I read this paper as a good example of case study research. Paper builds a model from the concept of the level of difficulty of Clinical Information Systems (CIS) implementation. The model explains the cultural explanations as antecedents of the level of difficulty. Two other building blocks of the model are CIS artifact and implementation practices.
The paper follow’s Yin’s case study approach in hospital settings. It starts answering the problem of the implementation difficulty by deriving 5 propositions (from literature) that can be reduced into two statements: 1) if there is a match between all the user values , CIS characteristics and implementation practices, the implementation will be successful, 2) If there is not a match in characteristics or subgroups (nurses, physicians, administrators) have different user values, by changing characteristics the implementations will be successful. Unit of analysis is the implementation process. For data collection and triangulation semi-structured interviews, project documentation and observation notes were used. The specific interview questions are not published, which makes the it harder to repeat the study. The authors link the data to propositions with within- and cross-case analysis. Criteria for interpreting findings can be found at least in coding scheme under category “effects on CIS implementation process” as codes “hindered” and “facilitated”.
The results show strong support for the propositions that stated that the implementation process is hindered if the values don’t match characteristics or the user values don’t match. Based on the results the authors make 3 more propositions: 1) ambiguity of characteristics (difficulty to learn) hinders the implementation process, 2) mismatch between the consistency of implementation process and actor’s status hinders the implementation process, and 3) implementation practices can affect the degree of saliency of the perspectives.
As noted in the beginning the paper is a good example of case study research. The paper provides a lot of raw data linked to the propositions, which make the research more transparent.