POWER CALCULATIONS FOR LIKELIHOOD RATIO TESTS IN GENERALIZED LINEAR-MODELS
POWER CALCULATIONS FOR LIKELIHOOD RATIO TESTS IN GENERALIZED LINEAR-MODELS
复制标题
DOI：
10.2307/2532736
复制
发表时间：
1992-03-01
期刊：
影响因子：
1.9
通讯作者：
OHARA, J
中科院分区：
文献类型：
--
作者：
SELF, SG;MAURITSEN, RH;OHARA, J
最近，赛尔夫和毛里塔尼亚(1988，Biometrics 44，79-86)描述了一种在广义线性模型框架内计算样本大小/功率的方法，该方法基于对分数统计分布的非中心卡方近似。在这项工作中，我们提出了另一种方法来解决同样的问题，它基于对似然比统计量分布的非中心卡方近似。这种方法比以前的方法更容易实现，仿真研究表明，它在更宽的参数值和数据配置范围内比以前的方法更准确。新方法也被应用于1：K配对病例对照研究的权力计算问题，该问题与广义线性模型框架有关，但形式上不在广义线性模型框架之外。
Recently Self and Mauritsen (1988, Biometrics 44, 79-86) described an approach for sample size/power calculations within the framework of generalized linear models that is based on a noncentral chi-square approximation to the distribution of score statistics. In this work, we present another approach to this same problem that is based on a noncentral chi-square approximation to the distribution of the likelihood ratio statistic. This approach is easier to implement than the previous one and simulation studies indicate that it is accurate over a much wider range of parameter values and data configurations than the earlier method. The new method is also applied to the problem of power calculations for 1:K matched case-control studies which is related to, but formally outside of, the generalized linear model framework.