Masterthesis
Continue to work out!
How attaining physical goals influences individuals to continue using
fitness apps.
Ruud Bluyssen (s4482174)
Master’s Thesis marketing
14-06-2020
Prof B. Hillebrand
Continue to work out!
How attaining physical goals influences individuals to continue using
fitness apps.
Masterthesis Marketing
Current version:
Final Masterthesis
Student name:
Ruud Bluyssen
Student number:
s4482174
Submission Date:
14-06-2020
Supervisor:
Prof B. Hillebrand
Second supervisor:
Dr. V. Blazevic
Educational program:
Master in Business Administration, specialization: Marketing
Faculty:
Nijmegen School of Management
Preface
Finally I can say this is the final version of my master thesis. The past half year I have been working hard on my biggest university project yet. This project has learned me to think and write differently as a marketer. In these strange times of 2020, it has been a real challenge where everything was different than normal, including writing a master thesis. Eventually, after a busy period, everything worked out and I am proud of the result.
I would like to thank my supervisor Prof. Bas Hillebrand for all the critical feedback and support (via Skype) in the whole process of writing this thesis. Because of his feedback I was able to think differently and dig deeper into the right subjects. I would also like to thank the second
examiner dr. Vera Blazevic for her time and effort. Last, I would like to thank all the respondents that took the time to fill in the survey. They helped to shape this master thesis as it is right now.
I hope you enjoy reading my master thesis!
Ruud Bluyssen
Abstract
Objective: Lack of physical activity is a growing global problem. Fitness apps are a
technology that can facilitate physical health improvement. Initial adoption of fitness apps is
a great start, but continuance usage intention is necessary to actually change physical
health. Attaining physical goals are a main behavior change technique that is being used in
the fitness apps. There is a gap in knowledge how attaining physical goals has an influence in
the continuance intention of fitness apps. This research tries to close this gap by developing
a conceptual model based on the post-acceptance model (PAM) that incorporates attaining
physical goals and fitness app continuance intention among the variables fitness app
satisfaction, habit, and technostress to explain the relationships.
Method: A survey was conducted resulting in a sample of 168 native Dutch fitness app users.
A partial least square analysis was done to test the proposed model and find and interpret
the strength of the relationships.
Results: The results show a significant positive relationship of attaining physical goals on
fitness app satisfaction, that in turn has a positive relationship with fitness app continuance
intention. Habit was also found to have a positive relationship with fitness app continuance
intention. Next to that, the results show that technostress is not a significant moderator of
the relationship between attaining physical goals and fitness app satisfaction.
Discussion/conclusion:
Main findings from this study contribute by developing and empirically testing a model that
explains the constructs that motivate individuals to keep using fitness apps. Theoretical
contributions are given and practical implications for fitness app developing firms are stated.
Keywords: Continuance research, Post-adoption research, Goal attainment, Mobile health,
Fitness app, habit, technostress
Contents
Chapter 1 – Introduction ... 1
1.1 Introduction ... 1
1.2 Problem statement and research question ... 3
1.3 Relevance ... 5
1.4 Structure ... 6
Chapter 2 – Theoretical framework ... 7
2.1 Previous research ... 7
2.2 Expectation-confirmation theory ... 8
2.3 Post-acceptance model ... 9
2.5 Key variables & hypotheses ... 10
2.5.1 Dependent variable fitness app continuance intention ... 10
2.5.2 Independent variables ... 10
2.5.3 Moderator technostress ... 12
2.6 Conceptual model ... 14
Chapter 3 – Methodology ... 15
3.1 Research design ... 15
3.2 Participants ... 15
3.3 Pre-test ... 16
3.4 Survey design ... 17
3.5 Research Ethics ... 18
3.6 Operationalization ... 18
3.6.1 Control variables ... 20
3.7 Analysis design ... 21
Chapter 4 – Data and results ... 22
4.1 Sample description ... 22
4.2 Discriminant validity and convergent validity ... 24
4.3 Reliability analyses ... 26
4.4 Partial least square path modeling analysis ... 27
4.4.1 Measurement model ... 27
4.4.2 Structural model ... 28
4.5 Path coefficients and effect sizes ... 28
4.6 Post-hoc analysis ... 30
Chapter 5 – Discussion ... 33
5.1 Conclusion and discussion ... 33
5.3 Practical implications ... 37
5.4 Limitations and future research ... 38
References ... 40
Appendix 1 – Operationalization table ... 44
Appendix 2 – Survey ... 47
Appendix 3 – Frequency of apps ... 51
Appendix 4 – Factor analysis (first attempt) ... 52
Appendix 5 – Factor analysis (final attempt) ... 55
Appendix 6 – Factor analysis for convergent validity ... 58
Appendix 7 – Reliability analyses ... 59
Appendix 8 – Adanco model ... 61
Appendix 9 – Adanco output ... 62
Appendix 10 – Posthoc analysis output c ... 67
Chapter 1 – Introduction
1.1 Introduction
Lack of physical activity has been clearly shown to be one of the biggest risk factors for a lot of health conditions. Physical inactivity is the fourth leading risk factor for premature death across the world (Kohl et al., 2012). Currently, we are living in the most health engaged era ever. This happened due to the fact that individuals are better informed and educated about the risks of an unhealthy lifestyle. On the other side, daily information sources, such as social media, TV, and magazines, have also informed individuals of the benefits that a healthy lifestyle can bring both physically, mentally, and socially (Gilhuly-Mandel, 2018). Not only do these sources inform individuals, they also set a reference for a perfect body image that individuals inspire to be like. Muscular men and skinny female models are displayed multiple times in daily lives, resulting in social pressure to achieve those ideal body images depicted by media (Jung, 2011). While this is happening, the number of individuals who are physically inactive is still growing in many countries, bringing their general health in danger (Romeo et al., 2019).
Because of the high popularity of physical health and the ideal body trend that is showing in recent years, many information systems have been developed and used to promote and monitor physical activity. These include internet-based devices and applications (apps), such as fitness-tracking apps on smartphones and smartwatches (Hsiao & Chen, 2019). The high popularity of these apps also attracted the attention of researchers to study how people use these apps and to what effect (Vaghefi & Tulu, 2019).
Devices such as the smartphone and smartwatch offer the perfect opportunity to monitor physical activity. Globally, activated smartphones outnumber citizens for a couple of years now (Romeo et al., 2019). On top of that, individuals do not just own a smartphone, most of them also carry their smartphone on them almost all the time, giving it a ubiquitous nature not only for communication purposes but also for health purposes. The smartwatch is even more close, giving notifications and haptic feedback available at all times on your own wrist (Baretta, Bondaronek, Direito, & Steca, 2019).
Due their powerful communication and interactive features, devices like smartphones and smartwatches offer the ideal opportunity to bring an unbounded amount of health information at any place and time. This is the reason that a lot of mobile health apps have been developed for these devices. The number of available health apps in the major app stores is continuously growing, as are the number of downloaded apps itself (Baretta et al., 2019). In 2017, already more than 325,000
mobile health apps were available and in the same year 3.7 billion mobile health apps were downloaded worldwide (Research2Guidance, 2017; Statista, 2019a). These apps are a great way to target lifestyle-related behaviors, such as fitness and physical activity or nutrition management (Chen & Allman-Farinelli, 2019). Fitness apps are the most downloaded type of mobile health apps (Beldad & Hegner, 2018). All mobile fitness apps fall in the “Health and Fitness” category in the iTunes App Store or Google Play store and are defined as applications on mobile devices that are designed specifically to assist with exercise, other types of physical training, nutrition, and diet (Beldad & Hegner, 2018). Individuals download these apps with the goal to increase their health or physical activity. In 2016, 33% of the global population used an app to track their fitness, and the forecast to the fitness app market size is expected to reach over USD 15.96 billion by 2026, giving it great opportunities for marketers and app developing companies to research, develop, and invest even more in fitness apps for the massive global target group (PolarisMarketResearch, 2020; Statista, 2019b).
In these fitness apps, many people like to be encouraged to start exercising and prefer objectives and coaching to motivate them to reach their goals. Being able to see positive outcomes of health goals during attempts of physical health change can be very rewarding and have a positive impact on the self-efficacy and satisfaction of an individual. This in turn will have a positive influence on the attainment of the target health goal (Liu & Willoughby, 2018). Goal attainment has become more noticeable with the assistance of mobile health apps, and in particular fitness apps, due to external incentives such as virtual badges and prizes, push notifications, and social training comparisons with friends motivating the user to attain their goals every time (Chuah, 2019).
Depending on the specific nature of the fitness app, the goal can be a daily, weekly, monthly, or even longer-term goal. Examples of these goals for which you can get “virtual prizes” can include number of steps to walk and monthly training days, but also number of glasses of water to consume a day or a daily calorie consumption. If the goal is not yet obtained, notifications like “It’s not too late, take a walk for 12 minutes to attain your daily goal” can be given as an extra last-minute motivation. In other words, you set your own personal goals but the apps will motivate you to attain your goals through a planning, evaluation of your performance according to the pre-set goal, feedback if necessary, social support and finally the provided rewards (Huang & Zhou, 2019). The apps therefore offer a number of incentives to motivate you to attain your physical goals (Y. Wang, Wang, Greene, & Sun, 2020).
Fitness apps make clever use of behavior change techniques. These techniques include for instance goal setting, self-monitoring and performance feedback. It is proven that these techniques facilitate health behavior change (Middelweerd, Mollee, Van Der Wal, Brug, & Te Velde, 2014). Taken together these techniques can be called goal management and this is the way in which goals are
attained (Oinas-Kukkonen & Harjumaa, 2009). The self-tracking ability of fitness apps will increase motivation for goal-setting, leading to the goal of physical behavior changes and eventually even improving the quality of the users life (Chuah, 2019). Empirical research has shown that fitness apps on smartphones have been used for not only the effectiveness of monitoring health conditions such as heart rate and blood pressure during training, but are also proven to effectively facilitate health behavior change and improve key lifestyle habits, such as daily physical activity (Huang & Ren, 2020; Middelweerd et al., 2014; Patel, Asch, & Volpp, 2015).
An advantage of the goals that can be set on mobile health apps is that they can be adapted to the specific person, even after they are adopted. This includes different steps, beginning with action planning, followed by goal evaluation and resulting in goal re-evaluation. Once a personal goal has been set, these different steps in the process allow personal adjustment of the level and
direction of the physical goals and the effort that is needed, which increases the potential effectiveness of the goal-setting and goal attaining (Baretta et al., 2019).
Following goal setting theory (Locke, 1981) attaining physical goals is here defined as
attaining a specific standard of proficiency on a personal physical task, usually within a specified time limit. This definition holds that these goals are recurring goals for fitness apps. These apps usually have goals that are for example daily and come back over and over again. Therefore, the goal of overall better health is a long-term overarching goal. (Locke, Shaw, Saari, & Latham, 1981)
1.2 Problem statement and research question
While initial adoption of fitness apps is a great first step, the ultimate success of a technology relies on the continuance use of it (Bhattacherjee, 2001). Fitness app continuance intention is defined based on the definition of Bhattacherjee (2001): an individual’s intention to continue using a fitness app (in contrast to initial use or acceptance). Although billions of mobile health apps were
downloaded, a national survey in the US revealed that the attrition rate of health app users was 45.7%, which means they downloaded health apps they currently no longer use (Krebs & Duncan, 2015). Another study unveiled that a quarter of mobile health apps is used only once (Vaghefi & Tulu, 2019). Although it is proven that mobile health apps, and therefore fitness apps, are effective to monitor and improve physical health data, for mobile health apps to produce their intended effect, users need to continue to use the apps for a longer period of time, which will incorporate the desired behavior changes into their daily life routines (Huang & Ren, 2020). Therefore, understanding how to increase user retention rate is a very important issue for not only fitness app developers, but also health practitioners and society at large (Molina & Sundar, 2020).
Prior studies examined factors that drive users to adopt a new information technology such as fitness apps. These studies were mainly based on innovation diffusion theory (Rogers, 1995), the theory of planned behavior (Ajzen, 1991), and Davis’ technology acceptance model (Davis, Bagozzi, & Warshaw, 1989). These studies all examined the initial acceptance of a new information system, rather than factors that influence users to continue the use the technology after they have adopted it (Bhattacherjee, 2001; Hong, Thong, & Tam, 2006). The increased market penetration of fitness apps has shifted the interest of both academics and practitioners from initial adoption to continued use (Chuah, 2019).
Since health improvement is a long-term process, and this is the physical goal that the user wants to achieve with the app, the fitness apps must be used for a longer period of time to be the factor facilitating the actual health improvement. It is expected that attaining physical goals over and over again is a way to keep individuals motivated to change and improve their health. The apps are developed to achieve small recurring goals in order to achieve the overarching goal. Therefore, the long-term success, and thus continuance use of a fitness app also depends on attainment of the physical goals. On the other hand, if goals in fitness apps are not attained over and over, there will be no motivation to continue using the app, since the physical improvement is not happening.
There currently is a need to dive deeper in mobile health apps users’ post-adoption behaviors (Cho, 2016). Despite awareness of global health issues of physical inactivity is very high, and continuance use of mobile health apps is relatively low, little empirical work has been done to explore the relationship between physical goal attainment and continuance intention (Chuah, 2019). For a part, not much research has been done due to the fact that mobile health apps are still
relatively new (Cho, 2016). Only if app developers know if this relationship really exists in practice, they will focus more on the behavior change techniques that motivate individuals to attain their goals and therefore keep using the fitness apps.
The goal of the current study is to help the global health issue by trying to understand the connection between attaining physical goals and continuance usage intention of fitness apps, in order to give managerial suggestions on how to keep retention rates of fitness apps high and contribute to continuance information systems usage literature that can eventually lead to an improvement of physical health due to fitness apps.
This can be translated to one main research question that has to be researched:
“What influence does physical goal attaining have on the continuance usage intention of fitness apps?”
1.3 Relevance
The current study is one of the first attempts to research the relationship of goal attainment on continuance usage intention in the growing market of mobile health apps. The relevance for researching this topic is three folded.
First, the study has social relevance, which is seen from the perspective of the users and society. The need for more physical activity is clear, it reduces the chance of premature death. Therefore it is also necessary to research ways to improve physical activity. Using fitness apps more and thereby improving health is such a way. If goal attaining theory is related to that, this has to be researched more extensively to make sure individuals actually keep using mobile health apps and attain their physical goals with them. This study therefore gives insights what fields to focus on in the mobile health app market to eventually work towards a healthier society.
Second, the research is academically relevant, because empirical research on the billion-dollar industry of fitness apps is scarce and this research extents knowledge to this topic. Fitness apps fit in the industry of IoT systems, and it is important to gain knowledge about this, because this industry is still rapidly developing and becoming even more important in the near future (Cho, 2016). Most research has been done about the initial acceptance and adoption of information systems like fitness apps, while the long-term viability and success of the apps lies in the continuance usage intention (Bhattacherjee, 2001). This study increases knowledge about post-adoption behavior of the apps. Some existing models have already empirically proven to explain a part of the variance of continuance intention multiple times, but do not take into account goal attainment, which is a variable that plays a big role in the fitness app context. An extended and adapted model fits better in the context of fitness apps post-adoption behavior and might explain more variance of continuous usage intention than the existing models. This extended model can then be generalized to contexts of other industries and cultures.
Third, there is a managerial relevance, which is seen from the perspective of the app developing firms. Literature about the continuance usage intention of mobile health apps is very important for business processes, because infrequent, inappropriate, and ineffective long-term use of fitness apps by consumers will contribute to less growth of the fitness app market and app firm failures (Bhattacherjee, 2001). The original functions of a technology tend to be adapted trough individuals continuous use of it, because the behavioral aspects and societal meaning related to a technology change over time. By focusing on what individuals like about the technology and on how to improve the public opinion about it, the chances of continuous usage are higher. Long term usage shows that customers are satisfied with the apps, making it more interesting to develop, update, and invest in fitness apps and motivate people more with improved apps to keep a healthy lifestyle. This
is again especially important for IoT products, because the quickly changing market can result in reduced life cycles of health apps, which makes it necessary to examine people’s reasons for continuance usage intention of current health apps (Cho, 2016).
On top of that, the forecast to the mobile health app market is very positive and growing every day, making the competitive landscape to grow as well. This makes is even more important to gain knowledge about this topic for fitness app companies itself to grow and beat the competition. If the companies know how to retain customers, this will let investors stay interested to keep investing in app developing companies. Next to that, it is also interesting for app developing companies on another level, because the cost of acquiring a new customer is five as high as that of retaining a current customer (Nascimento, Oliveira, & Tam, 2018). Taken together, this research helps app developers to make their products better by gaining knowledge on the factors that increase the retention rate of consumers. This knowledge can then be used as a practical focus point for fitness app companies in practice (Cho, 2016). Eventually this knowledge can then be used to make the society more physically active by continuous usage of health apps.
1.4 Structure
The research is reported using five chapters in order to answer the main research question. In chapter two the theoretical framework is presented to elaborate on the current literature about the topic and to explain the theories that are used in this study. Next to that, the hypotheses are
explained leading to a conceptual model. After that, the methodology chapter follows. In this chapter the research design, participants, survey design, research ethics, and operationalization are
elaborated and discussed. Chapter four presents the data and results. Here the results obtained from the survey and an analysis of the data is presented, followed by an analysis of the main relationships between the variables. In the last chapter the results are interpreted, and the research question is answered with the help of the outcome of the hypotheses. Finally the contributions, implications, and limitations of the study are stated.
Chapter 2 – Theoretical framework
First of all, in this chapter previous research is be briefly discussed. Next, 3 models are explained as a theoretical framework for this research. Using the theory hypotheses are stated with the help of the explanation of the key variables. These hypotheses form the conceptual model.
2.1 Previous research
Although there are studies that provide valuable insights on the adoption of mobile health apps and a few about the particular category of fitness apps, the question remains what guarantees the continued use of fitness apps. On one hand some studies looked at the design, functions, and persuasive nature of the apps itself (Vaghefi & Tulu, 2019). These studies were often done to categorize the numerous health apps currently available in a structured way. However, since these studies were not done with primary data obtained from health app users, they don’t explain the behavioral reasons why and how users adopt and keep using the app. On the other hand, another group of studies looked at the apps from a perspective of the user, to try to understand the behavior of the app user (Cho, 2016). A few studies regarding the use of mobile health apps have shown the importance of personal factors, such as user’s motivation, existing health conditions, and individual differences. Some studies used goal setting to study the effect of this on continuous usage (McEwan et al., 2016). These studies offered lots of information, but missed the point of another personal factor, attaining the physical goals, that the mobile health apps use to motivate you to actually improve your health. The whole aim of fitness apps is to attain your goal of being healthier and therefore the variable goal attainment is a behavior change technique that has to be included to guarantee continuance usage (Huang & Zhou, 2019).
Multiple theories can be applied to research this gap in knowledge. Two theories that are used in other information system continuance studies are expectation-confirmation theory originally developed in 1977 (Oliver, 1977) and the Post-acceptance model (Bhattacherjee, 2001). The acceptance model is an improved version of the expectation-confirmation theory. This post-acceptance model is adapted to fit the current study and context. In the following sections these theories and the use of them in the current study is explained.
2.2 Expectation-confirmation theory
Expectation-confirmation theory (ECT) is originally frequently used for research on consumers’ repurchase behavior from a marketing perspective, such as repurchase or complaining (Cho, 2016). Users’ intention to continue to use a technology (or repurchase it) is determined by their satisfaction with the technology (Bhattacherjee, 2001). The theory suggests that user satisfaction is influenced by the confirmation of expectations from prior technology use. Even before purchasing and using a technology, a consumer develops expectations about the product. (Thong, Hong, & Tam, 2006). These expectations form a baseline of satisfaction with an technology before usage (Bhattacherjee, 2001). After using a system, perceptions of the system are reconsidered and then confirmed or disconfirmed. This also means that the expectations of the system may have changed. If the system does not meet the expectations, i.e. the expectations are disconfirmed, this will result in a negative post-adoption belief. In other words, satisfaction with the system is not likely without the
confirmation of pre-adoption expectations of the system. In turn, dissatisfaction will not result in the continuance usage of an information system (Bhattacherjee, 2001; Ogbanufe & Gerhart, 2018).
In the context of fitness apps, confirmation of expectations is the confirmation of goals that are set, which is the attainment of goals. Therefore, the confirmation concept is replaced by goal attainment. Mobile health apps can help to motivate individuals to attain their physical goals and become healthier. This is most likely also the reason why they adopted the app in the first place (Chuah, 2019). Individuals start using a fitness app with the expectations that they will reach their physical goals. Following the logic of the ECT this means that if the physical goals are attained, the user of the fitness app is more satisfied with the app, which results in continuance usage intention of the app.
Fitness app satisfaction is in this study a concept that is a post-purchase feeling, since it is a concept that is about perceptions that can only be present after the actual use of the product. In this research we use Ogbanufe’s (2018) definition of satisfaction because he used it in a comparable context of mobile health apps on smartwatches. He defines it as an “ex-post positive affective evaluation of the smartwatch“. Therefore, satisfaction is defined as “ex-post positive affective evaluation of the fitness app” in this study. In this definition fitness app satisfaction cannot happen before adoption and it is a result of an evaluation of expectations and confirmation of the fitness app goals (Ogbanufe & Gerhart, 2018). Satisfaction is viewed as the key to obtaining a loyal and long-term relationship with consumers and is therefore important for continuance intention
2.3 Post-acceptance model
The ECT was originally developed for repurchase intention. To use a model that was made especially for continuance intention, an new model based on the ECT was developed called the
post-acceptance model (PAM) (Bhattacherjee, 2001). An adapted version of this model is used as the conceptual model in the current research. The original PAM model takes the concepts confirmation and satisfaction from the ECT to explain the continuance intention of the technology. In other words, the extent to which a technology confirms the user’s expectations of the technology result in the extent to which the user becomes satisfied and this in turn results in the continuance usage intention. In the PAM, the concept of perceived usefulness is added to this as a mediator to strengthen the logic of the relationship between confirmation and satisfaction on the continuance intention (Cho, 2016). Perceived usefulness is “the degree to which a person believes that using a particular system would enhance his or her job performance” (Davis, 1989, p. 320). The model is adapted to the context of continuance intention, by replacing ECT’s expectation with post-usage perceived usefulness. Perceived usefulness is referred to as a post-usefulness perception that results from accumulated usefulness perceptions. In this way it is not a single perception, but a long-term usefulness, resulting in long term satisfaction. Following this logic, it is used in continuance theory (Nascimento et al., 2018).
The usefulness of a fitness app is to attain physical goals and facilitate health improvement. In other words, when physical goals are not attained, the app is perceived as useless, since health improvement is the whole basic aim the app. As explained before, the confirmation of expectations is the confirmation of goals that are set, which is the attainment of goals. Therefore, the concepts perceived usefulness and attaining physical goals are very similar. If both the concepts confirmation and perceived usefulness are used in one conceptual model, there is a high chance that they are not distinguishable in analysis. Therefore, if we use fitness apps in the PAM, it is not possible to use both concepts at the same time. The two concepts of confirmation and perceived usefulness can however be taken together as the concept attaining physical goals. To explain it in one sentence, the
confirmation of the expectations is attaining physical goals, and this is the usefulness component of the fitness app.
Summarizing this whole model, after a period of usage of fitness apps, physical goals are attained or not attained. This degree of goal attainment (confirmation) form new feelings about the technology, which is basis for the fitness app satisfaction. The user is satisfied that he attains his fitness goals, because that is why he downloaded the app in the first place. Finally, the fitness app satisfaction has an influence on the users continuance intention of the technology, in this case the fitness app (Sørebø & Eikebrokk, 2008). This model plays a major role in the current research. A few
existing studies have used and validated the post-acceptance model in the context of information systems and one even regarding mobile health apps (Cho, 2016).
2.5 Key variables & hypotheses
To answer the research question an adapted version of the PAM is used that fits the context of fitness apps. This model contains several independent variables and a moderator that together influence the dependent variable. In the next paragraphs the key variables are explained and hypotheses are formed in order to build towards the conceptual model of this research.
2.5.1 Dependent variable fitness app continuance intention
The dependent variable is fitness app continuance intention and is based on the definition of
Bhattacherjee (2001, p. 352): “an individual’s intention to continue using a fitness app (in contrast to initial use or acceptance)” (Bhattacherjee, 2001). It is possible that individuals have the intention to use the fitness app every day or do not want to use the app anymore.
2.5.2 Independent variables
The first independent variable is fitness app satisfaction. This is defined as “ex-post positive affective evaluation of the fitness app” (Ogbanufe & Gerhart, 2018). This means there is a positive and
pleasurable state of attitude and emotion about the technology. In general marketing literature, satisfaction is considered as a key predictor of building and retaining a loyal base of long-term consumers (Limayem, Hirt, & Cheung, 2007). According to the ECT and the PAM, a users’ information system continuance intention is directly influenced by satisfaction with prior use. A lot of existing research regarding the ECT confirms this relationship (Bhattacherjee, 2001). If a user feels satisfied, he or she will have a stronger intention to continuously use the fitness app (Li, Liu, Ma, & Zhang, 2019). Multiple recent comparable studies researching the continued use of health apps also confirmed this relationship (Cho, 2016; Hsiao & Chen, 2019; Li et al., 2019; Nascimento et al., 2018; Paré, Leaver, & Bourget, 2018). Given this foundation, this study proposed the following hypothesis:
H1: Fitness app satisfaction is positively related to fitness app continuance intention.
The second independent variable is attaining physical goals. This variable based on the definition of Locke (1981) and is defined as “attaining a specific standard of proficiency on a personal physical task, usually within a specified time limit”. Since the type of goals differ for every mobile health app, it depends on the app what that specific time limit is. This can for example mean that the user manages to attain his goal of number of daily steps or weekly calorie intake. Physical goals for
fitness apps have a repetitive character and are short-term, which also makes goal attainment a variable that is short term. Repetitive attaining fitness app goals should result in health
improvement, which is long-term.
An app user will be satisfied with the fitness app when it helps attaining his or her goals. When the app meets the users’ expectations and the user thus attains his goals, he is likely to positively evaluate his feelings about the fitness app (Cho, 2016). In other words, the better the physical goal set in the fitness app are attained, and expectations of the app are thus confirmed, the more satisfied the user will be (Limayem et al., 2007).
As stated previously, attaining physical goals is a variation of the confirmation variable in the original ECT and PAM. The relationship between attaining physical goals and fitness app satisfaction has not been tested yet in existing literature. However, literature does confirm that the positive relationship between confirmation and satisfaction regarding health apps (Cho, 2016; Hsiao & Chen, 2019; Paré et al., 2018). Next to that, the positive relationship between goal setting and performance on satisfaction has been confirmed multiple times, which can also be somewhat compared to goal attainment (Clay Hamner & Harnett, 1974). Therefore, it is expected that the relationships between attaining physical goals and fitness app satisfaction exists, because it is a confirmation of the goals set. The following hypotheses is stated:
H2: Attaining physical goals is positively related to fitness app satisfaction.
The next independent variable that is included in this study is habit. Prior technology continuance research proposes that much of continued use of technology is habitual (Bhattacherjee & Lin, 2015; Limayem et al., 2007). Habit is defined as “the extent to which people tend to perform behaviors (use the fitness app) automatically because of learning” (Limayem et al., 2007, p. 705). If individuals are using a fitness app for a longer time, which is the case because the physical goals are repetitive and becoming healthier is a long-term process, they will get used to it. When the users incorporate the physical goals in their daily life, it might get very normal to attain the goals every day and become a habit. The cognitive process to use the app will diminish and the apps are used in an automatic manner (Bhattacherjee & Lin, 2015). The satisfaction of the app will also not be evaluated every time but rather take on an automatic character that appears to be reproduced under
subsequent, similar circumstances from habit rather than thought (Limayem et al., 2007). Research on technology continuance intention suggests that habit has a positive influence on continuance intention (Bhattacherjee & Lin, 2015; Limayem et al., 2007). Based on that the following hypothesis is proposed:
2.5.3 Moderator technostress
Because of the ubiquitous character of mobile phones and apps, there is a risk to experience technostress. Stress is usually defined as “a situation in which the demand threatens to exceed the person’s capabilities, in the conditions where he or she expects a substantial reward and cost when a task is met versus not met” (McGrath, 1976, p. 1351). It can have a negative impact on an individual because it creates a bodily or mental tension. Technostress is a particular form of stress and is defined as “an IT user’s experience of stress when using technologies” (Nathan, Tarafdar, Ragu-Nathan, & Tu, 2008). It was developed for work environments, but since technologies have changed in the last decades, it is relevant in other contexts as well.
In the context of fitness apps the original definition is adapted and technostress is defined as: a fitness app user’s experience of stress when using the fitness app. Technostress is usually caused by the inability to cope with constantly changing physical, social, and cognitive responses demanded by a technology such as a fitness app. The smartphone has become a kind of intrusion in everyone’s daily life, and a fitness app demands for goal attainment every time, causing a situation with information overload that is a source of stress (Horwood & Anglim, 2019; Hsiao, Shu, & Huang, 2017). Technostress leads to dissatisfaction with a technology when the expectations, in this case the physical goals, are not met as planned and is therefore a risk for app developers to not retain users. Individuals naturally try to avoid stress by changing their behavior to minimize the negative
consequence. One such a behavior changing strategy is to totally withdraw from the threatening situation, which means discontinuous usage intention (Maier, Laumer, Weinert, & Weitzel, 2015). Technostress has been acknowledged to be a moderator for goal setting and goal attainment in some existing studies (McEwan et al., 2016).
Previous research distinguished technostress in five dimensions: overload, techno-invasion, techno-insecurity, techno-complexity, and techno-uncertainty. In the case of fitness apps, the two dimensions techno-invasion and techno-overload are the most relevant. The other
dimensions are considered less important for the context of this study because fitness apps are not complicated, do not have very frequent updates and are not involved in work related situations (Wang, Shu, & Tu, 2008).
Techno-overload describes a situation where a pressure of too much information is felt because of the technology (Gaudioso, Turel, & Galimberti, 2017). This can happen in the context of fitness apps. If users do not attain their physical goals with the fitness apps, they might get stressed and frustrated and have a dissatisfied feeling about the fitness app, resulting in discontinuous usage of the app. This can however also be the case when they do attain their goals. If goals are attaint, but it takes too much effort, individuals will also not be satisfied with the fitness app, because they experience technostress. This moderation is especially the case for physical goals, since the physical
goals are not a single event, but are long-term and repeated daily or on a less frequent base,
depending on the particular fitness app. If the physical goal is attained or not attained with too much stress multiple times, the user will feel more dissatisfied with the app every time. The fitness app adds a constant stream of information and motivational feedback every day, adding an overload of information repetitively. This leads to a higher level of technostress, no matter if the goal is attained or not attained, and the satisfaction with the fitness app will go down every time the “deadline” for the next goal comes back. The user becomes stressed that he or she will attain the goal the next time but get less satisfied with the app at the same time because of the overload of information. When the technostress is not available the relationship between attaining physical goals and fitness app satisfaction is always positive. Technostress therefore moderates this relationship in the case of techno-overload.
Techno-invasion is a feeling that the technology invades personal life. It the case of fitness apps this is translated to a social pressure that causes technostress. If individuals feel the social pressure of having a healthier looking body, it is possible that they feel like they have to attain the fitness goals every time, even though they are not even satisfied with the fitness app. Individuals also naturally compare their goals to others. They can have a constant feeling of attaining their daily goals to not fall out in the social surrounding and want to achieve harder goals if they are aware that others around them are also achieving harder goals (McEwan et al., 2016). The individual might feel like he is never “free” from the technology, leading to technostress (Wang et al., 2008). This
influences the relationship between goal attainment and fitness app satisfaction. When the users compare their fitness app goals with others and experience a feeling of technostress, the positive relationship between goal attainment and fitness app satisfaction is lower, while it would have been a stronger positive relationship when technostress was not present. Without the social comparison of technostress, goal attainment would result in more satisfaction because goal attainment is proven to be a behavior change technique (McEwan et al., 2016).
Technostress has an influence on the relationship between these two constructs. Techno-overload and techno-invasion moderate the relationship as explained above. Studies researching discontinues usage intention often include technostress, but mixed conclusions are found (Maier et al., 2015). A meta study with the comparable context of researching the effectiveness of goal setting for changing physical activity behavior concluded that situational constraints, like technostress, moderate the relationship of goal setting and physical activity (McEwan et al., 2016). Other studies found no significant influence of technostress on the relationship for continuance intention, because technostress is determined by personality attributes (Hsiao et al., 2017). However it is expected that technostress is a plausible moderator negatively influencing the relationship between attaining physical goals and fitness app satisfaction in the context of fitness apps, because of the
techno-invasion and techno-overload in the repetitive character of fitness apps goals as explained above. Therefore, the following hypothesis is stated:
H4: The positive relationship between attaining physical goals and fitness app satisfaction is negatively moderated by technostress.
2.6 Conceptual model
Based on the expectation-confirmation theory, the post acceptance model was created, which is adapted for the context of this specific study. Four hypotheses were stated based on five main variables and one moderator. Figure 1 gives a visual representation of the definitive conceptual model based on the hypotheses.
Method:
Attaining
physical goals
Fitness app
Continuance
intention
mHealth apps
Fitness app
satisfaction
Figure 1: Conceptual model
Techno stress
eSatisfaction
Habit
Chapter 3 – Methodology
This chapter explains how the data is collected and what methodological choices have been made. First the research design is introduced. Furthermore, the reasoning for the target sample and the procedure of the pilot study are elaborated. The actual survey is explained in detail and the operationalization of the constructs is described. Finally, the explanation of the most suitable data analysis method is given.
3.1 Research design
To test the hypotheses in this studya quantitative research is most suitable. This type of research fits best to research what relationships exist and how strong the effects are. It is an approach to
empirically collect, analyze, and display data in numerical rather than native form (Given, 2008). The data gathering method that is used in this research is a survey. A survey is a good research design to capture emotions, feelings and perceptions of individuals (Vennix, 2016). Besides, a survey is a good tool to get a large and varied population in a relatively short period of time. Qualtrics is used as the software to conduct the survey. This is an online questionnaire tool that can distribute the survey using a weblink. With this research technique it is possible to get data on interval or ratio level, which makes it possible to conduct the analysis that reveals the strength of the relationships.
3.2 Participants
The research population appropriate for this study consists of all individuals that have installed a fitness apps on their phone more than 3 months ago. This way it is possible to try to get a high variation in the dependent variable continuance intention. It is possible that individuals do not want to use the app anymore, but it is also possible that they still have the intention to use the app every day. Using Partial Least Square as an analysis method, a rule of thumb for the minimal sample size is that the sample size should at least be 10 times the maximal number of indicators of a latent variable. This would imply a minimum sample size of 40 respondents in this research. However, a larger sample size increases the reliability and validity and therefore a larger sample was obtained.
Due to the COVID-19 virus that is currently going on, it is not possible to target random individuals on the streets or at sport clubs. Therefore, a convenience sample has been used to attempt to get the most participants. By using a convenience sample, participants are selected on their ease of availability and social surrounding of the researcher (Given, 2008). Participants were invited to take part in the survey with an online link. The researcher sent the link to as much people as possible and also asked the respondents to send the link to other individuals. This way the sample
variance depends on the contacts of the researcher. The link was distributed on WhatsApp, Facebook, and LinkedIn with an explanation that was different based on the platform it was
distributed on. The explanation written for WhatsApp and Facebook are rather informal and the one for LinkedIn more formal. They were all written to politely ask for help to convince individuals to take part in the study. This was done using goodwill and a monetary incentive. Goodwill was used by describing that it will help the researcher to graduate if they took part in the survey. Since almost all the possible respondents personally knew the researcher or got the survey via a friend of the researcher, goodwill was an incentive to make participants feel like it was a moral duty to fill in the survey. Next to that, the introduction text highlighted the fact that 2 gift carts will be randomly given away to two respondents that completed the survey. These are €25 bol.com gift cards, which is external motivator that is useful and well known for almost everyone in the Netherlands. Next to the incentives, the small explanation was written using a clear call to action. If an individual saw the text, they would know in seconds what they had to do. This way, it was made as easy as possible for the respondents to increase the response rate.
Because the study focusses on continuous use of fitness apps, only respondents that indicate to have downloaded a fitness app more than three months ago were asked to fill in the
questionnaire. If individuals indicate that they have not downloaded a fitness app from the category health&fitness on their smartphone more than three months ago, it is explained that they cannot take part in the survey and can close it. Respondents are free in selecting a fitness app: there is a wide variety of fitness apps available and almost all fitness apps have fitness goals and continuance intention is the goal of app developers. This can vary from a running app to a calorie intake fitness app etc. Based on the provided definition of fitness apps, these can all be included.
3.3 Pre-test
To test if the survey has no mistakes and to confirm the full comprehension of the items for the participants, a small-scale pre-test was conducted. This also guarantees that the items are correctly translated to Dutch. First, six individuals evaluated the questionnaire using the plus-minus method. Participants were asked to take the questionnaire and read it out loud. For every question and text parts in the survey they had to add a plus mark in clear items or pieces of text and a minus mark in unclear items or pieces of text. When an item or piece of text was not completely clear after reading it one single time or when the participant stuttered, a minus mark had to be selected (Sienot, 1997). The researcher was also present and noted when a question was difficult to read. At the end of the survey, a short interview was conducted to discuss the pluses and minuses that the participant wrote down. The participant could elaborate on parts that were not clear or should be formulated
be very successful in detecting many different kinds of unclear text parts and other reading problems (Sienot, 1997).
Based on the interviews with the six respondents, some parts of the introduction text and some of the items were rewritten to prevent misunderstanding. After this check, it was possible to start the full-scale survey.
3.4 Survey design
The survey consists of three different parts. It starts with an introduction and guarantee of confidentiality for the participants. An explanation that there are no good or wrong answers is included to avoid socially desirable answers and overthinking. Next to that, the chance to win the monetary incentive is presented. Two €25 bol.com gift cards will be randomly given away to two respondents. The introduction ends with thanking the respondents for their participation.
The participants are then asked to think of a fitness app that they have installed on their mobile phone more than three months ago and keep this app in mind for the remainder of the survey. It is possible that the participant already does not use the app very frequently, but it is also possible that he or she uses the app every day. Individuals are most of the time very satisfied with apps they have downloaded very recently. By only including apps that are downloaded more than three months ago, these first impressions of adoption are taken away and continuance intention can be researched. This way, there is a higher chance of more variation in the dependent variable. According to research agency Loyalistics the retention rate for apps in general after three months is only 29% (Perrow, 2018). Therefore, there is a high chance that the participants do not have a very high continuance intention anymore. The fitness app that the participants choose needs to be an app that falls in the category of “health and fitness-” in the app store. Therefore, in the survey it is
explained that “fitness app” is a very broad category and can include many different apps. The goal of the fitness app is not important, as long as there is one. Some examples of different fitness apps and the goals of those apps are given to clarify what is expected. Examples of fitness apps are for instance a running app or a 7Minuteworkout-app, but it can also be a calorie counter or a sleeping tracker. Examples of the fitness goals can then include running 2 times a week, 10,000 steps a day, 8 hours of sleeping every nights, 2000 calories a day etc. Next, it is explained that if individuals have no fitness app installed on their phone more than three months ago, they cannot take part in the survey and can close the online survey. After this explanation, three general questions about the fitness app are asked, to get some information about what specific app is chosen, how long ago it is
downloaded, and the usage frequency of the app.
Subsequently the questions of the main variables follow. This consists of 20 items which are elaborated on in paragraph 3.6 (Operationalization). The last questions are demographic questions.
These are at the end of the survey, because individuals usually do not like to answer these very personal questions. These questions are about age, gender, and education level. This helps to find out what kind of individuals the group consists of and to see if the group is representative for a larger population. These items are also be used as control variables, to make sure the research is internally valid and not biased.
Finally, the participants are thanked for their time and effort. Participants have the choice to write down their email address if they want to have a chance of winning the giftcard. There is also a box where they can fill in any questions or comments they have about the survey. Last of all, it is explained that if participants want to obtain the results at the end of the study, they can send an email to ask for that. The results are sent to them as soon as they are available. The email address of the researcher is mentioned. In appendix 2 the design of the actual survey can be found.
3.5 Research Ethics
Research ethics are considered in the introduction and end of the survey. In the introduction, participants are assured that the data of the participants is completely anonymous and for the research purposes of this study only. No data that can lead back to the respondent is recorded. Next to that, to avoid social desirable answers, it is clearly assured that there are no good or wrong answers, since the answers are opinions and feelings. Respondents should not think about their answers too long, but just follow their feelings and intuition. The duration of the survey is given to assure participants the survey will not take long.
At the end of the survey, the participants can indicate if they have any questions or comments about the survey and the purpose of the study is briefly explained. The purpose is explained at the end to avoid that participants gave different answers if they knew what the main relationship and potential expected outcome of the study were. The participants can fill in their email address if they want to have a chance of winning the giftcard and are guaranteed that this email address will only be used to contact the winner. It is not obligated to fill in their email address.
Last, the researcher made sure the research was as accurate, honest, and truthful as possible during whole procedure of the research form the beginning till the end. The research integrity form can be found in appendix 11.
3.6 Operationalization
This study used multiple scales for the main study variables. All items that are used in the survey are based on measurement items from existing literature. Most of them were slightly adapted to fit the current context. The original items are all borrowed from English literature and since the participants
are all native Dutch speakers, the items were translated to Dutch. In order to do this correctly, the back translation technique is used. In this technique, the items are translated from English to Dutch, and then back into English. If any discrepancies occur between the two English versions, the
translations are compared and resolved. Since this study involves a different context than the existing studies, most items were not exactly translated but slightly adapted.
In total the five constructs of this study are measured with 20 items. A summary of the operationalization of all the constructs is included in Appendix 1. Most items were measured on a five-point Likert scale, ranging from “totally disagree” (1) to “totally agree” (5). Only the items for the construct fitness app satisfaction are measured on a 5-point semantic-differential scale. In the following paragraphs, the items for all the constructs are elaborated.
The dependent variable, “fitness apps continuance intention”, is measured with four items. Three items were obtained from Hsiao et al. (2016) and one from Nascimento et al. (2018). The items used in those two studies are almost the same. The questions were originally developed for
continuance intention of mobile social apps and smartwatches. The items were adapted to the context of fitness apps for this study.
For fitness app satisfaction a semantic-differential scale is used. This scale is developed by Bhattacherjee (2001) and consists of 4 sets of adjuncts relations to satisfaction with a technology. Participants are asked to report the extent to which they felt like the level of response was most appropriate in their perception. The items all consist of opposites perceptions of the fitness app like very frustrated/very contented. This scale is measured on a five-point scale. The items were originally developed to measure continuance intention for online banking, but many other studies have used and validated these items in different technology contexts.
Attaining physical goals is measured with four items. One item is borrowed from Cho (2016), one item from Huang and Ren (2020), and two items from Bhattacherjee (2001). The items obtained from Bhattacherjee (2001) and Cho (2016) were originally made for measuring confirmation, but can be used, after adaption, for attaining physical goals. Since Attaining physical goals is a construct that has not been used in the context of continued use of fitness apps, this scale has not been validated before. In the study by Cho (2016), the items for confirmation were also for the purpose to find out to which extent the users were hoping to achieve multiple specific purposes, such as health goals, with the fitness apps. Therefore, these items can be used for goal attainment.
Four items for habit are inspired by the scale of Bhattacherjee and Lin (2015). The original items were developed to measure habit in the context of broad IS continuance intention and are slightly adapted for the fitness app context.
The items for technostress are based on items by Tarafdur et al. (2007) and Westerman (2017). Four items are used in this study based on both references. These items were used before to
measure technostress in other studies researching the continuous intention of technologies. Therefore, they are estimated to be applicable in this study as well. Only the items from the two relevant technostress dimensions Techno-overload and techno-invasion are used. The items are slightly adapted to match the fitness apps context.
3.6.1 Control variables
As control variables, age, gender, and educational level were included, because these might have an influence on the fitness app continuance intention. A study researching the mobile health use based on demographic background of users concluded that male, high educated millennials used mobile health apps the most (Bol, Helberger, & Weert, 2018). Next to that, younger individuals are engaged with their phone every hour of the day and have one of the highest fitness app use rates (Statista, 2019b). Therefore, it is expected that millennials have a higher continuance intention and see their phone use as habitual. Higher educated individuals are more likely to use fitness apps than less educated individuals (Bol et al., 2018). This used to be because higher educated individuals had better access to technology, but nowadays technology is available to almost everyone. The specific kind of fitness app is also expected to differ. Higher educated individuals use more mindfulness fitness apps than less educated individuals (Carroll et al., 2017). Based on literature, it is expected that men have higher continuance intention for fitness apps than women. Men have different motives to engage in fitness apps then women. Men are more motivated by strength and
competition, whereas women are more motivated by appearance and weight management (Klenk, Reifegerste, & Renatus, 2017). Therefore, the kind of app is also expected to be different for gender. Men are more likely to use workout apps, while women are more likely to use nutrition apps (Bol et al., 2018). For woman, goal-setting more important, but based on the fact that men use the apps more and competition makes them motivated to attain their goals, it is also expected that men have higher continuance intention for fitness apps (Klenk et al., 2017).
All control variables were transformed to dummy variables, with the dominant group as the reference category; Age was presented as an open question which was dichotomized into two categories for the analysis. These categories are 0=under 25 and 1=25 or above to make both groups almost the same size. Gender was measured on a nominal scale with the categories “female” and “male”. This was dichotomized into the values 0=female and 1=male. Educational level was measured with 5 levels: “primary school”, “high school”, “vocational education”, “university of applied science”, and “University”. This scale was dichotomized into the categories 0=Higher education (University and university of applied science) and 1=Lower education (primary school, high school, and vocational education).
3.7 Analysis design
To analyze the results obtained from the survey, the quantitative analysis programs IBM SPSS Statistics and Adanco are used. First of all, a factor analysis was conducted to be sure the items load on the right constructs. The scale validity and reliability are checked using IBM SPSS Statistics. Then the relationships between the dependent and independent variables are examined by doing Partial least squared structured equation modeling analysis (PLS-SEM) with Adanco. With this analysis technique the results regarding the hypotheses were obtained and explained. PLS a good technique to estimate the relationships in complex models with latent variables (Henseler, Hubona, & Ray, 2016).
Chapter 4 – Data and results
This chapter covers the results of the data analysis. First the sample is described and data of the chosen fitness app is presented. Thereafter, a factor analysis and a reliability analysis are described. Then the results of the two conducted PLS analyses are presented and discussed.
4.1 Sample description
The data was collected in the 20th and 21st week (11-21 may) of 2020. The survey link was opened
298 times. However, the data shows that 130 respondents immediately closed the survey at the introduction text or at the question where they had to fill in their chosen fitness app. This can be due to the fact that they did not have a fitness app installed on their mobile phone more than three months ago. It is also possible that they did not want to fill in the survey, they filled in the survey later, or they had other reasons to abort it. In the survey settings a technical rule was set that an answer was required for every question in order to continue to the next question. Therefore, none of the remaining responses were incomplete. Next to that, none of the respondents were flagged as speeders, filling in the survey in an impossibly quick amount of time. Additionally, none were flagged straightlining, which is answering all the questions the same. Therefore, the quality of the answers was considered good. Eventually, 168 usable native Dutch responses were collected.
The demographics of the sample can be seen in table 1. In terms of gender 80 (47.6%) respondents were male and 88 (52.4%) were female (see table 1). The median of the age of the respondents was at 25, which means that age was highly skewed to the left. With 48.8% of respondents, the age group 18-24 was overrepresented, which can be attributed to the fact that most of researcher’s network consists of students. Even though this could mean that the results are not generalizable for the whole Dutch population, it could still provide valuable insights for the current study. Looking at the educational level, 48.2% of respondents answered the university category and 35.5% university of applied science. The sample thus consisted of relatively highly educated respondents and this is again due to the researcher’s network.
The respondents were asked to choose a fitness app. A very broad variety of fitness apps was recorded, as can be seen in appendix 3. Four fitness apps were chosen most frequently. These were Strava (28), Apple Health (18), Samsung Health (17), and MyFitnessPal (16). To give a summary of the different apps that were chosen, six subcategories were created. The subcategories are
walking/running (60), general activity tracker/standard smartphone or smartwatch health app (52), diet (28), workout/gym (19), meditation/sleep (7), other (2). Most of the apps were in the category
walking/running and general activity tracker/standard smartphone or smartwatch app. The summary of the chosen fitness apps can be seen in table 2.
Looking at how long ago the respondents downloaded the fitness app, most respondents downloaded the fitness app more than 2 years ago (54). 34 respondents downloaded the app
between 3 and 4 months ago, 21 between 4 and 6 months ago, 20 between 6 months and 1 year ago, and 39 between 1 year and 2 years ago (see table 3). This means most apps are already relatively long on the respondents phones. Regarding the usage frequency, the category multiple times a week was chosen 37.5% of the time (63), making it the most chosen category. 41 respondents used the app daily, 47 multiple times a month and 17 less than 1 time a month. Taking these last two statistics together, it can be concluded that even after more than a year, individuals still use the chosen app frequently.
Table 1: Demographic overview
Category Frequency Percentage
Gender Male 80 47.6% Female 88 52.4% Other 0 0% Age <18 1 0.6% 18-24 82 48.8% 25-34 30 17.9% 35-44 6 3.6% 45-54 20 11.9% 55-64 28 16.7% >65 1 0.6%
Educational level Primary school 0 0% High school 13 7.7%
MBO 15 8.9%
HBO 59 35.1%
University 81 48.2%
Table 2: Fitness app categorization
App subcategory Frequency Percentage
Walking/Running 60 35.7%
General activity
tracker/Standard smartphone or smartwatch health app
52 30.9%
Workout/Gym 19 11.3%
Meditation/sleep 7 4.2%
Other 2 1.2%%
Total: 168 100%
Table 3: Time downloaded ago
Time downloaded: Frequency Percentage
Between 3 and 4 months ago 34 20.3% Between 4 and 6 months ago 21 12.5% Between 6 months and 1 year ago 20 11.9% Between 1 year and 2 years ago 39 23.2% More than 2 years ago 54 32.1%
Table 4: Usage frequency
Usage frequency Frequency Percentage
Daily 41 37.5
Multiple times a week 63 37.5 Multiple times a month 47 28.0 Less than 1 time a month 17 10.1
4.2 Discriminant validity and convergent validity
To assess discriminant validity, all items of the five constructs were entered in an exploratory factor analysis to check if the items loaded on the right latent variables (Field, 2013). The KMO-test was 0.851 (exceeding the recommended level of 0.5 to assure item coherence) and Barlett’s test of sphericity was significant (0.000 (x2 (190) = 1544.926, p<0.001), see Appendix 4), thus ensure that the
hypothesis that the original correlation matrix was an identity matrix can be rejected (Field, 2013; Hair, Black, Babin, & Anderson, 2014). Hence, the requirements for factor analysis were met.
Five factors were extracted with an eigenvalues higher than 1, together explaining 64.371% of the variance. All the communalities were above the threshold value of .20 (Field, 2013). However, there were two items that had the lowest value. These were the items Continuance intention 4 (CI4) and Techno stress 4 (TS4). Looking at the pattern matrix (appendix 4), these two items also cross-loaded on two different factors than the intended ones. Satisfaction 2 (SA2) also cross-loaded on two factors but this item still had a much higher loading on the one intended main factor and this could be resolved by first looking at the other two issues. Items should exceed the minimum level of |.30| in the pattern matrix to be correctly interpreted for the factor structure (Field, 2013). Looking at which items CI4 and TS4 were, potential reasons can be given why these two items did not fully loaded on the right construct. CI4 is “I am going to use this fitness app as many times as I do now”. The answer “completely agree” can apply for people that use the app every day as well as for people that never use the app. Therefore, it does not fit well in the construct fitness app continuance intention. The item is excluded for the rest of the analysis. TS4 cross-loaded relatively low on factor 2 as well as factor 5. TS4 is: “I feel my personal life is being invaded by this fitness app”. This item is much more extreme than the other three items measuring technostress. Therefore, it is expected that TS4 differs too much and is deleted for the rest of the analysis.
After exclusion of both CI4 and TS4, a new factor analysis was conducted and no items clustered on multiple factors above a value of |.30| anymore. Five factors were extracted with an eigenvalue above 1 and explained 68.637 of the variance (see Appendix 5). All items now loaded on the construct that they were expected to do. The assumptions for KMO (=.852) and Barlett’s test of sphericity (x2 (153) = 1470.142, p<0.001) were still met. Table 5 shows a summary of the final factor
analysis.
Table 5: Summary final factor analysis Pattern Matrix Factor Communality 1 2 3 4 5 Continuance intention 1 -.850 .845 Continuance intention 2 -.799 .828 Continuance intention 3 -.584 .692
Attaining physical goals 1 -.575 .710
Attaining physical goals 2 -.636 .642
Attaining physical goals 3 -.652 .547
Attaining physical goals 4 -.787 .605
Habit 1 .814 .762
Habit 2 .528 .493
Habit 3 .808 .775
Habit 4 .853 .815
