In message <9921C1171939D3119D860090278AECA20206DC14@EXCHANGE>, Sarah Delaney <[log in to unmask]> writes >Try triangulation - use a number of different approaches - if patterns agree >using a number of different methods you have a strong case? > Although I'm relatively new to qualitative research I've been thinking about issues such as validity and reliability for a long time. As I become familiar with different research paradigms I am more and more struck by how each paradigm would see the reality differently. In the case of the patients with atrial fibrillation one way of examining the decisions would be to accept a positivist view - "the correct treatment for certain patients with atrial fibrillation is warfarin" - in other words that the evidence for effectiveness is true and should be applied. I could then compare real decisions with the recommendations of a guideline and interview GPs where they conflict. I could ask GPs to take a test or questionnaire to measure their knowledge of the evidence and see whether their score correlates with their proportion of "correct" decisions. I could see whether "wrong" decisions correlated with any particular characteristics of patients such as age, sex or co-morbidity. (this is not a joke - I know people who would think this a perfectly sensible way of going about things). I could also interview the GPs, do a straightforward thematic analysis, and then say that I have used triangulation. The problem is that I don't think this would improve understanding at all, and would certainly change the "reality" that I ultimately reported. By looking for ways to triangulate I would already have changed the way I looked at the problem so that the question itself changes from "how do GPs make decisions...." to "what factors influence GPs' to make incorrect decisions..." So, because each paradigm (and therefore method) necessarily views reality in a different way, I am wary of triangulation. I think that the strongest studies are those that stick to one paradigm whatever that is - but that one paradigm is never sufficient to explain "everything". As a researcher I need to find the paradigm that is closest to the "reality" that I want to investigate. If I want to find out the most effective treatment for a disease then the strongest research design is a randomised controlled trial - firmly positivist. But, as a user of research in practice, if I want to see what is the most effective treatment for an individual with a certain disease then my use of "positivist" evidence from an RCT must be post-positivist, as one is translating population effects to an individual and the effects for that individual can only be expressed as probabilities. If I want to study how the decision is made, then I must use a constructivist paradigm. One of the major issues in "service delivery" research in the health services is the place of different methods and paradigms. Developers of guidelines use RCTs to produce recommendations for each particular condition ("patients with angina below the age of 65 should be prescribed...") and use RCTs to measure the effects of implementation programmes, with outcome measures being "proportion of patients on drug X in intervention group compared to control". This approach suits the Department of Health very well as it produces a means of measuring "performance indicators" and fits in with a rational technical approach to quality control. However many of us feel that service delivery is complex, messy and ambiguous, and that even the management of a particular disease, for which there appears to be strong evidence for effective management from RCTs, has many ramifications to do with individuals' culture, their working environment, their personal circumstances and so on. Since the social environment is so complex, is it not better to focus on a particular viewpoint, explore it in depth and be explicit about its narrow focus? My own preference is look at a wider area (in this case "use of research evidence in primary care practice") by asking questions that are narrowly but clearly focussed one at a time, and accepting that while each question requires a clear theoretical stance, understanding of the social environment as a whole requires many questions to be studied, each in its appropriate paradigm. It's a bit like putting a book of photographs together about "Newcastle upon Tyne". One photographer might produce beautifully crafted architectural studies, another spontaneous shots of the club scene, another might spend a day at the races, and yet another might spend weeks getting to know people in a deprived neighbourhood in order to illustrate their lives. All would show the "reality" of Newcastle, but all would be different and none would show the whole - nor (and this is really my point) is it likely that any one photographer would be able to show all these aspects of Newcastle equally well, particularly if they tried to do it as a single project. Toby -- Toby Lipman General practitioner, Newcastle upon Tyne Northern and Yorkshire research training fellow Tel 0191-2811060 (home), 0191-2437000 (surgery) Northern and Yorkshire Evidence-Based Practice Workshops http://www.eb-practice.fsnet.co.uk/