论文部分内容阅读
四、样本例数估算样本例数估算是在保证科研结论具有一定可靠条件下,确定最小观察例数。一般样本例数估算见文献[7]。现介绍常用的两样本比较的样本例数估算。(一)两样本均数比较按正态近似原理估算公式为:=(u_α+u_β)~2[(K+1)/K]σ~2/δ~2 (4.1)式中总体方差σ~2可用样本方差 s~2估计,s~2=(s_e~2+ks_c~2)/(1+K),差值δ=|_e-_c|,_e、_c与 s_e、s_c 分别为试验组、对照组的均数、标准差,试验组样本例数为 n,对照组样本例数为 kn,当 k=1时两组样本例数相等。一般按正态分布估算的样本例数加2,即近似用 t 分布估算的样本例数。例2某医师研究吲酰胺治疗高血压的疗效,经预试验得治疗前后舒张压差值(kPa)资料如下(与安慰剂比较,两药治疗前后差异均有显著性,P<0.01),当α=0.05,β=0.1时需治疗多少例可以认为吲酰胺有效(表2)。
IV. Estimation of sample size The estimation of sample size is to determine the minimum number of observations under the condition that the scientific research conclusions are reliable. For the estimation of the number of typical samples, see [7]. Now we introduce the estimation of the number of sample cases that are commonly used for comparing two samples. (a) Comparison of the mean of the two samples is estimated by the normal approximation principle: =(u_α+u_β)~2[(K+1)/K]σ~2/δ~2 (4.1) where the overall variance σ~ 2 Available sample variance s~2 estimation, s~2=(s_e~2+ks_c~2)/(1+K), difference δ=|_e-_c|, _e, _c and s_e, s_c are experimental groups, respectively The mean and standard deviation of the control group were n in the experimental group and kn in the control group. When k=1, the number of samples in the two groups was equal. Normally, the number of sample cases estimated by the normal distribution is increased by 2, that is, the number of sample cases approximated by the t-distribution. Example 2 A physician studied the efficacy of guanamine in the treatment of hypertension. The diastolic pressure difference (kPa) before and after treatment was obtained as follows (Compared with placebo, there was significant difference between the two drugs before and after treatment, P<0.01). The number of cases to be treated when α=0.05 and β=0.1 can be considered as effective as decanecarboxamide (Table 2).