Atkinson Anthony C
The London School of Economics, London, WC2A 2AE, UK.
Biostatistics. 2003 Apr;4(2):179-93. doi: 10.1093/biostatistics/4.2.179.
The paper compares randomized rules of the biased-coin type for the sequential allocation of treatments in a clinical trial. An important characteristic is the loss, which measures the increase in the variance of parameter estimates due to the imbalance caused by randomization. Simulations are used to find the small-sample distribution of loss. For some rules a simple chi-squared approximation to the asymptotic distribution holds well down to very small sample sizes.
本文比较了临床试验中治疗顺序分配的偏币型随机化规则。一个重要特征是损失，它衡量了由于随机化导致的不平衡而使参数估计方差增加的情况。通过模拟来找到损失的小样本分布。对于某些规则，对渐近分布的简单卡方近似在非常小的样本量下也能很好地成立。
