Abstract

Outliers can have a negative impact on the traditional Horvitz-Thompson (H-T) estimator by increasing its variance, though the estimator remains unbiased. These Outliers can have extremely high or low values, and when combined with large survey weights, they can significantly inflate the overall variance of the survey estimates and make these estimates unreliable for official statistics. The trade-off between minimum variance and bias in survey estimates has been widely investigated. This research, based on data from the Annual Financial Statistics (AFS) survey, presents an in-depth analysis on the impact that outliers in the AFS data have on the point estimates for turnover and their associated standard errors generated by the H-T estimator before and after outliering.