Table 1.
An overview of four common accounts of causation
| (1) Regularity account | (2) Manipulability account | (3) Mechanism-based account | (4) Intra-action-based account | |
|---|---|---|---|---|
| Line of inquiry | Which later events are associated with the specified causes, or which earlier events are associated with the specified effects? | Does the effect change when the cause is manipulated? | How does the cause bring about the effect? | Why did a cause emerge in a particular way? How does this influence its effect? |
| Underlying basic idea of causation | Causation involves regular succession between types of events. The best way to learn about causal relations is to look for regularities between variables | Regular succession is not sufficient for causation; causal dependence can be best demonstrated with a controlled experiment. If we can produce B by producing A, A is a cause of B | Identifying regularities or causal dependencies is not enough to establish causation. We should also understand how the cause brings about the effect. If we understand the mechanisms, i.e., causal chains and configurations, that underlie observed or experimentally produced regularities, we can provide better explanations and better-justified causal claims | Understanding causation requires understanding why elements materialize the way they do. This draws attention to the material processes and discourses that together shape what elements are and what they can do. It also emphasizes that elements do not pre-exist an analysis but are constituted within the process |
| Key causal concepts | Independent variable, dependent variable, variance | Intervention, sensitivity, invariance, counterfactual dependency, confounders | Mechanism, causal process, generative mechanism, interdependent behavior, internal organization | Intra-action, process, relation, phenomena |
| Common methods | Multivariate statistical analysis | Randomized control trials, controlled experiments, observations about “natural experiments” | Multiple methods from (1) and (2), process-tracing, agent-based modeling | Multiple, often qualitative, methods that allow uncovering material-discursive practices |
| References | Suppes (1970), Hume (1981) | Pearl (2000), Holland (2001), Woodward (2006) | Craver (2007), Hedström and Ylikoski (2010) | Barad (2007) |