«IMPLEMENTING INFORMATION SYSTEMS TO SUPPORT KNOWLEDGE WORK: AN EXPLORATION OF WORK MOTIFS Davis, Christopher J., College of Business, University of ...»
IMPLEMENTING INFORMATION SYSTEMS TO SUPPORT
KNOWLEDGE WORK: AN EXPLORATION OF WORK MOTIFS
Davis, Christopher J., College of Business, University of South Florida St Petersburg.
Hunagel, Ellen M., College of Business, University of South Florida St Petersburg.
The research presented in this paper is part of a larger study of the organizational impacts of information systems on knowledge work. The phase of the research reported in the paper explores current understanding of information systems implementation issues and asks whether it is adequate to effectively manage the wide range of IS that support knowledge work. Studies that have focused on systems that automate core business functions such as manufacturing, logistics, inventory management, and accounting have improved our understanding of both the range of issues involved and the factors most often associated with successful implementation. However this research questions whether such understanding is sufficient to optimize the use of knowledge assets in the modern economy. As organizations turn their attention to automating knowledge work, it makes sense to reexamine what we know about systems implementation and ask ourselves what new challenges this particular class of systems poses for organizational managers.
The paper demonstrates the existence and significance of shared cognitive schemas or work motifs that influence the way knowledge workers respond to information systems designed to automate their work. Findings from the study of a forensic information system show how these work motifs were used to reduce the level of uncertainty during the implementation of the information system and improve the effectiveness of the knowledge management processes it supported. The paper suggests a number of avenues for future KMS implementation research.
Keywords: restructuring expertise; work motifs; shared cognitive schemas; conversational technology.
1 INTRODUCTIONFor many decades, information systems researchers have investigated a host of issues that affect the success or failure of large scale implementation projects. A majority of those studies have focused on systems that automate core business functions such as manufacturing, logistics, inventory management, and accounting. Through these efforts, our discipline has come to better understand both the range of issues involved and the factors most often associated with successful implementations. As organizations now turn their attention to automating knowledge work, it makes sense to reexamine what we know about systems implementation and ask ourselves what new challenges this particular class of systems poses for organizational managers.
This research focuses on the implementation of a forensic information system in an effort to understand how knowledge workers experience and react to significant changes in their task environment. The data discussed here arise from a longitudinal study involving the implementation of an automated fingerprint identification system in the U.K. This study incorporated multiple methods including Repertory Grid Analysis, which was used to elicit and refine a set of overarching constructs or "work motifs" that could be understood and discussed across organizational levels.
The aim of this paper is to demonstrate the existence and importance of these shared cognitive schemas or work motifs that influence the way knowledge workers interpret and respond to information systems designed to automate their work. Also provided is evidence of how these work motifs were used by project participants -- managers, developers, and users -- to stimulate meaningful discussion of the changes faced by fingerprint experts and to probe more deeply into the issues of greatest concern to these knowledge workers.
The preliminary results of this investigation suggest that understanding the subjective experiences of knowledge workers as they adapt to altered work flows and changes in their work environment is crucial to the successful implementation of new systems and tools to support knowledge workers.
Furthermore, the identification of shared cognitive schemas suggests new ways of eliciting and analyzing these experiences to gain a deeper insight into the process and organization of knowledge work and the criteria used to judge it.
2 THEORETICAL DEVELOPMENT
This research builds on theories of situated learning (McLellan, 1996) which treat organizational knowledge as a critical resource that is embedded in and inseparable from the specific social context in which it occurs. This social context exerts a powerful influence on the ways in which specific knowledge is developed, transformed, shared, and applied (Nidumolu, Subramani et al. 2001; Postrel 2002). From a situated learning perspective, implementation cannot be studied in isolation. To gain a deep understanding of the issues that affect both knowledge workers' responses to the implementation process and the eventual outcomes of that process, we need to tap into "the social meanings that emerge from the process of designing and using information technology" (Sahay and Robey 1996, p258).
Prior research in the interpretivist tradition suggests that social meanings are intimately connected to the ways in which people within a community of practice communicate. Thus, what people say in the context of work has "situational significance" and can only be appreciated in the particular milieu in which they speak (Suchman 1987). The social meanings associated with knowledge work can be especially difficult to study because the people who do this work often develop a unique technical language to describe the essence of their work activities (Barley 1996; Bechky 2003). Researchers and managers who are not conversant in this technical language can easily overlook or misinterpret important issues, or may fail to recognize the true significance of the knowledge workers' concerns.
Furthermore, because work-related knowledge is largely tacit, identifying the criteria subject matter experts use to judge their own work and their work tools can be an especially challenging task.
3 STUDY BACKGROUND AND RESEARCH CHALLENGE
Forensic information systems have evolved rapidly to exploit scientific breakthroughs such as DNA profiling, facial recognition and psychological profiling. A key factor in the success of these systems is the extent to which they extend and exploit the knowledge of the forensic experts who use them.
Fingerprints represent unique knowledge assets in forensic investigation because they can provide positive identification of a criminal directly from material left at the crime scene. Like the majority of knowledge work, fingerprint work is complex and time consuming. It must be conducted with the utmost care in order to avoid mistakes. Because this work is highly technical, it takes 5 years to gain the expertise required to qualify as a fingerprint expert in the UK. Over the years, technologies such as image comparators, coding schemes and databases have enabled fingerprint experts to adapt their work processes so as to increase the likelihood and speed of identification. Automated Fingerprint Recognition (AFR) represents the latest generation of technology to offer opportunities for change to the process and organization of fingerprint work.
The information system studied for this phase of the research was the National Automated Fingerprint Identification System (NAFIS) which, in the few years since it was implemented, has already had a significant impact on crime investigation and detection (Leith 2000). The research reported here spanned the development, implementation and operational use of NAFIS at eight of the 43 police forces in the UK in order to assess its impact on fingerprint work, fingerprint experts, and their managers.
NAFIS consists of a central system; fingerprint bureau front-end systems for the 43 police forces in England and Wales; an integrated communications system connecting the central system to the bureau systems; two training systems; and a test and development system. NAFIS utilizes the Police National Network (PNN), to carry data between the central site and the police forces. The NAFIS Central Site is co-located with the Police National Computer (PNC). PNC contains the national database of criminal records. NAFIS is one of the largest image storage systems in Europe holding up to 90 million images and 3 million scenes of crime (latent) marks. Its search speed is one million fingerprint comparisons per second. The NAFIS architecture is designed to provide a foundation for future growth and technology insertion.
Investigating the impact of NAFIS on the knowledge work it supports represented a significant research challenge due to the scale of the NAFIS architecture and the multi-site context in which it is used. In addition, the methods chosen for this investigation had to provide a perspective that was both broad and deep to ensure that the results not only captured the essence of this particular type of knowledge work, and were potentially generalizable to other types of knowledge work.
4 RESEARCH METHODS
Recognizing fingerprint work as a form of knowledge work suggested that technologies such as AFR would impact the organization at a number of levels. However, in this phase of the research, the primary focus was the knowledge workers who routinely process, analyze, and evaluate fingerprint evidence. In order to optimize the reliability and validity of the data, three data gathering techniques (observation, interview and Repertory Grid Analysis) were used. These techniques provided complementary data sets that were combined in a substantial data archive (Yin 1993; Davis 2001).
Repertory Grid Analysis (RGA) provided a content-free means of gathering data from individuals about their experience of changes to fingerprint work during the implementation of NAFIS.
Three rounds of RGA were conducted with each participant. Using the same protocol each time, participants were asked to explain the repertoire of tasks that comprised their work. The task elements (typically 12-15) were recorded and used to elicit bi-polar constructs (typically 15-20) that differentiated the tasks. Each participant was then presented with a grid with columns and rows labeled with the element and construct names they had given, and asked to rate each task in relation to each construct. Two dimensional cluster analysis was used to re-order the grid using the values in the grid cells. (for a fuller discussion of the technique, see Fransella and Bannister 1971; Shaw 1994;
Stewart 1997; Davis 2001).
The data collected via RGA were analyzed and interpreted by the fingerprint experts themselves via the ‘talkback’ analysis protocol proposed by Thomas and Harri-Augstein (1985). Talkback involved presenting the final cluster analysis results to the participant who provided the original input and asking that individual to assign labels to each construct. Having the participants themselves explain the significance of the relationship between the tasks and constructs ensures that the researcher's a priori theories and hypotheses do not bias the results. The talkback process elicited a number of superconstructs from the RGA data that enabled fingerprint workers to articulate changes to the process and organization of fingerprint work in terms that could be shared with the developer and management communities.
5 RESEARCH FINDINGS
The RGA analyses identified a substantial number of constructs from individuals’ experience of changes to their work. Although collectively the grids contain over eight hundred constructs, the need to rely on the participants’ own use of language during the RGA and talkback processes means that many are duplicated. The organization corollary to Kelly’s (1955) fundamental postulate supports the use of Thomas and Harri-Augstein’s (1985) talkback protocol to develop a series of overarching constructs (based on the ordinal relationships in the re-ordered grids) that articulate cognitive schemas shared by the fingerprint experts. These schemas used language that made their concerns about changes to their work understandable by non-experts. To date, nine over-arching constructs have been dawn the research data.
• Fingerprint comparison scope (macroscopic-microscopic)
• Fingerprint comparison continuity (inclusion-exclusion)
• Fingerprint image utilization (fingerprint/administrative work balance)
• Automation, autonomy and control
• Performance, training and trainability
• Fingerprint expert status and task competence
• Exploration, discovery and confirmation
• Nature of output
• Awareness and integration (fingerprint process) These super-constructs or work motifs describe the essence of the contextually-based, shared cognitive schemas that provide a basis for the intellectual exchange among participants in the community of practice. Thus, they provide a unique, situation-specific interpretation of the fingerprint workers' perspectives, protocols, and frameworks. From a research perspective, the identification of work motifs enabled issues and concerns articulated through the RGA analyses to be categorized and prioritized. More importantly, however, they provided a deep insight into how these particular knowledge workers judged the value of their work and work processes. Work motifs arise from the transformational process at the heart of knowledge work and represent components in the iterative process of knowledge creation described by Nonaka (1994). They helped to articulate concerns and considerations that were shared between the community of fingerprint experts and the wider organizational community within which they were managed. The following vignette highlights the utility of the fingerprint comparison "scope" and "continuity" motifs.