From the Phd Chat page
Probably the most important issue is about how to anticipate problems and incorporate them within the research design. You may find it odd when, at the very beginning of your design process, your supervisor is asking you what you would do if all or part of the process goes wrong – but it is an essential question. Can you salvage a PhD if you end up with far less data than you anticipated? There is an important balance to be struck between being ambitious and realistic. I think it’s good to make ambitious plans to, say, engage with multiple and disparate literatures, interview 50 people or do a significant piece of survey work. It’s also good to consider the end result if, say, only 20 will speak to you. In most cases, I think a good supervisor would ask you to prepare to mitigate risk.
Let’s take a hypothetical example. You want to answer the ‘what is policy?’ question, and you set up three parts: (1) documentary/ textual analysis to identify what policymakers define as policy; (2) interviews with 30-50 practitioners to determine policy in practice; and, (3) a survey to explore policy outcomes from the perspective of service users. This is super-ambitious and, while task 1 is relatively straightforward (at least if the documents are in the public domain), a lot could go wrong with 2 (it takes ages to set up, conduct, transcribe and analyse interviews, and people may not be forthcoming) and 3 (it takes a long time to design and conduct, few may respond to the survey, or data analysis becomes unmanageable). In that case, a supervisor may advise you to focus on either 2 or 3 – or to accept that, when you engage in both, only one of them may work out. The big question is: if one of them goes wrong, can you still get a PhD from what you have? Or, from the beginning, should you put your efforts into fewer tasks? This is not an easy question to answer, but it is important to ask it at the beginning.