样本容量切实实在定(国外英语资料).doc

样本容量的确定 The determination of the sample size Classification: the Statistics ? In the estimation of parameter interval, there is a certain difference between the estimate and the general parameters, which is generated by the randomness of the sample. In the case of the sample size, the confidence interval will be expanded and the estimated accuracy will decrease if the estimated reliability is increased. To increase the accuracy of an estimate without reducing reliability, only by expanding the sample size. Increase the sample size is, of course, such as manpower, material resources and time limitations, so I need to meet certain precision condition, to determine the sample size as right as possible. A factor affecting the sample size (1) the overall variance (population variance) In the case of the other conditions, the sample size should be larger, and the sample should be smaller. For example, in the estimation of the normal population mean, the sampling mean error reflects the dispersion of the sample mean relative to the population mean. Therefore, when the variance of the population is large, the size of the sample is correspondingly larger, so that it will be smaller to ensure the accuracy of the estimate. (2) the size of the error Allow error refers to allow sampling error, as, for example, between the sample mean and the population mean error can be expressed as, without the permission of the margin of error in the form of absolute value shows the range of sampling error, the error is also known as. Allow illustrates the precision of estimation error, so, in the case of other conditions unchanged, if asked to estimate the high precision and allowed error is small, then the sample size is larger; If the accuracy is not high, the error can be large, and the sample size can be smaller. (3) the probability of 1 minus alpha The probability assurance indicates the reliability of the estimate. So, in the case of other conditions, if you ask for higher reliability,