on digital signage content to willingness-to-pay and word-of-mouth. Figure 2 shows the conceptual diagram of model 1 and 5.
Figure 2: Hayes PROCESS macro model 1, 4 and 5.
24 IV. Results
The Results chapter will discuss the results of the analysis of the conducted experiment. We will start with presenting the sample characteristics. Next, the convergent validity of the measurement scales will be discussed. Finally, the results of the regression analysis will be discussed.
a. Descriptives
The sample consists of 136 participants, of which 30 male and 106 female. Since there were no restrictions on a minimum or maximum age, the ages differentiate from 13 to 82 years (M = 34.04, SD = 15.282). The most participants did some type of higher education, the average education of all participants is HBO bachelor (M = 5.63, SD = 1.874). Table 1 summarizes the sample characteristics.
Table 1: Descriptive statistics.
Sample characteristics N = 136 Age 13-22 28 20.6% 23-32 56 41.2% 33-42 4 2.9% 43-52 25 18.4% 53-62 18 13.3% 63-72 4 2.9% 73-82 1 0.7% Gender Male 30 22.1% Female 106 77.9%
Education No schooling completed 2 1.5%
High school, no diploma 1 0.7%
High school 19 14%
MBO 23 16.9%
Some college, no degree 16 11.8%
HBO Bachelor 24 17.6%
University Bachelor 22 16.2%
University Master 27 19.9%
Other advanced degree beyond Master’s degree 2 1.5%
b. Convergent validity
25 (2015) an acceptable level of reliability is an Cronbach’s alpha from 0.6 to 0.7. The Cronbach’s alpha of the neuroticism scale (α = 0.593) is close to this threshold, so we have not adjusted this scale.
The scale of conscientiousness was changed by Srivastava et al. (2015) after performing a factor analysis. However, since other literature (Gosling et al., 2003) has also mentioned this item as driver of conscientiousness, we have still used this item in our experiment. Performing an reliability analysis tells us that our scale has a Cronbach’s alpha of 0.382, which means that the scale is not reliable. We could increase Cronbach’s alpha by leaving the item ‘dependable’ out of the scale to 0.791. Saucier (1994) has used the opposite of the dependable item: undependable. We have transformed our item to an undependable item to check if this had an influence on the Cronbach’s alpha. The results showed us that the conscientiousness scale with the item ‘undependable’ has a Cronbach’s alpha of 0.527, and when deleting the item ‘undependable’ Cronbach’s alpha is 0.791. Since both Cronbach’s alpha’s were too low to create a reliable scale and the Cronbach’s alpha when the item is deleted meets the required 0.7, we have deleted the item dependable from our scale. This means that the scale of conscientiousness consists of the items self-disciplined and organized. The Cronbach’s alpha of this scale is 0.791, which is higher than the criterion of 0.7, and therefore this scale is reliable.
For the scales of positive and negative emotions, we also did a reliability test to see if the scales are reliable. The Cronbach’s alpha of positive emotions was 0.847, while that of negative emotions was 0.815. The results show that both scales have a Cronbach’s alpha higher than 0.7. Therefore, the scales of positive and negative emotions are reliable scales. We have also checked if the word-of-mouth scale we have used is a reliable scale. We found that the Cronbach’s alpha of the word-of-mouth scale is higher than the required 0.7. Word-of-mouth has a Cronbach’s alpha of 0.94. Therefore, this scale is reliable.
The results of these reliability tests can also be found in Appendix B.
c. Analysis
We will first dive into the relationship between digital signage content and consumer behavior (hypotheses 1 and 2). Following this, we will add the mediating effect of emotions (hypotheses 3, 4 and 5). Finally, we will take consumer characteristics in account as moderating influence in the conceptual model (hypotheses 6 to 13).
26 have conducted a one-way ANOVA on the relationship between digital signage content and willingness-to-pay. These results can be found in Table 3. A one-way ANOVA (digital signage content: affective vs. cognitive vs. mixed) on willingness-to-pay also demonstrated that no main effect of digital signage content on willingness-to-pay was found (F(2,133) = 0.427, p = 0.653). Furthermore, the results indicated that the willingness-to-pay of affective digital signage content is not significantly different than the willingness-to-pay of cognitive or mixed digital signage content. This result indicates that hypothesis 2a and 2c are not supported.
The second regression analysis performed had word-of-mouth as criterion variable and digital signage content as predictor. These results can also be found in Table 2. A linear regression analysis (digital signage content on willingness-to-pay) demonstrated that the results are not in line with the hypothesis. That is, the results did not show a main effect of digital signage content on word-of-mouth. Therefore, hypothesis 1b is not supported. At the same time, we have conducted a one-way ANOVA on the relationship between digital signage content and word-of-mouth. These results can be found in Table 3. A one-way ANOVA (digital signage content: affective vs. cognitive vs. mixed) on willingness-to-pay gave the same result as the linear regression; that no main effect of digital signage content on word-of-mouth was found (F(2,133) = 0.826, p = 0.440). The results also indicated that the word-of-mouth intentions of affective digital signage content are not significantly different than the word-of-mouth intentions of cognitive or mixed digital signage content. Therefore, hypothesis 2b and 2d are not supported by our results.
Table 2: Regression results of digital signage on consumer behavior.
β se p-value R2 F p-value
Willingness-to-pay 2.876 3.190 0.369 0.006 0.813 0.369
Word-of-mouth -0.069 0.054 0.210 0.012 1.586 0.210
Note: * p < 0.01, **p < 0.05, p < 0.10
Table 3: ANOVA mean desciptives.
Willingness-to-pay Word-of-mouth Mean SD Mean SD Affective 61.441 97.757 3.114 1.617 Mixed 49.260 82.097 3.397 1.530 Cognitive 44.156 91.772 3.526 1.495 Total 51.512 90.208 3.348 1.545
27 After running a Hayes (2017) PROCESS macro, we first looked at the relationship between digital signage content and emotions. Results indicated that digital signage content was not a significant predictor of positive emotions (β = -0.0011, se = 0.0298, p = 0.9706) and negative emotions (β= 0.0196, se = 0.0269, p = 0.4674). Therefore, these results do not support hypothesis 3a and 3b.
The results of the Hayes Model 4 could be found in Table 4. The results indicated that the model of willingness-to-pay is not statistically significant. Approximately 1.7% of the variance in willingness-to-pay was predicted by the variables we have included in the model. Results further indicated that positive emotions were not a significant predictor of to-pay, just as negative emotions were not a significant predictor of willingness-to-pay. Since these relationships are non-significant, hypothesis 4a and 4b are not supported. The second Hayes (2017) PROCESS macro we have run, focused on the dependent variable word-of-mouth. The predictors of this model accounted for approximately 7.4% of the variance in word-of-mouth. Furthermore, the results indicated that the model is statistically significant. Results showed that positive emotions are a significant positive predictor of word-of-mouth intentions. An interesting finding is that negative emotions turned out to be a significant positive predictor of word-of-mouth intentions, while we expected a negative relationship. Therefore, hypothesis 5a is supported by the results, while hypothesis 5b is rejected.
Table 4: Mediating effect of emotions on the relationship between digital signage content and willingness-to-pay and word-of-mouth.
β se p-value R2 F p-value
Willingness-to-pay 0.0172 0.7689 0.5134 Digital signage content 2.8943 3.2031 0.3679
Positive emotions 11.2008 10.2583 0.2769 Negative emotions -0.2865 11.3719 0.9799
Word-of-mouth 0.0742 3.5287 0.0168** Digital signage content -0.0780 0.0533 0.1454
Positive emotions 0.4075 0.1706 0.0183** Negative emotions 0.5008 0.1891 0.0091***
Note: * p < 0.01, **p < 0.05, p < 0.10
Having determined the relationship of digital signage content and consumer behavior with the mediating effect of emotions, we only need to add the moderating effect of consumer characteristics to this analysis. Since the mediating influence on willingness-to-pay is non-significant, we will use Hayes (2017) PROCESS model 1. This model describes the conditional direct effect of consumer characteristics on digital signage to willingness-to-pay. We will use model 1 with a bootstrap sample of 5000. Since we have five moderating variables, we have run the PROCESS macro five times.
28 significant. This implies that none of the variables is influencing willingness-to-pay. Therefore, hypotheses 9a, 10a, 11a, 12a and 13a are not supported by our dataset.
A Hayes (2017) PROCESS model 5 on willingness-to-pay has been run as well, the results of this regression analysis can be found in Appendix E. The results indicated that there were no differences between the results of model 1 and model 5.
Table 5: Moderating effects of Big Five Traits on the relation between digital signage content and willingness-to-pay.
β se p-value R2 F p-value
BFT: Extraversion 0.0074 0.3279 0.8052 Digital Signage Content -3.2952 14.9742 0.8262
Extraversion -5.3283 13.8815 0.7017 Digital Signage Content
* Extraversion
1.3184 3.1180 0.6731
BFT: Neuroticism 0.0087 0.3866 0.7628 Digital Signage Content -2.5155 9.6164 0.7941
Neuroticism -8.0517 15.1284 0.5955 Digital Signage Content
* Neuroticism
1.8460 3.0968 0.5521
BFT: Conscientiousness 0.0156 0.6987 0.5545 Digital Signage Content 15.3540 11.7440 0.1934
Conscientiousness 12.9132 11.9448 0.2816 Digital Signage Content
* Conscientiousness
-2.7145 2.4456 0.2690
BFT: Agreeableness 0.0179 0.8023 0.4947 Digital Signage Content 20.9224 21.0964 0.3231
Agreeableness 4.4642 17.7833 0.8022 Digital Signage Content
* Agreeableness
-3.3110 3.8522 0.3916
BFT: Openness 0.0263 1.1896 0.3163 Digital Signage Content 11.4594 13.7110 0.4048
Openness -3.8433 12.1659 0.7526
Digital Signage Content * Openness
-1.8124 2.7924 0.5174
Note: * p < 0.01, **p < 0.05, p < 0.10
