From (Powell et al. 2024)

How to collect causal claims from which to draw causal maps?

(There is also a whole chapter about this task: Task 1 — Introduction)

There are a wide variety of options, including in-depth individual interviews (Ackermann et al. 2004), reuse of open-ended questions in structured surveys (Jackson & Trochim 2002), literature reviews (in which ‘sources’ can be documents rather than individuals) and archival or secondary material within which pre-existing causal claims are already made (Copestake, 2020). Other approaches aim to build consensus by using structured collaborative processes, including Delphi studies and PSM (Penn & Barbrook-Johnson 2019). Guidelines for causal mapping may include procedures for collecting primary data, with forms of elicitation including back-chaining (‘what influenced what?’) and forward-chaining (what resulted, or could result, from this?)

With primary data collection, we can distinguish between relatively closed and open approaches and whether respondents are forced to choose between pre-selected optional answers or can formulate their own (see Table 2). Interviewers may also be guided by a chaining algorithm; for example, they may be instructed to iteratively ask questions like ‘You mentioned X, please could you tell me what were the main factors that influenced X or led to it happening.’

Table 2. Different approaches within primary data collection for causal mapping, with example questions.