University of Groningen Faculty of Economics and Business MSc Business Administration, Change Management 21 August 2013

COPING WITH ADOPTION BEHAVIOURS DURING THE IMPLEMENTATION OF AN

ELECTRONIC PATIENT RECORD

University of Groningen

Faculty of Economics and Business

MSc Business Administration, Change Management

21 August 2013

P. Cordes

W.A. Scholtenstraat 12-5

9712 KW Groningen

(06) 50518534

p.cordes@student.rug.nl

student number: 2196530

Supervisors at the university

dr. M.A.G. van Offenbeek

dr. J.F.J. Vos

2

COPING WITH ADOPTION BEHAVIOURS DURING THE IMPLEMENTATION OF AN

ELECTRONIC PATIENT RECORD

Abstract

Recent research in information system (IS) adoption found acceptance and resistance,

formerly assumed to be two opposite ends of one dimension, to be positions on two

separate dimensions. This study applied a newly designed framework, which separates

these dimensions, to an Electronic Patient Record (EPR) implementation programme and

confirms the multiple dimensionality of acceptance and resistance behaviours. Although

the findings indicate that both dimensions are related, because antecedents which relate to

both dimensions are found. This study also expands the research on adoption behaviours

by focusing on the pre-implementation phase of this programme. Besides looking at the

intended user behaviours, this study also investigated the expected user behaviours by

implementers to see how they differ. This is important as success of IS adoption depends

on the expectations of implementers. If implementers are aware of differences between

their expectations and intended user behaviours, they can develop more focused

implementation strategies to achieve better results. The findings show that in the

pre-implementation phase, implementers’ expectations of the intended user behaviours are

besides global and diffuse also more sceptical than future users intended. Due to a lack of

measurable standards, instead of planning interventions, implementers planned to setup

these measurable standards, partly by acquiring more information. Doubt and updating,

which are elements of sensemaking, help implementers to interpret the intended user

behaviours.

Keywords: expected user behaviours; intended user behaviours; acceptance; resistance; implementers; IS adoption

INTRODUCTION

3

They argue that managers need to succeed in adopting the technology within an organizational context

and in establishing appropriate conventions for its use.

Introducing the term EPR raises the question of what it actually covers. There seems to exist

some ambiguity in literature about the term EPR as it is just one among many, which is depicted as a

central technology in supporting the examination, treatment, and care of the patient. Defined as the

electronic version of the paper-based medical record, the content of an EPR is not defined in a universal

manner (Jensen & Aanestad, 2007). As a result, different terms are used, some of which are (Lærum,

2004); Electronic Patient Record (EPR); Electronic Health Record (EHR); Electronic Medical Record

(EMR). This paper sticks to the term EPR. Greenhalgh et al. (2008) and Greenhalgh et al. (2009) argue

that an EPR is a complex innovation that must be accepted by individual patients and staff and should be

embedded in organizational and system level routines. This process is heavily influenced by material

properties of the technology, people’s attitudes and concerns, and interpersonal influence. An EPR can

range from an isolated file of computer-held information on a single patient, with or without decision

support functions, to a nationally networked database offering built-in interoperability functions with

other technologies and systems. The purpose and scope of EPR develops along with technology

(Greenhalgh et al., 2009). According to Jensen and Aanestad (2007), in recent years, there has been an

increasing demand for exploiting the possibilities of IS in healthcare. In many hospitals, the focus is on

EPR, hospital managers perceive IS as the key tool for achieving a better information flow and better

services, as well as for complying with organizational objectives regarding high quality in patient care

and treatment (Jensen & Aanestad, 2007). In the IS literature stream there is wide agreement that

acceptance and resistance are crucial factors in IS adoption (Van Offenbeek et al., 2013).

The research of Van Offenbeek et al. (2013) and Seo et al. (2011) on acceptance and resistance

behaviour in IS adoption, disentangles and combines these two perspectives, formerly assumed as being

opposite ends of a single dimension (Lapointe & Rivard, 2005), in a two dimensional framework (Figure

1, p.7) for describing and analysing complex behaviours during IS implementation. Their research

focused on users’ intentions and behaviours, because the behaviours that users display are crucial to the

eventual adoption. Implementers can use the framework as a tool to assess and monitor people’s

behaviour during IS implementation and to develop differentiated interventions (Van Offenbeek et al.,

2013). Lapointe and Rivard (2005) state that individual behaviour that arises independent from one

another, early in an implementation process, may later converge into group behaviours. User reactions

interact due to social influence and individual behaviour may result in group level phenomena (Van

Offenbeek et al., 2013). So, (intended) adoption behaviours are also important in the early phases of an IS

implementation project as behaviours evolve. The study of an EPR implementation programme will prove

useful as it comprises the concept of an IS.

4

implementers’ responses to user resistance, and more specifically, when this resistance has already

occurred.

So, as new ISs are not always adopted and used as intended by implementers, the latter can use

the framework of Van Offenbeek et al. (2013) to assess and monitor people’s (intended) behaviours

during IS implementation. Understanding the antecedents of user reactions offers cues on how to manage

them. Van Offenbeek et al. (2013) propose that resistance needs managerial intervention directed at the

context in which the system is being used and in the case of non-acceptance, interventions need to address

the system’s functionality and design. Support and acceptance should be viewed differently by

implementers as supportive behaviour is a resource they can use and acceptance is more a performance

indicator of system implementation (Van Offenbeek et al., 2013). It will be interesting to understand how

implementers reshape their expectations of intended user behaviours during different implementation

phases, once they obtain knowledge about the intended user behaviours. This is interesting to know,

because if implementers act upon a difference between their expected user behaviours and intended user

behaviours, through interventions, expected and intended user behaviours will change, and so will the

outcome and success of these interventions. But how do implementers interpret the intended behaviour of

others: how can we explain the expectations implementers have? Sensemaking might help to understand

this process.

Sensemaking is the process of social construction that occurs when discrepant cues interrupt

individuals’ on-going activity, and involves the retrospective development of plausible meanings that

rationalize what people are doing (Weick, 1995; Weick et al., 2005). In other words, sensemaking is the

process by which people give meaning to experience. Central to the development of plausible meanings is

the bracketing of cues from the environment, and the interpretation of those cues based on salient frames.

Sensemaking is thus about connecting cues and frames to create an account of what is going on (Maitlis

& Sonenshein, 2010). Thus, in situations where implementers become aware of intended user behaviours,

sensemaking might determine how implementers interpret these behaviours and adjust their expectations

accordingly. The workings, and relevance, of sensemaking in change situations will be discussed in more

detail in the Sensemaking chapter.

5

first question can be answered by adapting and applying the framework, proposed by Van Offenbeek et

al. (2013), for identifying and categorizing reactions to IS implementation, to suit an EPR implementation

case study. In doing so, the expected user behaviours of implementers and intended user behaviours can

be reflected upon, to analyse how they relate to each other on the user group level.

Applying the framework of Van Offenbeek et al. (2013) in the context of a pre-implementation

phase of an EPR programme will test its usability. By also applying the framework to find expected user

behaviours by implementers, this research will contribute to the relevance of acceptance and resistance

theories for IS adoption. The outcome is expected to be relevant for IS implementers, who should be

aware of the potential differences between their assumptions about, and apparent reality of, the intended

behaviours by future users. So that they can develop more focused implementation strategies, through

interventions, that, hopefully, will lead to better outcomes. This paper is structured as follows. In the next

chapter, important concepts in the theoretical framework of Van Offenbeek et al. (2013) are reviewed.

Subsequently, the relevance of sensemaking for coping with the intended user behaviours by

implementers is explained. The fourth chapter describes how, through an embedded case study of an EPR

implementation programme, quantitative and qualitative data was gathered and analysed and how is

reflected upon the results. The findings of these analyses are presented in the fifth chapter. The paper

concludes with a discussion about the most important findings, the contributions of this research for IS

adoption, and directions for future research.

IS ADOPTION

This chapter reviews relevant acceptance and resistance literature in the IS adoption stream, to clarify the

attempt of Van Offenbeek et al. (2013) to integrate these two research streams. Van Offenbeek et al.

(2013) used the term ‘adoption’ to cover acceptance and resistance behaviours, while in prior research

(e.g. Lapointe & Rivard, 2005, 2007; Meissonier & Houzé, 2010; Marcus, 1983) the term

‘implementation,’ is used often. Adoption goes beyond implementation, in the sense that it also comprises

the use of an IS, subsequent to its implementation. An IS encompasses much more than just the computer

based technology, in which data can be entered, processed and retrieved. Keen (1981) states that ISs,

when applied in an organizational context, increasingly alter relationships, patterns of communication and

perceived influence, authority and control, and the development of such a system involves an intensively

political and technical process. He also recognized the link between information and power and suggested

that adoption of an IS required a strategy to deal with the politics of data (Keen, 1981). Thus, IS adoption

is also a political process (e.g. Keen, 1981), where different interests of user groups causes competition

among them to increase power, control information, obtain a greater share of computing resources, and

achieve preferred task allocation (e.g. Markus, 1983; Mumford & Pettigrew, 1975).These political issues

primarily come forward when a system cuts across multiple user departments (Markus, 1983).

6

conceptualized resistance as a separate reaction next to acceptance, involving different behaviours and

underlying mechanisms. The next two sections will elaborate on these two dimensions, to understand

their differences following the two-factor view of Van Offenbeek et al. (2013). To assist in the

understanding of the different models, it is useful to note that most acceptance and resistance models have

their roots in psychology (Van Offenbeek et al., 2013).

Acceptance

One of the most prominent IS acceptance models is the Technology Acceptance Model (TAM),

introduced by Davis (1989). This model proposes that perceived usefulness and perceived ease-of-use are

determinants of IS use. According to Van Offenbeek et al. (2013), these two variables have become the

core variables in technology acceptance research, although later studies better explained use behaviours

by extending the range of variables and refining their measurement. Venkatesh and Davis (2000), for

instance, saw subjective norms, images, job relevance, output quality and result demonstrability as

determinants of perceived usefulness. Later on, Venkatesh et al. (2003) developed the Unified Theory of

Acceptance and Use of Technology, known as the UTAUT, which is an integrated model of user

acceptance models. It sees performance expectancy, effort expectancy, and social influence as

determinants of the intention to use and facilitating conditions as a direct determinant of use behaviour.

The inclusion of environment-based voluntariness as a moderator in shaping an individual’s intention to

use, in the UTAUT, proved to be an important contextual dimension in the attempt to disentangle

acceptance and resistance (Van Offenbeek et al. (2013).

Acceptance is conceptualized at the individual level as it explains the intentions of individual

users to use a system and by implicitly restricting acceptance behaviours to ‘system usage’, the

behavioural component of acceptance is equivalent to use (Lapointe & Rivard, 2007; Van Offenbeek et

al., 2013). Acceptance studies have not set out to explain resisting behaviours, as the acceptance’s

behavioural component is often positioned on an unipolar continuum from non-use to high use (Liang &

Xue, 2009).

Resistance

Markus (1983) made an important contribution to resistance research when she identified three different

views on overcoming resistance of IS implementation. The people-determined view argues for

HR-oriented interventions, the system-determined view asks for modification of the technical features of the

system, and the interaction theory identifies an interaction between users and the system. The interaction

theory has several distinct variations of which the sociotechnical and the political variants are most

important. The sociotechnical variant focuses on the distribution of responsibility for organizational tasks

across various roles and on the work-related communication and coordination around this division of

labour. The political variant explains resistance as a product of the interaction of system design features

with the intra-organisational distribution of power (Joshi, 1991). The political variant also raises the

interesting view that wider contextual issues may affect IS adoption (Van Offenbeek et al., 2013).

7

emphasizes that users’ reactions to IS implementation are determined by evaluation of the contextual

consequences, rather than of the system itself (Van Offenbeek et al., 2013).

The Multilevel Model of Resistance to ISs, developed by Lapointe and Rivard (2005), argues that

resistance behaviours occur following perceived threats that result from the impact of an IS in terms of

the interaction between individual and/or organizational level initial conditions and features of the system.

Important implications are that resistance behaviours can vary over time and may vary from apathy to

aggressive resistance. Lapointe and Rivard (2005) see resistance as a social phenomenon were individual

resistance behaviours may converge into group resistance and power and status loss can lead to resistance.

Besides including contextual issues, resistance research also explains the resistance behaviours of all the

actors involved, not just the users. These characteristics of resistance research reflect how resistance

theories have a broader scope than acceptance theories. Unfortunately, Lapointe and Rivard did not

examine if impact also works the other way around, in that impact of the system may also result in

supportive behaviours.

According to Van Offenbeek et al. (2013), the resistance continuum with aggressive resistive

behaviours on the one end includes supportive behaviours on the other end of the continuum. Judson

(1991) argues that support is important to achieve maximum benefits from change. Research on

supportive behaviours is limited. Coetsee (1999) is one of the few who addresses support and proposes a

dimension ranging from apathy, support, involvement to commitment, and passionate commitment. From

this review on prominent acceptance and resistance theories, it can be concluded that acceptance and

resistance are two separate dimensions, as both are triggered by a different set of causative mechanisms

(Van Offenbeek et al., 2013). Therefore, Van Offenbeek et al. (2013) introduced a framework to enable

the connection and probable combination of the two research streams.

Relating acceptance and resistance

Van Offenbeek et al.’s (2013) earlier review also shows most research on IS adoption theories to focus on

either acceptance or on resistance and they argue that the relationship between acceptance and resistance

has received little attention. Their research tried to understand the relationship between and the

co-occurrence of acceptance and resistance in IS adoption. Therefore, the authors propose a two-dimensional

adoption framework for mapping user reactions. Figure 1 depicts the continuums of

acceptance/non-acceptance and support/resistance that are viewed as separate behavioural dimensions. The

acceptance/non-acceptance dimension is defined as the user’s employment, or not, of an IS to perform a

task (Van Offenbeek et al., 2013). The support/resistance dimension is defined as support or opposition

by an actor, or a group of actors, to the change associated with IS adoption (Van Offenbeek et al., 2013).

8

(non-acceptance). For the resistance dimension the concepts can lead to activities aimed at supporting the

system, its implementation, its use and its consequences (support) or block or hinder the systems

implementation, its use and its consequences (resistance) (Van Offenbeek et al., 2013).

Figure 1. A two-factor view on user reactions: degrees of acceptance and support/resistance. Adapted

from “Towards integrating acceptance and resistance research: evidence from a telecare case study,” by

M. Van Offenbeek, A. Boonstra and D. Seo, 2013, European Journal of information systems, 22, p.438.

Van Offenbeek et al.’s (2013) as well as Seo et al.’s (2012) research shows that acceptance and

support/resistance behaviours can co-exist and that the variance in these behaviours may differ by user

group and over time. This creates the somewhat ambivalent categories of supporting non-users and

resisting users. The former category, that of supporting users, differs considerably from resisting

non-users. Supporting non-users might be positive about the initiative, but do not feel the immediate need to

use the system, or do not know how to imbed it in their routines. The latter category, that of resisting

users, brings a closer understanding of the significance of voluntariness in the intra-person interaction

between acceptance and support/resistance reactions. Voluntary use occurs when users perceive

technology adoption to be a wilful choice, whereas in a mandated environment, users perceive use to be

organizationally compulsory (Venkatesh & Davis, 2000). There is a continuum from voluntary to

mandatory use (Karahanna et al., 1999; Brown et al.,2002) and in a fully mandatory environment,

workers will end up being either supporting or resisting users. Although enthusiastic supportive and high

use behaviours are highly desirable, these are often an utopia. Which adoption behaviours suffice depend

on the behaviours implementers expect from future users.

Expected behaviours

9

the intended and the expected behaviours. When intended behaviours of future users become apparent to

implementers, during adoption, this will likely change their earlier expectations. How implementers

interpret this information and thus how they cope with their expectations, may depend on their

sensemaking.

SENSEMAKING

As stated earlier, sensemaking is the process by which people give meaning to experience. It is an active

process that involves the interaction of information seeking, meaning ascription, and associated responses

(Thomas et al., 1993). Sensemaking includes extracting particular behaviours and communications out of

streams of ongoing events (i.e., bracketing), interpreting them to give them meaning, and then acting on

the resulting interpretation. This occurs in conversations that involve giving accounts or self-justifying

explanations of events and activities (Ford et al., 2008). “An account is a linguistic device employed

when action is subject to evaluation, particularly when there is a gap between action and expectation or

between promise and performance” (Scott and Lyman,1968: cited in Ford et al., 2008: 364). Sensemaking

is mainly used to describe the process of meaning ascription by the workforce whereby managers act as

sensegivers. However, sensemaking can also be used to explain process of information interpretation

between and among managers. And although sensemaking is triggered by any interruption to on-going

activity, change is a condition that, because of the degree of disruption it incurs, offers a particularly

powerful occasion for sensemaking (Maitlis & Sonenshein, 2010). Maitlis and Sonenshein (2010)

identified ‘shared meanings’ and ‘emotions’ as two core elements present in the sensemaking context of a

change situation.

Shared meanings

Expectations is a kind of shared meanings that can be especially important in change situations as they

connect with cues to create meaning (Weick, 1995). Individuals then filter subsequent cues against this

meaning and gradually build up confidence about a definition of the situation. Shared meanings are vital

to sensemaking, but also potentially destructive. Expectations can be both enabling and constraining.

Both overly optimistic (i.e., never get achieved) and overly negative (i.e., cascade down to the work floor)

expectations can have potential harmful effects. However, individuals can, of course, update their

expectations in situ based on new cues. Nevertheless, expectations are sticky and this is where the danger

lies, as individuals hold onto familiar meanings (Maitlis & Sonenshein, 2010).

10

environments in which they are constructed. Besides shared meanings, emotion is a second, although less

explored, element important for sensemaking in change situations identified by Maitlis and Sonenshein

(2010).

Emotion

Since sensemaking involves bracketing cues in ways prompted by certain frames, others’ expressions of

panic or dismay can powerfully frame our interpretation of an ambiguous situation (Schachter & Singer,

1962). Further, these emotional expressions can be contagious, significantly affecting group sensemaking

processes (Hatfield et al., 1994). In organizational change, members’ emotions provide valuable data to

leaders about how a change is being received and so contribute to its enactment (Rubin et al., 2005).

Further, managers who notice and attend to the emotions of their employees are more likely to succeed in

their change implementation efforts (Huy, 2002; Sanchez-Burks & Huy, 2009).

Turning to positive expressed emotions, in organizational change situations, leaders often express

excitement and enthusiasm to signal their commitment to the new direction (Maitlis & Sonenshein, 2010).

Also, expressed positive emotions act as a sensegiving resource to influence employees’ understandings

of the value of the change (Huy, 1999; Huy, 2002). However here is also a downside, sensegiving that is

positive and optimistic can create blinkers for those in teams and organizations, causing them to overlook

or collectively reinterpret cues that signal potential danger (Kayes, 2004; Weick, 1993).

So, the intended acceptance/non-acceptance and resistance/support behaviours by future users are

clearly important for the adoption of an IS by its (future) users. Also, these dimensions can be used to

reflect on differences between the intended user behaviours and expected user behaviours by

implementers as these are important for the successful management of the project. Moreover, it will be

interesting to see if implementers act upon these differences as this might change the course and outcome

of the project. Furthermore, the intended user behaviours may influence the expected user behaviours by

the implementers through acts of sensemaking. The next section elaborates on how this study examined

this in the case of an EPR implementation programme; how data is collected and analysed.

METHOD

For understanding implementers’ expectations of future user behaviours, and whether they match the

intended behaviours of future users, during an EPR implementation project, a case study can provide rich

data. According to Eisenhardt (1989) case studies can be used to generate theory and is a research strategy

which focuses on understanding the dynamics present within single settings, this particularly suits this

research. Greenhalgh et al. (2009) state that a case study approach, in relation to EPR research can

illuminate how contextual factors shape, enable, and constrain new technology supported models of

patient care.

Research design and case context

11

This research expands the recent findings from Seo et al. (2011) and Van Offenbeek et al. (2013)

on acceptance and resistance behaviours in IS adoption, in that it tried to find out if there is a difference

between the intended behaviours of future users and expected behaviours by implementers during the

pre-implementation phase of a multiple year covering EPR pre-implementation programme. No prior hypotheses

exist, the premise of this research is that both behaviours match. This embedded case study reflects on

qualitative data about implementers’ expectations and quantitative data about future users. The initial

number of approached future users was over 1500, so quantitative data gathering seemed the only feasible

option for the time available. The advantage of a larger number of quantitative data is that it adds to the

reliability of the research, if the data is representative. Since the implementers were in lesser number,

qualitative data provided richer information. This research used a positivist approach, in that knowledge is

gained from observable experience. This approach is in part contextually bounded, as it necessarily

depends on the researchers’ appreciation of the situation (Van Offenbeek et al., 2013).

A healthcare institution in The Netherlands is preparing the implementation of an organization

wide EPR. This multiple year covering EPR programme was an IS implementation that comprises fifteen

coherent projects, only some of which are directly IS related. All users of the multiple legacy health care

systems should work together in one Patient Record. This is expected to involve large changes in the

daily work of many employees. Future users which should actively use the system are grouped, by the

implementers, into the following user groups: doctors; nursing staff; paramedics/perimedics (para/peri);

management; and administrative personnel.

Quantitative data collection

Data collection, for answering the first sub question; how and why in the case of EPR

implementation the intended behaviours of potential user groups differ from those expected by the

implementers, proceeded as follows. To measure the intended behaviours of user groups a survey was

conducted among four user groups. The survey was determined through consultation between researchers

from the University of Groningen and project staff of the EPR programme (as part of a larger research

conducted by researchers from the University of Groningen). The survey comprised 25 statements, see

Appendix B, which were derived from eight concepts relevant in IS adoption; acceptance, support, power,

impact, emotion, ease-of-use, usefulness, and facilitating conditions. The concepts, apart from acceptance

and support, are all recognized by Van Offenbeek et al. (2013) as antecedents of either acceptance or

support. Changes in power lead to either support or resistance, depending on an increase or decrease in

autonomy, power position, or financial consequences. Impact is not used as an antecedent by Van

Offenbeekt et al. (2013), but, as discussed in the IS adoption chapter, impact might lead to resisting or

supporting behaviours depending on the interaction of features of the system with individual and/or

organizational level initial conditions. Emotions may also lead to supportive, if positive, or resisting

behaviours, when negative. Ease-of-use (using the system would be free of effort or not), facilitating

conditions (an infrastructure exists to support use of the system or not), and usefulness (the system

produces useful outcomes or not) relate to actual usage of the system and therefore to acceptance.

12

apprehend (Converse & Presser, 1986). A not applicable box was added as a result from the pilot, to

exclude respondents checking the neutral box if they had no opinion about a statement. The pilot and the

final survey, including the statements’ codes and underlying constructs a priori, can be found in Appendix

A respectively Appendix B. All management staff of the three users groups were selected to participate in

the survey and all employees of the departments ophthalmology, orthopaedics, geriatrics, and the

oncological branch of obstetrics and gynaecology, from which staff fell under the three user groups.

Further, a random sample was drawn from the remaining employees of the user groups; doctors, nursing

staff, and para/peri. This came down to a total number of 1,579 participants in the survey. The digital

survey was managed by staff from the EPRs’ change management project, so the targeted user groups,

targeted departments, and the sample size were not up for discussion or alteration.

Quantitative data analysis

The quantitative data was checked for significant differences in age and gender compared to the sample,

to indicate if a non-response bias would exist. An exploratory factor analysis was performed using a

principal component analysis (PCA) with Varimax rotation to see if the statements loaded on the same

factors as a priory expected. Items were excluded from next analyses when they did not load well on a

factor, the reliability of the factor increased when the item got removed, and when the item did not match

on a theoretical basis. For the resulting factors, multi-item scales were created by summing up the items

(using inverted ones for the negatively loaded items) comprising each factor and dividing them by their

number. By doing so, the data is made quasi interval, allowing different parametric statistics to be used.

Significant differences between the means of the first and fourth quartile per factor were tested, using a

T-test, as a final check to see if the factors are relevant, in that they measure variation.

Regressions were used to discover relations between the remaining constructs resulting from the

factor analysis with acceptance respectively support. If relationships are supported, between acceptance

and/or support and other constructs resulting from the state-of-mind, this has implications for the

interpretation of the two dimensions. Prior to the regression, any outliers, per factor, were removed from

the data set. Outliers influence the mean and increase the standard deviation, which is undesirable (Field,

2009). A record was seen as an outlier if its value deviated more than 2.5 times the standard deviation

away from the mean. This method is known as the Schweinle method. Significant differences in mean

scores between user groups, on acceptance and support, were tested using one way ANOVA.

Qualitative data collection

13

(Eisenhardt, 1989). Implementers’ expectations of the intended behaviours of user groups were measured

by using the specific antecedents for the acceptance and resistance dimensions as used in the framework

of Van Offenbeek et al. (2013). To find out if and how sensemaking is involved in the meaning ascription

of the reported intended behaviours and how implementers update their expected behaviours through

sensemaking, the attended meetings were also observed for occurrences of shared meanings and emotions

in the reactions of the implementers to the intended user behaviours.

For the second sub question; how implementers act upon differences between their expected user

behaviours and the intended user behaviours per user group, two meetings of the change management

project team were observed in which possible actions or interventions based on differences between the

expectations of implementers and the apparent intended user behaviours might be discussed. These

meetings took place in the weeks after the presentation of the intended user behaviours (i.e. the results of

the survey) to the implementers.

Qualitative data analysis

Quotes from the three interviews, quotes in the document of the change management project of the

programme, and quotes from the observed meetings were coded based on a start-list of codes.

Acceptance, non-acceptance, support, and resistance formed the main categories and their specific

antecedents formed subcategories (coding schema in Appendix C and example codes in Appendix D). If

no user group was specified in a quote, the quote was ascribed to all user groups. The identified quotes

were held up for review with a member from the Change Management project group, to check their

internal validity. Besides this deductive method, for the attended meetings, attention was given to other

issues raised by participants regarding the expectations of implementers and apparent intended user

behaviours, to not impose derived codes on data on which they do not apply, and allow relevant specific

concepts, cultural references or contextual issues to come forward (Hennink et al., 2011). The quotes

from the expected user behaviours by the implementers are specified on user group level, they were used

to position expected user groups’ acceptance and resistance towards the EPR in terms of the framework of

Van Offenbeek et al. (2013).

Implementers’ reactions, which came to the fore when they became aware of, and discussed, the

intended behaviours of user groups, during the meetings were also observed. Shared meanings and

emotions were used as concepts, and their elements as categories, to code these reactions (see also

Appendix C), in order to discover how implementers interpreted the intended behaviours of future users.

For answering the second sub question, any action or intervention that was proposed, during the meetings,

in light of the revealed differences between intended behaviours and expected behaviours per user group,

or overall, are reported in the Findings section.

FINDINGS

14

planned interventions that were proposed, based on perceived differences, by the implementers, between

the intended and expected user behaviours, in the meetings of the programs’ change management project

team.

Descriptive statistics

In total 525 participants responded to the survey (33%). The data showed no extreme entries, however

three duplicate respondent IDs were removed. Unfortunately, the mean age confirmed a significant

difference (see Appendix E) with the sample. Kaiser-Meyer-Olkin’s measure showed a sampling

adequacy of .895 and Bartlett’s Test of Sphericity showed a significance of .000, both indicating a factor

analysis to be suitable (Field, 2009). The scree plot showed the best result when five factors were used,

see Appendix F. The PCA resulted in the following five factors: usefulness, impact, support, facilitating

conditions, and acceptance. Inverted variables were used for statement 4 and statement 10, because they

loaded negatively on their respective factors. This means that these statements (i.e. their interpretation)

were inverted also. Unfortunately, statement 5, measuring resistance, did not load well on any factor, so

for the support/resistance dimension only the support side is measured. Some theoretical concepts were

freely interpreted to fit the case context as impact stands for the impact of the EPR on the work of

employees and facilitating conditions comprises the implementation process by the organisation.

The rotated component matrix of the final PCA can be found in Appendix F. Each final factor

was reliable, as all factors had a Crohnbach alpha higher than .70 and lower than .90, see Appendix G (the

coded items comprising each factor can also be found here). Histograms of the factors minus the outliers

(Appendix H) showed near normal distributions for all factors. Significant differences between the means

of the first and fourth quartile per factor were tested as a final check to see if the factors are relevant, in

that they measure variation (Appendix I).

Table 1 shows some interesting associations between the constructs. As expected, impact is

related to support and acceptance is related to usefulness and facilitating conditions.

Table 1 Descriptive statistics and correlations between the research variables

Means (m), standard deviations (SD), and correlations between the research variables (n=525)

Variable

m

SD

1

2

3

4

5

6

1 Gender

65% = F

2 Age

44.38

11.54

-.143**

3 Usefulness

3.46

.71

-.011

-.108*

4 Acceptance

3.99

.79

-.066

-.122**

.508**

5 Impact

3.44

.76

-.125*

.093

.052

.133*

6 Fac. cond.

3.34

.59

-.077

.140**

.484**

.346**

.072

7 Support

2.87

1.09

-.072

.160**

.318**

.370**

.214**

.472**

* Significant at the 0.05 level (2-tailed)

15

But, somewhat surprisingly, facilitating conditions is also associated with usefulness and support, and

impact is also related to acceptance. Furthermore, support also relates to usefulness and acceptance, this

indicates that in this case, users who intend to use the EPR also support it and vice versa. Also, the table

shows 65% of the respondents were female. Mean age is presented in years of age. The mean scores on

the research variables range can range from 1, respondents strongly disagreed with the statements

comprising the variable, till 5, respondents strongly agreed with the statements. A score of 3.46 on the

usefulness variable for instance means respondents believe the EPR to be moderately useful and a score

of 2.87 on the support variable means respondents have a neutral/slightly not-supportive attitude towards

the system.

Intended behaviours of user groups

Linear regression of the independent variables and the dependent variable acceptance shows (Table 2)

that a positive relationship exists between them. 31.1% of the variation in acceptance is explained by

usefulness, facilitating conditions, and impact. Facilitating conditions and impact also show to have a

significant positive relationship with support and usefulness does not, 21.5% of the variation in support is

explained by these independent variables (Table 2).

Table 2 Linear regressions for acceptance and support with the independent variables

Variable

Acceptance

Variable

Support

_{β step 1 β step 2}

_{β step 1 β step 2}

Gender

-.044

-.028

Gender

-.073

-.052

Age

-.140*

-.136

Age

.063

.021

Usefulness

.382***

Usefulness

.086

Facilitating conditions

.217***

Facilitating conditions

.356***

Impact

.138*

Impact

.173*

R²

.020

.331***

R²

.010

.225***

∆R²

.311*** ∆R²

.215***

n=220

n=157

* Significant at the 0.05 level

** Significant at the 0.01 level

*** Significant at the 0.001 level

16

user groups, see Appendix J. All user groups scored significantly higher on acceptance than they did on

support. The major difference in number of respondents, per user group, between acceptance and support,

stems from fact that the acceptance construct consists out of 2 statements and the support construct out of

3 statements. If for one statement the not applicable box was selected, the respondent was excluded from

the analysis. Logically, the chances of this happening with 3 statements is greater than with 2 statements.

Table 3 ANOVA results for acceptance per user group

Mean scores (m), standard deviations (SD), and differences in mean between user groups

Acceptance

_{User group}

_n

_x̅

_s

₁

₂

₃

₄

1

_{Nursing staff}

_{156 3.99 .80}

_x

2

_Doctors

_{159 4.11 .78}

_.11

_x

3

_Para/Peri

_{92 3.46 .97 -.53* -.64*}

_x

4

_Management

_{30 4.18 .94}

_.19

_.08

_.72*

_x

* Significant differences in mean at the 0.05 level

Scale Acceptance: 1 = intend to not actively use the system, 5 = intend to actively use the system

Table 4 ANOVA results for support per user group

Mean scores (m), standard deviations (SD), and differences in mean between user groups

Support

User group

n

x̅

s

1

2

3

4

1

_{Nursing staff}

_{95 2.96 1.05}

_x

2

_Doctors

_{102 2.90 1.02 -.06}

_x

3

_Para/Peri

_{76 2.44 1.07 -.53* -.46*}

_x

4

_Management

_{25 3.77 .92}

_{.81* .87* 1.33*}

_x

* Significant differences in mean at the 0.05 level

Scale Support: 1 = not support implementation of system, 5 = active support implementation of

system

Expected user behaviours by implementers

17

to take this step at once, changing their roles in the process. “Make nursing care electronic at once; A new

mind set for nursing staff, for the nursing staff that will be a real change.”

Table 5 number of coded quotes per user group

Acceptance/non acceptance Code Nursing staff Doctors Para/Peri Management useful - not useful a-1 vs na-1 2 1 3 2 1 1 1 1 motivated - not motivated a-2 vs na-2 3 3 3 3 easy to use - not easy to use a-3 vs na-3 1 1 1 1 1 capable to use - not capable a-4 vs na-4 infrastructure - no infrastructure a-5 vs na-5 1 1 1 1 1 will use the system - will not use a-6 vs na-6 2 1 1 2 1 1 1 1 required to use - not required a-7 vs na-7 5 5 5 5

Total 11 7 10 8 8 6 8 6

Support/resistance Nursing staff Doctors Para/Peri Management increase of power - decrease s-1 vs r-1 1 1 1 1 1 consistent with norms - conflicts s-2 vs r-2 2 better quality of life - lower s-3 vs r-3 4 positive emotion - negative s-4 vs r-4 impact - no impact s-5 vs r-5 9 7 7 7 activities to support - block or

hinder s-6 vs r-6 1 5 1 6 1 4 1 4

Total 1 17 6 14 1 12 1 12

Doctors’ expected user behaviours were also, specifically, mentioned by the implementers. While

the implementers expected that doctors would perceive the EPR as useful, they also expected that their

lack of interest in, and preparation for, the system would hamper their acceptance. “They don’t put energy

into it, because they are not interested.” But also, like the nursing staff, their resistance level was

expected to be quite high due to the impact the system will have on their work: “A change of this

magnitude generates resistance almost by definition.”

18

however quality is also a big project with different themes and issues which keeps everyone occupied. But

for the whole hospital the impact is noticeably larger. Everything will change at once. Work methods will

change and you already have to think about that now, to be able to make the transition.”

One quote in the programs’ change management document best sums up implementers’ situation

regarding their expected behaviours of future users: “At this moment, the knowledge about the (potential)

resistance, willingness to change and the expectations which live in the organisation is global and diffuse.

However, we have the intention to actively direct and act on this and also look at the effects of the actions

undertaken.”

Reflecting on intended and expected user behaviours

When looking at the expected user behaviours by implementers and intended behaviours per user group,

both for the nursing staff and the doctors, implementers expected quite some resistance, while the survey

showed that these groups have a neutral attitude toward the system. Also, for para/peri and management

some resistance was expected by the implementers, while this was partly true for para/peri, management

did intent to support the system as the survey showed. Unfortunately, for the intended behaviours by

future users, a statement measuring resistance in the survey, was removed in the factor analysis as it did

not load well on a single factor. For the intended behaviours by future user groups only the support side

of the resistance/support dimension was measured. Although all user groups showed a neutral attitude

towards the system, this indicates that they did not intended to resist the system, at least not at that point

in time.

Looking at the intended acceptance/non-acceptance dimension, three user groups scored around 4

(on a scale of 1 till 5) and para/peri scored 3.46, while the implementers expected some problems with the

acceptance of the system for all user groups. It is important to note that the expected user behaviours by

the implementers were not phase specific, meaning that ‘when’ they would expect these future user

behaviours was not specified. Although the above analysis indicates that a difference exists between the

expected behaviours, at any point in time, and the intended behaviours of future users, at one point in

time. In the discussion and meetings subsequent to the presentation of the intended user behaviours,

implementers stated that the intended user behaviours were as they expected: “There are no surprises for

me; It confirms exactly how we are in it; I am quite satisfied with it.” Which is remarkable as there were

no measurable expected behaviours. “…I see this state-of-mind monitor as a baseline measurement,

because we have no prior measurement, so now we know how people think about it.”

Sensemaking

19

were found: “So we have to have a product first; If that is a goal than you have to say what we should do

to get those people in that mind-set.”

Another finding, in the discussion after the presentation, was that first an occurrence of negative

expectations was observed: “In itself it is not that bad, in that I thought that there would be more

resistance.” While later on in the discussion, on multiple occasions, implementers stated that the intended

user behaviours confirmed their expectations. “There are no surprises for me; …this is in reasonable

agreement with my expectations now I am here.” So, while the implementers had negative expectations or

no clear expectations at first, later on, when the implementers discussed the, in light with their

expectations, positive results, the implementers acted like this was what they expected. In doing so, the

implementers updated their expected user behaviours

Actions

To the extent to which implementers had expectations about the intended user behaviours, after the

presentation of the intended user behaviours, these prior expectations corresponded with the intended

behaviours, based on the statements of the implementers reported in the previous section. The

implementers however, had not specified as to what kind of intended behaviours they desired in the

upcoming phases of the program. Therefore, the implementers proposed to render their desired user

behaviours concrete by means of what they called a change management calendar, which should serve as

an indicator as to what kind of intended user behaviours the implementers want, at certain points in time,

during the program. “…concretise to change management calendar.”

To determine what kind of intended user behaviours to pursue for, the implementers planned to

speak with the user groups to acquire information as to what the users meant with their scores. “The

state-of-mind monitor… the quantitative measurement is a start for us to inquire what the user groups mean

with that.” Also, this information should serve a new version of the programs’ change management

document, which will be used as a plan of action for the change management project team in the

following phases of the program. “…this is great for in the PID (name of the change management

projects’ document); …no, there is no interpretation yet, we have to describe this clearly in the PID.” So,

the implementers did not act upon the intended behaviours of user groups, at that time, in the form of

(planned) interventions in the implementation program. They did plan to use the presented intended user

behaviours as input and convert these into quantifiable expected, or rather desired, user behaviours, so

that in the next phases of the EPR implementation program the implementers might monitor the intended

user behaviours against these desires to see if these match with their goals.

DISCUSSION

20

this research tried to find out if implementers acted on differences between their expectations and the

apparent intended user behaviours.

The positive relationship between facilitating conditions and support as well as with acceptance

in the pre-implementation phase of an EPR implementation is relatively easy to explain as facilitating

conditions in this phase of the programme relates to the organizational infrastructure to support

implementation, instead of use, of the system. And the support/resistance dimension is determined by

wider contextual issues. Impact of the system on the work of the future users relates to both acceptance

and support. If future users expect the changes the new system will entail, will improve how they carry

out their work, they will likely intent to use it more often and also support successful implementation.

These findings show that acceptance and support behaviours, although being different, are related to each

other. This gives a different view on the acceptance and support dimensions than Van Offenbeek et al

(2013) propose, so the context and/or phase of the implementation programme in which these adoption

behaviours occur seem important.

In this case, all future user groups intended to use the system. Although, it is important to mention

that use of the system is mandatory as all legacy systems will disappear. In other words, future users did

not have a choice but to accept it. Implementers already beard this in mind as quotes related to the

mandatory nature of the system, account for almost half of the acceptance/non-acceptance quotes.

Therefore the support/resistance dimension is interesting as future users do have a choice in whether they

support the new system or not. While all user groups intended to accept the system, support for the

system was significantly lower for all user groups. Para/peri and management showed notable differences

between their scores on acceptance as well as on support, in relation to the other user groups,

management scoring higher on both, para/peri lower on both. These differences might be caused by the

expected degree of future use of the system as para/peri scored lowest and was also expected to use the

system least, vice versa for management. However, the acceptance/non-acceptance and support/resistance

dimensions are not completely unrelated as facilitating conditions and impact are found to positively

relate to both dimensions.

Implementers expected quite some resisting user behaviours for all user groups while their

expectations about the degree of acceptance was neutral. Implementers expected some issues with the

motivation of future users to use the system. In this case, implementers tried to motivate future users to

use the system by letting them have a say in the design of the system, so using the system, will be

rewarding for them. At the time (i.e. pre-implementation) this seemed to work, as all user groups intent to

accept the system. One of the reasons EPR systems do not fulfil their potential is that EPR systems are

not able to adapt to clinical practices. Healthcare professionals find that the systems do not meet their

needs and require workarounds in order to complete work procedures (Jensen and Aanestand, 2007).

Giving future users a say in the design of the system might solve this, but the proof of the pudding will be

in the practicability of their desires.

21

intended user behaviours became apparent to the implementers. They seemed to close the ‘gap’ between

their expectations and intended user behaviours, in favour of the intended user behaviours, which seemed

logical as these were more positive. Doubt and updating revised the expected behaviours of the

implementers up to a point where they ‘understood’ the intended user behaviours.

Because the expected user behaviours by the implementers were still global and diffuse, the

actions of the implementers on the apparent intended user behaviours were primarily aimed at acquiring

more information to concretise, and translate, their expectations of the intended user behaviours into

measurable values as to have a baseline for future state-of-mind monitors. The still early phase of the EPR

implementation programme, pre-implementation, may account for the lack of concrete expectations of

future user behaviours by the implementers. Although this research showed that they did already took the

intended user behaviours into account, hence the state-of-mind survey. Although no interventions were

planned, based on the intended user behaviours, this is explicable as the intended user behaviours

surpassed the expectations of the implementers and the programme was still in the pre-implementation

phase. It seems that sensemaking is of influence on the interaction between intended and expected user

behaviours in that it is used to close a gap between the intended and expected user behaviours, although

the evidence is too sparse to draw inferences to other contexts.

Theoretical contribution

This study contributes to IS adoption theory by confirming the multiple dimensionality of acceptance and

resistance behaviours as proposed by Van Offenbeek et al. (2013). Although both support and acceptance

dimensions are found to be related as this research provides evidence that in the case of an EPR

implementation, the facilitating conditions, i.e. the degree to which an infrastructure exists to support use

of the system, and impact, i.e. how well the system creates interplay with initial individual and group

level conditions, are of influence to intended support/resistance behaviours as well as intended acceptance

behaviours. In a broader context this would partly contradict the findings of Van Offenbeek et al. (2013),

however in the specific case of an EPR implementation these behaviours make sense as for impact an

EPR might bring about major changes in the work processes of healthcare professionals, which seem to

stretch from individual usage to contextual issues. This new view is supported by Joshi’s (1991)

Equity-Implementation model, which states that an user looks at the changes in equity for him or herself in

relation to other users when welcoming or resisting a change, including contextual consequences. In

understanding the relation between impact and acceptation of an EPR, the normalization process theory of

May and Finch (2009) might help as this theory states that for an IS to be adopted the system needs to be

embedded in social work practices. In this theory also the cognitive aspect (what users think) plays a role

in the normalization process. An important aspect in IS acceptance, also mentioned by Van Offenbeek et

al. (2013) is the positive affect and awareness of the IS. If users believe the EPR will be instrumental to

their jobs, they are more likely to use the system.

22

Practical contributions

This study showed that during the pre-implementation phase of an EPR programme implementers’

expectations about the intended user behaviours are besides overly general, also quite sceptical. The

chance exists that implementers will act upon unsubstantiated or unconfirmed expected user behaviours,

which may differ wildly from the intended user behaviours. Implementers need to check their

expectations, and complement their information, about the intended behaviours in the pre-implementation

phase of the programme, at least before planning interventions as this research shows that future users are

already able to express their intentions. Making the expected user behaviours phase specific and

measurable will assist in monitoring the intended user behaviours throughout the implementation

programme, which will hopefully lead to better planned interventions. Furthermore, if implementers

monitor intended user behaviours in the first phases of an IS implementation programme, they can act

upon undesirable intended user behaviours in an early stage. This is important as Lapointe & Rivard

(2005) recognized that individual behaviours tend to converge into group behaviours over time.

The multidimensionality of acceptance and resistance/supportive behaviours means implementers

have to pay attention to different aspects for future users when trying to implement an IS. Van Offenbeek

et al. (2013) proposed that for the acceptance dimension, the system’s functionality and design are

important and for the resistance/supportive dimension the context in which the system is being used is

important. Because the intended supportive behaviours are lower than the intended acceptance behaviours

for all user groups, in the pre-implementation phase of an EPR implementation programme, implementers

should direct managerial interventions at the context in which the EPR is being used.

Limitations and future research

Although this study yields interesting findings, there were limitations that need to be mentioned. For the

quantitative data a non-response bias might exist, due to a difference in average age between the response

group and the sample. This non-response bias could imply that more people who hold a positive attitude

towards the programme responded to the survey as opposed to cynical employees. A way to prevent this

in future research would be to actively encourage people to respond. Unfortunately, the organisation

under study did not allow this. Furthermore, the support factor resulting from the quantitative data, only

measured the support side of the support/resistance dimension, so resistance was not measured directly.

Although a very low score on the support factor, meaning the programme is not supported by the

respondent(s), indirectly indicates resistance behaviour might exist as the opposite of support.

23

the influence of sensemaking on the interaction between the intended and expected user behaviours can

also be better explored to understand how elements of shared meanings play a role in discussions between

implementers about the intended user behaviours during the implementation phases of a new IS.

Conclusions

This research applied the new two-factor view on user reactions of Van Offenbeek et al. (2013) in the

pre-implementation phase of an EPR pre-implementation programme and confirms the multiple dimensionality of

acceptance and resistance/supportive behaviours. Although facilitating conditions and impact are found to

be antecedents of both support and acceptance, which indicates that although acceptance and resistance

are multidimensional, they are not completely unrelated to each other. Based on the data in this case the

two dimensions are located closer to each other than the framework of Van Offenbeek et al (2013)

presupposes. Although the non-response bias in this research ensures that this cannot be concluded with

certainty. Furthermore, this study set out to find out if implementers have expectations about the intended

behaviours of future users and if these expectations and actual intentions differ from each other. The data

showed that implementers have only global expectations and are more sceptical about the adoption

behaviours of future users, in the pre-implementation phase, than future users actually are. Implementers

planned to act upon differences between their expected user behaviours and apparent intended adoption

behaviours of user groups by gathering more information about the how and why of user’s intentions and

setting up measurable standards. This study contributes to the IS adoption literature stream by showing

the importance of implementers’ expectations of the intended user behaviours in the early phase of an

EPR implementation programme.

Acknowledgments

I thank dr. M.A.G. Van Offenbeek for her committed guidance and approachable stance, this certainly

helped me in conducting my research and writing this paper. Also, I thank dr. J.F.J. Vos for her useful

feedback on my draft version and I am grateful to the persons who helped me to conduct my research at

the research site.

REFERENCES

Asch, S. E. (1956). ‘Studies of independence and conformity: I. A minority of one against an unanimous

majority’. Psychological Monographs, 70, 9 (No. 416).

Avison, D. & Young, T. (2007). Time to Rethink Healthcare and ICT? Communications of the ACM,

50(6), 69-74.

Brown, S.A., Massey, A.P., Montoya-Weis, M.M. & Burkman, J.R. (2002). Do I really have to? User

acceptance of mandated technology. European Journal of Information Systems, 11(4), 283–295.

Burton-Jones, A. & Straub, D. (2006). Reconceptualizing system usage: an approach and empirical test.

Information Systems Research, 17(3), 228–246.

Chin, W.W., Johnson, N. & Schwarz, A. (2008). A fast form approach to measuring technology

acceptance and other constructs. MIS Quarterly, 32(4), 687–703.

Coetsee, L. D. (1999). From resistance to commitment. Public Administration Quarterly, 23(2), 204–222.

Compeau, D.R. & Higgins, C.A. (1995). Computer self-efficacy: development of a measure and initial