中文摘要：

在进行毒性分析的试验中，反应变量是试验中怀孕母鼠受精卵对毒性反应的可能性大小．笔者的研究目标是探求这个概率变量的一阶矩和剂量之间的函数关系．采用了以下两种方法：一是假设反应个数是服从Beta-二项分布的随机变量，通过矩方法把分布中的未知参数估计出来，从而得到反应个数取某个值时的概率与反应个数的一阶矩及窝内相关性的估计；二是先建立关于试验的剂量-反应模型，然后用极大似然法得到μ（d）=θ[1＋exp（-α-β1d）]^-1＋（1-θ）[1＋exp（-α-β2d）]^-1中的四个参数α，β1，β2，θ的估计α^^，β1^^，β2^^，θ^^．从而得到μ（d）的估计μ^^（d）．

英文摘要：

In the experiment of developmental toxicants consisting of mice, the response variable is the response in implants of pregnant female mice. It is a probability. Our purpose is to study the relation of response and dose, i.e. , to get a regression function μ（d）. One project is to assume the parameters of the Beta- binomial distribution of the response as a variable by moment method. Then we can get the probability when the response number is equal to some integer, the estimator of one moment and the estimator of the intra - litter correlation. The other project given is first to establish a dose - response model μ（d）=θ[1＋exp（-α-β1d）]^-1＋（1-θ）[1＋exp（-α-β2d）]^-1, then the estimators α^^,β1^^,β2^^,θ^^ of the four parameters α,β1,β2,θ can be given by maximum likelihood method.μ^^（ d）, the estimator of μ（ d）, as well as the estimator of the intra - litter correlation is gained.