Answers are grounded in this paper and cite their sources.
The seemingly obvious path to evaluating interaction process components via their correlations with measures of the interaction's outcomes is blocked. Process‐outcome correlations can be positive, negative, or null for reasons that have nothing to do with the process component's efficacy or importance. Describing the interaction's structure and dynamic prosperities is a more promising, though less direct, alternative route.
No takes yet. Share an insight, caveat, or question.
William B. Stiles (1993) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: