The effect of tailored notification messages on the exercise
intention of four DISC-types and the moderating role of
subliminal priming
Master’s thesis
Msc. in Business Administration – Marketing Track
Amsterdam Business School
Student: Tiffany Valk
Student ID: 11369108
Supervisor: Tina Dudenhöffer
2
Statement of originality
This document is written by Tiffany Valk who declares to take full
responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and
that no sources other than those mentioned in the text and its references have
been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision
of completion of the work, not for the contents.
3
Abstract
The concept of consciously presenting a message in a certain way (framing) or adapting a message to the needs and wants or the personality of individuals (message tailoring) have been investigated for years now. This, however, has not yet been applied to something more recent, like mobile exercise applications. Additionally, the usage of the DISC-model to assess one’s personality to tailor messages has not been researched either.
Therefore, the present study investigates the effect of tailored notification messages on the exercise intention of an mobile application user. Moreover, the potentially moderating effect of a subliminal prime has also been explored.
To research these effects, an online survey has been conducted. 210 participants filled in the questionnaire. A part of the survey was a video, which contained a subliminal prime and a notification message. Participants were randomly assigned to one of the four conditions (2x2 design).
The results showed that there were no significant effects for the overall attitude towards the notification message. The same was applicable to the effects of the subliminal prime.
Therefore, in the current research, no indication is found that tailoring messages to someone’s personality contributes to a more positive attitude towards this message, or to an increase in exercise intention. A subliminal prime does not reinforce this effect either.
4
Acknowledgements
Now this thesis has been completed, I want to thank several people for the support that helped me to come to this point.
First of all, thank you Ms Dudenhöffer, for all the help you offered me through the whole process. It was a tough time, but you eventually got me through it. Thank you for all the feedback, time and effort that helped me to finish this thesis successfully.
Also, I want to thank my family and friends for their support, encouragement and conversations that made me stay on track. I am lucky to have them around me.
5
Table of Contents
Statement of originality ………...………. 2 Abstract ………....…………..… 3 Acknowledgements ……….4 Introduction ………..…. 6 1. Literature Review ……….………..…… 91.1 Framing ………..….… 9 1.2 Message Tailoring ………..……... 10 1.3 Priming ………..… 12 1.3.1 Goal-priming ………..………….. 13 1.4 Exercise apps ………. 14 1.5 DISC-model ……….………….. 16 1.6 Conceptual Framework ……….………..18 1.6.1 Gamification ……….………... .19
1.6.2 The ‘’Influence’’ type ……….………….. 20
1.6.3 The ‘’Steadiness’’ type ………...….…….… 22
1.6.4 The ‘’Conscientiousness’’ type ……….….……….….. 23
1.6.5 Subliminal priming and Message Tailoring ……….……….….... 24
2. Data and Method ………..……….……. 25
2.1 Respondents ………..………….…….25 2.2 Research Design ……….….…...….……... 25 2.3 Pre-test ………...……….……….…….. .25 2.4 Procedure ……….………….…...……26 2.5 Measures ……….……….……...… 27 3. Results ……….……….……….... 30 4. Discussion ……….……….…...……... 36
4.1 Limitations and Future Research ……….……….………….. 38
5. Conclusion ……….…….…... 40
References ……….….…………... 41
Appendices ……….….……….... 47
6
Introduction
Physical activity is important for the human body. When lacking exercise, there is a higher risk of chronic health conditions, such as cancer, depression or metabolic syndrome. Still, 30% of people throughout the world do not exercise (Litman et al., 2015). Over the last few years, there has been a rise in health related and physical activity apps for activities such as fitness or exercise. Already in 2013, there were 40,000 health-related apps available in the Apple iTunes store (Krebs & Duncan, 2015). It turns out that these apps do help to change behaviour: in research by Glynn et al. (2014), it was found that, after using an application for 8 weeks that promotes physical activity, the activity was increased.
However, 31% of the adults worldwide do not work out enough (Belmon et al., 2015).
Therefore, there is room for improvement. Exercise apps already motivate a certain amount of people, but to reach more, another approach may be needed. Middelweerd et al. (2014) mentions that using a health behaviour change theory, for example by tailoring feedback (feedback based on the characteristics of the user), is more likely to be effective than general feedback. This seems interesting, but there has been limited research on this. A notable element of an app are notifications. These are small messages that inform the user about a variety of events (Sahami Shirazi et al., 2014). Because these messages are short, users are easily reached and it does not take a lot of effort to absorb the information. This is why they are ideal to use as a means to motivate people to use the application. It was previously mentioned that tailored feedback is effective, but what if notification messages could also be tailored? Message tailoring already exists as a technique to capture people’s attention. It originates from message framing. By framing a message, certain events or features are highlighted and accented as more salient than other elements. Frames could also be used as a strategic tool, to reach a certain goal (Benford & Snow, 2000). Message tailoring goes even further and narrows the content of a message to the individual, instead of simply generating a general message for all users. This will boost the relevance of the content for the recipient and eventually leads to a desired reaction (Hawkins et al., 2008). The content can be personalized by incorporating the individual’s characteristics, like personal wants, interests or specific personality traits (Rimer & Kreuter, 2006). Hirsh, Kang & Bodenhausen (2012) focused their research on tailoring a persuasive message to the personality profile of the receiver. They used the Five Factor Model (FFM) to assess their personalities. This model has five dimensions (Extraversion, Agreeableness, Conscientiousness, Neuroticism and Openness) and each dimension represents a personality trait. An individual receives a certain score on all five
7
traits after filling in a questionnaire. This gives an overview of which personality traits are applicable and thus a description of one’s personality. The FFM is often used by researchers in the psychology field (Mccrae & John, 1992). In the present study, however, another personality model will be used: the DISC-model. The DISC-model was founded by W.M. Marston and distinguishes four dimensions of personality: Dominance, Influence, Steadiness and Compliance. After filling in a questionnaire, individuals are assigned to one of the four personality types. Each type is different. The D-type is results-oriented, straight-forward and likes to take initiative. The I-type is outgoing, sociable and tries to delegate and persuade others. The S-type likes cooperating with others, security and work routines. Finally, the C-type thinks ahead, is accurate and prefers to have less contact with others (Slowikowski, 2005). The DISC-model is often used in business by the management to assess what type their employees are. Then, they can take their personality into account in working teams.
Furthermore, it will make the behaviour of workers more predictable. In this research, using the DISC-model might benefit managers, because the results can eventually be used to figure out what engages workers and how to stimulate them to be healthy.
Goal-priming might reinforce the effect of message tailoring, using the DISC-model. When using a goal-prime, people are influenced in order to let them behave in a desired way (Förster, Liberman & Friedman, 2007). Therefore, with message tailoring the goal is to capture attention and this could be further amplified by using a prime to impel to act as desired. A way to prime people is to use a subliminal prime, which is when an unconsciously perceived stimuli influences the individual. This could be executed by showing a picture or a word very briefly and masking it by letting the person focus on something else (like a video or a fixation point) (Bargh & Chartrand, 2000). However, it is important to keep in mind that the prime will only work if it is important to the individual, thus giving enough motivation to behave in the way the researcher wants (Strahan, Spencer & Zanna, 2002). In this case, it means that priming respondents will mainly affect people who already exercise or have already installed a fitness application. Some people, however, want to exercise, but they lack motivation. They develop a feeling of guilt, which eventually leads them to exercise. These types of people have an introjected regulation profile, meaning that they exercise because they want to avoid feelings of guilt (Lindwall et al., 2017). However, this driving force does not create a sustainable situation, which could eventually prove detrimental to one’s well-being and exercise routine (Hurst et al., 2017). The feeling that follows is shame. Shame is an affective response that may come from failing to carry out social or personal expectations. This eventually leads to a desire to avoid exercise, which in turn relates to less frequent
8
energetic and moderate workouts, or even not working out at all (Kang et al., 2017). Building on existing research, this study will integrate message tailoring with notification messages of fitness apps to investigate if this will trigger positive behaviour. Combining the DISC- model with the question of tailoring notifications of exercise apps, the main question is formulated as:
How does tailoring notification messages increase exercise intention of users of a mobile fitness application?
Previous research has not yet looked into notification messages as a tool to influence people. However, with the upcoming usage of mobile phones, people are every day exposed to mobile applications and messages. It has an increasing role in daily life. Therefore, notification messages might be the perfect tool for marketers to use more in the future.
Further, researchers have already studied the effect of tailoring the message to recipient’s needs or personality, but they have not used the DISC-model before. This might be interesting to research, since the DISC-model is closer to practice than the FFM. It is used by managers in organizations.
The thesis is set up as follows: first, an overview of the existing literature about framing, message tailoring, priming, mobile apps and notifications, the DISC-model and the
hypotheses are presented. Second, the methodology and research design will be discussed. Third, the results will be presented, followed by a discussion where the results will be linked to the theory. Lastly, the strengths and limitations and the overall conclusion shall be
9
1. Literature Review
1.1 Framing
Framing is selecting some elements of a perceived reality and making them more important in a message that is communicated. It is used by organizations, such as social movements, by selecting certain pieces of information that are in line with their beliefs and highlighting this information to get bystander support, to inspire and to justify their campaigns and activities. These frames are called Collective Action frames. It can also be used as a strategic instrument, using frames to achieve a specific purpose (Benford & Snow, 2000). With framing, some items are selected and highlighted to create an argumentation (Entman, 1993). The salience of specific parts of information is enhanced, making them more meaningful, noticeable or memorable to the receiver of the message. Examples of methods used to enhance are
repetition, placement or associating them with culturally familiar symbols (Entman, 1993). A famous example of framing comes from Tversky & Kahneman (1981). They proposed a framed problem and asked 152 university students to choose one out of two possible solutions. The first time, the problem was formulated like a disease which would kill 600 people. There were four solutions (A,B,C,D), the only difference was the formulation: either the number of lives saved or number of lives lost was emphasized. This experiment showed how framing can dictate which elements people will notice. Certain aspects of the ‘reality’ were specifically described and thus directed attention away from other aspects which were not described.
However, individuals all have their own existing schemata (a mental cluster of ideas that influences the processing of information). They might focus on different aspects of a message as opposed to what is highlighted. Therefore, it is not guaranteed that the presence of frames always has an effect on every message receiver (Entman, 1993). In the context of framing in the media, especially the news, Scheufele (1999) distinguishes three dimensions of processing news. The first is active processing, meaning that an individual seeks additional sources because they assume that information from mass-media is biased or incomplete. The second dimension is reflective integrating, which indicates people who think about and reflect on the information of mass-media or discuss it with others in order to fully understand the
information they have read. The last dimension holds selective scanning, which includes individuals who only seek information that they themselves find interesting. They skim over the message and ignore irrelevant content. This shows that different kinds of people react differently to messages.
10
There has been research done on specifically framing the promotion of Physical Activity (PA). It was found that gain-framed messages (messages that describe the gains someone gets from PA) are more effective when promoting low-risk behaviours (in this case maintaining a healthy lifestyle), whereas loss-framed messages (description of costs or the missed benefits of not taking action) have more of an effect for risky behaviours (for example illness). In this case, gain-frames messages would fit by promoting PA (Li, Cheng & Fung, 2014). In the article by Jones et al. (2003) the researchers did a similar experiment. The respondents received either a message containing strong arguments advocating regular physical exercise and promoting the gains, or a message emphasizing the risks of neglecting regular physical exercise. The results showed that the positive frame condition led to positive exercise intentions, whilst increased engagement in exercise and a negative frame condition led to reactance (Jones et al., 2003). These studies indicate that framing the positive sides of exercising in a message will have a positive effect on the exercise intention or even actual exercise behaviour of individuals. This is relevant information for this study because it demonstrates that using positive content can be effective to promote exercising. The effect of a positive response might be bigger if the message could be more personalized, so that people individually feel addressed. This might be possible with message tailoring. Where framing does not guarantee to cause an effect because of the existing schemata that differ per person, message tailoring can improve the effectiveness by adapting to this schemata. In these cases, there would be controlled for the dimensions of the processing the news (active processing,
reflective integrating and selective scanning).
1.2 Message tailoring
Related to framing is message tailoring. This approach focuses on the typical features of people. These features could be specific needs or personality traits (Rimer & Kreuter, 2006). It assumes that messages are not effective for all individuals and therefore should be tailored to the unique characteristics of the message recipient. Message tailoring can be defined as ways to enhance the relevance of the information presented and evoke positive changes in response to the communicated message. (Hawkins et al., 2008). This is more effective than a generic message with undifferentiated information. Especially in health communication, personal aspects can affect peoples’ decisions to engage in sustainable, healthy behaviour (Rimer & Kreuter, 2006; Hirsh, Kang & Bodenhausen, 2012; Pope, Pelletier & Guertin, 2017). There are two features that separates tailoring from other used message methods: 1. it is assessment-based (looking at what values are important to the recipient) 2. messages are
11
individually-focused (Kreuter, Strecher & Glassman, 1999). It only contains information that is important to the recipient and therefore there is a higher probability that people pay more attention to the message. This importance could be derived from a lot of characteristics, like demographic information, such as food intake, age, race, stage of readiness to change diet (Campbell et al., 1994), or personality of the individual (Hirsh, Kang & Bodenhausen, 2012). Because it is adapted to the specific needs of the person, it thus has the effect of contributing to staying motivated, acquiring new skills and sustaining desired change in life-style (Kreuter, Strecher & Glassman, 1999).
The effect of message tailoring can be explained by the Elaboration Likelihood Model (ELM). This model, proposed by Petty & Cacioppo (1981), states that there are two routes to attitude change: the central route and the peripheral route. Individuals taking the central
route carefully consider information in messages to comprehend, to learn and to eventually
evaluate. They look at the pros and cons of the issue. Attitude changes through this route tend to be enduring and predictive. The peripheral route, on the other hand, elicits attitude change because the information is related to cues and heuristics. These cues and heuristics (like an expert source or the amount of arguments that are given) are more superficial but are
unrelated to the core arguments in the message. Overall, the results of processing through the
central route are more likely to relate to future behaviour and be more permanent compared to
the peripheral route (Noar, Harrington & Aldrich, 2009). Which route someone takes depends on the level of involvement. The level is high when the message is personally relevant or has personal consequences and therefore becomes more important for the recipient. He/she is motivated to make a cognitive effort to evaluate the information and will follow the central
route. In the study by Petty, Cacioppo & Schumann (1983), participants got to see an
advertisement of razor blades. A high level of involvement was operationalized by informing that the razor was available in their own city. The participants who were classified in the group with low level of involvement, were informed that the razor was only available in another area. With this difference in treatment, there was a consequent difference in how concrete, and therefore how personally relevant, the product was for them. Since for one group, the product was not available, the importance would decrease (Petty, Cacioppo & Schumann, 1983).
An example in communicating health messages comes from Kreuter & Wray (2003). They first let 201 respondents who were overweight complete a behavioural assessment. After, they were randomly assigned to three conditions: 1) receive weight loss materials that were
12
Association (AHA) brochure 3) receive an AHA brochure that only looked like it was tailored. It was found that the respondents who received the tailored materials reacted more positively and also had a more positive self-assessment compared to the respondents who received the AHA brochure. The authors concluded that tailored messages stimulated greater central route processing, because the information provided was important to them (Kreuter & Wray, 2003).
Thus, according to the ELM, there are two routes (the central and peripheral) When applying message tailoring, there is a higher probability that the central route will be taken, because the level of involvement will increase. This is because the message contains important
information and attracts the attention of the recipient. Eventually, the convincing message might lead to an attitude change (Noar, Harrington & Aldrich, 2009). An attitude change is a desired goal of a notification message in a fitness application.
1.3 Priming
Priming is an effect whereby the individual is not aware of the influence a stimulus has. It unconsciously activates information that a person already knows (Bargh, 2016). When one is exposed to such a marketing tactic, it can activate mental constructs that are associated with it, leading consumers to behave in a manner which is implied by the tactic (Laran, Dalton & Andrade, 2010). Priming techniques can be a potentially powerful tool for marketers (Wheeler & Berger, 2007). There are generally two ways to prime: supraliminally and subliminally. Depending on the kind of priming, an individual can be conscious or
unconscious of the stimulus itself. When conscious of the stimulus, it is called supraliminal
priming. This can, for example, come in the form of specific words in text (a small reading
section), auditory signals (an audio fragment) or visual objects (pictures or a film clip) that are shown to individuals (Bhagwatwar, Massey & Dennis, 2013). On the other side of the
‘’stimulus spectrum’’, there is subliminal priming. In this case, the individual is unaware of the stimulus that they are exposed to, because often words or information are flashed on the screen, so that it is not easily detected. The concept of subliminal priming became well-known in 1957, when a market researcher named James Vicary claimed that he increased sales of popcorn and Coca Cola drinks in a cinema, by flashing the messages ‘’Eat popcorn’’ and ‘’Drink Coca Cola’’. In the end, it became apparent that the experiment was fake. Still, from then on, researchers did research the subliminal priming phenomenon. A famous, more recent example comes from Karremans, Stroebe & Claus (2006). They subliminally primed thirsty as well as non-thirsty people, either to ‘’Lipton ice’’ (an ice tea brand) or to a neutral
13
prime. After priming, the choice for Lipton Ice Tea increased, but only for the respondents that were thirsty. Thus, the subliminal prime turned out to be effective. Subliminal priming can be used to prime a specific goal, in combination with a motive, and can reach that goal by persuading that target effectively (Strahan, Spencer & Zanna, 2002). When using this
technique, an unconscious activation of meaning is caused. However, the effect of the prime is short-lived and therefore the target message, word or action must appear immediately after (Greenwald, Draine & Abrams, 1996).
1.3.1 Goal-priming
Goal-priming focuses on activating the end-goal state by leading an individual to behave in ways that are consistent with the eventual goal (Minton et al., 2016). This can be
accomplished by taking action which is consistent with the end goal, by priming a stereotype, trait or a behaviour. Eventually, it will increase the likelihood that the primed person will behave in the desired way (Förster, Liberman & Friedman, 2007). To reach this, the
individual has to have motivation to pursue the primed goal. As written above, the study by Karremans, Stroebe & Claus (2006) shows that the prime of Lipton Ice Tea only worked for people who were thirsty. In other words, the prime has to be applicable or relevant to the current motivation of the person, also called goal-relevancy. The more important the prime is to the person, the stronger the priming effect will be (Strahan, Spencer & Zanna, 2002; Bargh, 2016). However, there are people who want to work-out, but cannot get themselves to do it. They do not have the motivation to exercise, but eventually do it to avoid a feeling of guilt or to enhance their self-worth. This mechanism is called introjected regulation (Lindwall et al., 2017). This kind of ‘’motivation’’ may work for a while, but is actually not sustainable and can eventually harm the well-being of an individual (Hurst et al., 2017). According to Kang et al. (2017), feelings can impact how someone responds to persuasive health messages. When a message is personally relevant, but the person has a negative view on their self-competence and do not have the motivation, it can eventually create counterproductive effects, thus not exercising at all. A negative affective response that may arise from the previously mentioned effect is shame: the feeling that personal and social expectations are not met. This feeling leads to avoidance and thus not exercising at all (Kang et al., 2017). Therefore, in the current study, it was assessed that people did exercise at least a couple hours per week, so they do not feel threatened by the messages.
It may also be that goal primes do not have an influence on behaviour. For example, thirst primes only affect thirsty people, not on hydrated people. The internal physiological state has
14
to be in line with the prime to influence someone (Laran, Janiszewski & Cunha Jr., 2008). Further, the primed goal has to raise a positive affect, because this signals that the goal is worth pursuing. This nudges the recipient of the prime into a state of readiness to pursue the goal (Aarts, Custers & Veltkamp, 2008).
The same prime may also have a different effect on diverse individuals. According to the article of Wheeler & Berger (2007), one prime can activate different associations among different individuals. This has to do with personal and situational characteristics. One group of people may have different associations with a prime than another group. This could lead to a segmentation in the eventual effect of priming. A given example is that there could be a difference between males and females and the associations they have with shopping. Men are expected to act more ‘’purposeful driven’’, while women might view shopping as an
experience itself. Primes could activate different subsequent behaviour. Linking this to message tailoring, by adjusting the notification message to the personal characteristics (one of the DISC types), there should be a higher probability that the respondent is more motivated to absorb the information and thus will be primed and behave according to the goal.
Additionally, by filtering the participants on people who exercise, the variation in effect of the prime is reduced as much as possible.
1.4 Exercise apps
In recent years, smartphones are becoming a significant part of daily life. People cannot live without a phone and have developed habits of checking e-mail, social media, or other applications (Oulasvirta et al., 2012). There has been a growth in health and medical related apps. The most popular kind of apps are exercise and fitness apps, which account for around 39% of the fitness-related apps (Litman et al., 2015). These apps try to promote health and improve physical activity. This has potential, because exercise apps intend to focus on specific facets of cognition, behaviour and affect. For example, it gives feedback on previous exercises, gives motivational memo’s, gives advice about the right way to exercise, or sets and monitors goals.
A mobile app has a number of advantages over other health promoting appliances; such as internet websites. One is the degree of customizability: the app can adapt to the wants and needs of the user. Also, a smartphone can provide (health) information efficiently and quickly (Dennison et al., 2013). Further, the app is installed on a smartphone and most people carry their phone with them – thus, it can provide service continuously and can also keep a record of all and any physical activity carried out (Litman et al., 2015).
15
Health apps have likewise created techniques to engage people in exercising and motivate them to use the app (more often). These techniques are called Behaviour change techniques. To analyze what elements are used to motivate users to work out, apps use behaviour change techniques to change behaviour. This could be in the form of feedback, self-monitoring or reinforcement (Belmon et al., 2015). In their research, Belmon et al. (2015) drew up a list of behaviour change techniques, used by mobile phone apps to promote physical activity. Via a questionnaire, they asked Dutch young adults to rate which techniques were most effective. Techniques perceived as most effective were ‘’goal setting on the outcome of behaviour’’, ‘’self-monitoring of behaviour’’ and ‘’self-monitoring of the outcome of behaviour’’. Overall, in another study by Conroy, Yang & Maher (2014), it was found that top-ranked mobile apps for physical activity emphasize a limited number of behaviour change techniques and
therefore all tend to use the same limited number of techniques, namely ‘’providing instruction on how to perform behaviour’’, ‘’providing feedback on behaviour’’ and also goals setting, as corresponding to the results of Belmon et al. (2015) (Conroy, Yang & Maher, 2014). Edwards et al. (2016) also found several behaviour change techniques in smartphone apps, but more specifically focused on gamification (see below for more about gamification). The most frequently used techniques were ‘’feedback and monitoring’’, ‘’comparison of behaviour’’, ‘’reward and threat’’ and ‘’self-belief’’. Thus, goal setting is less seen in gamification apps, and there is more focus on monitoring and comparison. By using these techniques, the self-efficacy of app users will increase and, as a consequence, the motivation to exercise too. Eventually, their behaviour will change in a positive way (Belmon et al., 2015).
Dharia et al. (2016) did not look specifically at behaviour change techniques, but built a specific app named ‘’pro-fit’’, which is a personalized fitness application that utilises several ways to engage the user. Thus, the authors already tried to fit in several techniques to change the behaviour of the user in a positive way; personalizing it too. The app has four main components: an activity tracking and classification module, a calendar integration manager, an activity recommendation engine, and a fitness buddies recommendation engine.
In the present study, the behaviour change techniques will be applied, but will be specifically aiming at short messages. In mobile applications these messages are called notifications and can contribute to motivating the user to keep using the app and eventually trigger a desired behaviour.
To inform the users of an exercise application, they can be notified by a pop-up message, called a notification. A notification informs the user of a smartphone about several events,
16
such as a new message, a comment on a social media platform, or an application update. These messages are often short, so the user can process the information quickly (Sahami Shirazi et al., 2014). This makes it ideal for implementing behaviour change techniques into the message (for example a message about feedback or goal-setting), because there is a bigger probability that the user will absorb this small bit of information and will be triggered by it. Depending on the importance of the memo, the user immediately takes action or ignores it. Elements that lead to low importance are: too many notifications, information about system updates and actions initiated by the user. An important notification contains the context (e.g. when it is an app for work, it contains a message that makes it clear immediately) or are about the contacts of the user. In the case of a fitness app, it informs the user about their exercise results or recommendations. Therefore, the trick is to find a balance between disruption and notifying valuable information (Sahami Shirazi et al., 2014).
According to Warren & Meads (2013), push notifications should be used only when users must be informed in a timely manner. For example, a notification of an airline informing the user of flight delays or cancellations. Further, it is recommended to minimize the amount of notifications by personalizing the messages that are narrowed down to the individual interests and needs. With notifications, an application can efficiently inform users. As proposed by Warren & Meads (2013), by personalizing the messages, the amount of notifications can be reduced because only the most important information will be notified. When a message is important to the receiver, there is a higher probability that he will read it and eventually act on it. Furthermore, by using preferred Behaviour Change Techniques, like goal-setting or self-monitoring, the notification message could be yet more effective still.
1.5 DISC Model
All users of applications are different. The DISC model distinguishes four types of people with different characteristics and priorities. These characteristics and priorities influence the behaviour they show in daily life. The model is often used by managers in organisations to assess the personality profile of employees. This knowledge can be used to effectively communicate with employees, using everyone’s strengths to enhance the group and work more as a team. The ‘’D” and ‘’I’’ types are seen as appropriate leaders during times of growth and change, whereas the ‘’S’’ and ‘’C’’ types are seen as the foundation and are appropriate leaders during maintenance and stability (Slowikowski, 2005). According to the creator of the DISC model, W.M. Marston, there are two kinds of environments, which depend on circumstances: a favourable environment and a hostile environment. In these
17
environments, there are two roles that the four types can take on: an active or a passive role. These roles can already make a division between two groups. The ‘’D’’ and ‘’I’’ types are the ones that take on the active role; they like to take control. This characteristic can be described as being extraverted. On the other hand, the ‘’S’’ and ‘’C’’ types take on a passive role and wait before taking action. They are more introverted (Turnasella, 2002; Puccio & Grivas, 2009; aDot, 2007, Geyer, 2007; Michaelson, 2010).
The ‘’D’’ stands for Dominance, which refers to being direct and forceful. People who lean towards this DISC type push to keep things moving and like to have control (Sugerman, 2009). They are strong-minded, like accepting challenges and get immediate results (Reid & Reid, 2003). With a challenge, they can be competitive (Turnasella, 2002). D-type persons prefer to take an active role in a hostile environment, because they like taking risks and it is a challenge for them to solve problems, especially when things are not going well. They may become bored when there is no change, and therefore like variety and adventure. Also, they are demanding of themselves and others. If an effort eventually comes out without a reward or unnoticed, it will decrease their motivation. To let this type of people perform, it is necessary to change goals and criteria once in a while to prevent boredom. The goals should be difficult, but not impossible. A challenge motivates them as long as they have control over it. Also, they like to get a reward once they reached something (Turnasella, 2002). Terms to describe the D-type include goal-oriented, likes delegating, forceful, seeks change and competitive (Grant, 2017).
The ‘’I’’ stands for Influence. These type of people are lively and outgoing and like to be around other people. They are talkative and like to work with others and persuading them. (Sugerman, 2009). Also known as ‘’people people’’ (Reid & Reid, 2003). The I-type likes, similarly to the D-type, to take an active role, but more in a favourable environment, where they have an influential personality. This fits them because they are optimists and try to look at the positive side of problems. They have a large desire for acceptance and this is shown by impressing others with their network of friends and professional connections. Ideally, an I-type person performs best when the criteria are based on social interaction, thus in groups or teams. However, they also can be competitive, but mostly if the achievement is publicly recognized (Turnasella, 2002). Terms to describe the I-type include charismatic, encourager, enjoyer of social encounters, cheerful and persuasive (Grant, 2017).
The ‘’S’’ stands for Steadiness. People who are this type prefer to let other people be the centre of attention. They want a stable and predictable environment where everybody gets along (Sugerman,2009). They are characterized by being helpful, patient and good listeners
18
(Reid & Reid, 2003). The S-type prefers to take a passive role in a favourable environment. Changes are not an issue for these individuals once they are prepared in advance, but they do not particularly like challenges. What they like is working at a steady pace to achieve their goal and dislike pressure and deadlines because these aspects can compromise quality. To let the S-type perform, performance criteria should be objective (when rules are clear),
measurable and should allow them to set goals in order to avoid change and pressure
(Turnasella, 2002). Terms that can be used for the S-type are agreeable, negotiator, mediator, security and dependable (Grant, 2017).
The ‘’C’’ stands for Conscientiousness. These kind of people focus on accuracy and are analytical. They prefer to focus on logic rather than on emotions (Sugerman, 2009), and prefer to plan ahead (Reid & Reid, 2003). The C-type tries to avoid problems at all costs and takes on a passive role in a hostile environment. These individuals tend to be insecure and therefore find confidence in set rules that cannot suddenly change. They want to know details and performance criteria should be objective and measurable. Goals should not be too
challenging (Turnasella, 2002). Terms to describe the C-type include reticent, respectful, systematic, precise and calculating (Grant, 2017).
Often, elements of each DISC type are combined in daily life, but one type will eventually stand out. Assessing which DISC profile a user of a health app is can help to tailor the notification message of the app to their personality. This may increase the probability that the user will process the information and eventually intent to work out. Therefore, the
research question is:
How does tailoring notification messages increase exercise intention of users of a mobile exercise application?
1.6 Conceptual Framework
To answer the research question, we have to look at how tailoring notification messages can be executed. Earlier, it was discussed that the DISC-model can be used to assign individuals to one of four personality types. The personality could, in turn, be used to tailor the message by taking into account the characteristics of each type. To research this, two hypotheses are established to examine what the tailored message should contain.
Furthermore, a hypothesis is set up to research the additional effect of a subliminal prime. This could strengthen the tailored message by persuading the recipient further.
19
1.6.1 Gamification
An emerging trend in apps is gamification. Gamification is a technique whereby game play elements are used in a non-game context. Typical game elements that are used include: explicit rule systems, competition, striving towards goals or outcomes (Deterding et al., 2011). There are three main parts that gamification has: 1. Motivational affordance (using game elements to motivate individuals, like points, leaderboards, achievements/badges, etc.) 2. Psychological outcomes (how the affordances influence the mind of individuals) and 3. Behavioural outcomes (how the affordances influence the behaviour of individuals) (Hamari, Koivisto & Sarsa, 2014). In the context of physical activity, it encourages users to exercise by including elements such as: competition, a progress bar, a leaderboard, awards/badges, rules etc. (Wong & Kwok, 2016).
An early example of a gamified health app is Nexercise (see image 1).
This application used medals, rankings and points to keep a record of the exercise pattern of the user (however, nowadays it is not in the app stores anymore). An even more extreme version is the application ‘’Zombie, run!’’. It is set up as a zombie apocalypse where the user has to run for their life and can collect items in order to survive (image 2).
Gamification strategies (such as goal setting, providing feedback on performance, reinforcement, comparing progress and social connectivity) share key elements with
established health behaviour change techniques. They add a competitive element to the app (Edwards et al, 2016). Companies have already used this concept as an instrument to increase the initiation and retention of the aspired behaviours (Lister et al., 2014).
20
The first type of the DISC model, the ‘’D’’ type, is focused on results and challenges.
Gamification in apps specifically adds more competition to exercising. In the article of Sepehr & Head (2013), it was found that adding competition as a game element can be a successful approach to engage people in learning. Because the D-type is typically triggered by
competition, this engagement can be even stronger. Further, Ferro, Walz & Greuter (2013) distinguished several player types in games, eventually classifying overarching personality types and game mechanics that would work specifically for them. For individuals classified as ‘’Dominant’’, game mechanics like achievements, progress bars, leaderboards or points will motivate this type of person more than other elements and therefore there is a higher chance that a message related to gamification will be perceived as positive. Because D-type
individuals need challenges to prevent boredom, including signs of competition in a notification message, text that is related to gamification might convince the user to use the app. Using words related to a leaderboard or badges will help to create a more favourable message that stands out more for ‘’D’’ types, in comparison to notifications that are related to other kinds of focuses or words.
The notification messages for other DISC-types
For ‘’D-’’ type individuals, gamification is an already established and used behaviour change technique to trigger their attention and extraverted behaviour (Puccio & Grivas, 2009; aDot, 2007). However, for the other types of the DISC profile, it is not that straightforward,
connecting behaviour change techniques with the three types. Still, there are some elements of fitness apps that can be incorporated in the notification message and be linked to the I-, S- and C- type.
1.6.2 The ‘’Influence’’ type
The ‘’I’’ type person likes to be around people and is outgoing. They prefer to work in groups. This type might need another approach to be engaged in using the application. The
fitness buddies part of the ‘’Pro-fit’’ app by Dharia et al. (2016) searches for other users in the
social network or contacts of the person and encourages them to work out together. This type of element in an exercise app, thus socializing, can be mentioned in a notification, so that the user gets a reminder that there is an opportunity to work out with others. The previously mentioned app called ‘’Nexercise’’ also uses the idea of a ‘buddy’ to encourage exercising together (see image 3).
21
Translating this concept to a message, Hirsh, Kang & Bodenhausen (2012) tailored messages to the personality types of respondents. They used the Big Five personality dimensions (also known as the Five Factor Model (FFM)) to distinguish five descriptors of personality
(openness, conscientiousness, extraversion, agreeableness and neuroticism). Translating this to the DISC model, Jones & Hartley (2013) found that the ‘’I’’ type of the DISC model corresponds to extraversion of the Big Five Personality dimensions. The extraverted
respondents turned out to find the messages, tailored to their personality, significantly more interesting in comparison to messages that were tailored to other personalities (Hirsh, Kang & Bodenhausen, 2012). It is important when communicating with an Influence type of person, that there is a feeling of being connected to the greater good and of being a productive
member of society (Turnasella, 2002). Therefore, using words related to working out together with an individual or with a group will attract the attention of a I-type person.
As mentioned earlier, the D-type and the I-type are both extraverted, meaning that they take an active role and turn their interests outward to other people, instead of inward (Turnasella, 2002; Geyer, 2007;Michaelson, 2010). Because of this overlapping trait, it is possible to combine a gamification element as well as a socializing element in one message, as both involve other people or users, thus turning their interests outward. The elements might
eventually reinforce each other and lead to a tailored message, taking into account the traits of the D- and I-type.
It is expected that:
H1: A notification message tailored to the D- and I- type will have a more positive influence on the attitude of the D- and I-type towards the message, compared to the attitude of the S- and C-type
Image 3 Screenshot of ‘’Nexercise’’,
22
1.6.3 The ‘’Steadiness’’ type
The ‘’S’’ type is focused on steadiness and prefers a stable and predictable environment. They should be given realistic goals to strive for and should not be put under pressure. Linking this to exercising, this person likes to have a goal, but not too much of a challenge or a strict exercise regime. According to the research by Jones & Hartley (2013), Steadiness is related to Agreeableness in the FFM. Further to this factor, it appears that when a message is tailored to Agreeableness, the individual is significantly more interested in that message compared to messages that are tailored to one of the other factors of the FFM (Hirsh, Kang &
Bodenhausen, 2012). By setting a goal in a health app and getting regular exercise based on the goal, the ‘’S’’ type will be satisfied. This is loosely already applied to a number of fitness apps. An example is Google Fit. In this health app, there a possibility of picking goals and keeping track of the progress (see image 4).
In communication, ‘’Stable’’ type individuals like it when appointments are planned in
advance (Turnasella, 2002). The S-type tends to direct their interests inward, upon themselves (Geyer, 2012) instead of outward to other, making them introverted (Michaelson, 2010). The goal-setting configuration found by Belmon et al. (2015) and Conroy, Yang & Maher (2014) can be suited for the S-type, as well as the example of a calender integration manager of Dharia et al. (2016) to integrate the calendar of the user with an exercise scheme. The notification message should therefore contain some text about goal-setting, like making a schedule.
23
1.6.4 The ‘’Conscientiousness’’ type
The ‘’C’’ type is the most analytical and likes accuracy. When talking about exercise, notification messages should be detailed. This is also already applied in some fitness apps, like Endomono (see image 5). This application tracks the performance of the user and displays it.
Turnasella (2002) states that when communicating with a Conscientious type, you should give precise information and great detail. The feature of self-monitoring in fitness apps seems to fit to this DISC model type of person. By constantly keeping track of the results of exercises and including this in a notification message, there is a high probability to keep this type engaged to the application. Adding some text about analyzing results will attract the attention of the C-type towards the notification message.
As well as the S-type, the C-type is also an introvert (Geyer, 2012; Michaelson,2010), meaning that they also focus more on themselves, rather than outward to others. This creates commonality. Therefore, as well as with the D- and I-type message, it is possible to make a notification message, containing an element for the S-type (adding a goal-setting/scheduling element) as well as an element for the C-type (analyzing results). Both elements focus more on the self, than on others and this helps adapting one message to the S-type as well as to the C-type to positively influence their attitude.
24
It is expected that:
H2: A notification message tailored to the S- and C- type will have a more positive influence on the attitude of the S- and C-type towards the message, compared to the attitude of the D- and I-type
1.6.5 Subliminal priming and message tailoring
With goal-priming, the individual will be primed so that he will behave in a desired way (Förster, Liberman & Friedman, 2007). In this case, the goal of the prime is to promote physical activity, which is also the goal of an exercising app. In the article by Mannetti et al. (2013), it was found that a subliminal prime before an either promotion or prevention framed political messages (about nuclear power and immigrants) reinforced the effect of the framing. The respondents were first subliminally primed with words related to promotion, for example immigrants, and then had to read a message that was framed with promoting immigrants. After, they had to complete a word-completion task. The respondents eventually chose more words related to promotion rather than words related to prevention. Therefore, it seems that subliminally priming guide individuals in the desired direction (Mannetti et al., 2013). In this research, the subliminal prime will be related to sports (which is a likely activity that is promoted by exercise or fitness applications). This should help to motivate the user to exercise and will perceive the notification message even more positively. It is expected that:
H3: A subliminal goal-prime related to physical activity will enhance the effect of a tailored notification message on the attitude of all DISC-types towards that message.
25
2. Data and Method
2.1 Respondents
The experiment was in the form of an online survey. Amazon Mechanical Turk was used to collect the majority of responses. Other tools were Facebook and surveycircle.com. All these websites together created a random mix of participants and therefore would enhance the external validity of the results. In total, 276 responses were recorded of which 66 respondents did not finish the questionnaire and therefore were excluded from the analysis. Of the
respondents, the majority was male (56,2 %). The mean age was 30 years old and most were employed (68,1%).
They all exercised, because according to Karremans, Stroebe & Claus (2006), priming only works if there is goal-relevancy, which is doing exercise in this case. The majority of the respondents exercised around 3 to 4 hours per week (33.3%), but also around 1 to 2 hours per week (29%). Furthermore, most people were satisfied with their work-out routine (78.6%) and the people that were not, mainly wanted to exercise more.
Of the 210 respondents, 47.6% currently uses a fitness-app. The most named fitness app was Fitbit. Of the 52.4% that currently did not use an fitness application, 27.3% used one before. Fitbit was again the most named fitness app that people used before.
2.2 Research design
The research design consists of an online experiment (online survey). It is between-subjects, comparing each DISC group which each other. It is a 2 (either tailored or non-tailored message)X 2 (subliminal prime : prime or no prime) between-subjects, factorial design. Therefore, there are four possible conditions where respondents were randomly assigned to.
1.Tailored message Subliminal prime (word ‘’exercise’’)
2.Tailored message No prime
3.Non- tailored message Subliminal prime (word ‘’exercise’)
4.Non- tailored message No prime
Table 1 all possible conditions
2.3 Pre-test
A pre-test was conducted to ensure that the subliminal prime would not be seen, so that the unconscious influence of the prime would work. 28 people participated in this pre-test. 5
26
participants saw something flashing, with 1 person recognizing the primed word. Therefore, the duration of the prime was reduced from 0.30 milliseconds to 0.16 milliseconds.
Eventually, of the 210 respondents, only one person saw the prime.
2.4 Procedure
First, there were general questions related to the existent work-out routine of the respondents. After, the respondents took a short DISC-test, consisting of 18 multiple choice questions where the respondents had to choose which option was most descriptive of their personality. Further, in a video of approximately 15 seconds, an introduction to a new fictional mobile fitness application was shown with some short clips of people running, hiking and cycling. Depending on the experimental or control condition, the respondent was subsequently subliminally primed with the word ‘’exercise’’(or was not exposed to a subliminal prime). The word was chosen because according to Wheeler and Berger (2007), the effect of the prime is dependent on what associations the individual has in mind. Because all respondents had to exercise to fill in the questionnaire, it is expected that they all have some exercise associations in their mind. These will be activated once the word ‘’exercise’’ is primed. At the end of the video, one notification message was randomly shown, either tailored or not tailored to the DISC-profile of the respondent. This creates two groups: one group that sees a
messages tailored to their personality and one group that sees a non-tailored message .When creating only two messages, this might give clearer results or correlations compared to creating four messages. To create two messages, elements were combined: one notification message contained tailored elements for the D- and I-type and one message for the S- and C-type. Combining was possible because these pairs have commonalities: D and I are more extraverted, while S and C are more introverted (Turnasella, 2002; Puccio & Grivas, 2009). Therefore, it was expected that tone of the message would correspond better when combining elements for the D- and the I-type, as well as for the S- and C-type. For the D-type, an
element of gamification was added to the notification message, namely the text ‘’…climb to the top of the leaderboard’’ (see appendix 1.1). Wong & Kwok (2016) argued that adding elements like a leaderboard or badges in a mobile fitness app encourages people to exercise. Since the D-type likes competition, this seems like an element that would trigger them. For the I-type, a social element was added to the message. They were given the ability to challenge their friends (see appendix 1.1). They like to be around people and have social interaction (Sugerman, 2009) and therefore adding the text ‘’Challenge your friends now…’’
27
is likely to get their attention. For the S-type, the message contained a feature of goal-setting and making a schedule (see appendix 1.2), because they prefer stability and like to be
prepared for what is coming by planning things (Sugerman,2009). This was done by adding the text ‘’Make your schedule now, set goals..’’. Last but not least, for the C-type, the message included the text ‘’…analyze your results.’’ (appendix 1.2). This is because the C-type likes analyzing and accuracy and likes things to be measurable. Similarly to the S-C-type, they like to plan ahead (Reid & Reid, 2003). Thus, by adding an element about analyzing as well as the text about the scheduling possibility (which also focuses on the S-type), it is likely that they will like the message. In the questionnaire, the participant could either receive a tailored notification message or a not tailored message, meaning that they received a message that was tailored the other types of the DISC-model. This was shown at the end of the video. After the video, the respondent had to answer questions regarding their attitude towards the notification message. This part contained 8 statements and the respondent had to state to what degree they agreed to each statement. Lastly, questions were asked about demographic details (gender, age, employment).
2.5 Measures
DISC-type
Normally, a full DISC-questionnaire has about 80 questions. Due to time limits within the online survey of this Master thesis, the questionnaire was shortened down to 18 questions. These questions would terminate which of the four DISC types the respondent was. Each question had the phrase ‘’Which of the following is most descriptive of you’’ followed by four options to choose from. Each option described two characteristics, belonging to either the D-, I-, S- or C-type (Grant, 2017). For example one could choose between ‘’trusting,
enthusiastic’’, ‘’consistent, thorough’’, ‘’courageous, adventurous’’ or ‘’agreeable,
accommodating’’. Each of the options were scored, so that after all the questions would lead to certain scores for each type of the DISC and the highest score would lead to someone’s personality type.
Subliminal prime
In the middle of the questionnaire, there was a 15 seconds-long video of an introduction to a fictional fitness app called Fit+. It contained a subliminal prime around the 10th second,
28
This had to do with the short-lived effect of the prime and therefore the notification message had to appear immediately after (Greenwald, Draine & Abrams, 1996).
Attitude towards the notification message
The attitude towards the notification message that was shown at the end of the video, was measured right after. The attitude was measured with 8 statements and on a 5-point Likert scale. The question were for example ‘’What did you think of the notification message?’’ and for each of the following statements, the respondents had to choose how much they agreed with the statement, choosing between (1) strongly agree, (2) somewhat agree, (3) neither agree nor disagree, (4) somewhat agree and (5) strongly agree.
The eight statements were: 1.’’I liked the notification message’’, 2. ‘’I found it easy to understand’’, 3.’’ The message was user friendly’’ 4. ‘’The message stood out to me’’, 5. ‘’The content of the message fits my personality’’, 6. ‘’The message motivates me to
exercise’’, 7. ‘’I think the message will motivate others to exercise’’ and 8. ‘’After reading the message, I would use the Fit+ app more’’.
The statements were derived from the articles of Hirsh, Kang and Bodenhausen (2012) and Kreuter, M. W., Bull, F. C., Clark, E. M., & Oswald, D. L. (1999). A couple of statements of each article were taken to form a new scale for this article to measure attitude. Because this scale did not exist before, it was checked if there were eventual underlying factors between the items. Also, the reliability was tested.
An exploratory factor analysis, also known as principal axis factoring analysis (PAF) was conducted on the scale. The Kaiser- Meyer-Olkin (KMO) test confirmed the sampling adequacy for the analysis, KMO= 0.878. Barlett’s test of sphericity χ² (28) = 862.163, p< 0.001, pointed out that correlations between the items were large enough for the PAF. Another analysis was run to get eigenvalues for each item in the data. Two items had eigenvalues over Kaiser’s criterion, >1, and together explained 60,28% of the variance. The examination of the scree plot reported a levelling off after the second factor. Therefore, two factors were retained and rotated with a Oblimin with Kaiser normalization rotation. This further analysis actually showed that there were five items that clustered on factor 1 and only one item that clustered on factor 2. The individual item clustering with factor 2 was the statement’’I found it easy to understand’’. However, according to Hair et al. (2010), a factor ideally has three items, to ‘’provide minimum coverage of the construct’s theoretical
domain’’. This factor and single item was excluded from further analyses. The statements ‘’The message was user-friendly’’ and ‘’The message stood out to me’’ showed
cross-29
loadings on factor 1 as well as factor 2. This could be due to the content of the items. It was decided to exclude these items as well in the following analyses, because it does not clearly belong to one factor.
Looking at the five remaining statements, the scale had a high reliability with a Crohnbach’s alpha of .911. Additionally, the corrected item-total correlations demonstrated that all the items had a good correlation with the overall score of the scale (all above .30). Further, the items would not affect reliability if they were deleted.
All five items together are one reliable scale to measure the attitude of the respondents towards the notification message.
30
3. Results
Hypothesis 1 and hypothesis 2 test if tailoring a notification message to the personality of the DISC-types would influence their attitude towards that message. Hypothesis 1 focuses on the D- and I-type and hypothesis 2 on the S- and C-type.
To analyze these hypotheses, a one-way Analysis of Variance (ANOVA) was performed. The ANOVA was used to determine if there were differences between the means of different, independent groups. In this case, there were two groups, namely a group that was exposed to a message tailored to their personality type and a group that was exposed to a message that was personalized to the other personality types (so not tailored to their personality). The
measurement of the attitude had five statements and together formed one scale.
Table 2 shows the frequency of each DISC-type and table 3 the frequency of the tailored and not tailored conditions.
Looking at the Levene’s test, there was homogeneity of variances (p=.125). Consequently, the standard one-way ANOVA was interpreted for hypothesis 1 and 2.
To analyze the data separately for each condition, the SPSS configuration ‘’split file’’ was used to split the output in the type of message (tailored to the D-/I-type or to the S-/C-type) to examine the differences.
DISC-type Count % D-type 38 18,1 I-type 27 12,9 S-type 94 44,8 C-type 51 24,3 Total 210 100,0 Count % not tailored 98 46,7 tailored 112 53,3 Total 210 100,0
31
H1: A notification message tailored to the D- and I- type will have a more positive influence on the attitude of the D- and I-type towards the message, compared to the attitude of the S- and C-type
There were no statistically significant differences between attitude of the DISC-types towards the message that was tailored to the D- and I-type, F(3,99)= 1.580, p=.199.
Since the hypothesis researches the influence between specific DISC-types, the next step was to execute planned contrasts to see if there was a significant difference between attitudes of the D- and I-type versus the S- and C-type towards the notification message that was tailored to the D- and I-type. It revealed that there was no significant difference between the attitude of the D-type (M=3.8, SD= 1.0) and S-type (M=3.7, SD= 0.8) towards a message tailored to the D- and I-type, t(99)= .425, p=.672. However, comparing the attitude of the D-type and the C-type (M=3.2, SD=1.1), there is a tendency towards significance, t(99)= 1.920, p=0.058, with a value of contrast of .591. This means that the D-type’s attitude was higher than the attitude of the C-type, as was expected.
The contrast between the attitude of the I-type (M=3.5, SD=1.2) and the S-type did not turn out significant, t(99)= -.638, p=.525. And finally, there was also no significant difference in attitude between the I-type and C-type, t(99)= .901, p=.370.
Overall, there were no significant results to support the hypothesis that the attitude of the D- and I-type was more positive than the attitude of the S- and C-type towards the notification message tailored to the D- and I-type. There was, however, a tendency towards significance comparing the attitude of the D-type and C-type. Still, hypothesis 1 was not supported. Table 4 summarizes the results of the planned contrasts.
Table 4 Contrasts DISC-types D- and I-type message
DISC-types Value of Contrast Std. Error t df Sig. (2-tailed) D- vs S-type 0,1156 0,27176 0,425 99 0,672 D- vs C-type 0,5913 0,30794 1,920 99 0,058 I- vs S-type -0,1844 0,28912 -0,638 99 0,525 I- vs C-type 0,2913 0,32336 0,901 99 0,370
32
H2: A notification message tailored to the S- and C- type will have a more positive influence on the attitude of the S- and C-type towards the message, compared to the attitude of the D- and I-type
The one-way ANOVA showed that there were no significant differences between the attitude of the DISC groups towards the notification message that was tailored to the S- and C-type,
F(3, 103)= .511, p=.675.
Because this hypothesis also specifically investigates the influence in attitude between specific DISC-types, planned contrasts were executed.
There was no significant difference between the attitude of the S-type (M=3.4, SD=.88) and the D-type (M=3.5, SD=.98) towards the notification message tailored to the S- and C-type, t(103)=-.409, p=.684. The same was true for the attitude of the C-type (M=3.2, SD=1.23) compared to the D-type, t(103)= -1.150, p=.253. The contrast of the attitude between the S-type and the I-S-type (M=3.4, SD= .86) towards the S-/C-S-type tailored message was also not significant, t(103)= .104, p=.918. Finally, the attitude of the C-type compared with the attitude of the I-type was also not significant, t(103)= -.553, p=.581.
Given these results, hypothesis 2 was not supported. Table 5 summarizes the results of the contrasts.
Table 5 Contrasts DISC-types S- and C-type message
H3: A subliminal goal-prime related to physical activity will enhance the effect of a tailored notification message on the attitude of all DISC-types towards that message.
For this hypothesis a two-way ANOVA (also known as factorial ANOVA) was used to analyze, because this type of ANOVA can compare mean differences between groups that have been divided between two independent variables and see if there is an interaction
between these variables. For this hypothesis, the moderator, the subliminal prime, is treated as an independent variable and is compared with the effect of tailoring a notification message. Table 6 below reports the times each condition was randomly assigned to the participants.
DISC-types Value of Contrast Std. Error t df Sig. (2-tailed)
S- vs D-type -0,1100 0,26920 -0,409 103 0,684
C- vs D-type -0,3406 0,29607 -1,150 103 0,253
S- vs I-type 0,0345 0,33234 0,104 103 0,918
33
There was homogeneity of variances, as assessed by Levene’s test for equality of variances, p=.068, so no additional action had to be taken before analyzing.
To see if the subliminal prime would have an effect on the attitude of the DISC-types, the results were analyzed separately for the D-,I-, S- and C-type.
For the D-type, no statistically significant interaction was found between message tailoring and subliminal priming for the attitude score of the D-type, F(1,34)=1.966, p= .170, partial η2 = .055. Further, looking at the main effects, there was no significant main effect of the subliminal prime on the attitude score, F(1,34)= 1.607, p= .214, partial η2= .045 .
Message tailoring did not give a statistically significant main effect for the D-type either, F(1,34)= .969, p= .332, partial η2= .028.
It seems like there is no additional effect of a subliminal prime on the attitude of the D-type towards the notification message.
For the I-type, there was also no statistically significant interaction between message tailoring and subliminal priming for their attitude score, F(1,23)= 1.886, p= .183, partial η2= .076. Next, there was no significant main effect of the subliminal prime on the attitude of the I-type, F(1,23)= .141, p=.710, partial η2= .006. Message tailoring also did not give a significant main effect, F(1,23)= .202, p= .657, partial η2= .006.
Therefore, there was no interaction or main effect of the subliminal prime found for the I-type.
The interaction between subliminal priming and tailoring the notification message was not statistically significant for the S-type, F(1,90)= .386, p= .536, partial η2= .004. There was no significant main effect of subliminal priming on the attitude towards the notification message, F(1, 90)= 1.105, p=.296, partial η2= .012. Also message tailoring had no main effect on the attitude score, F(1,90)= 2.115, p=.149, partial η2= .023.
Condition Count % D+I prime 51 24,3 S+C prime 52 24,8 D+I no prime 52 24,8 S+C no prime 55 26,2 Total 210 100,0
34
For the S-type, subliminal priming did not have a significant effect on the attitude towards the notification message.
For the C-type, the interaction between message tailoring and subliminal priming was not significant, F(1,47)= .040, p= .842, partial η2= .001. There was no significant main effect of priming for the C-type, F(1, 47)= 1.289, p=.262, partial η2= .027. The tailoring of the notification message did not give a significant main effect, F(1,47)= .021, p= .886, partial η2= .000.
Thus, for the C-type, it seems like the subliminal prime did not have an effect on the ‘’attitude towards the notification message’’ score.
Investigating the effect of the subliminal prime on the attitude of all DISC-types, there was no statistically significant interaction between subliminal priming and message tailoring for the ‘’attitude towards the notification message’’ score, F(1,206)= .167, p=.683, partial η2= .001. Additionally, there was no statistically significant main effect of subliminal priming on the attitude score towards the notification message, F(1,206)= 2.954, p=.087, partial η2= .014. There was also no significant main effect of message tailoring, F(1,206)= .081, p=.776, partial η2= .000.
In conclusion, the results show that, for all DISC-types, there is no interaction between the subliminal prime and tailoring the notification message on the attitude towards that message. Also, when looking at the overall effect of subliminal priming, it seems that there is no significant effect.
Table 7 shows the overall effect message tailoring and subliminal priming. It seems like subliminal priming has the opposite effect on the attitude towards the message than expected: a subliminal prime causes a less positive attitude towards the notification message than without a prime. However, because the results were not significant, it is only a speculation. Overall, all results suggest that the hypothesis was not supported.
35 Table 7 Overall effect message tailoring and subliminal priming
