Using Organismic Integration Theory to Explore the Associations Between Users’ Exercise Motivations and Fitness Technology Feature Set Use1

RESEARCH ARTICLE

USING ORGANISMIC INTEGRATION THEORY TO EXPLORE THE ASSOCIATIONS BETWEEN USERS’ EXERCISE MOTIVATIONS AND FITNESS TECHNOLOGY FEATURE SET USE1

Tabitha L. James, Linda Wallace, and Jason K. Deane Pamplin College of Business, Virginia Tech, 1007 Pamplin Hall, Blacksburg, VA 24061 U.S.A. {[email protected]} {[email protected]} {[email protected]}

Wearable devices and applications (apps) that offer a variety of features intended to support exercisers have flooded the marketplace. Organismic integration theory (OIT) proposes that motivations to exercise can vary along a spectrum of self-determination. To best serve exercisers and assist organizations that are developing and promoting fitness technologies, we need a better understanding of how individuals’ exercise motivations influence their fitness technology feature set use. We also need to determine the impact of fitness technology features on enhancing or undermining wellness outcomes—such as subjective vitality. Our results suggest that almost every subtype of exerciser, where the subtype is defined by OIT motivations toward exercise, has a unique use profile. Our findings also suggest that the social interaction and data management features of current fitness technologies show promise in assisting well-being outcomes, but only for the more self- determined and amotivated subtypes of exercisers. This leads us to suggest that providing every type of exerciser the motivational support that best fits their motivational profile may not be a trivial task, but it ultimately may be necessary for fitness technologies to be universally useful in supporting wellness outcomes.

Keywords: Fitness technology, wearables, health information technology, self-determination theory, organismic integration theory, intrinsic motivation, extrinsic motivation

Introduction 1 agement, and competition. Fitness technologies are one of the most popular examples of technologies supporting the Wearable devices and applications (apps) (e.g., Fitbit, “quantified self” or “personal informatics” (Epstein,Kang et MapMyFitness) that are designed to support fitness are inun- al. 2016; Patel et al. 2015; Rapp and Cena 2016; Shin and dating the technology marketplace. These fitness technol- Biocca 2017), and they aspire to ultimately improve well- ogies provide features that allow users to perform activities being by providing functions that support a user’s motivation such as compiling, analyzing, and monitoring performance to exercise. Fitness technologies generated approximately data; searching for fitness support information; obtaining 238 million dollars in revenue during 2013 (Economist 2014), reminders and rewards; managing goals; gaining access to and projections are for revenues to reach 2.8 billion dollars by coaching; and integrating social sharing, comparison, encour- 2019 (Dolan 2014). Onetime leaders in the fitness tracker market, such as Fitbit, Nike, and Jawbone (Economist 2014), have seen increasing competition from a host of companies 1Sue Brown was the accepting senior editor for this paper. Stacie Petter with new product entries. served as the associate editor. Fitness technologies are novel and emerging, with analysts The appendices for this paper are located in the “Online Supplements” suggesting they are still in the diffusion phase and that the section of MIS Quarterly’s website (https://misq.org).

DOI: 10.25300/MISQ/2019/14128 MIS Quarterly Vol. 43 No. 1, pp. 287-312/March 2019 287 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

next phase of their development will be determining how to competence, and relatedness that are crucial to intrinsic best use collected data to lead to better wellness (IDC 2017). motivation. Specifically, Furthermore, one early report suggested that fitness technologies were being targeted at “fitness fanatics” (Economist The concept of psychological needs provides the 2014). These statements not only suggest different potential basis for describing characteristics of the environ- uses of fitness technologies but also the possibility they may ment that support versus undermine the organism’s appeal differently to different types of exercisers, which could attempts to master or engage each new situation. To ultimately have ramifications for both sales and wellness out- the extent that an aspect of the social context allows comes. If individuals with different motivations toward need fulfillment, it yields engagement, mastery, and exercise do use fitness technologies differently, such insight synthesis; whereas, to the extent that it thwarts need could be instrumental in determining how best to design and fulfillment, it diminishes the individual’s motivation, deploy future generations of such technologies to obtain better growth, integrity, and well-being (Ryan and Deci wellness outcomes. 2002, pp. 8-9).

While fitness technologies are becoming increasingly popular, Hence, environmental or social factors can support or thwart little is known about how individuals use them and how that an individual’s basic psychological needs and thereby en- use may contribute to well-being outcomes. Motivation hance or diminish the impact of the individual’s motivation on theory, specifically self-determination theory (SDT), tells us their activity outcomes (e.g., well-being, mastery of an that individuals have different motivations toward exercise. activity). Stated more precisely, individuals have particular What has not been determined is whether individuals with motivations to perform an activity that may lead to different different exercise motivations use fitness technologies differ- activity outcomes, and environmental factors are variables ently. Hence, our first research question is: How do individ- that are inserted into the environment in which the activities uals’ exercise motivations influence their fitness technology are being performed that may influence that relationship. In feature set use? The ultimate goal in employing fitness tech- the case of fitness technologies, the hope is that their features nologies is to improve the well-being of the user. Information are environmental factors that will support better activity regarding the impact of fitness technology feature set use on outcomes for all subtypes of exercisers. enhancing or undermining wellness outcomes—such as subjective vitality—is also limited. Thus, our second research Much of the early work in SDT focused on rewards as one question is: Does the use of fitness technology feature sets such environmental factor and found that rewards given to individuals for performing an activity thwarted their need for moderate the relationships between users’ motivations toward autonomy, undermining their intrinsic motivation (Deci et al. exercise and subjective vitality? 1999; Deci and Ryan 2012). Another environmental factor, positive feedback, has been suggested to support individuals’ Our study draws from SDT (Deci and Ryan 2012; Ryan and need for competence and thus foster or enhance intrinsic moti- Deci 2000), which offers a high-level picture of human motivation for an activity and improve activity outcomes (Deci et vation and states that there are three primary states of al. 1999; Deci and Ryan 2012). In our study, we suggest that motivation to perform an activity (e.g., exercise): amotiva- the fitness technologies provide a selection of environmental tion, extrinsic motivation, and intrinsic motivation. An factors (e.g., rewards, reminders, directives, feedback, social intrinsically motivated individual performs the activity for the encouragement) from which exercisers can choose to help pure enjoyment of it, an extrinsically motivated individual support their exercise. We will refer to these features as envi- responds to external pressure, and an amotivated individual ronmental motivational supports because they are intended by lacks any motivation. Organismic integration theory (OIT) is the fitness technology makers to be environmental factors to a sub-theory of SDT that delves into the subtypes of extrinsic support exercise and thereby to help improve exercisers’ well- motivation, suggesting that some forms of extrinsic motiva- being. tion can be more self-determined than others. Each subtype of extrinsic motivation presented in OIT is distinguished by The previous discussion of SDT and environmental motiva- the degree to which the desire to perform the activity for the tional supports illustrates a unique aspect of our fitness tech- individual’s own sake is internalized. nology context. SDT suggests that exercisers may have different types of motivation toward exercise, but unlike many SDT also proposes that there are characteristics of the envi- contexts in which SDT has been studied, our exercisers with ronment or social context in which the individual is per- varying motivations can choose their own selection of envi- forming the activity that can be supportive (or unsupportive) ronmental motivational supports. Contrast this scenario with of the individual’s basic psychological needs of autonomy, a classroom context within which a student is performing an

288 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

assignment (i.e., the activity), and the teacher decides if envi- challenges” (Deci and Ryan 2012, p. 427). Following SDT, ronmental motivational supports, such as rewards, threats, or the more self-determined motivations toward exercise should reminders, are a part of the student’s environment. Hence, a lead to higher subjective vitality, and the environmental unique aspect of our context is that the exercisers get to motivational supports may moderate this relationship. That decide which features of fitness technologies they will use to is, the environmental motivational supports may enhance or support their exercise. diminish the relationships between the different motivations and subjective vitality. Thus, our second research question Drawing from work in information systems (IS) on afford- examines the effect of the fitness technology features on ances, Leonardi (2013) argues that people may exhibit differ- enhancing or undermining this wellness outcome for different ent patterns of use of the same technology. His argumentation subtypes of exercisers. Exploring this question provides builds on work by DeSanctis and Poole (1994) that examines insight into which feature sets may successfully support exer- the use of technology at the feature set level and finds varia- cisers’ wellness outcomes—specifically, subjective vitality. tion in feature use among different people. We suggest that These findings are helpful because they provide a partial this is true for different subtypes of exercisers, where the indication as to why fitness technologies may be effective for exerciser subtype is defined by OIT motivations toward some exercisers and not for others as well as which fitness exercise. For example, users that are intrinsically motivated technology features may be most useful for different types of to exercise may especially prefer environmental motivational exercisers. supports that promote competence and allow them to gauge their exercise performance. To intrinsically motivated users, support that attempts to control their activity may be viewed as annoying, unhelpful, or unnecessary and may diminish Model Development their enjoyment. However, users that are extrinsically motivated to exercise may be looking for environmental motiva- Our research is grounded in SDT, which is a meta-theory of tional supports that add additional layers of control to help human motivation that has been an active area of research for sustain their exercise because they do not enjoy it and are several decades and has been separated into several sub- responding to external pressure. theories, including OIT. We provide a brief overview of SDT and the key sub-theories below. Based upon this reasoning, our study examines which subtypes of exercisers use which environmental motivational supports, when given the choice. In answering our first Self-Determination Theory (SDT) research question, we gain an improved understanding of the relationship between users’ motivations to exercise and their At a high level, SDT distinguishes between autonomous (i.e., use of the feature sets offered by leading fitness technologies. self-determined) and controlled (i.e., non–self-determined) Such insight can help fitness technology makers understand motivation (Ryan and Deci 2000). Intrinsic motivation, how to develop their technologies or customize usage sugges- where an individual performs an activity for the pure enjoy- tions for exercisers with different motivations toward exer- ment or satisfaction derived from its performance, is autono- cise. Prior research on exercise motivation has primarily mous or self-determined. Alternatively, extrinsic motivation, looked at the extent to which motivations predict exercise where an individual performs an activity due to perceived behavior (for a review of such research, see Teixeira, Carraça external pressure to do so and in order to attain some separate et al. 2012). Our study is unique in that it examines the outcome (e.g., praise), is controlled or non–self-determined. influence of motivations to exercise on the way fitness technologies are used. OIT is a sub-theory of SDT that resulted from consideration of more autonomous forms of extrinsic motivation and led to Many entities (e.g., parents, physicians, fitness technology a “differentiation of the varied types of internalized extrinsic makers, the exercisers themselves) would ultimately like to motivations” (Deci and Ryan 2012, p. 421). Similar to intrin- see fitness technologies provide environmental motivational sic motivation, Deci and Ryan (2012) suggest that supporting supports that enhance well-being. One measure of well-being the basic psychological needs of autonomy, competence, and that is commonly employed in SDT studies is subjective relatedness are important to facilitating internalized extrinsic vitality, which is defined as “a positive feeling of aliveness motivations, whereas thwarting the basic psychological needs and energy” (Ryan and Frederick 1997, p. 529). Subjective can impair these types of motivations. vitality “results from satisfaction of the basic psychological needs, is an important indicator of health, and provides the Early work on SDT focused on intrinsic motivation and necessary energy for effective self-regulation and coping with considered it to be “an inherent characteristic of human

MIS Quarterly Vol. 43 No.1/March 2019 289 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

beings” that “could be either undermined or enhanced tion, where the stimulus is extrinsic but the individual starts depending on whether the social environment supported or to want to perform the activity to some degree (i.e., inter- thwarted the needs for competence and self-determination” nalizes the behavior) rather than simply complying with the (Deci and Ryan 2012, p. 417). As studies of intrinsic motiva- outside force. tion attempted to explain “increasingly complex experimental phenomena” (Deci and Ryan 2012, p. 418), it became necessary to theorize how factors in the social environment Fitness Technology Feature Sets contributed to the satisfaction of the basic psychological needs and consequently shaped motivational outcomes. OIT provides the foundation to classify exercisers based upon Hence, the SDT sub-theory called the cognitive evaluation their motivation toward exercise. Fitness technologies theory (CET) was proposed “to explain the effects of extrinsic provide features that exercisers can choose to insert into their factors on intrinsic motivation” (Deci and Ryan 2012, p. 418). environment to support their exercise. Our context is novel because the environment is not a static feature in which the The previous discussion illustrates that central to work on activity is taking place or a variable being manipulated by the intrinsic and extrinsic motivations is the idea that to enable researchers (Wilson et al. 2008); rather, the exercisers can the preferred (i.e., more self-determined) types of motivation customize the environment in which their exercise takes that lead to better well-being, the social environment must place. Following the logic of the theory of affordances, we support individuals’ basic psychological needs. SDT eventu- argue that different OIT subtypes of exercisers will choose to ally proposed three basic psychological needs: autonomy, use different fitness technology feature sets. competence, and relatedness. Autonomy “refers to being the perceived origin or source of one’s own behavior” (Ryan and Deci 2002, p. 8). Competence “refers to feeling effective in Theory of Affordances one’s ongoing interactions with the social environment and experiencing opportunities to exercise and express one’s We explore fitness technology use at a “feature level of capacities” (Ryan and Deci 2002, p. 7). Relatedness is analysis” and view fitness technologies as “a collection of defined as “feeling connected to others, to caring for and specific feature sets” (Jasperson et al. 2005, p. 529) that pro- being cared for by those others, to having a sense of belong- vide environmental motivational support. Following previous ingness both with other individuals and with one’s com- literature, we suggest that fitness technology features can pro- munity” (Ryan and Deci 2002, p. 7). Another sub-theory of vide multiple utilities that individuals with different exercise SDT, the basic psychological needs theory, focuses on how motivations can enact (Leonardi 2013). In his work on satisfactions of the basic psychological needs are associated affordances, Leonardi (2013) argues that people may exhibit with well-being outcomes (Deci and Ryan 2012). different patterns of use of the same technology. In addition, Leonardi incorporates the theory of affordances credited to the ecological psychologist Gibson, which proposes that “we Organismic Integration Theory (OIT) perceive in order to operate on the environment, [that] perception is designed for action, [and] called the perceivable possi- While early SDT research concentrated on intrinsic motiva- bilities for action affordances” (Ware 2013, p. 18). Leonardi tion, the OIT sub-theory was developed to explore the idea of (p. 751) describes the concept of affordances as “objects have autonomous extrinsic motivation (Deci and Ryan 2012). OIT properties (or features, in the context of IT use) and animals resulted in the identification of four types of extrinsic moti- that make use of objects have their own physical character- vation (see Figure 1), which are differentiated by the extent to istics and a host of needs” and suggests that “IS researchers which they are internalized (Deci and Ryan 2012). To ex- who adopt a relational view of affordances stress that people’s plore the idea of internalization, it is useful to look at the two goals shape what they come to view the features of the tech- extremes. On one end of the spectrum, an individual’s nology as affording them the ability to do.” Based on SDT behavior can be completely externally regulated; the indi- and the theory of affordances, we propose that users’ specific vidual is simply complying with a demand and thus fully exercise motivations (i.e., characteristics) will shape which of controlled, without any interest in the activity itself or an the feature sets of fitness technologies they ultimately use. internal drive to partake in it. On the other end, an individual can be performing the activity simply for the intrinsic satis- Following DeSanctis and Poole (1994) and Poole and faction that results. This distinguishes pure extrinsic motiva- DeSanctis (1990), many researchers have examined technol- tion (the former) from pure intrinsic motivation (the latter). ogy use at the feature set level instead of the artifact level OIT proposes that these extremes do not paint the entire (e.g., Jasperson et al. 2005; Leonardi 2013; Maruping and picture and that there are actually levels of extrinsic motiva- Magni 2015). Researchers have focused on the range of sys-

290 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Figure 1. Organismic Integration Theory Spectrum of Motivations (R. M. Ryan and E. L. Deci, “Self- Determination Theory and the Facilitation of Intrinsic Motivation, Social Development, and Well-Being,” American Psychologist (55:1), 2000, pp. 68-78; adapted with permission; © American Psychological Association)

tem features that an individual will use when they add tech- We grouped the fitness technology features into three feature nology to a task or take into account when evaluating a sets,2 as shown in Figure 2. technology’s usability (Hoehle and Venkatesh 2015). Focusing on features rather than the technology as a whole is The first feature set, social interaction features (SIFs), is essential for understanding how technology can contribute to made up of features that allow the exerciser to interact with the behavioral outcomes of use (Markus and Silver 2008), and other people. The SIFs enable social interaction for the pur- researchers (Hoehle and Venkatesh 2015; Hong et al. 2013; poses of data sharing, encouragement, competition, compari- Venkatesh et al. 2012) have argued that being considerate of son, and coaching. The second feature set, exercise control contextual differences (e.g., differences in the form and func- features (ECFs), consists of features that are intended to tion of mobile, web, and PC applications) is important control the users’ exercise. ECFs offer rewards and reminders because there is “a general tendency to seek causal explana- that prompt a user to exercise, and goal management allows tions at lower rather than higher levels of analysis, a tactic the user to create and manage the goals (i.e., rule set) the referred to unflatteringly as explanatory reductionism” (Johns fitness technology uses to determine when the rewards and 2006, p. 403). Our exploration of the use of fitness technol- reminders are triggered. The third feature set, data manage- ogies at the feature set level is further justified by these ment features (DMFs), collects and manages the exerciser’s arguments for contextualized consideration of features that data. The DMFs collect, analyze, and feed data from the may provide more granular explorations of the use of tech- device/app to the users regarding their exercise activity as nologies that are highly customizable and differ from each well as provide data from other sources (e.g., maps of running other in function and form. and walking routes). The feature sets were developed to be generic representations of environmental motivational support categories available to exercisers by pairing a wearable, a Categorizing Fitness Technology Features mobile phone, or both with one or more apps.

Fitness technologies typically combine the use of some type of hardware (e.g., a wearable bracelet, token, or mobile phone) with the use of software (e.g., an app provided with 2In developing the feature sets, we first considered whether the environ- the wearable, third-party apps that can be paired with a device mental motivational support is provided solely by the fitness technologies (i.e., ECFs and DMFs) or by a combination of the fitness technologies and the exerciser owns, or both). Not only can a user choose other people (i.e., SIFs). The motivation literature provides a basis to further between multiple form factors for the hardware, but they can classify the features where the environmental motivation support is provided also pair the hardware with multiple software options, and solely by the fitness technology. Prior research suggests threats, deadlines, these software options may cooperate. Because the wearables directives, rewards, and imposed goals are all similar types of environmental motivational supports (Ryan and Deci 2000). The ECFs mimic this type of can be paired with multiple apps, exercisers have a great deal environmental motivational support. Effectance-promoting feedback is part of flexibility in customizing their environment. of a second type of environmental motivational support (Ryan and Deci 2000), which the DMFs most closely resemble.

MIS Quarterly Vol. 43 No.1/March 2019 291 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Figure 2. Fitness Technology Feature Sets

Exercise Motivations and Fitness Technology conversely, how externally regulated—the performance of the Feature Set Selection activity is.

There have been indications in the popular media that dif- “Externally regulated behaviors are dependent on the con- ferent types of exercisers may perceive the utility of or use tinuous presence of the controls” (Ryan and Deci 2006, pp. fitness technologies differently. For example, it has been sug- 1569-1570), suggesting that less self-determined exercisers gested that the fitness technology companies are targeting may see utility in fitness technology features that are more “fitness fanatics” (individuals that are intrinsically motivated controlling because these add continuously present controls toward exercise) (Economist 2014), implying that the fitness to their environments. SDT suggests that more self-deter- technology makers think the technologies may be more attrac- mined exercisers should demonstrate a preference for environ- tive to the intrinsically motivated exerciser. mental motivational supports that are supportive of the basic psychological needs. SDT has also explored the relationships Ultimately, the selection of features implemented as environ- between the exercise motivations proposed in OIT and sub- mental motivational supports is up to the exerciser. CET jective vitality, which is a commonly employed measure of describes how environmental motivational supports relate to well-being. In what follows, we expand this discussion to the psychological needs of individuals. This line of research introduce our hypotheses and research models. suggests that the environmental conditions can be manipulated to support someone’s motivation toward an activity either positively or negatively (where positive support refers Exercise Motivations and Social to environmental conditions conducive to self-determined Interaction Features (SIFs) motivations for an activity through the nourishment of the individual’s basic psychological needs) (Wilson et al. 2008). Fitness technologies have features that incorporate other What fitness technologies are attempting to do, in simplified people into exercisers’ fitness regimes, such as fitness data terms, is to manipulate the environmental conditions for exer- sharing, encouragement, competition, comparison, and cisers in all motivational subtypes. OIT provides a categori- coaching. A straightforward description of the SIFs would be zation of motivations based upon how self-determined—or, that they facilitate external social pressure, which should

292 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

transfer the locus of control to outside the person and reduce nication with a desired social group, but also autonomy- autonomy. However, the use of these features facilitates a thwarting to some degree, because they may facilitate external social connection between a user and their exercise com- pressure to exercise. This leads us to suggest that users who munity. That community could include both people who are are amotivated or extrinsically motivated to exercise will be trying to pressure the user to exercise (e.g., a parent watching more likely to use the SIFs because the external social how many steps their child takes) and people the exerciser is pressure may be viewed as an additional control on their socializing with for entertainment (e.g., sharing data to take exercise. However, we also argue that intrinsically motivated part in a challenge with friends). In the former case, the fit- exercisers will be more likely to use this feature set because ness technology facilitates external social pressure. In the they may derive enjoyment from the social aspect (i.e., their latter case, the fitness technology facilitates socialization that relatedness need is being met to some extent). could satisfy the need for relatedness (i.e., encouraging feelings of belongingness or connectedness with the individ- H1a: Intrinsic regulations for exercise will be positively ual’s exercise group). In fact, socially motivated exercise associated with the use of SIFs. goals have been harder to classify in the exercise SDT litera- H1b: Integrated regulations for exercise will be positively ture (Ryan et al. 1997). Previous literature has suggested that associated with the use of SIFs. social motives for exercise were extrinsic to exercise but that H1c: Identified regulations for exercise will be positively the social interaction may lead to increased enjoyment (Ryan associated with the use of SIFs. et al. 1997) which may enhance intrinsic motivation. We H1d: Introjected regulations for exercise will be posi- suggest that the social features of fitness technologies may be tively associated with the use of SIFs. simultaneously supporting exercise and socialization, and in H1e: External regulations for exercise will be positively doing so, they are at least partially supporting the exerciser’s associated with the use of SIFs. relatedness need. H1f: Nonregulations for exercise will be positively associated with the use of SIFs. Hence, the SIFs can be supportive of the relatedness need of exercisers as well as facilitate the connection between the exerciser and extrinsic social pressures that may reduce Exercise Motivations and Exercise autonomy. Including a social aspect to exercise is not a new Control Features (ECFs) concept, but technology-mediated social support for exercise is somewhat novel. Prior research has provided some support Research has long suggested that rewards decrease individ- for the idea that exercising with another person can lead to uals’ autonomy (i.e., are controlling) and hence are ineffective increased enjoyment and psychological functioning (Plante et means to motivate behavior because they do not support al. 2003; Plante et al. 2011). In addition, previous work has intrinsic motivation for an activity (Deci 1971; Deci et al. found that social motives for exercise may increase enjoyment 1999). Research has also suggested that “threats, deadlines, (Ryan et al. 1997; Wankel 1993), which enhances intrinsic directives, pressured evaluations, and imposed goals diminish motivation and can lead to better adherence to the exercise intrinsic motivation because, like tangible rewards, they con- program (Spink and Carron 1992); however, the social sup- duce toward an external perceived locus of control” (Ryan port features offered by the current generation of fitness tech- and Deci 2000, p. 70). The goal management and reminder nologies are somewhat asynchronous. That is, it is may be features of fitness technologies are deadlines, directives, and before or after exercise that the user can share data, obtain imposed goals, and so, as with rewards, they transfer the locus coaching, set up a competition, compare statistics, or receive of control to the fitness technology. The fitness technology feedback or acknowledgment from another person. For is set up to control the user’s actions, the user feels pressured example, a user can go for a run with the fitness technology by it to complete the activity and reach the goal, and such and then post the length and path of that run to Facebook control reduces the exerciser’s autonomy. through the app. While the technology enables relatedness with an exercise group before or after exercise, the features Reports imply that ECFs may be perceived as controlling by may not truly support the completion of a simultaneous some users. For example, one user stated to a reporter that exercise act. However, it has been suggested that the social using a Fitbit-like device was akin to “having a tiny, over- features of fitness technologies can serve as successful bearing mother always on your person” (Garber 2015, p. 1). environmental motivational support (Kelly 2013). Another fitness technology user stated

In agreement with how social motives for exercise have been My walks were no longer delightful excursions to considered in previous literature, we consider the SIFs to be marvel at nature or delight in old back alleys. They relatedness-supporting, because they may facilitate commu- were forced marches to accumulate steps. The fun

MIS Quarterly Vol. 43 No.1/March 2019 293 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

of it all took a dramatic, stage-left exit and I was left H2e: External regulations for exercise will be positively alone to take ever-more steps (Garber 2015, p. 1). associated with the use of ECFs. H2f: Nonregulations for exercise will be positively asso- In the latter example, the exerciser expresses enjoyment of the ciated with the use of ECFs. walks, and this intrinsically motivated exerciser perceives the ECFs as controlling and expresses disdain for them. Exercise Motivations and Data Intrinsically motivated exercisers enjoy the activity and are Management Features (DMFs) more likely to adhere to their exercise routines (Seghers et al. 2014; Wilson et al. 2008; Wilson et al. 2004) without seeing SDT research has suggested that environmental motivational a need to use a fitness technology to remind or reward them supports “that conduce toward feelings of competence during to exercise. In fact, it has been suggested that the quantified action can enhance intrinsic motivation for that action” and aspects of fitness technologies as they are currently designed that “optimal challenges, effectance-promoting feedback, and may take the fun out of exercising (Garber 2015) by under- freedom from demeaning evaluations were all found to mining the exerciser’s intrinsic motivation (Etkin 2016). The facilitate intrinsic motivation” (Ryan and Deci 2000, p. 70). environmental motivational supports that the ECFs mimic Research has confirmed (Deci et al. 1999) that a social envi- have been found to diminish intrinsic motivation (Ryan and ronment that provides information that promotes competence Deci 2000), and so it is unlikely that users who are intrin- for autonomous activity will meet the competence and auton- sically motivated to exercise would be drawn to such features. omy needs of the individual to some extent, and this will positively support intrinsic motivation (Deci and Ryan 2012). All extrinsic motivations rely at least partially on external Autonomy and competence-supportive environments have pressures to motivate the exercise and are thus more likely to been shown to be important to exercise and wellness out- depend on the constant presence of controls for motivational comes (Hsu et al. 2013; Moustaka et al. 2012; Teixeira, Silva support (Ryan and Deci 2006). We suggest that extrinsically et al. 2012), although studies manipulating the socio- motivated users will be more likely to use the ECFs because contextual environment of exercisers have been limited and these features provide controlling environmental motivational the results somewhat mixed (Wilson et al. 2008). Several fit- supports similar to those in non-technological environments ness technology features support exercisers’ competence and (e.g., rewards, reminders, goal setting). These features mimic autonomy needs; for example, fitness technologies provide or reinforce other external pressures (e.g., a parent reminding effectance-promoting feedback through the collection and a child to exercise), and since non–self-determined users are analysis of exercise data that helps users better understand turning outward for their motivation to exercise, fitness tech- their performance. Features of fitness technologies such as nologies are another instantiation of external pressure and fit exercise progress updates or analysis of a user’s exercise data their exercise motivational profile. For example, one user serve as feedback that promotes competence, and an stated that the “Fitbit’s real secret is to make you more inten- information-searching feature provides the user with auton- tional about parts of your life you’d normally ignore” (John omy in planning, carrying out, and evaluating their exercise 2014). Any exercisers that are classified as integrated, identi- routines. Therefore, data analysis, data collection, progress fied, introjected, externally regulated, or amotivated are likely updates, and information searching are all informational relying on the fitness technologies, which act as continuously supports afforded by fitness technologies that are supportive present controls, for “pressure” to exercise. While the ECFs of exercisers’ competence and autonomy needs. may not be the ideal feature set for externally regulated or amotivated individuals, these features are very likely to be the What distinguishes the DMFs from the other feature sets is ones extrinsically motivated and amotivated exercisers are that they are not instructing users to do something. Rather, drawn toward. these features simply provide information about what users have already accomplished or allow them to look for infor- H2a: Intrinsic regulations for exercise will be negatively mation to support their exercise, which makes these similar in associated with the use of ECFs. nature to effectance-promoting feedback in real life. Thus, H2b: Integrated regulations for exercise will be positively the key difference between the DMFs and the other feature associated with the use of ECFs. sets is that the fitness technologies are not providing or facili- H2c: Identified regulations for exercise will be positively tating external pressure; instead, they are serving as an associated with the use of ECFs. informational source that exercisers can use to gauge compe- H2d: Introjected regulations for exercise will be posi- tence or to help make future exercise decisions. Therefore, tively associated with the use of ECFs. the locus of control is still with the exerciser.

294 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

We would expect intrinsically regulated exercisers to be more pation are not likely to be seen as providing that easy likely to use the DMFs because these features will provide stimulus. them with information to evaluate their competence (i.e., how well they are performing) and help them make decisions about H3a: Intrinsic regulations for exercise will be positively what to do next (e.g., new trails to run, how much more or associated with the use of DMFs. less exercise they would like to do in a week). Self- H3b: Integrated regulations for exercise will be positively determined exercisers should prefer the DMFs because these associated with the use of DMFs. support autonomy and competence and provide environmental H3c: Identified regulations for exercise will be positively motivational support conducive to intrinsic motivation (Ryan associated with the use of DMFs. and Deci 2000). H3d: Introjected regulations for exercise will be negatively associated with the use of DMFs. While we would expect intrinsically motivated exercisers to H3e: External regulations for exercise will be negatively be the most likely users of the DMFs, we also propose that associated with the use of DMFs. individuals who are extrinsically motivated but fall on the H3f: Nonregulations for exercise will be negatively asso- more self-determined end of the scale (i.e., individuals with ciated with the use of DMFs. integrated or identified exercise regulations) will be more likely to use them. Exercisers with integrated regulations, Figure 3 illustrates the model to explore our first research while perhaps still exercising in part because of external question: How do individuals’ exercise motivations influence pressures, feel that exercise or its outcomes are a part of their their fitness technology feature set use? self-concept. Individuals with integrated regulations see value in the exercise itself and view exercise as personally important, and this is the closest of the extrinsic motivations Exercise Motivations, Environmental to intrinsic motivation. Exercisers with identified regulations Motivational Supports, and Well-Being value the benefits of exercising even though they may be exercising because of external pressures, and so they have Subjective vitality assesses a person’s aliveness and energy in only partially integrated exercise into their self-concept. We a physical activity context and is thought to reflect eudae- argue that these more self-determined extrinsically motivated monic well-being (Gunnell et al. 2014; Ryan et al. 2013). exercisers should also prefer the DMFs, which are supportive Eudaemonic well-being, focusing “on meaning and self- of autonomy and competence. realization and defin[ing] well-being in terms of the degree to which a person is fully functioning,” is contrasted with Moving through the spectrum in Figure 1 from the more self- hedonic well-being that “focuses on happiness and defines determined exercise regulations (intrinsic, integrated, and well-being in terms of pleasure attainment and pain avoid- identified) into the more non–self-determined exercise regula- ance” (Ryan and Deci 2001, p. 141). Researchers using OIT tions (introjected and extrinsic) and finally into a state of non- have shown that more self-determined motivational regula- motivation (amotivation), we would expect individuals at the tions are positively associated with well-being (McDonough non–self-determined or amotivated end of the spectrum to be and Crocker 2007). Ryan and Deci (2002) have also sug- less likely to use the DMFs. Exercisers with introjected regu- gested that more self-determined activity leads to greater lations are those who are exercising to avoid guilt or feelings well-being. Furthermore, they describe integrated and identi- of failure. They have integrated exercising into their self- fied regulations as more self-determined and introjected concept sufficiently to feel guilty for failing to do so, but they regulations as more controlling or closest to purely extrinsic. are still exercising because of external pressures. These exer- Thus, we expect the more self-determined regulations (intrin- cisers would be most closely related to purely extrinsically sic, integrated, and identified) to lead to higher levels of motivated exercisers, who are exercising solely because of subjective vitality when exercising. We would similarly external pressure. Amotivated exercisers are not motivated to expect the more non–self-determined regulations (introjected exercise. These individuals simply do not want to exercise or and external) to lead to decreased subjective vitality when do not see the point of exercising. When the exerciser is not exercising. We would not expect amotivation to be associated partaking for the pleasure of the activity, they are less likely with subjective vitality when exercising. to be interested in spending time considering the activity (e.g., analyzing their performance or searching for information on H4a: Intrinsic regulations for exercise will be positively new exercise locations or methods). Similarly, amotivated associated with subjective vitality. exercisers are likely looking for something to help stimulate H4b: Integrated regulations for exercise will be positively their exercise, and DMFs that require user interest or partici- associated with subjective vitality.

MIS Quarterly Vol. 43 No.1/March 2019 295 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Figure 3. Exercise Regulations and Fitness Technology Feature Set Selection

H4c: Identified regulations for exercise will be positively Prior research has indicated that environments supportive of associated with subjective vitality. the basic psychological needs lead to better exercise and H4d: Introjected regulations for exercise will be nega- wellness outcomes (Hsu et al. 2013; Moustaka et al. 2012; tively associated with subjective vitality. Teixeira, Silva et al. 2012). ECFs most closely mimic envi- H4e: External regulations for exercise will be negatively ronmental motivational supports that are said to be controlling associated with subjective vitality. and to thwart the autonomy need. Thus, we expect the ECFs H4f: Nonregulations for exercise will not be associated to diminish the relationships between the exercise motivations with subjective vitality. and subjective vitality because they do not help provide an environment supportive of the basic psychological needs. Subjective vitality has been conceptualized as “a feeling of personal energy associated with agency, which can be H5a–f: Use of the SIFs will positively moderate the relation- diminished by factors that block or hinder autonomy or ships between the exercise motivations and subjec- competence” (Ryan and Frederick 1997, p. 560). We expect tive vitality. the use of feature sets that are supportive of the basic psycho- H6a–f: Use of the ECFs will negatively moderate the logical needs of autonomy, competence, and relatedness to relationships between the exercise motivations and subjective vitality. enhance subjective vitality outcomes. In other words, these H7a–f: Use of the DMFs will positively moderate the feature sets should provide positive environmental motiva- relationships between the exercise motivations and tional support to exercisers. We argue that the SIFs and the subjective vitality. DMFs positively support at least some of the basic psychological needs. Therefore, we expect these feature sets to have Figure 4 illustrates the model to explore our second research a positive effect on the relationships between the exercise question: Does the use of fitness technology feature sets motivations and subjective vitality because they help provide moderate the relationships between users’ motivations toward an environment supportive of the basic psychological needs. exercise and subjective vitality?

296 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Figure 4. Exercise Regulations, Fitness Technology Feature Sets, and Subjective Vitality

Methods and Analysis developed for the current study. The process we used to identify these features and develop the items is described in Scale Development, Data Collection, Appendix A. The subjective vitality scale (SVS) was devel- and Participant Profile oped and tested in previous research by Ryan and Frederick (1997) and Bostic et al. (2000). The original SVS contained We used existing scales to create our survey instrument seven items, but work by Bostic et al. found that the removal whenever possible. The behavioral regulation in exercise of one item increased the scale’s effectiveness, so we used the survey version 3 (BREQ-3) (Markland and Tobin 2004; improved six-item scale. The survey items along with the Wilson et al. 2006) was used to measure the six OIT exercise mean and standard deviation for each item are provided in motivations. The latest version of the BREQ scale, version 3, Appendix B. contains items that measure the exercise motivations for each of the categories in Figure 1 (Wilson et al. 2006). BREQ’s An online survey was developed using the Qualtrics3 online direct mapping to the OIT spectrum seen in Figure 1 provides survey platform provider. Data was collected on Amazon’s “a focus on underlying, source level motives for exercise, Mechanical Turk (mTurk)4 crowdsourcing platform. The represented by the behavioral regulation continuum, rather mTurk platform has been adopted by many behavioral than surface level motives (such as weight control, sociali- researchers because it has been demonstrated to provide a sation, and fitness),” which “may increase our understanding somewhat more diverse demographic sample than using of the way in which perceived self-determination for action influences behavior” (Mullan et al. 1997, p. 751). The BREQ-3 was used without modification because it satisfied 3http://www.qualtrics.com/ our needs and has been very widely tested and used in previous exercise motivation research (Markland and Tobin 4Amazon’s mTurk allows researchers to easily provide a small payment to 2004; Teixeira, Carraça et al. 2012; Wilson et al. 2006). respondents (we paid $0.50 USD) in return for completing an anonymous survey. The use of a randomized survey completion token and an Amazon- assigned worker identification allows for payment without associating the The items to measure fitness technology feature use had to be response with a particular person.

MIS Quarterly Vol. 43 No.1/March 2019 297 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

college students and data that is as reliable (or more so) as contain six reflective exogenous motivations toward exer- traditional methods (Buhrmester et al. 2011; Lowry et al. cising constructs: nonregulation, external regulation, intro- 2016; Mason and Suri 2012). For our purposes, respondents jected regulation, identified regulation, integrated regulation, needed to currently use or have used a fitness technology (a and intrinsic regulation. Subjective vitality when exercising device, an app, or both). A filter question was posed at the is specified as a reflective endogenous construct. beginning of the survey asking respondents to indicate if they met this requirement. Those that were not current or past Three second-order factors represent the fitness technology users of a fitness technology were ejected from the survey. feature sets: SIFs, ECFs, and DMFs. These three second- Those respondents indicating that they were current or past order factors are formative (Petter et al. 2007). Formative users of a fitness technology were asked to specify which second-order factors in PLS are handled using a repeated devices, apps, or both that they use (or had used). The results indicator approach (Lowry and Gaskin 2014) such that first- of this query for the full data collection are given in Table 1. order factors are created that contain the indicators for each of the first-order subconstructs. Those first-order subconstructs The full survey was subject to a pilot test to examine the are used to predict the second-order construct, which contains quality of the new items and confirm the soundness of the all the items for all the first-order subconstructs. When the previously implemented scales. For the pilot test, data was second-order constructs are endogenous, the R-squared values collected on mTurk until 100 usable responses were obtained. for these become 1.0 (i.e., are “swamped out”) because the An exploratory factor analysis was conducted on the pilot first-order factors perfectly predict the second-order factors. data in IBM’s SPSS. The pilot data possessed the desired To overcome this issue, we followed the recommended two- factor structure, and the full data collection was allowed to step approach to test our models (Lowry and Gaskin 2014). proceed. For the full data collection, a total of 1,432 In the first step, the measurement model is tested using the responses were obtained from mTurk. Of those 1,432 repeated indicator approach to obtain latent variable scores. responses, 154 respondents did not complete the survey, In the second step, the latent variable scores are used as the resulting in unusable data that was discarded. Another 212 indicators of the constructs to test the structural models. respondents did not pass the filter question and were elim- inated. There were an additional 186 responses where at least Before we could examine the structural model, the measure- one of the “attention trap” questions was incorrectly ment model needed to be explored to determine convergent answered, which were also discarded. After this data cleaning validity, discriminant validity, and reliability. The statistics process, the resulting data set included 880 responses. reported in Appendix C suggest that all three are acceptable. We also checked for multicollinearity and performed checks Demographic information for the sample is provided in Table to establish a lack of common method bias. The results indi- 2, and some basic device and app demographics are provided cate that multicollinearity and common method bias are not in Tables 1 and 3. Table 1 shows that many of our respon- issues for our model. Detailed results for all tests are pro- dents are Fitbit, Nike Fuelband, or Samsung Gear device vided in Appendix C. The measurement model provides users (past or present) and Fitbit, Nike+ MyFitnessPal, or statistics (path coefficients and p-values) for the formative RunKeeper app users, which is to be expected because these second-order constructs. The path coefficients are shown in devices/apps were the most popular at the time of the data Appendix D, and all are significant at p < 0.001. collection. The fitness technologies included in this study are relatively new to the marketplace, so it was not surprising to The structural models were also tested using SmartPLS find that the length of ownership was typically reported as Version 3.2.1 (Ringle et al. 2015). Our structural models with less than two years (87%). path coefficients, R-squared values, and significance are shown in Figures 5 and 6. A summary of the results can be found in Tables 4, 5, and 6. Model Specification and Testing

We used SmartPLS Version 3.2.1 (Ringle et al. 2015) to test Moderation Testing our model. Partial least squares (PLS) regression is a form of structural equation modeling (SEM) that can be used to test We tested for moderation of the relationships between the multiple relationships between multiple independent and OIT exercise motivations and subjective vitality by the use of dependent variables simultaneously (Anderson and Gerbing the fitness technology feature sets by employing the boot- 1988; Gefen et al. 2000; Lowry and Gaskin 2014). In SEM strapping method described in Vance et al. (2015), Hayes analyses, a two-stage approach, first examining the measure- (2009), and Shrout and Bolger (2002). The results of this ment model and then the structural model (Anderson and analysis are presented in Table 6. We also tested for modera- Gerbing 1988; Kline 2011), is recommended. Our models tion of the relationships between the OIT exercise motivations

298 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Table 1. Current and Past Device and App Usage as Reported by Respondents Device Current Past Device Current Past Device Current Past Adidas miCoach Fit Smart 36 29 Jawbone Up3 11 7 Razer Nabu 6 5 Basis B1 Band 15 14 Jawbone Up Move 15 6 Razer Nabu X 8 7 Basis Carbon Steel 13 10 Jawbone Up24 14 14 Runtastic Orbit 18 13 Basis Peak 16 14 Jaybird Reign 6 7 Samsung Gear 30 23 BodyMedia LINK armband 12 9 LeFun Fit 13 7 Samsung Gear S 21 18 BodyMedia CORE armband 11 10 LG Lifeband Touch 16 11 Samsung Galaxy Gear Fit 41 23 Excelvan MTK 6260 7 8 LifeTrak Brite R450 7 7 Soleus Go 9 4 Fitbit Charge 68 26 LifeTrak Move C300 7 8 Sony SmartBand SWR10 17 8 Fitbit Charge HR 45 14 LifeTrak Core C200 4 8 Sony SmartBand Talk SWR30 20 8 Fitbit Flex 162 56 LifeTrak Zone C410 7 6 Spire 6 6 Fitbit One 80 47 Microsoft Band/Smart Band 12 9 Striiv Band 7 7 Fitbit Surge 13 7 Mio Fuse 5 6 Striiv Play 5 5 Fitbit Zip 30 17 Misfit Flash 10 6 Striiv Touch 9 6 Fitbug Orb 8 8 Misfit Shine 11 4 Striiv Fusion 6 8 Forestfish fitness tracker 7 2 Moov 13 12 SYNC Burn 7 9 Garmin Fenix 3 6 3 Motorola MotoACTV 10 6 SYNC Fit 12 11 Garmin Vivoactive 5 6 Motorola Moto 360 18 16 TomTom Multisport Cardio 9 5 Garmin Vivofit 18 11 Nike+ SportWatch 75 50 TomTom Runner Cardio 13 7 Garmin Vivosmart 16 5 Nike FuelBand 66 64 Wellograph 7 6 Garmin Vivofit2 4 3 Nike FuelBand SE 20 21 Withings Activite 6 10 HiveTech Fitness Tracker 9 10 OUMAX T2 6 5 Withings Activite Pop 5 5 iFit Active 31 26 Pivotal Living Tracker 4 5 Withings Pulse 12 4 Jarv Smart BT watch 5 7 Polar Loop 15 13 Withings Pulse O2 8 4 Jawbone Up 34 31 Polar M400 10 7 Xfit Fitness tracker watch 11 6 Jawbone Up2 9 12 Polar V800 11 4 Xiaomi Mi Band 24 12 Other 56 29 App Current Past App Current Past App Current Past 2Peak 11 10 Garmin Connect 37 27 Run.GPS! 16 19 Addaero 8 11 IFTTT 13 11 RunKeeper 115 91 BodyMedia Fit 27 24 Jetfit Workout 16 12 Runstastic 37 26 Crosscoach 17 19 JogMap 18 19 SparkPeople 21 31 Daily Burn 74 56 Lose it! 51 50 Sportanalytix 5 4 Digifit 18 16 MapMyFitness 100 60 Sportlyzer 10 4 EGYM 16 12 Matchup 9 9 Strava 18 14 Endomondo 30 18 Microsoft HealthVault 20 8 Tactio Health 10 6 Fitbeast 15 13 MY Asics 12 10 The Beautiful Walk 17 8 Fitbit 321 110 MyFitnessPal 169 95 ToBeSport 13 8 Fitboard 13 5 Nike+ 139 109 Trainerize 13 9 Fitstar 36 23 Polar 23 19 Training Peaks 17 14 FitTrend 27 20 Restwise 12 14 Up by Jawbone 36 18 GAIN Fitness 15 15 rubiTrack 8 8 Withings HealthMate 17 5 Other 80 32

MIS Quarterly Vol. 43 No.1/March 2019 299 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Table 2. Sample Demographics (n = 880) Age Distribution Employment Education 18–20 yrs. 35 Employed full time for salary or wages 532 Grade school (k–8 grade) 2 21–25 yrs. 187 Employed part time for salary or wages 121 High school or equivalent (e.g., GED) 61 26–30 yrs. 235 Out of work and looking for work 45 Some college credit, no degree 154 31–35 yrs. 166 Out of work but not current looking for work 5 Trade/technical/vocational training 22 36–40 yrs. 98 A homemaker 58 Associate degree 86 41–50 yrs. 95 Military 6 Bachelor’s degree 380 51–60 yrs. 47 Retired 14 Master’s degree 146 61+ yrs. 17 Unable to work 4 Professional degree 17 A full time student 59 Doctorate degree 12 A full time or part time student also 24 employed part time for salary or wages Other 12 Gender Ethnicity Male 396 White/Caucasian 549 Female 484 Black/African American 59 Asian 208 Pacific Islander 4 Latino 33 Native American Indian 13 Middle-Eastern 5 Other 9

Table 3. Sample Device Demographics (n = 880) Frequency of Use Length of Ownership Device/App Proficiency Non-Exercise Uses Multiple times per day 287 Less than 6 months 228 Novice 171 Nutrition 389 Once per day 236 6 months to 1 year 295 Intermediate 475 Weight Loss 546 Multiple times per week 213 1 to 2 years 245 Advanced 191 Sleep Monitoring 248 Once per week 58 2 to 4 years 83 Expert 43 Other 98 Multiple times per month 28 5 or more years 29 Once per month 17 Less than once per month 41

300 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

***p # 0.001; ** p # 0.10; *p # 0.05, n/s = not significant Figure 5. Exercise Regulations and Fitness Technology Feature Set Selection Results

***p # 0.001; ** p # 0.10; *p # 0.05, n/s = not significant

Figure 6. Exercise Regulations, Fitness Technology Feature Sets, and Subjective Vitality Results

MIS Quarterly Vol. 43 No.1/March 2019 301 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Table 4. Model 1 Detailed Results of Tested Hypotheses and Control Variables Tested Hypothesis/Path Β t-statistic Support Hypotheses (Model 1) H1a. Intrinsic Regulations  Social Interaction Features 0.108 2.092* Yes H1b. Integrated Regulations  Social Interaction Features 0.157 2.826** Yes H1c. Identified Regulations  Social Interaction Features (-0.125) 2.423* No† H1d. Introjected Regulations  Social Interaction Features 0.023 0.564 (n/s) No H1e. External Regulations  Social Interaction Features 0.315 7.082*** Yes H1f. Nonregulations  Social Interaction Features 0.188 4.445*** Yes H2a. Intrinsic Regulations  Exercise Control Features 0.066 1.223 (n/s) No H2b. Integrated Regulations  Exercise Control Features 0.167 2.920** Yes H2c. Identified Regulations  Exercise Control Features (-0.029) 0.476 (n/s) No H2d. Introjected Regulations  Exercise Control Features 0.014 0.311 (n/s) No H2e. External Regulations  Exercise Control Features 0.296 6.478*** Yes H2f. Nonregulations  Exercise Control Features 0.066 1.530 (n/s) No H3a. Intrinsic Regulations  Data Management Features 0.101 1.872 (n/s) No H3b. Integrated Regulations  Data Management Features 0.068 0.967 (n/s) No H3c. Identified Regulations  Data Management Features 0.135 1.961* Yes H3d. Introjected Regulations  Data Management Features (-0.000) 0.006 (n/s) No H3e. External Regulations  Data Management Features 0.064 1.311 (n/s) No H3f. Nonregulations  Data Management Features (-0.137) 3.013** Yes Controls (Model 1) Age  Social Interaction Features (-0.001) 0.021 (n/s) No Age  Exercise Control Features (-0.041) 1.299 (n/s) No Age  Data Management Features (-0.140) 4.177*** Yes Gender  Social Interaction Features (-0.038) 1.416 (n/s) No Gender  Exercise Control Features 0.033 1.094 (n/s) No Gender  Data Management Features 0.046 1.532 (n/s) No Frequency of Use  Social Interaction Features (-0.073) 2.531** Yes Frequency of Use  Exercise Control Features (-0.134) 4.070*** Yes Frequency of Use  Data Management Features (-0.117) 3.179** Yes Length of Ownership  Social Interaction Features 0.046 1.527 (n/s) No Length of Ownership  Exercise Control Features (-0.007) 0.219 (n/s) No Length of Ownership  Data Management Features 0.010 0.289 (n/s) No Device/App Proficiency  Social Interaction Features 0.139 4.247*** Yes Device/App Proficiency  Exercise Control Features 0.061 1.675 (n/s) No Device/App Proficiency  Data Management Features 0.139 4.044*** Yes

*** p # .001; ** p # 0.01, * p # 0.05, n/s = not significant. †We hypothesized a positive relationship between identified regulations and the SIFs.

302 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Table 5. Model 2 Detailed Results of Tested Hypotheses and Control Variables Tested Hypothesis/Path Β t-statistic Support Hypotheses (Model 2) H4a. Intrinsic Regulations  Subjective Vitality 0.369 8.185*** Yes H4b. Integrated Regulations  Subjective Vitality 0.193 3.602*** Yes H4c. Identified Regulations  Subjective Vitality 0.083 1.625 (n/s) No H4d. Introjected Regulations  Subjective Vitality (-0.117) 3.221** Yes H4e. External Regulations  Subjective Vitality (-0.105) 2.552** Yes H4f. Nonregulations  Subjective Vitality 0.126 3.339*** No† Controls (Model 2) Age  Subjective Vitality 0.034 1.227 (n/s) No Gender  Subjective Vitality (-0.035) 1.289 (n/s) No Frequency of Use  Subjective Vitality 0.016 0.510 (n/s) No Length of Ownership  Subjective Vitality (-0.001) 0.030 (n/s) No Device/App Proficiency  Subjective Vitality 0.043 1.312 (n/s) No

***p # 0.001; **p #; *p # 0.05, n/s = not significant. †We hypothesized no significant relationship between nonregulations and subjective vitality.

Table 6. Bootstrapped CI Tests for Moderation

Interaction 2.5% lower bound 97.5% upper bound Zero included Support Moderation of Exercise Motivations  Subjective Vitality by Social Interaction Features H5a. Intrinsic Regulations × Social Interaction Features  Subjective Vitality 0.0794 0.0207 No Yes H5b. Integrated Regulations × Social Interaction Features  Subjective Vitality 0.0509 0.0081 No Yes H5d. Introjected Regulations × Social Interaction Features  Subjective Vitality -0.0041 -0.0320 No Yes H5e. External Regulations × Social Interaction Features  Subjective Vitality -0.0018 -0.0319 No Yes H5f. Nonregulations × Social Interaction Features  Subjective Vitality 0.0309 0.0052 No Yes Moderation of Exercise Motivations  Subjective Vitality by Exercise Control Features H6a. Intrinsic Regulations × Exercise Control Features  Subjective Vitality 0.0411 -0.0200 Yes No H6b. Integrated Regulations × Exercise Control Features  Subjective Vitality 0.0213 -0.0117 Yes No H6d. Introjected Regulations × Exercise Control Features  Subjective Vitality 0.0069 -0.0139 Yes No H6e. External Regulations × Exercise Control Features  Subjective Vitality 0.0055 -0.0144 Yes No H6f. Nonregulations × Exercise Control Features  Subjective Vitality 0.0148 -0.0072 Yes No Moderation of Exercise Motivations  Subjective Vitality by Data Management Features H7a. Intrinsic Regulations × Data Management Features  Subjective Vitality 0.0661 0.0099 No Yes H7b. Integrated Regulations × Data Management Features  Subjective Vitality 0.0414 0.0041 No Yes H7d. Introjected Regulations × Data Management Features  Subjective Vitality -0.0022 -0.0248 No Yes H7e. External Regulations × Data Management Features  Subjective Vitality -0.0010 -0.0246 No Yes H7f. Nonregulations × Data Management Features  Subjective Vitality 0.0279 0.0024 No Yes

Note: We did not run moderation tests for the path between perceived identified regulations and subjective vitality because the primary path was not significant.

MIS Quarterly Vol. 43 No.1/March 2019 303 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

and the use of the feature sets by several of the controls, and between exercisers with intrinsic regulations and either the we report these results in Appendix E. ECFs or the DMFs. The latter reveals that our findings do not support the argument that exercisers with intrinsic regulations The analysis is conducted by first running the SmartPLS would be more likely to use features that would assist them in bootstrap method, which calculates a path coefficient for each gauging their competence and informing their exercise. resample and provides these as outputs, for a large number of Nevertheless, use of both the SIFs and the DMFs were found iterations (we used 5,000). The path coefficients for the vari- to positively moderate the positive relationship between ables of interest for each examined interaction are multiplied intrinsic regulations and subjective vitality, which implies that together, and these values are sorted in descending order for the use of both of these feature sets can help promote positive each interaction. Bounds are calculated using the formulas well-being outcomes for intrinsically motivated exercisers. k(.5-ci/200) for the lower bound and k(.5+ci/200) for the These findings suggest that the use of both the SIFs and the upper bound, where k is the number of resamples and ci is the DMFs may be beneficial, although not necessarily attractive, desired confidence interval (we used a 95% ci). If zero is to users with intrinsic regulations. found between the bounds, the interaction is not significant. Exercisers with integrated regulations are the closest on the spectrum to intrinsically motivated users. These exercisers have integrated the activity into their value and belief system, Discussion but while their exercise is more self-determined, it is still done to obtain separate outcomes they value (e.g., rewards) rather We posed two research questions: than purely for enjoyment (Ryan and Deci 2000). Our findings reveal that these exercisers are more likely to use the (1) How do individuals’ exercise motivations influence their SIFs and the ECFs, as hypothesized. Although we hypothe- fitness technology feature set use? sized that exercisers with integrated regulations would be more likely to use the DMFs, we found no relationship. (2) Does the use of fitness technology feature sets moderate Exercisers with integrated regulations also have a positive the relationships between users’ motivations toward exer- relationship with subjective vitality, and as with the intrinsi- cise and subjective vitality? cally motivated exercisers, the use of the SIFs and the DMFs positively moderated this relationship. Hence, exercisers with Answering the first question determines the fitness technology integrated regulations have a use profile similar to that of use profile for individuals with different motivations toward exercisers with intrinsic motivations, with the addition of the exercise. Answering the second question determines whether use of the more controlling ECFs that we argued would the use of the feature sets enhances or diminishes the known appeal to all the extrinsically motivated exercisers. relationships between exercise motivations and subjective vitality, which enables us to provide insights that may explain Exercisers with identified regulations endorse the activity and why the current generation of fitness technologies are more the value it brings (Ryan and Deci 2002). While they are less effective for some exercisers. An overview of our results for self-determined than exercisers with the previous two regula- each exercise motivation subtype is provided in Table 7. tions, exercise is considered part of these individuals’ identity, and its benefits are recognized (Markland and Tobin 2004; Broadly speaking, we argued that more self-determined Wilson et al. 2006). Our findings reveal that exercisers with exercisers would gravitate toward less controlling and more identified regulations have a very different use profile than informative feature sets (i.e., prefer the feature sets that are the other subtypes (see Table 7). These exercisers are the supportive of at least some basic psychological needs and only subtype to be more likely to use the DMFs, supporting avoid those that thwart these needs). Similarly, we argued hypothesis H3c, but less likely to use the SIFs, contrary to our that the less self-determined and amotivated exercisers would hypothesis. Exercisers with identified regulations were also prefer the more controlling feature sets. Our results indicate found to have no relationship with the ECFs or subjective that the story is more nuanced than this. Table 7 shows that vitality, rather than the positive ones hypothesized. These most exercise motivation subtypes had quite different feature results suggest that exercisers with identified regulations use set use profiles. fitness technologies simply for data analysis and information.

Individuals with intrinsic regulations exercise for the enjoy- Introjected regulations are tied to the exerciser’s self-esteem ment and satisfaction derived from it. Table 7 shows that and feelings of guilt about not exercising (Deci and Ryan exercisers with intrinsic regulations were more likely to use 2012; Gillison et al. 2009). Researchers have suggested that the SIFs, as hypothesized. However, there was no association introjected regulation is “a relatively unstable form of interna-

304 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Table 7. Summary of Results by Exercise Motivation Subtype Exercise Motivation Feature Set Relationship with Feature Set Moderation Subtype Associations Subjective Vitality (Subtype  Subjective Vitality) Intrinsic • SIFs (+) (+) • SIFs (+) Regulations • DMFs (+)

Integrated • SIFs (+) (+) • SIFs (+) Regulations • ECFs (+) • DMFs (+)

Identified • SIFs (-) • None • None Regulations • DMFs (+)

Introjected • None (-) • SIFs (-) Regulations • DMFs (-)

External • SIFs (+) (-) • SIFs (-) Regulations • ECFs (+) • DMFs (-)

Nonregulations • SIFs (+) (+) • SIFs (+) • DMFs (-) • DMFs (+)

lization…in which people adopt an ambient value or practice largely confirm our expectations that extrinsically motivated and are motivated to maintain it, as they ‘should,’ in order to exercisers would gravitate toward features that inserted maintain self-approval or avoid guilt” (Deci and Ryan 2012, additional controls on their exercise. p. 421). We argued that these exercisers would be more likely to use the ECFs and SIFs, which integrate external Amotivated exercisers are simply not motivated in any way to factors into the exercise regime, but less likely to use the exercise. Exercisers with nonregulations typically do not DMFs, which support competence. Table 7 reveals that exer- perform the activity, or do so passively, and may not value the cisers with introjected regulations have no significant rela- activity or the outcomes (Ryan and Deci 2002). Our findings tionships with the use of any of the feature sets. Furthermore, reveal that, as hypothesized, amotivated exercisers are more as expected, introjected regulation had a negative relationship likely to use the SIFs and less likely to use the DMFs (see with subjective vitality, and the use of the SIFs and DMFs Table 7). However, we hypothesized a positive relationship negatively moderated this relationship. Our results suggest between amotivated exercisers and the ECFs that was not sup- that no feature sets of current fitness technologies provide ported. We expected neither the positive relationship between support that is useful to this subtype. nonregulation and subjective vitality nor the positive moderation of this relationship by both the SIFs and DMFs. The Exercisers with external regulations are purely extrinsically results suggest that amotivated exercisers may benefit from motivated. “Externally regulated behaviors are dependent on using some features of fitness technologies. the continuous presence of the controls” (Ryan and Deci 2006, pp. 1569-1570), and therefore, we argued that exercisers with external regulations would be more likely to use Contributions to Research and Theory the feature sets that placed additional controls on their exercise. The ECFs add pressure in the form of beeping and Drawing on the theory of affordances, we examined the use buzzing notifications to prompt the user to stand or walk, and of fitness technologies at a feature set level of analysis, the SIFs integrate others into the users’ exercise regimes, arguing that individuals with different motivations toward creating the possibility of perceived social pressure. The exercise would use fitness technologies differently. The results show that there is a positive relationship between exer- foundation of this argument is supported by prior work cisers with external regulations and the use of both the ECFs exploring how individuals view the utilities of technology, and the SIFs, as hypothesized (see Table 7). No relationship perhaps best summarized by the following statement: “IS between external regulations and the use of the DMFs was researchers who adopt a relational view of affordances stress found. As expected, external regulation had a negative rela- that people’s goals shape what they come to view the features tionship with subjective vitality, and the use of the SIFs and of the technology as affording them the ability to do” (Leon- DMFs negatively moderated this relationship. These results ardi 2013, p. 751). Our results indicate that there are unique

MIS Quarterly Vol. 43 No.1/March 2019 305 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

patterns of use for almost all subtypes of exercisers. Our are performing activities, as well as to further explore the findings that those with different motivations toward exercise ramifications of using such technologies as environmental have unique fitness technology use profiles provide a theo- motivational supports. retical contribution, and the novelty of our study’s context elevates the importance of this discovery. Advocates of SDT have long extolled the virtues of creating “adaptive” environments via providing Fitness technologies are currently the most prominent sociocontextual supports that promote the fulfillment example of the increasingly popular (Lamkin 2017) set of of basic needs, facilitate more self-determined technologies supporting the “quantified self” or “personal regulation of behavior, and contribute to overall informatics” (Epstein, Kang et al. 2016; Patel et al. 2015; feelings of eudaemonic well-being and health Rapp and Cena 2016; Shin and Biocca 2017; Ware 2013). As (Wilson et al. 2008, p. 253). these terms imply, the overarching idea of these technologies is that they serve as environmental motivational supports for Fitness technologies provide users with such an adaptive specific “personal” activities (e.g., exercise, finances, educa- environment, and our study contributes to the motivation tion) to encourage individuals in obtaining desirable outcomes literature by revealing what kinds of environments users with associated with the performance of the activities (e.g., better different motivations toward exercise construct when left wellness, more money, a degree). These technologies offer unguided and how that use contributes to their well-being. support by collecting and analyzing data related to the acti- The implication for researchers is that studies may be limited vities, providing options to socialize or add gamification to by both excluding user characteristics when exploring the use activities in new ways (Liu et al. 2013; Liu et al. 2017), and of highly personalized technologies and ignoring that users offering electronic stimulation (automated notifications) as a can customize their use to construct their own personalized partial substitute for a life coach or assistant (Spence and environments, which necessitates a feature set level of Grant 2007). Motivational theory states that individuals differ analysis. in how they think about or approach an activity or even in their desired outcomes (Deci and Ryan 2012; Ryan and Deci In addition to these broader implications to research and 2000; Ryan et al. 2008). However, early generations of “per- theory, our findings also provide granular insights that may be sonal informatics” technologies are concerned with building helpful to researchers in the study of such personal and a user base (IDC 2017), because these technologies are still customizable technologies. First, the hypothesized positive quite new and quickly evolving (Lamkin 2017) and have not relationship between intrinsic regulations and the DMFs was yet turned to customizing their use (IDC 2017), leaving it to not supported. One implication of this result may be to par- the user to select which features to employ. This one-size- tially explain prior findings that suggested that “by high- fits-all approach may provide a partial explanation for some lighting a quantitative outcome of enjoyable activities, discouraging early findings regarding the use of fitness tech- measurement makes such activities seem more like work, nologies and certain outcomes (e.g., losing weight, enjoy- which undermines intrinsic motivation” (Etkin 2016, p. 980). ment) (Etkin 2016; Jakicic et al. 2016). Research is only Specifically, purely intrinsically motivated exercisers may not beginning to examine how to present the collected informa- need or even want the quantification provided by the fitness tion to users, frame messages, or best add gamification technology for at least two possible reasons: (1) it may take (Epstein, Caraway et al. 2016; Epstein, Kang et al. 2016; Liu the fun out of an activity they enjoy by making it feel like it et al. 2013; Liu et al. 2017; Shin and Biocca 2017). Our study is something they have to do (Etkin 2016) or (2) it may contributes to this emerging area by illustrating that to achieve become boring because the current data analytics mostly the most effective outcomes from the use of such personal inform them of what they already know (Cotgreave 2015; technologies, it may be wise to consider personalizing use to Rapp and Cena 2016). Hence, it is important for future the users’ characteristics. researchers to address such mismatches between the users’ motivational characteristics and the available features or the Our study extends SDT in the exercise context by combining guidance users are given for the technologies. Such mis- it with the theory of affordances to examine (1) how fitness matches could even lead to annoyance or anger at the tech- technology users with different motivations toward exercise nology if not properly managed (Lowry et al. 2015). construct their environments and (2) how the insertion of the fitness technology feature sets into users’ exercise environ- We should caution that our findings do not suggest that exer- ments impacts a wellness outcome. Our results suggest that cisers with intrinsic regulations should not use fitness tech- traditional motivational theories will need to be extended to nologies. In fact, the benefit of our feature-level approach lies accommodate advances in personal informatics technologies in pinpointing which feature sets naturally attracted different that allow users to customize the environments in which they subtypes of exercisers and examining the impact of the feature

306 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

sets on a well-being indicator. The ECFs and DMFs may not with introjected regulations are particularly notable. These appeal to the intrinsically motivated exerciser, but the SIFs individuals are exercising to avoid guilt or to buoy self- did, indicating that socializing or adding gamification may be esteem, and our findings suggest none of the current feature appealing. Research in IS calls for more investigation of sets may appeal to them. Furthermore, studies that examined gamification (Liu et al. 2013; Liu et al. 2017), and fitness the influence of introjected exercise motivations on exercise technologies offer such elements (e.g., challenges and leader- participation found inconsistent results, leading to the sugges- boards). Furthermore, both the use of the SIFs and the DMFs tion that “null or unreliable results from introjection are theo- positively moderated the relationship between intrinsic regu- retically expected within SDT, in which introjection is seen as lations and subjective vitality, which suggests that using both an unstable basis for motivation without positive long-term feature sets may enhance well-being outcomes for intrinsi- utility” (Teixeira, Carraça et al. 2012, p. 20). Prior research cally motivated exercisers. Thus, researchers should explore has also found gender differences in introjected regulations, how socializing or adding gamification may benefit wellness suggesting that self-worth gauged by peer acceptance and outcomes for intrinsically motivated exercisers and how sport may be more critical to males (Gillison et al. 2009). DMFs could be redesigned to be more appealing to this type Exercisers that are motivated by self-esteem issues and guilt of exerciser (Rapp and Cena 2016; Shin and Biocca 2017). may have a mismatch with all the current feature sets, which may reinforce such feelings by encouraging negative social The SIFs were the feature set that four of the six subtypes of comparisons, for example. Hence, research needs to consider exercisers were more likely to use, although those with iden- that personal informatics technologies as currently designed tified regulations were less likely to use them. These results may not be effective for all types of users. Our findings for suggest that carefully designed social and gamification fea- exercisers with introjected regulations reinforce the impor- tures that are mindful not to decrease the exerciser’s need for tance of extending SDT to the study of fitness technologies to competence and autonomy may prove a fruitful avenue to further clarify the motivational characteristics that are central assist the greatest number of subtypes of exercisers. Previous to achieving beneficial outcomes for all users. literature has found that social motives for exercise may increase enjoyment (Ryan et al. 1997; Wankel 1993), perhaps For exercisers with identified regulations, exercise reflects leading to better adherence to the exercise program through values or goals that are important to them but distinct from the satisfaction of the relatedness need (Spink and Carron 1992), pure enjoyment of the activity. An example of the uniqueness which our results indicate may be more attractive to more of identified regulation in the exercise context is given by types of exercisers than any other currently available feature Teixeira, Carraça et al. (2012a, p. 22), referring to Edmunds set. An editorial in the Journal of the American Medical et al. (2006), who Association suggests that fitness technologies need to provide environmental motivational support that helps sustain extrin- suggested that because sustaining a physically active sic motivations, although it advises that this is difficult to lifestyle presumably requires a high degree of effort, accomplish (Patel et al. 2015). The authors conclude that fit- often for mundane or repetitive activities, regulation ness technologies need to be combined with engagement by identification with the outcomes may be more strategies, which carefully designed socialization and gamifi- important than exercising for fun and enjoyment. cation could provide. However, our findings were not universally positive with regard to all exercise motivation subtypes Thus, one explanation for our findings for this subtype may be and the use of the SIFs. It may be that some of the extrin- that the outcomes associated with exercise are valued over sically motivated exercisers see the SIFs as a way for people enjoyment. Individuals with identified regulations may feel to view and further control their fitness progress rather than that health and well-being are important objectives and use as a source of enjoyable social interaction. Research needs to the DMFs to measure progress toward these goals. This result delve into perceptual differences of technology-mediated may actually illuminate an option for new features that inte- socialization to inform customized designs that provide bene- grate exercise goals (e.g., appearance, health, social, enjoy- fits to all types of exercisers. ment) (Etkin 2016; Ryan et al. 1997; Ryan et al. 2008) in more depth rather than relying primarily on simpler quanti- Our findings also provide insights for each subtype of exer- fication. This discussion also calls attention to the need for ciser, broadly suggesting that research needs to build on both researchers to extend our work to examine other motivational SDT and our work to determine what guidance can be pro- characteristics (Deci and Ryan 2012; Ryan et al. 2008), such vided to users with different motivational characteristics on as goal content, in relation to the use of fitness and other per- how to use the technologies as well as to inform technology sonal informatics technologies. Research is only beginning creators regarding how to design new features that are better to explore the personal and societal ramifications of personal for different types of exercisers. Our results for exercisers informatics technologies. Our study used only one sub-theory

MIS Quarterly Vol. 43 No.1/March 2019 307 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

of SDT; there is much to be gained by further integration of to use this feature set. There are some indications in the liter- SDT into the study of personal informatics technologies. ature that the DMFs may become boring after a while as exercisers become familiar with their normal levels of daily activity and that they may be discouraging to those whose use Implications for Society and Practice of fitness technologies has lapsed (Epstein, Caraway et al. 2016; Epstein, Kang et al. 2016; Rapp and Cena 2016; Shin Our results provide insights into the use and design of the and Biocca 2017). Changes in how the data is displayed or to current generation of fitness technologies that may be useful the framing of messages may help improve the DMFs to make to both users and developers. The ECFs are arguably the them more effective (Epstein, Caraway et al. 2016; Epstein, most prominent features of the current generation of fitness Kang et al. 2016; Rapp and Cena 2016; Shin and Biocca technologies. The ECFs are often the default in fitness tech- 2017). Fitness technology makers may be well served to im- nologies to, for example, send exercisers a reward badge prove the DMFs, but users could also explore changes in how when they exceed a certain number of steps or to remind them they use their own fitness data. For example, adding friends to stand, breathe, or take more steps. It is not especially and using a leaderboard to view their step counts compared to promising that our results revealed only two of the extrinsic others would be a way to add gamification to the data analy- motivation subtypes, integrated and external, were more likely tics. The SIFs appealed to the greatest number of subtypes of to use the ECFs. Moreover, motivation theory has long sug- exercisers (intrinsic regulations, integrated regulations, exter- gested that rewards, imposed goals, deadlines, and directives nal regulations, and nonregulations) and also positively all steer individuals’ focus toward external controls, dimin- moderated the relationships between the intrinsic, integrated, ishing intrinsic motivation (Ryan and Deci 2000). Further- and nonregulations and subjective vitality. This suggests that more, our findings show that the ECFs had no moderating the SIFs may be a very promising feature set for both users impact on the relationships between the exercise motivation and makers to explore. Furthermore, gamification elements subtypes and subjective vitality. The theory and our results such as challenges and leaderboards are often a part of the both suggest that the ECFs, as currently designed, may not be SIFs, which provides another valuable option for engagement the best feature set for most exercisers. (Liu et al. 2013; Liu et al. 2017; Patel et al. 2015).

One suggestion for fitness technology users would be to spend time exploring the other feature sets that may offer Limitations and Future Research more effective environmental motivational support. The SIFs and the DMFs are becoming increasingly sophisticated as Our study was limited because our data collection was con- fitness technologies mature. These features often require ducted at a single point in time. We could only observe more setup, learning, and customization from the user than the associations between current exercise motivations and current ECFs, but our results suggest this additional time and effort fitness technology feature set use. Future research should may be worth it. Given our findings for the ECFs and that too include longitudinal studies, which could capture and confirm many notifications may overwhelm or annoy users, fitness changes in the use of fitness technologies, users’ perceptions technology makers may want to carefully consider how they regarding fitness technologies, and users’ exercise behaviors. use notifications. It may be imprudent to send rewards or For example, mismatches between motivational characteris- reminders that may not be as effective, when notifications tics and fitness technology features could lead to continuance could be used to provide feedback supportive of the basic issues. Hence, future longitudinal studies could explore the psychological needs through carefully crafted messaging from influence of fitness technology feature sets on exercise the DMFs and SIFs that may lead to better outcomes for many adherence, regimen changes, and habit as well as investigate users. Furthermore, gamification elements should be gran- discontinuance among users with different motivational char- ularly investigated in future research, because some may acteristics. Such studies could inform design approaches; function more as rewards (e.g., badges) where others may be early research on discontinuance and design is already found to be more supportive of the competence or relatedness yielding interesting results (Epstein, Caraway et al. 2016; needs (e.g., leaderboards, challenges). Research has begun to Epstein, Kang et al. 2016). explore messaging and customizing analytics (Epstein, Cara- way et al. 2016; Epstein, Kang et al. 2016; Rapp and Cena Our study was limited to the feature sets of fitness technol- 2016; Shin and Biocca 2017), but there is much work ogies at the time of data collection. These features are likely remaining for both researchers and fitness technology makers. to expand substantially with further maturation of the product Furthermore, while our results for the DMFs showed more lines, and future research may benefit from a broader set of promise in assisting well-being outcomes for some exercisers, features to examine. As the number and sophistication of the only exercisers with identified regulations were more likely features of fitness technologies expands, future research may

308 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

also need to employ more formal methods of developing design and use may not be the most effective. Our findings feature sets (e.g., a card sort procedure). In addition, the form reveal that providing every type of exerciser with the of fitness technologies is changing rapidly. Technology to motivational support that best fits their motivational profile measure and collect exercise and health indicators, as well as may not be a straightforward task, but it ultimately may be nutrition and sleep data, is being integrated into a wide variety required for fitness technologies to be universally useful in of items (e.g., mobile phones, smart watches, bands, tokens, supporting wellness outcomes. shoes). We did not restrict our respondents to the use of a particular combination of devices and apps because our interests lie in determining what features they would imple- References ment if the choice were left up to them. However, different technologies have different characteristics. For example, Anderson, J. C., and Gerbing, D. W. 1988. “Structural Equation perceptions of usefulness, ease of use, privacy, and trust could Modeling in Practice: A Review and Recommended Two-Step all vary for different devices and affect their use and the well- Approach,” Psychological Bulletin (103:3), pp. 411-423. ness outcomes obtained thereby. Bostic, T. J., Rubio, D. M., and Hood, M. 2000. “A Validation of the Subjective Vitality Scale Using Structural Equation Future research may also want to delve into other antecedents Modeling,” Social Indicators Research (52:3), pp. 313-324. Buhrmester, M., Kwang, T., and Gosling, S. D. 2011. “Amazon’s of use, such as trust in the accuracy of the device, which has Mechanical Turk a New Source of Inexpensive, Yet High- recently been highlighted in the media (McPhate 2016). Quality, Data?,” Perspectives on Psychological Science (6:1), pp. Another interesting antecedent to different wellness outcomes 3-5. or to use may be to explore motivation to use fitness technol- Cotgreave, A. 2015. “Why I Ditched My Fitbit and What This ogies (e.g., Ke et al. 2013; Malhotra et al. 2008) instead of or Means for Analytics,” InfoWorld (http://www.infoworld.com/ in addition to the exercise motivations. Future research could rticle/2951751/personal-technology/why-i-ditched-my-fitbit-and- also explore an analysis of use that is more granular; for what-this-means-for-analytics.html). example, frequency of feature use (i.e., adapting our scales to Deci, E. L. 1971. “Effects of Externally Mediated Rewards on obtain frequency of use information rather than merely use), Intrinsic Motivation,” Journal of Personality and Social Psych- manipulations of the social environment, consideration of ology (18:1), pp. 105-115. various social psychological constructs in models of motiva- Deci, E. L., Koestner, R., and Ryan, R. M. 1999. “A Meta-Analytic tion and fitness technology use, or implementation choices Review of Experiments Examining the Effects of Extrinsic (i.e., form factors, combinations of apps and devices). Rewards on Intrinsic Motivation,” Psychological Bulletin (125:6), pp. 627-668. Deci, E. L., and Ryan, R. M. 2012. “Self-Determination Theory,” in Handbook of Theories of Social Psychology, P. A. M. Van Lange, A. W. Kruglanski, and E. T. Higgins (eds.), London: Conclusion Sage Publishers Ltd. DeSanctis, G., and Poole, M. S. 1994. “Capturing the Complexity Our study, grounded in SDT, provides a fine-grained exam- in Advanced Technology Use: Adaptive Structuration Theory,” ination of the associations between exercise motivations and Organization Science (5:2), pp. 121-147. fitness technology feature set use. Our results suggest that Dolan, B. 2014. “Prediction: Fitness Tracker Market to Hit $2.2 almost every subtype of exerciser has a unique use profile. In Billion This Year,” MobiHealthNews, May 19 (http:// addition, we examine the impact of fitness technology feature mobihealthnews.com/33256/prediction-fitness-tracker-market-to- set use on the relationships between users’ exercise motiva- hit-2-2-billion-this-year/). tions and subjective vitality. Our findings suggest that two of Economist. 2014. “Shaking Up the Wearables,” The Economist, the feature sets (SIFs and DMFs) show promise in enhancing Business and Finance Section, August 26 (http://www.economist. well-being outcomes, but only for the more self-determined com/news/business-and-finance/21613925-potential-market- and amotivated subtypes of exercisers. personal-fitness-tracking-devices-over-hyped-shedding- wearables). Fitness technologies are still relatively new to the market- Edmunds, J., Ntoumanis, N., and Duda, J. L. 2006. “A Test of Self Determination Theory in the Exercise Domain,” Journal of place, and little is known regarding their ultimate effective- Applied Social Psychology (36:9), pp. 2240-2265. ness. Some recent research even suggests they may not Epstein, D. A., Caraway, M., Johnston, C., Ping, A., Fogarty, J., and support other wellness outcomes, such as weight loss (Jakicic Munson, S. A. 2016. “Beyond Abandonment to Next Steps: et al. 2016). Our findings regarding subjective vitality Understanding and Designing for Life after Personal Informatics indicate that there is some promise to the technologies as Tool Use,” in Proceedings of the 2016 CHI Conference on environmental motivational supports, but our study also pro- Human Factors in Computing Systems, New York: ACM, pp. vides the insight that a one-size-fits-all approach to their 1109-1113.

MIS Quarterly Vol. 43 No.1/March 2019 309 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Epstein, D. A., Kang, J. H., Pina, L. R., Fogarty, J., and Munson, Johns, G. 2006. “The Essential Impact of Context on Organi- S. A. 2016. “Reconsidering the Device in the Drawer: Lapses zational Behavior,” Academy of Management Review (31:2), pp. as a Design Opportunity in Personal Informatics,” in Proceedings 386-408. of the 2016 ACM International Joint Conference on Pervasive Ke, W., Tan, C.-H., Sia, C.-L., and Wei, K.-K. 2013. “Inducing and Ubiquitous Computing, Heidelberg, Germany, pp. 829-840. Intrinsic Motivation to Explore the Enterprise System: The Etkin, J. 2016. “The Hidden Cost of Personal Quantification,” Supremacy of Organizational Levers,” Journal of Management Journal of Consumer Research (42:6), pp. 967-984. Information Systems (29:3), pp. 257-289. Garber, M. 2015. “The Ennui of the Fitbit,” The Atlantic, Kelly, H. 2013. “Fitness Tools that Tap the Power of Your Technology Section, July 10 (http://www.theatlantic.com/ Friends,” CNN Business, June 3 (http://www.cnn.com/2013/ technology/archive/2015/07/the-ennui-of-the-fitbit/398129/). 06/03/ tech/mobile/fitness-gadgets-motivation/). Gefen, D., Straub, D., and Boudreau, M.-C. 2000. “Structural Kline, R. B. 2011. Principles and Practice of Structural Equation Equation Modeling and Regression: Guidelines for Research Modeling (3rd ed.), New York: The Guilford Press. Practice,” Communications of the Association for Information Lamkin, P. 2017. “Fitbit’s Dominance Diminishes but Wearable Systems (4:1), pp. 2-77. Tech Market Bigger Than Ever,” Forbes, March 3 (https://www. Gillison, F., Osborn, M., Standage, M., and Skevington, S. 2009. forbes.com/sites/paullamkin/2017/03/03/fitbits-dominance- “Exploring the Experience of Introjected Regulation for Exercise diminishes-but-wearable-tech-market-bigger-than-ever/#1bb65 across Gender in Adolescence,” Psychology of Sport and Exer- 69d7f4d). cise (10:3), pp. 309-319. Leonardi, P. M. 2013. “When Does Technology Use Enable Gunnell, K. E., Crocker, P. R., Mack, D. E., Wilson, P. M., and Network Change in Organizations? A Comparative Study of Zumbo, B. D. 2014. “Goal Contents, Motivation, Psychological Feature Use and Shared Affordances,” MIS Quarterly (37:3), pp. Need Satisfaction, Well-Being and Physical Activity: A Test of 749-775. Self-Determination Theory over 6 Months,” Psychology of Sport Liu, D., Li, X., and Santhanam, R. 2013. “Digital Games and and Exercise (15:1), pp. 19-29. Beyond: What Happens When Players Compete,” MIS Quarterly Hayes, A. F. 2009. “Beyond Baron and Kenny: Statistical Media- (37:1), pp. 111-124. tion Analysis in the New Millennium,” Communication Mono- Liu, D., Santhanam, R., and Webster, J. 2017. “Toward Meaning- graphs (76:4), pp. 408-420. ful Engagement: A Framework for Design and Research of Hoehle, H., and Venkatesh, V. 2015. “Mobile Application Gamified Information Systems,” MIS Quarterly (41:4), pp. Usability: Conceptualization and Instrument Development,” MIS 1011-1034. Quarterly (39:2), pp. 435-472. Lowry, P. B., D’Arcy, J., Hammer, B., and Moody, G. D. 2016. Hong, W., Chan, F. K., Thong, J. Y., Chasalow, L. C., and “‘Cargo Cult’ Science in Traditional Organization and Informa- Dhillon, G. 2013. “A Framework and Guidelines for Context- tion Systems Survey Research: A Case for Using Nontraditional Specific Theorizing in Information Systems Research,” Informa- Methods of Data Collection, Including Mechanical Turk and tion Systems Research (25:1), pp. 111-136. Online Panels,” Journal of Strategic Information Systems (25:3), Hsu, Y.-T., Buckworth, J., Focht, B. C., and O’Connell, A. A. 2013. pp. 232-240. “Feasibility of a Self-Determination Theory-Based Exercise Lowry, P. B., and Gaskin, J. 2014. “Partial Least Squares (PLS) Intervention Promoting Healthy at Every Size with Sedentary Structural Equation Modeling (SEM) for Building and Testing Overweight Women: Project Change,” Psychology of Sport and Exercise (14:2), pp. 283-292. Behavioral Causal Theory: When to Choose It and How to Use IDC. 2017. “Xiaomi and Apple Tie for the Top Position as the It,” IEEE Transactions on Professional Communication (57:2), Wearables Market Swells 17.9% During the First Quarter, pp. 123-146. According to IDC,” June 5, International Data Corporation, Lowry, P. B., Gaskin, J., and Moody, G. D. 2015. “Proposing the Framingham, MA (https://www.idc.com/getdoc.jsp?containerId= Multi-Motive Information Systems Continuance Model (MISC) prUS42707517). to Better Explain End-User System Evaluations and Continuance Jakicic, J. M., Davis, K. K., Rogers, R. J., King, W. C., Marcus, Intentions,” Journal of the Association for Information Systems M. D., Helsel, D., Rickman, A. D., Wahed, A. S., and Belle, (16:7), pp. 515-579. S. H. 2016. “Effect of Wearable Technology Combined with a Malhotra, Y., Galletta, D. F., and Kirsch, L. J. 2008. “How Endog- Lifestyle Intervention on Long-Term Weight Loss: The Idea enous Motivations Influence User Intentions: Beyond the Randomized Clinical Trial,” JAMA (316:11), pp. 1161-1171. Dichotomy of Extrinsic and Intrinsic User Motivations,” Journal Jasperson, J. S., Carter, P. E., and Zmud, R. W. 2005. “A Compre- of Management Information Systems (25:1), pp. 267-299. hensive Conceptualization of Post-Adoptive Behaviors Asso- Markland, D., and Tobin, V. 2004. “A Modification of the Behav- ciated with Information Technology Enabled Work Systems,” ioral Regulation in Exercise Questionnaire to Include an Assess- MIS Quarterly (29:3), pp. 525-557. ment of Amotivation,” Journal of Sport and Exercise Psychology John, E. 2014. “Fitbit Flex Wristband Review: A Weight Loss (26:2), pp. 191-196. Tracker That Makes Fitness a Habit,” The Guardian, U.S. Markus, M. L., and Silver, M. S. 2008. “A Foundation for the Edition, Technology Section, July 14 (https://www.theguardian. Study of It Effects: A New Look at Desanctis and Poole’s Con- com/technology/2014/jul/14/fitbit-flex-wristband-review-weight- cepts of Structural Features and Spirit,” Journal of the Associa- loss-tracker-fitness-habit). tion for Information Systems (9:10), pp. 609-632.

310 MIS Quarterly Vol. 43 No. 1/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Maruping, L. M., and Magni, M. 2015. “Motivating Employees to Ryan, R. M., and Deci, E. L. 2002. “Overview of Self- Explore Collaboration Technology in Team Contexts,” MIS Determination Theory: An Organismic Dialectical Perspective,” Quarterly (39:1), pp. 1-16. in Handbook of Self-Determination Research, E. L. Deci, and Mason, W., and Suri, S. 2012. “Conducting Behavioral Research R. M. Ryan (eds.), Rochester, NY: The University of Rochester on Amazon’s Mechanical Turk,” Behavior Research Methods Press, pp. 3-33. (44:1), pp. 1-23. Ryan, R. M., and Deci, E. L. 2006. “Self Regulation and the McDonough, M. H., and Crocker, P. R. E. 2007. “Testing Self- Problem of Human Autonomy: Does Psychology Need Choice, Determined Motivation as a Mediator of the Relationship Self Determination, and Will?,” Journal of Personality (74:6), Between Psychological Needs and Affective and Behavioral Out- pp. 1557-1586. comes,” Journal of Sport & Exercise Psychology (29), pp. Ryan, R. M., and Frederick, C. M. 1997. “On Energy, Personality, 645-663. and Health: Subjective Vitality as a Dynamic Reflection of Well McPhate, M. 2016. “Just How Accurate Are Fitbits? The Jury Is Being,” Journal of Personality (65:3), pp. 529-565. Out,” The New York Times, Technology Section, May 25 (http:// Ryan, R. M., Frederick, C. M., Lepes, D., Rubio, N., and Sheldon, www.nytimes.com/2016/05/26/technology/personaltech/fitbit- K. M. 1997. “Intrinsic Motivation and Exercise Adherence,” accuracy.html?_r=0). International Journal of Sport Psychology (28:4), pp. 335-354. Moustaka, F. C., Vlachopoulos, S. P., Kabitsis, C., and Theo- Ryan, R. M., Huta, V., and Deci, E. L. 2013. “Living Well: A dorakis, Y. 2012. “Effects of an Autonomy-Supportive Exercise Self-Determination Theory Perspective on Eudaimonia,” in The Instructing Style on Exercise Motivation, Psychological Well- Exploration of Happiness: Present and Future Perspectives, A. Being, and Exercise Attendance in Middle-Age Women,” Delle Fave (ed.), Dordrecht, The Netherlands: Springer, pp. Journal of Physical Activity and Health (9:1), pp. 138-150. 117-139. Mullan, E., Markland, D., and Ingledew, D. K. 1997. “A Graded Ryan, R. M., Patrick, H., Deci, E. L., and Williams, G. C. 2008. Conceptualisation of Self-Determination in the Regulation of “Facilitating Health Behaviour Change and its Maintenance: Exercise Behaviour: Development of a Measure Using Confir- Interventions Based on Self-Determination Theory,” European matory Factor Analytic Procedures,” Personality and Individual Health Psychologist (10:1), pp. 2-5. Differences (23:5), pp. 745-752. Seghers, J., Vissers, N., Rutten, C., Decroos, S., and Boen, F. 2014. Patel, M. S., Asch, D. A., and Volpp, K. G. 2015. “Wearable “Intrinsic Goals for Leisure-Time Physical Activity Predict Devices as Facilitators, Not Drivers, of Health Behavior Children’s Daily Step Counts through Autonomous Motivation,” Change,” The Journal of the American Medical Association Psychology of Sport and Exercise (15:3), pp. 247-254. (313:5), pp. 459-460. Shin, D.-H., and Biocca, F. 2017. “Health Experience Model of Petter, S., Straub, D., and Rai, A. 2007. “Specifying Formative Personal Informatics: The Case of a Quantified Self,” Computers Constructs in Information Systems Research,” MIS Quarterly in Human Behavior (69), pp. 62-74. (31:4), pp. 623-656. Shrout, P. E., and Bolger, N. 2002. “Mediation in Experimental Plante, T., Bogdan, R., Kanani, Z., Babula, M., Ferlic, E., and and Nonexperimental Studies: New Procedures and Recommen- MacAskill, E. 2003. “Psychological Benefits of Exercising with dations,” Psychological Methods (7:4), pp. 422-445. Another,” Journal of Human Movement Studies (44:2), pp. Spence, G. B., and Grant, A. M. 2007. “Professional and Peer Life 93-106. Coaching and the Enhancement of Goal Striving and Well-Being: Plante, T. G., Gustafson, C., Brecht, C., Imberi, J., and Sanchez, J. An Exploratory Study,” The Journal of Positive Psychology 2011. “Exercising with an Ipod, Friend, or Neither: Which Is (2:3), pp. 185-194. Better for Psychological Benefits?,” American Journal of Health Spink, K. S., and Carron, A. V. 1992. “Group Cohesion and Adher- Behavior (35:2), pp. 199-208. ence in Exercise Classes,” Journal of Sport & Exercise Poole, M. S., and DeSanctis, G. 1990. “Understanding the Use of Psychology (14:1), pp. 78-86. Group Decision Support Systems: The Theory of Adaptive Teixeira, P. J., Carraça, E. V., Markland, D., Silva, M. N., and Structuration,” in Organizations and Communication Technol- Ryan, R. M. 2012. “Exercise, Physical Activity, and Self- ogy, J. L. Fulk and C. W. Steinfield (eds.), Newbury Park, CA: Determination Theory: A Systematic Review,” International Sage Publications, pp. 173-193. Journal of Behavioral Nutrition and Physical Activity (9:78), pp. Rapp, A., and Cena, F. 2016. “Personal Informatics for Everyday 1-30. Life: How Users Without Prior Self-Tracking Experience Teixeira, P. J., Silva, M. N., Mata, J., Palmeira, A. L., and Engage with Personal Data,” International Journal of Human– Markland, D. 2012. “Motivation, Self-Determination, and Long- Computer Studies (94), pp. 1-17. Term Weight Control,” International Journal of Behavioral Ringle, C. M., Wende, S., and Becker, J.-M. 2015. “SmartPLS 2,” Nutrition and Physical Activity (9:1), pp. 1-13. SmartPLS GmbH, Bönningsted, Germany Vance, A., Lowry, P. B., and Eggett, D. 2015. “Increasing (http://www.smartpls.com/). Accountability through User-Interface Design Artifacts: A New Ryan, R. M., and Deci, E. L. 2000. “Self-Determination Theory Approach to Addressing the Problem of Access-Policy Viola- and the Facilitation of Intrinsic Motivation, Social Development, tions,” MIS Quarterly (39:2), pp. 345-366. and Well-Being,” American Psychologist (55:1), pp. 68-78. Venkatesh, V., Thong, J. Y., and Xu, X. 2012. “Consumer Accep- Ryan, R. M., and Deci, E. L. 2001. “On Happiness and Human tance and Use of Information Technology: Extending the Potentials: A Review of Research on Hedonic and Eudaimonic Unified Theory of Acceptance and Use of Technology,” MIS Well-Being,” Annual Review of Psychology (52:1), pp. 141-166. Quarterly (36:1), pp. 157-178.

MIS Quarterly Vol. 43 No.1/March 2019 311 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Wankel, L. M. 1993. “The Importance of Enjoyment to Adherence and Expert Systems with Applications. She serves on the board of and Psychological Benefits from Physical Activity,” Interna- editors of Engineering Applications of Artificial Intelligence. tional Journal of Sport Psychology (24:2), pp. 151-169. Tabitha served as the corresponding author for this article. Ware, C. 2013. Information Visualization: Perception for Design (3rd ed.), Waltham, MA: Morgan Kaufmann. Linda Wallace is an associate professor and the John & Angela Wilson, P. M., Mack, D. E., and Grattan, K. P. 2008. “Under- Emery Junior Faculty Fellow in the Department of Accounting and standing Motivation for Exercise: A Self-Determination Theory Information Systems at Virginia Tech. She obtained her Ph.D. in Perspective,” Canadian Psychology (49:3), pp. 250-256. Computer Information Systems from Georgia State University in Wilson, P. M., Rodgers, W. M., Fraser, S. N., and Murray, T. C. 1999. Her research interests include online communities, fitness 2004. “Relationships between Exercise Regulations and Motiva- technologies, software project risk, and crowdfunding. Her research tional Consequences in University Students,” Research Quarterly has been accepted for publication in Journal of Management Infor- for Exercise and Sport (75:1), pp. 81-91. mation Systems, Decision Sciences, Communications of the ACM, Wilson, P. M., Rodgers, W. M., Loitz, C. C., and Scime, G. 2006. Information & Management, IEEE Security & Privacy, Decision “It’s Who I Am Really! The Importance of Integrated Regulation Support Systems, and others. She has served as an associate editor in Exercise Contexts,” Journal of Applied Biobehavioral for Information Systems Journal, Decision Sciences, and Informa- Research (11:2), pp. 79-104. tion & Management.

About the Authors Jason Deane is an associate professor of Business Information Technology in the Pamplin College of Business at Virginia Tech. Tabitha L. James is an associate professor of Business Information He received a Ph.D. in Information Systems and Operations Man- Technology in the Pamplin College of Business at Virginia Tech. agement from the University of Florida, and an M.B.A. and B.S. in She holds a Ph.D. from the University of Mississippi in Manage- Business Administration from Virginia Tech. His current research ment Information Systems. Her current research interests include interests are in the areas of computer aided decision support systems, information privacy and security, psychological impacts of technol- information system security, information system adoption analysis, ogy use, and combinatorial optimization. Her research has been and project management. He has articles published in Decision published in outlets such as Journal of Management Information Support Systems, Annals of Operations Research, Omega, Opera- Sysems, European Journal of Information Systems, European Jour- tions Management Research, The Journal of Computer Information nal of Operational Research, Information & Management, Com- Systems, and Journal of Information Technology Management puters & Security, IEEE Transactions on Evolutionary Computation, among others.

312 MIS Quarterly Vol. 43 No. 1/March 2019 RESEARCH ARTICLE

USING ORGANISMIC INTEGRATION THEORY TO EXPLORE THE ASSOCIATIONS BETWEEN USERS’ EXERCISE MOTIVATIONS AND FITNESS TECHNOLOGY FEATURE SET USE

Appendix A

Developing the Fitness Technology Feature Sets

Fitness technology feature use items were not available in the literature, and thus, were developed for the current study. Items were developed for each feature set following the procedure described next. We used a four-step process to determine the fitness technology features that make up our first-order subconstructs. First, we compiled a list of currently available fitness devices and their associated apps using lists of wearables published in popular media outlets. The original list contained 72 devices and was compiled using lists of wearables from CNET, PC Magazine, The Wall Street Journal, Engadget, Gizmodo, and others. While not comprehensive, the redundancy across multiple lists suggests that our list, at the very least, contains the most popular devices in the wearables category at the time the data was collected. Second, the three researchers independently visited every website of each device/app on the list and collected the features the company advertised for the fitness technology. Third, all three of the researchers’ feature lists were compared, discussed, and used to create an integrated list. Fourth, once survey items were created, an expert panel was convened to examine the feature list and scales as described below.

Accepted procedural methods (Churchill 1979; MacKenzie et al. 2011) were followed in developing the fitness technology use items. Once the items were developed, an expert panel was convened. The expert panel consisted of two faculty members who are well versed in survey- based methodologies, two faculty members who were active users of fitness technologies, and two employees of a fitness technology company. The expert panel was instructed to examine the entire survey instrument for clarity and to provide feedback on wording and note if any fitness technology features were missing. The expert panel did not provide any new fitness technology features, which indicated our list was reason- ably comprehensive. The expert panel did suggest wording changes to the fitness technology features set use items and the addition of a few more fitness device and apps to our list (primarily new versions of devices already present in our list). We compiled the expert panel wording suggestions and considered each suggestion. Most wording suggestions from the expert panel were implemented, improving the clarity of the survey items.

MIS Quarterly Vol. 43 No. 1—Appendices /March 2019 A1 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Appendix B

Items

Table B1. Items with References and Descriptive Statistics Std. Name Item Mean Dev. Motivation Items (BREQ-3) (from Markland and Tobin 2004; Wilson et al. 2006) Prompt: Why do you engage in exercise? (5-point Likert Scale; 1 = Not at all true for me to 5 = Completely true for me) Amot1 I don't see why I should have to exercise. 1.76 1.174 Amot2 I can't see why I should bother exercising. 1.73 1.158 Amot3 I don't see the point in exercising. 1.69 1.160 Amot4 I think exercising is a waste of time. 1.65 1.163 ExtReg1 I take part in exercise because my friends/family/partner say I should. 2.26 1.321 ExtReg2 I exercise because others will not be pleased with me if I don't. 2.07 1.265 ExtReg3 I feel under pressure from my friends/family to exercise. 2.19 1.312 ExtReg4 I exercise because other people say I should. 2.14 1.257 InjReg1 I feel ashamed when I miss an exercise session. 2.86 1.266 InjReg2 I feel like a failure when I haven't exercised in a while. 3.12 1.302 InjReg3 I would feel bad about myself if I was not making time to exercise. 3.30 1.225 InjReg4 I feel guilty when I don't exercise. 3.20 1.243 IdReg1 It's important to me to exercise regularly. 3.78 1.079 IdReg2 I value the benefits of exercise. 4.03 0.982 IdReg3 I think it is important to make the effort to exercise regularly. 3.91 0.968 IdReg4 I get restless if I don't exercise regularly. 3.15 1.285 IngReg1 I consider exercise part of my identity. 3.15 1.343 IngReg2 I consider exercise a fundamental part of who I am. 3.21 1.303 IngReg3 I consider exercise consistent with my values. 3.54 1.116 IngReg4 I exercise because it is consistent with my life goals. 3.65 1.093 IntReg1 I enjoy my exercise sessions. 3.59 1.098 IntReg2 I find exercise a pleasurable activity. 3.49 1.146 IntReg3 I exercise because it's fun. 3.26 1.240 IntReg4 I get pleasure and satisfaction from participating in exercise. 3.69 1.068 Fitness Technology Use Items: Developed for Current Study Prompt: I use (or have used) an exercise device and/or application (app) to: (5-point Likert Scale; 1 = Strongly Disagree to 5 = Strongly Agree) Share1 share my exercise statistics with other people. 2.74 1.337 Share2 share my exercise information with other people. 2.78 1.338 Share3 share my exercise data in a public forum (e.g., leaderboard, ranking, social media). 2.56 1.328 Share4 share my exercise accomplishments for other people to see. 2.71 1.321 Encourage1 have other people encourage my exercise activities. 2.76 1.339 Encourage2 receive encouraging messages regarding my exercise activities from others. 2.67 1.337 Encourage3 have my exercise accomplishments acknowledged by other people. 2.83 1.323 Encourage4 receive moral support for my exercise activities from others. 2.80 1.304 Coach1 get coaching from a live personal trainer. 2.31 1.267 Coach2 receive expert advice about my exercise regimen from a live coach. 2.40 1.287 Coach3 obtain feedback from a live coach about how my exercise activities are going. 2.35 1.281 Coach4 have a live coach guide me through my exercise regimen. 2.34 1.278

A2 MIS Quarterly Vol. 43 No. 1—Appendices/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Compare1 compare my exercise activities to other people's exercise activities. 2.79 1.326 Compare2 see how other people's exercise activities compare to mine. 2.84 1.329 Compare3 compare my exercise activities to the exercise activities of others. 2.80 1.329 Compare4 track my exercise activities with respect to how they compare to others. 2.84 1.323 Compare5 rank my exercise activities relative to others' exercise activities. 2.72 1.306 Compete1 compete with other people. 2.69 1.339 Compete2 challenge other individuals to exercise competitions. 2.61 1.319 Compete3 have exercise contests with other individuals. 2.66 1.363 Compete4 enter into exercise competitions with others. 2.59 1.332 Goals1 set my exercise goals. 3.95 0.946 Goals2 establish my exercise goals. 3.93 0.941 Goals3 develop goals for my exercise. 3.91 0.954 Goals4 create my exercise goals. 3.89 0.963 Remind1 remind me to do an exercise activity. 3.28 1.245 Remind2 notify me to perform an exercise activity. 3.18 1.261 Remind3 provide me with reminders when I need to do an exercise activity. 3.22 1.243 Remind4 prompt me when I need to perform an exercise activity. 3.19 1.221 Rewards1 receive rewards (e.g., discounts, points, badges, etc.) for my exercise activities. 2.69 1.323 Rewards2 obtain rewards (e.g., discounts, points, badges, etc.) for my exercise activities. 2.63 1.312 Rewards2 win prizes (e.g., discounts, points, badges, etc.) for my exercise activities. 2.56 1.321 Rewards4 earn prizes (e.g., discounts, points, badges, etc.) for my exercise activities. 2.62 1.332 Analyze1 manage my exercise data. 3.97 0.903 Analyze2 observe patterns in my exercise data. 3.76 0.996 Analyze3 analyze my exercise data. 3.90 0.949 Analyze4 calculate trends from my exercise data. 3.73 1.043 Analyze5 graph my exercise data. 3.86 1.009 Collect1 gather my exercise data. 4.06 0.840 Collect2 collect my exercise information. 4.11 0.802 Collect3 record my exercise data. 4.17 0.801 Collect4 accumulate my exercise data. 4.04 0.835 Updates1 provide me with messages about my exercise progress. 3.51 1.103 Updates2 give me visual cues (e.g., status bar, colors) about my exercise progress. 3.85 1.032 Updates3 provide me with exercise progress updates. 3.76 1.010 Updates4 update me with the status of my exercise progress. 3.82 0.984 Search1 search for exercise information(e.g., exercise routes, new exercise routines, etc.). 3.30 1.244 Search2 access exercise information (e.g., exercise routes, new exercise routines, etc.). 3.38 1.223 Search3 find exercise information (e.g., exercise routes, new exercise routines, etc.) that is 3.37 1.235 relevant to me. Search4 browse exercise information (e.g., exercise routes, new exercise routines, etc.). 3.35 1.218 Subjective Vitality Items (from Bostic et al. 2000; Ryan and Frederick 1997) Prompt: Please respond to each of the following statements by indicating the degree to which the statement is true for you when engaged in exercise. (5-point Likert Scale; 1 = Not at all true for me to 5 = Completely true for me) Vitality1 I feel alive and vital. 3.60 1.059 Vitality2 Sometimes I feel so alive I just want to burst. 2.93 1.282 Vitality3 I have energy and spirit. 3.61 1.062 Vitality4 I look forward to each new day. 3.63 1.066 Vitality5 I nearly always feel alert and awake. 3.39 1.130 Vitality6 I feel energized. 3.55 1.103

MIS Quarterly Vol. 43 No. 1—Appendices /March 2019 A3 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Appendix C

Details of Statistical Testing

Convergent Validity

To confirm convergent validity, all items “thought to reflect a construct converge, or show significant, high correlations with one another, particularly when compared to the items relevant to other constructs” (Straub et al. 2004, p. 391). In order to establish convergent validity in PLS, a bootstrap is run and the outer loadings and associated t-statistics are examined, along with the cross-loading matrix. The outer-loadings and t-statistics for our measurement model are given in Table C1 and the cross-loading matrix is shown in Table C2. The outer-loadings for most of our items in Table C1 are above 0.7, which is recommended, although for large sample sizes loadings above 0.3 are adequate (Hair et al. 2006). All items were retained because their loadings were adequate and their t-statistics indicated that the loadings were significant, indicating convergent validity.

Table C1. Outer Loadings and t-statistics Outer Outer Construct Item Loading t-statistic Construct Item Loading t-statistic Amot1 0.886 69.113 Compare1 0.915 108.942 Amot2 0.897 98.405 Compare2 0.914 115.230 Nonregulation Social Amot3 0.906 86.352 Compare3 0.916 111.796 Comparison Amot4 0.880 66.840 Compare4 0.887 78.675 ExtReg1 0.874 73.189 Compare5 0.888 76.823 ExtReg2 0.883 96.558 Compete1 0.918 131.936 External Regulation ExtReg3 0.860 62.212Social Compete2 0.932 140.902 ExtReg4 0.842 59.523Competition Compete3 0.937 179.126 InjReg1 0.829 46.229 Compete4 0.930 126.877 Introjected InjReg2 0.846 62.949 GoalMgmt1 0.866 66.425 Regulation InjReg3 0.791 36.455Goal GoalMgmt2 0.874 70.996 InjReg4 0.862 71.751Management GoalMgmt3 0.816 43.943 IdReg1 0.859 67.568 GoalMgmt4 0.849 52.303 Identified IdReg2 0.788 45.230 Remind1 0.887 73.660 Regulation IdReg3 0.832 52.795 Remind2 0.911 118.120 Reminders IdReg4 0.746 38.647 Remind3 0.915 118.187 IngReg1 0.818 50.315 Remind4 0.910 111.337 Integrated IngReg2 0.887 113.148 Rewards1 0.931 130.175 Regulation IngReg3 0.894 119.111 Rewards2 0.941 172.066 Rewards IngReg4 0.850 67.871 Rewards3 0.935 140.809 IntReg1 0.852 66.475 Rewards4 0.937 159.143 IntReg2 0.908 120.280 Analyze1 0.756 35.250 Intrinsic Regulation IntReg3 0.922 147.234 Analyze2 0.771 39.391 IntReg4 0.864 75.770Data Analysis Analyze3 0.809 54.218 Sharing1 0.911 112.122 Analyze4 0.797 50.604 Sharing2 0.908 101.344 Analyze5 0.677 24.767 Social Data Sharing Sharing3 0.877 77.880 Collect1 0.827 52.722 Sharing4 0.910 116.294 Collect2 0.807 39.880 Data Collection Encourage1 0.878 80.641 Collect3 0.791 41.066 Social Encourage2 0.904 124.769 Collect4 0.841 65.058 Encouragement Encourage3 0.902 96.897 Updates1 0.709 26.035 Encourage4 0.915 128.505 Updates2 0.681 24.448 Data Updates Coach1 0.935 125.135 Updates3 0.794 45.586 Live Coaching Coach2 0.938 153.218 Updates4 0.795 50.254

A4 MIS Quarterly Vol. 43 No. 1—Appendices/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Coach3 0.937 144.606 Search1 0.895 72.747 Live Coaching Coach4 0.940 170.583Information Search2 0.875 63.403 Searching Search3 0.895 81.968 Search4 0.912 116.501 Vitality1 0.881 94.025 Vitality2 0.717 29.922 Vitality3 0.884 101.583 Vitality Vitality4 0.778 44.163 Vitality5 0.826 61.132 Vitality6 0.873 71.204

Discriminant Validity

To establish discriminant validity, the cross-loading matrix can be examined for troublesome cross-loadings between the indicators. Discrim- inant validity is confirmed if it can be illustrated that “measurement items posited to reflect (i.e., ‘make up’) that construct differ from those that are not believed to make up the construct” (Straub et al. 2004, p. 389). Loadings should be an order of magnitude greater than the nearest cross-loading (i.e., the difference between the primary loading and any other loading should be greater than 0.1) (Lowry and Gaskin 2014). This is the case for all of the items seen in Table C2 with the exception of encourage2, compare5, and compete1. However, these three items all load highest on their primary factor and cross-load with other first-order subconstructs of the same second-order construct where we might expect some correlation, so we retained these items and performed the second check for discriminant validity. For the second check, we examined the square root of the average variance extracted (AVE) for a construct in comparison to the construct correlations of that construct with every other first-order construct in the model. These results are shown in Table C3. The right portion of Table C3 contains the construct correlations. The bolded values that appear down the diagonal of the table are the square roots of the AVEs found in the second column for each construct. Any correlation below an bolded value should be lower than that bolded value (Fornell and Larcker 1981), which is the case for all of our constructs. Taken together, these results indicate discriminant validity.

Reliability

Reliability was examined using the AVE, composite reliability, and Cronbach’s alpha for each construct. These values were calculated by the PLS algorithm and provided as output. For our model, these values are provided in Table C3. Reliability scores are intended to provide an indication of how reliable the scales will be over time (Straub 1989). Ideally, the composite reliability should be above 0.7 (Hair et al. 2006) and greater than the AVE. Both are true for all constructs: all composite reliabilities are above 0.7 and the AVE is less than the composite reliability. It is recommended that the AVE be 0.5 or above (Fornell and Larcker 1981; Hair et al. 2006), which is the case for all of our constructs. Cronbach’s alphas above 0.7 are recommended and above 0.5 are acceptable (Davis 1964; Peterson 1994). The Cronbach’s alphas are above 0.7 for all of the constructs in our model. Thus, reliability was confirmed for all of the scales used in the study.

Multicollinearity

Multicollinearity refers to the situation where predictors are highly correlated with each other. To check for multicollinearity, the variance inflation factors (VIF) obtained from SmartPLS Version 3.2.1 can be examined. It is suggested that the VIF be below 10 (Hair et al. 2006; Neter et al. 1996). A VIF of greater than or equal to 5 has been suggested to be indicative of moderate multicollinearity and greater than or equal to 10 suggestive of severe multicollinearity (Larose and Larose 2015). VIF values for the items for this study are given in Table C4. All of the VIFs are below 10 and most are below 5, which suggest that multicollinearity is not an issue in our model.

MIS Quarterly Vol. 43 No. 1—Appendices /March 2019 A5 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Table C2. Cross-Loading Matrix Amot ExtReg InjReg IdReg IngReg IntReg Sharing Encourage Coach Compare Compete Goal Mgmt Remind Rewards Analyze Collect Updates Search Vitality Amot1 0.886 0.628 0.105 -0.138 0.041 0.001 0.322 0.291 0.469 0.300 0.342 -0.069 0.206 0.314 -0.059 -0.218 0.007 0.112 0.076 Amot2 0.897 0.625 0.107 -0.113 0.077 0.013 0.391 0.359 0.536 0.347 0.390 -0.034 0.247 0.357 -0.048 -0.233 0.024 0.147 0.142 Amot3 0.906 0.604 0.099 -0.145 0.036 -0.038 0.324 0.297 0.448 0.294 0.345 -0.082 0.185 0.330 -0.094 -0.266 -0.045 0.088 0.063 Amot4 0.880 0.587 0.116 -0.127 0.020 -0.020 0.339 0.290 0.440 0.300 0.343 -0.076 0.185 0.326 -0.068 -0.245 -0.017 0.093 0.054 ExtReg1 0.591 0.874 0.199 -0.029 0.104 -0.008 0.386 0.358 0.427 0.348 0.394 0.009 0.223 0.368 0.040 -0.100 0.099 0.071 0.014 ExtReg2 0.585 0.883 0.221 0.019 0.147 0.067 0.416 0.413 0.468 0.386 0.407 0.052 0.290 0.373 0.039 -0.119 0.114 0.127 0.111 ExtReg3 0.648 0.860 0.242 0.026 0.166 0.084 0.398 0.366 0.471 0.361 0.389 0.012 0.277 0.388 0.022 -0.162 0.035 0.119 0.129 ExtReg4 0.542 0.842 0.243 -0.029 0.081 -0.011 0.361 0.339 0.403 0.339 0.355 0.051 0.228 0.351 0.017 -0.131 0.072 0.072 0.006 InjReg1 0.026 0.172 0.829 0.522 0.445 0.371 0.111 0.118 0.025 0.136 0.116 0.106 0.114 0.075 0.124 0.144 0.188 0.079 0.172 InjReg2 0.201 0.284 0.846 0.435 0.465 0.352 0.214 0.213 0.197 0.229 0.228 0.120 0.190 0.197 0.158 0.098 0.183 0.166 0.238 InjReg3 0.120 0.253 0.791 0.403 0.377 0.289 0.145 0.161 0.108 0.170 0.140 0.115 0.116 0.122 0.084 0.088 0.144 0.116 0.099 InjReg4 0.043 0.171 0.862 0.624 0.589 0.483 0.214 0.209 0.138 0.229 0.186 0.202 0.195 0.130 0.193 0.170 0.233 0.165 0.309 IdReg1 -0.151 0.001 0.492 0.859 0.698 0.619 0.112 0.123 0.045 0.109 0.069 0.227 0.120 0.024 0.170 0.233 0.196 0.204 0.415 IdReg2 -0.229 -0.096 0.406 0.788 0.550 0.558 0.039 0.038 -0.081 0.051 -0.001 0.252 0.050 -0.042 0.209 0.293 0.220 0.145 0.341 IdReg3 -0.220 -0.081 0.466 0.832 0.599 0.557 0.044 0.043 -0.055 0.032 -0.002 0.251 0.099 -0.036 0.213 0.317 0.255 0.171 0.356 IdReg4 0.109 0.152 0.581 0.746 0.660 0.629 0.165 0.165 0.158 0.191 0.154 0.130 0.147 0.132 0.144 0.160 0.171 0.190 0.407 IngReg1 -0.077 0.054 0.505 0.723 0.818 0.637 0.169 0.167 0.049 0.162 0.145 0.249 0.161 0.081 0.232 0.288 0.260 0.201 0.414 IngReg2 0.167 0.232 0.515 0.626 0.887 0.686 0.300 0.261 0.316 0.269 0.249 0.156 0.235 0.215 0.136 0.093 0.202 0.258 0.521 IngReg3 0.088 0.136 0.504 0.664 0.894 0.688 0.245 0.220 0.251 0.225 0.206 0.128 0.243 0.186 0.149 0.153 0.217 0.241 0.508 IngReg4 -0.033 0.062 0.479 0.700 0.850 0.667 0.181 0.184 0.107 0.181 0.124 0.168 0.140 0.091 0.205 0.260 0.241 0.216 0.459 IntReg1 0.050 0.072 0.392 0.581 0.643 0.852 0.183 0.154 0.140 0.171 0.139 0.128 0.158 0.123 0.188 0.190 0.179 0.170 0.472 IntReg2 -0.007 0.051 0.412 0.658 0.710 0.908 0.216 0.195 0.159 0.204 0.152 0.169 0.163 0.099 0.226 0.212 0.200 0.215 0.566 IntReg3 -0.001 0.042 0.407 0.675 0.715 0.922 0.213 0.183 0.171 0.197 0.148 0.186 0.178 0.097 0.218 0.224 0.224 0.232 0.547 IntReg4 -0.080 -0.019 0.433 0.691 0.685 0.864 0.134 0.116 0.084 0.133 0.071 0.231 0.130 0.046 0.203 0.257 0.202 0.212 0.508 Sharing1 0.338 0.398 0.208 0.119 0.246 0.200 0.911 0.786 0.522 0.781 0.743 0.185 0.310 0.472 0.219 0.023 0.257 0.250 0.250 Sharing2 0.338 0.398 0.203 0.115 0.246 0.206 0.908 0.793 0.510 0.772 0.719 0.173 0.294 0.475 0.191 -0.001 0.249 0.227 0.234 Sharing3 0.392 0.446 0.187 0.092 0.236 0.176 0.877 0.734 0.552 0.714 0.693 0.107 0.330 0.516 0.134 -0.067 0.211 0.238 0.235 Sharing4 0.333 0.392 0.182 0.086 0.223 0.180 0.910 0.789 0.507 0.759 0.703 0.170 0.295 0.467 0.178 -0.006 0.229 0.247 0.223 Encourage1 0.295 0.375 0.163 0.073 0.196 0.131 0.742 0.878 0.495 0.728 0.694 0.188 0.377 0.489 0.166 -0.016 0.272 0.256 0.191 Encourage2 0.335 0.410 0.235 0.122 0.272 0.200 0.820 0.904 0.550 0.775 0.744 0.204 0.347 0.488 0.186 -0.033 0.241 0.247 0.246 Encourage3 0.296 0.366 0.188 0.133 0.205 0.166 0.740 0.902 0.498 0.731 0.701 0.234 0.338 0.478 0.167 -0.040 0.259 0.250 0.228 Encourage4 0.329 0.388 0.201 0.096 0.202 0.163 0.793 0.915 0.521 0.772 0.727 0.217 0.334 0.495 0.182 -0.033 0.252 0.261 0.235 Coach1 0.522 0.505 0.131 0.008 0.205 0.128 0.533 0.529 0.935 0.486 0.492 0.081 0.403 0.544 0.044 -0.188 0.137 0.394 0.237 Coach2 0.484 0.469 0.150 0.026 0.191 0.141 0.553 0.548 0.938 0.516 0.502 0.094 0.395 0.528 0.071 -0.180 0.164 0.409 0.263 Coach3 0.492 0.464 0.152 0.043 0.215 0.167 0.543 0.545 0.937 0.496 0.495 0.113 0.415 0.519 0.084 -0.149 0.164 0.382 0.276 Coach4 0.504 0.486 0.139 0.022 0.209 0.154 0.543 0.531 0.940 0.494 0.495 0.091 0.426 0.520 0.068 -0.182 0.141 0.393 0.261 Compare1 0.322 0.378 0.223 0.099 0.227 0.183 0.780 0.756 0.468 0.915 0.792 0.178 0.283 0.489 0.191 -0.005 0.231 0.220 0.254 Compare2 0.326 0.380 0.210 0.097 0.209 0.159 0.772 0.767 0.481 0.914 0.783 0.167 0.271 0.456 0.176 -0.024 0.200 0.223 0.232 Compare3 0.331 0.399 0.234 0.120 0.238 0.192 0.771 0.766 0.490 0.916 0.802 0.199 0.259 0.492 0.183 -0.008 0.236 0.221 0.264 Compare4 0.279 0.355 0.211 0.116 0.222 0.183 0.724 0.749 0.493 0.887 0.776 0.225 0.335 0.470 0.228 0.043 0.268 0.215 0.280 Compare5 0.320 0.364 0.203 0.120 0.215 0.185 0.746 0.738 0.469 0.888 0.801 0.176 0.318 0.503 0.195 0.016 0.245 0.213 0.253 Compete1 0.383 0.418 0.180 0.060 0.186 0.116 0.738 0.739 0.477 0.828 0.918 0.172 0.292 0.506 0.140 -0.040 0.201 0.179 0.223 Compete2 0.350 0.402 0.191 0.069 0.197 0.133 0.741 0.756 0.490 0.803 0.932 0.177 0.300 0.492 0.133 -0.047 0.208 0.202 0.221 Compete3 0.363 0.415 0.195 0.055 0.188 0.141 0.734 0.733 0.503 0.823 0.937 0.175 0.298 0.530 0.129 -0.033 0.206 0.200 0.222 Compete4 0.389 0.428 0.218 0.084 0.224 0.148 0.733 0.733 0.497 0.796 0.930 0.170 0.308 0.553 0.134 -0.042 0.200 0.193 0.218 GoalMgmt1 -0.072 0.003 0.122 0.235 0.171 0.155 0.138 0.191 0.067 0.170 0.145 0.866 0.274 0.130 0.410 0.376 0.355 0.288 0.201 GoalMgmt2 -0.076 0.029 0.147 0.228 0.176 0.184 0.143 0.198 0.068 0.175 0.160 0.874 0.271 0.114 0.404 0.373 0.371 0.288 0.177 GoalMgmt3 -0.029 0.066 0.159 0.219 0.161 0.183 0.164 0.205 0.128 0.209 0.172 0.816 0.264 0.159 0.395 0.351 0.368 0.264 0.191 GoalMgmt4 -0.066 0.025 0.155 0.218 0.173 0.165 0.157 0.202 0.081 0.157 0.160 0.849 0.280 0.123 0.396 0.316 0.383 0.301 0.185 Remind1 0.196 0.240 0.186 0.122 0.181 0.152 0.293 0.333 0.354 0.265 0.273 0.299 0.887 0.236 0.190 0.047 0.374 0.263 0.188 Remind2 0.228 0.285 0.182 0.120 0.242 0.194 0.333 0.373 0.431 0.317 0.310 0.293 0.911 0.292 0.238 0.046 0.413 0.330 0.214 Remind3 0.218 0.300 0.173 0.119 0.200 0.148 0.320 0.368 0.408 0.306 0.304 0.288 0.915 0.292 0.211 0.051 0.400 0.300 0.214

A6 MIS Quarterly Vol. 43 No. 1—Appendices/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Remind4 0.201 0.246 0.167 0.115 0.205 0.150 0.286 0.328 0.389 0.283 0.281 0.280 0.910 0.259 0.213 0.031 0.384 0.301 0.184 Rewards1 0.334 0.393 0.163 0.035 0.164 0.099 0.512 0.505 0.516 0.512 0.530 0.155 0.272 0.931 0.181 0.004 0.230 0.249 0.142 Rewards2 0.331 0.388 0.149 0.028 0.153 0.102 0.494 0.508 0.507 0.493 0.523 0.156 0.272 0.941 0.171 0.001 0.220 0.236 0.125 Rewards3 0.377 0.413 0.151 0.020 0.160 0.091 0.501 0.512 0.554 0.496 0.525 0.123 0.291 0.935 0.136 -0.046 0.190 0.265 0.140 Rewards4 0.354 0.411 0.152 0.021 0.164 0.093 0.495 0.503 0.530 0.495 0.520 0.145 0.282 0.937 0.154 -0.027 0.207 0.248 0.156 Analyze1 -0.083 0.017 0.187 0.219 0.192 0.209 0.138 0.143 0.015 0.141 0.092 0.407 0.165 0.105 0.756 0.574 0.458 0.160 0.234 Analyze2 -0.066 -0.008 0.097 0.164 0.148 0.176 0.140 0.111 0.057 0.153 0.078 0.328 0.205 0.116 0.771 0.504 0.508 0.218 0.174 Analyze3 -0.078 0.042 0.158 0.202 0.179 0.205 0.177 0.198 0.071 0.187 0.138 0.369 0.193 0.163 0.809 0.567 0.543 0.206 0.209 Analyze4 -0.012 0.059 0.154 0.145 0.168 0.167 0.169 0.162 0.104 0.185 0.124 0.380 0.209 0.147 0.797 0.511 0.494 0.197 0.169 Analyze5 -0.041 0.019 0.079 0.129 0.090 0.139 0.139 0.127 0.020 0.154 0.119 0.311 0.120 0.122 0.677 0.493 0.463 0.138 0.146 Collect1 -0.237 -0.139 0.121 0.264 0.208 0.219 0.000 -0.019 -0.149 0.023 -0.023 0.332 0.043 -0.037 0.599 0.827 0.481 0.114 0.190 Collect2 -0.223 -0.131 0.095 0.244 0.129 0.152 -0.024 -0.043 -0.165 -0.012 -0.046 0.356 0.034 -0.033 0.540 0.807 0.414 0.073 0.134 Collect3 -0.231 -0.132 0.123 0.246 0.173 0.205 -0.037 -0.044 -0.159 -0.028 -0.076 0.328 0.057 0.008 0.533 0.791 0.430 0.085 0.131 Collect4 -0.189 -0.084 0.162 0.249 0.212 0.233 0.015 -0.007 -0.138 0.028 -0.001 0.343 0.023 0.002 0.593 0.841 0.450 0.072 0.155 Updates1 0.079 0.172 0.179 0.167 0.219 0.164 0.263 0.285 0.232 0.242 0.224 0.297 0.477 0.232 0.399 0.278 0.709 0.232 0.181 Updates2 -0.061 0.004 0.155 0.150 0.105 0.105 0.127 0.134 0.034 0.131 0.112 0.317 0.180 0.123 0.463 0.435 0.681 0.124 0.061 Updates3 0.016 0.096 0.198 0.227 0.238 0.193 0.225 0.243 0.143 0.217 0.186 0.333 0.365 0.181 0.523 0.417 0.794 0.230 0.197 Updates4 -0.046 0.018 0.158 0.223 0.221 0.211 0.176 0.193 0.085 0.192 0.141 0.345 0.289 0.148 0.534 0.477 0.795 0.193 0.189 Search1 0.160 0.150 0.166 0.201 0.254 0.217 0.251 0.255 0.412 0.234 0.208 0.297 0.279 0.266 0.194 0.063 0.215 0.895 0.272 Search2 0.083 0.061 0.150 0.207 0.231 0.206 0.215 0.237 0.345 0.187 0.151 0.295 0.294 0.205 0.218 0.108 0.245 0.875 0.259 Search3 0.089 0.087 0.138 0.187 0.228 0.207 0.253 0.276 0.367 0.228 0.191 0.307 0.299 0.251 0.214 0.088 0.224 0.895 0.257 Search4 0.119 0.114 0.144 0.197 0.242 0.210 0.236 0.240 0.382 0.217 0.196 0.299 0.307 0.233 0.238 0.116 0.247 0.912 0.254 Vitality1 0.010 -0.007 0.209 0.446 0.487 0.534 0.167 0.147 0.161 0.197 0.148 0.169 0.125 0.069 0.199 0.213 0.153 0.215 0.881 Vitality3 0.239 0.193 0.267 0.282 0.397 0.443 0.302 0.298 0.317 0.321 0.281 0.150 0.233 0.215 0.193 0.062 0.163 0.248 0.717 Vitality4 0.047 0.037 0.209 0.428 0.493 0.503 0.227 0.223 0.235 0.255 0.222 0.209 0.194 0.116 0.234 0.185 0.200 0.273 0.884 Vitality5 0.007 0.036 0.242 0.439 0.481 0.486 0.176 0.162 0.167 0.169 0.133 0.226 0.193 0.107 0.228 0.183 0.209 0.235 0.778 Vitality6 0.112 0.105 0.191 0.343 0.410 0.435 0.222 0.210 0.266 0.241 0.207 0.157 0.198 0.138 0.172 0.130 0.153 0.213 0.826 Vitality7 0.085 0.048 0.215 0.401 0.475 0.530 0.213 0.213 0.241 0.236 0.201 0.187 0.166 0.117 0.189 0.148 0.178 0.259 0.873

MIS Quarterly Vol. 43 No. 1—Appendices /March 2019 A7 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Table C3. Construct Correlations, AVES, Composite Reliabilities, and Cronbach’s Alphas AVE C.R. C.A. Amot ExtReg InjReg IdReg IngReg IntReg Sharing Encourage Coach Compare Compete Goal Mgmt Remind Rewards Analyze Collect Updates Search Vitality Amotivation 0.796 0.940 0.915 0.892 External 0.748 0.922 0.888 0.685 0.865 Regulation Introjected 0.693 0.900 0.856 0.120 0.262 0.832 Regulation Identified 0.652 0.882 0.821 -0.146 -0.002 0.607 0.807 Regulation Integrated 0.744 0.912 0.885 0.051 0.146 0.580 0.782 0.863 Regulation Intrinsic 0.787 0.936 0.909 -0.011 0.041 0.463 0.735 0.777 0.887 Regulation Social Data 0.813 0.946 0.923 0.388 0.452 0.217 0.115 0.264 0.212 0.902 Sharing Social Encour- 0.810 0.945 0.922 0.349 0.428 0.219 0.118 0.244 0.184 0.860 0.900 agement Live 0.879 0.967 0.954 0.534 0.513 0.153 0.026 0.219 0.157 0.579 0.574 0.938 Coaching Social 0.817 0.957 0.944 0.350 0.415 0.239 0.122 0.246 0.200 0.839 0.836 0.531 0.904 Comparison Social 0.863 0.962 0.947 0.399 0.448 0.211 0.072 0.214 0.145 0.793 0.797 0.529 0.875 0.929 Competition Goal 0.725 0.913 0.873 -0.072 0.036 0.171 0.265 0.200 0.202 0.177 0.234 0.101 0.209 0.187 0.851 Management Reminders 0.821 0.948 0.927 0.233 0.296 0.195 0.131 0.229 0.178 0.340 0.387 0.437 0.324 0.322 0.320 0.906 Rewards 0.876 0.966 0.953 0.373 0.428 0.164 0.028 0.171 0.103 0.535 0.542 0.563 0.533 0.560 0.155 0.298 0.936 Data 0.583 0.874 0.820 -0.074 0.034 0.179 0.226 0.205 0.236 0.201 0.195 0.071 0.215 0.144 0.471 0.235 0.172 0.764 Analysis Data 0.667 0.889 0.833 -0.269 -0.148 0.154 0.308 0.222 0.249 -0.013 -0.034 -0.186 0.004 -0.043 0.416 0.048 -0.018 0.695 0.817 Collection Data 0.557 0.833 0.733 -0.007 0.093 0.231 0.260 0.264 0.228 0.263 0.284 0.162 0.261 0.219 0.434 0.434 0.226 0.647 0.544 0.746 Updates Information 0.800 0.941 0.916 0.125 0.115 0.167 0.222 0.267 0.235 0.267 0.282 0.421 0.242 0.208 0.335 0.330 0.266 0.242 0.106 0.261 0.894 Searching Vitality 0.687 0.929 0.907 0.097 0.079 0.267 0.473 0.554 0.591 0.261 0.250 0.277 0.284 0.238 0.222 0.221 0.151 0.245 0.188 0.213 0.291 0.829

Common Method Bias

Our study design incorporated recommendations to reduce common method bias following leading literature (MacKenzie et al. 2011; Podsakoff et al. 2003). The survey was implemented on the Quatrics platform. The Qualtrics survey platform was used because it is an approved survey administration tool by the researchers’ institutional review board (IRB) and allows for data to be anonymously collected on the Amazon Mechanical Turk (mTurk) platform. The survey items were randomized within blocks based upon the Likert-scale response anchors for the items (e.g., strongly disagree to strongly agree). Providing anonymity to the survey respondents has been recommended as an approach to reduce common method bias (Podsakoff et al. 2003) by reducing the tendency of respondents to answer in a way that they think the researchers would prefer. Randomizing the survey items has also been suggested as a way to decrease common method bias (Podsakoff et al. 2003). “Attention trap” items were inserted throughout the survey. Attention trap items ask the respondent to select a particular response from the Likert-scale responses (Oppenheimer et al. 2009). For example, the respondent may be asked to “Please answer ‘Agree’ to this question.” The purpose of the trap items is to identify those respondents that are not cognitively engaged in responding to the survey and to discard those responses.

In addition, the construct correlation matrix can be examined to determine if any constructs are correlated above 0.90, which could indicate a common method bias issue (Pavlou et al. 2007). An examination of the construct correlations in Table C3 reveals that none of our constructs are correlated above 0.90. Harmon’s single-factor test (Lowry and Gaskin 2014; Podsakoff et al. 2003) was also employed to check for common method bias. We examined the unrotated factor solution in SPSS for all the items of our first-order constructs. The factor analysis revealed 12 distinct factors with the largest factor accounting for only 26.109% of the variance. This further suggests a lack of common method bias (Lowry and Gaskin 2014).

A8 MIS Quarterly Vol. 43 No. 1—Appendices/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Table C4. Variance Inflation Factors Construct Item VIF Construct Item VIF Amot1 2.790 Compare1 4.465 Amot2 2.722 Compare2 4.295 Nonregulation Amot3 3.328Social Comparison Compare3 4.591 Amot4 2.774 Compare4 3.474 ExtReg1 2.490 Compare5 3.698 ExtReg2 2.206 Compete1 4.380 External Regulation ExtReg3 2.150 Compete2 4.736 Social Competition ExtReg4 2.530 Compete3 5.034 InjReg1 1.899 Compete4 4.613 InjReg2 1.935 GoalMgmt1 2.340 Introjected Regulation InjReg3 1.900 GoalMgmt2 2.498 Goal Management InjReg4 2.177 GoalMgmt3 1.869 IdReg1 2.108 GoalMgmt4 2.201 IdReg2 1.752 Remind1 2.845 Identified Regulation IdReg3 2.056 Remind2 3.461 Reminders IdReg4 1.426 Remind3 3.588 IngReg1 2.873 Remind4 3.429 IngReg2 3.027 Rewards1 4.581 Integrated Regulation IngReg3 2.270 Rewards2 5.197 Rewards IngReg4 2.009 Rewards3 4.769 IntReg1 3.164 Rewards4 4.792 IntReg2 3.617 Analyze1 1.786 Intrinsic Regulation IntReg3 2.355 Analyze2 1.779 IntReg4 2.448Data Analysis Analyze3 2.008 Sharing1 3.896 Analyze4 1.874 Sharing2 3.867 Analyze5 1.517 Social Data Sharing Sharing3 2.995 Collect1 2.040 Sharing4 3.835 Collect2 1.863 Data Collection Encourage1 2.980 Collect3 1.811 Encourage2 4.044 Collect4 2.133 Social Encouragement Encourage3 3.325 Updates1 1.441 Encourage4 4.003 Updates2 1.420 Data Updates Coach1 4.738 Updates3 1.684 Coach2 4.967 Updates4 1.677 Live Coaching Coach3 5.136 Search1 3.046 Coach4 5.113 Search2 2.587 Information Searching Search3 3.001 Search4 3.301 Vitality1 3.081 Vitality2 1.586 Vitality3 3.204 Vitality Vitality4 1.887 Vitality5 2.378 Vitality6 3.087

MIS Quarterly Vol. 43 No. 1—Appendices /March 2019 A9 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Power

With a sample size of 880 and a probability level of 0.05, using the post hoc statistical power calculator for multiple regression (http://www.danielsoper.com/statcalc/calculator.aspx?id=9) our power is sufficient for each of our endogenous variables (i.e., > 0.80).

Appendix D

Path Coefficients for Second-Order Formative Constructs

***p # 0.001; **p # 0.01; *p # 0.05; n/s = not significant.

Figure D1. Path Coefficients for Second-Order Formative Constructs

A10 MIS Quarterly Vol. 43 No. 1—Appendices/March 2019 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

Appendix E

Additional Moderation Testing

Table E1. Bootstrapped CI Tests for Moderation

Interaction 2.5% lower bound 97.5% upper bound Zero included? Support? Moderation of Exercise Motivations  Social Interaction Features by Controls Nonregulation × Age  Social Interaction Features 0.0033 -0.0217 Yes No External Regulation × Age  Social Interaction Features 0.0110 -0.0178 Yes No Identified Regulation × Age  Social Interaction Features 0.0081 -0.0080 Yes No Integrated Regulation × Age  Social Interaction Features 0.0090 -0.0101 Yes No Intrinsic Regulation × Age  Social Interaction Features 0.0077 -0.0064 Yes No Non-Regulation × Device/App Proficiency  Social Interaction Features 0.0473 0.0110 No Yes External Regulation × Device/App Proficiency  Social Interaction Features 0.0668 0.0228 No Yes Identified Regulation × Device/App Proficiency  Social Interaction Features -0.0034 -0.0364 No Yes Integrated Regulation × Device/App Proficiency  Social Interaction Features 0.0415 0.0066 No Yes Intrinsic Regulation × Device/App Proficiency  Social Interaction Features 0.0329 0.0010 No Yes Non-Regulation × Frequency of Use  Social Interaction Features -0.0031 -0.0273 No Yes External Regulation × Frequency of Use  Social Interaction Features -0.0052 -0.0431 No Yes Identified Regulation × Frequency of Use  Social Interaction Features 0.0223 0.0008 No Yes Integrated Regulation × Frequency of Use  Social Interaction Features -0.0015 -0.0264 No Yes Intrinsic Regulation × Frequency of Use  Social Interaction Features -0.0001 -0.0110 No Yes Non-Regulation × Length of Ownership  Social Interaction Features 0.0207 -0.0030 Yes No External Regulation × Length of Ownership  Social Interaction Features 0.0349 -0.0045 Yes No Identified Regulation × Length of Ownership  Social Interaction Features 0.0018 -0.0168 Yes No Integrated Regulation × Length of Ownership  Social Interaction Features 0.0205 -0.0024 Yes No Intrinsic Regulation × Length of Ownership  Social Interaction Features 0.0140 -0.0020 Yes No Moderation of Exercise Motivations  Exercise Control Features by Controls External Regulation × Age  Exercise Control Features -0.0003 -0.0433 No Yes Integrated Regulation × Age  Exercise Control Features 0.0061 -0.0314 Yes No External Regulation × Device/App Proficiency  Exercise Control Features 0.0397 -0.0037 Yes No Integrated Regulation × Device/App Proficiency  Exercise Control Features 0.0255 -0.0023 Yes No External Regulation × Frequency of Use  Exercise Control Features -0.0187 -0.0628 No Yes Integrated Regulation × Frequency of Use  Exercise Control Features -0.0061 -0.0433 No Yes External Regulation × Length of Ownership  Exercise Control Features 0.0332 -0.0045 Yes No Integrated Regulation × Length of Ownership  Exercise Control Features 0.0213 -0.0025 Yes No Moderation of Exercise Motivations  Data Management Features by Controls Non-Regulation × Age  Data Management Features 0.0329 0.0010 No Yes Identified Regulation × Age  Data Management Features 0.0369 0.0061 No Yes Non-Regulation × Device/App Proficiency  Data Management Features -0.0057 -0.0358 No Yes Identified Regulation × Device/App Proficiency  Data Management Features 0.0404 0.0002 No Yes Non-Regulation × Frequency of Use  Data Management Features 0.0324 0.0040 No Yes Identified Regulation × Frequency of Use  Data Management Features -0.00004 -0.0356 No Yes Non-Regulation × Length of Ownership  Data Management Features 0.0018 -0.0173 Yes No Identified Regulation × Length of Ownership  Data Management Features 0.0196 -0.0025 Yes No

MIS Quarterly Vol. 43 No. 1—Appendices /March 2019 A11 James et al./Exploring Associations Between Exercise Motivation & Fitness Technology

References

Bostic, T. J., Rubio, D. M., and Hood, M. 2000. “A Validation of the Subjective Vitality Scale Using Structural Equation Modeling,” Social Indicators Research (52:3), pp. 313-324. Churchill Jr., G. A. 1979. “A Paradigm for Developing Better Measures of Marketing Constructs,” Journal of Marketing Research (16:1), pp. 64-73. Davis, F. B. 1964. Educational Measurements and Their Interpretation, Belmont, CA: Wadsworth Publishing Company. Fornell, C., and Larcker, D. F. 1981. “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error,” Journal of Marketing Research (18:1), pp. 39-50. Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., and Tatham, R. L. 2006. Multivariate Data Analysis, Upper Saddle River, NJ: Pearson Prentice Hall. Larose, D. T., and Larose, C. D. 2015. Data Mining and Predictive Analytics (2nd ed.), Hoboken, NJ: Wiley. Lowry, P. B., and Gaskin, J. 2014. “Partial Least Squares (PLS) Structural Equation Modeling (SEM) for Building and Testing Behavioral Causal Theory: When to Choose It and How to Use It,” IEEE Transactions on Professional Communication (57:2), pp. 123-146. MacKenzie, S. B., Podsakoff, P. M., and Podsakoff, N. P. 2011. “Construct Measurement and Validation Procedures in MIS and Behavioral Research: Integrating New and Existing Techniques,” MIS Quarterly (35:2), pp. 293-334. Markland, D., and Tobin, V. 2004. “A Modification of the Behavioral Regulation in Exercise Questionnaire to Include an Assessment of Amotivation,” Journal of Sport and Exercise Psychology (26:2), pp. 191-196. Neter, J., Kutner, M. H., Nachtsheim, C. J., and Wasserman, W. 1996. Applied Linear Statistical Models (4th ed.), Chicago: McGraw- Hill/Irwin. Oppenheimer, D. M., Meyvis, T., and Davidenko, N. 2009. “Instructional Manipulation Checks: Detecting Satisficing to Increase Statistical Power,” Journal of Experimental Social Psychology (45:4), pp. 867-872. Pavlou, P. A., Liang, H., and Xue, Y. 2007. “Understanding and Mitigating Uncertainty in Online Exchange Relationships: A Principal–Agent Perspective,” MIS Quarterly (31:1), pp. 105-136. Peterson, R. A. 1994. “A Meta-Analysis of Cronbach's Coefficient Alpha,” Journal of Consumer Research (21:2), pp. 381-391. Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., and Podsakoff, N. P. 2003. “Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies,” Journal of Applied Psychology (88:5), pp. 879-903. Ryan, R. M., and Frederick, C. 1997. “On Energy, Personality, and Health: Subjective Vitality as a Dynamic Reflection of Well Being,” Journal of Personality (65:3), pp. 529-565. Straub, D. W. 1989. “Validating Instruments in MIS Research,” MIS Quarterly (13:2), pp. 147-169. Straub, D. W., Boudreau, M.-C., and Gefen, D. 2004. “Validation Guidelines for IS Positivist Research,” Communications of the Association for Information Systems (13:1), pp. 380-426. Wilson, P. M., Rodgers, W. M., Loitz, C. C., and Scime, G. 2006. “It’s Who I Am Really! The Importance of Integrated Regulation in Exercise Contexts,” Journal of Applied Biobehavioral Research (11:2), pp. 79-104.

A12 MIS Quarterly Vol. 43 No. 1—Appendices/March 2019 Copyright of MIS Quarterly is the property of MIS Quarterly and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.