On certain consequences of averaging observations derived from normal population

Rok, strany: 2003, 391 - 397

It is assumed that we have a large number of observations from normal population which can be used in classical statistical inference about the mean of the population (confidence interval, testing hypothesis). Sometimes the investigator initially takes means of a fixed number of observations and makes statistical inference based on this “sample of means”. It is a good solution when he is not sure that the population is normal. But if it is normal, this averaging procedure is superfluous and causes loss of information. The aim of this paper is to check how much the quality of inference based on “the sample of means” is worse in comparison to the quality of inference based on “the raw sample”. It turns out that the length of confidence intervals and the power of tests are rather insensitive to averaging.

ISO 690:

Tarasińska, J. 2003. On certain consequences of averaging observations derived from normal population. In Tatra Mountains Mathematical Publications, vol. 26, no.2, pp. 391-397. 1210-3195.

APA:

Tarasińska, J. (2003). On certain consequences of averaging observations derived from normal population. Tatra Mountains Mathematical Publications, 26(2), 391-397. 1210-3195.