June 2014
User-group dependent adoption interventions:
Opening up the black box
MARGRIET PENNINKHOF
S1806858
Faculty of Economics and Business
MSc Business Administration, Change Management
University of Groningen, Groningen, the Netherlands
Supervisor: prof. dr. A. Boonstra
Academics suggest that adoption interventions can be introduced to give direction to a change management process. The effect of adoption interventions on user-group level is however a black box. This study aims to open up this black box by examining how different user-groups perceive the effectiveness of different adoption interventions in the context of an IS implementation. Based on a two-factor view framework of user reactions, four behavioral categories are determined which represent the different user-groups. This study shows that 1) different user-groups perceive a different combination of adoption interventions as effective 2) different motivations might underlie the perceived effectiveness towards a set of interventions such as the enhancement of trust towards the tool or gained understanding 3) different user-groups hold different combination of preferred ways how to introduce the perceived effective interventions. It demonstrates that discussion and (re)design are effective for all user-groups, while coercion, training and incentives are partially effective to user-groups. It concludes with promising areas for further research on the topic of user-group dependent adoption interventions.
Keywords: change management, IT enabled change, adoption interventions, IS implementation, user reactions, support, acceptance, resistance, user-group dependency
Word count: 14.234
INTRODUCTION
2 demonstrate high support and high usage (Ross & Weill, 2002; Jasperson et al., 2005). Researchers recently argued that an appropriate way to influence individual-level IS adoption behaviors and use is the utilization of adoption interventions (Benbasat & Barki, 2007; Goodheu, 2007; Venkatesh & Bala 2008; Venkatesh, Davis, & Morris, 2007; Rivard & Lapointe, 2012). They suggest that interventions, such as training or redesign, augment user acceptance and IS use.
Academic press calls for further research on this topic (Venkatesh, 2006; Venkatesh & Bala, 2008; Rivard & Lapointe, 2012) and indicate the following promising area: explore if interventions are user-group dependent. Research thus far considered the effect of interventions to be equal for all individual users. In other words, they neglect the potential dependency of interventions on user-groups in order to facilitate the adoption process. This fascinates us since we assume that not all interventions are equally effective for all users. We strongly believe that because of a huge diversity in users’ attitudes and reactions, distinct interventions may be effective for different user-groups based on users’ reactions. For instance users who do not understand a system might need other interventions than users who perceive the system as a threat to their jobs. In order to enhance end-users’ adoption behaviors and use we need to find out which interventions will move user-groups to use as intended by managers. Given the fact that recent findings illustrated the need for considering acceptance and resistance as different dimensions because of existing ambivalent behaviors (Van Offenbeek, Boonstra & Seo, 2013), we use this bipolarity view in our research to define distinct user-groups, each covering their own users’ reactions.
The goal of this paper is to advance our understanding of user-group dependent adoption interventions in the context of IS implementations by exploring how different user categories, drawing on a model of four behavioral categories (Van Offenbeek et al., 2013), perceive the effectiveness of different interventions. In light of this research objective the main question of this research is: How do different user-groups perceive the effectiveness of different adoption
interventions?
3 this study is to diagnose users’ reactions based on the degree of resistance. This will enable us to map the individual end-users in one of the four behavioral categories (Van Offenbeek et al., 2013). These different categories are equivalent for different user-groups. Second, users’ perceptions towards particular types of interventions will be identified and classified. Thereafter, an explorative review will take place to find out how user-groups perceive the effectiveness of different interventions. Finally, we strive to develop propositions on how interventions are user-group dependent, and, if possible, extend and further develop the framework by indicating which interventions, according to the four behavioral categories, will lead to category 1: supportive use. We acknowledge the fact that acceptance and resistance are very complicated concepts and cannot be painted as simple black-and-white dimensions. For instance, people can have different behavioral scores on the dimensions. Moreover, we are aware of the fact that user-groups do not necessarily have to be categorized according to behavior, but can also be formed according to rank or job-role. Yet we choose an intervention-sensitive approach where we focus explicitly on the relationship between interventions and behavioral user-groups based on the model elucidated later on (Van Offenbeek et al., 2013).
This research will contribute to both change management theory, especially to IT enabled change, and also to practice. The former as this paper advances our understanding of adoption interventions during IS implementations by exploring to what extent interventions are user-group dependent and examining how interventions are related to user-groups. Moreover, it shows which interventions will lead to supportive use according to four behavioral categories. Besides, this paper contributes to the literature by applying the two-factor view framework in the context of a new information system of a global company, and more importantly extending it by considering user-group dependent interventions. Overall, findings of this research paper are valuable from a change management perspective since it gives insight into the ways certain change agents respond to a change in their organization.
4 according to distinct behavioral categories, lead to supportive use, can help in building more focused implementation strategies. In this way ambivalent user-groups can be taken into account. Subsequently, the focused strategies will facilitate the implementation process, enhance IS adoption behaviors and supportive use, and ultimately attain the maximum business value from IS.
The paper will proceed with a review of adoption interventions as well as users’ reactions. This is followed by the research method section. Thereafter, a description of the case study will be presented followed by the findings of our research. The findings will be compared to the existing literature in the discussion chapter. In the final section the theoretical and practical implications and conclusions will be addressed.
LITERATURE REVIEW
Relevant theoretical concepts need to be covered in order to clarify the research question. Accordingly, two relevant concepts are defined: interventions and users’ reactions.
Interventions
Since the concept of interventions is very broad and applied in a variety of contexts it can be defined in different ways. However, for this research paper the concept will only be used in the change management context. Venkatesh and Bala (2008, p. 292) define interventions as follows:
‘a set of organizational activities that can potentially lead to greater acceptance of the system’.
Boonstra and de Caluwé (2006) add to this definition the argument that interventions are used to give direction to the change management process. We integrate both definitions and will stick to the following definition: ‘concrete activities to support individual actors and give direction to the
change management processes’. This definition is an integration of various definitions described
5 As pointed out before, researchers (e.g. Venkatesh, 2006) assume that interventions can ease the change transition. Though available evidence already demonstrated the critical importance of developing and implementing interventions (Cohen, 2005; Jasperson et al., 2005), evidence is limited and researchers call for future research on this topic (Venkatesh, 2006). Moreover, even less studies thus far focused on interventions in a user-group context (Seo, Boonstra, & Van Offenbeek, 2011) and as far as we know none of the studies examined user-group dependent interventions according to the four behavioral categories explained in more detail later on.
In this study we want to include different types of interventions which are representative for the taxonomy of adoption interventions mentioned in the current IS literature (Markus, 1983; Venkatesh & Bala, 2008; Rivard & Lapointe, 2012; Van Offenbeek et al., 2013). The following five types of interventions will be subject of this study: coercion, discussing, training, (re)design, and incentives.
Based on the analysis of 89 cases, Rivard and Lapointe (2012) presented a taxonomy of interventions which is because of the richness of case studies and the large quantity of data a valuable and representative overview of interventions. We based our interventions on their study which resulted in four types of interventions: coercion, discussing, training and (re)design. Next to these four interventions, many scholars mention one last important type of intervention which is not included in Rivard and Lapointe’s (2012) study: incentives (Markus, 1983; Beer & Nohria, 2000ab; Ba, Stallaert, & Whinston, 2001). In order to guarantee a representative range of interventions and attempt to give a holistic overview of the relation between user-groups and adoption interventions, we add this fifth type of intervention. Note that the types of interventions may represent different forms of participation ranging from light (e.g. incentives include inform, discussion includes involvement) to heavier forms (e.g. collaboration in redesign) (Mao & Markus, 2004; Venkatesh & Bala, 2008). We will now further elucidate all different types of interventions.
6 inform users and finally insist on compliance towards the system of all (potential) end-users (Markus, 1983). In other words, managers tell their subordinates that they should use a system. We suggest that coercion also incorporates threats of future layoffs, demotion or transfer. Moreover, Beer and Nohria (2000) state that a presence of negative consequences in case of not using the system may facilitate supportive use. For example, if your performance is assessed according to the degree of system use, you feel threatened to move towards supportive use in order to avoid a bad performance assessment which can result in demotion, layoff or transfer. Note that the use of coercive power may reduce trust between different stakeholders (Allen, Colligan, Finnie, & Kern, 2000). Moreover, according to Latham, Erez and Locke (1988) assigning a goal tersely without explanation, so using coercive power and telling people ‘do this’, leads to lower goal performance than if a goal is set participatively.
The second intervention is discussing issues. Rivard and Lapointe (2012) suggested discussing issues as an intervention to increase usage and support. The intervention refers to discussing issues that come up during the implementation process, for instance issues about essential customizations, are openly discussed (Rivard & Lapointe, 2012). With discussing issues you can strive for improvements (such as customization) or for understanding on the tool. Discussion is of course highly related with communication. There are different ways of discussing issues; Rivard and Lapointe (2012) mention round tables, task forces and focus groups. Kotter and Schlesinger (2008) add one-on-one discussions as a way to overcome resistance and enhance acceptance. While investigating users’ perceptions towards adoption interventions we will also address the preferred way of discussion.
7
commands and tools embedded in IS applications; business context knowledge covering the use of IS applications to effectively perform business tasks; collaborative task knowledge covering how others use the application in their tasks’ (Sharma & Yetton, 2007, p. 220). Beside these
knowledge areas, different forms of training exist such as e-learning and traditional classroom training (Zhang et al., 2004). We suggest a third form of training: on the job training which refers to training at someone’s desk while facing real scenarios.
Fourth, redesigning the system is a type of intervention where system features might cause non-acceptance and resistance (Rivard & Lapointe, 2012). Potential end-users may reject the system because it is more of an obstacle than an aid (ibid, 2012). In these cases redesigning the system may enhance use and support (Rivard & Lapointe, 2012). Moreover, Venkatesh and Bala (2008) suggest that besides redesigning a system, user participation in design might facilitate the implementation process and contribute to highly supportive usage. In addition, user involvement in the requirements analysis and testing may influence adoption and use due to emergent causal processes (Markus and Mao, 2004). These four activities do have the potential to move users towards supportive use. We call this type of intervention (re)design, covering four different subtypes: requirements analysis (1), design (2), testing (3) and redesign (4). To be clear we not only measure perceptions towards redesigning the system but also to user participation in (re)designing the system. To clarify, although the system already had its go live, users’ participation in future design, testing or requirements analysis for e.g. new functionalities within this case study might move users towards supportive use. We suggest that possible forms to redesign a system might be more friendly filters or extra functionalities.
Fifth, incentives are a type of interventions which can enhance supportive use by motivating commitment (Markus, 1983; Ba, Stallaert, & Whinston, 2001; Beer & Nohria, 2001; Knowles & Linn, 2004). Markus argues that in some cases neither changing people nor changing technical features of the system will lead to supportive use. For example, users may support the system but may not demonstrate high use if ‘there are no incentives in place for the users for using the
system effectively’ (Venkatesh & Bala, 2008, p. 297). In these cases incentives may facilitate
supportive use by rewarding desired behavior (Beer & Nohria, 2000) or ‘adding extra
8 distinguished in monetary and non-monetary rewards. Illustrations of reward incentives are stock ownership, gifts, bonuses, variable pay, day-to-day recognition and awards (Beer & Nohria, 2000). We will research reward incentives focusing on three types of rewards: bonuses, day-to-day recognition and gifts.
Next to reward incentives, there are two other incentives important in the context of IS implementation: peer advocate and manager advocate (Seo et al., 2011). First, peer advocate refers to coworkers, who hold favorable perceptions towards a new system, acting as advocates in a group. Acting as advocate means that coworkers share and discuss their positive experiences with non-supporting users (Seo et al., 2011). As a result, users’ perceptions and behaviors can be influenced and they can be motivated to support and use the system (Venkatesh & Davis, 2000; Seo et al., 2011). Exactly the same story is valid for management. When managers possess favorable perceptions towards the system and share this with their team members, they can be encouraged to perform the same actions. In sum we subdivide incentives in rewards (gifts, bonuses and day-to-day recognition) (1), peer advocate (2) and manager advocate (3).
To sum up, for the purpose of answering the research question we focus on these five types of interventions summarized in the table below.
Intervention Definition Sub types Example Literature
Coercion “Use of coercive power during an intervention”
“Insist on compliance towards the system of all (potential) end-users”
“If my boss insists on system usage/told me to use the system I would use the system”
(Rivard & Lapointe, 2012; Markus, 1983)
Discussing “Discussing issues that come up during the implementation process”
“Discussing issues to gain our understanding of the system would enhance my use”
(Rivard & Lapointe, 2012)
Training “Educating users and providing them with the required knowledge to help users develop positive perceptions towards the system and influence users’ reactions”
“If we were provided with required knowledge about the system during e-learning or class-room training I would be encouraged to adopt and use the system;
(Markus, 1983; Venkatesh, 1999; Venkatesh & Bala, 2008)
(Re)design “(Re)designing system features that might cause non-acceptance or resistance” Requirements analysis Design Testing Redesign
“If I could participate in the requirements analysis my adoption and use of the system would increase”; “If an extra feature could be added to the system my usage would be enhanced”;
9 Table 1. Adoption Interventions
In contribution, these five types of interventions are highly related with another theory which needs our attention. Knowles and Linn (2004) provided the approach-avoidance model of persuasion where they presented two strategies which can be used to promote organizational change: Alpha strategies and Omega strategies. On the one hand, Alpha strategies refer to increasing the approach forces: ‘an offer or a message can be made more attractive by adding
incentives, creating more convincing reasons, finding more credible sources, and so on’ (Knowles
& Linn, 2004, p. 117). On the other hand, Omega strategies refer to decrease the avoidance forces: ‘removing or disengaging someone’s reluctance to change’ (Knowles & Linn, 2004, p. 118). So, Alpha strategies strive to increase motivation to move towards the change objective while Omega strategies aim to reduce the resistance forces to move away from the change objective (Knowles & Linn, 2004). Knowles and Linn (2004) argue that both strategies need to be included to facilitate organizational change.
Reason for elaborating on this theory is that the intervention types, which are subject of this study, are also covered by the strategies of Knowles and Linn (2004). Alpha Persuasion Strategies cover all types of interventions except for coercion. For instance the strategy ‘adding extra
inducements for compliance’ (Knowles & Linn, 2004, p. 120) refers to incentives which can
increase motivation to move towards the objective. Another example is the increase of source credibility. This refers to (re)design: ‘making the source more attractive to increase their
persuasiveness’ (Knowles & Linn, 2004, p. 120). Omega Strategies incorporate discussion,
training and incentives. For example, Knowles and Linn (2004, p. 127) mention the raise of self-esteem to address resistance indirectly: ‘praising people’. This of course is highly related with day-to-day recognition, subtype of incentives. Another example is the suggestion of Knowles and Linn (2004) to counter argue resistance by using two-sided communication, which in this paper refers to discussion. To conclude, Knowles and Linn’s theory provides useful knowledge and this Incentives “Incentives may facilitate
supportive use by rewarding desired behavior” Rewards: bonuses, gifts, day-to-day recognition Peer advocate Manager advocate
“I would be using the system if my manager would be using it as well”; “If I would receive a bonus for using the system I would be stimulated to use the system”
10 research can confirm or disconfirm if both strategies are needed to move users towards supportive use.
Users’ reactions
The last decades many research is done on the determinants of IS adoption behaviors and use. The majority of the research looked at either user acceptance (e.g. Venkatesh & Davis, 2000) or user resistance (e.g. Markus, 1983; Rivard & Lapointe, 1999). However, recent research (Seo et al., 2011; Van Offenbeek et al., 2013) showed that user behaviors may consist of both user resistance - and user acceptance aspects. Accordingly, ‘ambivalent behaviors such as ‘supporting
but no or low usage’ and ‘resisting but high usage’ do occur and may even change over time (Seo
et al., 2011:68). An example of ambivalent behavior is fervent supporters who do not use a system because of several reasons, e.g. time constraints or personal difficulties (Nagy, Yassin, & Bhattacherjee, 2010). Thus, we must acknowledge that users’ intentions and adoption behaviors cover a range of ambivalence (Nah, Tan, & Teh, 2004; Seo et al., 2011). Therefore, acceptance and resistance should be considered as two different dimensions instead of opposites on one dimension. In light of this finding, Van Offenbeek et al. (2013) developed a two-factor view framework on users’ reactions which shows adoption behaviors of (potential) end-users focused
11 Figure 1. A two factor view on user reactions: degrees of acceptance and support/resistance
As illustrated in the framework, acceptance can be positioned on a ‘unipolar continuum from
non-use to high use’, whereas resistance can be positioned on a ‘unipolar continuum from enthusiastic support to aggressive resistance’ (Van Offenbeek et al., 2013, p. 436; Marakas &
Hornik, 1996). Resistance is here defined as ‘opposition by an actor, or a group of actors, to the
change associated with IS implementation’ (Van Offenbeek et al., 2013, p. 438). Since acceptance
12 These categories present four different user-groups. We assume that during all IS implementations the objective is to move all end-users to behavioral category 1: supporting users (Seo et al., 2011) as shown by the arrows in figure 1. To sum up, the five different types of interventions elucidated in this chapter will be used to investigate users’ perceptions towards interventions, ultimately to find out how user-groups based on the two-factor view framework perceive the effectiveness of different adoption interventions.
METHODOLOGY
This section covers a research design which explains the research approach and methods used. With these applied methods a deeper insight in the relations between the variables is aimed, since the theoretical section already provided theoretical background of the variables. This chapter will describe how data was collected, questionnaires were designed and results were processed. Finally it will set forth how data was analyzed.
Data collection
For this research a case study was conducted at a global company. Reasons for choosing a case study strategy are the following: 1) ‘a case study focuses on understanding the dynamics present
within single settings’ (Eisenhardt, 1989, p. 534) 2) it provides rich data (Yin, 2003). Next to that
and more important, a case study design offers the possibility to advance or refine existing theory through the intimate connection with empirical validity (Glaser & Strauss, 1967). Within the context of the specific case study, the unit of analysis is dual: first, the individual user in order to map individual users in one of the four groups (behavioral categories); second, the user-group (based on the behavioral categories) to find out which interventions lead to supportive use according to the four user-groups. To be clear the individual end-user is defined as ‘the
individual causes the system to act or serve a purpose, brings the system into service, and avails himself of the system’ (McLeod Jr & Clark, 2007, p. 3). In accordance with this definition, people
within the business are the individual end-users within this research paper.
13 2007). First, available documentation from the case study was consulted to further develop an understanding of the project and the context in which the IS was implemented. These documents illustrated the project strategy, -structure, and –vision. Second, conversations with various project stakeholders took place in order to clarify ambiguities regarding the project and its progress. Third, 14 interviews were conducted with a selection of (potential) users. The selection of interviewees was based on the list of target users established by the project team. The interviewees (N=14) were all target users. Moreover, they were intentionally selected from different hierarchical positions (e.g. team leads versus team members), different job roles and classes of business (e.g. finance department and supply department). The respondents list, demonstrated in Appendix B, shows that users represent four different direct user-groups based on job roles. In this way representativeness among the target-user group could be guaranteed (Markus & Mao, 2004). Unfortunately, the different job roles could not be further specified because of confidential information.
For the reason that interviewees needed to be able to speak freely without consequences on for instance their job performance, all interview- and survey questions were conducted on the basis of strict anonymity. Before having the interviews, all participants were re-engaged in the system and moreover received an email with a short description of the research in order to set expectations clearly. In order to enhance understanding and support from the participants there was some time for a face to face introduction just before the start of the interviews.
During the interviews, first survey questions, using a Likert scale, were asked to assess the users’ degree of acceptance and resistance to map the users in one of the four behavioral categories. To be clear, these categories present the four different user-groups. By using a Likert scale ‘a
favorable or unfavorable attitude towards to object of interest’ could be measured (Cooper &
14 For the reason that the interviewer was in the same room as the respondents when conducting the surveys, the participants could ask for clarification if needed.
After conducting survey questions, open-ended questions (Appendix D) were used to find out the users’ perceptions towards the six different types of interventions. We chose for semi structured interviews with open-ended questions since it encourage respondents to tell their experiences (Van Offenbeek et al., 2013) and to gain a deeper understanding of the users’ perceptions towards interventions (Cooper & Schindler, 2003). Next to asking for the established interventions in this paper, the interviewees were asked if they suggest any other interventions which would move them towards supportive use. In this manner we acknowledge the subjectivity of our chosen interventions and leave room for identifying other interventions. The interviews were also conducted face to face which has ‘the obvious benefit of being able to observe and record
nonverbal behavior as well’ (Cooper & Schindler, 2003, p. 205).
Data analysis
This research is primarily based on qualitative research techniques since these provide the opportunity to gain an understanding of users’ perceptions and intentions and adoption behaviors. Moreover, it creates a better understanding of reasons that govern such behavior. In this way, we could find out how different user categories perceive the effectiveness of different interventions. Ultimately, the right path towards supportive use by introducing user-group dependent interventions could be further clarified. In contrary, we used quantitative survey questions, using a Likert scale, to quickly map the users in one of the four behavioral categories. To clarify, in order to find out which interventions lead to supportive use according to the behavioral categories, mapping the users was a prerequisite but not our main research object, and should therefore not go at the expense of available time to ask users’ perceptions towards interventions.
15 started with the acceptance dimension; for the reason that in two cases two questions were asked to measure a concept we calculated the average of the two questions to decide on a final result for this concept. The same is valid for the average of all the concepts under the category acceptance vs. non-acceptance, and the category support vs. resistance. For instance if someone gave a 1 on use of the dashboard, this was equivalent to a 1 on the acceptance dimension. In addition we have to note that some questions were formulated in a reversed (R) way, so instead of ‘I find using the system easy – difficult’ we formulated ‘I find using the system difficult – easy’. This was done in order to guarantee neutrality and avoid pushing respondents in a specific direction, either to positive or to negative answers. If someone gave a 1 on the above mentioned question this was clearly equivalent for a 5 on the acceptance dimension. When the results of both the acceptance and resistance dimension were mapped, the intersection of both points illustrated the final mapping of the specific respondent in one of the four behavioral categories.
16 Besides deductive coding, we adopted an inductive approach to allow ‘research findings to
emerge from the frequent, dominant or significant themes inherent in raw data, without the restraints imposed by structured methodologies (deductive data analysis)’ (Thomas, 2003, p. 2).
During the coding one type arose and only one data-driven code has been added to the coding scheme: communication, which we categorized as discussion (Fereday & Muir-Cochrane, 2006). In addition, findings that caught our eye were taken into account to allow contextual issues to come forward. Accordingly, conclusions could be drawn about which combination of interventions for each different-user-group lead to supportive use. Consequently, recommendations were proposed about the most effective package of interventions for different behavioral categories and the preferred ways how to introduce the interventions. The results from the case analysis were compared with existing literature. In this way ‘conclusions about
relationships between the variables are suggested’ (Aken, van et al., 2007, p. 164).
RESULTS
In this section the findings of our study will be presented. For the purpose of answering our research question, this chapter is split in three parts. First, a case study description will be given. The second part asks whether users’ reactions can be mapped in one of the four behavioral categories and the third part will discuss how the user categories perceive the effectiveness of different adoption interventions.
Case description
17 accessible, timely, accurate and up-to-date. Unlocking the value of data and showing the proper data about stocks and performance measures along the supply chain in an easy, transparent, consistent and user-friendly way aid to better decision making.
18 Figure 2. Business Case
The dashboard itself operates inside an internet browser, in the so called BI launch pad, the standard web portal for users of the SAP BusinessObjects Business Intelligence (BI) platform. The dashboard is fed by multiple databases from multiple systems and applications. The dashboard is updated twice a day and users can log in by single sign-on once they have received access.
Users’ reactions
19 users in category 2 we leave this category out of scope in this research and we will not draw any conclusions about adoption interventions regarding this category. Rather we focus on how supporting users (1), resisting non-users (3) and supporting non-users (4) perceived the effectiveness of different adoption interventions. Noteworthy is that the figure not only shows the mapping of the users but also gives us information about the level or function from the specific respondent. For instance, the triangles are team leads whereas the circles refer to channel optimizers and product optimizers. Although it was not our objective to find out how user categories based on job-role perceive the effectiveness of different interventions we will not exclude the chance that any remarkable things could jump out at us. However, once again, our main focus will be on user categories based on our behavioral two-factor view framework as showed below.
Figure 3. User reactions of respondents12
The figure tells us that no more than 3 out of 14 respondents are mapped as supporting users. Only R3, R9 and R11 showed moderate use and moderate support. Clearly, this indicated low usage and adoption behaviors of the target users, and confirmed the business case as mentioned
1
Respondent 2 was exactly in between category 3 and category 4. Based on the verbal and nonverbal behavior during the interview I concluded that the respondent tends to fit in category 4 rather than in category 3. 2
20 in the method section. To find out how we could move different user-groups towards supportive use as asked by management, we will discuss now the perceptions towards different adoption interventions in the context of this IS implementation, summarized in Appendix F.
User-group 1: supporting users
Before analyzing how supporting users perceived the effectiveness of different adoption interventions, we first take a look at the position of the respondents in figure 5. As illustrated, supporting users showed moderate use on the acceptance dimension and are between constructive cooperation and neutral on the support vs. resistance dimension. Although respondents in this group were already categorized as supporting users, there is still room for improvement on both dimensions to increase supportive use. Regarding usage, the users indicated that they use the dashboard on a weekly basis. Reason for usage is to have a clear overview and visualization: ‘a
clear overview and visualization of the stocks and liftings and be able to present graphs and visuals to other people’ (R11).
Interventions
As shown in Appendix F, supporting users perceived the following interventions as effective: discussion, training, (re)design and peer and manager advocate (two subtypes of incentives). On the contrary, coercion was not perceived as an intervention which increases supportive use. Two out of three respondents argued that coercion would not help them because they see no value in the tool, while only one said that she would have no other choice then to obey ‘yeah what else
can you do’ (R3).
Discussion was positively perceived by three out of three respondents because it gains understanding on how specific functionalities work, for instance the download function: ‘if
someone shows the tool to you and says ‘it has this features’ it is more likely that someone is going to use it then after reading manuals’ (R3); ‘that helped a lot when I had issues with the download, that I could just quickly call person Y or X… and X shared his screen and told me you have to click here, here, here and then it works’ (R9). Moreover, discussion would be an opportunity to get system-related questions answered: ‘clear name not an inbox or a desk, just to
21 about the tool (showing concrete examples of value) and creating awareness by involving people would move them towards supportive use: ‘yeah, I think it is really first communicating about the
tool, make people aware that the tool is existing …the best way for this case is that you have some users who can already derive and demonstrate value and you can start build examples’
(R11). We categorized clear communication as discussion because communication and engaging people is a form of discussing issues during the implementation process. A preferable way of discussion is one-on-one discussions, not necessarily face-to-face but a clear name. One respondent mentioned that in order for discussion to be effective, a clear structure is needed: ‘you
need a clear structure to be sure you can change things and take actions following from the feedback’ (R11). So discussion is only effective if the person you discuss issues with actually has
the power to answer your questions and take action.
Besides discussion, training was also considered to be a preferable intervention by two out of three respondents. Motivation is that training helps to see potential value, provides required knowledge on the tool and helps to get used to it: ‘helps you to see what you can do with the tool,
with the distinct tabs, because maybe you can get some nice graphs out of it and I don’t have any idea’ (R3); ‘and to know where you have to go, to get used to the tool’ (R11). In despite of the
effectiveness of training to increase users’ usage and acceptance, supporting users argued that training in this case was not of huge importance because the tool is easy to use: ‘self-explanatory’ (R3) and ‘no rocket science’ (R11). Preferred ways of training are classroom training (R3) and e-learning (R11).
A third preferable intervention is (re)design consisting of the following subtypes: requirements analysis, design, testing and redesign. All respondents agreed on the fact that requirements analysis, design and testing would move them towards supportive use. That is to say it would enhance their usage and acceptance because they could give feedback on what is needed to bring value for them out of the tool: ‘sometimes IT tools do not understand or fit with the business part,
design is very nice but it is not fit for purpose’ (R9); ‘if you can participate in creating something you can give them feedback on what is important to you’ (R3). However, these subtypes would
22
asked. You just asked them to test it, just does it work yes/no’ (R11); ‘which has quite an impact at usage afterwards is how issues that pop up during testing are handled… you would not believe that they really get it fixed by the time it is working’ (R9). Subtype four (redesign) was found
useful by all supporting users as well because a redesign of certain system features (filters including download function) would make the tool more user-friendly, trustable, and valuable to use, and would remove negative emotions while using the tool: ‘it is absolutely user unfriendly.
That is why I marked the more negative emotions. If you want to change the date rate, I would say in two or three cases the cursor just get stuck’ (R9); ‘to download the stuff and translate it into normal stuff takes hours’ (R3). All subtypes of this intervention were perceived as effective;
nevertheless, in despite of (re)designing the system supporting users would consider it as a ‘nice
tool to have’ (R9).
As mentioned earlier, supporting users considered peer and manager advocate as effective subtypes of incentives while rejecting rewards incentives. They argued that peer and manager advocate, i.e. managers and peers sharing their positive experiences, would enhance acceptance and usage because it creates visibility and awareness: ‘it creates visibility and awareness when
you have a team manager in a meeting saying that he makes use of the tool and that he finds this and this really good, it encourages you to use it’ (R11). Moreover, it would encourage them to
use the tool in order to be able to talk on the same level: ‘talk about the same things. I know that
if I talk to him and I know he is checking that, I would check it as well, just to be on the same level’ (R9);
User-group 3: resisting non-users
None of the participants in this category used the IM dashboard on a daily basis. As shown in figure 5 users in this category were demonstrating on average low use and passive resistance.
Interventions
23 their further career growth. Yet they would make a remark that it doesn’t make any sense to use it: ‘of course I would. And I would ask him to make a remark that I think it doesn’t make sense’ (R8).
The second preferable intervention is discussion. Five out of five users were positive because discussion opens up the possibility to complain and think about solutions: ‘a really clear focal
point which I can call, to whom I can complain and together we can think about solutions’ (R12).
Moreover, through discussion the tool would come closer and the tendency to use is higher:
‘what I think does work is, so to speak the way we are sitting here together and conducting an interview and discussing the tool. As a result I think afterwards I will have a look at the dashboard again. All of a sudden it comes closer’ (R12); ‘it was not advertised enough to us obviously and I was waiting for more news’ (R10). On top of that, respondents thought that
concrete examples of main stakeholders who extracted value from the tool by relying on the tool to make their decisions would stimulate them to use it: ‘if there are good examples in Asia where
it is being used a lot, like some proven record. If the CO community would feel that this is a good way to base their decisions on, they would be stimulated to use it’ (R10). Preferred way of
discussion is one-on-one discussion through either interviews (R12, R7, R13) or through business focal points (R10, R12, R8). Note that all users stated that the person with who you discuss issues should have the power to do something: ‘actually you are kind of talking to a vacuum’ (R10).
The third effective intervention according to resisting non-users is training (four out of five). According to them training would enhance their usage and acceptance because in this way they would be urged to learn about the tool and they would understand better what the reasons and possible values are: ‘at least you already know what the dashboard is about and you understand
better the reasons and the possible value… at my desk with real scenario’s’ (R8); ‘yeah, to tell us why is there, what is it, how it could be useful, where is it already useful’ (R10). Preferred way of
training is on the job (three out of four) in small groups since it would be easier then to focus one’s attention on the tool instead of in a call with 30 other people.
24 would be valuable to involve people from the business from the beginning to gain understanding and enhance trust. Respondents did not necessarily want to participate themselves but proposed that one from each team, who has all the knowledge and can share his/her experiences with the others, would participate: ‘one person who understand it completely, and that is also the point
who goes to his colleagues saying ‘yes it is a great thing’ or ‘no I don’t like it’. So but you just have to concentrate you on this specific user which is a lot easier’ (R7). In addition, they
perceived (four out of five) redesign as effective because extra functionalities (including the functionality to see real-time data) would change the tool into a useful tool for their jobs: ‘it
would be helpful because then you can see the stocks…I need the difference between partners. So I need functionality. Additionally I have more possibilities to have different views on the data’
(R7); ‘more details on what happened. For example I am missing the distribution channels which
are really important to our jobs’ (R8).
The perceptions along the resisting non-users towards different subtypes of incentives varied extremely. None of the five users was positive towards subtype 1: rewards. However, three out of five users perceived peer advocate (subtype 2) as effective for the reason that it would build a common language and it would enable users to talk on the same level: ‘it becomes more common,
more normal to use it. Then you have a common language, you talk together about something you saw in the tool, or about the green line so to speak’ (R12). In addition, manager advocate was not
perceived as effective by three out of five users because ‘a manager can use it for whole other
purposes’ (R13). Hence, we conclude that only peer advocate as subtype of incentives is a
preferable intervention for resisting non-users.
User-group 4: supporting non-users
First, users from this category affirmed they use the IM dashboard seldom to never. Reason for use is that the tool provides an overall picture. As shown in figure 5 they are mapped slightly above low use and neutral on the support dimension.
Interventions
25 would not work: four out of six respondents were very clear in stating that they would not use the system if their line manager would insist on usage for the reason they would not see any value in using it. However, two respondents perceived coercion as effective in the view of further career growth: ‘if my team lead says we have to do it, I like to have a good performance also for my
further development’ (R6).
Discussion was indicated as a valuable intervention by four out of six respondents for the reason that it would increase awareness, improve understanding of the tool and enhance trust and respondents’ feeling that the tool is user-friendly: ‘it helps us to get closer to the tool and feel that
the tool is user-friendly for us’ (R4). Quick discussion and support is important because issues
hinder usage and if discussion could not take place quick enough people rather tend not to use the tool: ‘some people just don’t bother and will generally not use the tool’ (R14). Preferred way of discussion is one-on-one discussion: through IT employees or through business focal points. Two respondents out of six who rejected discussion indicated that redesign would be a precondition to move them towards supportive use: ‘no, discussion is not effective; it got to be usefulness enough
to make me move’ (R1).
Training was not mentioned as an effective intervention since only three out of six supporting users perceived the intervention as effective. In general, this user category found training not effective because the tool is easy to use: ‘relatively straightforward to use’ (R1) and ‘pretty easy
to work with’ (R5). The other three respondents liked training because in this way they would
gain knowledge on the system: ‘learning by doing it’ (R14).
Contrary to training, supporting non-users were convinced of the effectiveness of (re)design. Requirements analysis is a preferable intervention for supporting non-users (four out of six). In contrary, design was not perceived as effective by four out of six users because it could slow down the process: ‘it would also slow down the process because people are also cultural stuck in
their way of what they are doing’ (R1). Five out of six users shared the opinion that testing would
enhance their usage and support (subtype 3). Their motivation is that ‘it should be helpful to get
26
we can’t fix, this is something we can’t fix but this is something that will be resolved. This will work definitely’ (R14). In addition, four out of six respondents agreed on subtype four: redesign;
things that would need to be redesigned are more friendly filters and extra functionalities (including real-time data). Redesign would help moving supporting non-users towards supportive use because after redesigning the system, the system would be sufficiently good for their daily jobs: ‘I know that the timing of refreshing the system is going up from one time a day to two
times. That is not sufficient for the things we are doing. Real time, yes, then we can talk about this tool could be used in the future tool’ (R6). Moreover, redesign would help restore the trust in data
which would move users towards supportive use: ‘there were some problems with the links and since I have not trusted the data anymore. This is my main issue’ (R14). Hence, requirements
analysis, testing and redesign were perceived as effective by supporting non-users.
Regarding incentives, four out of six rejected reward incentives. Four out of six participants argued that manager- and peer advocate would facilitate supportive use because it would show the value and the bigger picture: ‘yes, if you see someone else using it, then you see ok let’s also
try it. If he sees some value in it, you may also see value in it. It may increase the chances for usage’ (R2). Moreover, top management support would help according to two respondents in
order to ‘show the bigger picture’ (R2).
27 intervention should be introduced. So, for instance, although redesign was perceived effective by all categories, preferences about how the redesign should be introduced differ per user-category; user category 1 prefers friendly filters while user category 3 prefers real-time data and extra functionalities.
‘yes’/’no’= effective/ineffective intervention (…/…) = number of respondents who perceive the interventions as effective/ineffective followed by an explanation
Cursive text= preferred way how to introduce the intervention
Table 2. Summary of analysis
Coercion Discussion Training (Re)design Incentives User-group 1 Supporting users No (2/3) No value in the tool Yes (3/3): one-on-one Gains understanding on the tool, shows value
Yes (2/3): e-learning and class-room
Helps to see potential value, to get used to the tool and provides required knowledge on the tool
Requirements analysis: Yes (3/3) Design: Yes (3/3)
Testing: Yes (3/3)
Redesign: Yes (3/3): friendly filters Requirements analysis, design and testing: opportunity to give feedback on what is needed in the tool, enhances trust
Redesign: makes the tool easier, user-friendly, easily accessible, trustable, and more valuable
Rewards: No (3/3) Peer advocate: Yes (3/3) Manager advocate: Yes (3/3)
Peer - and manager advocate: creates visibility and awareness, be able to talk on the same level User-group 3 Resisting non-users Yes (3/5) Future career growth Yes (5/5): one-on-one Enhances trust, possibility to complain and think about solutions, tendency to use higher
Yes (4/5): on the job
Urged to learn about the tool, gains understanding of possible values
Requirements analysis: Yes (5/5) Design: Yes (5/5)
Testing: Yes (5/5)
Redesign: Yes (4/5): extra functionalities Requirements analysis, design and testing: gains understanding, enhances trust
Redesign: makes the tool useful for daily jobs
Rewards: No (5/5) Peer advocate: Yes (3/5) Manager advocate: No (3/5)
Peer advocate: builds a common language, be able to talk on the same level User-group 4 Supporting non-users No (4/6) No value in the tool Yes (4/6): one-on-one Increases awareness, enhances trust, improves understanding of the tool and enhances respondents’ feeling that the tool is user-friendly, shows value
No (3/6)
Straightforward and easy to use
Requirements analysis: Yes (4/6) Design: No (4/6)
Testing: Yes (5/6)
Redesign: Yes (4/6): extra functionalities and
friendly filters
Requirements analysis and testing: gains understanding
Design: good as is, can slowdown the process Redesign: restores trust, makes the tool useful for daily jobs
Rewards: No (4/6) Peer advocate: Yes (4/6) Manager advocate: Yes (4/6)
28 DISCUSSION AND CONCLUSION
In this section the discussion and conclusions will be addressed. The discussion is threefold; first, the analytical findings will be compared with the existing literature in adoption interventions and user reactions. In which way do our findings support existing findings and how do they build further on existing theory? This will result in suggested propositions that describe how different user-groups perceive the effectiveness of different adoption interventions. Second, theoretical and managerial implications will be analyzed. Third, research limitations and further research will be examined.
Theoretical interpretation of results
In line with the work of Van Offenbeek et al. (2013) and Seo et al. (2011) on users’ reactions, this paper took a two-factor view considering acceptance and resistance as two different dimensions and recognizing ambivalent behaviors. The expectation was that ambivalent behaviors would be found along the respondents. The findings support the existing theory by showing the presence of ambivalent behaviors (six out of fourteen users were mapped in user category 4: supporting non-users). In addition, this research acknowledged the four different user categories as mentioned in the two-factor view framework and took this as starting point for its analysis. Moreover, in accordance with the work of inter alia Benbasat and Barki (2007), Venkatesh and Bala (2008) and Rivard and Lapointe (2012) this paper assumed that adoption interventions could be introduced to move users towards supportive use: ‘interventions can
support individual actors and give direction to the change management process’ (Boonstra & de
29 The findings suggest that 1) adoption interventions can be introduced to facilitate supportive use 2) groups perceive a different combination of adoption interventions as effective 3) user-groups hold a different combination of preferred ways regarding how their perceived effective interventions should be introduced 4) different motivations might underlie the perceived effectiveness towards a set of interventions. The first suggestion supports the work of many academics (e.g. Benbasat & Barki, 2007; Venkatesh & Bala, 2008; Rivard & Lapointe, 2012) by showing that respondents admit the perceived effectiveness of certain adoption interventions in moving them towards supportive use. So, interventions are indeed a way to give direction to the change management process and support therefore the statement of Boonstra and de Caluwé (2006).
In addition, the second suggestion builds further on existing theory by showing that adoption interventions, as suggested by literature, are user-group dependent based on the two-factor view framework (Van Offenbeek et al., 2013). In other words, user-group dependency should be taken into account when aiming to effectively introduce adoption interventions. This research confirms the two-factor view framework of Van Offenbeek et al. (2013) by providing evidence for ambivalent behaviors and by articulating the ability to use the framework as foundation for the introduction of user-group dependent interventions. This framework as foundation for the categorization of user-groups is used before by Seo et al. (2011). However, they only mentioned strategies to promote IS adoption by user-group and did not research the relationship between adoption interventions, as suggested by literature, and user-groups. This study therefore opens up the black box of user-group dependent adoption interventions by showing that user-groups perceive a different combination of adoption interventions as effective.
30 (Zhang et al., 2004) is present in all IS implementations regardless of the content of the case study. Hence, the framework can be expanded by mentioning the effective set of adoption interventions per user-category and the preferred ways of introducing these interventions. This results in the following illustration:
Figure 3. Set of perceived effective interventions per user-category and preferred ways of introducing these interventions3
First, the figure illustrates that different user-groups perceive a different combination of adoption interventions as effective. Second, the figure demonstrates that different user-groups hold also a different combination of preferred ways regarding how their perceived effective interventions should be introduced. Consequently we suggest the following propositions:
3
31
Proposition 1. Different user-groups perceive a different combination of adoption interventions as effective
Proposition 2. Different user-groups hold a different combination of preferred ways regarding how their set of perceived effective interventions should be introduced
Before continuing with the discussion of the fourth suggestion resulting from the findings, the former propositions should be viewed in light of the work of Knowles and Linn (2004). As showed in figure 3, each set of effective adoption interventions as perceived by the user categories entails both Alpha and Omega strategies. Since discussion was incorporated by both the Alpha and Omega strategies and this type of intervention is indicated as preferable by all three user categories, we can conclude that both types of strategies need to be included to facilitate the change management process. Therefore, this study contributes to the work of Knowles and Linn (2004) by confirming their argument that both strategies are critical.
The fourth suggestion resulting from the findings, that different motivations might underlie the perceived effectiveness towards adoption interventions by user-groups, is a novel learning. No specific ideas were developed about how user-groups perceive the effectiveness of the interventions. Moreover, academics did not provide evidence before on this topic. Therefore, it is valuable that this research showed how different user categories perceive the effectiveness of different adoption interventions. Several propositions suggested later on in this chapter will refer to how effectiveness is perceived by different user-groups. This learning can be used to further build theory in the context of user-group dependent adoption interventions.
non-32 users. On the one hand, we thought that supporting non-users might have a lack of knowledge on the tool which would hinder them to use the tool. Subsequently, we thought that training could solve this gap. On the other hand, we assumed that resisting non-users would not be encouraged by training because providing knowledge on how to use the tool would not reduce their resistance. Our research thus supports the academics in recommending discussion and (re)design while disconfirming the value of training for all user-groups. However, we think the reason that training is not perceived as effective by supporting non-users has to do with the easiness of the tool. In this case study no required knowledge on the tool was needed because the tool was straightforward. Besides, we think that training is perceived as effective by resisting non-users because training reduced resistance by facilitating conditions by providing guidance for assistance on the tool to see the possible values which they could not see themselves without training. So, seeing the value in the system by the use of training would reduce resisting non-users’ resistance.
33 side note in mind when introducing this intervention. This brings us to the following proposition:
Proposition 3. Coercion is effective towards resisting non-users because not complying to coercive power would hinder their further career growth
According to Rivard and Lapointe (2012), discussion of issues during an implementation process encourages usage and support as well. The findings support Rivard and Lapointe’s argument by showing that discussion is perceived as effective by all user categories. So both resisting non-users, supporting non-users and supporting users shared the opinion that discussion would move them towards supportive use. This leads to the following proposition:
Proposition 4a. Discussion is effective towards supporting users, supporting non-users and resisting non-users
While discussion was perceived as an effective intervention for all user-groups, reasons for this effectiveness slightly differ per user-group. Supporting users, category 1, stated that discussion is effective for them because it would increase users’ understanding of systems functionalities, clarify ambiguities regarding system usage and shows value on the tool. Resisting non-users, category 3, preferred discussion because it would be a possibility to complain about issues and think about solutions, and to come closer to the tool. Supporting non-users, category 4, argued that discussion would move them towards supportive use because it increases awareness, enhances trust, improves understanding of the tool and enhances respondents’ feeling that the tool is user-friendly. Commonality among these reasons is that discussion on the one hand helps to gain more understanding on the tool, and on the other hand increases more trust on the tool. As a result users would move towards supportive use. For the purpose of exploring how user categories are receptive towards interventions we articulate the following proposition:
Proposition 4b. Discussion is effective for supporting users, supporting non-users and resisting non-users because it helps to gain more understanding on the tool and it increases trust on the tool
34
Proposition 4c. Discussion is preferred in the form of one-on-one discussions by supporting users, supporting non-users and resisting non-users
The third type of intervention which is recommended by literature (e.g. Markus & Mao, 2004; Sharma & Yetton, 2007) to give direction to a change management process is training. The findings suggest that supporting users and resisting non-users were positively receptive towards training, also illustrated in figure 3 and table 2. So, against expectations, even for resisting non-users training would be a way to move them towards supportive use. Motivation is that training reduces resistance because training facilitates conditions by providing guidance for assistance on the tool to see the possible values which they could not see themselves without training. Motivation for the effectiveness for both user-groups is threefold: 1) training provides the required knowledge on how to use the tool 2) it helps to see potential values 3) it helps to get used to the tool. These three reasons confirm the emphasis of training stated by Markus (1983) and Venkatesh (1999): educating users and providing them with the required knowledge to effectively use the system. If we look at the former mentioned motivation, training is effective for these users when it covers the following knowledge areas: application knowledge area and the business context knowledge area (Kang & Santhanam, 2003-2004; Sharma & Yetton, 2007). Furthermore, findings show that preferred ways of training for supporting users are e-learning and classroom training while resisting non-users prefer on the job training.
In contrary to user category 1 and 3, supporting non-users (category 4) did not perceive training as effective in this case study for the reason that the tool is straightforward and easy to use. We therefore add three propositions regarding the intervention training:
Proposition 5a. Training is effective towards supporting users and resisting non-users are because 1) it provides the required knowledge on how to use the tool 2) it helps to see potential values 3) it helps to get used to the tool
Proposition 5b. Training is preferred in the form of e-learning and class-room training by supporting users and in the form of on the job training by resisting non-users
35 Rivard and Lapointe (2012) recommend (re)designing system features as a way to move users towards supportive use. As findings show, (re)design was, besides discussion, found effective by all three user categories. Therefore we articulate the following proposition:
Proposition 6a. (Re)design is effective towards supporting users, supporting non-users and resisting non-users
However, the question arises how the different user categories perceive the different subtypes of (re)design. Markus and Mao (2004) and Venkatesh and Bala (2008) argue that requirements analysis, design and testing (subtype 1-3) facilitate the adoption process and can enhance usage and support. Findings validate the former argument by showing that these three activities open up the possibility to give feedback on what is needed in the tool and to get to know the tool. However, while requirements analysis and testing were found effective by all three user categories, design was not indicated as preferable intervention by all three user categories. The latter was found effective by supporting users and resisting users but not by supporting non-users. So, our research shows that academics’ suggestions are not valid for all user-groups. This leads us to the following suggestions:
Proposition 6b. Requirements analysis is perceived effectively by supporting users, supporting non-users and resisting non-users because in that way feedback can be given on what is needed to extract value out of the tool and to get to know the tool
Proposition 6c. Design is perceived effectively by supporting users and resisting non-users because in that way feedback can be given on what is needed to extract value out of the tool and to get to know the tool
Proposition 6d. Testing is perceived effectively by supporting users, supporting non-users and resisting non-users because in that way feedback can be given on what is needed to extract value out of the tool and to get to know the tool
