Section 64 The “why question” as a generic method in social research

Up to now we have looked at how to process fragments of causal information. But overlapping that is a prior questions of how to get those fragments in the first place.

Here I’ll briefly highlight what seems to me the minimum research strategy for eliciting causal maps, stripping off other “nice-to-have” elements. It seems to me this minimum research strategy (“the why question”) is interesting in its own right.

The “why question” method: formulating a single question about what influences what, and posing it to many sources or respondents to elicit many fragments of causal maps.

There are various different, related, methodologies in the literature. Markiczy and Goldberg (1995), Trochim (2017), Copestake, Morsink, and Remnant (2019).

64.1 Steps to asking “the why question”

64.1.1 Formulate …

… a single question about what causally influences what. There are various different scenarios.

  • QuIP-like: “What factors influence this important thing, and what influences those factors?”
  • Appreciative: “Tell us a story about the most important positive effect the project had (what were the enabling factors, what were the consequences)”
  • Impact: “Tell us a story about most important positive or negative effects the project had (what were the enabling factors, what were the consequences)”
  • Research: “What do you think are the most important causes and consequences e.g. of this thing [e.g. climate change]”
  • Goalfree (Sensemaker-like) “What is the most important issue right now, and what are its causes and consequences”

In each case, you can tweak in various ways

  • emphasise causes and/or consequences,
  • restrict respondents more or less to predefined items (etic versus emic)
  • specifically ask about causes of causes and/or consequences of consequences (i.e. more than one link away from the main focus – though people often mention these things anyway).
  • specifically ask people for how beneficial (or detrimental) the variables are. For example asking at the end, “… and which of these things are most important to you?”
  • ask people for another example when they have finished the first (and even then ask for a third or fourth example)
  • code only for structure (“X influences Y”) and parameters (“X is a strong but negative influence on Y”) of the causal network, or you can also code information about actual levels of variables (“right now, X is very high”) and even contrasting or counterfactual levels of variables (“X would be low if Z was happening”).

64.1.2 Pose the question

… to several sources or respondents. You can always add, for example, official documents or even your own considered opinion to the list of evidence and perhaps assign them a high level of trust. You’ll usually ask a couple of additional questions about each source like gender, age, status.

64.1.3 Get the answers into a spreadsheet

… with a column with the replies and a columns for the additional questions. (If you have allowed people to go back and provide another example, it would be good to have an additional column to record the ID of the source)

The subsequent steps (Code, Simplify / synthesise / Analyse) have already been covered in earlier sections.

64.2 When would you not want to use this approach?

  • If you can guess the main answers - which / how many sources mention which things, you might be better off just formulating those issues as closed questions
  • If this is a theory or theory of change which has already been discussed a lot, so there is already pretty much a consensus

So this works best with causal structures which we don’t yet know too much about. This probably comes under the buzzword “complexity-aware”.

“To reduce the redundant data, semi-structured CCM (Laukkanen, 1994, 1998) uses a different format. The interviewed participants answer a series of questions about what factors they perceive to influence the focal phenomenon (anchor theme) and what consequences follow from it.”