Small-sample bias and corrections for conditional maximum-likelihood odds-ratio estimators.
Small-sample bias and corrections for conditional maximum-likelihood odds-ratio estimators.
复制标题
DOI：
10.1093/biostatistics/1.1.113
复制
发表时间：
2000-03-01
期刊：
影响因子：
--
通讯作者：
Greenland, S
中科院分区：
文献类型：
--
作者：
Greenland, S
一些小样本校正已被提出的条件最大似然估计的比值比匹配对二分暴露。我在这里对比了几种校正的原理和性能，特别是那些容易推广到多元条件Logistic回归的校正。这些修正或具有信息先验的贝叶斯分析可以作为小样本问题的诊断。点说明了一个小的确切的性能比较，并与一个例子，从电线和儿童白血病的研究。前一个比较表明，小样本偏倚可能比通常认识到的更普遍。
A number of small-sample corrections have been proposed for the conditional maximum-likelihood estimator of the odds ratio for matched pairs with a dichotomous exposure. I here contrast the rationale and performance of several corrections, specifically those that generalize easily to multiple conditional logistic regression. These corrections or Bayesian analyses with informative priors may serve as diagnostics for small-sample problems. Points are illustrated with a small exact performance comparison and with an example from a study of electrical wiring and childhood leukemia. The former comparison suggests that small-sample bias may be more prevalent than commonly realized.