Causal Reasoning with Continuous Outcomes
Causal Reasoning with Continuous Outcomes
复制标题
具有连续结果的因果推理
作者：
A. A. Rashid;M. Buehner
Twenty experiments investigated how people reason about causal relations where a binary cause (present/absent) influences the continuous magnitude of a target outcome. The experimental design was based on a conceptual mapping of probabilistic influences in binary causation to deterministic influences on continuous effects. Doing so preserved the computational properties related to binary causation, and allowed me to test applicability of well-established causal reasoning strategies in continuous causation. The investigation employed three methods: the first one involved asking participants the standard causal questions on strength rating; the second method asked other participants to make judgments in accordance to counterfactual questions; and the third method required participants to identify the direction candidate cause influenced effect magnitude. Results reveal that when reasoning about binary causes that reduce a continuous outcome magnitude, the support is for proportional reasoning approach, which is conceptually equivalent to the Power PC theory of binary causation. When reasoning aboutcauses that increase a continuous magnitude, however, the results did not converge to any
prominent strategy because of various moderating factors. Moreover, under certain circumstances, reasonsers also appear to adopt a strategy based on a multiplicative reasoning, which has not been documented in the literature before. The evidently low consistency of results within participant and within condition across experiments suggests that neither approach properly explains this type of reasoning.
