next up previous
Next: The underlying principle Up: Lectures Previous: Lectures


Situation of resampling in contemporary statistics

Fisher's Classical Paradigm

This will be explained as a flow chart showing where Fisher's initial hypotheses come in and how the data are studied under the conjectured distribution : usually multivariate normal. There are several books on Multivariate Analaysis that follow this paradigm and I will not go into any details, but it makes sense to bear in mind that if one does know from previous studies for instance that the data will be normal multivariate, there is only a simple estimation problem left, all the possible information from the data is summarised by the mean and variance covariance matrix and no more is to be said.

Tukey and Mallow's Exploratory-Confirmatory Paradigm

Susan Holmes 2004-05-19