摘要：Extreme value, in Sample Survey Theory, is termed outlier in General Statistical Theory. Extreme value data analysis would yield over-estimation or under-estimation in statistical estimation process. Non-linear data transformation method and Sarndal’s correction factor method in Sample Survey Theory have been confirmed to correct extreme value effect in an estimate. However, since the two methods work towards the same objective, there is need to ascertain the efficient estimate between the two methods. This study has empirically compared the two extreme value correction methods. Results revealed that non-linear data transformation method, though associated with back transformation challenge, had lower Percentage Coefficient of Variation (PCV) over correction factor method. Hence, non-linear data transformation method proved efficient over correction factor method. It was recommended that Survey Statisticians should improve on the Sarndal’s developed correction factor method and/or developed new improved estimators for the correction of extreme value in Sample Survey Theory.
关键词：Extreme value;non-linear data transformation;correction factor;regression estimator;percentage coefficient of variation.
