INTERACTION BETWEEN DISCRETE CAUSES
INTERACTION BETWEEN DISCRETE CAUSES
复制标题
DOI：
10.1093/oxfordjournals.aje.a113153
复制
发表时间：
1981-01-01
影响因子：
5
通讯作者：
KOOPMAN, JS
中科院分区：
文献类型：
--
作者：
KOOPMAN, JS
当充分成分离散原因模型中没有交互作用的假设成立时，疾病发生率的交互反差(ICDR)始终为零或略为负，ICDR是对无交互作用的相加模型的偏差的度量。由于人们可能假设的离散原因之间的所有物理相互作用来源都暗示着对该模型的偏离，ICDR是筛选因果相互作用的一个很好的参数。当ICDR不同于零时，应寻找偏离充分成分离散原因模型中没有相互作用的假设的具体模型。文中给出了这种模型中相互作用的参数化实例。当研究两个原因时，加法模型与乘法模型不一致；乘法模型中的负交互作用或无交互作用可能代表正交互作用，加法模型中的正交互作用可能代表正交互作用。因此，使用乘性模型来筛选因果交互作用可能会导致关于是否需要为交互作用寻求因果解释的不适当决定。由于乘法模型中的正交互作用意味着加法模型中的正交互作用，所以当乘法模型中观察到正交互作用时，就不会出现这种不恰当的决定。
The interaction contrast of disease rates (ICDR), a measure of deviation from an additive model of no interaction, is always zero or slightly negative when the assumptions of no interaction in the sufficient-component discrete causes model hold. Since all physical sources of interaction which one might postulate between discrete causes imply a deviation from this model, the ICDR is a good parameter to screen for causal interaction. Specific models which deviate from the assumptions of no interaction in the sufficient-component discrete causes model should be sought when the ICDR differs from zero. An example of parameterizing interaction in 1 such model is presented. The additive model is inconsistent with the multiplicative when 2 causes are studied; negative or no interaction in the multiplicative model might represent positive interaction in the additive model might represent positive interaction in the additive model. Use of the multiplicative model to screen for causal interactions could thus lead to inappropriate decisions regarding the need to seek causal explanations for interaction. Since positive interaction in the multiplicative model implies positive interaction in the additive, there will be no such inappropriate decisions when positive interaction is observed in the multiplicative model.