A tale of two frames: Exploring the role of framing in the use discontinuance of volitionally adopted technology

The discontinuance of volitional IS (i.e., information systems adopted, used and discontinued at will) has recently attracted remarkable attention from academics and practitioners alike. However, most research to date has been ahistorical. Ignoring the temporal progression can be problematic when the phenomenon under investigation is dynamic and evolving. To balance this, we adopt a stage modelling approach to understand the process ending with the technology use being discontinued by users of a popular crowdsourcing platform. Two questions guided our investigation: (1) Why do users discontinue using an IS they have volitionally adopted and used? (2) How does IS discontinuance occur over time in such context? We develop a stage model demonstrating that five stages are critical in understanding IS discontinuance: IS framing, goal pursuit, frame disruption, dormancy and quitting, after which possible switching denotes a new cycle. Furthermore, we identify two frames that help us understand why different users interpret and evaluate the technology differently – namely, the gain frame and the hedonic frame. On one hand, a gain frame is linked to the goal of improving one's resources and thus directs the user's attention to the technology's instrumental value. On the other hand, a hedonic frame is linked to the goal of having fun and thus directs the user's attention to the technology's enjoyment value. But, most importantly, we show that the technology's use lifecycle as a whole from initial use to discontinuance is shaped and guided by the user's dominant frame. Our insights elicit a number of important theoretical and practical implications.

attention to the technology's enjoyment value. But, most importantly, we show that the technology's use lifecycle as a whole from initial use to discontinuance is shaped and guided by the user's dominant frame. Our insights elicit a number of important theoretical and practical implications.
IS discontinuance, qualitative study, stage model, volitional IS
In volitional contextsthe main interest of this articlediscontinuance is far less understood. Since end users have full discretion to use or abandon a given technology at will, IS discontinuance has emerged as one of the most troublesome issues for modern digital services targeting individual users. Market reports indicate that high churn rate and low user retention are problems that many if not most digital platforms struggle with. For instance, while many media reports praised Twitter for the massive growth rate of its user base during its early years (1382% between 2008Ostrow, 2009), it has also been shown that nearly 90% of Twitter users who joined in 2012 had already stopped tweeting by 2013 (Koh, 2014). In a similar fashion, recent usage reports suggest that Facebook is experiencing a steep rise in the number of users who are abandoning the platform (Anderson & Jiang, 2018;Hsu, 2018). Retaining users can be even more difficult for businesses less affluent than the likes of Facebook and Twitter. Usage statistics show that the average retention rate for smartphone applications (i.e., mobile apps) on the Android platform three months after signing up is less than 5%, meaning that more than 95% of new users of an average app will quit the use within the first 90 days (Chen, 2015). These observations highlight the importance of understanding how and why users change their minds about IT/IS products and services that they once chose to adopt and use.
Surprisingly, despite its relevance and its impact on practice, the research on IS discontinuance remains for the most part static and attributes-focused: most IS studies to date treat discontinuance as a static construct (i.e., a dependent variable), and the ultimate aim of most of these studies is to identify the 'objective attributes' or characteristics that make an IT/IS more or less likely to be discontinued or rejected (Cenfetelli, 2004). While insightful, this perspective leaves important questions unanswered, including how and why users decide to abandon an IS/IT they once adopted. Questions of this nature are better suited to a process approach since it accommodates change and development (Langley, 1999) and has the capacity to construct a view based on the analysis of users' accounts of their lived experiences (Pentland, 1999;Ramiller & Pentland, 2009). Along these lines, the objective of this study is, first and foremost, to understand the process (i.e., stages) leading to the discontinuance of a volitionally adopted technology. More specifically, we set out to answer the following research questions: (1) Why do users discontinue the use of an IS they have volitionally adopted and used? (2) How does ISdiscontinuance occur over time in such context? To answer these questions and to capture the dynamic nature of use and discontinuance behaviour, we conducted a longitudinal study over a span of two years in the context of a crowdsourcing platform called Scoopshot, which by design reflects a business tool (i.e., utilitarian IT/IS) but is viewed by many users as an application for leisure time (i.e., hedonic IS/IT). The rich insights we gained from interviewing the study participants led to several important conclusions, including that different users evaluate their usage experience from very different perspectives, their goals for the use and perspectives on the technology may change over time and the process from initial exposure to discontinuance involves several stages. Aided by the theoretical lens of framing (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007;Orlikowski & Gash, 1994), we develop a stage model that explains why and how the content of these stages can be radically different depending on the users' goals and framing of the IS/IT in question.
The remainder of the article is organised as follows. In the next section, we first present a review of earlier research on IS/IT discontinuance, and then in Section 3, we introduce our research approach and elaborate the theoretical framework adopted for this study. In Section 4, we present the key findings of the study, which in turn will serve as the building blocks for discussion and theoretical elaboration in Section 5. Finally, Section 6 is dedicated to outlining the concluding remarks of the study.

| LITERATURE REVIEW
Two perspectives may be used to define IS discontinuance. First, from the technology perspective, discontinuance reflects the state of not being in use. This implies that when an IS/IT is not used, it becomes a mere assembly of hardware and software, and it is no longer an information system (Paul, 2010). Second, from the user perspective, IS discontinuance represents a decision to temporarily or permanently refrain from using a given technology. This decision, however, varies in meaning depending on the context and temporal stage at which this decision is taken (Soliman & Rinta-Kahila, 2020). In this study, we focus on the user perspective and discontinuance as a user's postadoption decision/behaviour.
To gain a broad perspective on how IT/IS discontinuance is discussed in the extant literature, we engaged in extensive reading in the IS domain, as well as in other relevant fields, such as service (Hogan et al., 2003;Rosenbaum & Wong, 2015), marketing (Butcher et al., 2020;Hand et al., 2009), communication (York & Turcotte, 2015) and education research (Geri & Naor-Elaiza, 2008). Appendix Table A1 provides a list of the reviewed articles and some of the most important insights they provide. Below, we present a summary of this body of literature with an eye on contextual and ontological dimensions.
While the notion of context reflects a wide variety of facets, such as geographical, cultural and organisational characteristics (Davison & Martinsons, 2016), situational characteristics (Mallat et al., 2009) and technological characteristics (Salo & Frank, 2017), in our review, we focus on a socio-technical facet of context that is commonly captured in IS research by a distinction made between organisational IS (i.e., technologies developed for organisations to be used by its members) and non-organisational IS (i.e., technologies developed to be used by the general public at their own discretion). Even though not all use contexts of organisational IS are mandatory and not all nonorganisational contexts are volitional, this classification is helpful in organising the key insights from the literature.
The ontological dimension, in turn, reflects here what the conception of IS discontinuance in a given study is. In effect, this dimension captures whether discontinuance is conceptualised as the outcome of a process by which a user goes through qualitatively different stages, or as a (dependent) variable that varies only in quantity (e.g., intention to discontinue using an IS on a scale of 1-to-7) but preserves its meaning across time. Table 1 provides an overview of the reviewed literature based on these two dimensions.

| Discontinuance of organisational IS/IT
Nearly one-third of the identified IS discontinuance articles focus on organizationally oriented IS. This body of research makes a distinction between the technology buyers (typically referred to as 'managers' or the 'organisation') and technology users (typically referred to as the organisational members such as 'employees'). The most dominant narrative here is that discontinuance represents a conflict between what the management envisions and what employees do, and the studies typically explore mechanisms that managers can utilise to steer the employees in the desired direction.
In many cases, the emphasis is on understanding why employees reject or resist technologies the organisation has invested in. Most notable antecedents include perceived ease of use and usefulness (Aggarwal et al., 2015;Polites & Karahanna, 2012;Recker, 2016), compatibility (Geri & Naor-Elaiza, 2008;Tully, 2015), supportability (Furneaux & Wade, 2010, as well as trialability and flexibility (Tully, 2015). We also identified three articles that aim at capturing the IS discontinuance from a process perspective (Mehrizi et al., 2019;Miller et al., 2009;Pollard, 2003) and draw our attention to the temporal complexity inolved in the IS discontinunace process within organisations.
T A B L E 1 Overview of the literature on IS/IT discontinuance a Organisational IS Non-organisational IS IS Discontinuance as a Variable: IS discontinuance treated as a dependent variable 13 out of 50 articles The dependent variable is captured either quantitatively, typically measured on an 1-to-7 scale, such as discontinuance intentions, switching intentions, and so on (e.g., Polites & Karahanna, 2012), or qualitatively, as a binary decision, such as reject, quit, unadopt, and so on (e.g., Goode, 2005). Methods used: Surveys (Furneaux & Wade, 2017), field experiments (Aggarwal et al., 2015), interviews (Charki et al., 2017), as well as conceptual modelling (Furneaux & Wade, 2010). IT/IS studied: Office automation tools (Cooper, 1991), organisational inventory IS (Recker, 2016), inter-organisational IT (Power & Gruner, 2015).
IS Discontinuance as a Process: Discontinuance treated as a consequence of temporally ordered stages 3 out of 50 articles Identified stages depend on the theoretical perspective. For instance, from a pathbreaking perspective, the discontinuance of a legacy system is seen as a sequence stages, such as, realisation, reversion, handover, then marginalisation (Mehrizi et al., 2019). Methods used: Interviews (Pollard, 2003) and participant observations (Miller et al., 2009). IT/IS studied: Legacy IS (Mehrizi et al., 2019), CAD in the construction domain (Miller et al., 2009) and group support systems (Pollard, 2003).
3 out of 50 articles Identified stages depend on the theoretical perspective. For instance Tsohou et al. (2020) theorise that following technical failure, users of IT-based service go through the five stages of blaming, bypassing, tolerating, abandoning and overcoming. Methods used: Interviews (Cho, 2015;Tsohou et al., 2020) and focus groups (Butcher et al., 2020). IT/IS studied: Facebook (Cho, 2015), ITbased services in general (Tsohou et al., 2020) and mobile games (Butcher et al., 2020). a A more detailed summary of the identified 50 articles on IS/IT discontinuance is in Appendix Table A1. b Among the 31 articles categorised as non-organisational use-context are four articles where the reporting of intended usecontext has been rather ambiguous (Cenfetelli, 2004;Spiller et al., 2007;Whitacre & Rhinesmith, 2016;Zhu & He, 2002).

| Discontinuance of non-organisational IS/IT
With IS/IT targeting the general publicthe main focus of our researchthe individual user is ultimately solely responsible for her or his own choices and behaviours to the degree that they serve her or his self-set goals. Two-third of the identified articles (34 out of 50) in our review focus on non-organisational IS, the majority of which (31 out the 34) conceives IS discontinuance as a variable. Among these, we could identify two main narratives: discontinuance of social networking sites (specifically) and the discontinuance of IT innovations (more generally). On the one hand, in studies on the use discontinuance of social networking sites (mainly Facebook), discontinuance is typically presented as a form of coping with (and adapting to) negative encounters and disturbances, or what is generally described as the 'dark side of IT' (Tarafdar et al., 2015). From this perspective, the likelihood of discontinuing social media is associated with negative and undesired experiences, such as addiction (Turel, 2015;Vaghefi et al., 2020), guilt (Turel, 2016), technostress (Luqman et al., 2017;Maier, Laumer, Weinert, et al. 2015;Maier, Laumer, Eckhardt, et al. 2015), fatigue (Ravindran et al., 2014), as well as social media overload (Maier, Laumer, Weinert, et al. 2015;Maier, Laumer, Eckhardt, et al. 2015;Zhang et al., 2016). On the other hand, the main objective of studies focusing on IT-based service/innovation is to explain or predict the discontinuance (e.g., rejection and un-adoption) of IT-based services and innovations beyond the narrow focus of social media, such as the Internet service (Kim, 2011), broadband (Whitacre & Rhinesmith, 2016), mobile data subscription (Kim et al., 2008), web browsers (Bhattacherjee et al., 2012), and online shopping platforms (Hand et al., 2009). These studies generally aim at identifying the key attributes that facilitate or hinder the adoption of the IT innovation or service. In addition to these predominantly attribute-oriented (i.e., static) studies, a few sets out to shed light on the dynamic nature of the process leading up to a discontinuance decision. For instance, Tsohou et al. (2020) inductively develop a theory that proposes a series of stages that IT service users pass through (i.e., blaming, bypassing and tolerating) after they experience a technical failure, and before they make the final decision to abandon the service.
Our review of IS/IT discontinuance literature reveals that while we know a great deal about the significant factors that explain variance in the 'users' intentions to discontinue IT' construct, very little is known about the process or the stages leading up to (actual) discontinuance. Two approaches to explaining phenomena are worth noting briefly here: a static approach (i.e., variance, stage-less theories) and a dynamic approach (i.e., process or stage theories). In the static approach, the phenomenon of interest (e.g., technology use) is composed of entities (e.g., dependent and independent variables) that maintain a fixed meaning and identity through time, even if their values change (Van de Ven & Engleman, 2004). For instance, in the life course of technology usage, the attribute 'usefulness' is assumed to hold the same meaning for all actors and across time whether the reported values is 1 out of 7 or 7 out of 7 (see e.g., Burton-Jones et al., 2015). By contrast, in the process or stage perspective, the phenomena of interest are assumed to be developing and changing, actors are assumed to live qualitatively different experiences in the form of stages, and entities (e.g., factors, events, attributes, etc.) are expected to have different roles and exert different influences at different stages (Weinstein et al., 1998). For instance, 'user dissatisfaction' emanating from a disconfirmed expectation for a first-time user of a given technology differs from 'user dissatisfaction' emanating from evolution of a long-time user (e.g., maturing or reprioritizing life goals). Although 'user dissatisfaction' might be partially responsible for 'discontinuance' in both cases, the meaning of these entities is different.
Hence, we maintain that understanding the history an IS/IT shares with its user helps in distinguishing between the different use experiences without reducing them to simple notions of non-use (intentions). We will address this gap with the help of dynamic lenses provided by process and stage modelling where temporality is at the core (Langley, 1999;Langley et al., 2013;Pentland, 1999).

| RESEARCH APPROACH
The work at hand belongs to the interpretation-centric, inductive research genre (Sarker et al. 2018). More specifically, we aim to develop a theory-informed narrative explaining the process of IS discontinuance in the studied context. The resulting narrative corresponds to what Gregor (2006) calls 'Type II theory', a conception of theory that promotes greater insights into the phenomenon of interest by advancing an explanation of 'how and why things happened in some particular real-world situation' (p. 624). Hence, the study participants' narrated experiences with the IS, as well as our own personal experiences, are at the core of our analysis (Langley, 1999;Pentland, 1999). Although interpretive research is not interested in testing theory in the verificationist sense (Klein & Myers, 1999), an existing theory can provide a lens through which to make sense of the world or phenomenon studied and afford a scaffold for further development. Our interpretations, and the theoretical model we develop, are informed by framing theory and its implications on individuals' interpretations and behaviour (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007;Orlikowski & Gash, 1994).
This conception of theory is consistent with the theory as narrative view (Pentland, 1999), in which 'explanation requires a story and [...] stories can be understood as process theories' (p. 717). Narrative-oriented inquiry belongs to a retrospective genre of research (Cox & Hassard, 2007), and, as such, it is acknowledged that a personally narrated account (i.e., narratives-as-told) is not a perfect account of undisputed objective reality but a representation of someone's point of view (Bold, 2012). Work of this nature typically utilises process or stage theorising and fits well within the interpretive paradigm, where the theory itself is an end product and is not expected to lead to predictive, deterministic theory (Gregor, 2006).
Whereas the variance perspective might be well suited to answer 'what' questions (e.g., what are the most statistically significant antecedents of IS discontinuance intentions?); the process perspective is more suited to answer 'how' questions (e.g., how does IS discontinuance occur over time?). (Langley et al., 2013;Van de Ven & Engleman, 2004). In the specific domain of IS research, the process perspective is often found to be more relevant when 'the agency of the users is given much more importance' (Karjalainen et al., 2019, p. 697). It also takes into account temporality and 'focuses empirically on evolving phenomena, and it draws on theorising that explicitly incorporates temporal progressions of activities as elements of explanation and understanding' (Langley et al., 2013, p. 1). This approach matches well with our aim to understand why users discontinue the use of an IT/IS they have volitionally adopted and theorising about how IS discontinuance occurs over time. Progression assumed in stage theorising, however, does not imply that the development is inevitable or irreversible as understood for example in biological development theories (Weinstein et al., 1998). Indeed, whereas one user might decide to discontinue an IS immediately after the first trial, others may move on to further stages and become active (i.e., continued) users. Yet, at a later point in time, things could change and these active users might start contemplating quitting a once enjoyed technology. As such, in line with Weinstein et al. (1998), stage-based theorising at its core recognises that 'different issues are important at different times, and at any point, the process can be halted, reversed, or abandoned' (p. 291).
Considering the interpretive nature of our study, and the theoretical ambition discussed earlier (Gregor, 2006), our main objective is to understand a specific phenomenon in a specific context rather than to make causal inferences from sample to population (Seawright & Gerring, 2008). Such orientation reflects what is often described as an ideographic (context specific), rather than a nomothetic (general law), type of research (Gerring, 2006;Lee, 1991).

| The theoretical concept of framing
The premise of framing is simple, but its implications are vast. In everyday life discourse, we often talk about the notion of frame of reference and the idea that one might see matters differently depending on how a situation or a problem is 'framed' and what 'labels' we attach to these situations. Orlikowski and Gash (1994) introduced the notion of technological frames to IS to explain how different organisational groups perceived and interpreted the same technology differently, thus leading to different outcomes. In that work, technological frames were defined as 'subset of members' organisational frames that concern the assumptions, expectations and knowledge they use to understand technology in organisations' (Orlikowski & Gash, 1994, p. 178). The roots of framing, however, has been attributed to various influential writings, mainly, Wittgenstein's (1953) notion of family resemblance (Orlikowski & Gash, 1994;Rosch, 1978), Berger and Luckmann's (1967) work on social construction (Orlikowski & Gash, 1994), Bijker et al.'s (1987) work on social construction of technology (Davidson, 2002;Davidson & Pai, 2004), as well as Goffman's (1974) work on frame analysis (Druckman, 2001;Su, 2015). The core premise of framing theory, as highlighted by Chong and Druckman (2007), is that 'an issue can be viewed from a variety of perspectives and be construed as having implications for multiple values or considerations' (p. 104). In this sense, framing is often described as 'the process by which people develop a particular conceptualization of an issue or reorient their thinking about an issue' (ibid). The powerful implication of this simple idea has led to the popularity of framing theory in various fields of research, including psychology (Mervis & Rosch, 1981;Rosch, 1978), communication science (Entman, 1993), political science (Chong & Druckman, 2007;Druckman, 2001) and organisation science (Weick et al., 2005) as well as in information systems science (Davidson, 2002;Hsu, 2009;Orlikowski & Gash, 1994;Su, 2015).
To explain the underlying mechanism of frame formation at the individual level, Lindenberg (2001) complements framing theory by emphasising the crucial role of competing goals on dominant frames. This body of work is often referred to as goal framing theory (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007). Goal framing theory suggests that in a given situation individuals may be faced with a number of goals that compete to dominate the individual's limited cognitive resources, or simply 'compete for the privilege of being on centre stage' (Lindenberg, 2001, p. 322). The winning goal dominates the frame, which in turn shapes how we see, interpret and evaluate what we experience in our environment. The theory points to three alternative frames that may be relevant in different contexts: (a) a hedonic frame, which is often linked to the goal 'to have fun' or 'to feel better'; (b) a gain frame, which is often linked to the goal 'to improve one's resources'; and (c) a normative frame, which is often linked to the goal 'to act appropriately' (p. 339). The normative frame is more relevant to contexts where moral quandaries are salientsuch as environmental protection behaviour (Lindenberg & Steg, 2007) or social order in public spaces (Keizer et al., 2008). Therefore, in our work, we focus on the hedonic and gain frames.
As will become evident in our findings, recognising and understanding the role of goals are critical because they serve two functions. First, goals motivate people to develop strategies and take actions that help them to achieve these goals. Second, a goal acts as a standard for judging satisfaction (Lunenburg, 2011). Indeed, research on human motivation acknowledges that 'to say that one is trying to attain a goal of X means that one will not be satisfied unless one attains X' (Locke & Latham, 2002, p. 709). We argue that use of technology is always goal orientedespecially in volitional contextseven if the goal is not clearly defined and consciously acknowledged by the user. For example, an individual might come across a new smartphone app that s/he decides to try. Her or his initial goal could be to satisfy curiosity or simply to pass time while waiting for a bus; but after the first try, the goal might develop into something else, say, to provide entertainment.
To summarise, informed by goal framing theory, we argue that a given IS/IT can be viewed from a variety of perspectives depending on what goals they serve for a given user. Furthermore, based on this theoretical lens, we realise that different users, while performing the same activity, may experience and evaluate the use of the same technology differently. In other words, what a user experiences from a hedonic frame (with a hedonic goal and motivation) may constitute a different reality to what a user sees from a gain frame (with a utilitarian goal and motivation).

| Empirical context of the study
The empirical context of our study comprises the Scoopshot platform (https://www.scoopshot.com/) and its users.
Scoopshot is a crowdsourcing platform dedicated primarily to the trade of crowd-or user-generated content (UGC). With the first author having been a participating user in the platform and given the unique opportunity to access the organisation and its management (Eisenhardt & Graebner, 2007), Scoopshot was initially chosen as the research setting to explore the specific phenomenon of crowdsourcing post-adoptive behaviour. Long and indepth exposure to the case and its community inspired us to expand the research scope to include discontinuance as an integral part of the IS usage lifecycle to understand how and why discontinuance happened in the specific context of Scoopshot.
During the empirical research period, Scoopshot was successful in attracting a large user base. Having been available for Apple, Android and Windows smartphones, it had been installed by over 500 000 users across 177 countries. The platform facilitates and manages the exchange of UGC, such as photos and videos, between content seekers (e.g., media agencies) and the Scoopshot users. Scoopshot users, who are also called Scoopshooters, participate either by responding to a pre-set task by a seeker firm or by submitting content that they believe to be publishing-worthy. The submitter may get compensated if the content is chosen to be purchased by a seeker. Sold content is typically used for publishing purposes in different visual media (e.g., TV, newspapers or online media). The company's CEO summed up how Scoopshot works: 'Media can send targeted tasks to the Scoopshooters anywhere in the world. In other words, I select a region and I direct and send a push notification to all the Scoopshooters in that region asking them to take photographs of an event. At the same time I tell them how much I'm willing to reward them for that photo' (Salz, 2012).
For the empirical research duration, the Scoopshot platform did not facilitate online social impression capabilities or features (e.g., liking, following, commenting, etc.). This was an intentional choice by Scoopshot's management to position themselves as a 'serious service'. Scoopshot's COO told us, 'We are not doing social media; we are helping media to become social'. In other words, the management at that point did not envision Scoopshot to be competing with other photo-based social media platforms, such as Instagram and Snapchat. Instead, Scoopshot was designed to offer a unique opportunity for media organisations to utilise the crowd as a constant source of fresh content and for the crowd members to be compensated for their efforts.

| Data collection
In addition to the initial interviews with Scoopshot's management, our primary data consist of 'conversation-as-narrative' (Bold, 2012) produced through semi-structured interviews with various actors in the studied context. Interviews are central to qualitative research 'since they enable researchers to step back and examine the interpretations of their fellow participants in some detail' (Walsham, 1995, p. 78). Altogether, 30 interviews were conducted over the span of two years with 20 informants: two Scoopshot top management members and 18 Scoopshot users.
The 18 Scoopshooters included 15 males and three females of ages ranging between 17 and 46 years and with different educational and professional backgrounds (see Appendix Table A2 for details). The first round of interviews took place between April 2012 and May 2013, with the emphasis on understanding the process by which the participants became active users. The second interview round was conducted between May 2014 and December 2014, with the emphasis on understanding the process by which active participation transformed into discontinuance.
Since our focus has been on post-adoptive behaviour(s), our selection rationale was purposive in nature rather than random (Guest et al., 2006). Specifically, considering the non-probabilistic design of our research, the guiding principle for eligibility was active participation and discontinuance well beyond the experimentation or initial use stages. We communicated this objective with our contacts at Scoopshot, who in turn provided us with a list of user contacts who fit the description. Of the 18 Scoopshooters, 10 participated in both interview rounds, five in the first interview round only and three in the second interview round only. All 18 were relevant for the purpose of user profiling, while the 13 participating in the second round of interviews were instrumental in understanding the latter stages of the use lifecycle. Additionally, to enhance our understanding of the user experience, the first author has been a registered user of the service since 2011. He has participated in several announced photography tasks and has also been following the development of the service from a user's viewpoint. Our secondary data include press reviews and a wide range of online materials concerning how Scoopshot is presented and discussed in press releases (see Appendix Table A3 for a list of sources of our secondary data).
Considering the dispersed geographic locations of the Scoopshooters, synchronous computer-mediated communication tools were used to allow for flexibility in choosing the time and place for the interviews. While the core interview themes remained the same for all interviews, some questions were added or omitted depending on the flow of the conversation, and the exact wording and order of some questions differed from one interview to another.
The guiding rule was to allow the participants to reflect on their use experience and to give them a chance to reconstruct their stories based on how they had lived them (Pentland, 1999), with careful attention to avoid imposing any preconceived assumptions. After establishing that the interviewee had actually discontinued his or her use of the Scoopshot app for a period of time, the conversation was organised around elaborating the sequence of events and reasons that resulted in discontinued use. The core questions were used to elicit explanations of how the user went from being an active user to a non-user, about disappointments and features that s/he missed in Scoopshot and about advice on how Scoopshot could be improved (see Appendix Table A4 for an example of the interview protocol).

| Data analysis
In line with our process theorising objective, our analysis framework is influenced by narrative theory (Bold, 2012;Pentland, 1999;Schwarz et al., 2014) and its synthesis of the stratification of narratives (see Figure 1). From a structural point of view, narratives are seen to be composed of multiple levels. At the surface level, text (or narrative-astold) represents the most descriptive account of people's experiences. This is typically the level at which we observe what people say, conduct interviews and generally collect data (Pentland, 1999). Additionally, at this level, we typically expect thick descriptions and accept that each individual experience is unique. Below (narrated) text reside stories, which are more abstract than text and occupy a deeper layer than text in narrative theory. Whereas text preserves the uniqueness of each individual experience, stories preserve shared perspectives. As such, stories may be seen as denser and more abstract conceptions of the observed data (Pentland, 1999). In a theory-building exercise such as ours, stories are essential because they provide the foundational building blocks of theorising. The next level down is the fabula, or those aspects of stories that do not change from one telling to another. Fabula can be thought of as a level of abstraction that preserves the essence of stories and, as such, provides the best example of what is often referred to as abstract process or stage models.
The dynamic nature of the process theorising perspective led us to utilise a combination of both static and dynamic data analysis approaches. By static analysis, we mean the actions we took to identify the key themes in our data (i.e., the building blocks) based on an iterative process of within-case and cross-case analysis (Braun & Clarke, 2006), where the case here refers to an individual user. Thematic analysis centres on the cyclic iteration between coding the textual data, searching for themes, reviewing and refining themes and developing an overall story that accentuates the most important themes (Braun & Clarke, 2006). Thus, thematic analysis affords the procedural flexibility needed especially when dealing with narratives which contain 'partial stories' (Bold, 2012) as well as different conversation styles in terms of, for example, a narrative's richness or vividness. This notion was clear in our data; some participants were highly lucid and elaborate in telling their stories, while others were much briefer during the conversation. All interview transcripts were coded using both semantic and latent themes (Braun & Clarke, 2006), reflecting the distinction that Van Maanen (1979) makes between first-order and second-order concepts. Specifically, first-order concepts in our analysis are seen to be more reflective of the empirical data and represent as close as possible the narratives as generated by the participants themselves. Second-order concepts (or latent themes), in turn, are more analytic in nature as they reflect the analysts' interpretation of certain concepts, which are influenced by the theoretical lenses we adopt. For instance, we consider the code 'Framing' to be analytic, second-order because, consistent with goal framing theory (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007), it provides an adequate label for the process by which the same technology is perceived differently by different users (for coding examples, see Appendix Table A5).
To complement this coding procedure, and to capture the dynamic nature of the process, we maintained a database to tabulate the key events experienced by each participant in a temporal progression manner. The three logically ordered stages of a typical IS use lifecycleinitial use, continued use and discontinuanceserved as the foundation for our temporal bracketing strategy (Langley, 1999) and helped us capture the activities pertaining to each stage. Key insights from each interview were documented in this database with a high-level timeline of each user's own story following a logical temporal sequence (e.g., the first time the interviewee heard of Scoopshot, the first time s/he used it, the key motivations for initial use, the key motivations for continued use, the main reasons for disappointment, the time span s/he had the app before deleting it, and eventually the time when s/he decided to remove it from her/his smartphone). Even though this procedure was predominantly descriptive in nature, it was in fact the first step towards the more theoretical analysis that followed (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007). In addition to the identification of the stages, we paid careful attention to the salient stage-specific experiences (Schwarzer, 2008;Tsohou et al., 2020;Weinstein et al., 1998) as well as the triggers that influenced users to either move forward or backward between stages, which we called forward and relapse triggers (Schwarzer, 2008;Tsohou et al., 2020;Weinstein et al., 1998).
The iterative data-analysis process continued until we reached an adequate level of theoretical saturation, where considering more data would not lead to new or different understanding. Figure 1 illustrates our data collection and analysis framework.

| FINDINGS
The distinction our analysis framework ( Figure 1) makes between the empirical and the theoretical corresponds to the demarcation narrative theory makes between stories and fabula. Specifically, whereas the empirical level corresponds to the richness of narratives in terms of participants' accounts of their stories, the theoretical level F I G U R E 1 Analysis framework corresponds to the parsimony of fabula and extrapolating the mechanisms at play (Pentland, 1999). Accordingly, we first present our findings by focusing on the surface level of narratives, that is, by accentuating and contextualising the participants' voices and their stories (as told). This descriptive account will then serve as the foundation for the theoretical elaboration that we present in the section that follows, where the emphasis will shift towards theoretical explanation. In terms of reporting, we find Soh and Markus's (1995) advice regarding backward narrating (i.e., process synthesis) very useful. Starting from the end, we first report our findings regarding discontinuance itself and the different forms of discontinuance we identified. Next, we present the dominant reasons that users attributed to explaining why they discontinued using Scoopshot. We also highlight the important role of competing alternatives in making that decision. Finally, we show how goal framing is the starting point based on which users interpreted, used and evaluated Scoopshot.

| Different forms of IS discontinuance
The first key finding of our study supports a recent development in the literature suggesting that IS discontinuance is a multifaceted phenomenon constituting distinct stages and forms (Soliman & Rinta-Kahila, 2020;Tsohou et al., 2020). In the specific context of our study, we identify two key stages, namely, dormancy and quitting.
Whereas quitting denotes the final stage of decisively abandoning Scoopshot, dormancy reflects an intermediary stage between use and non-use (e.g., passive use and lurking). The following illustration ( Figure 2) summarises the Scoopshot app usage lifecycle for the participants that have discontinued their use.
All discontinuing users in our study had gone through dormancy or were still in this stage at the time of the second interview, suggesting that dormancy is a critical stage in understanding post-adoption discontinuance. For some Scoopshooters, going dormant may be described as a conscious decision which involves a level of deliberation. Even though these users were not contributing content to the platform, they were quietly lurking. To summarise, these empirical findings point to the importance of distinguishing between the stages of dormancy and quitting in post-adoption discontinuance. Both of these stages, in turn, constitute stage-specific activities and decisions. On one hand, dormancy reflects an intermediary stage between use and non-use, where users may be lurking (in anticipation of possible return) or simply unintentionally and indecisively inactive. Quitting, on the other hand, reflects the concluding stage where a user either simply abandons altogether the purpose that the app once served or decides to abandon the app in favour of another IS (i.e., switching).

| Reasoning behind discontinuance
To better understand post-adoption discontinuance, we probed deeper into why users had stopped using Scoopshot.
Two main findings emerged: one highlights different sources of dissatisfaction with Scoopshot, and the other reveals the catalytic role of attention to alternatives.

| Different sources of dissatisfaction
Our findings suggest that dissatisfaction with Scoopshot as a primary reason for discontinuance derives from a misalignment between how the user frames Scoopshot and how their experiences actually unfold. Disruption to the gain frame is rooted Scoopshot's disappointing utility, while disruption to the hedonic frame is rooted in Scoopshot's lack of feedback and networking.
First, we discuss the findings regarding disruption to the gain frame is linked to the goal of 'improving one's resources' (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007). As a source of dissatisfaction, disruption to this frame reflects Scoopshot's inability to meet the user's envisioned instrumental value. To Jari and Kaisa, Scoopshot was only as useful as its ability to provide tasks that would lead to sales of their photos. Jari, for instance, thought that 'things have got worse with this app [Scoopshot]… First, I thought this is [an] easy way to gain some extra income and there was lots of missions. But for the last year, missions have mostly been weird and there hasn't been so many of them'. To Jari, 'weird missions' (e.g., 'send us a photo of your BBQ!') reflect a disruption to his framing of Scoopshot as a business-oriented or serious IT application. Interestingly, this finding also demonstrates how financial rewards can act as a two-edged sword. On one hand, financial rewards work as a very effective approach to attract users to a crowdsourcing service for as long as there are enough resources to reward users' acceptable contributions.
However, once the service gains popularity, it is inevitable that the mechanisms of competition will take over, and thus many contributors will be unjustifiably unrewarded. In a similar vein, Joost's participation was predominantly utility-driven although different in nature. For him, Scoopshot was a marketing channel, with its instrumentality determined by its ability to get his family business photos viewed by the public. Even though he had earlier successfully sold some photos via Scoopshot, as it consistently failed to deliver on his main objective, he switched to an alternative service (i.e., a newspaper app with a narrower scope on local matters).
Second, we discuss our finding regarding disruption to the hedonic frame is linked to the goal of 'having fun' or 'feeling better' (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007

| Seeking alternatives
One of the key findings of our study is the realisation that dissatisfaction alone was not sufficient to immediately trigger a decision to quit Scoopshot, as would be expected by most IS discontinuance research (Bhattacherjee, 2001;Najmul Islam, 2014). In our case, dissatisfaction encouraged many Scoopshooters to pay close attention to available alternatives. Consistent with Kim and Son (2009) To summarise, in order to understand the users' reasons to discontinue using Scoopshot, we predominantly focused on why the study participants made the discontinuance decision in its various forms (e.g., dormancy, quitting, switching). Two critical findings emerged. First, disappointing utility, lack of feedback and lack of social networking emerged as the main sources of dissatisfaction with Scoopshot. More specifically, those users who focused on Scoopshot's joyful experience were more disappointed with the lack of hedonic features (e.g., 'likes' and connecting with friends) than with Scoopshot's instrumentality (e.g., earning money). By contrast, those who focused on Scoopshot's serious experience were more disappointed in Scoopshot's perceived utility than with its hedonic value.
Second, attention to alternatives played a critical role in triggering a transition from dormancy to quitting. The user's ability to find a suitable alternative facilitated a discontinuance decision, while the inability to find one prolonged Scoopshot's usage span, despite not being fully satisfied with it. We emphasise the notion of suitability to stress the fact that different users had drastically different perceptions of what constituted a suitable alternative to Scoopshot.
For those who focused on Scoopshot's joyful experience, in their search for alternatives, hedonism was primary. For those who focused on Scoopshot's instrumental 'serious' experience in their search for alternatives, utility was primary.

| Different frames, different experiences
Differences in the identified sources of dissatisfaction, and perceptions of suitable alternatives, led us to the realisation that understanding the process of discontinuance begins with understanding the goal a user envisions for the focal technology. The distinction goal framing theory makes between hedonic and gain frames (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007) helped us categorise our participants into two distinct groupsinstrumentalists and hobbyists (see Appendices 5.1 and 5.2).
Instrumentalist users see Scoopshot through a gain frame (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007). They are mainly driven by Scoopshot's instrumental or utilitarian value, with their main motivation stemming from the possibility to receive financial rewards and/or to utilise the platform to advance their career. The instrumentalists appear to consider Scoopshot as a 'serious' business application and accordingly attach more value to its utility than to its aspects related to having fun, which would be seen by some of them as 'weird'. To the instrumentalist users, features such as 'liking' and 'sharing', highly valued by the hobbyists, make no difference either to their perception of or their participation in Scoopshot tasks. In fact, instrumentalists seem to adopt the frame intended by Scoopshot's management, who described Scoopshot as a 'serious service' that is trying to help media companies become social rather than offering social media services themselves. The primary use motivations and the main sources of dissatisfaction for the instrumentalist users in our sample are illustrated with example quotes in  Lindenberg, 2001;Lindenberg & Steg, 2007): they are primarily motivated by Scoopshot's entertainment value and enjoy photography for its own sake. They find more appeal in social aspects (i.e., social aspects of photography) than in financial rewards or career-related purposes. The hobbyists' primary use motivations as well as their main sources of dissatisfaction are illustrated with example quotes in Table 3.
The fact that these two framings reflect the two contrasting perspectives for making sense of Scoopshot does not, however, mean that any user would necessarily be driven by a pure set of hedonic or utilitarian motivations.
Rather, an interplay between the various motivations is expected while prioritising some motives over the others. This is well exemplified by Jackie, a hobbyist who predominantly enjoyed Scoopshot 'just for fun', but who also The prioritised motivation and goals determine the user' dominant frame (Lindenberg, 2001), which for Jackie is hedonic.

| DISCUSSION
In this section, we first outline our proposed model and reflect its relevance to existing research, then we discuss its core theoretical and practical implications, as well as its limitations and boundary conditions.

| Towards a stage theory of discontinuance of volitionally adopted IS
Moving from the surface level of narratives (i.e., text and stories) to the deeper level of fabula and generating mechanisms (Pentland, 1999;Schwarz et al., 2014), we now proceed to the theorising level and propose a stage model that explains the process by which discontinuance of a volitionally adopted technology occurs. This model is the result of an iterative process of confronting empirical data with the literature and relevant theories (see Figure 1), and it conceptualises the process by which the discontinuance of volitionally adopted IS takes place as a five-stage model, as follows: IS framing, goal pursuit, frame disruption, dormancy and quitting, after which possible switching denotes a new cycle (see Figure 3). In Figure 3, dashed arrows reflect theoretically possible paths that were not observed in our empirical data within the data collection time frame.
Stage 1: IS framing. The starting point in our model is IS framing. The salient question reflecting this stage may be articulated as 'does the new IS/IT align with the user' goals'? Consistent with goal framing theory (Lindenberg, 2001;Lindenberg & Steg, 2007), the outcome of IS framing, whether a hedonic or gain frame, plays a critical role in shaping the overall user experience. Central to the hedonic frame is the goal of enjoyment and having fun, while a gain frame is mainly driven by utilitarian (e.g., financial or career-oriented) objectives. This is in line with F I G U R E 3 Stage model of volitional IS discontinuance framing research in IS which has explained how different social actors may frame, or make sense of, the same technology differently (Hsu, 2009;Orlikowski & Gash, 1994;Su, 2015;Webster & Martocchio, 1993). From a temporal perspective, IS framing typically takes place during the initial use period, when the user is either searching for a potential IS to fulfil an envisioned goal or is experimenting with a new IS out of sheer curiosity. Either way, during this early stage, a user makes sense of the extent to which the IS/IT may serve his or her specific goals ( Van de Ven & Poole, 1995). Our findings reflect this notion clearly: not only do different frames shape how hobbyists and instrumentalists interpret Scoopshot's use purpose differently, but these framings also regulate the overall Scoopshot user experience.
More specifically, framing shaped why Scoopshot was adopted and used (e.g., for what purpose), how it was evaluated (e.g., sources of dissatisfaction) and how it was eventually discontinued (e.g., switching to an alternative IS or simply abandoning the goal). Considering the highly experimental nature of this stage, not all users will find or recognise a need that the technology can fulfil. That is, whereas frame alignment (i.e., a match between the IT/IS and a goal) triggers starting actual use of the IT/IS, rejecting it is also possible if a relevant goal could not be identified (i.e., the IT/IS will be rejected due to irrelevance) or if the usage experience was below what the user imagined it would be (i.e., negatively disconfirmed expectations). Rejection at this early stage corresponds to what IS research discusses as 'outright rejection' (Cenfetelli, 2004) or 'early discontinuance' (Aggarwal et al., 2015;Bhattacherjee, 2001).
Stage 2: Goal Pursuit. Following initial experimentation, the user's goals and the resultant dominant frame will impact how a technology is categorised cognitively (Orlikowski & Gash, 1994;Webster & Martocchio, 1993) and which aspects of it become more salient during the goal pursuit stage. The salient question at this stage may be articulated as 'How to best fulfil the user's goal(s) using the IS/IT'? The goal pursuit stage is typified by continued or repetitive use, where the user's aim is to fulfil an envisioned goal that s/he deemed the technology suitable for attaining.
As demonstrated by our findings, continued or repetitive use entails having a satisfactory usage experience, which reflects the degree to which the goal is being fulfilled according to the user's dominant frame. In our study, the dominant gain frame is linked to the goal of improving one's resources and thus directs the user's attention to features in the technology that are conducive to a productive experience, whereas a dominant hedonic frame is linked to the goal of having fun and thus directs the user's attention to features in the technology that are conducive to a joyful experience. Hence, the usage experience will be deemed satisfactory to the extent the goal is fulfilled and the IT/IS functions in line with the adopted frame. This means that while a given technology feature (e.g., sharing photos on social media) is considered essential to satisfy a hobbyist user, it can at the same time be irrelevant or even cause annoyance to the instrumentalist users, and cause frame misalignment. For as long as the goal is being fulfilled according to the adopted frame, the usage experience is deemed satisfactory, and the user is expected to remain in continued or repetitive usage (i.e., goal pursuit) stage. A dominant frame need not be permanent, especially in volitional use contexts. Thus, reframing is expected to occur depending on the individual, contextual and technological circumstances at play. For instance, returning to trialling and reframing (i.e., relapsing to Stage 1) may occur due to changes in the user herself (e.g., recognising new needs), changes in the environment (e.g., political tensions) or changes in the focal technology (e.g., new updates).
Stage 3: Disruption. The user moves to the frame disruption stage when misalignment occurs between the user's dominant frame and the actual usage experience. This disruption reflects a notable misalignment between what the user's goal is and what the technology delivers. The disruption experienced at this stage is characterised by discontent that is not necessarily rooted in a particular service failure in the technical sense (Tsohou et al., 2020).
Rather, depending on the dominant frame, users will experience disruption differently. To instrumentalists, disruption to their view of the system's utility (e.g., gaining less money than before) was a key source of dissatisfaction. This frame supports IS literature emphasising the importance of IT/IS's performance and usefulness as precursor to (dis)continued use of utilitarian systems, such as self-service banking systems (Prendergast & Marr, 1994), grocery shopping systems (Hand et al., 2009), as well as knowledge management systems (Najmul Islam, 2014). To hobbyists, disruption to the hedonic frame was a key source to dissatisfaction (e.g., the desire to share their photos with friends and receive comments). This frame, in turn, is in line with IS research highlighting the importance of peer feedback and social networking in motivating the use of hedonic systems, such as online communities (Burke et al., 2009), blogging platforms (Tang & Chen, 2020), as well as mobile gaming (Merikivi et al., 2017). Having said that, frame disruption should not be confused with expectation disconfirmation as described in expectation disconfirmation theory (EDT) (Oliver, 1977;Oliver & DeSarbo, 1988). EDT posits that the key predictor to consumer complaining and nonrepurchase intentions is dissatisfaction emerging from negatively disconfirmed beliefs when the first-hand experience with a product or service turns out to be poorer than what was expected. As such, the salient question at the disruption stage is not whether the user is satisfied or dissatisfied, rather the question asked is 'can the frame disruption be tolerated'? To some users, frame disruption will be tolerated, and they will continue using the IS/IT. It is only when a user concludes that the frame disruption is intolerable that s/he is triggered to transition into the next stage.  (Maier, Laumer, Weinert, et al. 2015;Maier, Laumer, Eckhardt, et al. 2015), fatigue (Ravindran et al., 2014) and time wasting (York & Turcotte, 2015); or in certain situations, when a user may prefer to take a technological pause and refrain from using technology, for example for the duration of a holiday (Rosenbaum & Wong, 2015). By contrast, dormancy in our case reflects a stage during which the user questions the underlying need for the IS/IT itself, whether only very briefly or over some length of time.
The salient question here is 'how important is the goal to the user'? As such, the dormancy stage represents a major fork on the usage lifecycle path depending on how the user answers this question. Finding an alternative, however, is not necessarily immediate, or even possible. Our findings suggest that, for lack of alternatives, or awareness thereof, users who deemed the goal important had decided to continue using Scoopshot for varying periods of time, although it was not an entirely satisfactory experience (i.e., they relapsed to Stage 3). This finding supports Tsohou et al.'s (2020) work which shows that users who experience intolerable technical service failure may eventually decide to remain with the service provider because it was their only option.
Stage 5: Quitting. Deeming the goal no longer important leads up to abandoning not only the IT/IS but also the underlying goal initially driving its use. For those users, the usage lifecycle of the focal IT/IS comes to an end, and no further efforts are made to pursue the goal by other means (Stage 5). In our case, this is exemplified by users who decided to quit Scoopshot without any indication that they were interested in finding an alternative. Much of IS discontinuance research focuses on quitting as a coping mechanism to addictive and stressful technologies (Luqman et al., 2017;Turel, 2015), and very little attention is paid to the role of goal abandonment in making the quitting deci- sion. An exception to this is the study of Hand et al. (2009) that shows how in a volitional use context (or 'discretionary use mode'), users may quit the system when the driving goal (or 'initial trigger') disappears.
New cycle with Switching. By deeming the goal important, dormant users start paying more active attention to available alternatives in the market in hopes of finding one that better matches their goal. Consistent with Tsohou et al.'s (2020) insights from post-service failure experience, the main motive driving the user at this point is to find a better alternative. Here, the dominant frame focuses the user's attention towards a pool of alternatives that is in congruence with this frame. In our study, a hobbyist user considered a photo sharing platform as an alternative to Scoopshot, whereas an intstrumentalist user considered a local newspaper platform. Finding a better alternative triggers the user to switch to another IT/IS, and s/he can resume fulfilling her or his goal (New Cycle). From the user's perspective, the goal is still being met although by a different means, reflecting what Van de Ven and Poole (1995) would describe as equifinality. Effectively, an IS use lifecycle has come to an end, and a new one has begun. At any point, however, interest in the previously used IS/IT may re-emerge, thus bringing the user back to experimenting with a once-abandoned IS.

| Theoretical implications
Our study makes three main contributions to IS theory: we propose a novel a stage model for the discontinuance of volitionally adopted IS, highlight the importance of goal-directedness in volitional use contexts, and empirically establish the role of goal framing in volitionally adopted IT/IS contexts.
First, our work contributes to the IS discontinuance literature by proposing a stage theory. IS discontinuance is, by definition, a change phenomenon, and therefore dynamic theorising is needed to develop and refine explanations for how and why this change happens. Our study contributes to the IS literature by developing a stage model for the discontinuance of volitionally adopted IS, which corresponds to what Gregor (2006) characterises as theory for explanation or understanding (Type II). Our nascent theory explains how IS discontinuance occurs in volitional contexts as a sequence of five qualitatively different stages; namely, IS framing, goal pursuit, frame disruption, dormancy, and quitting, after which the user might start a new use cycle via switching. Each stage has a defining question, stage-specific experience, as well as triggers for moving forward or backward between stages. Although each of these stages contributes to our understanding of the discontinuance process as a whole, we find the dormancy stage to be particularly critical. During the dormancy stage, the user begins to question the importance of the underlying goal driving the IT/IS usage. Deeming the goal unimportant terminates, not only the use lifecycle but also the goal itself. Those who deem the goal still important will begin to actively look for alternatives and finding a suitable one will trigger switching. Those who fail to find an alternative, however, may return to usage as dissatisfied users. If this cycle continues (i.e., fluctuating between stage 3 and 4) without observable improvement in the IT/IS to move them to satisfied usage (i.e., stage 2), it becomes a matter of time before the user either gives up the goal or finds an alternative. In practice, the length of time the user spends dormant before quitting or switching can vary from a very brief moment to a prolonged period of time. Our conceptualization of dormancy, and the emphasis on seeking alternatives during this stage, offers additional empirical support to the early findings on trialling as an important activity following a dissatisfactory experience (Salo & Makkonen, 2018) and as a necessary precursor to switching (denoting discontinuance of the original IS) especially in volitional contexts.
Second, our work enriches the IS discontinuance literature by highlighting the importance of goaldirectedness in volitional use contexts. In line with IS research approaches that give the agency of the user greater importance in explanation (see e.g., Karjalainen et al., 2019;Boudreau & Robey, 2005), our model recognises the important role of goals in shaping how technologies are selected, used and evaluated. Embracing the human agency in IS research reminds us that what users do with the technology is not determined by the features of the technology itself but rather by the human's 'ability to form and realize one's goals' when using the technology (Leonardi, 2011, pp. 147-8). Our proposed model emphasises the importance of goal-directedness and thus acknowledges three core assumptions. First, IS/IT users are autonomous, goal-directed agents (Van de Ven, 1992; Van de Ven & Poole, 1995).
Second, human agents recognise, and adapt to, the limits on action (ibid). Third, goals may be achieved via different equally effective routes (ibid). As such, our model contributes to theory by shedding new light on IS discontinuance in volitional contexts in which IS users are self-determined agents who choose to use a technology because of a personally defined goal rather than as a response to external pressures dictating how and when a given technology is to be used (e.g., organisational policies, management directives, etc.). Furthermore, our model recognises that in volitional contexts, human agents are aware that achieving their goals is not guaranteed and that abandoning the goal itself is always an optiona path not commonly feasible in mandated organisational settings. We know that in organisational settings, when an incongruence between a technology and work routines occurs, employees will modify the technology to align it with their work routines (Leonardi, 2011) if the technology is flexible, or they will modify their own work routines, for example, by developing new workarounds (Recker, 2016), if the technology is inflexible.
By contrast, in a volitional context, the lack of compulsoriness makes discontinuance a more viable option. Rather than changing their own routines, users who fail to achieve their envisioned goals with an IS/IT that is not malleable enough to meet their needs, will initiate a process of IS discontinuance. While our proposed theory shares with affordance theory the assumption of goal-directedness of actors (Leonardi, 2011;Volkoff & Strong, 2017), they provide different accounts of explanations. That is, whereas affordance theory introduces affordances as qualities of an object (e.g., visual cues) that define a spectrum of actions made possible by the technology (Norman, 1999); our work introduces frames as cognitive models (of the users) which may or may not align with the affordances of a given technology. Research on affordances of IS has advanced our understanding of how users perceive specific functions of different technologies and their possibilities (Norman, 1999;Volkoff & Strong, 2017) and how to use them effectively (Burton-jones & Volkoff, 2017); however, we maintain that frames are conceptually better suited for analysing and understanding the whole use life cycle from exposure to abandonment. Volitional user makes sense of the technology's affordances by framing and decides if and how the technology is used, regardless of how (or for what) it was designed.
Third, our work contributes to the technology frames literature by empirically establishing the role of goal framing in volitionally adopted IT/IS contexts. Our work enriches the current understanding of technology framing research by highlighting the relevance of goal framing in understanding technology use beyond the organisational context (Davidson, 2002;Orlikowski & Gash, 1994). Informed by goal framing theory (Keizer et al., 2008;Lindenberg, 2001;Lindenberg & Steg, 2007), we introduce IS framing as the starting point for understanding IS discontinuance. The argument we advance here is that understanding how the use lifecycle unfolds begins with understanding how a user frames and adapts the technology to their own goals and needs, which may or may not coincide with the IS/IT's intended purpose. In our empirical setting, goal framing theory sensitised us to identify two dominant frames: the hedonic frame and the gain frame. The findings demonstrate how the same technological artefact, depending on how it was framed by the users, could be interpreted, used and evaluated differently. We should note that framing is context-dependent and therefore in different contexts, other frames may be more relevant or applicable. Our view is in line with earlier IS research adopting the framing lens (Alvarez, 2008;Bartis & Mitev, 2008;Hsu, 2009;Orlikowski & Gash, 1994;Su, 2015) illustrating that what is often considered a single-meaning IS artefact in organisational context may produce radically different perceptions and behaviours among different organisational members. Hence, our work lends support to Orlikowski's (1992) notion of the interpretive flexibility of technology, where IS in use mode does not necessarily reflect its design mode.
Furthermore, by identifying the gain and hedonic frames in our work, we shed new light on the widely applied classification that labels IS/IT artefacts depending on their intended design or 'functional capacity' (Wu & Lu, 2013, p. 155) as either hedonic IS (intended to be used for leisure purposes) or utilitarian IS (intended to be used for productive purposes) (Gerow et al., 2013;Lowry et al., 2018;van der Heijden, 2004). This artefact-oriented classification has been very useful in giving the IS community a language and a theoretical foundation to discuss two admittedly distinct IS use experiences. However, this classification is insensitive to technology usage that does not fit the designer's (and often the researcher's) framing. For instance, framing a social networking service (SNS), such as Facebook, as a hedonic technology (Turel, 2015) desensitises us to its utilitarian usage and the fact that many users use it with utilitarian goals in mind (Xu et al., 2012). Instead of compartmentalising IS/IT into hedonic or utilitarian IS based on preconceived assumptions, we might gain new insights by looking into how different users, depending on their own framing, may creatively utilise the same technology to serve their own needs (Orlikowski, 1992). Recognising the importance of this shift not only affords a better understanding of IS use discontinuance but also a more holistic understanding the IS use lifecycle. With our study, we advocate a frame-oriented rather than an artefact-oriented research agenda, where instead of seeking only to explain why a class of technology is used or discontinued, we should also focus on understanding how users creatively utilise what is available to achieve their goals.

| Practical implications
Our findings also point to a number of practical implications. First and foremost, providers of volitional IS should understand that the users do not necessarily share the envisioned and designed use purposes. Therefore, effort should be put into understanding how the users adapt and appropriate the IS to serve their own goals. In the specific context of Scoopshot, the identification of the two dominant user types of instrumentalists and hobbyists and their needs and expectations readily suggest two strategies to alleviate the different sources of dissatisfaction.
The first strategy could be to improve the instrumental value of the service, which could be achieved by expanding the number of partners on the buying side of the network. This is supported by what information economists refer to as cross-side network effects, suggesting that an increase in the number of users on one side of a platform (here the demand side partners) increases the attractiveness of the platform for the users on the other side (here the supply side Scoopshooters) (see e.g., Rysman, 2009;Gawer, 2014).
The second strategy would be to improve the entertainment value of the service. This strategy would target the hobbyists who seem to be affected less by the cross-side network effects and more by the same-side network effects, which describes how an increase in the number of users on one side increases the attractiveness of that side (see e.g., Eisenmann et al., 2006). This calls for Scoopshot to cultivate a community rather than just a marketplace for selling and buying content.
Furthermore, different gamification features (Morschheuser et al., 2016), such as a scoring mechanism  with points awarded to all contributions (content, peer-evaluation, etc.) (see e.g., , could be utilised to address the users' need for feedback, feelings of competence and social connectedness.
It is important to note, however, that choosing between these two strategies or employing them in tandem depends to a large extent on the available resources and the strategic direction of the IS provider. Under conditions of limited resources, it can be prudent to focus on a single strategy.
Lastly, we believe that the identification of the dormancy stage in our study has major practical implications. As such, the dormancy stage should be seen as an early warning sign for IT/IS providers. We found that Scoopshot users who were once active typically went through a dormancy stage for periods ranging between three months to over one year before quitting or switching. If this is indeed a typical pre-termination stage in the wider user-base, the IS provider should have sufficient time to retain those dissatisfied users. Assuming that the provider has gained sufficient understanding of the dominant frames guiding the use of their IS, they should align their offering and inform their users of the updates as soon as possible. After all, what drives a user of volitional IS is the fulfilment of an envisioned goal.

| Limitations and boundary conditions
In theory development work, generalizability commonly refers to the applicability of a theory in a setting different from the one where it was conceived (Lee & Baskerville, 2003). Generalising is, however, by definition an inductive (Seddon & Scheepers, 2012;Tsang & Williams, 2012) and ampliative (Ketokivi & Mantere, 2010) form of reasoning for which no amount of empirical evidence can logically justify making knowledge claims beyond what has been studied or observed. Considering this paradox, researchers are advised to make carefully articulated generalisation claims and accept that '[a]ll knowledge claims, including generalizations, are subject to revision' (Seddon & Scheepers, 2012, p. 7). With this in mind, we focus on a particular notion of generalizabilitygeneralisation from local theory to other settingswhich involves articulating the conditions under which the local theory may be useful in informing other contexts or situations. Here, we highlight three potential boundaries limiting the generalizability of our proposed theory. First, our study and the proposed theory are within the bounds of volitional contexts. As such, we do not expect our theory to be applicable to mandated use situations (e.g., in some workplace contexts), where users can be required to use a given technology to complete their work tasks. Second, our study and the proposed theory are within the bounds of functional self-control. That is, we do not expect our theory to explain the discontinuance behaviour of users who suffer from deficient self-regulation symptoms such as technology addiction (LaRose et al., 2003;Turel, 2015) or compulsive use (Caplan, 2010;Turel et al., 2011). Thirdly, the proposed model is conceived in a context in which the IS/IT is freely available. As such, economic lock-in factors (e.g., subscription fees, sunk costs) were irrelevant in our studied context. Nevertheless, we acknowledge that these issues are likely to contribute to the use behaviour of IS made available with such service monetization strategies.

| CONCLUSIONS
The main objective of this study was to understand the process leading to the discontinuance of a volitionally adopted technology. Specifically, we sought to answer the two following questions: (1) Why do users discontinue the use of an IS they have volitionally adopted and used? (2) How does IS discontinuance occur over time in such context? Informed by dynamic theorising and goal framing theory, we show how the same technology may be interpreted, used and evaluated from two distinct framesone that directs the user's attention to focus on the technology's instrumental value (the gain frame) and another that directs the user's attention to focus on the technology's enjoyment value (the hedonic frame). Analysing our data from this perspective helped us develop a stage model emphasising five key stages of IS framing, goal pursuit, frame disruption, dormancy and quitting, after which possible switching denotes a new cycle. One of the key insights our nascent theory reveals is that the sources of dissatisfaction for different users are attributed to their own goals and frames rather than the technology's intended purpose. Furthermore, it shows that while dissatisfaction is critical to disrupting a usage experience, dissatisfied users do not necessarily immediately make a decision to quit. Instead, they enter a stage of dormancy during which they reduce their activity, search for alternatives and reassess the goal the IS is expected to serve. Moving forward to the quitting stage indicates that a more suitable alternative is found (denoted by switching) or that the goal itself has been abandoned.
Our study indicates that IS discontinuance is a more complex phenomenon than generally portrayed in the literature. To understand the dynamics and the changing nature of the encounters and events leading users to continue or discontinue the use of these systems requires more theorising as well as further empirical research. In particular, future research should develop a deeper and more refined understanding of the stages preceding and succeeding discontinuance decisions. We hope that the ideas advanced in our study will serve as a starting point for future research to appreciate the importance of equifinality, which echoes the assumption that users continually find creative ways, based on what is available, to achieve their goals.

DATA AVAILABILITY STATEMENT
The raw interview data that support the findings of this study are available from the corresponding author upon reasonable request. The data are not publicly available due to privacy restrictions. The most frequent reasons for not using the assignment system were:

ORCID
system is not offered on the courses; the system is not compatible with students' needs; the system is not consistent with teachers' guidelines; complexity of the system; using the system is not mandatory.
14 Goode use of internet; the transition from adoption to use was out of the study scope. However, the article acknowledges that adoption and use of Internet are two distinct processes that are influenced by different forces. Perceived popularity and perceived characteristics of Internet were found to determine its adoption (or lack of), whereas perceived need for Internet impacted its continued use.