individual observations that fall well outside the
overall pattern of the data. If outliers are known to be atypical and not a fair
representation of the treatment it is then permissible to reject the observations and to
regard them as ‘missing’ in any analysis. Sometimes an observation seems very