Biometrical Letters vol. 45(2), 2008, pp. 57-64
|
INFLUENCE OF THE AVERAGING OBSERVATIONS ON ELLIPSOIDAL CONFIDENCE REGIONS IN BIVARIATE NORMAL SAMPLES
J.Tarasińska Department of Applied Mathematics and Computer Science, University of Life Sciences in Lublin, Akademicka 13, 20-934 Lublin, Poland, e-mail: joanna.tarasinska@up.lublin.pl |
It is assumed that a large number of observations from a bivariate normal population are given. These can be used for classical statistical inference about the mean. Sometimes the investigator averages data and makes the inference based on this "sample of means". Such an averaging procedure, when not justified by the non-normality of the data, causes loss of information.
The aim of this paper is to establish by how much the quality of an ellipsoidal confidence region based on the "sample of means" is inferior compared with the "raw sample".
The aim of this paper is to establish by how much the quality of an ellipsoidal confidence region based on the "sample of means" is inferior compared with the "raw sample".
bivariate normal distribution, confidence region for mean
