How Robotic Process Automation Augments Work:
A Mid-Range Theory of Affordances
Kristoffer Sandén (S3235866)
Abstract – Today, along with automation’s growing effects on society, some predict great disruptions to the
white-collar workforce. However, answers to what implications the collaboration of humans and robots has on organizational routines are still lacking. Since employees increasingly depend on novel Information technology (IT) solutions, we need to understand how it augments their everyday work. With the research question addressed in this paper, I respond to the call for more mid-range theories within Information Systems (IS) research. From a socio-technical perspective on organizational change, I studied an implementation project of Robotic Process Automation (RPA). In a qualitative case study, I utilized Critical Realism (CR) and the Theory of Affordances. The proposed explanatory mid-range theory is based on a network of six interacting affordances and four generative mechanisms. My insights contribute to both academics’ and practitioners’ understanding of how and why affordance actualization occurs, by illustrating how RPA augments employees’ organizational routines.
Key words: Affordances; Critical Realism; Mechanisms; RPA.
Supervisor: Dr. Benjamin Mueller Co-assessor: Dr. I. Maris-de Bresser
MSc. Business Administration – Change Management Faculty of Economics and Business
University of Groningen Date: 21st of March 2019
1 TABLE OF CONTENT
Introduction ... 3
Conceptual Foundations ... 6
Robotic Process Automation ... 6
Critical Realism ... 6
Affordances ... 8
Critical Realism and Affordances ... 9
Methodology ... 10
Case Description... 10
Data Collection ... 11
Data Analysis ... 13
Theorizing the Conceptual Framework ... 18
Individual Affordance Actualization ... 18
Outsourcing intersystem actions ... 19
Detecting process errors ... 20
Monitoring progress ... 20
Concealing confidential data ... 20
Accessing integrated process records ... 21
Identifying transactions and associated actors ... 21
The Affordance Network ... 22
The Generative Mechanisms ... 24
The inception mechanism ... 25
The resource reallocation mechanism ... 26
The transparency mechanism ... 27
The appreciation mechanism ... 27
Discussion ... 29
2
Future Research and Limitations ... 31
Conclusion ... 32
References ... 33
Appendices ... 37
Appendix 1 - Codebook Examples ... 37
Appendix 2 - Interview Guide Summary ... 38
3
Introduction
The impact of various types of automation has been around since the industrial revolution and today its effects on the society are vast. Brynjolfsson and McAfee (2014) believe that the age when robots substitute human labor is closer than one might think. Consequences have mainly been observed affecting blue-color workers, though concurrently we are starting to see a similar phenomenon within the area of knowledge work.
Some claim that the rise of Artificial Intelligence (AI) and other sophisticated technological advancements will disrupt the requirements of white-collar workers, eventually outdating today's job descriptions (West, 2015). Others believe we will see an increase in human collaboration with robots (Lacity & Willcocks, 2016) as their capabilities continue to advance, i.e. enhanced individual agency of robots. Consequently, this may lead to constellations where humans and machines are acting together with each other, collaborating towards a common goal, rather than acting on each other. While automation and AI utilization are expanding, we do not know how that affects organizations’ work.
Since workers progressively rely on using innovative software-based information technologies (IT) (Leonardi, 2011), we need to understand new technologies and how they augment individuals’ and teams’ work in organizations. A specific category of robotics called Software Robotics is currently on the forefront, often referred to as Robotic Process Automation (RPA). In essence, it implies the use of a software plugin that interacts with and operates on top of the existing applications on a computer to automate processes. As a service software, RPA has demonstrated benefits similar to previous automation, e.g. reduced costs, improved speed and accuracy (Lamberton, Brigo & Hoy, 2017). Still, the discussion around how this impact humans in a knowledge or administrative job setting seems to be virtually absent (Isaksson & Wennberg, 2016). Instead, studies on the interaction with robots have, for instance, taken a physical approach, such as within a factory setting (Meneweger, Wurhofer, Fuchsberger, & Tscheligi, 2015). The few studies within a white-collar job setting have incorporated a focus on automation interaction rather than cooperation, such as pilots’ interaction with autopilots (Wickens, Mavor, Parasuraman, McGee, 1998). Therefore, I argue that a deeper understanding of how organizations that utilize RPA, will decrease friction in robotic implementation, improving collaboration among a seemingly futuristic defined workforce which contains both humans and robots (Shave, 2017).
4
(2014) adjusted for some earlier raised concerns in exploiting the theory of affordances within IS research, by proposing their affordance actualization lens. They further argue that future research needs focus on theory development explaining IT-associated organizational change, so that it provides actionable recommendations. Today, such research initiatives associated with RPA is to the author’ knowledge absent. To address this gap, I engaged to explore the implementation of RPA. More specifically I asked,
how does Robot Process Automation augment the work of individuals and teams in organizations? To
address my question, I employed a case study using Critical Realism (CR) (Wynn & Williams, 2012). To investigate the identified events and changes to the work in question, I adopted the notion of organizational routines, defined as “repetitive, recognizable patterns of interdependent actions, carried out by multiple
actors” (Feldman & Pentland, 2003, p. 95). In that vein, I was able to study the RPA end-users’
organizational routines as a whole, delimiting my study from mapping the entire routines internal structures (Pentland & Feldman, 2005), as that would be a different study.
I furthermore applied the affordance theory owing to its suitability in providing explanations within IS research and organizational change (Strong et al., 2014). Volkoff and Strong (2017) recently expressed that studies of affordances using CR is an important avenue for future research. The advantage of utilizing this socio-technical perspective, lies in its ability to be specific about the technology yet importantly incorporating the social environment (Volkoff and Strong, 2017).
Therefore, in combination, the above-mentioned standpoints supported me in being congruently explicit in my examination, as I studied multiple instances where individuals interact with introduced RPA applications, to observe its impacts on organizational routines. In that manner, I provide explanations for how RPA augments the end-users’ work by revealing the story of how that transpired, in the pursuit of my research question. Accordingly, I contribute to both theory and practice with insights into how and why affordance actualization occurs. Specifically, details surrounding the evident affordances are explained and illustrated through a network of affordances. Accompanied by four generative mechanisms that together illustrate the entirety of the phenomena. Thereby, the core academic value lies in the response to academic calls, both for the development of mid-range theories (Strong et al., 2014; Hedström & Swedberg, 1998) and for extending the knowledge of how affordances are actualized (Strong et al., 2014; Bygstad, Munkvold, & Volkoff, 2016).
5
6
Conceptual Foundations
The purpose of this chapter is to illustrate a foundational understanding of the literature, later serving as the base in constructing my conceptual framework. Firstly, I lay out what the RPA technology can do and how that may look. Secondly, I outline the fundamentals of the CR philosophy and the theory of affordances separately, before I lastly interweave the two.
Robotic Process Automation
I first dedicate a brief introductory section to the concept or RPA in order to make the technology more digestible in the remaining parts of the paper. Software robotics implies the use of a software plugin that operates as an extra layer on top of the existing applications on a computer, to ultimately automate the process performed by human workers. While RPA is generally viewed as a fairly new conceptual product, its constructional components are less so, enabling it to virtually imitate human interaction.
The process of installing a new RPA software requires, firstly, to map out the process in question, secondly, to link applications, and thirdly, to schedule the steps which the robot follows automatically when required (Lamberton, et al., 2017). To briefly illustrate how RPA works; it is capable of extracting information from different applications such as PDF and Excel spreadsheets and enter it into another system such as Enterprise Resource Planning (ERP) and generate protocols of such entries (Willcocks, Lacity & Craig, 2015). The types and combinations of interactions that an RPA is capable of are many, making it proficient to carry out parts of, or complete business processes. With reviewing that short example of potential RPA related technical concerns, the stage is set to dive into the philosophy of CR.
Critical Realism
The philosophy of Critical Realism became popular through the works of Bhaskar (1998), merging realist ontology with interpretive epistemology. CR postulates the existence of a real world, independently of our knowledge of it. This real world contains structures (physical or social objects such as artifacts and social concepts) and mechanisms. This is the first of three domains within which CR interprets reality. Mechanisms within the real are thought to generate events in the actual, the second domain. These events may or may not be observed yet are the constituent parts of the actual. A subset of the observed events within the actual may then be identified within the empirical, the third domain.
7
Figure 1: A typology of social mechanisms (adopted from Hedström & Swedberg, 1998. p.22).
Identified mechanisms therefore explain causation of events, resulting in a causal chain (Wynn & Williams, 2012). Importantly, Hedström and Swedberg (1998) clarified that social events cannot be explained by single mechanisms. They adopted Coleman’s (1986) macro-micro-macro model to illustrate their typology of social mechanisms, including three types of mechanisms acting within a micro and macro-level context (see Figure 1). Simply put, macro-macro-level events cannot be explained without taking the micro-level context into account. Therefore, macro-micro-level change should always be explained via the micro-micro-level. This is what the three types of mechanisms show, as they are linking macro and micro. Hence, to show how two macro events are linked, one needs to determine the three mechanisms. First, the situational mechanism demonstrates how one macro event (A) affected an individual on a micro-level (macro-micro). Second, the
action-formation mechanism explains how the individual embraces the macro-level impact (micro-micro).
Finally, the transformational mechanism describes how individuals’ actions collectively generate a new outcome (micro-macro), i.e. another macro event (B) (Hedström & Swedberg, 1998).
Another crucial facet of CR is the perception of agency and structure (Carter & New, 2005; Mutch, 2010), being regarded as temporarily separate. Structures serve as the preceding foundation, enabling the potential for action. Nonetheless, structures may be altered over time as a consequence of such actions (Volkoff, Strong, & Elmes, 2007). Furthermore, their properties and powers are distinct, actions are temporal while structures are more or less permanent. Agents or actors such as people may, for instance, possess emotions, intentions, perceptions, and self-conciseness, making them capable of expressing and pursuing ideas. Thus, they can exercise their power that potentially alters or preserves structures. Contrastingly, structures possess the power to influence actions and may therefore enable or constrain them. This implies that change may vary if the structures were different or nonexistent (Carter & New, 2005).
Since socio-technical structures and their surroundings are crucial components in IS studies (Wynn & Williams, 2012), such understanding allows us to categorize observed artifacts into structures (e.g.
8
applications and components of software and hardware) (Fleetwood, 2005). In the same manner, we may classify observed social constructions as structures, for example, ideas, norms, or values (Elder-Vass, 2005; Wynn and Williams, 2012). With this background in CR, I now turn to discuss the fundamentals of Affordances.
Affordances
The Affordance theory was first put forward by Gibson (1979) in the field of ecological psychology. It refers to the potential actions of an object that could be utilized (afforded) by an actor towards pursuing a specific goal. Importantly, affordances are in this view not characteristic features of an object, rather, they arise from the interaction between the specific actor and object. In the field of IS research, however, affordances have become part of the ongoing discussion around socio-technical systems and IT-associated organizational change. In this context, affordances aid to understand the actor-object relation, representing the potential actions of technologies (Markus & Silver, 2008; Leonardi, 2011).
In their study of affordances, Strong et al. (2014) investigated the role of IT in organizational change, looking at an implementation of an electronic health record system. Their developed mid-range theory advanced the literature by discussing the process of affordance actualization, affordances in an organizational context, and bundles of interrelated affordances. Similar to action possibilities, Strong et al. expressed the need to separate an affordance into its opportunities, actions taken, and their consequences (i.e. affordance, actualization, and outcomes respectively). Furthermore, they recognized the goal related complexity. Employees may possess goals of various character, (e.g. task or individual goals) and with a different scope, (e.g. individual, group, or organizational goal). In order to address this issue, the authors focused on “immediate concrete outcomes” of the performed task.
Despite their seemingly rather narrow focus, affordance actualization is potentially more complex in reality. Strong et al. (2014) recognized this facet by acknowledging the bundles of affordances, the potential nature of several affordances arising from any interaction among an object and actor or actors. Taken together, their proposed theoretical contribution, the Affordance Actualization lens, offers a perspective to distinguish social and material. Particularly useful when studying the actualization process, and subsequent IT-associated change related theorizing.
9
affordance theory. Consequently, Leonardi proposed that the largest organizational implications arise from group level network change when individuals utilize the same subset of affordances.
To make the case of mutually enacted affordances, Leonardi (2013) introduced the notion of (1) individual, (2) shared, and (3) collective affordances, referring to their actualization. The individual category represents a specific actor that actualizes an affordance solitarily (1), whereas a shared affordance refers to the same affordance being similarly actualized by multiple individuals (2). Contrastingly, a collective affordance combines the idea that multiple individuals are carrying out a shared goal, yet individually executing different tasks, in order to pursue that goal (3).
Thus, based on the outlined discussion above, for the rest of this paper, I define an affordance as
the potential for behaviours associated with achieving an immediate concrete outcome and arising from the relation between an object (e.g., an IT artefact) and a goal-oriented actor or actors (Volkoff and Strong,
2013; Strong et al., 2014). With the above discussions in mind, I will move on to intertwine the philosophy with the theory, to conclude the chapter.
Critical Realism and Affordances
Combining the fundamentals from above enables utilization of the theory of affordances, aligned with its philosophical routes (Markus & Silver, 2008), as a practical tool for identifying mechanisms. By making use of retroduction and the notion that actualized affordances produce immediate concrete outcomes, one may abstract individual affordances from such events. While such actualization remains specific to each case, the associated affordance, nevertheless, is more generic. That is, since it is grounded in the case and therefore reflects the relationship among the type of organization and technological artifact. In this way, it is argued to be the basis for and a specific case of generative mechanisms, beneficial in developing better mid-range theories of IT-associated organizational change (Volkoff & Strong, 2013), reflecting the work of Hedström and Svedberg (1998). In other words, affordances are a narrower form of mechanisms, arising among a goal-directed actor and a structure or object. In contrast, generative mechanisms have the capacity to arise among different structures, without being triggered by an actor.
10
Methodology
To study the phenomena of interest, I have applied a critical realist case study approach after Wynn and Williams (2012), in order to address how individuals and groups within organizations actualize the use of a Robot Process Automation implementation. Accordingly, CR suits my study particularly well as it seeks to answer how and why events occur, rather than attempting to predict the future. This section is structured as follows, I first present the case, followed by my data collection and subsequent analysis.
Case Description
The case company of study in this research project, referred to as “Company A”, is a European based multinational manufacturer and service provider of specialized industrialized equipment. Company A has been engaged with internal process harmonization related projects for the last decade and has increasingly advanced their RPA investments over the last year. Considering that the application of RPA technology within the field is far from mature, the case company enabled a unique opportunity to study the implications of the novel technology. That is especially true considering that several RPA robots have been implemented at several different sites, spanning over multiple continents.
At the beginning of Company A’s RPA project (2017), an external consultancy was sourced and responsible for the development. However, six months later the decision was made to no longer pursue any development through the external consultancy. Instead, the entire development was moved in-house by establishing an RPA team. Today, the RPA initiative at Company A has grown into a team of four, the RPA manager and three business process analysts and developers. This entity serves as the center of the organization’s RPA engagements located at its Headquarters1. Nowadays, they also have dedicated RPA development hubs, locally operating at various locations of the organization’s establishments.
As a steering entity, the RPA team is in charge of identifying opportunities and develop solutions to support internal departments and their processes within the organization, utilizing the RPA technology. In the process of pursuing that mission, the developers and process analysts of the RPA team take on different roles in order to develop desired solutions. Based on each specific instance, they engage with the RPA software to construct the required setup. These RPA developments then ultimately serve to support the individuals (i.e. RPA end-users) located throughout the organization in their daily work. End-users may be of various types and therefore also have different reasons for utilizing the RPA. Consequently, depending on their specific role and function (such as an accountant, clerk, or manager), associated responsibilities and goals may vary as well.
Thus far, the RPA teams’ operations have covered various business functions at several locations worldwide, leading to multiple RPA robots globally in action at this time. Most of the implemented RPA’s
11
within Company A are located within the finance, human resources, and sales departments. Though, the main portion is established within finance related functions. The company has been using SAP as their central IT platform for more than a decade. Thus, process transactions within the organization are generally handled in SAP, although commonly supported by the use of Excel.
Altogether, Company A’s history and wide scope of utilizing the technology serves as a particularly suitable context for the study. The setting enables access to multiple perspectives within a single organization on emerging phenomena.
Data Collection
In my study of RPA, initial personal observations and conversations with consultants guided the first phase of my exploratory work. These observations along with two non-recorded informal interviews with consultants specialized in automation, served to inform me in understanding the fundamentals and complexities in the utilization of RPA in the field. Likewise, these insights both confirmed and supported me in the development of relevant questions. It appeared to me that experts understanding and experience in relation to how organizations and particularly employees were affected by its introduction was low. Mainly, since most of the RPA initiatives (at the time) was considered to be in a rather early phase.
The insights I gained from these initial observations supported me in refining my interview guideline. Subsequently, I pursued further exploration of RPA, its implications and how it evolved in the field through semi-structured interviews with one RPA specialized consultant at “Company C” and two RPA managers at Company A and “Company B” respectively. At this point in time, it became clear to me that Company A allowed for a comprehensive case that would suit my research particularly well and access for further exploration was negotiated. Using multiple data collection approaches to address my research objective allowed for triangulation, which contributes to assuring validity, reliability, and credibility of the gathered data (Volkoff et al., 2007). Semi-structured interviews served as my main data source, complemented by participant observation and company documents.
The initial interview with the RPA manager at Company A and complementing informal conversations regarding the RPA project, served as a foundational understanding of the organization’s structure and scope of technology adoption. This also enabled me to identify and discuss relevant stakeholders for further interviews and observations within the organization. Consequently, in parallel, I commenced further interviews at Company A.
12
associated details. In that sense, I conducted observations for 45 minutes of 3 members in the RPA team during their daily work. This equipped me with insights in two fundamental ways. Firstly, it enhanced my understanding of RPA’s technical capabilities, features, and properties. Secondly, I attained a better, more nuanced picture of their social environment in pursuing day to day work. More specifically, practices and activities in relation to how they cooperated, both within the team and with the rest of the organization, in developing RPA applications.
Thereby, the developers’ perspectives played an invaluable role in the study, shedding a unique light on the rather complex phenomena, i.e. RPAs influence on end-users’ organizational routines. That is because the developers themselves are a distinct type of RPA user. Namely, the developer’s goal in utilizing the RPA is to develop and enable customized solutions that empower the capabilities of RPA to support the end-user’s daily work. The data collected up until this point in time represents the first stage (referred to as Phase 1), and guided the remaining data collection (referred to as Phase 2), striving towards achieving a holistic view of RPA realization within Company A. See Table 1 for an overview of the interviews in relation to the phases.
One month after Phase 1 was finalized, I engaged in further data collection, Phase 2. Interviews were conducted with end-users of RPA in a total of five interviews with six interviewees (two interviewees were combined into one interview). Among the five interviews, three were conducted with employees located at different departments at the location of the headquarters, while two were located in different countries. The six users had different roles; two departmental process development managers, three accountants and one sales support clerk. The employees’ various roles and functions gave me detailed insights into the phenomena with specific RPA utilizations and associated impacts through their personal stories and observations.
In contrast to the developers, the different end-users embraced the RPA in ways that could leverage their own work-related processes and routines, to carry out their individual or team’s responsibilities and goals. Nevertheless, in order to do so, they were dependent on the developer’s ability to interact with the RPA. All things considered, besides the already outlined involved individuals, the introduced RPA agents themselves possess agency, too. Hence, they resemble an additional important actant. As such, developed RPA applications are required to be activated by any user possessing the skill to do so. Although when launched, it holds its own agency.
13
phenomena from multiple levels, enhancing the study’s analytical validity (Zachariadis, Scott & Barrett, 2013).
Data collection Inside case company Outside case company Position Quantity Position Quantity
Phase 1 RPA manager 1 RPA Expert 1
RPA developer / Business process analyst
3 RPA Manager 1
Phase 2 Departmental process development manager
2
Accountants 3
Sales support clerk 1
Table 1: Overview of the interviews.
The interviews were held in English, except for one expert interview that was held in Swedish. They lasted between 26 and 62 minutes. Four of the interviews were performed using Skype due to different practical reasons, the rest were performed in person. Each interview was tape recorded and successively transcribed after ensured interviewees’ informed consent. To increase the reliability, I used similar interview guides for all interviews, constructed and enhanced through the support of Wynn and Williams (2012) and Bygstad et al. (2016). Although the structure was adapted towards the interviewees’ position; expert, manager, developer, and clerk (see Appendix 2). All the collected data were constantly transferred to a research database for storage. For a complete overview of the data sources, see Table 2.
Data Analysis
Along with my critical realist approach, I adopted the recommendations made by Williams and Edge (2012) about how to pursue such research in the field of IS. That is, through an iterative data collection and analysis process, resembling grounded theory method (Strauss & Corbin, 1994). As a result from my initial observations and informal interviews, I utilized a socio-technical systems perspective (e.g. Leonardi, 2011)
Table 2: Overview of the data sources.
Data source Details Within case company External to case
Interviews 11 9 2
Recorded interview material 8 hours 7 hours 1 hour
Transcripts 15 pages 160 pages 17 pages
Observation 45 min 45 min -
14
as a basis to guide my further exploration of the phenomena, how the introduction of RPA influenced the organizational routines.
First, I used open coding to remain grounded in the data while developing a clear illustration of the primary events. This coding process continued until the last data were collected, resulting in a total of just over 230 open codes (See Appendix 1 for codebook examples). Through the initial open coding, several business drivers for the RPA technology and its general capabilities were identified. This phase provided me with a basic understanding and a preliminary idea of the managerial expectations, problems, and benefits in the field. Additionally, I gained a first glimpse of the RPA project within Company A, its timeline, major initiatives, and how the technology was exploited to date. These early indications guided the continued data collection.
After I completed the interviews of Phase 1 and the subsequent open coding, it emerged that the RPA technology was associated with specific benefits and consequential changes to the organization’s routines and employees’ work. On an individual level, changes to routine tasks, simplified processes, and interfaces were observed. From the perspective of the organization, general cost savings, changes to business processes, and increased employee satisfaction were evident. However, at this point in time, my data seemed to have the strongest explanatory power around an individual level focus.
In addition to above, CR not only guided me towards these insightful events but also helped me identify relevant structures, actors, and contexts, along with the exploration of those events. Table 3 outlines the main evident structures (including a description and examples of their related properties) among the key events from the case. In addition, the table contains an illustrative example of a specific event and its accompanying relevant structures, to increase contextual comprehension.
I turned to the extant literature for explanations and relevant theories adding to my understanding of the initial observations. The theory of affordances appeared as an appropriate perspective for my research endeavors. It supports me in determining and classifying the identified events and their consequences. That was especially true for Strong et al. (2014) development of the affordance lens, offering a multi-level perspective. Moreover, Volkoff and Strong (2013) argued for the benefits of utilizing affordances in the data analysis of CR research, finding it especially suitable to identify mechanisms arising in the relation between the technology and its user. Similarly, Bygstad et al. (2016) later developed a more detailed framework for utilizing affordance theory within CR research, based on well recognizable CR literature (Bhaskar, 1998; Danermark, Ekstrom, Jakobsen, & Karlsson, 2002; Wynn and Williams, 2012).
15
work of individuals and groups. Although, I was fundamentally wondering what mechanisms that could explain what my preliminary findings have demonstrated?
Following the data collection of Phase 2 and the continued open coding, several outcomes emerged reflecting the perspective of both individuals, teams, and the entire organization. However, again, most events revolved around the individual level, additionally providing the strongest degree of contextual details. Hence, my data suggested a primary focus on the individual level, with regards to the robustness of explanatory power. I embraced this by refining my scope to examine the organizational routines of the RPA end-users.
Main Structures Description and example of properties
Clerks/accountants RPA end-user, possesses individual needs,
responsibilities and goals towards/within team/department
Managers RPA end-user, holds team and individual
responsibilities
RPA developers Desire to develop RPA applications for end-users
RPA Material agency activated by human interaction
SAP Cross organizational work system software (ERP)
Microsoft Office Excel Data processing spreadsheet software Finance/sales support/human resource department Work system, routines, hierarchies, group
dynamics, communication
Description of event example Associated Structures (Related Properties)
An introduced RPA application supported a team with financial transactions and loading of Excel files to SAP. After its introduction, new
instructions for collaboration (new routines) with the RPA was issued for the accountants as a response to early interactional issues among the actants. The accountants collectively shared the responsibility to supervise the RPAs progress (new shared responsibility). Managers and controllers were able to make use of the history from the RPA process in achieving their individual goals.
Financial department (work system, hierarchy, procedures, habits)
Manager (desire to improve individual and team goal achievement)
Accountants (disbelief in RPA, frustration towards SAP, desire to achieve tight deadlines,
enabling/constraining RPA’s agency)
RPA developer (desire to interact with RPA, end-users and improve business processes)
RPA (material agency, interacts with other softwares)
SAP Excel
16
In the light of emphasized critical events, I engaged in retroduction, assuring analytical validity (Wynn & Williams, 2012; Zachariadis et al., 2013). Accordingly, I abstracted plausible affordances from immediate concrete outcomes, based on several instances of individuals in various positions. In doing so, CR allowed me to isolate each event and to identify the involved structures and entities (as per Table 3). By critically re-describing each event and analyzing the interactions among involved entities, I ended up with over 10 different affordances, each accompanied with deep contextual details. Subsequently, I went back to my data.
Through corroboration, the process of grounding conceptual developments by going back to the data to validate them (e.g. Wynn and Williams, 2012) and cross-sectional analysis, I merged some affordances and disconfirmed others. Eventually, I ended up with six main affordances. During the individual analysis, indications in the data pointed me towards potential links between the affordances and actors themselves, recognizing that affordances are existing simultaneously as highlighted by Elder-Vass (2010). Therefore, I commenced analyzing the affordances in relation to each other. A network of connections of various kinds began to emerge. As an outcome of analyzing the affordances along its associated contextual details, I identified two main categories through the contextual similarities within the network, generalizable into a group context.
As a consequence of the above, I was able to hypothesize generative mechanisms that could explain the observed outcomes and concepts in my data. Following the procedure of retroduction, the two affordance categories developed into two generative mechanisms, the resource reallocation mechanism and the
transparency mechanism. Similarly, through the abstraction of cross-case comparable enabling conditions,
I managed to recognize a third, the inception mechanism. The fourth, however, emerged from the corroboration process of another potential mechanism, related to management support.
17
mindset were found, pointing towards an increased proactive search for RPA opportunities, eventually resulted in the appreciation mechanism.
18
Theorizing the Conceptual Framework
This chapter serves to outline the progression of analysis, demonstrating the identified components, and show how the mechanisms eventually unfolded. In doing so, we first need to identify the individual affordances (see Table 4) to understand the individual’s interaction with the RPA. Although, by only understanding the affordances in isolation, we do not take into account the entire picture of the phenomenon and are therefore not yet equipped to answer the research question. As a matter of fact, I move on to scrutinize the relationships between affordances, enabling a more complete picture of how the RPA is utilized within the contextual environment.
Based on those findings, I subsequently constructed a theoretical framework (see Figure 2). However, the figure by itself does not answer the research question completely. Hence, I lastly illustrate the generative mechanisms that explain how the RPA related change unfolded. Ultimately, the generative mechanisms help to answer to how RPA augments organizational routines of its end-users, while simultaneously demonstrating how these findings were empirically evident.
Individual Affordance Actualization
While the affordances are presented in a structured list, the data from which they evolved was not. In reality, the identification of affordances rather resembled the art of mushroom picking, as oftentimes the awareness or identification of one potential affordance guided me, or at least pointed, in the direction of another. Therefore, telling the transparent story of how the six affordances appeared runs the risk of becoming uncomfortably repetitive. Instead, I decided to structure the affordances according to the list in Table 4 (i.e. from 1 to 6) to guide the reader through the scenic route of mushroom picking, layered with experience-based storytelling along the way. Therefore, affordances can be illustrated with respective highlights within their context in the case.
Basic Affordance
1. Outsourcing intersystem actions
Visibility Affordances
2. Detecting process errors 3. Monitoring progress 4. Concealing confidential data 5. Accessing integrated process records 6. Identifying transactions and associated actors
19
Outsourcing intersystem actions
At a low level, and perhaps the most fundamental affordance (why that might be the case is discussed later) afforded end-users with the possibility to outsource particular actions to the RPA. Throughout the case, this affordance emerged in various forms and was also most frequently observed among the affordances. It supported end-users in carrying out their work, mainly from a basic perspective. When an RPA was successfully actualized, it managed to perform the processes which earlier was carried out by an end-user, thanks to the RPAs ability to access and integrate with multiple systems. This affordance may be explained by illustrating its components or potential internal affordance hierarchy. On the most basic level, the RPA can access and integrate with various computer software’s. Thanks to its access, it may consequently be able to, for instance, compare, summarize, or synchronize information. Thus, the RPA software is able to replicate human actions, thanks to the combination of its capabilities and setup.
The variety of observed outsourced tasks took place within several distinct contexts and conditions. However, the characteristic or perception of a process itself was generally an influencing factor. Seemingly, it acted as a stimulating condition, such as a slow, mundane, frustrating, tedious or non-value adding task to carry out. Likewise, high workload or pressure on process owners seemed to be a stimulating condition. One accountant explained how a particular task was carried out before the introduction of one RPA: “it’s
for one hour that you are doing the same, you are copying from one excel you are putting it into excel, into the ERP system and click execute, so it’s for one hour you are just doing that” (User 5).
At the same time, individuals reported how a lack of trust in the RPA may hinder the affordance actualization. Furthermore, it occurred that clerks interfered with RPAs while they were running, causing unexpected errors. Temporary solutions were later developed to diminish the issue by restricting the clerk’s usage of certain SAP functions while the RPA was operating (releasing condition). Another accountant further explained how an RPA currently supports in their daily work:
“we used to load them, but nowadays the robot is doing that part of our work […] it wasn’t easy to get all those memos loaded because we had to check them and load them and do other things which are to be done every day so it nowadays helps quite a lot” (User 1).
In another instance, one developer explained an RPA use case that was actualized by a controller which was supported by simplifying a specific process: ” one automation […] was about exporting data from SAP and
20
Detecting process errors
When a process has been outsourced, the RPA might still be running into errors due to various reasons. For instance, the material which an individual provides the RPA with may be missing or incorrect. In such a case, the robots are able to detect and notify whoever is concerned. In one case, the manager received notifications that enabled him to contact the responsible individual or initiate further actions. Partly, this appeared as an essential function in allowing (releasing condition) to run and outsource the task from the beginning, making sure that errors would neither slip through without taking notice, nor without being solved when performing a specific process. A manager explained their current situation:
we still we have some error tracking that has to happen, so it reframed peoples work so that they are only looking at the exceptions so if […] the robot can’t process it for some reason it throws out an error and that’s where a human needs to get involved and look at it […] [so] we watch the errors that come out of it [the RPA] (User 3).
Monitoring progress
When a task was successfully outsourced and thus no longer needs to be executed by an individual (i.e. actualized), it opened the possibility to monitor the progress of that particular process. As one developer stated about a clerk’s usage of an RPA, “he’s only checking it […] he’s running it when he can simply look,
okay it’s running” (Developer 3). In such cases, the clerk was generally still in charge of, or responsible for,
the outcomes of the process. Assumingly, this is a stimulating condition to the actualization of observing the RPAs working progress. Contrastingly, at times when employees had a low trust in RPA’s capacity, it appeared to be an enabling condition to monitor the progress. Some accountants explained that they were monitoring more frequently in the beginning when they had a lower trust for its ability, while others admitted that sometimes it was more due to their curiosity. For instance, one accountant reflected on their current monitoring habits: “we might do some random checking that it’s okay” (User 1).
Concealing confidential data
21
“especially within HR because most of the data is very confidential, its usually easier […] from GDPR perspective that they are then running it on […] a computer simply for the team so nobody else can have access to it” (Developer 3).
Accessing integrated process records
When an RPA is actualized and carries out an (intersystem) process, it stores a record of the performed transactions. This record can be accessed if the affordance is actualized.
One manager summarized the use and some benefits of this affordance:
we see it brings additional layer of control […] from the audit perspective it creates a more robust audit trail than if you compare that only humans are doing it and of course you have basically the same information then in the system logs of separate systems, but RPA on the other hand creates audit trail for the whole process if it expands from one system to another or across several different systems, so in the RPA process log you see everything that what is happening that one process without having to look in to several different systems (Manager 1).
Another manager was performing a routine check of earlier transactions when realizing that the actualization of this affordance could simplify her task (stimulating condition). Simply, she then could see all transactions within the collective RPA track record, rather than checking multiple systems (enabling condition).
Identifying transactions and associated actors
After having actualized the RPAs integrated process records, one controller looked further into the history of files and the attached information along with each transaction. In doing so, a more detailed view of the process history, including supporting files and attached identification tags, became available to the controller. The stimulating condition here was the preceding unclear communication of identifying responsible individuals in the incumbent process. In other words, instead of contacting the department to find out who was responsible for specific transactions, the controller now had all the necessary details displayed on the screen. An accountant explained the outcome for controllers actualizing this affordance:
22
loaded, so everybody can check what is behind it what is behind the figures, where are they coming from (User 1).
The context and origin of the individual affordances serve as a fundamental basis in the pursuit of answering the research question. Standing alone, however, they are not capable of answering it. Instead, to comprehend the nature of RPA related organizational change and how it is exploited, we need to view the whole picture of how things transpired. Therefore, I now turn to analyze the affordances as a bundle to understand them as a network.
The Affordance Network
As the individual affordance actualization story outlines, each affordance was initially disclosed and analyzed in isolation. However, the power of the affordance lens surfaced when the bundle or network of affordances was taken into account (e.g. Strong et al., 2014). This further analysis benefited in exploring the how, allowing for relations and codependencies to emerge while laying the groundwork for the mechanism identification.
The attentive reader may already have recognized some connections among the affordances in the writing above. It was clearly evident in my data that dependencies existed between the actualized affordances illustrated by the arrows in Figure 2. In fact, an actualization of Affordance 1 appeared to be a basis for most affordances. Similarly, some of the affordances seemingly acted as a stimulating, releasing, or enabling condition for another. For instance, as noted earlier, an end-users awareness of the option to monitor RPA’s progress (Af. 3) or, being notified about unexpected errors (Af. 2). Both may have acted as the enabling, stimulating, or releasing condition in the actualization of Affordance 1.
23
However, individuals in possession of the knowledge to actualize a specific affordance may still be dependent on a factor outside of their control, that is, other end-users. As stated earlier, individual affordances have been the primary anchor in the data, as a result of my study’s end-user focus. However, along with the clear relations among affordances, dependencies between different actor’s actualization also arose. Thus, in order to get closer to an understanding of the entire picture, I will momentarily leave the notion of individual affordances behind. Instead, we can benefit from an analysis utilizing the concept of collective affordances.
The data indicates that even though each individual has distinct goals and responsibilities, they still occasionally depend on each other in achieving their goal. A specific finance-related RPA accurately demonstrates how different actors are dependent on each other to successfully actualize an affordance involving multiple parties. Three individuals with both different positions and locations each had tasks involved in a larger process. In order to successfully actualize Affordance 1 in this case, each of the three individuals had to perform their part of the process by uploading relevant information into an Excel file. The RPA was then capable to take over by carrying out Affordance 1, collectively actualized by the three individuals. After the completion of the process, the RPA automatically left an intersystem track record behind, which anyone could access by actualizing Affordance 52 if there was a perceived necessity to do so.
2 Potentially also Affordance 6
24
Another example reflected the dependency among actualized affordances of multiple end-users. An accountant was able to effectively actualize Affordance 1, a time consuming yet important task. However, the clerk’s manager had some monthly checking which was needed to be done. Simultaneously, a controller also had an interest in the same transaction history, although at a different level. While the manager looked at the general transactions of the process (Af. 5), the controller was rather interested in the details of the prework leading up to those transactions (Af. 6). In doing so, the controller was able to identify the attached documentation of the prework so that responsible actors were identifiable. In other words, despite having different final affordances in mind, both the manager and controller were dependent on the same clerk’s actualization of the outsourcing affordance.
Altogether, the main notion from the findings and discussion above has implications for our further analysis. Mainly, it indicates that after one end-user’s successful actualization of Affordance 1, several paths open up for other (not yet necessarily involved) actors to actualize related yet more specific affordances, i.e. Affordance 2 to 6. With that in mind and by examining each individual mechanism’s implication and fundamental meaning, I could abstract and subsequently organize them into two higher level categories, (compare Table 4). For instance, actualizing Affordance 1 allowed actors to initiate the use of the technology, hence, it constitutes the basic affordance. It served as the stimulating, releasing, and enabling condition for the visibility affordances. Phrased differently, it acted as the first fundamental step in every individual actualization journey. However, that did not hold true when affordances were collectively actualized.
After successful actualization of the basic affordance, the actualization of visibility affordances became possible. Enhanced visibility, by means of transparency and control, could be achieved by actualizing any of the visibility affordances, thanks to the available data from the earlier actualized affordance. The above abstracted categories assist in understanding the implications of affordance actualizations at a group level. Even though these higher order constructs bring us closer to answering the paper’s research question, we still need to comprehend the larger picture as well as its details. In the subsequent section, I demonstrate this through the identified underlying generative mechanisms.
The Generative Mechanisms
25
The inception mechanism
Based on the understanding of the timeline and progression of each event in the case data, I was able to reconstruct and retroduce generative mechanisms. The first mechanism unfolded as similar enabling conditions were noted across cases, in terms of the actualization of the basic affordance. For instance tedious and time-consuming processes. Likewise, this actualization process was also often shown to be a rather complex process. Eventually however, it allowed the end-users to initiate the use of the technology congruently across the cases.
Despite the user focus, the developers turned out to play a key role in the process of the end-user's actualization of the basic affordance. That was since successful actualization was partly dependent on the end-user's capability of communicating an accurate process path (enabling condition). Several instances reflected large amounts of information exchange between the end-user and developer, for example through multiple exchanges, such as email conversations and face to face meetings. However, such interactions could at times be both enabling and constraining the actualization progress. Notably, input was at times conflicting and discussions among process owners could surface multiple conflicting solutions to the same issues. Consequently, such input did sometimes also confuse the developers, constraining the actualization process even further. Likewise, stored process documentation was a resource that repeatedly was assumed to be a releasing condition, however, it often turned out to be outdated and therefore less valuable. Altogether, the developers understanding of other parties’ software (such as SAP) and the end-users processes acted as a contributing factor (releasing condition).
Apart from the above outlined contextual conditions, when the end-users attempt to actualize the basic affordance, some inhibiting factors may interfere, as we saw in some of the earlier examples. Fundamentally, the pressure to actualize, the initial trust, and the perceived potential benefits of actualizing will influence an individual at taking action. Additionally, actualizing was also shown to be impeded by other actors who unknowingly interfered with the RPAs’ work process.
To conclude, the inception mechanism explains the progression of how improvement to organizational routines came about via the basic affordance. Figure 3 demonstrates this process of the
inception mechanism (i.e. from a macro-micro-macro perspective).
26 Figure 3: The inception mechanism.
Figure 4: The resource reallocation mechanism.
The resource reallocation mechanism
27 Figure 5: The transparency mechanism.
The transparency mechanism
In a similar way to how the previous mechanism unfolded, the transparency mechanism also arose from the affordances. Although, this time around the visibility category and an in common enabling condition, namely the basic affordance, early hinted of another existent mechanism. Data generation through the track record, the possibility of being notified about errors at their occurrences, and to monitor performed processes all enabled a new extra layer of control. End-users’ recognition of the introduced additional layer of information through the RPA applications created new opportunities for them to exert control. Together, they represented different ways of overviewing business processes, thus referred to as the transparency
mechanism (see Figure 5).
The appreciation mechanism
This mechanism surfaced as a consequence of the corroboration of another proposed but not confirmed mechanism. More explicitly, management support was only found to be somewhat supported in the data, while neither completely disconfirmed. Later, the RPA team itself seemed to have some influence. The team’s dedication was evident throughout the data and also highlighted in the internal survey. Consequently, the word had begun to spread within the organization, for instance, employees had now begun to approach the developers in the corridors, asking if they had time to converse about new ideas for RPA developments. These observed changes to employee’s mindset were also found at multiple instances within the data, explaining an increased proactive search for opportunities with RPA. Bringing these notions together, the
appreciation mechanism explains how the escalation in development during the RPA implementation was
28 Figure 6: The appreciation mechanism.
Figure 7: RPA augmentation sequence of organizational routines.
29
Discussion
This research began by investigating IT-associated organizational change in the context of a novel technology and how it augments the work within organizations on a socio-material basis. From a critical realist standpoint, I took advantage of the applicability of the affordance actualization lens, in exploring how change evolved in the case of a globally operating manufacturer. I discuss in this section the theoretical and practical implications of the research.
To begin with, I engage with the identified affordances as the building blocks of my theory. Second, while grounded in affordances from an individual level, I show how affordances are not actualized in isolation and that dependencies are evident among them. I also discuss how my findings may compare to other affordance-based mid-range theories in the field. Third, I explain the value of my developed theory from an organizational view supported by the identified generative mechanisms. In doing so, I also confirm the value of utilizing the affordance lens in conjunction with CR. Lastly, I discuss practical managerial implications, limitations of this investigation and future research.
Firstly, the affordances are the building blocks in the development of my preliminary affordance-based theory (see Table 4). In the pursuit of contributing to the IS research and socio-technical mid-range theories, my affordances are anchored in the individual level and were abstracted from their outcomes as a result of their actualizations. Therewith, the data allowed me to explain how and why outcomes occur, along with the philosophical routes of affordances, as in CR. The exploration of each affordance in isolation supported the identification of contextual environments, such as identified enabling, constraining stimulating, and releasing conditions. These findings initially assisted me in the further analysis of the affordance network, their relations and potential dependencies. Likewise, it supported subsequent theory development and later served as beneficial clues towards potentially evident mechanisms.
30
My findings have parallels to Volkoff and Strong’s (2013) study of two different IS system implementations, because they identified dependencies among their affordances in similar ways. In terms of the affordance network, more basic affordances acted as enabling conditions in the actualization of the remaining more advanced affordances. Since my study confirms that particular finding, it seems that while details may vary across cases or technologies, different IS systems may still be similar in some ways. I furthermore confirm Strong et al. (2014) finding that actualization occurs over time, rather than affordances are all being actualized at once.
In difference to Leonardi’s (2013) work on when technology will bring about change, I instead explain how change occurs. While Leonardi proposed that change occurs when individuals use technology in homogenous ways, similarly, I explain that increased utilization and adoption of the technology likely intensifies change, through the appreciation mechanism. However, this differs from Leonardi in one important way. His work emphasizes that increased congruent actualization of shared affordances explains
when change occurs. In contrast, my study shows how congruent affordance actualization may augment
organizational routines, hence causing change. Although, this is not necessarily central to the explanation, because divergent utilization also may account for the appreciation mechanism.
Thirdly, my individual-RPA affordance actualization theory is to the author’s knowledge the first of its kind i.e. devoted to the RPA technology. This contributes to the IS literature by means of an empirically-based CR study. While primary anchored in the individual level, generalized results also account for the organizational context in the form of the basic and visibility affordance categories. Nonetheless, the main contribution from an abstract level lies in the four generative mechanisms. Together, they describe the progression of events in the studied phenomena, empowering our understanding of how RPA augments organizational routines, from the perspective of end-users.
In sum, my contribution can be regarded as a mid-range theory since its theoretical concepts reflect the context of the RPA technology and manufacturing organizations. As such, it equips researchers and practitioners with an understanding of how and why organizational change and associated outcomes occur, rather than predicting the future. Answering such questions resembles the core of CR.
With the above discussion in mind, my study outlines yet another application of the affordance theory and CR in the field. This has shown to be a valuable approach in earlier research (e.g. Volkoff & Strong, 2013; Bygstad et al., 2016). Importantly, it also fulfills the ambitions of CR by generating technology-mediated mid-range theories, while successfully avoiding the trap of generalizing into infinity. Managerial Implications
31
that some foundational affordances act as enablers for more advanced affordances is useful for managers. It is also valuable to explore managerial strategies that appropriately stimulate sources of desired outcomes and prevent impediments. This also assists in classifying issues and their sources as well as avoiding insufficient or undesired actualizations. After all, intervening or providing the right conditions for actualization journeys can assist in achieving better results.
Altogether, these insights support managers during implementations, such as with expectation management and adaptation. Such understanding allows managers to appropriately mediate RPA-related organizational change initiatives. Despite its grounding in a specific case, some generalizability seems reasonable since real affordances are more general in nature, as discussed earlier.
Future Research and Limitations
The data used in this study may be biased as the main focus was on a northern European case. Considering the highly collaborative environment in which affordance actualization seem to occur, geographic culture differences such as individualistic and collectivistic ones could have an influence on the findings (Hofstede, Hofstede & Minkov, 2005).
My study is in line with earlier statements regarding the limited generalizability of CR research (Wynn and Williams, 2012). Nevertheless, continued response to Wynn and Williams (2012) call for more CR research would contribute to the long-term benefits achieved through investigating mechanisms in various contexts, in order to leverage the promising potential of CR research.
Furthermore, the data and findings are primarily anchored in the individual level, hence this embeddedness should be taken into account.
The data collection took place during a short timeframe yet served to represent the entire RPA implementation period. Thus, potential memory retention due to the time between the events occurrences and the interviews may have hindered the providence of a thicker data set. However, this limitation could be addressed by future researchers undertaking an analysis of longitudinal data sets.
32
Conclusion
This critical realist case study explored how the introduction of RPA augments the work of individuals and the related organizational routines. My analysis presents four mechanisms, built on a network of affordances with accompanying contextual details, founded on six individual affordances. Thereby, I present an explanatory mid-range theory building on individually embedded actualized affordances and a focus on mechanisms for change, essential for such theories (Hedström & Swedberg, 1998; Volkoff & Strong, 2013). The affordances interact in a network represented by the captures of basic and visibility affordances. They are central to comprehend the four presented mechanisms; the inception mechanism, resource
reallocation mechanism, transparency mechanism and appreciation mechanism. Together, they describe
the progression of how RPA augments the organizational routines of its end-users.
33
References
Bhaskar, R. 1975. A Realist Theory of Science, Hassocks, England: Harvester Press.
Bhaskar, R. (1998). [1979] The Possibility of Naturalism: A philosophical critique of the contemporary human sciences, London: Routledge.
Brynjolfsson, E., & McAfee, A. (2014). The second machine age: Work, progress, and prosperity in a time
of brilliant technologies. WW Norton & Company.
Burton-Jones, A., & Grange, C. (2012). From use to effective use: a representation theory perspective. Information systems research, 24(3), 632-658.
Bygstad, B., Munkvold, B. E., & Volkoff, O. (2016). Identifying generative mechanisms through affordances: a framework for critical realist data analysis. Journal of Information Technology, 31(1), 83-96.
Carlsson-Wall, M., & Strömsten, T. (2018). Robotisation of Accounting in Multi-National Companies: Early Challenges and Links to Strategy. Per Andersson, Staffan Movin, Magnus Mähring, Robin Teigland,
and Karl Wennberg (eds.), 175.
Carter, B., & New, C. (2005). Making realism work: Realist social theory and empirical research. Routledge.
Coleman, J. S. (1986). Social theory, social research, and a theory of action. American journal of
Sociology, 91(6), 1309-1335.
Danermark, B., Ekstrom, M., Jakobsen, L. and Karlsson, J.C. (2002). Explaining Society. Critical Realism in the Social Sciences, London: Routledge.
Elder-Vass, D. (2005). Emergence and the realist account of cause. Journal of Critical Realism, 4(2), 315-338.
34
Feldman, M. S., & Pentland, B. T. (2003). Reconceptualizing organizational routines as a source of flexibility and change. Administrative science quarterly, 48(1), 94-118.
Fleetwood, S. (2005). Ontology in organization and management studies: A critical realist perspective. Organization, 12(2), 197-222.
Gibson, J. J. (1979). The theory of affordances. The ecological approach to visual perception.
Hartwig, M. 2007. Dictionary of Critical Realism, London: Routledge.
Hedström, P., Swedberg, R., & Hernes, G. (Eds.). (1998). Social mechanisms: An analytical approach to
social theory. Cambridge University Press.
Hofstede, G., Hofstede, G. J., & Minkov, M. (2005). Cultures and organizations: Software of the mind (Vol. 2). New York: Mcgraw-hill.
Isaksson, D., & Wennberg, K. (2016). Digitalization and collective value creation. In Bergström, A., & Wennberg, K. Machines, Jobs and equality. Technological Change and Labor Markets in Europe. European Liberal Forum, Brussels, Belgium
Lacity, M. C., & Willcocks, L. P. (2016). A new approach to automating services. MIT Sloan Management
Review, 58(1), 41.
Lamberton, C., Brigo, D., & Hoy, D. (2017). Impact of Robotics, RPA and AI on the insurance industry: challenges and opportunities.
Leonardi, P. M. (2011). When flexible routines meet flexible technologies: Affordance, constraint, and the imbrication of human and material agencies. MIS quarterly, 147-167.
35
Markus, M. L., & Silver, M. S. (2008). A foundation for the study of IT effects: A new look at DeSanctis and Poole's concepts of structural features and spirit. Journal of the Association for Information
systems, 9(10), 5.
Meneweger, T., Wurhofer, D., Fuchsberger, V., & Tscheligi, M. (2015, August). Working together with industrial robots: Experiencing robots in a production environment. In Robot and Human Interactive
Communication (RO-MAN), 2015 24th IEEE International Symposium on (pp. 833-838). IEEE.
Orlikowski, W. J. (2000). Using technology and constituting structures: A practice lens for studying technology in organizations. Organization science, 11(4), 404-428.
Shave, L. (2017). Your next assistant could be a robot!. IQ: The RIM Quarterly, 33(1), 18.
Strauss, A., & Corbin, J. (1994). Grounded theory methodology. Handbook of qualitative research, 17, 273-85.
Strong, D. M., Johnson, S. A., Tulu, B., Trudel, J., Volkoff, O., Pelletier, L. R., ... & Garber, L. (2014). A theory of organization-EHR affordance actualization. Journal of the Association for Information
Systems, 15(2), 53.
Thapa, D., & Sein, M. K. (2018). Trajectory of Affordances: Insights from a case of telemedicine in Nepal. Information Systems Journal, 28(5), 796-817.
Volkoff, O., Strong, D. M., & Elmes, M. B. (2007). Technological embeddedness and organizational change. Organization science, 18(5), 832-848.
Volkoff, O., & Strong, D. M. (2013). Critical realism and affordances: Theorizing IT-associated organizational change processes. Mis Quarterly, 819-834.
Volkoff, O., & Strong, D. M. (2017). Affordance theory and how to use it in IS research. In The Routledge
Companion to Management Information Systems (pp. 232-245). Routledge.
36
Wickens, C. D., Mavor, A., Parasuraman, R., & McGee, J. (1998). The future of air traffic control - Human operators and automation.
Willcocks, L. P., Lacity, M., & Craig, A. (2015). The IT function and robotic process automation.
Wynn, D. and Williams, C.K. (2012). Principles for Conducting Critical Realist Case Research in Information Systems, MIS Quarterly 36(3): 787–810.
37
