Page - 211 - in Applied Interdisciplinary Theory in Health Informatics - Knowledge Base for Practitioners

4. What makes up a goodtheory and which theory should be used? Different disciplines tend to have their own perspective of what constitutes a theory and the criteria for a good-quality theory [21]. Whilst there are some criteria that are likely to be universal (e.g., clarity of concepts, causality, testability, generalisability etc.), there are also likely to be other criteria specific to the needs of health informatics. We expect that a better appreciation of what exists in terms of theory will help to spark more attention to what constitutes a good health informatics theory. Evidence from related disciplines (e.g., implementation science, quality improvement) who face some parallel challenges to health informatics, suggest that the choice of theory is often made on an arbitrary basis, and usually based on expediency or previous exposure [7]. Clearly, this question merits attention in a research agenda for theory in health informatics. There are different approaches to how theory is developed and tested. For instance “adaptive” theory can be defined as a combination of pre-existing theory and incoming evidence [35;36]. Adaptive theory approaches can thus be shaped by research evidence, even while the pre-existing theoretical material (framework, concept) is helping to shape the course of evidence gathering. Alternatively, “grounded theory” is based on the notion that theory emerges from the research data. For grounded theory [26; 36] one of the measures for judging the relevance of a theory is whether or not it is comprehensible to the subjects of the research. 5. Limitations In addition to the limitations of scope discussed in section 2, it is helpful also to reflect on the explicit limitations of each theory. Table 3 summarises the limitations identified in each chapter by the authors. Table 3. Explicit limitations of the interdisciplinary theories in this textbook. Theory Limitations General System Theory and Process Mining “Anything could be seen as a system depending on the boundaries you set”. Shannon's Information Theory Need to explicitly model the “noise” that is inherent in the communication model. Shannon entropy, relative entropy and conditional entropy are non- intuitive concepts. Information Value Chain Theory Relatively new, with few applications. The theory does not attempt to provide detailed mechanistic explanations for the impact of information technology beyond the causality implied in the structure of the chain itself. As with any theory that relies on quantitative measurements, it is important to ensure that data used in any analysis actually measures what it is meant to. UserCentred Design and Activity Theory Software application needs to collect and infer relevant contexts to understand the user's situation. Users will invariably have different perceptions, understanding and expectations, influenced by social, cultural and historical context. P.J.Scott etal. /ReflectingandLooking to theFuture 211