Encoding people's narratives about what causes what always involves a certain amount of modelling or theory-construction. Most approaches — Causal mapping approaches differ in application, construction, analysis and how they deal with multiple sources. This might involve constructing a codebook of common factors, but it might also involve applying some kind of special logic of causation, for example:
- distinguishing between necessary and sufficient conditions
- identifying special packages of causes which somehow fit together
- coding the strength and/or polarity of causal links
Our experience, together with Bath SDR, of coding thousands and thousands of documents every year — manually and with AI support — is that most people don't use these special features in their language most of the time, not even scientists. You can go through a whole interview trying to work out if each cause is supposed to be a necessary or a sufficient condition of its effect, but you quite likely won't find a single case where the source actually uses the idea explicitly.
So mostly, we say: don't bother.
At Causal Map Ltd, in our consulting practice, we've taken this minimalist approach even further and mostly code initially without any kind of codebook at all.